Housing conditions and long-term care needs of older adults in Ghana: Evidence from the WHO SAGE Ghana Wave 1

The present study examined the association between housing conditions and long-term care needs of older adults in Ghana. We used data from 4,920 adults aged ≥50 years that participated in the World Health Organisation’s (WHO) Study on adult health and AGEing Ghana Wave 1. Housing conditions were assessed with drinking water, sanitation, cooking conditions and building materials, and long-term care needs were based on WHO Disability Assessment Schedule 2.0. Multivariable logistic regressions modelled the effect of housing conditions on long-term care needs. After full adjustment for all available potential confounders, older adults living in households with unimproved cooking conditions had higher odds of reporting long-term care needs (OR = 6.87, 95%CI: 5.04–9.37) compared to those in improved cooking condition households. Moreover, those in households with unimproved housing materials (OR = 1.27, 95%CI: 1.01–1.72) and those in unimproved sanitation households (OR = 1.26, 95%CI: 1.05–1.54) were more likely to experience long-term care needs after respectively controlling for demographic and health-related covariates. Poor housing conditions are risk factors of long-term care needs in Ghana. Efforts to improve housing conditions may benefit older age functional abilities and unmet long-term care needs.


Introduction
Robust geriatric assessment of the care needs of older adults requires an understanding of a person's physical and mental health status and their context, including the housing conditions. The United Nations Decade of Healthy Ageing (2021-2030) aims to prolong older adults' life expectancy with good health by addressing four areas of action [1]. The first three action areas a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 association between housing and long-term care needs, or health are limited. A systematic review on effects of the use of kerosene on papers from low-and middle-income countries revealed that kerosene use poses health risk to users including older adults [20]. In Ghana, a study reported that the use of unimproved water sources and unimproved sanitations are significantly associated with increased risk of depression, particularly among women. 37 The current study extends the available evidence in low-and middle-income countries by exploring the association between housing conditions and long-term care needs of older adults in Ghana.
Regarding the confounding variables, available evidence shows that wealth (measured in income) [21,22], advanced aged [23,24], sex, education, multimorbidity, and marital status [24] influence older adults' housing status and long-term care needs. Accordingly, this study examined the relationship between housing conditions and the long-term care needs of older adults, controlling for potential demographic and health-related factors. This study hypothesised that housing conditions will have an association with long-term care needs.

Study sampling
We used a nationally representative data from the Study on global AGEing and adult health (SAGE) Ghana Wave 1 conducted between 2007/2010 across all the 13 regions of Ghana (S1 Data). The sample size recorded in the study was 5,571 participants (aged 18 years and older); however, this study used a sample of 4,920 participants who responded to all 12 questions on functional disability, which served as a proxy for long-term care need for our analyses. Additional details about the study methodology and other relevant information can be found elsewhere [25]. In addition, we merged the household dataset with the individual dataset for inclusion of housing variables, including water supply source, sanitation, cooking conditions, and housing construction materials, used to define 'improved housing'. Ethical approval for this study was obtained from the WHO Ethical Research Committee.

Long-term care needs
Long-term care needs were defined by the 12-item version of WHO Disability Assessment Schedule (WHODAS) 2.0, originally measured using five ordinal scales (none, mild, moderate, severe, and extremely severe). Disability is considered the best proxy for measuring long-term care because evidence shows that disability necessitates older adults' long-term care needs [26]. The WHODAS 2.0 contains 12 questions from all six domains of the full version, namely: cognition, mobility, self-care, getting along, life activities, and participation in society [27]. S1 Appendix contains the 12 questions included in the analysis. We scored the WHODAS 2.0 on a scale of 0 to 100, with a lower score implying lower disability and a higher score, high disability. Similar to other published studies [10][11][12] we decided to take the top 10 th percentile as our cut-off point for determining the severity of the disability. That is, scoring <90.18% were considered "no disability" and "> = 90.18%" were considered as "with a disability" [11,12]. In our study, we used the disability score to determine the long-term care needs of older adults. That is, those "with a disability" > = 90.18%" disability score) were considered as "needing longterm care". Long-term care, in this case, will include assistance with personal care, household chores as a result of limitations experienced by individuals spanning from a longer period. Using this cut-off point is understandable as individuals reporting more than moderate to extremely severe on any of the functional activities usually require some form of care [15]. Individuals reporting "no disability" <90.18% disability score) were considered "not needing long-term care". It is not unusual for researchers to use disability severity as a proxy for measuring long-term care needs. For instance, a report by WHO measured long-term care needs by considering the limitations on the five components of the WHO-ICF including body functions and structures, activity limitations as well as participation restrictions [1].

Independent variables
Housing quality. Housing quality was measured using four variables, including questions about sanitation, water supply source, cooking conditions, and housing construction materials. As defined by the WHO Joint Monitoring Programme (WHO-JMP) [28], when all four variables met the criteria for improvement (detail given below), overall housing quality was considered as improved [29,30]. In this study and similar to Awuviry-Newton, Wales, Tavener, Kowal, Byles [14], we decided to measure the effects of the four separate variables on the longterm care needs of older adults, rather than using the overall measure of housing quality. This technique is relevant because it enlightens the unique impact and quantum of association on long-term care. Definitions of improved are provided for each of the individual components below.

a. Sanitation
In SAGE Ghana Wave 1, sanitation was measured through the type of toilet facility and whether the toilet is shared as categorised by the WHO-JMP [16]. The question "what type of toilet facility do members of your household usually use?" was used to determine "toilet facility type". Based on the WHO-JMP categorisation, 'improved sanitation' included: 1) flush/pour-flush to a piped sewer system, flush/pour-flush to a septic tank, flush/pour flush to pit latrine, flush/pour flush to other location, flush/pour flush to unknown place/not sure, ventilated improved pit latrine, pit latrine with slab, and composting toilet. On the other hand, households with the following types of toilets were considered 'unimproved sanitation': pit latrine without slab/open pit, bucket latrine, hanging toilet/hanging latrine, no facilities or bush or field. The question, "Do you share this facility with other households?" was used to assess the 'toilet shared' expressed in dichotomous response (yes or no). When participants responded 'no', the variable was defined as 'improved' whereas 'yes' was defined as 'unimproved'. Sanitation (improved vs unimproved) was measured by combining toilet shared (improved vs unimproved) and toilet facility type (improved vs unimproved). Combining these two variables, sanitation was considered as "improved" if the participant reported improved for toilet shared and toilet facility type. On the other hand, sanitation was considered unimproved if the participant reported unimproved in at least one of the two variables.
b. Drinking water source To determine the water supply source, the question, "What is the main source of drinking water for members of this household?" was asked, with 13 possible response categories. "Improved water supply" included piped water into dwelling, piped water to yard/plot, public tap/standpipe, tube well/borehole, protected dug well, protected spring, hygienic source bottled water, and rainwater collection. Unimproved water supply included unprotected dug well, unprotected spring, small scale vendor, tanker-truck, surface water (river, dam, lake, pond, stream, canal, irrigation channels) [16].
c. Cooking conditions Three variables were used to assess the quality of the cooking conditions in the household, including a question about where cooking takes place, what the food was cooked on, and cooking fuel used to cook food. Where cooking took place was measured with the response categories "cooking in the living or sleeping room", "outdoor", and "in a separate room".
Cooking in the living or sleeping room and outdoors was considered "unimproved", and cooking in a separate room was considered "improved". The second question was, "In this household, is food cooked on an open fire, an open or closed stove?". Again, we considered "open fire" as "unimproved" and "open stove" and "closed stove" as "improved". Lastly, the question "What type of fuel does your household mainly use for cooking?" "Unimproved" included coal/charcoal, wood, agriculture/crop, animal dung, shrubs/grass, while "improved" included gas, electricity, and kerosene/paraffin. Cooking conditions were therefore classified as "unimproved" when participants reported unimproved in at least any two of the three variables whereas "improved" referred to the opposite [16].

d. Construction material quality
Two questions were included about the floor and wall type to measure the quality of construction materials. Floor type was measured with the question "What type of floor does your dwelling have?' categorised as "hard floor" (tile, cement, brick, wood) and "earth floor".
Hard floor" was considered as "improved" and "earth floor" as "unimproved". For the wall type, responses to the question, "What type of wall does your dwelling have?" included 1) cement brick, stone, or wood; 2) mud/mud brick; 3) thatch and other; 4) plastic sheet; 5) metal sheet. Per the categorisation of WHO [16], we categorised them as "unimproved" (mud/mud brick; thatch and other; plastic sheet, metal sheet) and "improved" (cement, brick, stone, or wood). Construction material was considered as "unimproved" if participants reported unimproved in at least one of the two variables". Otherwise, we considered it as "improved" if we categorized both variables as improved.

Confounding adjustment.
A set of socio-demographic and health-related variables was considered as confounding factors. The socio-demographic characteristics considered were the age of participants (in years), education (no education, maximum junior high completed, at least senior high achieved), marital status (single, separated, divorced, married, cohabiting, widowed), sex (male, female), location of residence (rural/urban) and relative wealth (measured in quintiles (Q1-Q5) [31]. Relative wealth was measured using household assets and possessions [31,32]. Quintile 1 refers to the household with the poorest states, whereas Quintile 5 referring to the richest household.
The health variables used as covariates in the analyses were self-reported conditions compiled into multi-morbidity status (no chronic conditions, single condition, more than two conditions). Chronic conditions included in this analysis were stroke, arthritis, angina, diabetes, chronic lung disease/asthma, hypertension, cataract, oral health, and injuries. Body mass index (BMI) was also categorised as underweight (<18.5 kg/m 2 ), normal weight (18.5-24.9 kg/ m 2 ), overweight (25.0-29.9 kg/m 2 ) and obese ( 30.0 kg/m 2 ).
Analysis. Descriptive analyses, including frequency and percentages, were used to describe the categorical variables, whereas mean and standard deviation were used for the continuous variable, particularly age. Chi-square and t-test were used to test the relationships between independent variables and the generated dependent variable (long-term care needs). Bivariate and multivariate logistic regressions were performed to estimate the crude and adjusted Odds Ratios (OR) and 95% confidence intervals (CI) for the associations between housing variables and long-term care needs. All variables that were at p<0.05 on the bivariate analyses were included in the multivariable logistics regression model. We conducted a multivariate logistic regression adjusting for all available potential confounders to examine whether the associations between housing variables and long-term care needs were independent of sociodemographic and health factors. We developed four models of logistic regression. Model 1 -unadjusted relation between housing quality variables and long-term care needs; Model 2 -adjusted for sociodemographic variables (age, sex, marital status, location of residence, education, relative wealth; Model 3 -adjusting for health variables; and Model 4 -adjusting for all socio-demographic and health related variables at a p-value<0.20. STATA 16 was used as a statistical software package for the analysis.

Ethical consideration
Ethical approval for this analysis was obtained from the World Health Organization Ethical Research Committee. Written informed consent was obtained from all participants.

Descriptive statistics
Bivariate analysis of independent variables in relation to long-term care needs is presented in Table 1. Most participants were men (53.7%), with a mean age of 60 years. More than half were married/cohabiting (60.4%), lived in a rural area (59.3%), and 48.2% had no formal education. Nearly the same proportion of participants was distributed across the income quintiles, with a little over one-quarter reporting the highest income quintile (20.6%). About 68% reported having no chronic conditions, and nearly half had a normal BMI (55.7%). Nearly 12% of participants reported needing long-term care, with prevalence relatively significant.
Comparatively, very few of the participants were living in unimproved housing conditions: 17.6% lived in households with unimproved water source, 32.2% lived in households with an unimproved sanitation, and 13.5% with unimproved housing construction materials. Over 93% had unimproved cooking conditions (93.4%).
Health-related variables and all demographic variables except the location of residence were associated with long-term care needs. Older women reported high long-term care need (61.7%) compared to their older men counterparts. Demographic characteristics, such as being widowed (21.7%), having no formal education (17.1%), being in the middle-income quintile (59.7%), living with at least two chronic conditions (25.5%), and being underweight (18.5%) reported higher long-term care needs compared to their respective counterparts. A slight variation of prevalence of long-term care needs existing between households with unimproved water supply and households with an improved water supply source (11.9% vs 11.7%). Likewise, a relatively high prevalence (12.3%) of long-term care need was found among participants living in households with unimproved sanitation. A positive association was found between unimproved cooking conditions and long-term care needs. participants living in housing constructed with unimproved housing construction materials had a relatively high prevalence of long-term care need (12.0%). The prevalence of long-term care needs in relation to specific variables across the 12-items are relatively high (Table 2). Table 3 shows the multiple logistic regression analysis assessing the association between housing variables and long-term care needs. Although in model 1, the relationship was not significant, in model 2, after adjusting for significant socio-demographic variables, the relationship between the unimproved water supply and long-term care need remained statistically insignificant (p = 0.672). Similarly, in terms of relationship, after adjusting for health variables including BMI and multi-morbidity in model 3, an insignificant relationship between unimproved housing construction materials and long-term care needs was noticed. In the parsimonious model 4, after adjusting for significant variables from models 3 and 4, there was a slight  (Table 3). Participants living in households with the unimproved sanitation were 9% more likely to have long-term care needs in model 1. After controlling for socio-demographic variables, the relationship remained statistically insignificant (AOR, 1.15; CI: 0.93, 1.42). Adjusting for health variables in model 3, participants living with unimproved sanitation were 26% more likely to have long-term care need with a statistically significant relationship (p<0.01). When adjusted for all significant cofounding variables, participants living in households with us were 22% more likely to have long-term care need compared to their counterparts (p = 0.07).

Housing and long-term care needs
Participants in households with unimproved cooking conditions were 419% more likely to report long-term care need than those living in households with improved cooking conditions (p<0.001) in model 1. When adjusted for socio-demographic variables in model 2, the relationship remained significant at p<0.001 (AOR, 6.59; CI: 4.86, 8.94). When adjusted for health variables in model 3, participants in households with unimproved cooking conditions were 459% more likely to have long-term care need compared to that reported improvement (p<0.001). In model 4, when adjusted for all significant socio-demographic and health variables, a significant independent relationship was maintained between unimproved cooking conditions and long-term care need (AOR; 6.87, CI: 5.04, 9.37).
Participants living in households with unimproved housing construction materials were 23% more likely to have long-term care need (p = 0.346) in model 1. When we adjusted for the significant socio-demographic characteristics in model 2, participants living in households

Discussion
This study examined the relationship between housing conditions and long-term care needs in Ghana using data from the Study on global AGEing and adult health (SAGE) Wave 1.
The results provide initial information about the relationship between housing and health and point to specific housing areas to address and subsequently be used to moderate the long-term care needs of older adults in the country. The current study has begun a debate on an innovative area of research by focussing on the relationships between housing and long-term care needs. The prevalence of long-term care needs found in this current study was nearly 12%. The prevalence seems low in this current study but it is likely to be high as the Ghanaian population grows. Overall, housing conditions except water supply source were associated with long-term care needs in models that adjusted for health variables but not for socio-demographic variables. This finding implies that health factors better moderate the relationship between housing and long-term care needs. Health-related factors determining the significant association between housing and long-term care needs, as revealed in this study, are not surprising because older adults living with multiple chronic conditions will have more functional disabilities [14,33]. that may increase their needs for long-term care. While water is considered a vital basic necessity, the plausible explanation for its non-significant association with LTC needs could be that it may be the least prioritized housing condition for the sampling units. Sanitation was associated with long-term care needs when adjusted for health variables as opposed to demographic variables depicting that health-related factors may offer a better explanation on the relationship between housing (sanitation) and long-term care needs among older adults in Ghana. The finding that multi-morbidity and underweight led to a significant relationship between housing and long-term care needs increase our understanding of how the relationship may be health-related. This finding further implies that older adults living in poor sanitary conditions have a high need for long-term care as it harbours rapid transmission of diseases [34]. Evidence in Ghana shows that an appreciable number of older adults go to the toilet in the bush or fields, and some using open defecation around river bodies or bagged it in polythene during the day and throw it around the vicinity during the night [14,35]. The health of older adults is at risk should they remain in bad sanitary environments, which may increase the need for long-term care. Mandated state institutions should champion sanitation activities nationwide. Efforts geared towards ensuring that every household has an appropriate toilet and bath facility, as well as a proper waste disposal system, should be intensified as this may improve older adults' long-term care experiences.
We found evidence that unimproved cooking conditions associate with increased longterm care needs of older adults in Ghana, affirming the current evidence of the harmful impact of cooking in the living room, using charcoal and wood for cooking on older adults' health [36]. Cooking in household spaces like the living room instead of a kitchen, could expose older adults to serious air pollution and suffer respiratory complications [37,38]. It is significant to encourage the use of specified kitchens and modernized cooking apparatus like a gas stove in a household with older adults. A social services initiative specialised to ensure the affordability of closed stoves could increase the patronage, thereby reducing their long-term care need. A specialised strategy by stakeholders including governmental and non-governmental organisations to improve the cooking conditions of households with older adults can contribute to meeting older adults' long-term care needs in Ghana. A long-term care policy that can specifically provide access to caregivers of older adults with modernised cooking stoves either free or at reduced cost will help promote enhancing their ageing experience.
Housing material was associated with long-term care need when adjusted for socio-demographic factors implying that socio-demographic factors such as advanced age and income affect the relationship. The quality of the housing environment, the physical nature of the housing, and the presence of vital amenities are relevant factors to be considered in assessing a household's housing condition [39]. Nonetheless, building materials in Ghana are generally expensive relative to the income of many older adults [40,41], compelling many to purchase poor building materials that are likely to deteriorate faster. Additionally, when poor building materials are used in constructing a house, it could pose a threat to the safety of older adults. For instance, houses built with mud could have damp conditions, which may cause cracks in the wall [37]. Therefore, it would be useful if older adults are assisted in having access to proper building materials that would enable them to construct elderly-friendly houses to make their living comfortable. A voluntary amount of money can be contributed towards older adults' savings (irrespective of their employment status) by the state to cater to their housing needs, especially during the later years of life.
To the best of our knowledge, this study is the first to use a countrywide sample to examine the relationship between housing conditions and long-term care needs among older adults in Africa and developing countries. The current study has begun a new area of research interest into the association between housing and long-term care needs. The present study raises a new question that requires further examination. Specifically, how do the association of housing condition and long-term care needs differ according to gender, age, and chronic conditions? On a positive note, the current study provides baseline information towards the improvement of housing conditions that may enhance long-term care experiences among older adults.

Limitations of this study
Some limitations of this study need to be acknowledged. The first limitation is that we used functional disability to measure long-term care needs, which could have been measured from self-reported health, which elucidates our understanding of how older adults yearn for longterm care. Second, the approach we adopted to categorise responses for long-term care may be misclassified; however, we ensured to also classify those reporting needing no care and those needing mild care. Although the current study suggests demographic and health impacts of the association between housing conditions and long-term care needs, it did not examine how gender and chronic conditions influence the association. Moreover, the research did not capture the causal effect of the association between housing and long-term care needs.

Conclusions
The current study provides a baseline finding for older adults' housing and long-term care needs in this decade of healthy ageing. This research provides a basis for policymakers to focus attention on practical housing policies and programmes to ensure improved housing conditions to safeguard the health and well-being of older adults. Further studies on housing needs of older adults to their long-term care may benefit from longitudinal analysis and qualitative data to inform policymakers' understanding of the need to care for older adults.