Scolaris Content Display Scolaris Content Display

Base de datos Cochrane de Revisiones Sistemáticas Revisión - Intervención

Atención médica en hospitales de día para personas mayores versus formas alternativas de atención

Declaraciones de intereses de los autores

Versión publicada: 23 junio 2015 Historial de versiones

https://doi.org/10.1002/14651858.CD001730.pub3
Contraer todo Desplegar todo

Resumen

disponible en

Antecedentes

La proporción de la población mundial mayor de 60 años está aumentando. Por lo tanto, es necesario examinar diferentes métodos de prestación de servicios de salud para esta población. Los hospitales de día ofrecen servicios de salud multidisciplinarios a las personas mayores en un solo lugar.

Objetivos

Evaluar la eficacia de los hospitales de día para personas mayores en la prevención de la muerte, la discapacidad, la institucionalización y la mejora del estado de salud subjetivo.

Métodos de búsqueda

Se hicieron búsquedas en el Registro de Estudios del Grupo Cochrane para una Práctica y Organización Sanitaria Efectivas (Cochrane Effective Practice and Organisation of Care, EPOC), CENTRAL (2013, número 7), MEDLINE vía Ovidio (1950‐2013), EMBASE vía Ovidio (1947‐2013) y CINAHL vía EbscoHost (1980‐2013). También se realizaron búsquedas de referencias citadas, se buscaron las actas de conferencias y los registros de ensayos, se realizaron búsquedas manuales en revistas seleccionadas y se estableció contacto con los autores e investigadores pertinentes para solicitar datos adicionales.

Criterios de selección

Ensayos aleatorizados y cuasialeatorizados que comparan los hospitales de día con formas alternativas de atención para las personas mayores (media/mediana > 60 años de edad).

Obtención y análisis de los datos

Dos autores evaluaron de forma independiente la elegibilidad de los ensayos y el riesgo de sesgo y extrajeron los datos de los ensayos incluidos. Se utilizaron los procedimientos metodológicos estándar previstos por la Colaboración Cochrane. Los ensayos fueron subcategorizados como atención integral, atención domiciliaria o sin atención integral.

Resultados principales

Dieciséis ensayos (3689 participantes) compararon los hospitales de día con la atención integral (cinco ensayos), la atención domiciliaria (siete ensayos) o ninguna atención integral (cuatro ensayos). En general, hubo una calidad moderada de la evidencia de estos estudios para los siguientes resultados.

En cuanto al resultado de la muerte, no hubo pruebas sólidas a favor o en contra de los hospitales de día en comparación con otros tratamientos en general (odds ratio (OR) 1,05; IC del 95%: 0,85 a 1,28; P = 0.66), o a la atención integral (OR 1,26; IC del 95%: 0,87 a 1,82; P = 0,22), a la atención domiciliaria (OR 0,97; IC del 95%: 0,61 a 1,55; P = 0,89), o a la ausencia de atención integral (OR 0,88; IC del 95%: 0,63 a 1,22; P = 0,43).

Para el resultado de muerte o deterioro en las actividades de la vida diaria (ADL), no hubo evidencia contundente para la atención en hospital de día en comparación con otros tratamientos (OR 1,07; IC del 95%: 0,76 a 1,49; P = 0.70), o a la atención integral (OR 1,18; IC del 95%: 0,63 a 2,18; P = 0,61), a la atención domiciliaria (OR 1,41; IC del 95%: 0,82 a 2,42; P = 0,21) o a la ausencia de atención integral (OR 0,76; IC del 95%: 0,56 a 1,05; P = 0,09).

En cuanto al resultado de muerte o mal resultado (atención institucional, dependencia, deterioro de la función física), no hubo evidencia contundente para los hospitales de día en comparación con otros tratamientos (OR 0,92; IC del 95%: 0,74 a 1,15; P = 0,49), o en comparación con la atención integral (OR 1,05; IC del 95%: 0,79 a 1,40; P = 0,74) o la atención domiciliaria (OR 1,08; IC del 95%: 0,67 a 1,74; P = 0,75). Sin embargo, en comparación con la ausencia de atención integral, hubo una diferencia a favor de los hospitales de día (OR 0,72; IC del 95%: 0,53 a 0,99; P = 0,04).

Para el resultado de muerte o atención institucional, no hubo evidencia contundente para los hospitales de día en comparación con otros tratamientos en general (OR 0,85; IC del 95%: 0,63 a 1,14; P = 0,28), o con la atención integral (OR 1,00; IC del 95%: 0,69 a 1,44; P = 0,99), la atención domiciliaria (OR 1,05; IC del 95%: 0,57 a 1,92; P = 0,88) o la ausencia de atención integral (OR 0,63; IC del 95%: 0,40 a 1,00; P = 0,05).

Para el resultado de deterioro de la AVD, no hubo evidencia contundente de que la asistencia en el hospital de día tuviera un efecto diferente al de otros tratamientos en general (OR 1,11; IC del 95%: 0,68 a 1,80; P = 0,67) o en comparación con la atención integral (OR 1,21; 0,58 a 2,52; P = 0,61), o la atención domiciliaria (OR 1,59; IC del 95%: 0,87 a 2,90; P = 0,13). Sin embargo, los pacientes de los hospitales de día mostraron una reducción de las probabilidades de deterioro en comparación con los que no recibían atención integral (OR 0,61; IC del 95%: 0,38 a 0,97; P = 0,04) y diferencias significativas en los subgrupos (P = 0,04).

Para el resultado de requerir atención institucional, no hubo evidencia contundente para los hospitales de día en comparación con otros tratamientos (OR 0,84; IC del 95%: 0,58 a 1,21; p = 0,35), o con la atención integral (OR 0,91; IC del 95%: 0,70 a 1,19; p = 0,49), la atención domiciliaria (OR 1,49; IC del 95%: 0,53 a 4,25; p = 0,45), o la ausencia de atención integral (OR 0,58; IC del 95%: 0,28 a 1,20; p = 0,14).

Conclusiones de los autores

Hay evidencia de baja calidad que los hospitales de día parecen ser eficaces en comparación con la falta de atención integral para el resultado combinado de muerte o mal resultado, y para el deterioro en las AVD. No hay evidencia clara de otros resultados, ni de una ventaja sobre la prestación de otros tipos de atención médica.

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.

Resumen en términos sencillos

disponible en

Atención médica en hospitales de día para personas mayores frente a formas alternativas de atención

Los hospitales de día son una forma de proporcionar asistencia sanitaria a las personas mayores. Son instalaciones ambulatorias a las que los pacientes mayores acuden durante un día completo o casi completo y reciben atención médica multidisciplinar "bajo un mismo techo". En esta revisión se incluyeron 16 ensayos con 3689 participantes y se compararon los hospitales de día con otros servicios integrales (incluidos los servicios para pacientes hospitalizados y ambulatorios), la atención domiciliaria y ningún servicio integral. La asistencia a un hospital de día ofrece beneficios en comparación con la ausencia de tratamiento, entre los que se incluye la reducción del riesgo de necesitar más ayuda en las actividades de la vida diaria como lavarse o vestirse. Además, los pacientes tienen menos probabilidades de sufrir uno de los siguientes problemas: morir, ser institucionalizado o volverse más dependiente de los demás. No hay ningún beneficio aparente cuando se comparan los hospitales de día con otros servicios integrales o de atención domiciliaria. El valor económico de los hospitales de día en comparación con otros servicios de atención de la salud permanece incierto.

Authors' conclusions

Implications for practice

Any conclusions are limited by the relatively small amount of data available and the low quality of the data contributing to a number of important outcomes. Day hospital care appears to be an effective outpatient service for older people, but no more effective (at least for the outcomes examined in this review), and possibly more expensive, than other forms of comprehensive care. These findings do not support the closure of day hospital services but do support the exploration of alternative systems for delivering an equivalent or superior form of comprehensive care. Our findings support the view that day hospital attendance needs to be carefully monitored (George 1989) and the staff and facilities used as flexibly and efficiently as possible (Brocklehurst 1995).

Implications for research

The findings are limited by the relatively small amount of data available and overall low quality of the evidence; further research is likely to impact on our confidence in the estimate of effect. Further randomised trials are justifiable and should focus on comparing services which aim to provide an equivalent intervention to day hospital care (e.g. domiciliary care). Given the diversity of patients attending day hospitals and the corresponding diversity of day hospital interventions employed, future trials should be large, multi‐centre trials or should examine more focused questions. Outcomes should include subjective health status and carer well being. There is concern that commonly used measures of disability lack sensitivity to change in the outpatient setting of a day hospital due to their ceiling effect (Parker 1994). Future trials should incorporate measures of instrumental activities of daily living as a more relevant and potentially more sensitive outcome. Furthermore, future trials should incorporate adequate methods of randomisation and allocation concealment, and undertake blinded outcome assessment where possible, as well as ensuring that methods are adequately reported.

Summary of findings

Open in table viewer
Summary of findings for the main comparison. Day hospitals compared to alternative care or no care for rehabilitation needs

Day hospitals compared to alternative or no care for rehabilitation

Patient or population: patients with rehabilitation needs
Intervention: day hospitals
Comparison: alternative care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Alternative or no care

Day hospitals

Death by the end of follow up
Follow‐up: median 12 months

Study population

OR 1.05
(0.85 to 1.28)

3533
(16 studies)

⊕⊕⊝⊝
low1,2,3

127 per 1000

132 per 1000
(110 to 157)

Moderate

66 per 1000

69 per 1000
(57 to 83)

Death or institutional care by the end of follow up
Follow‐up: median 12 months

Study population

OR 0.85
(0.63 to 1.14)

3030
(13 studies)

⊕⊕⊝⊝
low1,2,3

303 per 1000

270 per 1000
(215 to 331)

Moderate

221 per 1000

194 per 1000
(152 to 244)

Death or deterioration in activities of daily living (ADL)
Follow‐up: median 12 months

Study population

OR 1.07
(0.76 to 1.49)

1268
(7 studies)

⊕⊕⊝⊝
low1,2,3

407 per 1000

423 per 1000
(343 to 506)

Moderate

430 per 1000

447 per 1000
(364 to 529)

Death or poor outcome (institutional care, disability or deterioration)
Follow‐up: median 12 months

Study population

OR 0.92
(0.74 to 1.15)

2831
(13 studies)

⊕⊕⊝⊝
low1,2,3

365 per 1000

346 per 1000
(299 to 398)

Moderate

241 per 1000

226 per 1000
(190 to 267)

Deterioration in ADL in survivors
Various ADL measures

Study population

OR 1.11
(0.68 to 1.8)

905
(7 studies)

⊕⊕⊝⊝
low1,2,3

251 per 1000

271 per 1000
(185 to 376)

Moderate

233 per 1000

252 per 1000
(171 to 354)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; OR: Odds ratio; ADL: activities of daily living

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Limitations for at least one risk of bias criterion, or some limitations for multiple criteria, sufficient to lower confidence in the estimate of effect
2 Whilst there was evidence of heterogeneity, this was anticipated due to the diversity of the populations and of the interventions
3 Wide CIs

Open in table viewer
Summary of findings 2. Day hospitals compared to no comprehensive care for rehabilitation needs

Day hospitals compared to no comprehensive care for rehabilitation needs

Patient or population: patients with rehabilitation needs
Intervention: day hospitals
Comparison: no comprehensive care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

No comprehensive care

Day hospitals

Death by the end of follow up
Follow‐up: median 12 months

Study population

OR 0.88
(0.63 to 1.22)

1345
(4 studies)

⊕⊕⊝⊝
low1,2,3

128 per 1000

114 per 1000
(85 to 152)

Moderate

131 per 1000

117 per 1000
(87 to 155)

Death or institutional care by the end of follow up
Follow‐up: median 12 months

Study population

OR 0.63
(0.4 to 1)

1177
(4 studies)

⊕⊕⊝⊝
low1,2,3

248 per 1000

172 per 1000
(117 to 248)

Moderate

307 per 1000

218 per 1000
(151 to 307)

Death or deterioration in ADL
Follow‐up: median 9 months

Study population

OR 0.76
(0.56 to 1.05)

651
(2 studies)

⊕⊕⊝⊝
low1,2,3

436 per 1000

370 per 1000
(302 to 448)

Moderate

446 per 1000

380 per 1000
(311 to 458)

Death or poor outcome (institutional care, disability or deterioration)
Follow‐up: median 12 months

Study population

OR 0.72
(0.53 to 0.99)

982
(3 studies)

⊕⊕⊝⊝
low1,2,3

347 per 1000

277 per 1000
(220 to 345)

Moderate

400 per 1000

324 per 1000
(261 to 398)

Deterioration in ADL in survivors
Follow‐up: median 9 months

Study population

OR 0.61
(0.38 to 0.97)

407
(2 studies)

⊕⊕⊝⊝
low1,2,3

277 per 1000

189 per 1000
(127 to 271)

Moderate

227 per 1000

152 per 1000
(100 to 222)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; OR: Odds ratio; ADL: activities of daily living

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Limitations for at least one risk of bias criterion or some limitations for multiple criteria, sufficient to lower confidence in the estimate of effect
2 Whilst there was evidence of heterogeneity, this was anticipated due to the diversity of the population and of the study design
3 Wide CIs

Open in table viewer
Summary of findings 3. Day hospitals compared to domiciliary care for rehabilitation needs

Day hospitals compared to domiciliary care for rehabilitation

Patient or population: patients with rehabilitation needs
Intervention: day hospitals
Comparison: domiciliary care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Domiciliary care

Day hospitals

Death by the end of follow up
Follow‐up: median 6

Study population

OR 0.97
(0.61 to 1.55)

901
(7 studies)

⊕⊕⊝⊝
low1,2,3

101 per 1000

98 per 1000
(64 to 148)

Moderate

64 per 1000

62 per 1000
(40 to 96)

Death or institutional care by the end of follow up
Follow‐up: median 6 months

Study population

OR 1.05
(0.57 to 1.92)

672
(5 studies)

⊕⊕⊝⊝
low1,2,3

187 per 1000

194 per 1000
(116 to 306)

Moderate

69 per 1000

72 per 1000
(41 to 125)

Death or deterioration in ADL
Follow‐up: median 9 months

Study population

OR 1.41
(0.82 to 2.42)

443
(4 studies)

⊕⊕⊝⊝
low1,2,3

392 per 1000

476 per 1000
(346 to 609)

Moderate

334 per 1000

414 per 1000
(291 to 548)

Death or poor outcome (institutional care, disability or deterioration)
Follow‐up: median 6 months

Study population

OR 1.08
(0.67 to 1.74)

581
(5 studies)

⊕⊕⊝⊝
low1,2,3

297 per 1000

313 per 1000
(221 to 424)

Moderate

364 per 1000

382 per 1000
(277 to 499)

Deterioration in ADL in survivors
Follow‐up: median 9 months

Study population

OR 1.59
(0.87 to 2.9)

349
(4 studies)

⊕⊕⊝⊝
low1,2,3

225 per 1000

315 per 1000
(201 to 457)

Moderate

188 per 1000

269 per 1000
(168 to 402)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; OR: Odds ratio; ADL: activities of daily living

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Limitations for at least one risk of bias criterion or some limitations for multiple criteria, sufficient to lower confidence in the estimate of effect
2 Whilst there was evidence of heterogeneity, this was anticipated due to the diversity of the population and the interventions
3 Wide CIs

Open in table viewer
Summary of findings 4. Day hospitals compared to comprehensive care for elderly persons requiring rehabilitation

Day hospitals compared to comprehensive care for older people requiring rehabilitation

Patient or population: older people requiring rehabilitation
Intervention: day hospitals
Comparison: comprehensive care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Comprehensive care

Day hospitals

Death by the end of follow up
Follow‐up: median 12 months

Study population

OR 1.26
(0.87 to 1.82)

1287
(5 studies)

⊕⊕⊝⊝
low1,2,3

144 per 1000

175 per 1000
(128 to 234)

Moderate

69 per 1000

85 per 1000
(61 to 119)

Death or institutional care by the end of follow up
Follow‐up: median 12 months

Study population

OR 1
(0.69 to 1.44)

1181
(4 studies)

⊕⊕⊝⊝
low1,2,3

426 per 1000

426 per 1000
(339 to 517)

Moderate

231 per 1000

231 per 1000
(172 to 302)

Death or deterioration in ADL
Follow‐up: median 12 months

Study population

OR 1.18
(0.63 to 2.18)

174
(1 study)

⊕⊕⊝⊝
low1,3

349 per 1000

387 per 1000
(252 to 539)

Moderate

349 per 1000

387 per 1000
(252 to 539)

Death or poor outcome (institutional care, disability or deterioration)
Follow‐up: median 12 months

Study population

OR 1.05
(0.79 to 1.4)

1268
(5 studies)

⊕⊕⊝⊝
low1,2,3

410 per 1000

422 per 1000
(355 to 493)

Moderate

221 per 1000

230 per 1000
(183 to 284)

Deterioration in ADL in survivors
Follow‐up: median 12 months

Study population

OR 1.21
(0.58 to 2.52)

149
(1 study)

⊕⊕⊝⊝
low1,3

243 per 1000

280 per 1000
(157 to 448)

Moderate

243 per 1000

280 per 1000
(157 to 447)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; OR: Odds ratio; ADL: activities of daily living

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Limitations for at least one risk of bias criterion or some limitations for multiple criteria, sufficient to lower confidence in the estimate of effect
2 Whilst there was evidence of heterogeneity, this was anticipated due to the diversity of the population and of the study design
3 Wide CIs

Background

The first geriatric day hospital was opened in the UK in 1952 (Farndale 1961). Day hospitals developed rapidly in the United Kingdom in the 1960's as an important component of care provision for older people designed to complement in‐patient services (Black 2005). The model has since been widely applied in New Zealand, Australia, Canada, the USA and several European countries.

Geriatric day hospitals provide multi‐disciplinary rehabilitation in an outpatient setting and operate in a pivotal position between hospital and home‐based services (Ames 1995; Black 2005; Brocklehurst 1973; Petermans 2011). They provide specialist services for older people, which can include examinations and consultations, all concentrated in one location (Bussche 2010).

Although there is considerable descriptive literature on day hospital care (RCP 1994), concern has been expressed that evidence for effectiveness is equivocal (Brocklehurst 1980; Donaldson 1986) and that day hospital care is expensive (NAO 1994).

Concern is often expressed about the most appropriate health and social services required to address the needs of an aging population. In the UK, for example, the largest population increase is seen in the over 85 age group. A range of different services models, of which the day hospital is one, may be appropriate to address these needs. This review sets out to examine the effectiveness and resource implications of geriatric medical day hospital attendance for older people and to compare it with other models of healthcare delivery for an older population. This is an updated Cochrane review first published in Forster 1999a.

Description of the condition

Geriatric day hospitals are not usually specific to one condition. However, many will provide rehabilitation services appropriate to conditions such as stroke that are likely to be seen in an older population.

Description of the intervention

Geriatric day hospitals are out‐patient healthcare facilities for older people living in the community. They provide multi‐professional treatment on a full or part time basis (Beynon 2009).They serve several functions, including assessment, rehabilitation, and medical, nursing, maintenance, social and respite care (Brocklehurst 1980). Rehabilitation and maintenance comprise the main work of the day hospital: 42% and 23% respectively (RCP 1994), with rehabilitation regarded as the most important function (Brocklehurst 1980). The specific features and services offered by individual geriatric day hospitals are subject to considerable variation. However, they usually include a combination of medical assessment with support from nurses and allied health professionals, often including physiotherapists and occupational therapists. There is no consensus on what types of healthcare professionals should make up the multi‐disciplinary teams (Petermans 2011). Additional services such as chiropody, social work, exercise classes and assistance with bathing and hair washing are offered by some hospitals.

How the intervention might work

Geriatric day hospitals offer a multidisciplinary approach to assessment and rehabilitation, with provision of a variety of services in one location. As a result of assessment and treatment occurring 'under one roof,’ the health requirements of older people should be identified and responded to in an appropriate and timely manner. The day hospital can provide out‐patient delivery of a Comprehensive Geriatric Assessment (CGA) which has a robust evidence base for inpatient setting use (Ellis 2011). A CGA addresses medical, physical, psychological and social needs, and includes the formation of a plan of care and rehabilitation, with a clear method of implementation. Day hospital staff have specific skills, knowledge and experience related to working with older people. Furthermore, the day hospital environment has the advantage of providing social interaction between patients, a factor which domiciliary services and usual care cannot provide. These factors could result in better outcomes for patients through the provision of effective rehabilitation and other healthcare delivery for an older population.

Why it is important to do this review

Between 1985 and 2010 the proportion of the world's population that is aged over 65 years grew by approximately a quarter from 6.0% (291 million) to 7.6% (524 million), and is expected to increase to 13% by 2035, exceeding a billion people globally (UN 2011). As a result of this increase, providing health care that meets the diverse needs of an older population and is cost effective and efficient will be ever more important. Day hospitals are one way of delivering multidisciplinary rehabilitation to older people in an outpatient setting. This review is necessary to assess the effectiveness of day hospitals across a number of health, cost and resource outcomes.

Objectives

The primary question was whether older patients attending a geriatric medical day hospital would experience better outcomes (in terms of death, dependency or institutionalisation) than those receiving alternative forms of care.

Secondary questions concerned the impact of day hospital care on patient satisfaction and subjective health outcomes, carer distress and resource use and costs.

Methods

Criteria for considering studies for this review

Types of studies

We included studies that were of a prospective, controlled design in which there was random assignment of participants to alternative treatment groups (one of which involved day hospital care), not as part of a complex multi‐service intervention. Studies which utilised ‘quasi’ randomisation procedures (for example allocation to groups based on date of birth) were also included.

Types of participants

We included patients receiving medical care (mean/median age of >60 years for individual studies). We are aware that day hospital descriptive studies have indicated that day hospital attendance is determined more by needs than age and that younger patients do attend day hospitals. Our pre‐specified participant criterion of age 60 years and over was chosen to pragmatically capture this clinical practice. Studies which were specific to psychiatric patients were excluded.

Types of interventions

We defined a day hospital as an out‐patient facility where older patients attend for a full or near full day and receive multidisciplinary rehabilitation in a healthcare setting. This is consistent with previous definitions (Siu 1994) and excluded trials evaluating social day centres, or other types of day hospitals such as psychiatric day hospitals for patients with dementia or psychiatric conditions. We excluded studies on day hospitals that only provided services for single, specific conditions (for example, arthritis). We wanted to assess the effects of providing typical general assessment and rehabilitation services relevant to older people. The inclusion of disease‐specific trials would risk incorporating the effects of very specific therapies for specific conditions, which were not the focus of this review.

We anticipated considerable heterogeneity, particularly in the control services, and so pre‐specified key subgroup comparisons prior to reviewing the trials.

1) Day hospital care versus comprehensive care ‐ where control patients had access to a range of geriatric medical services (both inpatient and outpatient).
2) Day hospital care versus domiciliary care ‐ where control patients were provided an approximately equivalent rehabilitation program within their own home or social day centre.
3) Day hospital care versus no comprehensive care ‐ where control patients did not routinely have access to outpatient rehabilitation services.

Types of outcome measures

We wished to identify outcomes which reflected a previous definition of the purpose of day hospital care: to facilitate and prolong independent living for older people in the community (Donaldson 1987). Effective day hospital care would thus be expected to reduce death, to maintain older people in their own home and to reduce admissions to hospital. The following outcomes were therefore selected, all of which were recorded at the end of scheduled follow up.

Primary outcomes

Primary outcomes were:

  • death;

  • the need for institutional care;

  • dependency;

  • global 'poor outcome' comprising death or one of the following (in order of preference): resident in institutional care, severe dependency at end of follow up, or deterioration in physical function during follow up; this outcome was included in anticipation of incomplete data sets.

Secondary outcomes

Secondary outcomes included:

  • dependency, measured by activities of daily living (ADL) scores;

  • patient satisfaction;

  • subjective health status (including mood);

  • resource use (in hospital or institutional care) plus overall cost analyses;

  • carer distress.

We considered all studies that met the eligibility criteria for study design, participants and interventions regardless of whether the pre‐specified primary or secondary outcomes were reported.

Search methods for identification of studies

For this edition of the review, D Andre, University of Leeds Library, developed search strategies in consultation with the authors. They were peer reviewed by M Fiander, EPOC Trials Search Coordinator. We searched the databases listed below for relevant studies.

Electronic searches

  • Effective Practice and Organisation of Care (EPOC) group register of trials (August 2013);

  • Cochrane Central Register of Controlled Trials (CENTRAL; The Cochrane Library, Issue 7, July 2013; Appendix 1);

  • MEDLINE (1996 to July 2013; Appendix 2);

  • Medline in Process (1996 to August 2013; Appendix 3);

  • EMBASE (1996 to August 2013; Appendix 4);

  • Cumulative Index to Nursing and Allied Health Literature (CINAHL; 1996 to August 2013; Appendix 5);

  • Allied and Complementary Medicine Database (AMED; 1996 to August 2013; Appendix 6);

  • Physiotherapy Evidence Database (PEDro; August 2008; Appendix 7);

  • Applied Social Science Index and Abstracts (ASSIA; 1996 to August 2013; Appendix 8);

  • International Bibliography of Social Sciences (IBSS; 1996 to August 2013; Appendix 9);

  • PsycINFO (1996 to August week 1, 2013; Appendix 10);

  • Health Management Information Consortium Database (HMIC; January 2008 to August 2013; Appendix 11);

  • NHS Economic Evaluation Database (NHS EED; searched October 2013; Appendix 12);

  • Health Technology Assessment (HTA) Database (searched October 2013; Appendix 12);

  • Web of Knowledge (1996 to August 2013; Appendix 13);

  • Web of Science, Conference Proceedings Citation Index ‐ Social Science (1990 to 2012; Appendix 14);

  • Google Scholar (searched August 2013; Appendix 15);

  • Index to Theses (1996 to August 2013; Appendix 16);

  • Proquest Dissertations and Theses (1996 to August 2013; Appendix 17);

  • Current Controlled Trials (searched August 2013; Appendix 18).

Search strategies were comprised of natural language (free text) terms and controlled vocabulary (index) terms. Language limits were not applied. Search strategies for this update have been revised in order to improve sensitivity and precision. Changes were made based on an analysis of indexing terms found on previously included studies and by testing terms from the original strategy for precision. Given these changes, searches have been run retrospectively. The results of this search have been de‐duplicated from searches we carried out for the previous update of this review in 2008. The reference list of reviews of potential relevance were also examined (Bours 1998; Mason 2007; Outpatient Service 2004; Prvu Bettger 2007; Petermans 2011).

Searching other resources

  • HSRProj (searched August 2013; Appendix 19);

  • National Research Register (searched September 2007);

  • Australian New Zealand Clinical Trials Registry (May 2008).

Data collection and analysis

Selection of studies

Two review authors independently assessed the titles and abstracts from the electronic searches and excluded obviously irrelevant studies. We obtained full text articles of the remaining studies and at least two review authors independently assessed these against pre‐specified inclusion criteria to determine which trials would be eligible for inclusion. Study authors were contacted for further details when necessary. Disagreements were resolved by discussion with other members of the review team.

Data extraction and management

At least two review authors extracted data independently. Disagreements were resolved through group consensus. When possible, we contacted study authors for additional information and data as required.

Assessment of risk of bias in included studies

Two review authors independently assessed risk of bias in the included studies using the tool for assessing risk of bias in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011). We scored each study as being at ‘high risk of bias’, ‘low risk of bias’ or ‘unclear risk of bias’ for each of the following domains, and reported them in the 'Risk of bias' tables.

  • Random sequence generation (selection bias).

  • Allocation concealment (selection bias).

  • Blinding of participants and personnel (performance bias).

  • Blinding of outcome assessment (detection bias).

  • Incomplete outcome data (attrition bias).

  • Selective reporting (reporting bias).

  • Other possible bias.

Measures of treatment effect

We calculated odds ratios with 95% confidence intervals for the dichotomous outcomes using standard methods. We used a random effects model as the subjects and interventions would have differed in ways which we anticipated would affect results and we could not assume a common effect size (Borenstein 2009).

We calculated inpatient resource use as the average (mean) use of hospital beds (in days) per patient recruited to each trial group. This figure was calculated for individual trials, and groups of trials, by dividing the total number of bed days by the total number of patients.

Unit of analysis issues

In cross‐over trials, we only included data from the first period of the trial in meta‐analyses to guard against carry‐over effects. Where cluster randomised studies presented an estimate of effect that properly accounted for the cluster design, this was used. Where this was not the case, we assumed that the intra‐cluster correlation coefficient (ICC) was the same as for other studies included in the review for that outcome. We calculated an average ICC for the outcome and corrected the values for each unadjusted study by the design effect (Higgins 2011).

Dealing with missing data

Where possible, studies were analysed on an intention‐to‐treat basis. Patients who were lost to follow up or for whom outcome data were not available were excluded from the initial analysis. However, they were included in ‘best case’ (all missing data in favour of day hospital care), intermediate and ‘worst case’ (all missing data in favour of alternative care) sensitivity analyses.

Assessment of heterogeneity

We assessed heterogeneity using I² and the Q statistic, with P < 0.1 determining significant heterogeneity (Higgins 2011).

Assessment of reporting biases

We attempted to reduce the risk of reporting bias by undertaking comprehensive searches of multiple databases and trials registers, and contacting authors. Where sufficient studies were included for individual outcomes, we undertook visual inspection of funnel plots to identify any obvious sources of publication bias.

Data synthesis

For patient outcomes, we undertook meta‐analyses at the end of follow up for the domains of death, death or institutional care, death or deterioration in ADL, death or poor outcome (institutional care, disability or deterioration) and deterioration in ADL in survivors. Analyses were based on the published summary data rather than individual patient data. For other patient outcomes ‐ ADL, subjective health status and patient satisfaction ‐ we present a narrative summary and a summary of the data is provided in the Data and analyses section. A summary of carer outcomes is also presented in the Data and analyses section. To investigate resource use, we performed a meta‐analysis for the domain of requiring institutional care at the end of follow up. For hospital bed use during follow up and cost we present a narrative summary in the Data and analyses section. We assessed the quality of the evidence using the GRADE approach which results in a quality score of high, moderate, low or very low (GRADEpro 2014).

Results

Description of studies

This review update includes 16 randomised controlled trials of medical day hospital versus alternative forms of care for older people. This includes three new studies in addition to the 13 studies from the previous version of this review (Forster 2008).

Results of the search

We screened over 25,000 unique citations and reviewed the full text of 190. 153 full papers and 34 abstracts were obtained and reviewed by a minimum of two reviewers to assess eligibility. Of these, three were included; five are awaiting assessment pending translation or availability of a published report (Studies awaiting classification); the majority of studies were excluded for reasons described in Characteristics of excluded studies; a further 131 studies were excluded for this update but not reported, as they were excluded early in the selection process. A flow diagram of the review update process can be found in Figure 1.

Included studies

Three new studies have been added to this review update (Crotty 2008; Masud 2006; Parker 2009). Five studies are currently awaiting assessment.

Interventions

The current analysis includes 16 trials comprising 37 individual day hospitals. In accordance with the definition of day hospital used, multidisciplinary outpatient rehabilitation was available at all sites. Several of the studies evaluated more than one day hospital; the pilot study undertaken by Vetter 1989 involved two, Masud 2006 and Crotty 2008 each involved three, while a further four trials (Hedrick 1993; Parker 2009; Weissert 1980; Young 1992) each evaluated four day hospitals and Roderick 2001 involved five. The studies were undertaken in various countries including the UK (Burch 1999; Gladman 1993; Masud 2006; Parker 2009; Roderick 2001; Vetter 1989; Woodford 1962; Young 1992), USA (Cummings 1985; Hedrick 1993; Weissert 1980), Australia (Crotty 2008), Canada (Eagle 1991), Hong Kong (Hui 1995), Finland (Pitkala 1991) and New Zealand (Tucker 1984). For further details see Characteristics of included studies.

Comparison groups

Attendance at a day hospital was evaluated against various comparison treatments which were grouped together in the following sub‐categories:

1) In five studies the comparison treatment was comprehensive care comprising a range of inpatient, outpatient and domiciliary geriatric medical services (Cummings 1985; Eagle 1991; Hedrick 1993; Pitkala 1991; Tucker 1984).

2) In seven trials the comparison treatment was domiciliary therapy. This was provided in the patient's home (Crotty 2008; Gladman 1993; Parker 2009; Roderick 2001; Vetter 1989; Young 1992) or day centre (Burch 1999). Three of these trials recruited stroke patients only (Gladman 1993; Roderick 2001; Young 1992) and a fourth was a pilot study (Vetter 1989). In the Nottingham trial patients were randomly allocated to domiciliary rehabilitation or hospital‐based rehabilitation in three strata according to discharge ward: health care of older people, general medical unit or stroke unit (Gladman 1993). Hospital‐based rehabilitation was provided during day hospital attendance for patients in the older people care stratum and only patients in this stratum have been included in our analysis.

3) Four trials compared day hospital attendance against a control group in which patients were eligible for, but not referred to, existing services (Hui 1995; Masud 2006; Weissert 1980; Woodford 1962). In Masud 2006 the control arm received information leaflets on falls prevention and usual care from the primary care service until outcome data was completed, after which time control participants were offered access to the day hospital intervention.

We initially allocated Cummings 1985 and Hui 1995 into their own individual sub‐categories according to their comparison group; day hospital versus inpatient care (Cummings 1985) and day hospital versus medical outpatient care (Hui 1995). However, in order to streamline the analysis, these two trials were incorporated into the above categorisation schemes prior to data analysis. The Cummings 1985 trial investigated a day hospital service designed to facilitate early hospital discharge. The service offered to the comparison group was equivalent to comprehensive care and the trial was re‐categorised accordingly. The Hong Kong trial (Hui 1995) recruited stroke patients admitted to the same ward and randomised to receive rehabilitation care led by a neurology team or by a geriatrician team. After discharge, patients assigned to a neurologist were followed up at a medical outpatient clinic and the geriatrician patients by day hospital attendance. There were no differences in length of inpatient stay or dependency at discharge and the main treatment difference at final follow up assessment was the type of supporting aftercare: day hospital or medical outpatients. Further discussion with the trialists indicated that this comparison group could best be categorised as 'no comprehensive care'.

Patient characteristics

This review includes studies with a total of 3689 participants. One trial (Hedrick 1993), which was run by the United States Department of Veterans Affairs, recruited largely (96%) male patients. The other trials had a mix of male and female patients. In all but one trial the mean patient age was over 70 years; the New York trial (Cummings 1985) had a mean patient age of 65 years. Four trials (Gladman 1993; Hui 1995; Roderick 2001; Young 1992) recruited only stroke patients. Masud 2006 specifically recruited participants considered at a high risk of falling. The remaining eleven studies recruited patients with a mixture of diagnoses (Burch 1999; Crotty 2008; Cummings 1985; Eagle 1991; Hedrick 1993; Parker 2009; Pitkala 1991; Tucker 1984; Vetter 1989; Weissert 1980; Woodford 1962). The participants usually had a degree of dependency at recruitment as judged by their ADL scores (for further details see Characteristics of included studies).

Excluded studies

The majority of studies were excluded for reasons including a lack of randomisation, intervention that did not meet our criteria for a day hospital, or participants who were not older patients receiving medical care. It should be noted that only those studies which initially appeared to meet the inclusion criteria, but on closer inspection did not, were reported in the Characteristics of excluded studies. For this update, a further 131 studies were excluded but not reported: 45 were not RCTs; in 45 the intervention did not meet our criteria; 29 were review, commentary or discussion papers; five were questionnaires or surveys; and in two the intervention was for a single condition.

Risk of bias in included studies

Ten studies had a low risk of selection bias (method of random sequence generation) of which four studies used a computer generated method (Burch 1999; Crotty 2008; Hedrick 1993; Roderick 2001), four used a random number table (Gladman 1993; Hui 1995; Tucker 1984; Woodford 1962) and two used external Internet/web based services (Masud 2006; Parker 2009). Pitkala 1991 had a high risk of bias as randomisation was by date of birth. In five studies, the method of random sequence generation was unreported  or unclear (Cummings 1985; Eagle 1991; Vetter 1989; Weissert 1980; Young 1992). For review authors' judgements about each risk of bias item presented as percentages across all included studies see Figure 2, and for review authors' judgements about each risk of bias item for each included study see Figure 3.

Figure 2
Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3
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

Nine studies had an adequate method of allocation concealment (Burch 1999; Crotty 2008; Gladman 1993; Hedrick 1993; Hui 1995; Masud 2006; Parker 2009; Vetter 1989; Young 1992). Methods were unclear in 6 studies (Cummings 1985; Eagle 1991; Roderick 2001; Tucker 1984; Weissert 1980; Woodford 1962). Pitkala 1991 presented with a high risk of bias as their method of randomisation was by date of birth which  meant allocation could have been foreseen.

Blinding

Performance bias was a feature of all studies as it was not possible to blind participants due to the nature of the intervention. As a result all studies had a high risk of bias in this domain. Six studies were considered at a low risk for detection bias (blinded outcome assessment; Burch 1999; Crotty 2008; Masud 2006; Roderick 2001; Tucker 1984; Young 1992). Masud 2006 stated that it was not possible to blind researchers to group allocation. However, the review authors considered that the relevant outcome measurements were unlikely to be influenced by a lack of blinding and therefore the risk remained low. The remaining studies were considered to have a high or unclear risk of bias (Cummings 1985; Eagle 1991; Gladman 1993; Hedrick 1993; Hui 1995; Parker 2009; Pitkala 1991; Vetter 1989; Weissert 1980; Woodford 1962).

Incomplete outcome data

Eight studies were considered to be at low risk of bias for attrition (Burch 1999; Crotty 2008; Eagle 1991; Gladman 1993; Hedrick 1993; Masud 2006; Roderick 2001; Young 1992). Three studies were judged to be at high risk of bias. For the Parker 2009 study, losses were similar across the groups but were in excess of 35% by final follow up. For Weissert 1980, 718 participants were excluded for missing data or due to non‐adherence. The numbers lost and reasons lost per group were not reported. For Woodford 1962, approximately a third of participants were lost and whilst numbers were balanced across groups,  the reasons were not reported. The remaining studies were unclear regarding attrition (Cummings 1985; Hui 1995; Pitkala 1991; Tucker 1984; Vetter 1989).

Selective reporting

Two studies were judged to be at low risk for reporting bias (Masud 2006; Parker 2009). Crotty 2008 was considered at high risk as not all the proposed outcomes reported in the study protocol were included in the available publication. For the remaining studies it was unclear whether selective reporting occurred, or pre‐study protocols were unavailable (Burch 1999; Cummings 1985; Eagle 1991; Gladman 1993; Hedrick 1993; Hui 1995; Pitkala 1991; Roderick 2001; Tucker 1984; Vetter 1989; Weissert 1980; Woodford 1962; Young 1992).

Other potential sources of bias

The majority of studies were considered at low risk for other sources of bias (Crotty 2008; Eagle 1991; Gladman 1993; Hedrick 1993; Hui 1995; Masud 2006; Parker 2009; Roderick 2001; Tucker 1984; Vetter 1989; Weissert 1980; Woodford 1962; Young 1992).The Cummings 1985 study was judged as unclear as this was an ‘artificial’ day hospital established for the purpose of the study and the under‐utilisation of the facility may introduced bias. Pitkala 1991 was judged as unclear as 23% of the day hospital group refused the care. Burch 1999 was considered to be at a high risk as 10 of 55 patients transferred from day centre to day hospital.

Further details on how individual studies were scored across the different domains of bias are reported in the risk of bias tables in the Characteristics of included studies.

Effects of interventions

See: Summary of findings for the main comparison Day hospitals compared to alternative care or no care for rehabilitation needs; Summary of findings 2 Day hospitals compared to no comprehensive care for rehabilitation needs; Summary of findings 3 Day hospitals compared to domiciliary care for rehabilitation needs; Summary of findings 4 Day hospitals compared to comprehensive care for elderly persons requiring rehabilitation

The 16 trials included in the review recruited a total of 3689 patients.

Patient outcomes

Nine studies provided final outcome data at 12 months (Burch 1999; Eagle 1991; Gladman 1993; Hedrick 1993; Masud 2006; Parker 2009; Pitkala 1991; Weissert 1980; Woodford 1962), four studies at six months (Crotty 2008; Hui 1995; Roderick 2001; Young 1992), one study at five months (Tucker 1984), one study at three months (Cummings 1985) and one at two months (Vetter 1989).

Death

All 16 trials published data, or provided data on request, for the combined outcome of death at the end of follow up. The pooled OR for all the trials for death at the end of scheduled follow up shows no difference between the day hospital and comparison interventions (odds ratio (OR) 1.05; 95% confidence interval (CI) 0.85 to 1.28; P = 0.66). There was no evidence of a difference when day hospital attendance was compared with comprehensive care (OR 1.26; 95% CI 0.87 to 1.82; P = 0.22), domiciliary care (OR 0.97; 95% CI 0.61 to 1.55; P = 0.89) or no comprehensive care (OR 0.88; 95% CI 0.63 to 1.22; P = 0. 43). There was no significant heterogeneity overall (Chi² = 12.04; df = 14; P = 0.60) or for any of the subgroups (P > 0.05) (Analysis 1.1). Outcome data were missing for a total of 102 day hospital patients and 54 controls (representing 3.2% of patients in the comprehensive care subgroup, 0% in the domiciliary subgroup and 7.8% in the no comprehensive care subgroup). Best and worst case sensitivity analyses include the possibility of significant benefit (P < 0.001) or harm (P < 0.01) from day hospital attendance. Visual inspection of funnel plots did not identify any obvious signs of publication bias.

Death or institutional care

Thirteen trials published data, or provided data on request, for death or institutional care by the end of follow up (Burch 1999; Crotty 2008; Eagle 1991; Gladman 1993; Hedrick 1993; Hui 1995; Masud 2006; Pitkala 1991; Tucker 1984; Weissert 1980; Vetter 1989; Woodford 1962; Young 1992). The pooled OR for all the trials for death or institutional care at the end of scheduled follow up shows no difference between the day hospital and comparison interventions. (OR 0.85; 95% CI 0.63 to 1.14; P = 0.28). There was no significant difference between day hospital patients and those receiving comprehensive services (OR 1.00; 95% CI 0.69 to 1.44; P = 0.99), domiciliary care (OR 1.05; 95% CI 0.57 to 1.92; P = 0.88) or no comprehensive services (OR 0.63; 95% CI 0.40 to 1.00; P = 0.05). There were no significant subgroup differences (P = 0.26). There was significant heterogeneity overall for all studies (Chi² = 25.4, df =11, P = 0.01; I² = 57%; Analysis 1.2). Outcome data were missing for a total of 224 day hospital patients and 110 controls (representing 4.2% of patients in the comprehensive care subgroup, 0% in the domiciliary care subgroup and 19.3% in the no comprehensive care subgroup). Best and worst case sensitivity analyses include the possibility of significant benefit (P < 0.0001) or harm (P < 0.0001) from day hospital attendance. Visual inspection of funnel plots did not identify any obvious signs of publication bias.

Death or deterioration in ADL

Seven trials published data on death or deterioration in ADL (Burch 1999; Gladman 1993; Hui 1995; Pitkala 1991; Vetter 1989; Weissert 1980; Young 1992). The pooled OR for all the trials at the end of scheduled follow up shows no difference between the day hospital and comparison interventions (OR 1.07; 95% CI 0.76 to 1.49; P = 0.70). Only Pitkala 1991 provided data for day hospital compared to comprehensive care and the difference was not significant (OR 1.18; 95% CI 0.63 to 2.18, P = 0.61). There was no difference between day hospital and domiciliary care (OR 1.41; 95% CI 0.82 to 2.42; P = 0.21) or no comprehensive care (OR 0.76; 95% CI 0.56 to 1.05; P = 0.09). There were no significant subgroup differences (P = 0.11) and no significant heterogeneity overall (Chi² = 10.25, df = 6, P = 0.11; I² = 41%; Analysis 1.3). Visual inspection of funnel plots did not identify any obvious signs of publication bias.

Death or poor outcome

Thirteen trials published data on death or poor outcome (Burch 1999; Cummings 1985; Eagle 1991; Gladman 1993; Hedrick 1993; Hui 1995; Pitkala 1991; Roderick 2001; Tucker 1984; Vetter 1989; Weissert 1980; Woodford 1962; Young 1992). Roderick 2001 reported data on poor outcome which they defined as "death, recurrent stroke and a six month Barthel score of < 14"; we determined that this was sufficiently similar to our own definition to include in the results. The pooled OR for all the trials at the end of scheduled follow up shows no significant difference between the day hospital and other interventions (OR 0.92; 95% CI 0.74 to 1.15; P = 0.49). There was no significant difference when day hospital was compared with comprehensive care (OR 1.05; 95% CI 0.79 to 1.40; P = 0.74) or domiciliary care (OR 1.08; 95% CI 0.67 to 1.74; P = 0.75). However, there was a significant difference in favour of the day hospital when compared with no comprehensive care (OR 0.72; 95% CI 0.53 to 0.99; P = 0.04), although subgroup results were not significantly different from each other (P = 0.17). There was no significant heterogeneity overall (Chi² = 17.27, df = 12, P = 0.14; I² = 31%; Analysis 1.4). Outcome data were missing for 55 day hospital patients and 121 controls (representing 4.6% of patients in the comprehensive care subgroup, 0.5% in the domiciliary care subgroup and 10.3% in the no comprehensive care subgroup). Best and worst case sensitivity analyses included the possibility of significant benefit (P < 0.0001) or harm (P < 0.05) from day hospital attendance. Visual inspection of funnel plots did not identify any obvious signs of publication bias.

Deterioration in ADL among survivors

We wished to examine the influence of day hospital attendance on the functional status of survivors. Although most trials described results in terms of ADL scores, seven different measures were used and reported in different ways. We therefore describe results in terms of recorded deterioration in ADL and the raw ADL results. Seven trials provided data on deterioration in ADL among survivors (Burch 1999; Gladman 1993; Hui 1995; Pitkala 1991; Vetter 1989; Weissert 1980; Young 1992). We judged the quality of the evidence for the following outcome as low (summary of findings Table for the main comparison). Overall there was no difference between day hospital and alternative care in ADL scores (OR 1.11; 95% CI 0.68 to 1.80; P = 0.67). However, day hospital attenders appeared less likely to deteriorate than those receiving no comprehensive care (OR 0.61; 95% CI 0.38 to 0.97; P = 0.04). Differences were not significant when comparing day hospitals with comprehensive care (OR 1.21; 95% CI 0.58 to 2.52; P = 0.61) or domiciliary care (OR 1.59; 95% CI 0.87 to 2.90; P = 0.13). There were significant subgroup differences (P = 0.04) and evidence of heterogeneity (Chi² = 11.94, df = 6, P = 0.06; I² = 50%; Analysis 1.5). Visual inspection of funnel plots did not identify any obvious signs of publication bias.

ADL score

Fourteen trials reported a standardised measure of ADL among survivors. However, various measures were used and data were insufficient to allow a statistical summary of the results. Two trials demonstrated significant but small improvements in functional ability with day hospital attendance which was not sustained at six month follow up (Hui 1995; Tucker 1984). One trial (Young 1992) reported an improved functional outcome for the comparison group. The other 11 trials (Burch 1999; Cummings 1985; Eagle 1991; Gladman 1993; Hedrick 1993; Masud 2006; Parker 2009; Pitkala 1991; Roderick 2001; Vetter 1989; Weissert 1980) found no difference in disability scores between the day hospital and comparison groups (Analysis 1.6).

Subjective health status

A number of studies investigated subjective health status. However, various measures were used and we were unable to incorporate data into a meta‐analysis. Three studies investigating day hospital versus comprehensive care found no significant difference between the groups (Cummings 1985; Eagle 1991; Hedrick 1993). In Tucker 1984 there was a significant improvement in mood measured by the Zung index in the day hospital group compared to the comprehensive care group at final follow up (P = 0.01). Pitkala 1991 provided no comparable data. There were no significant differences in any of the studies investigating day hospital versus domiciliary care (Burch 1999; Gladman 1993; Parker 2009; Roderick 2001; Vetter 1989; Young 1992). For day hospital versus no comprehensive care, Hui 1995 found no significant differences. Weissert 1980 and Woodford 1962 did not provide comparable data (Analysis 1.7).

Patient satisfaction

Data on patient satisfaction were only available from one study. Hui 1995 found no significant difference between the day hospital and no comprehensive care (Analysis 1.8).

Carer outcomes

Distress

There were no available or comparable data for day hospital versus comprehensive care or day hospital versus no comprehensive care. Data were available from three studies comparing day hospital with domiciliary care. Crotty 2008 and Gladman 1993 found no significant difference at follow up; Burch 1999 found a significant difference in the mean change between baseline and three months in the Caregiver Strain Index in both groups but no significant difference between groups. (Analysis 1.9).

Resource use

Requiring Institutional care at the end of follow up

Thirteen trials provided information about the number of patients requiring institutional care at the end of follow up (Burch 1999; Crotty 2008; Eagle 1991; Gladman 1993; Hedrick 1993; Hui 1995; Masud 2006; Pitkala 1991; Tucker 1984; Vetter 1989; Weissert 1980; Woodford 1962; Young 1992). In one trial (Weissert 1980) these data were available only for a subgroup of patients (384 patients of 552 recruited to the main study). There was no difference between day hospital and all other services (OR 0.84; 95% CI 0.58 to 1.21; P = 0.35), or for any of the subgroups: day hospital versus comprehensive care (OR 0.91; 95% CI 0.70 to 1.19; P = 0.49), day hospital versus domiciliary care (OR 1.49; 95% CI 0.53 to 4.25; P = 0.45) or day hospital versus no comprehensive care (OR 0.58; 95% CI 0.28 to 1.20; P = 0.14). Overall there was significant heterogeneity (Chi² = 20.03, df = 11, P = 0.04; I² = 45%; Analysis 2.1). On the basis of these data (95% CI 10 to 34) 21 patients (95% CI 12.3 to 70.9) would need to attend day hospital (as opposed to receiving no comprehensive service) to prevent one admission to long term institutional care. Visual inspection of funnel plots did not identify any obvious signs of publication bias.

Hospital bed use

Although hospital use was described in several ways in the trials, it proved possible to obtain a standardised measure for 14 trials of average (mean) hospital bed use per patient recruited (Burch 1999; Cummings 1985; Eagle 1991; Gladman 1993; Hedrick 1993; Hui 1995; Masud 2006; Pitkala 1991; Roderick 2001; Tucker 1984; Vetter 1989; Weissert 1980; Woodford 1962; Young 1992; Analysis 2.2). A measure of variance was not possible for this analysis and therefore confidence limits cannot be reported. The results show a small reduction in bed use by the day hospital patients compared to other treatment across all trials: 13.6 versus 14.6 (Analysis 2.2), with subgroup results as follows:

Day hospital versus comprehensive care ‐ 20.5 versus 21.5.

Day hospital versus domiciliary care ‐ 6.8 versus 9.2.

Day hospital versus no comprehensive care ‐ 9.3 versus 9.4.

Data from Parker 2009 was in a format that did not allow us to incorporate it into the above analysis. However, they reported the mean total length of stay in hospital for patients which was higher in the day hospital group compared to the home rehabilitation group (mean difference 9.3 days; 95% CI 12.5 to 31.1; P > 0.05).

Costs

A number of studies reported a comparison of treatment costs (Analysis 2.3), but methods for reporting data were not consistent and therefore the data could not be incorporated into a meta‐analysis. Seven studies reported that day hospital attendance was more expensive than the comparison treatment (Burch 1999; Gladman 1993; Hedrick 1993; Masud 2006; Tucker 1984; Weissert 1980; Young 1992). Three trials reported that the costs were similar (Hui 1995; Cummings 1985; Roderick 2001). Woodford 1962 reported that day hospital attendance was considerably less expensive than inpatient care (8% of weekly inpatient costs) but made no comparison of other costs incurred specifically by the comparison group. Parker 2009 reported that there was insufficient evidence to support the hypothesis that rehabilitation is less expensive in a home based setting.

For the sub category comparing mean treatment costs between day hospital care and other comprehensive care services, Cummings 1985, Hedrick 1993 and Tucker 1984 reported higher costs for the day hospital. The Cummings 1985 and Hedrick 1993 trials included the cost of nursing home care. There was no information from two trials for this comparison (Eagle 1991; Pitkala 1991). For the sub category comparing treatment costs between day hospital care and domiciliary care the Burch 1999; Gladman 1993 and Young 1992 trials found that day hospital was more expensive. In the Roderick 2001 trial, the day hospital was more expensive for rehabilitations costs but was less so when considering total health and social services costs. There was no information from three trials for this comparison (Crotty 2008; Parker 2009; Vetter 1989). For the sub category comparing treatment costs between day hospital care and no comprehensive care, day hospital was more expensive than no comprehensive care in Hui 1995, Masud 2006 and Weissert 1980. No formal costing data were provided by Woodford 1962.

Assessments of the quality of the body of evidence

Using the GRADE approach we judged the quality of the body of evidence to be low for the patient outcomes of death, death or institutional care, death or deterioration in ADL, death or poor outcome, and deterioration in ADL (summary of findings Table for the main comparison; summary of findings Table 2; summary of findings Table 3; summary of findings Table 4; Appendix 20). For each outcome the body of evidence was from randomised controlled trials but we reduced the quality rating because of a high likelihood of bias in the included studies and imprecision in the effect estimates (wide CIs). We did not reduce the quality rating despite evidence of heterogeneity because this was anticipated due to the diversity of the populations and of the interventions.

Discussion

Summary of main results

The majority of included studies have compared day hospital care with other services. Only four trials employed a comparison group of patients who received neither comprehensive care nor domiciliary rehabilitation (Masud 2006; Woodford 1962; Hui 1995; Weissert 1980). The results from this group were the most favourable to day hospital care, but these four trials are now quite old. Overall the quality of the evidence was low, therefore further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

For the outcome of death, there was no difference between day hospitals and other services, including when day hospitals were compared with any of the subcategories individually. For the combined outcome of death or institutional care there was no significant difference between the day hospital and all other services. For the combined outcome of death or deterioration in ADL, there was no significant difference between the day hospital and other services although there was a trend in favour of the day hospital compared with no comprehensive care. For the combined outcome of death or poor outcome, there was a significant difference in favour of the day hospital when compared with no comprehensive services. For the outcome of deterioration in ADL, there was a significant difference between attending day hospital and no comprehensive care. However, there was no difference between the day hospital and other services.

When considering resource implications among those requiring institutional care, there was no difference at the end of follow up between day hospital and other services. There was a slight reduction in hospital bed use overall for day hospital patients and particularly when day hospitals were compared with domiciliary services; however, whilst a summary statistic was not possible for hospital bed use or cost, individual studies suggest that day hospitals are predominantly as expensive or more expensive than others services.

Overall completeness and applicability of evidence

The day hospital trials included in this systematic review have predominantly employed a pragmatic design and have attempted to address broad questions of overall day hospital effectiveness. This review included 16 studies with 3689 participants, although within the analyses of specific outcomes these numbers were reduced as each study only contributed data for some comparisons. It is unfortunate that data was not available, or in a suitable format, to undertake statistical analyses for the patient outcomes of activities of daily living, subjective health status, patient satisfaction or carer distress, or for the resource outcome of cost. It also proved impossible to determine a summary statistic for disability because, although included as an outcome measure in 11 trials, different measurement instruments were used and variance data were not available. The outcome of death was reported by all studies, however other adverse events and effects were not reported consistently by all studies, thus compromising the overall completeness of findings.

We have based the systematic review on a broad comparison of day hospital care versus alternative services. We wanted to be able to generalise to a range of scenarios and not defined populations. As we anticipated considerable variations in the comparison services, these were identified and categorised prior to data collection and analysis. We have ensured that the treatment schedules described matched our pre‐determined definition of day hospital care. Thus, whilst the study by Weissert and colleagues (Weissert 1980) refers to ‘day care’ services, the intervention provided fitted our definition of day hospital care and was therefore included. A lack of consensus in terminology related to day hospitals has also been noted elsewhere (Petermans 2011).

The applicability of the findings from this review to various regions will depend on current health care provision and populations. As comprehensive care, in one form or another, is likely to be available to many older patients who require rehabilitation, the relevance of some comparisons may be limited in certain countries and populations; specifically the findings from the comparison between day hospitals and no comprehensive care (which was favourable for the day hospital on the combined measure of death or poor outcome, with a trend in favour of the day hospital for the combined measure of death or institutional care). This review found little evidence that day hospitals were better than alternative types of comprehensive service. However, the diversity in the content of alternative services and the populations being served (studies originated from seven different counties) means the external validity of this finding may be compromised. Furthermore, 10 of the studies were at least 20 years old and the types of health service and the populations being served may not reflect current practice or requirements. Services may need to be considered on a case by case basis regarding their applicability against current health provision.

Two main limitations of our review lie in the multinational settings of the studies and in the forty year time span of study publication, during which time health and social care policies inevitably changed. It is disappointing that there have been only four further evaluations of the effectiveness of day hospital care since this review was first published (Forster 1999a). Nonetheless, the data presented here probably represent the best evidence currently available upon which to base a judgment of day hospital effectiveness.

Quality of the evidence

There are limitations due to the lack of statistical power resulting from small, heterogeneous trials. As a result there is a danger of both false‐positive and false‐negative results being generated. Furthermore, the data was from a series of studies performed by researchers operating independently and the studies were therefore not functionally equivalent. This resulted in significant heterogeneity through variations in participants and in the interventions employed in the various day hospitals and comparison treatments. Participants and interventions likely differed in ways that impacted the results. Consequently, we cannot assume a common effect size (Borenstein 2009). A frequently encountered issue with pragmatic rehabilitation trials is that the methods to record subject characteristics, which might influence prognosis and treatment processes, are poorly developed. Considerable detail was recorded in the USA (Hedrick 1993) trial; it is important that any future trials also address this issue. Data were missing for a number of outcomes which could theoretically alter the scale and direction of the results. All of our conclusions must be qualified by this condition.

It is possible that biases have resulted in the overestimation or underestimation of the effects of the intervention. All studies were at a high risk of bias for blinding of participants and personnel; this is typical of such an intervention and effective blinding would be challenging to implement in view of ethical considerations which require participants to have prior knowledge of potential interventions. However, a lack of blinding of outcome assessment was also problematic and only half the studies were considered to have a low risk in this domain. Contamination can result in the underestimation of treatment effects and whilst not all studies were judged to be of high risk, contamination and deviation from the protocol (switching between interventions) did feature in some studies comparing day hospitals with other comprehensive services and domiciliary care. In addition, the risk of selective reporting was often unclear. Only the three studies which were reported in the last eight years had available protocols against which their reports could be assessed to determine if all outcomes had been reported. Finally, half of the studies were considered to have a low risk of bias for the domain of incomplete outcome data. This should be considered positive, taking into account the nature of the intervention and participants, post intervention follow up periods, and the age of the studies where attrition was less frequently reported.

Potential biases in the review process

The search strategy was extensive, taking into account multiple databases and sources, and this is reflected in the large number of titles identified. Two authors extracted all data, discussions were undertaken to achieve consensus, and a third person was utilised where disagreements remained. As a result, we are confident in the quality of the review. It was unfortunate, however, that we could not quantitatively combine data for a number of outcomes, due to the diversity of measures used (for example, ADL, subjective health status, patient satisfaction, carer distress, bed use and resource use).

Publication bias (Egger 1997) remains a possibility in any review process. However, our search strategy was extensive and included contacting the authors of papers relating to day hospital care around the world.

We were fortunate that many of the authors of the published papers or abstracts were able to provide additional information which has not been published previously.

Agreements and disagreements with other studies or reviews

Another systematic review drew similar conclusions to our review (Petermans 2011). They found that geriatric day hospitals were better for patients undergoing assessment and intervention than no comprehensive care. However, they found little benefit when compared to treatment in a geriatric ward or other geriatric services. They also found few studies reporting on the outcome of patient satisfaction. The authors did not include meta‐analysis and they did not report on any cost‐benefit outcomes; their conclusions were drawn from various sources including RCTs and cohort studies. Another systematic review, with meta‐analysis, found that comprehensive geriatric assessments linking geriatric evaluation with long term management are effective in improving survival and function in older people (Stuck 1993).

Individual results from the included studies suggest that day hospitals are probably as expensive or more expensive than other comprehensive or domiciliary services. Several costing studies have drawn attention to the expense of day hospital services (Eagle 1987; Gerard 1988; Gladman 1994; MacFarlane 1979; Young 1993).

original image
Figuras y tablas -
Figure 1
Ir a la figura de la revisiónAbrir en una pestaña nueva

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.
Figuras y tablas -
Figure 2

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.
Figuras y tablas -
Figure 3

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

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 1 Death by the end of follow up.
Figuras y tablas -
Analysis 1.1

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 1 Death by the end of follow up.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 2 Death or institutional care by the end of follow up.
Figuras y tablas -
Analysis 1.2

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 2 Death or institutional care by the end of follow up.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 3 Death or deterioration in activities of daily living (ADL).
Figuras y tablas -
Analysis 1.3

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 3 Death or deterioration in activities of daily living (ADL).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 4 Death or poor outcome (institutional care, disability or deterioration).
Figuras y tablas -
Analysis 1.4

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 4 Death or poor outcome (institutional care, disability or deterioration).

Ir a la figura de la revisiónAbrir en una pestaña nueva

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 5 Deterioration in activities of daily living (ADL) in survivors.
Figuras y tablas -
Analysis 1.5

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 5 Deterioration in activities of daily living (ADL) in survivors.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Study

Day hospital vs Comprehensive elderly care

Cummings 1985

No significant difference in the Kenny ADL score ‐
Day hospital: 27.1
Control: 26

Eagle 1991

No significant difference in the Geriatric Quality of Life Questionnaire ADL score ‐
Day hospital: 4.01
Control: 4.43

Hedrick 1993

No significant in the physical dimension of the Sickness Impact Profile (NB high score indicates increased disability) ‐
Day hospital: 29.0 (SD 18.6)
Control: 32.1 (18.8)

Pitkala 1991

No data in this form

Tucker 1984

No significant difference in the mean change in Northwick Park ADL score from baseline (NB high score indicates increased disability)
Day hospital: 0.63
Control: ‐0.64

Day hospital vs Domiciliary care

Burch 1999

No significant difference in Barthel index ‐
Day hospital: 14.5 (SD 4.9)
Control: 15.7 (sd 4.5)

Gladman 1993

No significant difference in median Barthel index ‐
Day hospital: 17
Control: 16

Parker 2009

No significant difference in mean (SD) Nottingham Extended Activities of Daily Living Scale

Day hospital: 31.6 (15.4)

Control: 28.1 (17.5)

Roderick 2001

Significant improvement in Barthel score in both groups.
No significant difference between groups.

Vetter 1989

No significant difference in mean Barthel index ‐
Day hospital: 13.2 (SD 3.8)
Control: 12.3 (SD 4.1)

Young 1992

Significantly (P=0.01) lower Barthel ADL score ‐
Day hospital: 15 (IQR 12‐18)
Control: 17 (IQR 15‐19)

Day hospital vs No comprehensive elderly care

Hui 1995

No significant difference in mean Barthel index ‐
Day hospital: 17.1 (SD 3.6)
Control: 15.6 (SD 5.6)

Masud 2006

No significant difference in median (IQR) Barthel index scores
Day hospital: 19 (17‐20)
Control: 19 (17‐20)

No significant difference in median (IQR) Nottingham Extended Activities of Daily Living (NEADL) scores

Day hospital: 53 (43‐62)
Control: 56 (43.8‐61)

Weissert 1980

No comparable data

Woodford 1962

No comparable data

Figuras y tablas -
Analysis 1.6

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 6 Activities of daily living (ADL) scores.

Ir a la figura de la revisión

Study

Day hospital vs Comprehensive elderly care

Cummings 1985

No significant difference in the modified Dupay assessment ‐
Day hospital: 30.4
Control: 25.9

Eagle 1991

No significant difference in the General Health Questionnaire ‐
Day hospital: 3.85
Control: 4.33

Hedrick 1993

No significant difference in the Sickness Impact Profile ‐
Day hospital: 34.1
Control: 34.5

Pitkala 1991

No comparable data

Tucker 1984

There was a statistically significant improvement in mood measured by the Zung index in the Day hospital group compared to the Comprehensive elderly care group at final follow up (P = 0.01)

Day hospital vs Domiciliary care

Burch 1999

No significant difference in the change in the Philidelphia Geriatric Morale scale during follow up ‐
Day hospital: 1.80
Control: 0.92

Gladman 1993

No significant difference in the number of patients with distress (Nottingham Health Profile >30) ‐
Day hospital: 17 (33%)
Control: 25 (48%)

Parker 2009

There was no significant difference in the EuroQol index (difference in means 0.147, P = 0.141) or the EuroQol VAS (difference in means 0. 6.315, P = 0.187)

Roderick 2001

No between groups significant difference for SF‐36 physical or mental health scales.
The Philadelphia Geriatric Morale Scale scores fell in both groups (indicating lower morale) but less so in the domiciliary group.

Vetter 1989

No significant difference in the Sickness Impact Profile

Young 1992

No significant difference in the number of patients with distress (Nottingham Health Profile >30) ‐
Day hospital: 19 (41%)
Control: 20 (39%)

Day hospital vs No comprehensive elderly care

Hui 1995

No significant difference in the Geriatric Depression Scale

Weissert 1980

No comparable data

Woodford 1962

No comparable data

Figuras y tablas -
Analysis 1.7

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 7 Subjective health status.

Ir a la figura de la revisión

Study

Day hospital vs Comprehensive elderly care

Cummings 1985

No available data

Eagle 1991

No available data

Hedrick 1993

No comparable data

Pitkala 1991

No comparable data

Tucker 1984

No available data

Day hospital vs Domiciliary care

Burch 1999

No available data

Gladman 1993

No available data

Roderick 2001

No available data

Vetter 1989

No available data

Young 1992

No available data

Day hospital vs No comprehensive elderly care

Hui 1995

No significant difference in the level of satisfaction

Weissert 1980

No available data

Woodford 1962

No available data

Figuras y tablas -
Analysis 1.8

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 8 Patient satisfaction.

Ir a la figura de la revisión

Study

Day hospital vs Comprehensive elderly care

Cummings 1985

No comparable data

Eagle 1991

No available data

Hedrick 1993

No comparable data

Pitkala 1991

No comparable data

Tucker 1984

No data available

Day hospital vs Domiciliary care

Burch 1999

There was a significant difference in the mean change between baseline and 3 months in the Caregiver Strain Index in both groups
Day hospital: ‐1.45 (95% CI ‐0.41, ‐2.49)
Control: ‐1.59 (‐0.62, ‐2.56)

There was no significant difference between between groups

Crotty 2008

No significant difference at 3 months in the Carer Strain Index
Day hospital: 4.9 (3.9)
Control: 4.3 (3.1)

Gladman 1993

No significant difference at 6 months in the median Life Satisfaction Index ‐
Day hospital: 18 (IQR 11‐22)
Control: 15 (IQR 10‐19)

Roderick 2001

No available data

Vetter 1989

No data available

Young 1992

Proportion of carers showing distress (General Health Questionnaire 28 score > 5) ‐
Day hospital: 14 of 33 scored >5 (42%)
Control: 8 of 30 scored >5 (27%)

Differences were not significant

Day hospital vs No comprehensive elderly care

Hui 1995

No available data

Weissert 1980

No available data

Woodford 1962

No available data

Figuras y tablas -
Analysis 1.9

Comparison 1 Day Hospital vs Alternative Care ‐ patient outcomes, Outcome 9 Carer Distress.

Ir a la figura de la revisión

Comparison 2 Day Hospital vs Alternative Care ‐ resource outcomes, Outcome 1 Requiring institutional care at the end of follow up.
Figuras y tablas -
Analysis 2.1

Comparison 2 Day Hospital vs Alternative Care ‐ resource outcomes, Outcome 1 Requiring institutional care at the end of follow up.

Ir a la figura de la revisiónAbrir en una pestaña nueva

Study

Day hospital vs Comprehensive elderly care

Cummings 1985

Day hospital: 206 bed days for 48 patients (4.3 days per patient)
Control: 274 bed days for 48 patients (5.7 days per patient)

Eagle 1991

Day hospital: 1388 bed days for 55 patients (25.2 days per patient)
Control: 1351 bed days for 58 patients (23.3 days per patient)

Hedrick 1993

Day hospital: 8020 bed days for 411 patients (19.5 days per patient)
Control: 8067 bed days for 415 patients (19.4 days per patient)

Pitkala 1991

Day hospital: 3538 bed days for 88 patients (40.2 days per patient)
Control: 3713 bed days for 86 patients (43.2 days per patient)

Tucker 1984

Day hospital: 472 bed days for 62 patients (7.6 days per patient)
Control: 800 bed days for 58 patients (13.8 bed days per patient)

Day hospital vs Domiciliary care

Burch 1999

Day hospital: 923 bed days for 50 patients (18.5 per patient)
Control: 1438 bed days for 55 patients (26.1 per patient)

Gladman 1993

Day hospital: 436 bed days for 76 patients (5.7 days per patient)
Control: 766 bed days for 79 patients (9.7 days per patient)

Roderick 2001

Day hospital: 296 bed days for 74 patients ( 4 days per patient)
Control:203 bed days for 66 patients (3 bed days per patient)

Vetter 1989

No bed days used in either group

Young 1992

Day hospital: 311 bed days for 61 patients (5.1 days per patient)
Control: 278 bed days for 63 patients (4.4 days per patient)

Day hospital vs No comprehensive elderly care

Hui 1995

Day hospital: 81 bed days for 59 patients (1.4 days per patient)
Control: 165 bed days for 61 patients (2.7 days per patient)

Masud 2006

Day hospital: 552 bed days for 172 patients (3.2 days per patient)
Control: 529 bed days for 171 patients (3.1 days per patient)

Weissert 1980

Day hospital: 3443 bed days for 313 patients (11.0 days per patient)
Control: 2868 bed days for 239 patients (12.0 days per patient)

Woodford 1962

Day hospital: 2534 bed days for 168 patients (15.1 days per patient)
Control: 2375 bed days for 163 patients (14.6 days per patient)

Figuras y tablas -
Analysis 2.2

Comparison 2 Day Hospital vs Alternative Care ‐ resource outcomes, Outcome 2 Hospital bed use during follow up.

Ir a la figura de la revisión

Study

Day hospital vs Comprehensive elderly care

Cummings 1985

Average treatment cost per patient to 3 months post discharge ‐
Day hospital: $ 16,966
Control: $ 14,082

Eagle 1991

No cost data

Hedrick 1993

Mean total costs over 12 months ‐
Day hospital: $ 28,709
Control: $ 26, 204

Pitkala 1991

No costing data

Tucker 1984

Average cost for 5 months care ‐
Day hospital: NZ$ 3052
Control: NZ$ 2083

Day hospital vs Domiciliary care

Burch 1999

Total annual cost per attendance ‐
Day hospital: £ 77.39
Control: £ 59.46

Gladman 1993

Mean total health service cost per patient ‐
Day hospital: £ 456.90
Control: £ 362.60

Parker 2009

(entries) Mean total cost at 213 days: Day hospital (21) £10,102; Home based rehab (25) £14,330 (note SDs not reported)

(entries) Mean total cost at 395 days: Day hospital (13) £23,812; Home based rehab (23) £26,105 (note SDs not reported)

(entries) Median total cost at 213 days: Day hospital (21) £5958; Home based rehab (25) £7679 (note IQRs not reported)

(entries) Median total cost at 395 days: Day hospital (13) £9842; Home based rehab (23) £18,432 (note IQRs not reported)

Roderick 2001

Median costs per patient: Rehabilitation costs: Day hospital:£1090 (IQR 513‐1475),
Control group: 933 (IQR 339‐2010).
Median Total health and social service costs: Day hospital 1568 (IQR 982‐3130), Control £2208 (IQR 694‐3849)

Vetter 1989

No cost data

Young 1992

Mean total costs for 8 weeks treatment ‐
Day hospital: £ 620 (IQR 555‐730)
Control: £385 (IQR 240‐510)

Day hospital vs No comprehensive elderly care

Hui 1995

Mean cost of treatment to the health service over 6 months ‐
Day hospital: $ 58,168 (SEM 25,898)
Control: $ 51, 809 (SEM 30,480)

Masud 2006

The mean falls programme cost was £349 per person. With additional health‐care costs, this resulted in a mean incremental cost of £578 for the intervention

Health service resource use and costs. Total cost per participant mean (SD):

Day hospital £2,238 (4957)

Control £1,659 (5100)

Weissert 1980

Average cost for 1 year ‐
Day hospital: $ 5813
Control: $ 3815

Woodford 1962

No formal costing

Figuras y tablas -
Analysis 2.3

Comparison 2 Day Hospital vs Alternative Care ‐ resource outcomes, Outcome 3 Resource use.

Ir a la figura de la revisión
Summary of findings for the main comparison. Day hospitals compared to alternative care or no care for rehabilitation needs

Day hospitals compared to alternative or no care for rehabilitation

Patient or population: patients with rehabilitation needs
Intervention: day hospitals
Comparison: alternative care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Alternative or no care

Day hospitals

Death by the end of follow up
Follow‐up: median 12 months

Study population

OR 1.05
(0.85 to 1.28)

3533
(16 studies)

⊕⊕⊝⊝
low1,2,3

127 per 1000

132 per 1000
(110 to 157)

Moderate

66 per 1000

69 per 1000
(57 to 83)

Death or institutional care by the end of follow up
Follow‐up: median 12 months

Study population

OR 0.85
(0.63 to 1.14)

3030
(13 studies)

⊕⊕⊝⊝
low1,2,3

303 per 1000

270 per 1000
(215 to 331)

Moderate

221 per 1000

194 per 1000
(152 to 244)

Death or deterioration in activities of daily living (ADL)
Follow‐up: median 12 months

Study population

OR 1.07
(0.76 to 1.49)

1268
(7 studies)

⊕⊕⊝⊝
low1,2,3

407 per 1000

423 per 1000
(343 to 506)

Moderate

430 per 1000

447 per 1000
(364 to 529)

Death or poor outcome (institutional care, disability or deterioration)
Follow‐up: median 12 months

Study population

OR 0.92
(0.74 to 1.15)

2831
(13 studies)

⊕⊕⊝⊝
low1,2,3

365 per 1000

346 per 1000
(299 to 398)

Moderate

241 per 1000

226 per 1000
(190 to 267)

Deterioration in ADL in survivors
Various ADL measures

Study population

OR 1.11
(0.68 to 1.8)

905
(7 studies)

⊕⊕⊝⊝
low1,2,3

251 per 1000

271 per 1000
(185 to 376)

Moderate

233 per 1000

252 per 1000
(171 to 354)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; OR: Odds ratio; ADL: activities of daily living

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Limitations for at least one risk of bias criterion, or some limitations for multiple criteria, sufficient to lower confidence in the estimate of effect
2 Whilst there was evidence of heterogeneity, this was anticipated due to the diversity of the populations and of the interventions
3 Wide CIs

Figuras y tablas -
Summary of findings for the main comparison. Day hospitals compared to alternative care or no care for rehabilitation needs
Ir a la tabla de la revisión
Summary of findings 2. Day hospitals compared to no comprehensive care for rehabilitation needs

Day hospitals compared to no comprehensive care for rehabilitation needs

Patient or population: patients with rehabilitation needs
Intervention: day hospitals
Comparison: no comprehensive care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

No comprehensive care

Day hospitals

Death by the end of follow up
Follow‐up: median 12 months

Study population

OR 0.88
(0.63 to 1.22)

1345
(4 studies)

⊕⊕⊝⊝
low1,2,3

128 per 1000

114 per 1000
(85 to 152)

Moderate

131 per 1000

117 per 1000
(87 to 155)

Death or institutional care by the end of follow up
Follow‐up: median 12 months

Study population

OR 0.63
(0.4 to 1)

1177
(4 studies)

⊕⊕⊝⊝
low1,2,3

248 per 1000

172 per 1000
(117 to 248)

Moderate

307 per 1000

218 per 1000
(151 to 307)

Death or deterioration in ADL
Follow‐up: median 9 months

Study population

OR 0.76
(0.56 to 1.05)

651
(2 studies)

⊕⊕⊝⊝
low1,2,3

436 per 1000

370 per 1000
(302 to 448)

Moderate

446 per 1000

380 per 1000
(311 to 458)

Death or poor outcome (institutional care, disability or deterioration)
Follow‐up: median 12 months

Study population

OR 0.72
(0.53 to 0.99)

982
(3 studies)

⊕⊕⊝⊝
low1,2,3

347 per 1000

277 per 1000
(220 to 345)

Moderate

400 per 1000

324 per 1000
(261 to 398)

Deterioration in ADL in survivors
Follow‐up: median 9 months

Study population

OR 0.61
(0.38 to 0.97)

407
(2 studies)

⊕⊕⊝⊝
low1,2,3

277 per 1000

189 per 1000
(127 to 271)

Moderate

227 per 1000

152 per 1000
(100 to 222)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; OR: Odds ratio; ADL: activities of daily living

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Limitations for at least one risk of bias criterion or some limitations for multiple criteria, sufficient to lower confidence in the estimate of effect
2 Whilst there was evidence of heterogeneity, this was anticipated due to the diversity of the population and of the study design
3 Wide CIs

Figuras y tablas -
Summary of findings 2. Day hospitals compared to no comprehensive care for rehabilitation needs
Ir a la tabla de la revisión
Summary of findings 3. Day hospitals compared to domiciliary care for rehabilitation needs

Day hospitals compared to domiciliary care for rehabilitation

Patient or population: patients with rehabilitation needs
Intervention: day hospitals
Comparison: domiciliary care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Domiciliary care

Day hospitals

Death by the end of follow up
Follow‐up: median 6

Study population

OR 0.97
(0.61 to 1.55)

901
(7 studies)

⊕⊕⊝⊝
low1,2,3

101 per 1000

98 per 1000
(64 to 148)

Moderate

64 per 1000

62 per 1000
(40 to 96)

Death or institutional care by the end of follow up
Follow‐up: median 6 months

Study population

OR 1.05
(0.57 to 1.92)

672
(5 studies)

⊕⊕⊝⊝
low1,2,3

187 per 1000

194 per 1000
(116 to 306)

Moderate

69 per 1000

72 per 1000
(41 to 125)

Death or deterioration in ADL
Follow‐up: median 9 months

Study population

OR 1.41
(0.82 to 2.42)

443
(4 studies)

⊕⊕⊝⊝
low1,2,3

392 per 1000

476 per 1000
(346 to 609)

Moderate

334 per 1000

414 per 1000
(291 to 548)

Death or poor outcome (institutional care, disability or deterioration)
Follow‐up: median 6 months

Study population

OR 1.08
(0.67 to 1.74)

581
(5 studies)

⊕⊕⊝⊝
low1,2,3

297 per 1000

313 per 1000
(221 to 424)

Moderate

364 per 1000

382 per 1000
(277 to 499)

Deterioration in ADL in survivors
Follow‐up: median 9 months

Study population

OR 1.59
(0.87 to 2.9)

349
(4 studies)

⊕⊕⊝⊝
low1,2,3

225 per 1000

315 per 1000
(201 to 457)

Moderate

188 per 1000

269 per 1000
(168 to 402)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; OR: Odds ratio; ADL: activities of daily living

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Limitations for at least one risk of bias criterion or some limitations for multiple criteria, sufficient to lower confidence in the estimate of effect
2 Whilst there was evidence of heterogeneity, this was anticipated due to the diversity of the population and the interventions
3 Wide CIs

Figuras y tablas -
Summary of findings 3. Day hospitals compared to domiciliary care for rehabilitation needs
Ir a la tabla de la revisión
Summary of findings 4. Day hospitals compared to comprehensive care for elderly persons requiring rehabilitation

Day hospitals compared to comprehensive care for older people requiring rehabilitation

Patient or population: older people requiring rehabilitation
Intervention: day hospitals
Comparison: comprehensive care

Outcomes

Illustrative comparative risks* (95% CI)

Relative effect
(95% CI)

No of Participants
(studies)

Quality of the evidence
(GRADE)

Comments

Assumed risk

Corresponding risk

Comprehensive care

Day hospitals

Death by the end of follow up
Follow‐up: median 12 months

Study population

OR 1.26
(0.87 to 1.82)

1287
(5 studies)

⊕⊕⊝⊝
low1,2,3

144 per 1000

175 per 1000
(128 to 234)

Moderate

69 per 1000

85 per 1000
(61 to 119)

Death or institutional care by the end of follow up
Follow‐up: median 12 months

Study population

OR 1
(0.69 to 1.44)

1181
(4 studies)

⊕⊕⊝⊝
low1,2,3

426 per 1000

426 per 1000
(339 to 517)

Moderate

231 per 1000

231 per 1000
(172 to 302)

Death or deterioration in ADL
Follow‐up: median 12 months

Study population

OR 1.18
(0.63 to 2.18)

174
(1 study)

⊕⊕⊝⊝
low1,3

349 per 1000

387 per 1000
(252 to 539)

Moderate

349 per 1000

387 per 1000
(252 to 539)

Death or poor outcome (institutional care, disability or deterioration)
Follow‐up: median 12 months

Study population

OR 1.05
(0.79 to 1.4)

1268
(5 studies)

⊕⊕⊝⊝
low1,2,3

410 per 1000

422 per 1000
(355 to 493)

Moderate

221 per 1000

230 per 1000
(183 to 284)

Deterioration in ADL in survivors
Follow‐up: median 12 months

Study population

OR 1.21
(0.58 to 2.52)

149
(1 study)

⊕⊕⊝⊝
low1,3

243 per 1000

280 per 1000
(157 to 448)

Moderate

243 per 1000

280 per 1000
(157 to 447)

*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (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; OR: Odds ratio; ADL: activities of daily living

GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

1 Limitations for at least one risk of bias criterion or some limitations for multiple criteria, sufficient to lower confidence in the estimate of effect
2 Whilst there was evidence of heterogeneity, this was anticipated due to the diversity of the population and of the study design
3 Wide CIs

Figuras y tablas -
Summary of findings 4. Day hospitals compared to comprehensive care for elderly persons requiring rehabilitation
Ir a la tabla de la revisión
Comparison 1. Day Hospital vs Alternative Care ‐ patient outcomes

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Death by the end of follow up Show forest plot

16

3533

Odds Ratio (M‐H, Random, 95% CI)

1.05 [0.85, 1.28]

1.1 Day Hospital vs Comprehensive elderly care

5

1287

Odds Ratio (M‐H, Random, 95% CI)

1.26 [0.87, 1.82]

1.2 Day hospital vs Domiciliary care

7

901

Odds Ratio (M‐H, Random, 95% CI)

0.97 [0.61, 1.55]

1.3 Day hospital vs No comprehensive elderly care

4

1345

Odds Ratio (M‐H, Random, 95% CI)

0.88 [0.63, 1.22]

2 Death or institutional care by the end of follow up Show forest plot

13

3030

Odds Ratio (M‐H, Random, 95% CI)

0.85 [0.63, 1.14]

2.1 Day hospital vs Comprehensive elderly care

4

1181

Odds Ratio (M‐H, Random, 95% CI)

1.00 [0.69, 1.44]

2.2 Day hospital vs Domiciliary care

5

672

Odds Ratio (M‐H, Random, 95% CI)

1.05 [0.57, 1.92]

2.3 Day hospital vs No comprehensive elderly care

4

1177

Odds Ratio (M‐H, Random, 95% CI)

0.63 [0.40, 1.00]

3 Death or deterioration in activities of daily living (ADL) Show forest plot

7

1268

Odds Ratio (M‐H, Random, 95% CI)

1.07 [0.76, 1.49]

3.1 Day hospital vs Comprehensive elderly care

1

174

Odds Ratio (M‐H, Random, 95% CI)

1.18 [0.63, 2.18]

3.2 Day hospital vs Domiciliary care

4

443

Odds Ratio (M‐H, Random, 95% CI)

1.41 [0.82, 2.42]

3.3 Day hospital vs No comprehensive elderly care

2

651

Odds Ratio (M‐H, Random, 95% CI)

0.76 [0.56, 1.05]

4 Death or poor outcome (institutional care, disability or deterioration) Show forest plot

13

2831

Odds Ratio (M‐H, Random, 95% CI)

0.92 [0.74, 1.15]

4.1 Day hospital vs Comprehensive elderly care

5

1268

Odds Ratio (M‐H, Random, 95% CI)

1.05 [0.79, 1.40]

4.2 Day hospital vs Domiciliary care

5

581

Odds Ratio (M‐H, Random, 95% CI)

1.08 [0.67, 1.74]

4.3 Day hospital vs No comprehensive elderly care

3

982

Odds Ratio (M‐H, Random, 95% CI)

0.72 [0.53, 0.99]

5 Deterioration in activities of daily living (ADL) in survivors Show forest plot

7

905

Odds Ratio (M‐H, Random, 95% CI)

1.11 [0.68, 1.80]

5.1 Day hospital vs Comprehensive elderly care

1

149

Odds Ratio (M‐H, Random, 95% CI)

1.21 [0.58, 2.52]

5.2 Day hospital vs Domiciliary care

4

349

Odds Ratio (M‐H, Random, 95% CI)

1.59 [0.87, 2.90]

5.3 Day hospital vs No comprehensive elderly care

2

407

Odds Ratio (M‐H, Random, 95% CI)

0.61 [0.38, 0.97]

6 Activities of daily living (ADL) scores Show forest plot

Other data

No numeric data

6.1 Day hospital vs Comprehensive elderly care

Other data

No numeric data

6.2 Day hospital vs Domiciliary care

Other data

No numeric data

6.3 Day hospital vs No comprehensive elderly care

Other data

No numeric data

7 Subjective health status Show forest plot

Other data

No numeric data

7.1 Day hospital vs Comprehensive elderly care

Other data

No numeric data

7.2 Day hospital vs Domiciliary care

Other data

No numeric data

7.3 Day hospital vs No comprehensive elderly care

Other data

No numeric data

8 Patient satisfaction Show forest plot

Other data

No numeric data

8.1 Day hospital vs Comprehensive elderly care

Other data

No numeric data

8.2 Day hospital vs Domiciliary care

Other data

No numeric data

8.3 Day hospital vs No comprehensive elderly care

Other data

No numeric data

9 Carer Distress Show forest plot

Other data

No numeric data

9.1 Day hospital vs Comprehensive elderly care

Other data

No numeric data

9.2 Day hospital vs Domiciliary care

Other data

No numeric data

9.3 Day hospital vs No comprehensive elderly care

Other data

No numeric data
Figuras y tablas -
Comparison 1. Day Hospital vs Alternative Care ‐ patient outcomes
Ir a la tabla de la revisión
Comparison 2. Day Hospital vs Alternative Care ‐ resource outcomes

Outcome or subgroup title

No. of studies

No. of participants

Statistical method

Effect size

1 Requiring institutional care at the end of follow up Show forest plot

13

3003

Odds Ratio (M‐H, Random, 95% CI)

0.84 [0.58, 1.21]

1.1 Day hospital vs Comprehensive elderly care

4

1181

Odds Ratio (M‐H, Random, 95% CI)

0.91 [0.70, 1.19]

1.2 Day hospital vs Domiciliary care

5

672

Odds Ratio (M‐H, Random, 95% CI)

1.49 [0.53, 4.25]

1.3 Day hospital vs No comprehensive elderly care

4

1150

Odds Ratio (M‐H, Random, 95% CI)

0.58 [0.28, 1.20]

2 Hospital bed use during follow up Show forest plot

Other data

No numeric data

2.1 Day hospital vs Comprehensive elderly care

Other data

No numeric data

2.2 Day hospital vs Domiciliary care

Other data

No numeric data

2.3 Day hospital vs No comprehensive elderly care

Other data

No numeric data

3 Resource use Show forest plot

Other data

No numeric data

3.1 Day hospital vs Comprehensive elderly care

Other data

No numeric data

3.2 Day hospital vs Domiciliary care

Other data

No numeric data

3.3 Day hospital vs No comprehensive elderly care

Other data

No numeric data
Figuras y tablas -
Comparison 2. Day Hospital vs Alternative Care ‐ resource outcomes
Ir a la tabla de la revisión