Scolaris Content Display Scolaris Content Display

Intervensi menyokong pesakit penyakit jantung koronari untuk kembali bekerja

Contraer todo Desplegar todo

Background

People with coronary heart disease (CHD) often require prolonged absences from work to convalesce after acute disease events like myocardial infarctions (MI) or revascularisation procedures such as coronary artery bypass grafting (CABG) or percutaneous coronary intervention (PCI). Reduced functional capacity and anxiety due to CHD may further delay or prevent return to work.

Objectives

To assess the effects of person‐ and work‐directed interventions aimed at enhancing return to work in patients with coronary heart disease compared to usual care or no intervention.

Search methods

We searched the databases CENTRAL, MEDLINE, Embase, PsycINFO, NIOSHTIC, NIOSHTIC‐2, HSELINE, CISDOC, and LILACS through 11 October 2018. We also searched the US National Library of Medicine registry, clinicaltrials.gov, to identify ongoing studies.

Selection criteria

We included randomised controlled trials (RCTs) examining return to work among people with CHD who were provided either an intervention or usual care. Selected studies included only people treated for MI or who had undergone either a CABG or PCI. At least 80% of the study population should have been working prior to the CHD and not at the time of the trial, or study authors had to have considered a return‐to‐work subgroup. We included studies in all languages. Two review authors independently selected the studies and consulted a third review author to resolve disagreements.

Data collection and analysis

Two review authors extracted data and independently assessed the risk of bias. We conducted meta‐analyses of rates of return to work and time until return to work. We considered the secondary outcomes, health‐related quality of life and adverse events among studies where at least 80% of study participants were eligible to return to work.

Main results

We found 39 RCTs (including one cluster‐ and four three‐armed RCTs). We included the return‐to‐work results of 34 studies in the meta‐analyses.

Person‐directed, psychological counselling versus usual care

We included 11 studies considering return to work following psychological interventions among a subgroup of 615 participants in the meta‐analysis. Most interventions used some form of counselling to address participants' disease‐related anxieties and provided information on the causes and course of CHD to dispel misconceptions. We do not know if these interventions increase return to work up to six months (risk ratio (RR) 1.08, 95% confidence interval (CI) 0.84 to 1.40; six studies; very low‐certainty evidence) or at six to 12 months (RR 1.24, 95% CI 0.95 to 1.63; seven studies; very low‐certainty evidence). We also do not know if psychological interventions shorten the time until return to work. Psychological interventions may have little or no effect on the proportion of participants working between one and five years (RR 1.09, 95% CI 0.88 to 1.34; three studies; low‐certainty evidence).

Person‐directed, work‐directed counselling versus usual care

Four studies examined work‐directed counselling. These counselling interventions included advising patients when to return to work based on treadmill testing or extended counselling to include co‐workers' fears and misconceptions regarding CHD. Work‐directed counselling may result in little to no difference in the mean difference (MD) in days until return to work (MD −7.52 days, 95% CI −20.07 to 5.03 days; four studies; low‐certainty evidence). Work‐directed counselling probably results in little to no difference in cardiac deaths (RR 1.00, 95% CI 0.19 to 5.39; two studies; moderate‐certainty evidence).

Person‐directed, physical conditioning interventions versus usual care

Nine studies examined the impact of exercise programmes. Compared to usual care, we do not know if physical interventions increase return to work up to six months (RR 1.17, 95% CI 0.97 to 1.41; four studies; very low‐certainty evidence). Physical conditioning interventions may result in little to no difference in return‐to‐work rates at six to 12 months (RR 1.09, 95% CI 0.99 to 1.20; five studies; low‐certainty evidence), and may also result in little to no difference on the rates of patients working after one year (RR 1.04, 95% CI 0.82 to 1.30; two studies; low‐certainty evidence). Physical conditioning interventions may result in little to no difference in the time needed to return to work (MD −7.86 days, 95% CI −29.46 to 13.74 days; four studies; low‐certainty evidence). Physical conditioning interventions probably do not increase cardiac death rates (RR 1.00, 95% CI 0.35 to 2.80; two studies; moderate‐certainty evidence).

Person‐directed, combined interventions versus usual care

We included 13 studies considering return to work following combined interventions in the meta‐analysis. Combined cardiac rehabilitation programmes may have increased return to work up to six months (RR 1.56, 95% CI 1.23 to 1.98; number needed to treat for an additional beneficial outcome (NNTB) 5; four studies; low‐certainty evidence), and may have little to no difference on return‐to‐work rates at six to 12 months' follow‐up (RR 1.06, 95% CI 1.00 to 1.13; 10 studies; low‐certainty evidence). We do not know if combined interventions increased the proportions of participants working between one and five years (RR 1.14, 95% CI 0.96 to 1.37; six studies; very low‐certainty evidence) or at five years (RR 1.09, 95% CI 0.86 to 1.38; four studies; very low‐certainty evidence). Combined interventions probably shortened the time needed until return to work (MD −40.77, 95% CI −67.19 to −14.35; two studies; moderate‐certainty evidence). Combining interventions probably results in little to no difference in reinfarctions (RR 0.56, 95% CI 0.23 to 1.40; three studies; moderate‐certainty evidence).

Work‐directed, interventions

We found no studies exclusively examining strictly work‐directed interventions at the workplace.

Authors' conclusions

Combined interventions may increase return to work up to six months and probably reduce the time away from work. Otherwise, we found no evidence of either a beneficial or harmful effect of person‐directed interventions. The certainty of the evidence for the various interventions and outcomes ranged from very low to moderate. Return to work was typically a secondary outcome of the studies, and as such, the results pertaining to return to work were often poorly reported. Adhering to RCT reporting guidelines could greatly improve the evidence of future research. A research gap exists regarding controlled trials of work‐directed interventions, health‐related quality of life within the return‐to‐work process, and adverse effects.

PICO

Population
Intervention
Comparison
Outcome

El uso y la enseñanza del modelo PICO están muy extendidos en el ámbito de la atención sanitaria basada en la evidencia para formular preguntas y estrategias de búsqueda y para caracterizar estudios o metanálisis clínicos. PICO son las siglas en inglés de cuatro posibles componentes de una pregunta de investigación: paciente, población o problema; intervención; comparación; desenlace (outcome).

Para saber más sobre el uso del modelo PICO, puede consultar el Manual Cochrane.

Intervensi membantu pesakit kembali bekerja selepas serangan jantung, pembedahan pintasan atau prosedur sten.

Apakah matlamat ulasan ini?

Tujuan kami adalah untuk mencari dan menganalisis keputusan kajian‐kajian yang menilai program‐program untuk membantu pesakit penyakit jantung kembali bekerja bagi menentukan samada program‐program ini benar‐benar membantu mereka untuk kembali bekerja, dan juga jika program‐program ini menjejaskan kualiti kehidupan atau mempunyai kesan‐kesan yang tidak diingini.

Mesej‐mesej utama

Program rehabilitasi jantung, meamsukkan dua komponen iaitu senaman dan kaunseling, berkemungkinan boleh memendekkan masa untuk kembali bekerja (bukti berkepastian sederhana) dan mungkin menambahkan bilangan pesakit yang kembali bekerja dalam enam bulan pertama selepas serangan jantung, pembedahan pintasan atau prosedur sten (bukti berkepastian rendah), tetapi program‐program ini mungkin mempunyai kesan yang sedikit ataupun tiada ke atas pengembalian bekerja selepas enam bulan. Program‐program yang mengandungi hanya kaunseling atau senaman mungkin memberi sedikit hingga ke tiada perbezaan kepada bilangan pesakit yang kembali bekerja atau dalam tempoh yang diperlukan untuk kembali bekerja (bukti berkepastian rendah hingga sangat rendah).

Apa yang dikaji dalam ulasan ini?

Pesakit yang sedang pulih dari serangan jantung atau dari prosedur untuk menambahbaik penyakit jantung mungkin mempunyai masalah untuk kembali bekerja. Prosedur‐prosedur ini adalah pembedahan pintasan (prosedur pembedahan untuk memintas arteri koronari yang sempit, juga dipanggil graf pintas arteri koronari atau CABG) atau intervensi bukan pembedahan, termasuk sebagai contoh implan sten (yang dipanggil intervensi koronari perkutaneus (PCI)). Kelemahan fizikal dan masalah emosi disebabkan oleh penyakit jantung boleh menyebabkan ketiadaan lama di tempat kerja atau membawa kepada persaraan ketidak upayaan. Keadaan di tempat bekerja mungkin juga menjadikannya sukar bagi pesakit untuk kembali bekerja. Ini boleh mempunyai impak jangka masa lama ke atas kualiti kehidupan mereka. Kami melihat kepada program‐program yang telah memudahkan pesakit kembali bekerja, contoh melalui modifikasi keadaan bekerja mereka, atau memberi perhatian kepada keresahan yang kerap didapati bersama penyakit jantung melalui pendidikan pesakit mengenai kesihatan jantung, membantu mereka untuk bersenam atau mengaplikasikan gabungan kaunseling dan senaman bagi membantu mereka menjadi cukup sihat untuk kembali bekerja.

Apakah keputusan‐keputusan utama ulasan ini?

Kami menemui sejumlah 39 kajian yang melihat kepada pengembalian bekerja dalam kalangan pesakit penyakit jantung dalam program yang direkabentuk untuk menyokong proses pemulihan atau menggalakkan pengembalian bekerja berbanding kepada pesakit yang menerima penjagaan biasa.

Kami tidak menemui kajian yang telah membuat perubahan kepada tempat bekerja atau polisi di tempat bekerja untuk memudahkan pengembalian bekerja, contoh dengan mengurangkan bilangan jam bekerja atau beban kerja pesakit, dan secara beransur‐ansur menambah bilangan jam bekerja dan beban kerja seiring dengan pemulihan kesihatan.

Kami menemui 11 kajian yang menilai program‐program yang memberi pengalamatan kepada ketakutan dan kemurungan yang lazim berlaku bersama penyakit jantung, melalui pengajaran kepada pesakit mengenai penyakit jantung. Kami tidak tahu jika program kauseling dan pendidikan kesihatan ini menambah bilangan pesakit yang kembali bekerja atau memendekkan tempoh pesakit tidak hadir bekerja (bukti berkepastian rendah hingga sangat rendah).

Kami menemui empat kajian menggunakan program yang mengesyorkan bila pesakit penyakit jantung patut kembali bekerja atau memberi kaunseling kepada rakan sekerja untuk mengurangkan kebimbangan mereka mengenai punca serangan jantung dan keupayaan pesakit untuk kembali bekerja. Intervensi kaunseling berlandaskan kerja mungkin memberi sedikit hingga kepada tiada perbezaan kepada tempoh keperluan pesakit kembali bekerja (bukti berkepastian rendah).

Kami menemui sembilan kajian yang memberikan program senaman sahaja. Program senaman mungkin memberi sedikit hingga tiada perbezaan dalam bilangan pesakit kembali bekerja antara enam bulan dan satu tahun (bukti berkepastian rendah) dan mungkin memberi sedikit hingga kepada tiada perbezaan dalam bilangan pesakit bekerja antara satu dan lima tahun atau dalam tempoh diperlukan untuk kembali bekerja (bukti berkepastian rendah).

Kami menemui 17 kajian yang menilai gabungan program‐program senaman dan kaunseling. Program‐program gabungan ini mungkin menambah bilangan pesakit kembali bekerja sehingga enam bulan selepas serangan jantung, pembedahan pintasan atau prosedur sten (bukti berkepastian rendah): untuk setiap lima pesakit yang berdaftar dalam program gabungan rehabilitasi jantung, satu pesakit tambahan mungkin kembali bekerja. Program‐program ini mungkin memendekkan masa yang diperlukan untuk kembali bekerja (bukti berkepastian sederhana) sebanyak lebih kurang satu bulan.

Sehingga bila ulasan ini dikemaskini?

Kami telah mencari kajian yang mempunyai penerbitan sehingga 11 Oktober 2018.

Authors' conclusions

disponible en

Implications for practice

We found low‐certainty evidence that cardiac rehabilitation, including both physical conditioning and psychological aspects, may promote return to work up to six months following coronary heart disease (CHD), but we also found low‐certainty evidence that these programmes may have little or no effect on the proportion of participants returning to work between six months and one year. Due to the very low certainty of evidence found, we do not know if these programmes increase the proportion of participants at work after a year.

Regarding single‐component, person‐directed interventions, we do not know if programmes including only a counselling component make any difference in return to work up to six months or between six months and one year (very low‐certainty evidence). We found low‐certainty evidence that work‐directed counselling alone may result in little to no difference in the time needed to return to work. We found very low‐certainty evidence regarding the effect of physical conditioning programmes up to six months, so we do not know if physical conditioning alone has any effect on return to work. Physical conditioning programmes may result in little to no difference in return to work between six months and one year (low‐certainty evidence).

Implications for research

Our review identified several aspects that future research could address.

Population

In our analysis, pooling the effect estimates of psychological interventions (including health education) and physical conditioning interventions resulted in risk ratios 1.24 and 1.17, respectively, for short‐term return to work, but the pooled confidence intervals were imprecise. According to our power analysis, the pooled confidence intervals for these two results should not have included a null effect 83% to 84% of the time. To find precise estimates of smaller effects 80% of the time with 95% confidence, such as the RR 1.06 we observed for medium‐term return to work following combined interventions, new studies need to recruit altogether 3774 study participants (compared to the 992 study participants included in our analysis). Since sick leave is costly for employers and paid sick leave may be limited or even unavailable for some workers, we consider even small increases in return to work to be relevant. However, detecting small effects requires conducting very large trials.

In addition, we still need high quality studies that directly address the return‐to‐work process and adequately report the vocational status and job characteristics of study participants prior to the onset of CHD. In a subgroup analysis of physical conditioning interventions, we found that physical conditioning lowered the time needed to return to work only among the two study populations where physically strenuous working conditions or blue‐collar occupations were predominant (Analysis 3.4). More information is needed to corroborate this finding and to determine if interventions may be more effective at promoting return to work for certain employee populations.

When working situations are beneficial and supportive of health, return to work can be considered an important component of regaining full health and improving health‐related quality of life. Delayed return to work or early retirement following CHD can have long‐lasting detrimental financial consequences on individuals and their families, especially where social systems are lacking to provide adequate financial support following a prolonged illness. For some people, such financial factors may be the main impulse to decide if and when they will return to work. Additional research is needed to determine if health outcomes are comparable between people who feel compelled to return to work and those who want to return to work of their own accord.

Interventions and comparisons

Additional evidence is also needed to determine if cardiac rehabilitation including both physical conditioning and psychological components truly promotes return to work up to six months following CHD. In addition to containing exercise, as well as anxiety and risk factor education, future combined interventions may also need to develop better ways to assist transitions back into the workforce without inadvertently promoting presenteeism. Returning to work is a complex and multi‐factorial process, and combined interventions that better address work‐related factors, possibly by providing return‐to‐work coordination, could eliminate further barriers to returning to work. Cardiac rehabilitation interventions also need to make accommodations for people who have to or want to return to work. There is a need to concurrently support the recovery process while alleviating any difficulties that can occur during the return‐to‐work process. This may require the development of strategies that improve access to cardiac rehabilitation centres.

None of the studies exclusively considered work‐directed interventions such as stepwise occupational reintegration (SOR). We also found no controlled studies on the effectiveness of coaching by an occupational physician or on the effects of structured communication between occupational physicians, employers, and the cardiac rehabilitation team. Few combined rehabilitation programmes (three studies) mentioned providing individual work‐directed recommendations to patients or employers as part of the rehabilitation programme. Similarly, only a few studies directly addressed the return‐to‐work process by offering a recommendation for when to return to work (three studies) or by counselling patients and their co‐workers to assuage their concerns about working with heart disease (one study). Although studies sometimes reported changes in working status (full versus part time), reductions in working hours seemed to have been initiated by the patients themselves and were not part of the intervention.

In view of the variation of the single interventions implemented to address either physical or psychological condition following CHD, more research is also needed. Effective single interventions are advantageous, because they are cheaper and simpler to organise than the combined interventions and can also take place outside cardiac rehabilitation centres. Studies considering single components of combined interventions also help explain how much return to work is impacted by either focusing on psychological or physical recovery following CHD among study participants with specific risks.

Outcomes

Return to work was often a secondary outcome of the studies, and as such, the results pertaining to return to work were often poorly reported. Providing the complete results of secondary analyses, at least as on‐line supplements (even when the results were not statistically significant), would help future assessments of return to work among people with CHD. Adhering to recommended reporting guidelines for RCTs could also greatly improve the evidence obtained from future research of return to work following cardiac rehabilitation programmes.

A priori registration of protocols in online RCT registries, which would assist in the objective assessment of selective reporting, may already be improving, as we found seven ongoing registered studies. We also encountered difficulties in identifying participant populations with comparable CHD severity due to the greatly varying selection of cardiac health measures and comorbidities reported. Using core outcome sets when assessing cardiac health of study populations will help alleviate this problem.

Summary of findings

Open in table viewer
Summary of findings for the main comparison. Psychological interventions (including health education) compared to usual care for people with coronary heart disease

Psychological interventions (including health education) compared to usual care for people with coronary heart disease

Patient or population: people with coronary heart disease
Setting: hospital/home
Intervention: psychological interventions (including health education)
Comparison: usual care

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with psychological interventions (including health education)

Proportion of participants returning to work in the short term (up to 6 months)
Follow‐up: range 3 months to 4 months

Study population

RR 1.08
(0.84 to 1.40)

375
(6 RCTs)

⊕⊝⊝⊝
Very low1,2,3,4

We do knot know if psychological interventions (including health education) increase the proportion returning to work in the short term (up to 6 months)

63 per 100

68 per 100
(53 to 88)

Proportion of participants returning to work in the medium term (6 months ‐ 1 year)
Follow‐up: range 6 months to 1 year

Study population

RR 1.24
(0.95 to 1.63)

316
(7 RCTs)

⊕⊝⊝⊝
Very low1,2,3,4

We do not know if psychological interventions (including health education) increase the proportion returning to work in the medium term (6 months ‐ 1 year).

63 per 100

78 per 100
(59 to 100)

Proportion of participants at work in the long term (> 1 to < 5 years)
Follow‐up: range 1.5 years to 4 years

Study population

RR 1.09
(0.88 to 1.34)

239
(3 RCTs)

⊕⊕⊝⊝
Low2,3

Psychological interventions (including health education) may make little or no difference in the proportion working in the long term (> 1 to < 5 years)

74 per 100

81 per 100
(65 to 99)

Days until return to work
Follow‐up: range 6 months to 1.5 years

The mean time to return to work was 9.7 days lower
(35.09 lower to 15.69 higher)

125
(2 RCTs)

⊕⊝⊝⊝
Very low1,2,3

We do not know if psychological interventions (including health education) lower the days needed until returning to work

*The risk in the Intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

1Downgraded one level due to substantial heterogeneity that we could not completely explain.
2Downgraded one level due to risk of bias.
3Downgraded one level due to imprecision (pooled confidence interval is wide and includes either a possible appreciable harm or benefit).
4Downgraded one level, because results of a funnel plot indicated possible publication bias.

Open in table viewer
Summary of findings 2. Work‐directed counselling compared to usual care for people with coronary heart disease

Work‐directed counselling compared to usual care for people with coronary heart disease

Patient or population: people with coronary heart disease
Setting: hospital/home
Intervention: work‐directed counselling
Comparison: usual care

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with work‐directed counselling

Days until return to work

The mean time to return to work was 7.52 days lower
(20.07 lower to 5.03 higher)

618
(4 RCTs)

⊕⊕⊝⊝
Low1,2

Work‐directed counselling may result in little to no difference in days until return to work

Adverse effects: cardiac deaths
Follow‐up mean: 6 months

2 per 100

2 per 100
(0 to 8)

RR 1.00
(0.19 to 5.39)

388
(2 RCTs)

⊕⊕⊕⊝
Moderate3

Work‐directed counselling probably results little or no difference in cardiac death rates

*The risk in the Intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

1Downgraded one level due to substantial heterogeneity that we could not completely explain.
2Downgraded one level due to imprecision (two of the four studies did not report the standard deviation).
3Downgraded one level due to imprecision (pooled confidence interval is wide and includes either a possible harm or benefit).

Open in table viewer
Summary of findings 3. Physical conditioning interventions compared to usual care for people with coronary heart disease

Physical conditioning interventions compared to usual care for people with coronary heart disease

Patient or population: people with coronary heart disease
Setting: hospital/home
Intervention: physical conditioning interventions
Comparison: usual care

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with physical conditioning interventions

Proportion of participants returning to work in the short term (up to 6 months)
Follow‐up: range 3 months to 5.5 months

Study population

RR 1.17
(0.97 to 1.41)

460
(4 RCTs)

⊕⊝⊝⊝
Very low1,2,3

We do not know if physical conditioning interventions increase the proportion returning to work in the short term (up to 6 months)

68 per 100

80 per 100
(66 to 96)

Proportion of participants returning to work in the medium term (6 months‐1 year)
Follow‐up: range 0.5 years to 1 years

Study population

RR 1.09
(0.99 to 1.20)

510
(5 RCTs)

⊕⊕⊝⊝
Low1 4

Physical conditioning interventions may result in little to no difference in proportion returning to work in the medium term (6 months‐1 year)

75 per 100

82 per 100
(74 to 90)

Proportion of participants at work in the long term (> 1 to < 5 years)
Follow‐up: range 3 years to 4 years

Study population

RR 1.04
(0.82 to 1.30)

156
(2 RCTs)

⊕⊕⊝⊝
Low1

Physical conditioning interventions may result in little to no difference in proportion at work in the long term (> 1 to < 5 years)

64 per 100

67 per 100
(53 to 84)

Proportion of participants at work in the extended long term (≥ 5 years)
Follow‐up: mean 5 years

Study population

RR 1.83
(1.26 to 2.66)

119
(1 RCT)

⊕⊕⊝⊝
Low5

Physical conditioning interventions may increase the proportion at work in the extended long term (≥ 5 years)

37 per 100

68 per 100
(47 to 99)

Days until return to work

The mean time to return to work was 7.86 days lower
(29.46 lower to 13.74 higher)

430
(4 RCTs)

⊕⊕⊝⊝
Low1 2

Physical conditioning interventions appear to result in little to no difference in mean time to return to work (days)

Adverse effects: cardiac deaths

Follow‐up: mean 4.8 years

8 per 100

8 per 100
(3 to 24)

RR 1.00
(0.35 to 2.80)

285
(2 RCTs)

⊕⊕⊕⊝
Moderate3

Physical conditioning interventions probably do not increase adverse effects (cardiac deaths)

*The risk in the Intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; RCT: randomised controlled trials; RR: risk ratio

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

1Downgraded one level due to risk of bias.
2Downgraded one level due to substantial heterogeneity that we could not completely explain.
3Downgraded one level due to imprecision (pooled confidence interval is wide and includes either a possible appreciable harm or benefit).
4Downgraded one level, because results of funnel plot indicated possible publication bias.
5Downgraded one level because only one study reported the proportion of study participants working five years after the intervention.

Open in table viewer
Summary of findings 4. Combined interventions compared to usual care for people with coronary heart disease

Combined interventions compared to usual care for people with coronary heart disease

Patient or population: people with coronary heart disease
Setting: hospital/home
Intervention: combined interventions
Comparison: usual care

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with combined interventions

Proportion of participants returning to work in the short term (up to 6 months)
Follow‐up: range 2.3 months to 4 months

Study population

RR 1.56
(1.23 to 1.98)

395
(4 RCTs)

⊕⊕⊝⊝
Low1,2

Combined rehabilitation interventions may increase the proportion returning to work in the short term (up to 6 months)

39 per 100

61 per 100
(48 to 78)

Proportion of participants returning to work in the medium term (6 months ‐ 1 year)
Follow‐up: range 6 months to 1 year

Study population

RR 1.06
(1.00 to 1.13)

992
(10 RCTs)

⊕⊕⊝⊝
Low3

Combined interventions may result in little to no difference in the proportion returning to work in the medium term (6 months ‐ 1 year)

72 per 100

76 per 100
(72 to 81)

Proportion of participants at work in the long term (> 1 to < 5 years)
Follow‐up: range 1.2 years to 3 years

Study population

RR 1.14
(0.96 to 1.37)

491
(6 RCTs)

⊕⊝⊝⊝
Very low1,3

We do not know if combined interventions increase the proportion working long term (> 1 to < 5 years)

53 per 100

60 per 100
(51 to 72)

Proportion of participants at work in the extended long term (≥ 5 years)
Follow‐up: 5 years

Study population

RR 1.09
(0.86 to 1.38)

350
(4 RCTs)

⊕⊝⊝⊝
Very low1,3

We do not know if combined interventions increase the proportion working after an extended term (≥ 5 years)

37 per 100

41 per 100
(32 to 51)

Days until return to work

The mean time to return to work in the intervention group was 40.77 days lower
(67.19 lower to 14.35 lower)

181
(2 RCTs)

⊕⊕⊕⊝
Moderate4

Combined rehabilitation interventions probably reduce mean time to return to work (days)

Health‐related quality of life assessed with: Angina Pectoris Quality of Life Questionnaire

The MD for HrQoL was 0.40 (‐0.03 lower to 0.83 higher)

87
(1 RCT)

⊕⊕⊝⊝
Low2,5

Combined interventions may result in little to no difference in HrQoL

Adverse effects: reinfarctions

Follow‐up: mean 3.8 years

10 per 100

6 per 100
(2 to 15)

RR 0.56
(0.23 to 1.43)

265
(3 RCTs)

⊕⊕⊕⊝
Moderate1

Combined interventions likely result in little to no difference in adverse effects

*The risk in the Intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; HRQoL: health‐related quality of life; RCT: randomised controlled trial; RR: risk ratio; MD: mean difference

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

1Downgraded one level due to imprecision (pooled confidence interval is wide and includes either a possible appreciable harm or benefit).
2Downgraded one level due to risk of bias.
3Downgraded two levels due to risk of bias.
4We detected substantial heterogeneity that we could not completely explain.
5Downgraded one level because only one study reported the effects of the intervention on health‐related quality of life.

Background

disponible en

Description of the condition

Coronary heart disease (CHD), also called coronary artery disease or ischaemic heart disease, is a narrowing or blockage of the blood vessels supplying the heart muscles (WHO 2012). The most common cause of CHD is atherosclerosis, which is a build‐up of cholesterol and fatty deposits (called plaques) on the inner walls of these arteries. A myocardial infarction (MI) may be the first manifestation of coronary artery disease, but it may also occur in people with established disease. Cardiac ischaemia, that is restriction in blood supply, can often cause chest pain known as angina pectoris when the myocardium, or heart muscle tissue, is starved of oxygen.

CHD is the most important cause of mortality and morbidity in Western industrialised countries. In 2016, with 9.4 million deaths (16.2% of total deaths, all ages) it was the leading cause of deaths in the world (WHO 2018a). In European countries it accounts for 13.6% of total disability adjusted life years (DALYs) and 7.6% of total DALYs internationally (WHO 2018b).

CHD morbidity has economic as well as social implications. Leal 2006 estimated the total costs for the European Union to be EUR 45 billion in 2003, with 51% incurred in health care, 34% in productivity losses and 15% in informal care. Anxiety and depression are often experienced after MI and can have major effects on quality of life and on return to work (Dickens 2006; O'Neil 2010).

People who have experienced cardiac events face many challenges, such as pain and discomfort, fatigue, anxiety, problems with physical activity, cardiac medication, or concerns about diet (Blair 2014). Furthermore, data from qualitative interviews with young patients show that their disease has an impact on establishing a career, meaningful relationships, family, and financial security, thus negatively affecting mental health and health‐related quality of life (Walsh 2018).

Cardiac rehabilitation plays an important role in the overall clinical management of cardiac patients. The National Institute for Health and Care Excellence (NICE) has defined cardiac rehabilitation as a "coordinated and structured programme designed to remove or reduce the underlying causes of cardiovascular disease, as well as to provide the best possible physical, mental and social conditions, so that people can, by their own efforts, continue to play a full part in their community. A healthier lifestyle and slowed or reversed progression of cardiovascular disease can also be achieved" (NICE 2015). Although physical activity is commonly recommended as a core element for people with MI or other acute coronary syndromes, combined (or comprehensive) cardiac rehabilitation consists of interventions with health education, lifestyle advice, stress management and physical exercise components (NICE 2013; Perk 2012; Piepoli 2014). According to the Agency for Healthcare Research and Quality (AHRQ), the programmes are "designed to limit the physiological and psychological effects of cardiac illness, reduce the risk for sudden death or re‐infarction, control cardiac symptoms, stabilise or reverse the atherosclerotic process, and enhance the psychosocial and vocational status of selected patients" (Wenger 1995; Wenger 2008).

The benefits of cardiac rehabilitation have been examined in several systematic reviews. A recently updated Cochrane Review concluded that exercise‐based cardiac rehabilitation for people with CHD is effective in reducing cardiovascular mortality in medium‐ to long‐term studies, and hospital admissions in short‐term studies, but not total MI or need for revascularisation by means of coronary artery bypass surgery (CABG) or percutaneous coronary intervention (PCI) including percutaneous transluminal coronary angioplasty and stents (Anderson 2016). Both PCI and CABG are used to treat blocked coronary arteries. CABG is a surgical procedure to bypass narrowed coronary arteries, whereas PCI is a nonsurgical procedure that opens blocked or narrowed coronary arteries. Another Cochrane Review that focused on psychological interventions for CHD found that psychological interventions may produce small to moderate reductions in depression and anxiety, and may also reduce cardiac mortality. The authors did not find evidence that psychological interventions reduced the rate of MI or the need for cardiac surgery, or total mortality (Richards 2017; Whalley 2011). A third Cochrane Review stated that there is not enough information available to fully understand the impact of educational interventions on mortality, morbidity and health‐related quality of life of people with CHD (Anderson 2017b; Brown 2011).

Although all patients should be offered a cardiac rehabilitation programme with an exercise component (NICE 2013), the majority of CHD patients eligible for cardiac rehabilitation do not enter into these programmes; this is especially true for women, older people, and people with a lower socio‐economic status (Sunamura 2017).

However, it is not sufficient to focus on mortality and morbidity alone. Returning to work is another important outcome of societal and economic significance, especially for younger patients. Although one goal of cardiac rehabilitation is to improve vocational status, it is not known how effective the various properties of cardiac rehabilitation programmes are at enhancing return to work among people with CHD, nor how effective interventions provided by the occupational physicians or other healthcare personnel are when there is no cardiac rehabilitation. According to Hämäläinen 2004 there are also large variations between countries in what proportion of patients (between 40% and 90%) return to work following a MI.

Returning to work is a complex and multi‐factorial process. It has been shown that there are a variety of predictors of returning to work among patient groups, for example, the medical seriousness of the disorder, work‐related factors, personal factors, national compensation policies, and the structure of the healthcare system (Cancelliere 2016; De Vries 2018; Den Bakker 2018). Recent studies examining generic factors that influence return to work found job control, work ability, perceived good health, higher self‐efficacy, the individual's own prediction of their return to work, high socioeconomic status, return‐to‐work co‐ordination, and multidisciplinary interventions facilitate return to work, while job strain, anxiety, depression, comorbidity, long‐term sick leave, older age and low education were identified to be barriers to returning to work (Cancelliere 2016; Gragnano 2018; Vooijs 2015).

Concerning people with CHD, important predictors of returning to work appear to be cardiac factors on admission to the hospital (heart failure, arrhythmia), recurrent cardiac events, and depression scores during hospitalisation (Bhattacharyya 2007), as well as occupational factors, such as the physical intensity of work (Dreyer 2016). Results of a systematic review suggest that depression recorded between admission and up to two months after discharge predicted poorer return to work six to 12 months after a cardiac event (O'Neil 2010). Furthermore peoples' beliefs and perceptions about their illness are considered key determinants of recovery after MI (Petrie 1996). More recently, a study suggested that when patients are satisfied with their job and perceive their work environment positively, they will be more likely to return to work early (Fiabane 2012). An interview survey of a random sample of 2000 people in the UK revealed that being able to work was judged to be the third most important aspect of quality of life for people suffering from an illness, whereas healthy people viewed it as only the sixth most important aspect (Bowling 1995).

While there is a high interest in increasing return to work, the adverse effects of returning to work too early, also called presenteeism, have to be considered (Järvholm 2012). A study by Kivimäki 2005 from the Whitehall II cohort examined the association between sickness absenteeism and the incidence of serious coronary events. The incidence of serious coronary events among unhealthy employees with no sickness absenteeism was twice as high as among unhealthy employees with moderate levels of sickness absenteeism.

Several authors in various countries have proposed additions or alterations to cardiac rehabilitation programmes that are important for work outcomes. In the Netherlands a new guideline on cardiac rehabilitation has been established which includes occupational checklists for determining the need for intervention (NVVC 2011). These checklists and interventions are based on the Dutch guideline for occupational physicians on how to deal with people with CHD (Verbeek 2006). The guidelines strongly advise to start supporting return to work during cardiac rehabilitation, and not after it has finished.

Usually, cardiac rehabilitation programmes focus on the use of aerobic exercise to restore functional capacity after an acute cardiac event. Also resistance training is nowadays standard practice. If the primary goal is return to work, the training programmes should be based on actual job‐related activities (Mital 2004). For example, studies with measurements of functional capacity requirements of firefighters and of police officers have found that a greater functional capacity is required than that typically attained in traditional cardiac rehabilitation programmes (Adams 2009; Adams 2010).

An example of a work‐directed intervention is the stepwise occupational reintegration (SOR) programme. It is an established instrument in Germany intended to support insured workers currently on sick leave to reintegrate back into work step‐by‐step after long‐term illness of more than six weeks duration (Bethge 2016; Bürger 2011). Another programme has been developed for people who were not able to return to work after finishing their regular cardiac rehabilitation called "Interdisciplinary Support Programme (INA)". INA is a combined support programme consisting of exercise training, health education, psychological intervention and expert advice concerning job‐related problems (Karoff 2000a).

Description of the intervention

Based on the International Classification of Functioning, Disability, and Health model (ICF) by the World Health Organization (WHO 1993) there are three opportunities for interventions to enhance return to work (Verbeek 2006):

  1. better treatment of the disease;

  2. work‐directed interventions; and

  3. person‐directed interventions.

This Cochrane Review aims to assess the effects of interventions directed at people with CHD or their environment, specifically their working environment, or combinations of the two, to enhance return to work.

Work‐directed interventions are defined in this review as: workplace adjustments such as modified work hours, modified work tasks, or workplace modifications and improved communication with or between managers, colleagues and health professionals.

Person‐directed interventions consist of:

  1. Physical conditioning interventions that include any type of physical training and physical exercises, and

  2. Psychological interventions that include any type of intervention such as patient counselling and health education; screening and treatment of comorbid psychological disorders; stress management and relaxation training; social support; and gender‐specific interventions.

How the intervention might work

Person‐directed interventions like physical conditioning interventions and intense, occupation‐specific training aim to equip patients with a level of functional capacity that is necessary to perform work tasks safely and successfully (Adams 2010; Adams 2009). Specific psychological interventions, on the other hand, can help by changing people's perception of their illness such that they see themselves again as capable workers and not just as recuperating patients (Petrie 2002).

Work‐directed interventions aim to facilitate return to work by reducing perceived or actual barriers to returning to work by implementing workplace design changes, pauses, etc.

Why it is important to do this review

A range of programmes has been developed to increase the return to work of people with CHD. There are also large variations between countries in the proportion of people that return to work following an MI (ranging from 40% to 90%) (Hämäläinen 2004). While varying cultural and sociopolitical factors may influence people's decisions to return to work (Perk 2004), the variation between countries also seems to suggest that some programmes may be more effective than others.

A number of Cochrane Reviews (Anderson 2016; Anderson 2017a; Anderson 2017b; Brown 2011; Heran 2011; Richards 2017; Whalley 2011) have already assessed the effects of cardiac rehabilitation consisting of: patient education, exercise and psychological interventions in reducing morbidity and mortality of people with CHD. However, none of these reviews have specifically assessed the effects on return to work, which is the aim of our review.

Objectives

disponible en

To assess the effects of person‐ and work‐directed interventions aimed at enhancing return to work in patients with coronary heart disease compared to usual care or no intervention.

Methods

disponible en

Criteria for considering studies for this review

Types of studies

We included all randomised controlled trials (RCTs) including cluster‐RCTs and quasi‐RCTs irrespective of publication language or publication status. Quasi‐RCTs are controlled trials that use inappropriate randomisation strategies, accompanied by inadequate allocation concealment, and are therefore at higher risk of bias (Higgins 2017).

Due to the difficulties of performing RCTs at workplaces, we originally intended to include controlled before‐after studies (CBAs). CBAs are non‐randomised studies with one group that receives the intervention and a control group that does not. For a CBA study to have been included, data must have been collected contemporaneously, both at baseline and post‐intervention, so that the timing of the study periods for the control and intervention groups are comparable. Although we found a large number of CBAs examining the effects of person‐directed interventions on return to work, none of the CBA studies that we identified used interventions conducted at workplaces. As CBA studies are more prone to bias than RCTs, and because the CBAs that we found did not contribute information on work‐directed interventions, we deviated from the published protocol and excluded CBAs from the review (see Differences between protocol and review). The CBAs excluded from the review can be found in the Characteristics of excluded studies table.

Types of participants

We included studies involving adults (18 years or older) who had been diagnosed with CHD, who experienced a MI, or a coronary revascularisation procedure like CABG or PCI, as well as people with angina pectoris or angiographically‐defined CHD. Within each study, at least 80% of participants had to fulfil these criteria.

Participants should also have been employed (either in paid employment or self‐employed) at the time of diagnosis and on sick leave or otherwise not working at the time of the study because of the CHD. This could have been a subgroup of a trial, but at least 80% of the participants should not have been working at the start of the trial.

Types of interventions

We considered all interventions in the following categories that aim to support the return‐to‐work process with individual or group approaches.

  1. Work‐directed interventions: these can include changes in the work environment, work tasks or working methods such as in a stepwise occupational reintegration (SOR) programme

  2. Person‐directed interventions:

    1. psychological interventions: all psychological interventions, such as counselling and health education; screening and treatment of comorbid psychological disorders; stress management and relaxation training; social support; gender‐specific interventions undertaken by any qualified professional (e.g. psychologist)

    2. physical conditioning interventions: any supervised or unsupervised inpatient, outpatient, or community‐ or home‐based intervention including some form of physical training or physical exercises that is applied to a cardiac rehabilitation patient population

  3. Any combination of the above

We included studies with a control group receiving no intervention, that is, usual care (as described in study reports). We considered studies involving any pharmacotherapeutic or dietary therapies only if both the intervention and control groups received the same treatment.

Types of outcome measures

Primary outcomes

The primary outcome was return to work, including return to either full‐ or part‐time employment, to the previous job, and to the same role or with changes in work status (change of duties, working location, function).

Return to work could be measured either as event data (e.g. return‐to‐work rates, disability pension rates), or as time‐to‐event data (e.g. time span between reporting sick and resumption of work, number of days on sick leave during the follow‐up period).

Secondary outcomes

  1. Health‐related quality of life within the return‐to‐work process, either measured with generic instruments (SF‐36 and SF‐12, EuroQol EQ‐5D™), or with disease‐specific instruments for participants with angina, MI or heart failure (SAQ, QLMI/MacNew, MLHF, MIDAS, CLASP; Thompson 2003)

  2. Number of participants who returned to work and were still working after an extended period of at least one year

  3. Adverse effects

As we encountered a number of studies reporting the number of participants who were still working after five years during the review process, we added working after five years to the list the secondary outcomes.

Search methods for identification of studies

Electronic searches

We searched the following electronic databases through October 2018 to identify potentially relevant studies:

  1. Cochrane Central Register of Controlled Trials (CENTRAL; 2018, Issue 10) in the Cochrane Library;

  2. MEDLINE (PubMed);

  3. EMBASE (OVID);

  4. PsycINFO (ProQuest);

  5. NIOSHTIC (OSH‐UPDATE);

  6. NIOSHTIC‐2 (OSH‐UPDATE);

  7. HSELINE (OSH‐UPDATE);

  8. CISDOC (OSH‐UPDATE); and

  9. LILACS (Virtual Library of Health).

We also searched ClinicalTrials.gov (ClinicalTrials.gov), and the World Health Organization trials portal (www.who.int/ictrp/en/), in May 2018 to identify ongoing trials. We searched all databases from their inception to the present, and we imposed no restriction on language of publication.

The search strategies used for each database and the day of the searches are available in Appendix 1, Appendix 2, Appendix 3, Appendix 4, Appendix 5, and Appendix 6.

Searching other resources

We checked the reference lists of all included studies and key review articles (Anderson 2016; Anderson 2017a; Anderson 2017b; Brown 2011; Heran 2011; O'Brien 2017; Whalley 2011), for additional references. We also contacted experts in the field to identify additional unpublished materials.

Data collection and analysis

Selection of studies

Two review authors (UE, UEW) independently screened titles and abstracts of all the studies we identified as a result of the initial search, and coded them as 'retrieve' (eligible or potentially eligible/unclear) or 'do not retrieve'. Two review authors (PH, AF or PH, JH) also independently screened later search updates. We retrieved the full‐text study reports or publication and two of the review authors (UE, UEW, or JH) independently screened the full‐texts, identified studies for inclusion, and recorded reasons for exclusion of the ineligible studies. We resolved any disagreement through discussion or, if required, we consulted a third person (JA or AS). We identified and excluded duplicates and collated multiple publications of the same study so that each study, rather than each report or publication, was the unit of interest in the review. We recorded the selection process in sufficient detail to complete a PRISMA flow diagram (Moher 2009), and Characteristics of excluded studies table.

We determined the inclusion of articles published in languages other than English or German by having documents professionally translated or with the help of native speakers.

Data extraction and management

We used a data collection form for study characteristics and outcome data, which was piloted on one study in the review. Two of the review authors (UEW, JH, PH) extracted the following study characteristics from included studies.

  1. Methods: study design, total duration of study, study location, study setting, withdrawals, and date of study

  2. Participants: number, mean age or age range, gender, severity of condition, diagnostic criteria if applicable, inclusion criteria, and exclusion criteria

  3. Interventions: description of intervention, comparison, duration, intensity, content of both intervention and control condition, and co‐interventions

  4. Outcomes: description of primary and secondary outcomes specified and collected, and at which time points reported

  5. Notes: references to review for inclusion, funding for trial, and notable conflicts of interest of study authors

Two of the review authors (UEW, PH or PH, JH) independently extracted outcome data from included studies. We noted in Characteristics of included studies if outcome data were not reported in a usable way. We resolved disagreements by consensus or by involving a third person (AF). We extracted multiple publications or reports describing a single study into a single data collection form.

We transferred extracted information into Review Manager 5 (Review Manager 2014), file via Covidence. We originally planned to enter the data directly into Review Manager 5, but during the review we decided to use Covidence to enter and compare extracted data. Two review authors (PH, JH) entered data into Covidence twice and compared entries before importing data into Review Manager 5. A second review author (AF) compared the data presented in the systematic review and study characteristics with study reports for accuracy. Where relevant data were missing or in case of uncertainties, we attempted to contact the authors of the original articles. Articles published in languages other than English, German, or Dutch were translated into English or German for the extraction and 'Risk of bias' assessment.

Assessment of risk of bias in included studies

Two authors (PH, JH) independently assessed the risk of bias in RCTs using the ‘Risk of bias’ tool recommended by Cochrane (Higgins 2017). In case of differences we consulted a third review author (AF). We assessed the risk of bias according to the following domains.

  1. Random sequence generation

  2. Allocation concealment

  3. Blinding of participants and personnel

  4. Blinding of outcome assessment

  5. Incomplete outcome data

  6. Selective outcome reporting

  7. Other bias

We graded each potential source of bias as high‐risk, low‐risk or unclear and provided quotes from the study reports together with a justification for our judgment in the 'Risk of bias' table. We summarised the 'Risk of bias' judgements across different studies for each of the domains listed. We considered blinding separately for different key outcomes where necessary (e.g. for unblinded outcome assessment, risk of bias for all‐cause mortality may be very different than for a patient‐reported health‐related quality‐of‐life scale). If information on risk of bias related to unpublished data or correspondence with a study author, we noted this in the 'Risk of bias' table.

We assessed the risk of bias in cluster‐RCTs with the six domains of the 'Risk of bias' tool as well as recruitment bias, baseline imbalance, loss of clusters, incorrect analysis and compatibility with RCTs randomised by individual.

We originally intended to have two authors (UE, UEW) independently assess the risk of bias in CBAs by using the checklist developed by Downs and Black (Downs 1998). We wanted to only use the items on internal validity and not those on reporting quality or external validity. The instrument has been shown to have good reliability, internal consistency and validity. The thirteen items of the checklist include the domains of the 'Risk of bias' tool recommended in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2017), listed above. We planned to modify the answers to the questions of the checklist so that they would fit the 'Risk of bias' tool as implemented in Review Manager 2014 by using 'high risk', 'low risk' or 'unclear' instead of 1 or 0 as proposed by the checklist authors. Due to their increased susceptibility to bias compared to RCTs, we deviated from our protocol and excluded CBA studies (Differences between protocol and review), making the assessment of bias with the Downs and Black checklist unnecessary.

Measures of treatment effect

We entered the outcome data for each study into the data tables in Review Manager 5 to calculate the treatment effects (Review Manager 2014). We expressed dichotomous outcome data as risk ratios with their 95% confidence intervals (CIs). When overall results were statistically significant, we calculated the number needed to treat for an additional beneficial outcome (NNTB).

For continuous variables, such as the number of days until returning to work, we used the mean difference (MD) when outcome measurements in all trials were made on the same scale. We converted results reported in months or weeks into days. If future updates of this review include studies that measure the same concept with different scales, we will calculate the standardised mean difference (SMD) with its 95% CI.

Unit of analysis issues

We originally planned to analyse data from cluster‐RCTs at the level of the individual by accounting for the clustering by using the intracluster correlation coefficient (ICC), as explained in the Cochrane Handbook for Systematic Reviews of Interventions (Deeks 2017). However, because the cluster‐RCTs that we identified did not report the number or size of clusters, it was impossible to include their results. We were unable to contact the authors of the cluster‐RCTs to obtain this information.

Dealing with missing data

We contacted investigators or study sponsors to verify key study characteristics and obtain missing numerical outcome data where possible (e.g. when a study was identified as abstract only). Where this was not possible, and the missing data were thought to introduce serious bias, we explored the impact of including such studies in the overall assessment of results with a sensitivity analysis (see Sensitivity analysis).

If numerical outcome data such as standard deviations (SDs) or correlation coefficients were missing, and we could not obtain them from the study authors within six weeks of request, we calculated them from other available statistics such as P values and t‐scores, according to the methods described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). In one case, we calculated the SD from the reported range and sample size using a formula for small studies where n ≤ 15 (Hozo 2005). Where only means and sample sizes were available, we imputed SDs from the pooled SD of the other studies in the same comparison group (Furukawa 2006).

Assessment of heterogeneity

We attempted to assess the clinical homogeneity of the results of included studies based on similarity of intervention, outcome and study designs. We did this by considering study populations with similar distributions of gender, severity of CHD, physically demanding occupational groups or alternatively blue‐collar and white‐collar workers as homogeneous.

During the review process, we found that the heterogeneous reporting of occupational characteristics made it difficult to objectively establish which study populations could be considered as having participant populations with similar physically demanding occupational groups. Therefore, we created a definition for categorising studies into groups with similar physically demanding working conditions that was not a part of the original protocol. We defined physically demanding occupational groups as studies where a majority of study participants (more than 50%) worked in physically demanding employment, manual labour or were described as blue‐collar workers. If 50% or fewer participants worked in physically demanding employment, manual labour or were considered blue‐collar workers, we categorised these study populations as having predominantly non‐physically demanding occupations. We considered all other studies not reporting the characteristics of occupations before the incident CHD to have unknown physical demands.

Likewise, the immense variation in how baseline cardiovascular health was reported made it necessary to create an objective framework for determining which studies could be considered to have study populations with similar CHD severity. We created this decision framework during the review process and it was not included in the original study protocol. We examined study exclusion criteria and the most commonly reported cardiovascular baseline characteristics, in order to create a framework for identifying studies with similar distributions of CHD severity. We categorised study populations as having less severe CHD if the study reported:

  1. excluding participants with one or more of the following:

    1. heart failure or systolic dysfunction (i.e. left ventricular ejection fraction (LVEF) < 40%),

    2. unstable or stable angina (often only reported as angina),

    3. positive exercise stress test (i.e. ≥ 2 mm ST segment change, ischaemia) using treadmill or bicycle ergometer,

    4. intracardiac defibrillator (ICD) or atrial fibrillation; or

  2. the study reports that either less than 25% of the participant population had heart failure or the mean LVEF in the study population was more than 40% at baseline.

We included stable angina in the criteria, because studies often used the term angina without explicitly differentiating between unstable and stable anginas. We considered study populations having more severe CHD when patients were not excluded based on cardiovascular criteria and when over 25% of the participant population had heart failure or the average LVEF in the study population was below 40% at baseline. We had a clinical occupational medical doctor specialised in occupational cardiology (JVD) assess and categorise studies that reported excluding participants based on some of the above criteria but including others. We categorised all other studies into a third category of unknown cardiovascular health or CHD severity where we could not determine the severity of CHD from the reported data.

We considered the following interventions as different from each other: work‐directed interventions, physical conditioning interventions, psychological interventions, work‐directed counselling, and combined interventions.

We considered both return‐to‐work outcomes and sick leave‐duration outcomes as similar return‐to‐work outcomes. We planned to combine overall quality‐of‐life outcomes, even if measured with different instruments, with the intention to specifically consider quality of life within the return‐to‐work process. Often studies reported results for subscales or aspects of quality of life (e.g. depression and anxiety) of all study participants, not just study participants in the return‐to‐work subgroups. Similarly, studies also reported adverse events for the entire study populations and not just for participants working prior to returning to work or who were in the return‐to‐work process. Therefore, we presented the results for health‐related quality‐of‐life outcomes and adverse events only for studies where at least 80% of the study participants were eligible to return to work.

For the assessment of statistical heterogeneity, we used the Chi² test with a significance level of P = 0.1 (because of low power of the test in most meta‐analyses), as well as the I² statistic (Higgins 2003). We adopted the values for interpretation proposed in the Cochrane Handbook for Systematic Reviews of Interventions, "0% to 40%: might not be important; 30% to 60%: may represent moderate heterogeneity; 50% to 90%: may represent substantial heterogeneity; 75% to 100%: considerable heterogeneity" (Deeks 2017).

Assessment of reporting biases

Where we were able to pool more than five studies in any single meta‐analysis, we created and visually examined a funnel plot to explore possible small study biases. Asymmetry of the plot may be due to publication bias. Where a sufficient number of studies were available, we additionally tested for funnel plot asymmetry with the test developed by Egger 1997 (Sterne 2017).

Where we detected publication bias, we adjusted for reporting bias using the 'Metatrim' command in Stata. We planned to calculate the failsafe N, which means the estimated number of studies needed to negate the results of the meta‐analysis. However, the results of the analyses where we detected publication bias were not statistically significant.

Data synthesis

Where more than one study provided usable data in any single comparison, we pooled data from studies judged to be clinically homogeneous using Review Manager 5 software (Review Manager 2014), and not version 5.2 as was stated originally in the review protocol. Where studies were statistically heterogenous, we used a random‐effects model; otherwise we used a fixed‐effect model. When using the random‐effects model, we conducted sensitivity checks by using the fixed‐effect model to reveal differences in results. We included a 95% CI for all estimates.

Where there was considerable unexplainable heterogeneity, we refrained from aggregating the studies and instead presented a narrative review.

Where multiple trial arms were reported in a single trial, we included only the relevant arms. Where two comparisons (e.g. intervention A versus usual care and intervention B versus usual care) were combined in the same meta‐analysis, we divided the control group in half to avoid double‐counting.

GRADE and 'Summary of findings' table

We planned to create a 'Summary of findings' table using the following outcomes: return to work, number of participants who were still at work after one year, number of participants still at work after five years, health‐related quality of life, and any adverse effects of interventions, if reported. We expanded the return‐to‐work outcomes to reflect the follow‐up times considered for each of the main comparisons (i.e. up to six months, between six months and one year, number of participants who were still at work after one year, number of participants still at work after five years) as well as the mean time until return to work, and any adverse effects of interventions (i.e. cardiac deaths, total mortality, reinfarctions).

We used the five GRADE considerations (i.e. study limitations, consistency of effect, imprecision, indirectness, and publication bias) to assess the quality of the body of evidence as it relates to the studies that contributed data to the meta‐analyses for the prespecified outcomes. We used methods and recommendations described in Section 8.5 (Higgins 2017), and Chapter 12 (Schünemann 2017), of the Cochrane Handbook for Systematic Reviews of Interventions using GRADEpro GDT software (GRADEpro GDT 2015). We justified all decisions to down‐ or upgrade the quality of studies using footnotes and made comments to aid readers' understanding of the review where necessary.

Subgroup analysis and investigation of heterogeneity

We stratified analyses according to the length of follow‐up and conducted subgroup analyses to examine how the gender of the study populations, physically demanding occupational groups or CHD severity in the study population influenced the impact of the interventions. Given sufficient trials in future updates of this review, we will also perform meta‐regression analyses (using Stata® software) to relate the following study characteristics to their sizes of effect:

  1. study population (age, gender, country);

  2. length of follow‐up;

  3. study date; and

  4. physically demanding occupational groups or alternatively blue‐collar versus white‐collar workers.

As we expect that the quality of the usual care applied in the comparison groups is continually improving over time to include forms of cardiovascular rehabilitation in accordance with available guidelines (Price 2016), we performed meta‐regression analysis considering study date with the Stata package metareg (Stata) for outcomes where five or more studies were available. We also ordered the studies in the forest‐plots according to their publication date to visually assess any change in effect over time.

Sensitivity analysis

We performed sensitivity analysis to see what effect study limitations, that is problems in sequence generation, allocation concealment, or blinding, or incomplete outcome data, or selective outcome reporting, might have had on the results by omitting studies we judged to have a high overall risk of bias from meta‐analyses. We considered studies to have a high risk of bias overall if we judged any of the domains: sequence generation, incomplete outcome data, or selective outcome reporting to have a high risk of bias.

Results

Description of studies

Results of the search

Running our systematic search strategies in the chosen electronic reference databases from inception to 11 October 2018 resulted in a total of 10,335 references. After removing duplicates, we screened 9561 titles and abstracts for eligibility. This title and abstract screening identified 199 records where the full text of the articles needed assessment, and we identified an additional 12 records through other sources. Of these 211 articles, we excluded 156 articles (147 studies) with reasons (see Excluded studies; Characteristics of excluded studies) and we were unable to obtain the full text of six arcticles (see Studies awaiting classification; Characteristics of studies awaiting classification). The qualitative synthesis included 39 RCTs described in 49 articles (see Included studies; Characteristics of included studies), and we included 34 of the 39 studies in the quantitative synthesis of data. Figure 1 depicts the study selection process as a PRISMA flow diagram (Moher 2009). We also identified six ongoing studies through searches of clinical trials registries (see Ongoing studies; Characteristics of ongoing studies).


PRISMA flow diagram of study selection process

PRISMA flow diagram of study selection process

Included studies

We included 39 RCTs in the review (see Characteristics of included studies).

Design

All of the included studies were RCTs. One study applied cluster randomisation (Geissler 1979), and four studies evaluated more than one form of intervention using a three‐armed design (Froelicher 1994; PRECOR 1991; Rivas 1988; Stern 1983).

Sample sizes

Thirty‐eight of the 39 studies (excluding the multicentre WHO 1983 study) randomised altogether 5944 people with CHD into intervention and control arms. The sample sizes of the studies ranged from 10 to 456 participants. Excluding two studies lacking any information on the number of study participants who had been working prior to CHD (Carson 1982; Hämäläinen 1991), the return‐to‐work subgroups of the studies comprised 3660 participants. The studies included in the quantitative analysis randomised altogether 4661 participants and the return‐to‐work subgroups followed up comprised 3290 participants.

Setting

Studies had been conducted mostly in North America and Europe (31 of 38 studies, excluding the international multicentre WHO 1983 study). The countries contributing the most studies were the USA (eight studies), Sweden (five studies), UK (five studies), and Australia (three studies). Finland, the Netherlands, New Zealand, and Norway each contributed two studies. We also found single studies originating from Canada, Cuba, Denmark, France, the former German Democratic Republic, Germany, Italy, Portugal, and Switzerland. Of the 39 included studies, 32 recruited patients admitted to hospitals or cardiac care units, where they were being treated for CHD. Of the remaining studies, three recruited PCI patients (Higgins 2001; Hofman‐Bang 1999; Pfund 2001), one recruited patients before elective CABG (Engblom 1997), one recruited participants from what seemed to be a post‐MI outpatient clinic (Holmbäck 1994), one recruited patients referred to the study by their attending cardiologist (Erdman 1986), and one study recruited study participants among patients surviving the first (possibly in‐hospital) rehabilitation phase (Geissler 1979). Six studies conducted inpatient interventions before the participants were discharged from hospital (sometimes beginning shortly before a planned cardiac procedure). Twenty‐four studies conducted interventions as outpatient programmes, and nine studies began their interventions in the hospital before discharge and continued the intervention with either outpatient rehabilitation sessions or some sort of post‐discharge contact with the participants.

The oldest study was published in 1974 and the most recent study was published in 2017. Six of the 39 included studies first published results in the 1970s. We also observed a peak in study publication in the 1980s (13 studies) and 1990s (12 studies) that subsided in the decades beginning in 2000 (six studies) and 2010 (two studies).

Participants

Most trials (24 of the 39 included studies) included both men and women, where women typically made up a smaller proportion of the recruited participant population. Andersson 2010 was the only study to include only women, whereas 12 studies included only men (Andersen 1981; Bethell 1990; Carson 1982; Engblom 1997; Erdman 1986; Fielding 1980; Geissler 1979; Picard 1989; PRECOR 1991; Vermeulen 1988; WHO 1983; Worcester 1993), and two studies did not report the sex of participants (Bertie 1992; Marra 1985).

Only 15 of the 39 studies provided any information regarding the types of employment prior to the intervention or how many of the study participants worked in physically strenuous jobs. Based on the information provided, we classified six studies as having examined interventions among a study population of predominantly manual (blue‐collar) workers (Dugmore 1999; Haerem 2000; Lidell 1996; Maeder 1977; Vermeulen 1988; Worcester 1993) and nine studies as having considered a more sedentary (white‐collar) working population (Burgess 1987; Engblom 1997; Higgins 2001; Holmbäck 1994; Horlick 1984; Marra 1985; Picard 1989; Pilote 1992; Rivas 1988). The remaining studies did not provide enough information to judge the physical demands of work among the study population.

Most studies had been conducted among people who had suffered an acute MI (34 of 39 studies). Three studies included only PCI patients, one study included CABG patients, and one study included patients who had either suffered a MI or had undergone CABG or PCI (Andersson 2010). The severity of CHD in the participant populations was difficult to assess with the information reported, however we judged 14 studies to have included only participants with less severe CHD (Andersson 2010; Bertie 1992; Burgess 1987; Erdman 1986; Hall 2002; Holmbäck 1994; Maeder 1977; Marra 1985; Oldridge 1991; Pfund 2001; Pilote 1992; PRECOR 1991; Stern 1983; Vermeulen 1988), and 12 studies to have included participants with more severe CHD (Bengtsson 1983; Carson 1982; Dugmore 1999; Engblom 1997; Froelicher 1994; Hofman‐Bang 1999; Petrie 2002; Picard 1989; Pozen 1977; Rahe 1979; WHO 1983; Worcester 1993). Although Pozen 1977 considered participants with less severe CHD separately, we categorised this study in the more severe category but examined the results of both categories separately in subgroup analyses of CHD severity. We could not determine the severity of CHD among participant populations of the remaining 13 studies.

Interventions

We compared interventions to usual care. Usual care for CHD may have sometimes also included some lesser forms of cardiovascular rehabilitation, and participants receiving usual care might have sought other sources of cardiac rehabilitation. Some studies described usual care as having included the provision of brochures on risk factors, individual risk factor counselling or recommendations for physical training, while other studies only described usual care as comprising the clinical care of patients or provided no further description of usual care. Descriptions of the care received by participants included in the control group are included in the Characteristics of included studies tables.

Comparisons

We compared studies according to the type of intervention(s) implemented compared to usual care. We defined categories of intervention comparisons as follows.

  1. Work‐directed interventions versus usual care

  2. Psychological interventions (including health education) versus usual care

  3. Work‐directed counselling versus usual care

  4. Physical conditioning interventions versus usual care

  5. Combined interventions applying both psychological counselling and physical conditioning versus usual care

We included four three‐armed RCTs. One study randomised participants into one of two combined intervention groups with varying intensities of exercise and a control group receiving usual care (Rivas 1988), and three randomised participants into an exercise intervention, a counselling intervention, and usual care groups (Froelicher 1994; PRECOR 1991; Stern 1983). We considered the study arms of the latter studies in the appropriate comparison groups and divided the control groups in half to avoid double counting.

The control group of one included study also received a light exercise programme instead of usual care (Worcester 1993), but the results of this study were comparable to the results of the other exercise intervention studies.

Work‐directed interventions

None of the studies implemented work‐directed interventions at the organisational level, meaning changes in the work environment, work tasks or working methods, or a stepwise occupational reintegration programme.

Person‐directed psychological interventions

Eleven studies examined the impact of psychological counselling, risk factor educational interventions or a combination of both on return to work compared to usual care (Broadbent 2009; Fielding 1980; Figueiras 2017; Haerem 2000; Hanssen 2009; Horlick 1984; Petrie 2002; Pozen 1977; PRECOR 1991; Rahe 1979; Stern 1983). We included in our meta‐analyses the return‐to‐work results for a total of 615 participants receiving psychological counselling interventions or usual care.

Person‐directed work‐directed counselling interventions

Four studies (641 participants) applied work‐directed counselling, either by recommending a time frame for return to work based on the results of a symptom‐limited treadmill test (Picard 1989; Pilote 1992), by recommending a specific workday for return to work (within a week of the counselling session) to participants and their family physicians (Pfund 2001), or by extending the counselling offered to address concerns regarding the causes of the CHD and return to work after CHD to include participants' co‐workers (Burgess 1987).

Person‐directed physical conditioning interventions

Ten studies evaluated the impact of some form of physical conditioning or physical exercises on return to work compared to usual care (Andersen 1981; Bethell 1990; Carson 1982; Dugmore 1999; Froelicher 1994; Holmbäck 1994; Maeder 1977; Marra 1985; Stern 1983; Worcester 1993). We included the return‐to‐work results of 920 participants altogether (nine studies) in our meta‐analyses. We excluded one study from the meta‐analysis because the authors did not report information regarding the number of participants returning to work in each arm of the study (Carson 1982).

Person‐directed combined interventions

Seventeen studies reported return to work following combined cardiac rehabilitation programmes including both counselling and exercise interventions compared to usual care studies (Andersson 2010; Bengtsson 1983; Bertie 1992; Engblom 1997; Erdman 1986; Froelicher 1994; Geissler 1979; Hall 2002; Hämäläinen 1991; Higgins 2001; Hofman‐Bang 1999; Lidell 1996; Oldridge 1991; PRECOR 1991; Rivas 1988; Vermeulen 1988; WHO 1983). We included the return‐to‐work results of 1230 study participants (13 studies) in our meta‐analyses.

We excluded four studies of combined interventions from our meta‐analysis (Geissler 1979; Hall 2002; Hämäläinen 1991; WHO 1983). We excluded Hall 2002 because they did not provide, and we could not obtain, the numbers of participants rejoining the workforce at various time points. We also excluded Hämäläinen 1991 from our meta‐analysis because it was unclear how many study participants had been in employment prior to the MI. We could not include the cluster‐randomised study by Geissler 1979 in our meta‐analysis, because we could not determine the number of clusters and the size of the clusters. We also excluded the WHO 1983 multicentre study from our meta‐analysis because the interventions and study methods varied greatly between centres, details about the study procedures, interventions, and characteristics of study participants of each individual centre were lacking, and results were ‐ at least in part ‐ published elsewhere by the individual studies.

Outcomes
Primary Outcomes

Most of the included studies reported the number or proportion of study participants working at follow‐ups using a subgroup of study participants who were working before their CHD. We did not include studies that did not consider return to work at least as a secondary outcome. In 10 studies, all of the participants were working or on sick leave prior to their CHD (Dugmore 1999; Fielding 1980; Froelicher 1994; Hofman‐Bang 1999; Marra 1985; Pfund 2001; Picard 1989; Pilote 1992; Rivas 1988; Vermeulen 1988). When authors reported the proportion of participants working only as percentages, we calculated the number of participants using the total number of participants in the return‐to‐work subgroups (working before CHD) where this was possible. We could not determine the number of participants working prior to CHD and at the follow‐ups in two studies (Hall 2002; Hämäläinen 1991), and the follow‐up time and number of participants who returned to work was unclear in one study that reported the mean time until return to work (Carson 1982). Although Hall 2002 applied a survival analysis to evaluate differences in return‐to‐work rates, the reported results included only the P values of Wilcoxon and log‐rank tests. Thirteen studies also reported mean time on sick leave or until return to work (Bengtsson 1983; Bethell 1990; Burgess 1987; Carson 1982; Fielding 1980; Hanssen 2009; Higgins 2001; Holmbäck 1994; Maeder 1977; Marra 1985; Pfund 2001; Picard 1989; Pilote 1992).

Secondary Outcomes

The studies reporting adverse effects and aspects of health‐related quality of life often reported results for the entire study population and not just among those eligible to return to work (health‐related quality of life within the return‐to‐work process). Therefore, we considered the adverse effects and health‐related quality of life results only among studies where the population eligible to return to work exceeded 80%.

Health‐related quality of life

For psychological intervention studies where more than 80% of the population were eligible to return to work, one study measured anxiety with a Catell Self‐Analysis Form and nine‐point rating scale (reporting only results of the paired t‐test; Fielding 1980), and a second study measured perceived health with a self‐developed personal adjustment questionnaire (Horlick 1984). We did not find enough studies reporting total health‐related quality of life to perform a meta‐analysis of health‐related quality of life for psychological interventions.

One study of work‐directed counselling assessed aspects of health‐related quality of life within the return‐to‐work process using the Impact of Events Scale, the Taylor Manifest Anxiety Survey, and the Zung Depression Scale at baseline and at the three‐ and 13‐month follow‐ups (Burgess 1987). A second study assessed health‐related quality of life with the EuroQoL Questionnaire at baseline and the four‐month follow‐up, but reported only the baseline values (Pfund 2001). All work‐directed counselling studies included only participants eligible for return to work. We did not find enough studies reporting total health‐related quality of life to perform a meta‐analysis of health‐related quality of life of work‐directed counselling interventions.

Two physical conditioning intervention studies assessed aspects of health‐related quality of life where at least 80% of the study population was considered eligible to return to work; one used the Toronto attitude scale (TAS) and the profile of mood states (POMS) checklists to assess depression, anxiety and vigour or activity, as well as a 10‐item quality‐of‐life questionnaire at the 12‐month follow‐up, stratifying the results according to prognosis (Dugmore 1999), and the other used a self‐report questionnaire on perceived physical performance and psychological well‐being but did not report the individual results (Holmbäck 1994). We were not able perform a meta‐analysis of health‐related quality of life of physical conditioning interventions.

Three studies of combined interventions where at least 80% of the study population was considered eligible to return to work assessed aspects of health‐related quality of life (Bengtsson 1983; Erdman 1986; Hofman‐Bang 1999). One study used the Minnesota Multiphasic Personality Inventory (MMPI) and questions on anxiety (Bengtsson 1983), a second study used a self‐developed well‐being questionnaire at the six‐month and five‐year follow‐ups to measure mean well‐being, feelings of disability, despondency, and social inhibition at the six‐month and five‐year follow‐ups (Erdman 1986), and the third study used the Angina Pectoris Quality of Life Questionnaire (APQLQ), Beck Depression Inventory, and Trait anxiety questionnaires, and reported the mean health‐related quality of life scores at the one‐ and two‐year follow‐ups (Hofman‐Bang 1999). We did not find enough studies reporting total health‐related quality of life scores to perform a meta‐analysis of combined interventions.

Adverse effects

We considered severe adverse effects, such as deaths, reinfarctions, cardiac surgeries, and hospital readmissions reported by studies where at least 80% of the study population was considered eligible to return to work.

Two studies of psychological and educational interventions considered adverse outcomes in study populations where at least 80% of all study participants were eligible to return to work. One reported total mortality up to six months (Broadbent 2009), and the other reported reinfarctions (Fielding 1980). We did not find enough studies to perform a meta‐analysis of adverse effects for psychological interventions.

Two studies assessed cardiac mortality or reinfarction rates up to six months following work‐directed counselling versus usual care (Picard 1989; Pilote 1992).

Three physical conditioning studies reported adverse effects as total mortality (Holmbäck 1994), cardiac deaths or fatal reinfarctions (Dugmore 1999; Marra 1985), and reinfarctions (Holmbäck 1994; Marra 1985) in study populations where at least 80% of all study participants were eligible to return to work (Dugmore 1999; Holmbäck 1994; Marra 1985). We considered fatal MI together with cardiac deaths in one meta‐analysis, and reinfarction rates in a second meta‐analysis.

Studies of combined interventions reported adverse effects as all deaths (Bengtsson 1983; Erdman 1986; Hofman‐Bang 1999; Rivas 1988), cardiac deaths (Vermeulen 1988), hospital readmissions (Hofman‐Bang 1999), or reinfarctions (Bengtsson 1983; Erdman 1986; Vermeulen 1988) in study populations where at least 80% of all study participants were eligible to return to work. We evaluated results for all deaths (total mortality) in one meta‐analysis and reinfarction rates in a second meta‐analysis.

Working after an extended period of at least one year

We found a total of 15 studies reporting on the rates of participants still working after an extended period of at least one year that could be included in a meta‐analysis (Andersen 1981; Andersson 2010; Bengtsson 1983; Bertie 1992; Burgess 1987; Dugmore 1999; Engblom 1997; Erdman 1986; Hanssen 2009; Hofman‐Bang 1999; Lidell 1996; Maeder 1977; PRECOR 1991; Rahe 1979; WHO 1983). Three studies reported extended working rates after psychological counselling and education programmes (Hanssen 2009; PRECOR 1991; Rahe 1979), one study reported extended working rates after work‐directed counselling (Burgess 1987), three studies reported working rates after physical conditioning interventions (Andersen 1981; Dugmore 1999; Maeder 1977), and eight studies reported extended working rates after combined interventions (Andersson 2010; Bengtsson 1983; Bertie 1992; Engblom 1997; Erdman 1986; Hofman‐Bang 1999; Lidell 1996; PRECOR 1991).

Follow‐up

The included studies reported return‐to‐work rates for various follow‐up times, so we categorised results into similar periods of time to examine the short‐term (up to six months), medium‐term (six to 12 months), long‐term (between one and five years), and extended long‐term (five years or longer) effects of the interventions on return to work. Where studies reported results for more than one time point we considered the data for the longest follow‐up in the range. For example, if a study reported the number of participants returning to work at both eight and 12 months, we only included the 12‐month results in the analysis of medium‐term results. Single studies sometimes provided data for more than one follow‐up range.

Five studies considered only shorter follow‐up times up to six months (Broadbent 2009; Froelicher 1994; Marra 1985; Petrie 2002; Pfund 2001), while nine studies reported both short‐term results and at least one additional follow‐up time (Andersen 1981; Bertie 1992; Dugmore 1999; Figueiras 2017; Higgins 2001; Horlick 1984; Rahe 1979; Rivas 1988; Worcester 1993).

A total of 22 studies reported return‐to‐work results for follow‐ups between six and 12 months: eight studies reported the six‐month follow‐up (Dugmore 1999; Erdman 1986; Fielding 1980; Horlick 1984; Picard 1989; Pilote 1992; Pozen 1977; Rivas 1988), eight reported the 12‐month follow‐up (Andersson 2010; Figueiras 2017; Higgins 2001; Hofman‐Bang 1999; Holmbäck 1994; Oldridge 1991; Stern 1983; Worcester 1993), and five studies reported both (Engblom 1997; Geissler 1979; Haerem 2000; Lidell 1996; Rahe 1979).

We also differentiated the secondary outcome of working after an extended period of time of at least one year to consider follow‐ups conducted between one and five years and at five years or later. Thirteen studies reported working rates between one and five years (Andersen 1981; Andersson 2010; Bengtsson 1983; Bertie 1992; Burgess 1987; Engblom 1997; Geissler 1979; Hanssen 2009; Hofman‐Bang 1999; Maeder 1977; PRECOR 1991; Rahe 1979; WHO 1983), and five studies reported results for five‐year follow‐up (Andersen 1981; Dugmore 1999; Engblom 1997; Erdman 1986; Lidell 1996).

Excluded studies

We excluded 147 studies (published in 156 articles), and listed the most critical reasons for exclusion in the Characteristics of excluded studies table. We excluded studies for the following reasons:

  1. 63 studies (64 articles) did not consider the outcome return to work or did not report the return to work results;

  2. 45 studies (53 articles) were not RCTs;

  3. 20 studies lacked a usual care control group;

  4. eight studies lacked an intervention;

  5. seven studies did not meet the requirement that at least 80% of study participants had to be employed at the time of diagnosis and on sick leave because of the CHD, or that study authors considered a subgroup of previously employed study participants (Bar 1992; Cay 1981; Gutschker 1977; Kittel 2008; Nelson 1994; Schiller 1976; Yonezawa 2009);

  6. two studies applied interventions that did not satisfy our inclusion criteria (Heller 1993; Kagan‐Ponomarev 1994);

  7. two study populations did not meet our requirements regarding CHD indications (Christensen 2017; Huber 2014).

Six futher studies reported in six articles are awaiting classification because our library staff could not locate or obtain the full‐text articles (see Characteristics of studies awaiting classification).

Risk of bias in included studies

Due to poor reporting, we often judged studies to have an unclear risk of bias for one or more domains. We considered studies to have an overall high risk of bias if we judged them to have a high risk of bias in any of the following domains: sequence generation (selection bias), incomplete outcome data (attrition bias), or selective outcome reporting (reporting bias). According to these criteria, 15 studies had an overall high risk of bias (Andersen 1981; Andersson 2010; Bertie 1992; Broadbent 2009; Carson 1982; Erdman 1986; Geissler 1979; Hanssen 2009; Hofman‐Bang 1999; Holmbäck 1994; Horlick 1984; Lidell 1996; Pozen 1977; WHO 1983; Worcester 1993). Interventions requiring the active participation of the participants, such as cardiac rehabilitation interventions are difficult, if not impossible, to conduct completely without the knowledge of the study participants (i.e. blinding of participants). Therefore, we did not consider the domains for blinding (performance bias and detection bias) in our determination of the overall risk of bias.

Of the 24 studies we considered not to have an overall high risk of bias, we assigned six studies a low risk of bias for random sequence generation, and low or unclear risk of bias for allocation concealment, blinding of outcome assessors (detection bias), incomplete outcome data (attrition bias) and selective outcome reporting (reporting bias) (Figueiras 2017; Maeder 1977; Petrie 2002; Picard 1989; Pilote 1992; Rivas 1988). We assigned the remaining 18 studies an unclear risk of bias for random sequence generation and low or unclear risk of bias for incomplete outcome data (attrition bias), incomplete outcome data (attrition bias), or selective outcome reporting (reporting bias) (Figure 2).


'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study

'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study

Allocation

We judged three studies to have a low risk of selection bias, that is, they used a suitable random sequence generation method and concealed allocation (Broadbent 2009; Picard 1989; Pilote 1992), and seven further studies used a suitable sequence generation method (Bethell 1990; Erdman 1986; Figueiras 2017; Maeder 1977; Petrie 2002; Rivas 1988; Worcester 1993). Usually studies did not describe the method of random sequence generation and did not mention allocation concealment, and we judged these studies to have an unclear risk of bias. One study was cluster‐randomised, by region according to hospital districts (Geissler 1979). We considered this study to have a high risk of bias, as they did not report the number and size of the clusters or further details regarding the method of randomisation. Also we judged the multicentre WHO 1983 study to have a high risk of bias, because the authors reported that only half of the centres appeared to have achieved suitable randomisation.

Blinding

We gave all but three of the included studies a rating of high risk of performance bias, because the study participants and personnel were aware of the rehabilitation intervention. One exception is the study by Figueiras 2017, where the authors reported that the caregivers did not know the group allocation and we judged the risk of bias to be low. Where the form of the intervention made it less likely that participants in either group would have realised their allocated group, we judged the risk of performance bias to be unclear. This was the case for the studies where follow‐up counselling was provided with telephone calls (Hanssen 2009), and the counselling intervention was integrated within the inpatient care (Petrie 2002).

We gave six studies a low risk of bias rating for blinding of outcome assessors (Engblom 1997; Froelicher 1994; Haerem 2000; Hofman‐Bang 1999; Lidell 1996; Picard 1989). Although only Picard 1989 reported that the data co‐ordinator assessing the outcome (employment status) was not involved with performing the intervention (and presumably blinded to group allocation), we also judged detection bias to be low if work status was obtained from official documents, registries or validated questionnaires (Engblom 1997; Froelicher 1994; Haerem 2000; Hofman‐Bang 1999; Lidell 1996). We judged studies to have a high risk of bias for blinding of outcome assessors if the study descriptions stated that the outcome assessors were aware of the group allocation. This applied to four studies (Bethell 1990; Marra 1985; Oldridge 1991; Worcester 1993).

Incomplete outcome data

We assigned studies a high risk of attrition bias rating if there were unbalanced losses to follow‐up (i.e. over 5%‐point difference between groups), overall attrition exceeded 10% (without information regarding group allocation), information pertaining to the number of participants in the return‐to‐work analyses or follow‐up times for the return‐to‐work analyses were incomplete, study participants who suffered adverse outcomes were excluded from the return‐to‐work analyses, or participants' reported reasons for dropping out of the study could have biased the results, and no intention‐to‐treat analysis was conducted. Altogether this pertained to 11 studies (Andersen 1981; Andersson 2010; Bertie 1992; Broadbent 2009; Erdman 1986; Hanssen 2009; Holmbäck 1994; Horlick 1984; Lidell 1996; Pozen 1977; WHO 1983). We judged 12 studies with low losses to follow‐up and balanced attrition or studies that conducted intention‐to‐treat analyses to have a low risk of attrition bias (Bengtsson 1983; Bethell 1990; Dugmore 1999; Engblom 1997; Figueiras 2017; Geissler 1979; Hall 2002; Marra 1985; Oldridge 1991; Picard 1989; PRECOR 1991; Vermeulen 1988). We judged the remaining 13 studies as having an unclear risk of attrition bias because not enough information was provided to determine if attrition was balanced or there where discrepancies in the reported number of persons followed.

Selective reporting

Reporting bias was difficult to assess, because none of the studies cited any prior study protocol or registration with a clinical trials database. Therefore, we considered studies reporting non‐significant results and without indications of unplanned subgroup analyses as having a low risk of selective reporting bias (Andersson 2010; Vermeulen 1988). We judged the remaining studies to have an unclear risk of reporting bias.

Other potential sources of bias

We evaluated the following additional sources of bias for the Geissler 1979 cluster‐RCT: recruitment bias, baseline imbalance, loss of clusters, incorrect analysis, comparability with individually randomised trials. We included our judgements and the reasons for these judgements in the risk of bias table under other bias. Otherwise, we did not find any other potential sources of bias among the studies.

Effects of interventions

See: Summary of findings for the main comparison Psychological interventions (including health education) compared to usual care for people with coronary heart disease; Summary of findings 2 Work‐directed counselling compared to usual care for people with coronary heart disease; Summary of findings 3 Physical conditioning interventions compared to usual care for people with coronary heart disease; Summary of findings 4 Combined interventions compared to usual care for people with coronary heart disease

1. Work‐directed interventions

We could not consider the effect of work‐directed interventions alone, as we found no studies examining only work‐directed interventions conducted at the organisational level. Only one study integrated a work‐directed intervention into their combined cardiac rehabilitation programme by providing employers with recommendations for work modifications when it was deemed necessary (Bengtsson 1983). We examined the results of this study in the combined interventions category.

2. Person‐directed psychological interventions

2.1 Psychological counselling and risk factor education versus usual care: primary outcomes

Eleven studies examined the impact of psychological counselling, risk factor educational interventions or a combination of both on return to work (Broadbent 2009; Fielding 1980; Figueiras 2017; Haerem 2000; Hanssen 2009; Horlick 1984; Petrie 2002; Pozen 1977; PRECOR 1991; Rahe 1979; Stern 1983).

2.1.1 Short term (less than six months)

Psychological interventions had little or no effect on the proportions of study participants returning to work up to six months (RR 1.08, 95% CI 0.84 to 1.40; I² = 69%; very low‐certainty evidence), and both the I² values and the Chi²‐test (P = 0.007) indicated substantial heterogeneity (Analysis 1.1). The severity of CHD in the study populations seemed to explain some of this heterogeneity (Analysis 1.2). To determine the possible impact of small‐study effects, we also conducted a fixed‐effect meta‐analysis, where smaller studies received less weight. The fixed‐effect model resulted in a summarised RR of 1.04 (95% CI 0.92 to 1.17). When we excluded the two studies with an overall high risk of bias (Broadbent 2009; Horlick 1984), the pooled effect of counselling interventions on the proportion of participants returning to work up to six months was RR 1.13 (95% CI 0.91 to 1.41; I² = 1%). Exclusion of these two studies also seemed to explain much of the observed heterogeneity. Considering changes to study results over time, we found that newer studies appeared less likely to find an effect. However, a meta‐regression considering the linear relationship between publication year and the log RR did not indicate any change in the impact of interventions over time (slope β = 0.006, P = 0.623). The asymmetry of the funnel plot for the short‐term results (Figure 3), indicated a presence of publication bias. However, the results of the Egger's test did not indicate any small‐study effects (P = 0.502), so we did not apply the 'trim and fill' method.


Funnel plot of comparison 2. Psychological interventions (including health education) vs usual care, outcome: 2.3 proportion returning to work (short term)

Funnel plot of comparison 2. Psychological interventions (including health education) vs usual care, outcome: 2.3 proportion returning to work (short term)

2.1.2 Medium term (six months to one year)

Seven studies reported the number of participants in work at follow‐ups from six to 12 months following psychological counselling and risk factor education (Analysis 1.1). These interventions resulted in a pooled RR for medium‐term return to work of 1.24 (95% CI 0.95 to 1.63; I² = 65%; very low‐certainty evidence). The I² value indicated substantial heterogeneity and a P value < 0.1 for the Chi²‐test was detected for the results of the medium‐term follow‐up period. A sensitivity analysis with fixed‐effect analysis to detect small‐study effects lowered the observed RR to 1.03 (95% CI 0.91 to 1.16). As a sensitivity analysis we also excluded the two studies we judged to have an overall high risk of bias (Horlick 1984; Pozen 1977), which increased the RR to 1.40 (95% CI 1.11 to 1.77; I² = 0%). Excluding the studies with an overall high risk of bias also appeared to explain some of the heterogeneity. Since interventions focused primarily on risk factor education may produce smaller effects, we also excluded the one study applying a predominantly informative intervention as a further sensitivity analysis. However, excluding Haerem 2000 lowered the pooled effect (RR 1.20, 95% CI 0.89 to 1.63; I² = 61%).

Visually, the intervention effects on return to work between six and 12 months appear to be decreasing with time (Analysis 1.1). However, the results of the meta‐regression considering the linear relationship between study year and the log RR did not indicate any time‐dependency (slope β = −0.004, P = 0.668).

A funnel plot of the seven studies included also indicated the presence of reporting biases, which was supported by the Egger test (P = 0.034). We applied a 'trim and fill' method to correct for the asymmetry, and the corrected random‐effects estimate was RR 0.97 (95% CI 0.74 to 1.27), after filling with four 'missing' studies. The pooled results of the seven studies were not statistically significant, so we did not calculate a failsafe N. We did not conduct a subgroup analysis for the sex of the study participants, because with the exception of two studies including only male participants (Fielding 1980; PRECOR 1991), all of the studies included both women and men. Similarly, we did not perform any subgroup analyses based on physically demanding occupations, because only two studies reported having study populations with either predominantly physically demanding occupations (Haerem 2000) or less physically active occupations (Horlick 1984). The remaining five studies did not describe the physical demands of the study populations' occupations (Fielding 1980; Figueiras 2017; Pozen 1977; Rahe 1979; Stern 1983).

Subgroup analysis

We also considered return to work at six to 12 months for subgroups of studies with similar severity of CHD, where we considered the low‐ and high‐risk subpopulations of Pozen 1977 separately (Analysis 1.3). Among the two study populations with higher severity of CHD (Pozen 1977; Rahe 1979), we found a summarised effect of RR 1.61 (95% CI 0.97 to 2.67; I² = 43%). The summarised effect was RR 1.17 (95% CI 0.67 to 2.03; I² = 0%) for the subgroup with less severe CHD (Pozen 1977; Stern 1983). Among the studies where we could not determine the severity of CHD among the study participants, we found a summarised effect was RR 1.12 (95% CI 0.82 to 1.53; I² = 67%; Fielding 1980; Figueiras 2017; Haerem 2000; Horlick 1984). The severity of CHD in the study populations also explained some of the heterogeneity, where both the Chi² tests and I² values indicated lower heterogeneity in the subgroups where we were able to classify the general severity of CHD.

2.1.3 Mean days until return to work

Two studies considering psychological interventions also reported the mean or median days until returning to work (Fielding 1980; Hanssen 2009). We pooled these two studies using SDs derived from the reported range (Fielding 1980), or imputed (Hanssen 2009). We observed a pooled MD for time until return to work of −9.70 days (95% CI −35.09 to 15.69, very low‐certainty evidence; Analysis 1.4).

2.2 Psychological counselling interventions versus usual care: secondary outcomes
Working after an extended period of at least one year

We found three studies reporting the rates of people working more than one year and up to four years (long‐term) after hospitalisation (Hanssen 2009; PRECOR 1991; Rahe 1979). Psychological interventions may have little or no effect on the proportion of participants working at follow‐ups between one and four years (RR 1.09, 95% CI 0.88 to 1.34; I² = 44%; low‐certainty evidence). Excluding the one study with overall high risk of bias (Hanssen 2009) from the analysis resulted in a RR of 1.28 (95% CI 0.61 to 2.67; I² = 67%). Pooling with the fixed‐effect model did little to change the summarised effect estimate (RR 1.08, 95% CI 0.94 to 1.23). There were not enough studies to perform a meta‐regression.

2.3 Work‐directed counselling versus usual care: primary outcomes

Four studies applied work‐directed counselling, either by recommending a time‐frame for returning to work based on the results of a symptom‐limited treadmill test (Picard 1989; Pilote 1992), recommending a specific workday for return to work to participants and their family physicians (Pfund 2001), or by extending the offered counselling to participants' social networks (including co‐workers) to address their concerns regarding the causes of CHD and the ability of participants to return to work (Burgess 1987). Due to the variation in follow‐up times, we could not summarise the effects of these interventions on the relative proportions of study participants returning to work (Analysis 2.1).

We pooled the MD in days of the four studies by using imputed SDs for two studies (Burgess 1987; Picard 1989). We observed a pooled MD of −7.52 days (95% CI −20.07 to 5.03; low‐certainty evidence; Analysis 2.2) for mean time until return to work following work‐directed counselling interventions compared to usual care. The results of the four studies showed considerable heterogeneity (Chi² = 20.36, df = 3 (P = 0.0001); I² = 85%). Excluding the one study population categorised as having a more severe CHD (and the only study population consisting of only men) from the analysis (Picard 1989), reduced the observed heterogeneity (Chi² = 2.48, df = 2 (P = 0.29); I² = 19%) and the observed effect estimate (MD −2.02 days, 95% CI −8.53 to 4.49). We considered none of the four work‐directed counselling studies to have a high overall risk of bias, and we found no visual indication of any time‐dependency (Analysis 2.2).

2.4 Work‐directed counselling versus usual care: secondary outcomes
Adverse effects

Two studies reported the rates of cardiac deaths (i.e. sudden death, death following MI) and reinfarctions up to six months after work‐directed counselling (Picard 1989; Pilote 1992). Work‐directed counselling probably makes little or no difference to cardiac death rate (RR 1.00, 95% CI 0.19 to 5.39, I² = 0%; moderate‐certainty evidence; Analysis 2.3). Work‐directed counselling may make little or no difference to reinfarction rate (RR 0.67, 95% CI 0.21 to 2.11; I² = 3%; Analysis 2.4).

3. Person‐directed physical conditioning interventions versus usual care

3.1 Person‐directed physical conditioning interventions versus usual care: primary outcomes

We included nine studies comparing the impact of some form of physical training or exercises versus usual care on return to work in the meta‐analysis shown in Analysis 3.1 (Andersen 1981; Bethell 1990; Dugmore 1999; Froelicher 1994; Holmbäck 1994; Maeder 1977; Marra 1985; Stern 1983; Worcester 1993).

3.1.1 Short term (less than six months)

Physical conditioning interventions resulted in a pooled RR estimate for short‐term return to work of 1.17 (95% CI 0.97 to 1.41; very low‐certainty evidence), with indications of substantial statistical heterogeneity (Chi² = 11.54, df = 3 (P = 0.009); I² = 74%). Excluding the Dugmore 1999 results, which we extracted from a graph, eliminated much of the observed heterogeneity (Chi² = 0.73, df = 2 (P = 0.69); I² = 0%) and reduced the pooled effect (RR 1.06, 95% CI 0.98 to 1.14). A sensitivity analysis excluding studies with overall high risk of bias (Andersen 1981; Worcester 1993), resulted in a pooled RR of 1.62 (95% CI 0.65 to 4.06; I² = 91%), due to the increased influence of the Dugmore 1999 study results.

3.1.2 Medium term (six months to one year)

Physical conditioning interventions made little or no difference in the medium‐term return‐to‐work rates (RR 1.09, 95% CI 0.99 to 1.20; I² = 5%; low‐certainty evidence). When we excluded studies with an overall high risk of bias (Holmbäck 1994; Worcester 1993) in a sensitivity analysis, physical conditioning interventions increased return‐to‐work rates at six to 12 months (RR 1.16, 95% CI 1.02 to 1.31; I² = 0%). The funnel plot of the five study results suggested some potential publication bias (Figure 4). The results of the Egger's test (P = 0.60), gave no indication of potential publication bias for this outcome, so we did not apply the 'trim and fill' method. However tests of publication bias may be underpowered when few studies are available. The meta‐regression considering changes of log RR over study time gave no indication of a time‐dependency (slope β = 0.00, P = 0.951).


Funnel plot of comparison 4. Physical conditioning interventions vs usual care, outcome: 4.3 proportion returning to work (medium term)

Funnel plot of comparison 4. Physical conditioning interventions vs usual care, outcome: 4.3 proportion returning to work (medium term)

Subgroup analysis

Pooling only study populations with similar CHD severity (Analysis 3.2) resulted in physical interventions having a bigger effect on the return‐to‐work rate in the two studies where the CHD was generally more severe (RR 1.12, 95% CI 1.00 to 1.25; I² = 0%; Dugmore 1999; Worcester 1993), and made little to no difference to return to work among the three study populations where the CHD was less severe (RR 1.04, 95% CI 0.84 to 1.29, I² = 36%; Holmbäck 1994; Marra 1985; Stern 1983).

3.1.5 Mean days until return to work

Four studies reported the mean time until return to work after MI either in weeks (Bethell 1990; Holmbäck 1994), or months (Maeder 1977; Marra 1985). We converted the reported results into mean days until return to work and pooled the mean differences. Using the SDs reported by Bethell 1990 and the interquartile ranges reported by Holmbäck 1994 we imputed the SD for the remaining two studies. Marra 1985 reported the results separately for study participants previously working in blue‐ or white‐collar professions, and we combined these results for Analysis 3.3. This analysis found that physical interventions made little or no difference in the time needed until return to work compared to usual care (MD −7.86 days, 95% CI −29.46 to 13.74; low‐certainty evidence). Due to the considerable statistical heterogeneity observed for this analysis (Chi² = 12.38, df = 3 (P = 0.006); I² = 76%), we also conducted a sensitivity analysis with the fixed‐effect model. The fixed‐effect model resulted in a smaller MD of −2.84 days (95% CI −10.43 to 4.75), giving some indication of the presence of a small‐study effect.

Subgroup analysis

We conducted subgroup analyses of study populations with more physically demanding occupations (blue‐collar workers) or less physically demanding occupations (white‐collar workers), using the stratified results reported by Marra 1985 (Analysis 3.4). We observed no mean difference in return‐to‐work times and considerably heterogeneity (Chi² = 4.50, df = 1; P = 0.03; I² = 78%) for white‐collar workers, while physical conditioning interventions reduced the mean time until return to work for blue‐collar workers (MD 28.29 days, 95% CI −48.68 to −7.91; I² = 0%).

3.2 Person‐directed physical conditioning interventions versus usual care: secondary outcomes
Working after an extended period of at least one year

Two studies reported the number or proportions of participants working between one and five years (Andersen 1981; Maeder 1977). Physical conditioning interventions had little to no effect on the proportion of participants at work in the long term (RR 1.04, 95% CI 0.82 to 2.66; I² = 0%; low‐certainty evidence). Excluding Andersen 1981, which we judged to have an overall high risk of bias, left only Maeder 1977, where the authors detected no effect of the intervention. However, Maeder 1977 applied an early, in‐hospital mobilisation intervention, and it is reasonable to expect more moderate effects on return to work by such a mild intervention. Only Dugmore 1999 reported the proportions of participants working five years after study completion. Dugmore 1999 alone reported the effects of a physical conditioning intervention on return to work compared to usual care at five years' follow‐up (RR 1.83, 95% 1.26 to 2.66; low‐certainty evidence).

Adverse effects

One study reported the rates of cardiac deaths (Marra 1985), and a second reported the rate of fatal MI (Dugmore 1999). We found that physical conditioning interventions may make little or no difference to the rate of cardiac deaths (RR 1.00, 95% CI 0.35 to 2.80; I² = 0%; moderate‐certainty evidence; Analysis 3.5). Two studies reported reinfarction rate (RR 0.70, 95% CI 0.26 to 1.88; I² = 47%; Analysis 3.6).

4. Combined interventions versus usual care

4.1 Combined interventions versus usual care: primary outcomes

We included 13 studies evaluating combined (comprehensive) cardiac rehabilitation programmes combining both counselling and physical exercise components in the meta‐analysis (Andersson 2010; Bengtsson 1983; Bertie 1992; Engblom 1997; Erdman 1986; Froelicher 1994; Higgins 2001; Hofman‐Bang 1999; Lidell 1996; Oldridge 1991; PRECOR 1991; Rivas 1988; Vermeulen 1988).

4.1.1 Short term (less than six months)

Four studies reported rate of return to work up to six months following a combined cardiac rehabilitation programme (Bertie 1992; Higgins 2001; PRECOR 1991; Rivas 1988). Combined cardiac rehabilitation programmes may increase the short‐term return‐to‐work rate (RR 1.56, 95% CI 1.23 to 1.98; I² = 20%; low‐certainty evidence; Analysis 4.1). This corresponds with a NNTB of 5, meaning one additional person will return to work up to six months after CHD hospitalisation for every five people receiving combined cardiac rehabilitation. Rivas 1988 considered two combined cardiac rehabilitation arms with varying intensities of the exercise component versus a single control group receiving usual care, and we combined the results of both intervention arms for the data synthesis. A sensitivity analysis excluding the one study with an overall high risk of bias (Bertie 1992) did not substantially alter the pooled estimate (RR 1.51, 95% CI 1.09 to 2.09, I² = 42%). The forest plot gave no visual indications of any time‐dependency for short‐term effects of combined interventions, and there were not enough studies considering short‐term return to work to conduct a meta‐regression.

4.1.2 Medium term (six months to one year)

Ten studies reported medium‐term return to work following combined interventions (Andersson 2010; Engblom 1997; Erdman 1986; Froelicher 1994; Higgins 2001; Hofman‐Bang 1999; Lidell 1996; Oldridge 1991; Rivas 1988; Vermeulen 1988). Combined interventions may make little to no difference in the medium‐term return‐to‐work rate (RR 1.06, 95% CI 1.00 to 1.13; I² = 0%; low‐certainty evidence; Analysis 4.1). As a sensitivity analysis, we omitted the four studies with an overall high risk of bias (Andersson 2010; Erdman 1986; Hofman‐Bang 1999; Lidell 1996) from the analysis, and this caused little change to the pooled effect estimate (RR 1.05, 95% CI 0.97 to 1.14; I² = 23%). Both a funnel plot (Figure 5), and the results of the Egger's Test (P = 0.843), showed no indications of publication bias for this outcome. We discerned no clear pattern of changing effect over time from the forest plot, and the meta‐regression of the log RR and study year also did not indicate any time‐dependency (slope β = 0.005, P = 0.409).


Funnel plot of comparison 5. Combined conditioning interventions vs usual care, outcome: 5.3 proportion returning to work (medium term)

Funnel plot of comparison 5. Combined conditioning interventions vs usual care, outcome: 5.3 proportion returning to work (medium term)

Subgroup analysis

The subgroup analysis of participant populations with similar severities of CHD (Analysis 4.2), resulted in a larger pooled effect estimate for participant populations with more severe CHD (RR 1.12, 95% CI 0.99 to 1.25; I² = 11%). We also considered study populations with similar physical work demands in a subgroup analysis (Analysis 4.3). The three study populations with more sedentary (white‐collar) workers (Engblom 1997; Higgins 2001; Rivas 1988), resulted in a pooled RR of 1.11 (95% CI 0.97 to 1.28; I² = 20%), while the two studies with study populations predominantly comprised of physical labourers (Lidell 1996; Vermeulen 1988), resulted in a pooled RR of 1.06 (95% CI 0.76 to 1.48; I² = 66%). Results did not differ notably according to the sex of the participants included in the studies (Analysis 4.4).

4.1.3 Mean days until return to work

Two studies reported the time until return to work following a combined intervention (Bengtsson 1983; Higgins 2001). We obtained SDs from the t‐test results of Higgins 2001 and applied this to both studies. Combined interventions shortened the mean length of time until return to work to a MD of −40.77 days (95% CI −67.19 to −14.35; I² = 66%; moderate‐certainty evidence; Analysis 4.5). We considered neither of these studies to have an overall high risk of bias. A sensitivity analysis with the fixed‐effect model resulted in a pooled MD of −39.32 days (95% CI −54.49 to −24.16).

4.2 Combined interventions versus usual care: secondary outcomes
Working after an extended period of at least one year

Aggregation of the six studies reporting results from long‐term follow‐ups of one to five years (Andersson 2010; Bengtsson 1983; Bertie 1992; Engblom 1997; Hofman‐Bang 1999; PRECOR 1991), resulted in a RR of 1.14 (95% CI 0.96 to 1.37; I² = 37%; very low‐certainty evidence; Analysis 4.1). Excluding the three studies with an overall high risk of bias (Andersson 2010; Bertie 1992; Hofman‐Bang 1999) increased the RR and the heterogeneity (RR 1.23, 95% CI 0.88 to 1.70; I² = 69%). Both a funnel plot (Figure 6), and the Egger's test (P = 0.406), showed no indications of publication bias or small‐study effects for long‐term return‐to‐work rates following combined interventions. We observed low heterogeneity for return‐to‐work rates following combined interventions, and sensitivity analyses with fixed‐effect models resulted in similar estimates. Regarding changes of effect over time, the meta‐regression of the log RRs and study year indicated no time‐dependency (slope β = 0.006, P = 0.614), and we discerned no clear pattern of changes over time from the forest plot.


Funnel plot of comparison: 5 combined conditioning interventions vs usual care, outcome: 5.6 proportion returning to work (long term)

Funnel plot of comparison: 5 combined conditioning interventions vs usual care, outcome: 5.6 proportion returning to work (long term)

Four studies reported the effects of combined interventions on the working status at five‐year follow‐up (Andersson 2010; Engblom 1997; Erdman 1986; Lidell 1996). Combined interventions resulted in a summarised RR for working after five years of 1.09 (95% CI 0.86 to 1.38, I² = 0%, very low‐certainty evidence; Analysis 4.1). When we excluded the three studies with an overall high risk of bias from the meta‐analysis, only Engblom 1997 remained. This study found the highest beneficial effect of the interventions on working status at five years (RR 1.66, 95% CI 0.76 to 3.61). There were also too few studies to be considered in a meta‐regression, and we discerned no pattern of changes over time from the forest plot.

Health‐related quality of life

One study reported a total score for health‐related quality of life using the Angina Pectoris Quality of Life Questionnaire among study participants primarily eligible to return to work at a follow‐up time of two years (Hofman‐Bang 1999). The studied combined intervention appeared to have little to no effect on health‐related quality of life score when compared to usual care group at two years (MD 0.40, 95% CI −0.03 to 0.83; low‐certainty evidence; Analysis 4.6).

Adverse effects

Four studies reported total mortality rates after combined interventions for follow‐up times of one to five years (Bengtsson 1983; Erdman 1986; Hofman‐Bang 1999; Rivas 1988). Combined interventions resulted in a summarised RR for total mortality of 1.43 (95% CI 0.59 to 3.51; I² = 5%; Analysis 4.7). Three studies also reported reinfarction rates after combined interventions for follow‐up times of one to five years (Bengtsson 1983; Erdman 1986; Hofman‐Bang 1999; Vermeulen 1988). Combined interventions resulted in a summarised RR for reinfarctions of 0.56 (95% CI 0.23 to 1.40; I² = 0%; moderate‐certainty evidence; Analysis 4.8).

Discussion

disponible en

Summary of main results

This review included 39 studies applying a randomised controlled study design to investigate the impact of various interventions on the rate and timing of return to work following a MI, CABG or PCI. We do not know if counselling interventions (including health education) that addressed fears or concerns related to CHD increase the proportion of people with CHD returning to work at follow‐ups of up to one year (very low‐certainty evidence), and these interventions may make little or no difference in the proportion working in the long term (low‐certainty evidence). We also do not know if psychological counselling interventions reduce the time needed to return to work due to the very low certainty of the evidence. We did not find studies reporting total health‐related quality of life during the return‐to‐work process or adverse effects following person‐directed psychological interventions, so we could not assess these secondary outcomes.

Counselling directed specifically at encouraging returning to work, for example by providing a physician‐sanctioned goal for returning to work based on the results of symptom‐limited treadmill testing or by attempting to assuage concerns of co‐workers regarding their colleague’s ability to return to work, may result in little to no difference in days until return to work. Work‐directed counselling probably results in little or no difference in cardiac death rates. Why counselling to encourage return to work did not have more of an impact on return to work is unclear. Perhaps interventions providing a concrete time frame for when it should be physically safe to return to work do not adequately address other personal and work‐related obstacles that may affect an individual's decision to return to work.

Cardiac rehabilitation comprising only some form of physical conditioning following CHD may result in little to no difference in the medium‐term (six to 12 months) return‐to‐work rate (low‐certainty evidence) and may result in little to no difference in the proportion at work after one year and up to five years (low‐certainty evidence). We do not know if physical conditioning interventions alone increase short‐term (up to six months) return to work (very low‐certainty evidence). Considering the timing of return to work among participants returning to work, physical conditioning interventions may result in little to no difference in mean time needed to return to work (low‐certainty evidence). Physical conditioning interventions appeared to reduce time away from work particularly among blue‐collar workers. Physical conditioning interventions probably do not increase adverse effects (cardiac deaths).

Combined (comprehensive) cardiac rehabilitation programmes combining physical conditioning with counselling and risk factor education appeared to have some effect on return to work. Combined cardiac rehabilitation programmes may increase the proportion of participants resuming work up to six months following hospitalisation for CHD (low‐certainty evidence). For about every five participants receiving combined rehabilitation, one additional participant returned to work within six months of hospitalisation (NNTB 5). Combined rehabilitation programmes may result in little to no difference in the proportion returning to work after six months and up to one year (low‐certainty evidence). We do not know if combined cardiac rehabilitation programmes increase the long‐term (one to five years) or the extended long‐term (five or more years) proportion of participants working after hospitalisation (very low‐certainty evidence). Combined interventions probably reduce the average time needed to return to work (moderate‐certainty evidence) by about 40 days when compared to receiving usual care. Combined interventions may result in little to no difference in health‐related quality of life and probably result in little to no difference in adverse effects (assessed as reinfarctions).

Overall completeness and applicability of evidence

We found 39 RCTs examining the effect of person‐directed interventions on return to work among people with CHD conducted since the 1970s. However, none of the studies focused on the evaluation of work‐directed programmes. Overall the evidence of the studies was directly applicable to the aim of this review and its study questions. Studies were conducted for the most part in North America, Western Europe, and Australia. Although women were more often included in the study populations (25 studies) than not, women generally comprised a very small portion of the included participants. Only one study explicitly examined the effect of an intervention on return to work among women (Andersson 2010). Studies predominantly considered people who had been hospitalised for a MI, and less frequently considered people undergoing CABG or PCI. This may mean that the results are less applicable to people undergoing revascularisation procedures.

The studies considering health‐related quality of life reported results for the entire study populations and not just those eligible to return to work. Therefore, to assess our secondary outcome of health‐related quality of life within the return‐to‐work process, we considered health‐related quality of life results of studies where at least 80% of study participants were eligible to return to work. We did not find enough studies conducted predominantly among participants eligible to return to work that had assessed health‐related quality of life to conduct a meta‐analysis of this secondary outcome. Likewise, we also considered cardiac death rates, reinfarction rates and total mortality as severe adverse effects of interventions among studies where at least 80% of study participants were eligible to return to work in order to increase the applicability of the results to the return‐to‐work process.

Quality of the evidence

We included 39 studies, but once we aggregated studies according to intervention form and follow‐up times, the highest number of studies that we could aggregate was 10. We judged the overall risk of bias to be high for 16 of the studies. However, excluding studies judged to have an overall high risk of bias did not seem to systematically alter the results. The way that studies described many aspects of the study characteristics of interest, including return‐to‐work results and details regarding the severity of CHD at baseline were heterogeneous and often lacking important details, such as the actual number of participants considered in the return‐to‐work analysis. Also studies rarely provided information regarding study participants' employment characteristics prior to CHD (i.e. how many participants worked in physically strenuous occupations). Studies considering return to work among a subgroup of previously employed participants for the return‐to‐work analyses, reported loss to follow‐up for the entire study populations, which made it difficult for us to determine the outcome‐specific losses to follow‐up in these studies. Some studies also reported that the desire to return to work reduced compliance with the rehabilitation programmes, however withdrawals during the interventions should not have affected the results among studies conducting intention‐to‐treat analyses, and we considered intention‐to‐treat in our 'Risk of bias' assessments.

Our assessment of the quality of evidence was also hindered by poorly reported methods. Even studies published after the publication of the first proposal of standards for reporting of RCTs (Standards of Reporting Trials Group 1994) lacked adequate descriptions of allocation methods to permit clear 'Risk of bias' judgements, and none of the included studies cited a study protocol that would have permitted more objective assessments of selective reporting bias. Although we initially intended to contact all study authors to obtain additional information to aid in the 'Risk of bias' assessments, this often proved to be impossible, as many studies were conducted more than 20 years ago.

We downgraded nearly all outcomes by at least one level due to risk of bias. We also downgraded many outcomes because of imprecision due to wide confidence intervals that could include possible appreciable harm or benefit, and because of inconsistency due to substantial heterogeneity that could not completely be explained. We did not downgrade any outcomes for indirectness.

Return to work was often a secondary outcome of the studies, and as such, the results pertaining to return to work were not always clearly reported. There may have been additional cardiac rehabilitation studies that considered return to work, but did not report these results in any published document. It is possible that such omissions may be more likely to involve results for secondary outcomes when these were not statistically significant, and this selective reporting could result in a form of publication bias. We found some indications of publication bias among the studies of psychological interventions considering short‐term (Figure 3), and medium‐term return to work. We also observed visual indications of publication bias among physical conditioning intervention studies reporting medium‐term return to work (Figure 4). We did not detect publication bias for the pooled analyses of combined interventions reporting medium‐term (Figure 5) and long‐term (Figure 6) return to work.

Potential biases in the review process

Although we conducted an extensive search, our review process may have some limitations. We excluded studies mentioning return to work somewhere in the abstract or introduction without reporting any return‐to‐work results. We included studies reporting return‐to‐work results as percentages without providing the absolute number of study participants working prior to their CHD. We tried to obtain unpublished data from study authors regarding numbers of people working prior to CHD, but we were not always able to contact them. While we still included these studies in the review, we could not include the results of these studies in the meta‐analysis.

We also found registered clinical trials mentioning return to work as an outcome, that according to their registered start dates, should have produced results by now. However, we did not find any results that we could link to these studies. Publication bias may be leading to an underreporting of return‐to‐work results.

Agreements and disagreements with other studies or reviews

Like the Anderson 2017a review, we found no evidence that psychological interventions for CHD had any impact, positive or otherwise, on adverse reactions such as mortality or non‐fatal MI in the studies examining return to work. The Anderson 2017a review also reported that four of 10 studies examining health‐related quality of life observed improvements in at least one dimension of health‐related quality of life in the intervention group receiving a psychological intervention that differed significantly from that observed in the comparison groups. We were unable to examine health‐related quality of life following psychological interventions among studies examining return to work.

In contrast to the Anderson 2016 review, which found evidence that exercise‐based rehabilitations reduced cardiac mortality, we did not observe any meaningful differences across study groups with regard to cardiac mortality in the studies examining return to work in populations predominantly eligible to return to work.

PRISMA flow diagram of study selection process
Figuras y tablas -
Figure 1

PRISMA flow diagram of study selection process

'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study
Figuras y tablas -
Figure 2

'Risk of bias' summary: review authors' judgements about each risk of bias item for each included study

Funnel plot of comparison 2. Psychological interventions (including health education) vs usual care, outcome: 2.3 proportion returning to work (short term)
Figuras y tablas -
Figure 3

Funnel plot of comparison 2. Psychological interventions (including health education) vs usual care, outcome: 2.3 proportion returning to work (short term)

Funnel plot of comparison 4. Physical conditioning interventions vs usual care, outcome: 4.3 proportion returning to work (medium term)
Figuras y tablas -
Figure 4

Funnel plot of comparison 4. Physical conditioning interventions vs usual care, outcome: 4.3 proportion returning to work (medium term)

Funnel plot of comparison 5. Combined conditioning interventions vs usual care, outcome: 5.3 proportion returning to work (medium term)
Figuras y tablas -
Figure 5

Funnel plot of comparison 5. Combined conditioning interventions vs usual care, outcome: 5.3 proportion returning to work (medium term)

Funnel plot of comparison: 5 combined conditioning interventions vs usual care, outcome: 5.6 proportion returning to work (long term)
Figuras y tablas -
Figure 6

Funnel plot of comparison: 5 combined conditioning interventions vs usual care, outcome: 5.6 proportion returning to work (long term)

Comparison 1 Psychological interventions (including health education) vs usual care, Outcome 1 Proportion returning to work (all studies).
Figuras y tablas -
Analysis 1.1

Comparison 1 Psychological interventions (including health education) vs usual care, Outcome 1 Proportion returning to work (all studies).

Comparison 1 Psychological interventions (including health education) vs usual care, Outcome 2 Proportion returning to work short term (< 6 months) by CHD severity.
Figuras y tablas -
Analysis 1.2

Comparison 1 Psychological interventions (including health education) vs usual care, Outcome 2 Proportion returning to work short term (< 6 months) by CHD severity.

Comparison 1 Psychological interventions (including health education) vs usual care, Outcome 3 Proportion returning to work medium term (6 months‐1 year) by CHD severity.
Figuras y tablas -
Analysis 1.3

Comparison 1 Psychological interventions (including health education) vs usual care, Outcome 3 Proportion returning to work medium term (6 months‐1 year) by CHD severity.

Comparison 1 Psychological interventions (including health education) vs usual care, Outcome 4 Mean time until return to work (days).
Figuras y tablas -
Analysis 1.4

Comparison 1 Psychological interventions (including health education) vs usual care, Outcome 4 Mean time until return to work (days).

Comparison 2 Work‐directed counselling vs usual care, Outcome 1 Proportion returning to work (all studies).
Figuras y tablas -
Analysis 2.1

Comparison 2 Work‐directed counselling vs usual care, Outcome 1 Proportion returning to work (all studies).

Comparison 2 Work‐directed counselling vs usual care, Outcome 2 Mean time until return to work (days).
Figuras y tablas -
Analysis 2.2

Comparison 2 Work‐directed counselling vs usual care, Outcome 2 Mean time until return to work (days).

Comparison 2 Work‐directed counselling vs usual care, Outcome 3 Adverse effects: cardiac deaths.
Figuras y tablas -
Analysis 2.3

Comparison 2 Work‐directed counselling vs usual care, Outcome 3 Adverse effects: cardiac deaths.

Comparison 2 Work‐directed counselling vs usual care, Outcome 4 Adverse effects: reinfarctions.
Figuras y tablas -
Analysis 2.4

Comparison 2 Work‐directed counselling vs usual care, Outcome 4 Adverse effects: reinfarctions.

Comparison 3 Physical conditioning interventions vs usual care, Outcome 1 Proportion returning to work (all studies).
Figuras y tablas -
Analysis 3.1

Comparison 3 Physical conditioning interventions vs usual care, Outcome 1 Proportion returning to work (all studies).

Comparison 3 Physical conditioning interventions vs usual care, Outcome 2 Proportion returning to work medium term (0.5‐1 year) by CHD severity.
Figuras y tablas -
Analysis 3.2

Comparison 3 Physical conditioning interventions vs usual care, Outcome 2 Proportion returning to work medium term (0.5‐1 year) by CHD severity.

Comparison 3 Physical conditioning interventions vs usual care, Outcome 3 Mean time until return to work (days).
Figuras y tablas -
Analysis 3.3

Comparison 3 Physical conditioning interventions vs usual care, Outcome 3 Mean time until return to work (days).

Comparison 3 Physical conditioning interventions vs usual care, Outcome 4 Mean time until return to work (days) by physically strenuous workgroup.
Figuras y tablas -
Analysis 3.4

Comparison 3 Physical conditioning interventions vs usual care, Outcome 4 Mean time until return to work (days) by physically strenuous workgroup.

Comparison 3 Physical conditioning interventions vs usual care, Outcome 5 Adverse effects: cardiac deaths.
Figuras y tablas -
Analysis 3.5

Comparison 3 Physical conditioning interventions vs usual care, Outcome 5 Adverse effects: cardiac deaths.

Comparison 3 Physical conditioning interventions vs usual care, Outcome 6 Adverse effects: reinfarctions.
Figuras y tablas -
Analysis 3.6

Comparison 3 Physical conditioning interventions vs usual care, Outcome 6 Adverse effects: reinfarctions.

Comparison 4 Combined interventions vs usual care, Outcome 1 Proportion returning to work (all studies).
Figuras y tablas -
Analysis 4.1

Comparison 4 Combined interventions vs usual care, Outcome 1 Proportion returning to work (all studies).

Comparison 4 Combined interventions vs usual care, Outcome 2 Proportion returning to work medium term (6 months‐1 year) by CHD severity.
Figuras y tablas -
Analysis 4.2

Comparison 4 Combined interventions vs usual care, Outcome 2 Proportion returning to work medium term (6 months‐1 year) by CHD severity.

Comparison 4 Combined interventions vs usual care, Outcome 3 Proportion returning to work medium term (6 months‐1 year) by physically strenuous work.
Figuras y tablas -
Analysis 4.3

Comparison 4 Combined interventions vs usual care, Outcome 3 Proportion returning to work medium term (6 months‐1 year) by physically strenuous work.

Comparison 4 Combined interventions vs usual care, Outcome 4 Proportion returning to work medium term (6 months‐1 year) by sex.
Figuras y tablas -
Analysis 4.4

Comparison 4 Combined interventions vs usual care, Outcome 4 Proportion returning to work medium term (6 months‐1 year) by sex.

Comparison 4 Combined interventions vs usual care, Outcome 5 Mean time until return to work (days).
Figuras y tablas -
Analysis 4.5

Comparison 4 Combined interventions vs usual care, Outcome 5 Mean time until return to work (days).

Comparison 4 Combined interventions vs usual care, Outcome 6 Health‐related quality of life.
Figuras y tablas -
Analysis 4.6

Comparison 4 Combined interventions vs usual care, Outcome 6 Health‐related quality of life.

Comparison 4 Combined interventions vs usual care, Outcome 7 Adverse effects: total mortality.
Figuras y tablas -
Analysis 4.7

Comparison 4 Combined interventions vs usual care, Outcome 7 Adverse effects: total mortality.

Comparison 4 Combined interventions vs usual care, Outcome 8 Adverse effects: reinfarctions.
Figuras y tablas -
Analysis 4.8

Comparison 4 Combined interventions vs usual care, Outcome 8 Adverse effects: reinfarctions.

Summary of findings for the main comparison. Psychological interventions (including health education) compared to usual care for people with coronary heart disease

Psychological interventions (including health education) compared to usual care for people with coronary heart disease

Patient or population: people with coronary heart disease
Setting: hospital/home
Intervention: psychological interventions (including health education)
Comparison: usual care

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with psychological interventions (including health education)

Proportion of participants returning to work in the short term (up to 6 months)
Follow‐up: range 3 months to 4 months

Study population

RR 1.08
(0.84 to 1.40)

375
(6 RCTs)

⊕⊝⊝⊝
Very low1,2,3,4

We do knot know if psychological interventions (including health education) increase the proportion returning to work in the short term (up to 6 months)

63 per 100

68 per 100
(53 to 88)

Proportion of participants returning to work in the medium term (6 months ‐ 1 year)
Follow‐up: range 6 months to 1 year

Study population

RR 1.24
(0.95 to 1.63)

316
(7 RCTs)

⊕⊝⊝⊝
Very low1,2,3,4

We do not know if psychological interventions (including health education) increase the proportion returning to work in the medium term (6 months ‐ 1 year).

63 per 100

78 per 100
(59 to 100)

Proportion of participants at work in the long term (> 1 to < 5 years)
Follow‐up: range 1.5 years to 4 years

Study population

RR 1.09
(0.88 to 1.34)

239
(3 RCTs)

⊕⊕⊝⊝
Low2,3

Psychological interventions (including health education) may make little or no difference in the proportion working in the long term (> 1 to < 5 years)

74 per 100

81 per 100
(65 to 99)

Days until return to work
Follow‐up: range 6 months to 1.5 years

The mean time to return to work was 9.7 days lower
(35.09 lower to 15.69 higher)

125
(2 RCTs)

⊕⊝⊝⊝
Very low1,2,3

We do not know if psychological interventions (including health education) lower the days needed until returning to work

*The risk in the Intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

1Downgraded one level due to substantial heterogeneity that we could not completely explain.
2Downgraded one level due to risk of bias.
3Downgraded one level due to imprecision (pooled confidence interval is wide and includes either a possible appreciable harm or benefit).
4Downgraded one level, because results of a funnel plot indicated possible publication bias.

Figuras y tablas -
Summary of findings for the main comparison. Psychological interventions (including health education) compared to usual care for people with coronary heart disease
Summary of findings 2. Work‐directed counselling compared to usual care for people with coronary heart disease

Work‐directed counselling compared to usual care for people with coronary heart disease

Patient or population: people with coronary heart disease
Setting: hospital/home
Intervention: work‐directed counselling
Comparison: usual care

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with work‐directed counselling

Days until return to work

The mean time to return to work was 7.52 days lower
(20.07 lower to 5.03 higher)

618
(4 RCTs)

⊕⊕⊝⊝
Low1,2

Work‐directed counselling may result in little to no difference in days until return to work

Adverse effects: cardiac deaths
Follow‐up mean: 6 months

2 per 100

2 per 100
(0 to 8)

RR 1.00
(0.19 to 5.39)

388
(2 RCTs)

⊕⊕⊕⊝
Moderate3

Work‐directed counselling probably results little or no difference in cardiac death rates

*The risk in the Intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; RCT: randomised controlled trial; RR: risk ratio

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

1Downgraded one level due to substantial heterogeneity that we could not completely explain.
2Downgraded one level due to imprecision (two of the four studies did not report the standard deviation).
3Downgraded one level due to imprecision (pooled confidence interval is wide and includes either a possible harm or benefit).

Figuras y tablas -
Summary of findings 2. Work‐directed counselling compared to usual care for people with coronary heart disease
Summary of findings 3. Physical conditioning interventions compared to usual care for people with coronary heart disease

Physical conditioning interventions compared to usual care for people with coronary heart disease

Patient or population: people with coronary heart disease
Setting: hospital/home
Intervention: physical conditioning interventions
Comparison: usual care

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with physical conditioning interventions

Proportion of participants returning to work in the short term (up to 6 months)
Follow‐up: range 3 months to 5.5 months

Study population

RR 1.17
(0.97 to 1.41)

460
(4 RCTs)

⊕⊝⊝⊝
Very low1,2,3

We do not know if physical conditioning interventions increase the proportion returning to work in the short term (up to 6 months)

68 per 100

80 per 100
(66 to 96)

Proportion of participants returning to work in the medium term (6 months‐1 year)
Follow‐up: range 0.5 years to 1 years

Study population

RR 1.09
(0.99 to 1.20)

510
(5 RCTs)

⊕⊕⊝⊝
Low1 4

Physical conditioning interventions may result in little to no difference in proportion returning to work in the medium term (6 months‐1 year)

75 per 100

82 per 100
(74 to 90)

Proportion of participants at work in the long term (> 1 to < 5 years)
Follow‐up: range 3 years to 4 years

Study population

RR 1.04
(0.82 to 1.30)

156
(2 RCTs)

⊕⊕⊝⊝
Low1

Physical conditioning interventions may result in little to no difference in proportion at work in the long term (> 1 to < 5 years)

64 per 100

67 per 100
(53 to 84)

Proportion of participants at work in the extended long term (≥ 5 years)
Follow‐up: mean 5 years

Study population

RR 1.83
(1.26 to 2.66)

119
(1 RCT)

⊕⊕⊝⊝
Low5

Physical conditioning interventions may increase the proportion at work in the extended long term (≥ 5 years)

37 per 100

68 per 100
(47 to 99)

Days until return to work

The mean time to return to work was 7.86 days lower
(29.46 lower to 13.74 higher)

430
(4 RCTs)

⊕⊕⊝⊝
Low1 2

Physical conditioning interventions appear to result in little to no difference in mean time to return to work (days)

Adverse effects: cardiac deaths

Follow‐up: mean 4.8 years

8 per 100

8 per 100
(3 to 24)

RR 1.00
(0.35 to 2.80)

285
(2 RCTs)

⊕⊕⊕⊝
Moderate3

Physical conditioning interventions probably do not increase adverse effects (cardiac deaths)

*The risk in the Intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; RCT: randomised controlled trials; RR: risk ratio

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

1Downgraded one level due to risk of bias.
2Downgraded one level due to substantial heterogeneity that we could not completely explain.
3Downgraded one level due to imprecision (pooled confidence interval is wide and includes either a possible appreciable harm or benefit).
4Downgraded one level, because results of funnel plot indicated possible publication bias.
5Downgraded one level because only one study reported the proportion of study participants working five years after the intervention.

Figuras y tablas -
Summary of findings 3. Physical conditioning interventions compared to usual care for people with coronary heart disease
Summary of findings 4. Combined interventions compared to usual care for people with coronary heart disease

Combined interventions compared to usual care for people with coronary heart disease

Patient or population: people with coronary heart disease
Setting: hospital/home
Intervention: combined interventions
Comparison: usual care

Outcomes

Anticipated absolute effects* (95% CI)

Relative effect
(95% CI)

№ of participants
(studies)

Certainty of the evidence
(GRADE)

Comments

Risk with usual care

Risk with combined interventions

Proportion of participants returning to work in the short term (up to 6 months)
Follow‐up: range 2.3 months to 4 months

Study population

RR 1.56
(1.23 to 1.98)

395
(4 RCTs)

⊕⊕⊝⊝
Low1,2

Combined rehabilitation interventions may increase the proportion returning to work in the short term (up to 6 months)

39 per 100

61 per 100
(48 to 78)

Proportion of participants returning to work in the medium term (6 months ‐ 1 year)
Follow‐up: range 6 months to 1 year

Study population

RR 1.06
(1.00 to 1.13)

992
(10 RCTs)

⊕⊕⊝⊝
Low3

Combined interventions may result in little to no difference in the proportion returning to work in the medium term (6 months ‐ 1 year)

72 per 100

76 per 100
(72 to 81)

Proportion of participants at work in the long term (> 1 to < 5 years)
Follow‐up: range 1.2 years to 3 years

Study population

RR 1.14
(0.96 to 1.37)

491
(6 RCTs)

⊕⊝⊝⊝
Very low1,3

We do not know if combined interventions increase the proportion working long term (> 1 to < 5 years)

53 per 100

60 per 100
(51 to 72)

Proportion of participants at work in the extended long term (≥ 5 years)
Follow‐up: 5 years

Study population

RR 1.09
(0.86 to 1.38)

350
(4 RCTs)

⊕⊝⊝⊝
Very low1,3

We do not know if combined interventions increase the proportion working after an extended term (≥ 5 years)

37 per 100

41 per 100
(32 to 51)

Days until return to work

The mean time to return to work in the intervention group was 40.77 days lower
(67.19 lower to 14.35 lower)

181
(2 RCTs)

⊕⊕⊕⊝
Moderate4

Combined rehabilitation interventions probably reduce mean time to return to work (days)

Health‐related quality of life assessed with: Angina Pectoris Quality of Life Questionnaire

The MD for HrQoL was 0.40 (‐0.03 lower to 0.83 higher)

87
(1 RCT)

⊕⊕⊝⊝
Low2,5

Combined interventions may result in little to no difference in HrQoL

Adverse effects: reinfarctions

Follow‐up: mean 3.8 years

10 per 100

6 per 100
(2 to 15)

RR 0.56
(0.23 to 1.43)

265
(3 RCTs)

⊕⊕⊕⊝
Moderate1

Combined interventions likely result in little to no difference in adverse effects

*The risk in the Intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

CI: confidence interval; HRQoL: health‐related quality of life; RCT: randomised controlled trial; RR: risk ratio; MD: mean difference

GRADE Working Group grades of evidence
High certainty: we are very confident that the true effect lies close to that of the estimate of the effect.
Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different.
Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect.
Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

1Downgraded one level due to imprecision (pooled confidence interval is wide and includes either a possible appreciable harm or benefit).
2Downgraded one level due to risk of bias.
3Downgraded two levels due to risk of bias.
4We detected substantial heterogeneity that we could not completely explain.
5Downgraded one level because only one study reported the effects of the intervention on health‐related quality of life.

Figuras y tablas -
Summary of findings 4. Combined interventions compared to usual care for people with coronary heart disease
Comparison 1. Psychological interventions (including health education) vs usual care

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Proportion returning to work (all studies) Show forest plot

11

Risk Ratio (IV, Random, 95% CI)

Subtotals only

1.1 Short term (< 6 months)

6

375

Risk Ratio (IV, Random, 95% CI)

1.08 [0.84, 1.40]

1.2 Medium term (6 months‐1 year)

7

316

Risk Ratio (IV, Random, 95% CI)

1.24 [0.95, 1.63]

1.3 Long term (> 1 to < 5 years)

3

239

Risk Ratio (IV, Random, 95% CI)

1.09 [0.88, 1.34]

2 Proportion returning to work short term (< 6 months) by CHD severity Show forest plot

6

397

Risk Ratio (IV, Random, 95% CI)

1.10 [0.85, 1.43]

2.1 CHD severity unknown

3

241

Risk Ratio (IV, Random, 95% CI)

0.98 [0.68, 1.40]

2.2 CHD more severe

2

67

Risk Ratio (IV, Random, 95% CI)

1.10 [0.83, 1.46]

2.3 CHD less severe

1

89

Risk Ratio (IV, Random, 95% CI)

1.87 [1.03, 3.38]

3 Proportion returning to work medium term (6 months‐1 year) by CHD severity Show forest plot

7

316

Risk Ratio (IV, Random, 95% CI)

1.23 [0.96, 1.59]

3.1 CHD severity unknown

4

208

Risk Ratio (IV, Random, 95% CI)

1.12 [0.82, 1.53]

3.2 CHD more severe

2

73

Risk Ratio (IV, Random, 95% CI)

1.61 [0.97, 2.67]

3.3 CHD less severe

2

35

Risk Ratio (IV, Random, 95% CI)

1.17 [0.67, 2.03]

4 Mean time until return to work (days) Show forest plot

2

125

Mean Difference (IV, Random, 95% CI)

‐9.70 [‐35.09, 15.69]

Figuras y tablas -
Comparison 1. Psychological interventions (including health education) vs usual care
Comparison 2. Work‐directed counselling vs usual care

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Proportion returning to work (all studies) Show forest plot

4

Risk Ratio (IV, Random, 95% CI)

Totals not selected

1.1 Short term (< 6 months)

1

Risk Ratio (IV, Random, 95% CI)

0.0 [0.0, 0.0]

1.2 Medium term (6 months‐1 year)

2

Risk Ratio (IV, Random, 95% CI)

0.0 [0.0, 0.0]

1.3 Long term (> 1 to < 5 years)

1

Risk Ratio (IV, Random, 95% CI)

0.0 [0.0, 0.0]

2 Mean time until return to work (days) Show forest plot

4

618

Mean Difference (IV, Random, 95% CI)

‐7.52 [‐20.07, 5.03]

3 Adverse effects: cardiac deaths Show forest plot

2

388

Risk Ratio (IV, Fixed, 95% CI)

1.00 [0.19, 5.39]

4 Adverse effects: reinfarctions Show forest plot

2

388

Risk Ratio (IV, Fixed, 95% CI)

0.67 [0.21, 2.11]

Figuras y tablas -
Comparison 2. Work‐directed counselling vs usual care
Comparison 3. Physical conditioning interventions vs usual care

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Proportion returning to work (all studies) Show forest plot

8

Risk Ratio (IV, Random, 95% CI)

Subtotals only

1.1 Short term (< 6 months)

4

460

Risk Ratio (IV, Random, 95% CI)

1.17 [0.97, 1.41]

1.2 Medium term (6 months‐1 year)

5

510

Risk Ratio (IV, Random, 95% CI)

1.09 [0.99, 1.20]

1.3 Long term (> 1 to < 5 years)

2

156

Risk Ratio (IV, Random, 95% CI)

1.04 [0.82, 1.30]

1.4 Extended long term (≥ 5 years)

1

119

Risk Ratio (IV, Random, 95% CI)

1.83 [1.26, 2.66]

2 Proportion returning to work medium term (0.5‐1 year) by CHD severity Show forest plot

5

510

Risk Ratio (IV, Fixed, 95% CI)

1.09 [1.00, 1.19]

2.1 CHD more severe

2

277

Risk Ratio (IV, Fixed, 95% CI)

1.12 [1.00, 1.25]

2.2 CHD less severe

3

233

Risk Ratio (IV, Fixed, 95% CI)

1.05 [0.92, 1.21]

3 Mean time until return to work (days) Show forest plot

4

430

Mean Difference (IV, Random, 95% CI)

‐7.86 [‐29.46, 13.74]

4 Mean time until return to work (days) by physically strenuous workgroup Show forest plot

4

Mean Difference (IV, Random, 95% CI)

Subtotals only

4.1 White‐collar/less strenuous

2

153

Mean Difference (IV, Random, 95% CI)

‐1.10 [‐52.79, 50.59]

4.2 Blue‐collar/more strenuous

2

148

Mean Difference (IV, Random, 95% CI)

‐28.29 [‐48.68, ‐7.91]

4.3 Type of work not reported

1

129

Mean Difference (IV, Random, 95% CI)

3.0 [‐5.81, 11.81]

5 Adverse effects: cardiac deaths Show forest plot

2

285

Risk Ratio (IV, Fixed, 95% CI)

1.00 [0.35, 2.80]

6 Adverse effects: reinfarctions Show forest plot

2

230

Risk Ratio (IV, Fixed, 95% CI)

0.70 [0.26, 1.88]

Figuras y tablas -
Comparison 3. Physical conditioning interventions vs usual care
Comparison 4. Combined interventions vs usual care

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Proportion returning to work (all studies) Show forest plot

13

Risk Ratio (IV, Random, 95% CI)

Subtotals only

1.1 Short term (< 6 months)

4

395

Risk Ratio (IV, Random, 95% CI)

1.56 [1.23, 1.98]

1.2 Medium term (6 months‐1 year)

10

992

Risk Ratio (IV, Random, 95% CI)

1.06 [1.00, 1.13]

1.3 Long term (> 1 to < 5 years)

6

491

Risk Ratio (IV, Random, 95% CI)

1.14 [0.96, 1.37]

1.4 Extended long term (≥ 5 years)

4

350

Risk Ratio (IV, Random, 95% CI)

1.09 [0.86, 1.38]

2 Proportion returning to work medium term (6 months‐1 year) by CHD severity Show forest plot

10

992

Risk Ratio (IV, Random, 95% CI)

1.06 [1.00, 1.13]

2.1 CHD more severe

3

293

Risk Ratio (IV, Random, 95% CI)

1.12 [0.99, 1.25]

2.2 CHD less severe

4

384

Risk Ratio (IV, Random, 95% CI)

0.96 [0.85, 1.08]

2.3 CHD severity unknown

3

315

Risk Ratio (IV, Random, 95% CI)

1.10 [0.99, 1.24]

3 Proportion returning to work medium term (6 months‐1 year) by physically strenuous work Show forest plot

10

Risk Ratio (IV, Random, 95% CI)

Subtotals only

3.1 White‐collar/less strenuous

3

357

Risk Ratio (IV, Random, 95% CI)

1.11 [0.97, 1.28]

3.2 Blue‐collar/more strenuous

2

167

Risk Ratio (IV, Random, 95% CI)

1.06 [0.76, 1.48]

3.3 Type of work not reported

5

468

Risk Ratio (IV, Random, 95% CI)

1.05 [0.98, 1.13]

4 Proportion returning to work medium term (6 months‐1 year) by sex Show forest plot

10

Risk Ratio (IV, Random, 95% CI)

Subtotals only

4.1 Men only

3

273

Risk Ratio (IV, Random, 95% CI)

1.09 [0.81, 1.45]

4.2 Women and men

6

623

Risk Ratio (IV, Random, 95% CI)

1.07 [1.00, 1.14]

4.3 Women only

1

96

Risk Ratio (IV, Random, 95% CI)

0.99 [0.76, 1.30]

5 Mean time until return to work (days) Show forest plot

2

181

Mean Difference (IV, Random, 95% CI)

‐40.77 [‐67.19, ‐14.35]

6 Health‐related quality of life Show forest plot

1

Mean Difference (IV, Fixed, 95% CI)

Totals not selected

7 Adverse effects: total mortality Show forest plot

4

438

Odds Ratio (M‐H, Fixed, 95% CI)

1.43 [0.59, 3.51]

8 Adverse effects: reinfarctions Show forest plot

3

265

Odds Ratio (M‐H, Fixed, 95% CI)

0.56 [0.23, 1.40]

Figuras y tablas -
Comparison 4. Combined interventions vs usual care