Results of the search

The Preferred Reporting Items for Systematic Reviews and Meta‐Analyses (PRISMA) study flow diagram of the search for the updated and extended review is shown in Figure 1. Electronic searches identified 966 records. On the basis of titles and abstracts, we excluded 550 records and we assessed 109 full‐text records for eligibility.

Figure 1

---

---

Study flow diagram.

Of 109 records examined in full, we considered 13 to be eligible for inclusion, and we found two additional studies by searching the reference sections of relevant reviews and by updating our search of databases of ongoing trials. These 15 records represented nine independent studies (De Villiers 2009; FLUCAD 2008; FLUVACS 2002; Govaert 1994; IVCAD 2009; Langley 2011a; NCT01945268; Phrommintikul 2011; Wu 2010) (some of the 15 records were conference abstracts and have not been listed in the reference section). One of the trials was available in abstract form only (IVCAD 2009), but we obtained additional information by contacting the study author. Six of these trials reported cardiovascular death (FLUCAD 2008; FLUVACS 2002; Govaert 1994; IVCAD 2009; Langley 2011a; Phrommintikul 2011) and could be included in a meta‐analysis of this outcome. Two had already been included in the previous version of the review (FLUCAD 2008; FLUVACS 2002), and the remainder were newly added.

Included studies

For details of the study characteristics of individual studies, see Characteristics of included studies.

Excluded studies

We excluded 95 records for the following reasons: The study was not an RCT (n = 59), eligible outcomes were not reported (n = 20) or the study did not have a placebo or 'no intervention' group (n = 14). One record could not be obtained (but was unlikely to be an RCT) and one ongoing trial was identified (MacIntyre 2007), but contact with study authors suggested that the trial was abandoned and replaced by a case‐control study.

Allocation

The randomisation process was clearly reported and adequate in five studies (De Villiers 2009; FLUCAD 2008; Langley 2011a; Phrommintikul 2011; Wu 2010); the others were unclear about method of randomisation. Only two studies had clearly adequate allocation concealment (FLUCAD 2008; Wu 2010).

Blinding

All of the studies reported blinding of outcome assessment. Participants and personnel were blinded in five trials (De Villiers 2009; FLUCAD 2008; Govaert 1994; Langley 2011a; Wu 2010). One trial did not clearly describe whether participants and personnel were blinded (FLUVACS 2002); in the IVCAD 2009 study, only participants were blinded; and in another study (Phrommintikul 2011), blinding was not possible as the control was 'no intervention'.

Incomplete outcome data

Attrition bias was adequately addressed in most trials, and two studies did not provide a clear description (IVCAD 2009; Langley 2011a). Dropouts and losses to follow‐up were generally well balanced between comparison groups. The proportion of participants who completed the study ranged between 89% and 100% (over 95% for most studies). All but two studies (Govaert 1994; IVCAD 2009) clearly used intention‐to‐treat analysis.

Selective reporting

Most studies reported outcomes as outlined in the Methods section of the respective paper. The FLUVACS 2002 study reported on certain composite outcomes that were subsequently not reported, and some of the outcomes were only selectively reported in the subgroups assessed. Information on the IVCAD 2009 study was available only in abstract form and from information provided by the study author, so whether outcomes were reported as originally planned could not be assessed.

Other potential sources of bias

Most studies reported that comparison groups were balanced at baseline with respect to the most important characteristics. A difference in the proportion of participants taking calcium channel blockers at baseline may have been seen in the FLUCAD 2008 study (significance not reported). The authors of the FLUVACS 2002 study stated that groups were balanced with respect to age, sex and signs of necrosis (undefined), but some of the other parameters reported look less well balanced (e.g. hypertension, distribution of types of MI). In the study by Phrommintikul 2011, a small significant difference was observed in participants taking angiotensin‐converting enzyme (ACE) inhibitors/angiotensin II receptor blockers at baseline. Wu 2010 reported a limited number of baseline characteristics.

A power analysis was carried out and power was adequate in the following studies: De Villiers 2009; FLUCAD 2008; Langley 2011a; Phrommintikul 2011. A power analysis was carried out in the FLUVACS 2002 study, but the actual number of participants suggests that the study was underpowered for measuring the primary outcome.

Cardiovascular death

Cardiovascular death was reported by all secondary prevention studies and by three of the studies reporting mortality as part of their safety analyses. Overall analyses for cardiovascular death are shown in Analysis 1.2.

In the secondary prevention studies, significantly fewer cardiovascular deaths occurred in the vaccine group than in the control group (risk ratio (RR) 0.45, 95% confidence interval (CI) 0.26 to 0.76, P value 0.003) with no significant heterogeneity between studies. Cardiovascular death occurred in 2.3% of participants in the vaccine groups and in 5.1% of those in the control groups. Among individual studies, only FLUVACS 2002 reported a significantly lower cardiovascular death rate in the vaccine group than in the placebo group (6.2% vs 17.7%, P value 0.002) at one year; this was not maintained at two‐year follow‐up (P value 0.14). Among subgroups of participants with acute coronary syndromes/MI and those with stable angina/percutaneous cardiac interventions in the FLUCAD 2008 and FLUVACS 2002 studies, no significant difference in cardiovascular death was seen in either of the subgroups overall.

In studies reporting cardiovascular death as part of their safety analyses, no significant difference in cardiovascular death was observed between vaccine and placebo groups. Cardiovascular death occurred in 0.7% of participants in the vaccine groups and in 0.8% of participants in the placebo groups.

The outcome reported by De Villiers 2009 was death from all causes; the distribution of causes of death was not given for the comparison groups, but study authors stated that the most frequent causes of death were MI, cardiac failure and cerebrovascular disorders, and that these occurred in statistically similar proportions of vaccine and placebo recipients (all P values ≥ 0.422; Analysis 1.1).

Other cardiovascular outcomes

Analysis 1.4 to Analysis 1.14 show results for other cardiovascular events as reported by study investigators.

Studies targeting a reduction in cardiovascular events reported a range of cardiovascular outcomes.

In the FLUCAD 2008 study, no significant differences in major adverse coronary events (MACE ‐ cardiovascular death, MI or coronary revascularisation) over 12 months were reported in the vaccine group compared with the placebo group (3.00% vs 5.87%, P value 0.13; Analysis 1.3). However, the rate of coronary ischaemic events (MACE or hospitalisation for myocardial ischaemia) over 12 months was significantly lower in the vaccine group than in the placebo group (6.02% vs 9.97%, P value 0.047; Analysis 1.4). This applied to the participant population overall and to the subgroup with an acute coronary syndrome, but no significant difference in coronary ischaemic events was noted in the subgroup with stable angina. No significant difference was reported between groups in terms of individual cardiac events, namely, cardiovascular death, MI (ST segment elevation MI (STEMI) or non‐ST segment elevation MI (NSTEMI)), coronary revascularisation or hospitalisation for myocardial ischaemia (see Analysis 1.15; Analysis 1.10; Analysis 1.14). This applied to the participant population overall and to both subgroups (those with an acute coronary syndrome and those with stable angina). Multi‐variate analyses identified female sex (hazard ratio (HR) 2.15, 95% CI 1.11 to 4.15%, P value 0.024) and recent acute coronary syndrome (HR 2.93, 95% CI 1.52 to 5.65, P value 0.001) as predictors of coronary ischaemic events, and influenza vaccination was found to be protective (HR 0.38, 95% CI 0.19 to 0.78, P value 0.009). No effects were seen for age, smoking, hypertension, elevated low‐density lipoprotein (LDL)‐cholesterol, low high‐density lipoprotein (HDL)‐cholesterol or presence of diabetes mellitus.

In the FLUVACS 2002 study (see Analysis 1.5; Analysis 1.6; Analysis 1.15; Analysis 1.10; Analysis 1.14), the triple endpoint (acute MI, rehospitalisation for recurrent angina, death) was seen less frequently in the vaccine group than in the placebo group for the whole population (RR 0.50, 95% CI 0.29 to 0.85, P value 0.009) and in the subgroup with acute MI (RR 0.42, 95% CI 0.21 to 0.83, P value 0.008) but not for the PCI stenting subgroup at six months. No significant difference in MI was reported in either subgroup. Also no significant differences in cardiac revascularisation were noted in the PCI stenting group. Among participants with acute MI, significantly fewer were rehospitalised for ischaemia in the vaccine group (RR 0.33, 95% CI 0.11 to 1.0, P value 0.03), and significantly fewer participants had a double endpoint (of reinfarction, hospitalisation for ischaemia and death) (RR 0.30, 95% CI 0.22 to 1.28, P value 0.03). These results were largely maintained at one‐year follow‐up. In the whole study group, 22% of participants had a triple endpoint in the vaccine group and 37% in the placebo group (RR 0.59, 95% CI 0.4 to 0.86, P value 0.004). The difference was also significant in the subgroup with acute MI at one year (19% vs 42%, RR 0.44, 95% CI 0.27 to 0.71, P value 0.0003) but not in the PCI stenting group. Significant benefit of the vaccination for occurrence of the double endpoint was seen only in the MI subgroup (10% vs 28%, RR 0.37, 95% CI 0.19 to 0.72, P value 0.002), not in the whole study population. No significant difference in MI or rehospitalisation for ischaemia was seen. Cox regression analyses were done for the combined triple endpoint at one year for the following subgroups: participants older or younger than 65 years, those with ST segment elevation or non‐ST segment elevation MI, participants with elevation or no elevation of enzymes at entry, those with diabetes or no diabetes, participants with a history or no history of smoking, those with thrombolysis in myocardial infarction (TIMI) risk score below or above 6 and participants with a history or no history of revascularisation. Significantly greater benefit with influenza vaccination was seen among participants with non‐ST segment elevation MI (RR 0.13, 95% CI 0.03 to 0.52, P value 0.004), those older than 65 years (RR 0.36, 95% CI 0.14 to 0.92, P value not given), non‐smokers (RR 0.18, 95% CI 0.05 to 0.57, P value not given) and those at high risk for future ischaemic episodes (TIMI risk score > 6) (RR 0.22, 95% CI 0.06 to 0.87, P value not given).

The IVCAD 2009 study reported no significant differences between the vaccination group and the placebo group in individual cardiovascular outcomes over one year (CABG, PCI, MI, episodes of unstable angina, cardiovascular death; see Analysis 1.8; Analysis 1.9; Analysis 1.10; Analysis 1.13). Significantly more participants with at least one cardiac adverse event (acute coronary syndrome, coronary revascularisation, cardiovascular death) were reported in the placebo group than in the vaccine group at six months, but this finding was not maintained at one‐year follow‐up (29% in the intervention group vs 26% in the placebo group, P value 0.6; Analysis 1.7). Study authors also reported that angina severity scores were more improved in the vaccine group than in the placebo group.

In the study by Phrommintikul 2011 (see Analysis 1.3; Analysis 1.11; Analysis 1.12; Analysis 1.14), participants in the vaccine group had significantly fewer major adverse coronary events (MACE, composite of cardiovascular death or hospitalisation for acute coronary syndrome, heart failure or stroke) at 12 months than did participants in the placebo group (9.5% vs 19.3%, unadjusted HR 0.70, 95% CI 0.57 to 0.86, P value 0.004). The rate of hospitalisation for acute coronary syndrome was also significantly lower in the vaccine group than in the placebo group (4.5% vs 10.6%, unadjusted HR 0.73, 95% CI 0.55 to 0.91, P value 0.032). Effects for these two outcomes remained significant with adjustments for age, sex, serum creatinine, ACE inhibitor treatment and coronary revascularisation. No significant difference was reported for cardiovascular death, hospitalisation for heart failure or hospitalisation for stroke. No significant difference in effects for MACE was observed for the following subgroups: participants younger than 65 years or older than 65 years, male or female participants, those with or without diabetes mellitus, participants with ST segment elevation or non‐ST segment elevation MI, those with serum creatinine levels less than 1.1 mg/dL or greater than 1.1. mg/dL and participants with or without revascularisation.

Limited information on non‐fatal cardiovascular adverse events was reported by studies that included cardiovascular events in their safety analyses. Govaert 1994 reported three cases of ‘intercurrent illness (undefined), cerebrovascular accident’ ‐ two in the placebo group (0.2%) and one in the vaccine group (0.1%) (RR 0.49, 95% CI 0.04 to 5.41). Wu 2010 reported five cases of angina in the intervention groups (0.8%) and none in the placebo groups (RR 1.83, 95% CI 0.1 to 32.85).

Adverse events

De Villiers 2009 reported that more reactogenicity events (including runny nose/nasal congestion, cough, sore throat, headache, muscle aches, tiredness and decreased appetite) occurred in the vaccine group 11 days after vaccination (P value 0.042). No significant differences in serious adverse events were observed between vaccination and placebo groups over four weeks (1% vs 1.5%, P value 0.27) or eight months (10.1% vs 8.6%) post vaccination.

In the study by Langley 2011a, no significant differences were observed between vaccine and placebo groups over one year in reporting of one or more serious adverse events (3.2% vs 4.0%) or reporting of one or more medically attended events (30% vs 30.4%).

Wu 2010 reported the following rates of adverse events for the different intervention groups over 42 days: whole virion + Al 5 µg 12.9%, whole virion + Al 10 µg 19.6%, split virion + Al 7.5 µg 14.9%, split virion + Al 15 µg 12.0%, split virion 15 µg 7.9%, split virion 30 µg 11.9% and placebo 7.0%. The highest rate was seen with whole virion + Al 10 µg vaccine (P value 0.016 vs whole virion + Al 15 µg), but no significant association with dosage, presence of aluminium adjuvant or type of vaccine (whole virion vs split virion) was described. The rate of mild adverse events ranged between 5.0% and 17.6%, that of moderate adverse events between 0 and 2.0% and that of severe adverse events between 0 and 1%.

No adverse events beyond the cardiovascular outcomes already summarised were reported in the following studies: FLUCAD 2008, Govaert 1994, IVCAD 2009 and Phrommintikul 2011.