Health effects of housing improvement: systematic review of intervention studies

Search strategy

We searched the following databases: ASSIA (Applied Social Science Index and Abstracts, 1987-2000), CAB Health (1973-2000), DHSS-DATA (1983-2000), Embase (1974-2000), HealthSTAR (1975-2000), Medline (1966-2000), PAIS (Public Affairs Information Service, 1976-2000), PsycINFO (1887-2000), SIGLE (System for Information on Grey Literature in Europe, 1980-2000), Social SciSearch (1972-2000), Sociological Abstracts (1963-2000), Social Science Citation Index (1981-2000), Urbadisc, Cochrane Controlled Trials Database 2000 Issue 2, IBSS (International Bibliography of the Social Sciences), SPECTR (Social, Psychological, Educational, and Criminological Controlled Trials Register, searched December 2000), and the world wide web. Full details of the search strategy are available from the authors.

We hand searched the bibliographies of all reports, papers, and text books that we reviewed. We also requested information on unpublished and ongoing studies from subscribers to the Housing Studies Association newsletter and email list and the Health Action Zone discussion group. HT contacted health authority housing departments, academic departments in the United Kingdom, local authorities, and housing associations. We also asked delegates at an international housing conference for details of suitable studies, either completed or ongoing.

Selection

We sought primary studies in any language that used experimental or quasi-experimental approaches to examine the effects of housing improvements. These included randomised controlled trials and observational studies that used prospective or retrospective measures of health. Our outcome measures were based on a social model of health and included socioeconomic changes and illness based outcomes. Housing interventions were defined as rehousing, and all physical changes to housing were defined as infrastructure—for example installation of heating, insulation, double glazing, and general refurbishment.

We excluded cross sectional studies that did not investigate the effects of housing improvement before and after the intervention. We did not include interventions to improve the indoor environment through furniture or indoor equipment (such as vacuuming, mattresses, and air purifiers) unless the evaluation measured changes in residents' health and the measures were part of a package of interventions that included improvements to the house itself. Environmental studies of the adverse effects of lead, urea formaldehyde foam, poor air quality, allergens, or radon were not included. At least two reviewers independently screened all abstracts identified by the searches.

Assessment of validity assessment and data abstraction

Three reviewers critically appraised the included studies according to the criteria (box). Studies graded as C were not considered in the final assessment of the evidence. When reviewers' conclusions differed, the study was reviewed jointly by three reviewers. Data were abstracted by one reviewer (HT) and checked by a second reviewer (MP). When data on the group of interest were not given in the publication, we calculated them if possible—for example, new P values were calculated using the relevant sample sizes.

Medical priority rehousing

All three studies of rehousing on the basis of medical need found improvements in self reported physical and mental health. However, the only prospective study was small,⁷,⁸ and no study controlled for the effects of possible confounding variables. One study examined the effects on use of health services and found no clear pattern.⁹

Rehousing, refurbishment and relocation or community regeneration

Two prospective controlled studies reported beneficial effects of rehousing or refurbishment on health outcomes, including improvements in mental health.^11–14 Only one study had controlled for confounding. This study showed an initial increase in illness episodes in the intervention group at 9 months. At 18 months, however, the intervention group showed a larger reduction in illness episodes compared with the control group, although the absolute difference was small (29 episodes/1000 people) and the rate of follow up was not stated.¹¹,¹² The other prospective controlled study reported improvements in mental and physical health, but the study was small and the comparability of the control group is unclear.¹³,¹⁴

Energy efficiency measures

Although the four studies that we identified all found that energy efficiency measures improve respiratory and other symptoms, only one study adjusted for potential confounding variables.²⁷ High rates of attrition in this and most other studies limit the generalisability of these findings.

Use of health services and social effects

Some studies assessed the effects of improving housing on use of health services; decreased visits to the general practitioner, reduced likelihood of inpatient and outpatient use of health services, and reduced prescribing of hypnotic and respiratory drugs were reported. None of the evidence for these effects came from methodologically robust prospective controlled studies.

Broader social impacts of housing improvement were reported in some studies, including improvements in social outcomes, such as perceptions of safety and social and community participation.¹⁶,¹⁷,¹⁹,²³ One small study reported a small increase in social support after the intervention.¹⁹ Two studies that examined the effects of housing improvement in the context of area regeneration, reported that residents' concerns about local crime were reduced.¹⁷,²³ Another small study reported that fewer days were lost from school because of asthma after heating improvements.²⁵

Difficulties in studying housing and health

Reasons for the lack of studies into the effects of housing on health may include methodological difficulties and political obstacles. There are many methodological difficulties inherent in assessing the health effects of housing interventions. Poor housing conditions often exist alongside other forms of deprivation, and housing interventions rarely occur in isolation. This may affect the sociodemographics of an area and make before and after comparisons problematic.²⁰ Moreover, response and follow up rates in studies of deprived areas are often low.

More generally, the experimental approach to housing research has been criticised for being reductionist and ignoring the multifactorial nature of causality in housing, deprivation, and health.³¹ Nevertheless, broad generalisations about the link between deprivation and ill health can have only a limited role in informing specific policy decisions.³² Evidence of the effectiveness and cost effectiveness of specific interventions is therefore particularly important. Assembling such evidence requires a holistic approach, combining quantitative and qualitative methods and taking into account a range of possible influences and mechanisms.³³ Although there is a long tradition of this type of evaluation in the United States, it has rarely been attempted in the United Kingdom.³⁴ In 1989, Smith recommended that housing research should embrace a public health approach and include more multidisciplinary studies.³⁵ This is now starting to happen. Of the ongoing studies we identified, eight are collaborations between housing and health agencies and academics.

Political obstacles to conducting housing research may also exist. Traditionally, policy makers in the United Kingdom have not had access to much evidence on the health effects of social interventions. This lack of evidence and the methodological limitations of existing studies may be used by governments to absolve themselves of responsibility for improving housing. However, the current Labour government's interest in identifying “what works,” and its emphasis on joined up decision making, may facilitate a less fragmented approach to tackling deprivation. The number of current collaborative housing studies suggests a greater willingness to use such joined up approaches. What is also needed is robust evidence of their effects.

Other evidence

Sources of evidence other than experimental studies are also important. Longitudinal studies have been recommended as a useful, if expensive, study design in evaluating complex interventions such as housing.³⁵ For example, recent results from the 33 year follow up from the longitudinal national childhood developmental study show that poor housing adversely affects health in later life. The study found a dose-response relation, with multiple housing deprivation leading to greater risk of disability or severe ill health in later life.³⁶ Data from the Boyd-Orr cohort also show that childhood housing conditions have an effect on adult health independent of the effects of socioeconomic deprivation.³⁷

Further research is also needed into the direction of the relation between health and housing. Previous work has suggested that poor health can negatively affect housing opportunities.³⁸

Although our review focused on major housing improvements, good evidence exists from systematic reviews that other interventions to improve health inside the home may be effective. Among these are interventions to reduce house dust mites and to reduce accidents among children and elderly people.^39–41

Conclusion

The basic human need for shelter makes the relation between poor housing and poor health seem self evident.⁴² Despite, or perhaps because of, this intuitive relation, good research evidence is lacking on the health gains that result from investment in housing. We know little about the mechanisms of interaction of social factors and the effects of poor housing over the lifecourse. There is also a lack of comparative information on the costs and effects of specific housing improvements, such as central heating or major refurbishment. It is this type of evidence that is likely to be most valuable to policy makers and housing providers. Large scale studies that investigate the wider social context of housing improvements and their comparative effectiveness and cost effectiveness are now required.