Background
In New Zealand infectious diseases emerged as an increasing public health problem during the 1990s. Hospitalisation rates from infectious diseases increased by about 50 % during that decade [
1]. New Zealand experienced a severe and prolonged meningococcal disease epidemic, which began in 1991 and resulted in disease rates that were about 10 times higher than pre-epidemic levels [
2]. In addition, New Zealand has relatively high rates of several diseases spread by respiratory routes and close physical contact, notably rheumatic fever [
3] and childhood pneumonia [
4].
A case–control study of meningococcal disease conducted in Auckland from 1997 to 1999 showed the risk of disease in children was highly associated with household crowding, as measured by the number of adults and adolescents per room [
5]. Other New Zealand studies identified an association between household crowding and the risk of rheumatic fever [
6], tuberculosis [
7] and childhood pneumonia [
4].
Despite the intuitive logic that household crowding contributes to the spread of infectious diseases there is a surprisingly small published literature on the health effects of this exposure and few published studies on the effects of housing interventions to reduce household crowding [
8]. A New Zealand review of the effects of crowding on health concluded, “The debate about the relationship between crowding and health is long standing and inconclusive” [
9].
The Housing and Health Research Programme (He Kainga Oranga) was formed in response to concerns about the health consequences of poor housing and how these could be addressed. One of its core projects, the Social Housing Outcomes Worth (SHOW) Study, was established to investigate the health effects of housing conditions, particularly household crowding [
10]. The study was actively supported by the Chief Executive of New Zealand’s social housing provider, Housing New Zealand Corporation (HNZC) [
11]. We know of no other studies that have used large administrative housing datasets to explore household crowding and health issues in this way.
The SHOW Study has the following aims: (1) To describe the characteristics of social housing applicants and tenants and their use of social housing over time; (2) To assess the relationship between household crowding and other household exposures and health outcomes; (3) To assess the impact of placement in social housing on the health status of tenants; (4) To assess the impact of housing improvements on the health of tenants.
Here we report on the methods and feasibility of this study and the first of its aims to describe the characteristics of social housing applicants and tenants.
Discussion
This research has demonstrated that it is possible to establish a large cohort study using administrative data collected by a public housing provider. It has also shown that a high proportion of subjects in this database can be linked to their health records in a way that is ethically acceptable and protects privacy. This population of social housing applicants and tenants is characterised by its young age, high proportion of Māori and Pacific peoples, and sole parent households. It has a low income and high level of receipt of Government income support.
These results also show that this population is highly exposed to household crowding and household smoking, compared with the New Zealand population more generally. It has also demonstrated that the majority of housing applicants who became tenants decrease their level of household crowding in the process. This study therefore has the potential to add to the relatively small evidence base of research on the health effects of housing conditions. A further advantage of this study is that by focussing on social housing tenants, who are a particularly vulnerable population, it partly redresses the balance in other New Zealand cohorts which are disproportionately European.
Using administrative databases, such as that collected by HNZC, has the advantage of providing a potentially large cohort population at relatively low cost. Similarly, using established national hospitalisation and mortality databases to identify health outcomes in this cohort allows a far larger study size than would otherwise by possible. This study design also lets the researchers investigate multiple health outcomes potentially allowing the study to quantify a larger proportion of the overall burden of disease that can be attributed to household conditions than could be done with individual disease studies. It also enables researchers to analyse the effect of important changes to social housing policy that have occurred during the course of the cohort.
This study design and methods have a number of potential weaknesses and limitations that are common to many observational epidemiological studies, notably confounding, selection bias, information bias, and issues of generalisability.
The high level of matching of cohort members with health data is important for minimising potential selection bias. Such a bias could occur if, for example, the relationship between crowding and disease risk was different for those included (matched) compared with those who were not included (unmatched) [
16]. Given the high proportion that is matched, this bias is unlikely to be important.
Household crowding is highly associated with other measures of socio-economic deprivation such as low income, unemployment, low education level and fewer material resources. Other risk factors for respiratory disease, notably active and passive smoking, are also more prevalent in crowded households [
17]. These confounders could have the effect of producing or increasing the measured cross-sectional association between household crowding and increased hospitalisations. Several potential confounders, notably age, sex, and ethnicity are well recorded and will be used in the analysis. Because the study population is defined on socio-economic grounds (as particularly deprived) some confounders are effectively controlled by restriction. Data on active and passive smoking are also recorded for more than half of the households, which will permit analysis of this effect on a substantial sub-group. But most importantly, the longitudinal component of this study, which follows participants as they change their levels of crowding over time, will effectively overcome the issue of confounding by other important and relatively fixed covariates, such as chronic disease and disability status.
Some study data are dependent on the accuracy of information that housing applicants and tenants supply to HNZC, which introduces potential for information bias of exposures and covariates. This particularly applies to information on the number of people living in the applicant and tenant households. Housing applicants may tend to over-state the number of people in their homes to increase their priority on the waiting list. Housing tenants may under-count the number of people staying with them to minimise their rent (though HNZC tenants can house two boarders without it counting as income for the calculation of income-related rent). Consequently, if there truly was an association of reduced hospitalization rates due to reduced household crowding after placement from an applicant to tenanted house, we may only detect such a reduction for apparent reductions in household crowding that exceeded those purely due to biased recording. This effect will be considered in a sensitivity analysis of the findings.
Because this study is restricted to social housing users, who are by definition a socio-economically deprived group in New Zealand, the generalisability of the findings to the total population may be limited. However, as noted previously, a major aim of this cohort is to track the impact of the move to social housing for this economically-deprived population rather than for the New Zealand population in general. Because of the cohort design, this study will also allow researchers to see whether the role of household crowding that has so far been mainly seen for infectious diseases in children, can be generalised to a wider range of infectious and non-infectious diseases where such a role seems plausible.
Conclusions
The SHOW Study has demonstrated that an administrative housing database can be used to form a large cohort population and successfully link cohort members to their health records in a way that meets confidentiality and ethical requirements. It confirms that social housing tenants are a highly deprived population with relatively low incomes and high levels of exposure to household crowding and environmental tobacco smoke.
This study provides a mechanism for investigating the role of housing conditions, such as household crowding, as a risk factor for a range of diseases and injuries. By taking advantage of a ‘natural experiment’ where large numbers of housing applicants are re-housed in public housing, this study could also add to the small literature on the health effects of housing improvements [
8].
This study also demonstrates the value of partnerships between researchers and service providers. By building research and evaluation into service deliver processes it is sometimes possible to provide research findings in a more efficient way than with ‘stand-alone’ research projects. Now that it is established, this study could be extended at relatively low cost to investigate other links between housing conditions and health outcomes. Such mechanisms provide benefits for the service provider as well as the research and policy end-users.
Acknowledgements
Establishment and operation of this cohort study has been support by funding from the Health Research Council of New Zealand and HNZC. We wish to particularly thank the following current and former staff at HNZC: Patricia Laing, Blair Badcock, Michael Lennon, and Helen Fulcher. Staff at the Ministry of Health who have assisted the study include Chris Lewis. We also wish to thank colleagues who have contributed to establishment of this study while working with us at He Kainga Oranga: Jasminka Milosevic, Robin Turner, and Charlotte Kieft.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (
http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MB initiated and designed this study, carried out some of the analyses, and drafted the manuscript. JZ carried out the data management and some of the analyses. TB, JC, KS-S, PH-C assisted with design and establishment of the SHOW Study. All authors read and approved the final manuscript.