A Systematic Review of Health Economic Analyses of Housing Improvement Interventions and Insecticide-Treated Bednets in the Home

Background Housing improvements have considerable potential for improving health. So does the provision of insecticide-treated bednets for malaria prevention. Therefore we aimed to conduct updated systematic reviews of health economic analyses in both these intervention domains. Methods and findings The search strategy included economic analyses of housing improvement interventions and use of insecticide-treated bednets for community-dwelling, healthy populations (published between 1 January 2000 and 15 April 2014). We searched the Cochrane Database of Systematic Reviews, MEDLINE, PubMed, EMBASE, and three health economics databases. Thirty-five economic analyses of seven types of intervention fulfilled the inclusion criteria. Most included studies adopted a health sector perspective and were cost-effectiveness analyses using decision analytic modeling or conducted alongside trials. The overall quality of the studies was generally likely to be adequate for informing policy-making (albeit with limitations in some areas). There was fairly consistent evidence for the cost-effectiveness/favorable cost-benefit of removing indoor lead to prevent lead poisoning and sequelae, and retrofitting insulation to prevent lung disease. But the value of assessing and improving home safety and providing smoke alarms to prevent injuries was more mixed and the economic evidence was inconclusive or insufficient for: home ventilation to prevent lung disease, installing heaters to prevent lung disease and regulating tap water temperatures to prevent scalding. Few studies (n = 4) considered health equity. The 12 studies of providing insecticide-treated bednets or hammocks to prevent malaria found these interventions to be moderately to highly cost-effective. Conclusions This systematic review provides updated evidence that several housing improvement interventions (such as removing indoor lead and retrofitting insulation) and also the provision of insecticide-treated bednets are cost-effective interventions. Nevertheless, for some interventions additional analyses are required to better clarify their health economic and health equity value.


Methods and findings
The search strategy included economic analyses of housing improvement interventions and use of insecticide-treated bednets for community-dwelling, healthy populations (published between 1 January 2000 and 15 April 2014). We searched the Cochrane Database of Systematic Reviews, MEDLINE, PubMed, EMBASE, and three health economics databases. Thirty-five economic analyses of seven types of intervention fulfilled the inclusion criteria. Most included studies adopted a health sector perspective and were cost-effectiveness analyses using decision analytic modeling or conducted alongside trials. The overall quality of the studies was generally likely to be adequate for informing policy-making (albeit with limitations in some areas). There was fairly consistent evidence for the cost-effectiveness/favorable cost-benefit of removing indoor lead to prevent lead poisoning and sequelae, and retrofitting insulation to prevent lung disease. But the value of assessing and improving home safety and providing smoke alarms to prevent injuries was more mixed and the economic evidence was inconclusive or insufficient for: home ventilation to prevent lung disease, installing heaters to prevent lung disease and regulating tap water temperatures to prevent scalding. Few studies (n = 4) considered health equity. The 12 studies of providing insecticide-treated bednets or hammocks to prevent malaria found these interventions to be moderately to highly cost-effective.

Introduction
Housing quality is an important social determinant of health [1]. Thus, interventions that improve housing quality have the potential to improve individual and population health [1]. If these interventions are targeted at populations disadvantaged by living in housing of relatively poor quality, then they also have the potential for improving health equity [1]. Recent systematic review evidence suggested that selected housing interventions effectively improve health [2]. It was concluded, for example, that "housing investment which improves thermal comfort in the home can lead to health improvements, especially where the improvements are targeted at those with inadequate warmth" (p. 2) [2]. Thus, the effectiveness of some housing interventions is now fairly evidence-based [2]. In addition, some such interventions also have considerable non-health benefits in domains such as climate change, energy use and income. For example, retrofitting insulation to improve thermal comfort may not only improve health, but at the same time also reduce domestic energy use and anthropogenic green-house gas emissions.
Economic analysis is "the comparative analysis of alternative courses of action in terms of both their costs and their consequences" (p. 1) [3]. Health economic analyses encompass three primary types of studies, i.e., cost-benefit, cost-effectiveness and cost-utility analyses. The World Health Organization (WHO) has noted the central importance of evidence from economic analyses of the health impact of interventions addressing the social determinants of health, including housing-related interventions, in making the economic case for such interventions. The WHO has also noted the scarcity of and called for additional such evidence to be produced [4]. Systematic reviews of health economic analyses remain scarce, but are increasingly gaining attention. While specific, standard methodology for such systematic reviews has been proposed, additional development and refinement to these methods is required [5]. Best practice guidelines [5] are clear that such systematic reviews should always review both economic analyses conducted alongside randomized controlled trials [6] (or, more broadly, intervention studies) and economic analyses using decision analytic modeling [7]. The main use of such systematic reviews is to inform decision model development; identify the most relevant study for a specific decision context; and understand the key economic trade-offs and causal relationships in a decision model or treatment area [5]. Because cost-effectiveness analytic evidence is commonly highly context specific, systematic reviews of economic evaluations of the health impact of interventions may have relatively little potential to produce (pooled) cost-effectiveness estimates that are generalizable [5], unless multiple methodologically and statistically homogenous studies from a comparable context are available for meta-analysis.
We identified only one previous systematic review specifically of economic analyses of the health impact of housing improvements [8]. This review synthesized evidence from economic analyses that were included in another systematic review [9] of the effectiveness of housing interventions that was published in 2009 (which covered studies published before 2008). It identified two such studies of community-dwelling, healthy populations conducted before 2008, namely one study each of installing heaters [10] and retrofitting insulation [11][12][13], plus one study [14,15] of a population with pre-existing conditions and another study [16] of an intervention that combined housing with major non-housing components. The review authors concluded that there is a "near absence of economic evaluation of housing improvements" (p. 843) [8]. Similarly, the authors of a cost-effectiveness study of enhanced home ventilation published in 2011 have argued that their study was only the second comprehensive economic analysis of a housing intervention [17].
An updated systematic review is therefore necessary for establishing the status of evidence from economic analyses of the health impact of housing interventions for four reasons. First, research activity on housing and health has grown substantially in the last seven years (i.e., [2008][2009][2010][2011][2012][2013][2014], and so new studies may have been published that the previous systematic review [8] did not include. Second, this previous systematic review [8] was fairly narrow in scope for included study types and interventions. For example, it excluded any economic analyses using decision analytic modeling [7] and may also have excluded some types of housing interventions altogether, such as home safety assessment and modification (HSAM) interventions (e.g., those for preventing injuries in community-dwelling older adults). Third, this previous systematic review [8] did not search dedicated health economics databases. Furthermore, the status of equity analysis conducted as part of economic analyses has not specifically been reviewed before, despite the WHO calling specifically for economic analyses of social determinants of health interventions to assess and value effects on health equity [4].
Another in-house intervention is the provision of insecticide-treated bednets (ITBNs) for malaria prevention. This intervention has existing evidence for cost-effectiveness from a systematic review [18] that covered interventions published between 2000 and 2010. But since this is an active area for on-going research, we considered it appropriate to consider an updated review of this literature.
Given the above, our study objective was to provide updated systematic reviews of health economic analyses of both housing interventions and the provision of insecticide-treated bednets for community-dwelling, healthy populations.

Study eligibility criteria
We developed a strict study protocol before we commenced the search stage, which is available from the authors on request. To be included in this systematic review a study had to be an economic analysis, i.e., a cost-benefit, cost-effectiveness and/or cost-utility analysis. Following best practice guidelines [5], we included both economic analyses conducted alongside randomized controlled trials [6] (or other intervention studies) and economic analyses using decision analytic modeling [7]. Any other study types, including studies of costs alone, were excluded.
Interventions were included if they were structural, physical changes to housing infrastructure and/or contents, or involved insecticide-treated bednets (ITBNs) or insecticide-treated hammocks (ITHs). Other included in-home health promoting interventions were smokealarms, and devices which removed health-hazards such as unflued gas heaters or open-fire cooking stoves. Interventions that removed lead from the home (e.g., lead paint) were also included. However, interventions removing radon from homes were excluded. This was because a detailed review of the cost-effectiveness of indoor radon control interventions has previously been published by the World Health Organization [19]. We excluded interventions for changing the outdoor physical environment (e.g., creation of parks and gardens); the social environment (e.g., neighborhood crime reduction) or economic environment of the house (e.g., improve affordability of housing). We also excluded multi-mode interventions that combined structural housing interventions with more predominant other interventions (e.g., HSAM provided as a minor intervention alongside major education or exercise programs). Rehousing interventions (e.g., for low-income families) were also excluded, because they may have acted primarily through the changed social environment, rather than a change to the physical environment.
All types of static, physical, permanent houses were included. The exception was that we excluded institutionalized housing such as hospitals or supported housing for vulnerable populations. Non-static, (potentially) non-permanent houses such as caravans and house boats were also excluded, because they may not be comparable to the houses included in the systematic review.
Included participants were community-dwelling (non-institutionalized), healthy (without major pre-existing conditions) populations residing in any country. Homeless people and people with major pre-existing conditions such as moderate or severe asthma, severe visual impairments and HIV and/or AIDS were excluded from the review, because housing interventions may have a different cost-effectiveness in these populations than in the healthy general population. However, general population samples that included some people with pre-existing conditions were included. Records written in any language were included.

Search and screening
One review author (FP) searched a total of seven electronic academic databases between 28 April and 15 May 2014 for economic analyses of the health impact of housing interventions in community-dwelling, healthy populations published between 1 January 2000 and 15 April 2014. Fig 1 presents the MEDLINE search strategy, and this search strategy was adapted to suit the searches of the other database. The Cochrane Database of Systematic Reviews (The Cochrane Library 2014, Issue 5), MEDLINE, PubMed, and EMBASE electronic academic databases were searched. Three specialized databases of health economic analyses were also searched, i.e., the Health Economic Evaluation Database, the NHS Economic Evaluation Database and the Tufts Cost-Effectiveness Analysis Registry. We also screened the first 30 hits on Internet search engine Google Scholar. When we were near the completion of the systematic review we searched the PubMed database again for records published after 15 May 2014 to identify any additional studies and records published since the original searches had been conducted.
One review author (FP) screened the titles of all potentially relevant records and then the abstracts of all records with potentially relevant titles. Both review authors then independently screened the full texts of all potentially relevant records in depth for the inclusion criteria. We also independently hand searched the reference lists of relevant previous systematic reviews [8,18] and of each included study record for additional studies or records. system or societal); time horizon; and effectiveness outcome measure (quality-adjusted lifeyears [QALY] gained, disability-adjusted life-years [DALY] averted, life-years saved [LYS], lives saved or deaths averted and other outcomes). We also documented as qualitative characteristics the sources for estimation of effectiveness, sources for estimation of resource utilization, discount rates used and whether sensitivity analysis had been conducted. Sources for estimation of effectiveness and resource utilization were classified as primary data collection (for example, questionnaires or trials), secondary data collection (for example, administrative records), literature (for example, systematic reviews of effectiveness) or expert opinion. Use of discount rates was classified into use on costs, effects or both costs and effects, or not used at all.

Quality assessment
Both review authors independently assessed the quality of the included economic analyses using Drummond and Jefferson's established checklist [21] as modified by Zelle and Baltussen [22]. This checklist included 29 items covering five categories, namely study design; effectiveness estimation; cost estimation; analysis; and interpretation of results. We independently assessed, for each study, each item along a three-point scale, allowing grading as to whether the respective item was fully considered (assigning 2 points), partially considered (1 point) or not considered (0 points) in the study. Items that were irrelevant for a study were excluded from the quality assessment. For example, if a study sourced effectiveness estimates from an individual study, then we did not assess whether the study provided details of the method of synthesis or meta-analysis of effectiveness estimates (item 11). The two reviewers discussed any disagreement between the quality assessments until resolution was reached. To provide a mean quality score, scores were summed and compared to the maximum attainable score, both for each assessment category and overall. We also screened all studies for potential disclosed and undisclosed financial conflicts of interest and note these as part of our quality assessment. In the absence of best practice guidelines for assessing publication bias in systematic review of economic analyses, we assessed whether the body of evidence included a plausible range of costeffectiveness estimates across the included studies.

Study findings
We extracted and documented each study's objective, the comparator intervention(s) and the main outcomes measures (i.e., cost-effectiveness or cost-benefit measures). The main outcome measures of this systematic review were (i) the study author's or authors' conclusion of the cost-effectiveness or cost-benefit of the housing intervention (e.g., differentiating not, likely or highly cost-effective, or cost-saving) and (ii) the average cost per relevant health outcome (e.g., average cost per DALY averted). If study authors did not reach a clear conclusion on the costeffectiveness of an examined housing intervention, we judged cost-effectiveness ourselves. We based our judgment on the WHO standard [23] of one per-capita gross domestic product (GDP; sourced from the World Bank [24]) per QALY gained or DALY averted as indicating high cost-effectiveness and two to three per capita GDP as indicating likely cost-effectiveness. In the absence of a global standard of cost-effectiveness for the cost of one malaria infection averted, we applied an arbitrary cost-effectiveness threshold of US$50. If a study reported costeffectiveness or cost-benefit measures for multiple health outcomes, then we prioritized measures for the health outcomes in the following order: DALYs averted over QALYs averted over LYS over lives saved or deaths averted over other outcomes. For studies that report results from two types of economic analyses (i.e., combined cost-effectiveness and cost-benefit analyses), we extracted and reported findings from both economic analyses types.

Search results
The search identified a total of 7,957 records (without duplicates). Fig 2 presents a PRISMA flow-chart of the selection of studies. After title screening, 374 records were considered potentially relevant. After abstract screening, 138 records of 135 studies were still considered potentially relevant and these progressed to in-depth full-text screening. This next screening phase identified 34 studies with 37 records (described below). Hand searching of the reference lists of relevant previous systematic reviews [8,18] identified one additional eligible study (Grimes et al 2012) [25]. The review included a final total of 35 studies with 38 records that fulfilled eligibility criteria, and they were included in the review and synthesized qualitatively. Study heterogeneity prohibited combining studies in meta-analysis.
analyses, one was a cost-utility analysis and two combined cost-effectiveness with cost-benefit analyses. The base year of cost data ranged between 1990 and 2014, although it was unclear in four studies. Most analyses used decision analytic modeling (n = 14 studies) or were conducted alongside randomized controlled trials or other experimental studies (n = 12), but some were conducted alongside observational studies (7), and two were conducted alongside intervention implementation. Slightly more studies took a health system perspective (22) than a societal perspective (13). Societal perspectives, in addition to health, covered one or more of the domains of climate change, energy, crime, education, income, productivity and wealth. The time horizon was short (i.e., 1-4 years) in 14 studies, medium (i.e., 5-10 years) in six studies, long (i.e., >10 years) in 11 studies and unclear in four studies.
The most common country setting was the United States (n = 11 studies), followed by New Zealand (4) and the United Kingdom (3). Two studies each were conducted for Australia, Eritrea, Kenya and Togo, and one study each for: Canada, France, India, Malawi, Myanmar, the Netherlands, Senegal, South Africa, Tanzania, Thailand, Vietnam, the Southern / Eastern African region and the Western African region. Study populations were the community-dwelling, healthy general population (n = 12 studies), children (13), older people (6) and/or specific population groups (e.g., persons residing in geographically deprived areas or persons earning low or middle incomes) (6). Study populations were generally well-tailored to the studied interventions. For example, most HSAM interventions were studied in older people, and indoor lead paint removal was studied in children.
HSAM interventions were studied in seven studies [39][40][41][42][43][44][45]. All of these interventions aimed to reduce falls in older people (except one study that aimed to reduce injury in the general  population [42]) by identifying and removing fall and other injury hazards in the home, for example by installing hand rails on stairs or in the bathroom and removing trip hazards. Removal of indoor lead was examined in five studies [47][48][49][50][51]. These interventions included removal of lead-containing window frames and other indoor materials, as well as enforcement of public policy for removing indoor lead. Retrofitting of insulation and/or installation of heaters was examined in five studies [10-13, 25, 52, 53]. Finally, giving away or installing smoke alarms was examined in three studies [54,55,58], regulating household tap water temperature to prevent scalds was examined in two studies [56,57], and increasing home ventilation was examined in one study [46]. Effectiveness outcome measures were QALYs gained (n = 3 studies), DALYs averted (8), life-years saved (LYS) (2), lives saved or deaths averted (10) and/or one or more other outcomes (24 studies). Other outcomes comprised morbidity ones (e.g., asthma, attention deficit hyperactivity disorder, lead poisoning, malaria infection, mental retardation or psychological distress), injury (e.g., fall or scald), medication and use of health services (e.g., GP visit and hospitalization). Effectiveness estimates were sourced primarily from the literature (n = 14 studies) and secondary data (15), and less commonly from primary data (7), expert opinion and/or an unclear source (1 each). Estimates of resource utilization were sourced from secondary data (19), literature (18 studies), primary data (3), expert opinion (2), unclear sources (2) and/or a field study or operational costs (1 each). If discount rates were used, they ranged between 3% and 5.3%, but were most commonly either 3% or 5% and applied to both effects and costs. However, just under half of all included studies (17) did not discount either or both of effects and costs, and four studies did not discuss discounting. Sensitivity analyses were conducted in most (27) studies. Only four studies reported some form of equity analysis, which were generally limited to comparisons of targeted (to disadvantaged populations) and untargeted strategies. For example, one study compared the cost-benefit of retrofitting insulation targeted to the low-and middle-income populations with that of the intervention targeted to the high-income population [11][12][13].  [53]. For interventions examined by four or more studies, studies on provision of smoke alarms received the highest average score (83%), followed by studies on provision of ITBNs or ITHs (73%) and retrofitting insulation and/or installing heaters (68%). Relatively low average scores were received by studies of removal of indoor lead (63%) and HSAM (62%). On average,

Quality assessment
Economic Analyses of Housing Interventions: Systematic Review studies scored highly on the categories of effectiveness estimation (94% of the maximum attainable score over all studies) and interpretation of results (86%). However, cost estimation, study design and analysis scored relatively poorly with 54%, 59% and 62%, respectively. We found the study [46] on home ventilation for preventing chronic lung disease to have a potential financial conflict of interest, because it received funding from "a society that aims to encourage technical, economical and scientific developments in the Dutch building services industry" (p. 384). The range over which the cost-effectiveness estimates ranged appeared plausible and study authors' conclusions included "not cost-effective" in some studies, which provided no evidence for publication bias. Table 4 presents study findings. Overall, there was a considerable body of evidence on the costeffectiveness of the health impact of several housing interventions (including for ITBNs) in several populations and country settings. In general, this body of evidence suggested that these interventions were moderately to highly cost-effective or cost-beneficial, although cost-effectiveness varied by different intervention types. There was fairly consistent evidence for the cost-effectiveness or favorable cost-benefit of ITBNs or ITHs, indoor lead removal and retrofitting of insulation. The 12 studies of ITBNs or ITHs for preventing malaria or other insect-borne infections found these interventions to be moderately to highly cost-effective [26][27][28][29][30][31][32][33][34][35][36]38]. This body of evidence included two studies [28][29][30] which found ITBNs to have an cost per malaria infection averted of less than US$50, and one study [35] which found ITBNs to have a cost per DALY averted of Int$51, which we regard to indicate high cost-effectiveness. The only exception was one study [26] that found ITBNs to have an cost per malaria infection averted of US$52, which could be of borderline cost-effectiveness assuming a cost-effectiveness threshold of US$50. Two studies found that, under certain conditions, long-lasting ITBNs were relatively more cost-effective than conventional ITBNs [34,37].

Study findings
Five studies on indoor lead removal for reducing lead poisoning and sequelae consistently found this intervention to be highly cost-beneficial, accruing considerable net benefits [47][48][49][50][51]. Three studies on retrofitting insulation found that retrofitting insulation was highly cost-effective or cost-beneficial [11-13, 25, 52].
The evidence was more mixed around the cost-effectiveness or cost-benefit for HSAM and provision of smoke-alarms. Seven studies of HSAM for reducing injury (predominantly falls) had mixed results, but tended towards being cost-effective/cost-beneficial overall. One study found HSAM not to be cost-effective among older people [43]. Another study found HSAM likely be cost-effective among older people [41]. Relatively high cost-effectiveness was found in two studies [40,44] for older people and one study [42] for the general population. Similarly, another study [39] suggested HSAM was a very cost-effective intervention among older people, producing health gain below the WHO's standard cost-effectiveness threshold [23] of the per-       [45]. The three studies of smoke alarm provision found mixed evidence favoring cost-effectiveness. One study found that giving away smoke alarms was unlikely to be cost-effective [54]. In contrast, another found that both giving away and installing smoke-alarms was cost-effective and cost-beneficial, with the give-away programs being relatively more cost-effective than the installation program [58]. And the third study found this intervention to be cost-saving [55].
Finally, the economic evidence was inconclusive for installing heaters and insufficient for home ventilation and regulating tap water temperatures. Three studies of installing heaters could either not establish the cost-effectiveness of this intervention [10,25] or found cost-effectiveness only if the intervention was targeted at households with high asthma rates [53]. One study found home ventilation to be cost-effective [46], but we note that this study may have had a financial conflict of interest. Both studies examining regulating tap water temperature found this intervention to be cost-effective [56,57], but we judged this body of evidence too small in size to be considered conclusive.

Main findings and interpretation
This review identified 35 economic analyses of housing improvement interventions, of which 12 involved ITBNs or ITHs. Most of the included studies were cost-effectiveness analyses that either used decision analytic modeling or were conducted alongside randomized controlled trials or other experimental studies and that adopted a health sector perspective. Overall, the quality of the body of evidence is probably acceptable for informing policy-making regarding some intervention types (i.e., provision of ITBNs or ITHs, lead paint removal and retrofitting of insulation). But there is still considerable scope for additional studies and for improvements in study quality for other types of housing interventions. Furthermore, few studies determined the relative cost-effectiveness of different housing interventions (i.e., HSAM versus home insulation), and no study determined the cumulative cost-effectiveness of implementing two or more housing interventions at once (e.g., HSAM plus home insulation).  ITBNs vs indoor residual spraying To estimate the cost-effectiveness of ITBNs in preventing malaria infection ITBNs were more effective and more costly at a cost per death averted of R11,718 (US$1,915).
Guyatt et al, 2002 [28,29] ITBNs vs standard practice To estimate the cost-effectiveness of ITBNs in preventing malaria infection [28] ITBNs had a cost per malaria infection averted of US$29 [28]. [Assuming a cost-effectiveness threshold of US$50 per malaria infection averted, ITBNs were highly cost-effective.] As above As above To estimate the cost-benefit of ITBNs in preventing malaria infection [29] ITBNs had a cost per person protected from malaria infection of US$2.34 [29].  In summary, this review found that several housing improvement interventions were costeffective or cost-beneficial in general. There was a considerable amount of consistent evidence of relatively high quality for the cost-effectiveness or favorable cost-benefit of providing ITBNs or ITHs, removing indoor lead paint or dust and retrofitting insulation. In contrast, evidence on providing HSAM and giving away or installing smoke alarms while mostly of acceptable quality had mixed results (though favoring cost-effectiveness) and, for providing smoke alarms, was relatively small (n = 3 studies). The relatively small body of evidence on installing Economic Analyses of Housing Interventions: Systematic Review heaters was inconclusive, with two studies completely unable to estimate cost-effectiveness or cost-benefit. Finally, evidence on home ventilation and regulating tap water temperature was insufficient in size to judge value from a health economic perspective (n = 1 study and n = 2 studies, respectively) and, for home ventilation, was also of relatively low quality and with a potential financial conflict of interest. The objective of this systematic review was to synthesize evidence on the cost-effectiveness of housing improvement interventions as policy tools for improving health and health equity, but the systematic review also found evidence that some housing interventions have considerable non-health co-benefits. For example, retrofitting insulation can cost-effectively improve health and at the same time also reduce domestic energy use and, in turn, anthropogenic greenhouse gas emissions [11][12][13]. Abating lead from the home can cost-effectively improve health over the life-course and can also result in savings from less special education, reduced crime and increased lifetime earnings and productivity [51]. Therefore, it could be argued that the economic benefits of housing interventions found in this review present an underestimate of their total benefits, and that the identified co-benefits further advance the economic case for the cost-effectiveness of housing interventions.
The only previous systematic review specifically of housing interventions covered only two studies [10][11][12][13] of two intervention types (installing heating and retrofitting insulation) in the community-dwelling, healthy population and concluded that there is a "near absence of economic evaluation of housing improvements" (p. 843) [8]. However, our updated review identified 23 housing improvement studies (in addition to the 12 on ITBNs) in community-dwelling, healthy populations. Similarly, we believe our review benefited from a broader scope for considering interventions.
Furthermore, in terms of ITBNs and ITHs, the most relevant previous systematic review [18] covered publications up to 2010, and included only seven studies [26, 27, 29-31, 36, 37]. However, our review identified 12 studies of this intervention type.

Review strengths and weaknesses
As well as its relatively broad scope, a strength of this systematic review was that we searched several electronic databases, including three databases of health economic analyses. This helped ensure that we captured relevant economic analyses published in both the academic and grey literature. Also, the two review authors independently screened potentially relevant records for eligibility criteria, extracted data and assessed the quality of the included studies. As a result, we are reasonably confident that we have identified nearly all relevant studies in the academic literature from 1 January 2000 to 15 April 2014, but some relevant reports in the grey literature may have been missed.

Potential implications for policy-making
The findings of this review provide support for the economic case [4] for addressing housing quality as a social determinant of health [1]. In particular, they help provide the economic case for interventions to remove lead paint and for retrofitting insulation.
In countries with malaria and donor countries supporting them, there is now even a stronger health economic case for the provision of ITBNs or ITHs. However, local circumstances (e.g., for ITBNs or ITHs, local levels of insecticide resistance and access to local malaria treatment services) will, of course, influence the cost-effectiveness of such interventions.

Implications for future research and guideline revisions
The existing evidence is currently limited in its ability to determine whether housing improvement interventions are cost-effective policy tools for improving health equity. Only four out of the 35 included economic analyses considered health equity, and generally such considerations were limited to comparisons of the cost-effectiveness or cost-benefit of housing interventions when they were targeted versus untargeted to disadvantaged populations. This is therefore an important area for further research.
Future research should also address the currently mixed evidence (favoring cost-effectiveness) for providing HSAM and giving away or installing smoke alarms; inconclusive evidence for installing heaters; and insufficient evidence for improving home ventilation and regulating tap water. Existing evidence on most housing intervention types (except for providing ITBNs and ITHs) is currently limited to high-income countries (mostly only the United States, New Zealand and the United Kingdom).
For all housing intervention types, additional economic analyses of the health equity impact of housing interventions are required. Although best practice guidelines for economic analyses of interventions addressing the social determinants of health suggest that such analyses should consider health equity [4], this review found that few existing economic analyses of housing interventions have considered health equity even in a minor way. This leaves a considerable gap in the current body of evidence that future research should fill.
Moreover, a considerable number of studies included in the review took a relatively narrow health sector perspective, which is likely to underestimate the full economic benefits for society that housing interventions are likely to have. In line with WHO guidelines for economic analyses of interventions addressing the social determinants of health [4], future economic analyses of housing interventions should try to adopt a societal perspective and carefully include all relevant potential co-benefits, if feasible.
Furthermore, homes are particularly important sites of public health intervention, not only because most people spend considerable amounts of their time at home, but also because it is potentially possible to cost-effectively address multiple health hazards at once, potentially with multiple interventions. For example, a recently trialed HSAM intervention that comprehensively removed injury hazards (rather than only falls hazards) for all household members (rather than only older people in the household) has been found to be highly cost-effective [42]. Taking this one step further, future economic analyses should determine the cumulative economic value of combining several housing interventions. For example, the cost-effectiveness of combining the relatively broader HSAM intervention with other housing interventions such as insulation retrofitting is currently unknown.
Finally, this review also has methodological implications for systematically reviewing economic analyses of the health and health equity impact of interventions. A previous systematic review [8] of economic analyses with comparable focus (albeit smaller scope) that was conducted alongside a review of effectiveness identified two studies of community-dwelling, healthy populations, whereas our systematic review identified 23 relevant studies. This highlights the relative limitations of conducting systematic reviews of economic analyses alongside systematic reviews of effectiveness, if only the economic analyses conducted alongside studies included in the systematic review of effectiveness are included. This places emphasis on best practice guidelines [5] calling for comprehensive systematic reviews of economic analyses to include economic analyses using decision analytic modeling [7]. Furthermore, searching specifically for economic analyses and searching electronic databases of health economic analyses can considerably increase the coverage of systematic reviews of economic analyses of the health and health equity impact of interventions.