Substandard housing conditions and hazardous indoor environmental exposures contribute to significant morbidity and mortality worldwide [1]. Despite their known adverse health effects, most national surveys and housing indices do not collect this information. Our study developed a national, multidimensional Housing and Environmental Quality Index (HEQI) informed by the WHO’s Housing and Health Guidelines and composed of ten domains addressing structural deficiencies, indoor environmental exposures, and building conditions associated with adverse health. Using the 2019 AHS data, the HEQI identified approximately 92 million (79%) U.S. households with one or more HEQI risk factors. Compared to established housing indices, the HEQI captured four new environmental health domains of household fuel combustion, high indoor temperatures, severe crowding, and high building leakage, which enabled the identification of 57.7 million (63%) more households at risk.
The multidimensional HEQI performed better than established housing indices at capturing both housing quality and environment health risk factors. Established indices focus primarily on physical deficiencies, costs of repair, or the deflation in home values as a result of these deficiencies [11,12,13, 47, 48]. In particular, the PQI and Adequacy Index failed to capture environmental risk factors like mold, cockroaches, household crowding, household fuel combustion, and higher building leakage. Moreover, although prior studies have used the AHS to characterize environmental risk factors, most have focused on single AHS items like thermal comfort [51], air exchange [52], wood combustion [53], mold [54], and pests [55]. To our knowledge, the HEQI is the first multidimensional index that captures a range of housing quality and environmental health risk factors.
Prevalent environmental risk factors identified by the HEQI and not well-captured by established housing indices were household fuel combustion and dampness and mold. Household fuel combustion sources include gas cookstoves, wood and kerosene heating fuel, and fireplaces. These sources emit carbon monoxide, particulate matter, nitrogen dioxide, and other hazardous air pollutants that are associated with adverse cardio-respiratory health effects, particularly among children and the immunocompromised [56,57,58]. Approximately 72 million (61.4%) U.S. households reported the presence of least one household fuel combustion source, particularly gas cookstoves and fireplaces. Due to limitations of the AHS data, we did not have information about modifiers of emission levels such as appliance type, frequency of appliance use, and the presence and use of ventilation controls in order to better quantify the level and duration of exposure. However, we did observe that homeowners, single-family households, and households with children reported higher prevalence of household fuel combustion sources and lived in more airtight buildings, suggesting that these households may have a higher risk of exposure and could be prioritized for intervention efforts. Even so, the high prevalence of fuel combustion sources across all U.S. households emphasizes the need for more questions about indoor air quality in the AHS and other national surveys to more accurately quantify residential exposure.
Mold and damp environments were reported by approximately 18.7 million (15.9%) households, particularly water leaks from the roof, basement, and pipes, and mold in bathrooms. Mold spores can enter the indoor environment through building openings (e.g., doorways, windows, cracks, HVAC systems) and thrive in damp areas with excessive moisture, leaks, and flooding events [18, 54]. Mold triggers allergic symptoms, eczema, respiratory infections, asthma, dyspnea, and other pulmonary diseases [18, 59, 60]. Given that U.S. households spend approximately 87% of their time indoors [61], the risk of chronic exposure to these residential hazards are high, and particularly in the wintertime when the building envelope is more sealed [62].
Our study found that the HEQI had good discriminant and criterion validity to capture unique dimensions of housing and environmental quality. The inverse correlation of household fuel combustion with the building leakage domain reflects known trade-offs in indoor air quality and energy efficiency. In homes with frequent combustion-source activities (e.g., smoking, cooking, or candle/incense use) and without proper ventilation controls, building airtightness can trap air pollutants resulting in higher indoor concentrations. At the same time, high building leakage increases the risk of dampness, mold, pest problems, and energy loss [63,64,65,66]. Structural deficiencies can also lead to moisture, mold, and physical injuries; energy loss resulting in lower indoor temperatures; and openings for pests. As such, we observed positive correlations between domains affected by building structural integrity, such as dampness and mold, low indoor temperatures, lead paint risk, and injury hazards. Surprisingly, dampness and mold was not correlated with building leakage. This may be due to spatial imprecision of the building leakage indicator, which was based on regional U.S. estimates, or its coarseness as a binary indicator.
Furthermore, the HEQI was associated with household characteristics such as unit rating, year built, and rent costs. Unit rating is a consumer rating index capturing residents’ perception of well-being and quality of life [67]. In our study, risk factors strongly associated with lower unit satisfaction generally included those that residents were able to directly observe or experience, such as pests and allergens, lead paint risk, and injury hazards, consistent with a previous study [13]. Older housing is a known risk factor for physical deficiencies and chemical hazards [68]. While we could not evaluate chemical hazards, we found that older housing was strongly associated with physical deficiencies like inadequate water and sanitation, higher building leakage, and high indoor temperatures attributed to no central air or window air conditioning units. Rent costs is a market value index that assigns a monetary value to the quality of housing and neighborhood amenities, with higher rents suggestive of better quality [48]. In our study, the modest associations between HEQI domains and monthly rent costs could be due to the omission of neighborhood amenities from our analyses [48]. However we still found significant negative associations with rent costs for four HEQI domains of pests and allergens, low indoor temperatures, injury hazards, and high building leakage. Additionally, severe crowding was significantly associated with higher rent costs, consistent with previous findings that cost-burdened residents doubled-up to save on rent [2, 69, 70].
Our study also yielded findings that inform areas for future research. Household fuel combustion and high building leakage were positively associated with higher unit satisfaction. Since we did not have direct measures of indoor air quality or building leakage, we used proxy measures such as cooking and heating appliances and building features (e.g., unit size and height, basement and foundation type, year built). In effect, the positive associations with unit satisfaction may reflect residents’ preferences instead of (or despite) an understanding of the potential health risks. Indoor air pollution levels and building ventilation are generally difficult to observe without the assistance of sensor technology [71]. In addition, both of these domains have not traditionally been included in housing quality indices. Our findings underscore the need for further education among residents and housing practitioners about the sources of and strategies for reducing indoor air pollution and improving building ventilation.
We also found positive associations of inadequate water and sanitation with unit satisfaction and year built, which was primarily driven by the high proportion of households with non-public drinking water sources such as individual wells (7.5%) (exclusion of this item switched the coefficient direction of this domain to be negative, data not shown). Non-public drinking water sources are not regulated by the U.S. Environmental Protection Agency under the Safe Drinking Water Act [72] and have been associated with a higher risk of waterborne illnesses [43, 44, 73]. The positive association between non-public water sources and higher unit satisfaction may be attributed to suburban housing status, since suburbs that have a higher percentage of newer construction [74]. Unfortunately, we were not able to investigate this given the lack of information about urban/suburban status in the public AHS data in recent survey cycles.
Next, we evaluate the utility of the AHS to capture healthy housing domains recommended by the WHO and make recommendations for areas of improvement. Five domains were not captured by the AHS due to the lack of data across survey years: radon, pesticides, asbestos, noise, and housing accessibility. These housing and environmental risk factors have been widely associated with adverse health effects (Table 1) [1] and should be ascertained in future national surveys. The four HEQI domains of dampness and mold, low indoor temperatures, household crowding, and inadequate water and sanitation were well-captured by the AHS and should be continued in future surveys to allow for longitudinal assessments. The remaining six HEQI domains, household fuel combustion, lead paint risk, pest and allergens, high indoor temperatures, injury hazards, and ventilation, were roughly approximated and likely imprecise. These domains could be improved with more questions added in future surveys.
In particular, the household fuel combustion domain could be improved with more direct questions about the frequency and intensity of source activities such as cooking, heating, smoking, and candle and incense use that can contribute to higher indoor air pollution concentrations [6, 17]. Questions about appliance efficiency, furniture and flooring types, kitchen size, and the types of ventilation controls like kitchen and bathroom exhaust fans are also important determinants of indoor air quality. Despite the potential imprecision of our household fuel combustion measure, studies have found an independent association between the presence of gas stoves and higher nitrogen dioxide concentrations [28,29,30,31]. In addition, the majority of U.S. households have access to piped or bottled natural gas fuel, with the number of natural gas consumers increasing since the 1980s [75]. As such, the risk of exposure to gas combustion by-products, particularly from cooking, could be even higher. Based on the 2019 AHS, approximately 70% of U.S. households consumed natural gas, with 57% of these households using it for cooking [76]. More homes may opt for gas cooking appliances in the future. In addition, there is no federal requirement for mechanical kitchen ventilation in residential spaces [31], and even in homes with stove exhaust vents or range hoods, the quality and the degree of use during cooking events are highly variable [77,78,79]. Similarly, an experimental evaluation of an enclosed wood fireplace still found elevated particulate concentrations emitted into the living space [32]. Therefore, the use of a surrogate measure based on cooking and heating fuel and appliance type provides a baseline estimate of potential households at risk for indoor air pollution exposure.
In addition, our estimate of lead-paint risk at 1.7% (1.95 million) of U.S. households is likely a conservative estimate of the proportion of households with lead-based paint hazards. We used a two-fold criteria based on whether the home was built prior to 1980 and the presence of peeling paint size 8 × 11-inches or larger. This latter criteria is likely too stringent because lead-based paint can peel and crack at smaller sizes and crumble into dust [38]. To better capture lead-paint risk, future surveys should consider adding response options for smaller surface areas of peeling/cracked paint or dust. In addition, our estimate likely underestimates the risk of lead exposure overall given the lack of AHS data on other residential sources of lead, such as in soil, dust, and drinking water. Prior field-based studies using residential dust wipe samples, paint measurements, and soil samples to measure lead-based paint hazards in U.S. housing found a much higher prevalence of at-risk households: 35% (38 million) in 2000 [80] and 22% (23.2 million) in 2005–2006 [81]. The later study also found that 93% of homes with lead-based paint were built before 1978 [81]. In our study, 53% of occupied housing units were built before 1980. Based on these findings, it possible that the more accurate estimate of U.S. households with lead-based paint risk may be between 1.7% and 53%. Our lower estimate could also reflect the turnover in old housing stock nationally from renovations and newer construction [74, 82], which may have also contributed to decreasing blood lead concentrations in the U.S. population over time [83].
For the pests and allergens domain, the AHS only asked about the presence and frequency of rodents and cockroaches. Future surveys should consider other pests and allergens such as bed bugs and pet dander [19]. For the injury hazards domain, only information about electrical hazards and building integrity were collected in the AHS across multiple survey years. Information about smoke and carbon monoxide detectors, stairs and window railings, pool safety, and chemical storage were asked in previous AHS cycles but discontinued in recent years or only asked among a sub-sample of households. Going forward, these questions should be asked on a routine basis and among all households. The domain for high indoor temperatures was inferred from AHS items on central air and window air conditioning. High indoor temperatures could also be influenced by ambient temperatures and humidity, which could not be ascertained in the AHS [41, 84]. Future surveys should include direct questions about heat stress (e.g., unit was uncomfortably hot for 24 + hours) and usual temperature in the home. The ventilation domain should also include more questions about the types and performance of natural and mechanical ventilation controls (e.g., bathroom and kitchen exhausts, number of doors and windows, frequency of window opening). In addition, information about climate conditions, basement foundation type, and weatherization are needed to more accurately estimate building leakage. Lastly, routine data collection about energy efficiency (e.g., insulation, solar panels, Energy Star ratings) is important to track cost-savings and understand adaptation strategies to address climate change.
The AHS data is also subject to limitations common to national surveys. AHS items were self-reported and may be susceptible to recall error or social desirability bias. The AHS survey design is based on a federally-sponsored in-person and telephone survey, which may underestimate households in precarious or temporary housing arrangements. These issues could impact the precision of our findings and/or underestimate the prevalence of the hazards identified. Lastly, the AHS is conducted predominantly in English and Spanish languages (95% of households in the 2019 AHS data). Findings may not be generalizable to the small proportion of U.S. households speaking other languages. In spite of these limitations, a major strength of the HEQI is its accessibility for widespread adoption. The HEQI is based on AHS data that is nationally-representative, publicly-available, and collected biennially by the U.S. Census Bureau. In addition, AHS items in the HEQI are available across survey cycles since 2011 and asked of all occupied households. Therefore, the HEQI can be used in longitudinal analyses to evaluate HEQI trends across U.S. households.