Socio-economic, infrastructural and health-related risk factors associated with adverse heat-health effects reportedly experienced during hot weather in South Africa

Introduction Poor urban communities are likely to bear the brunt of climate change impacts on health and well-being. The City of Johannesburg, South Africa, is predicted to experience an average increase in ambient temperature of 4°C by 2100. Focusing on the urban environment, this study aimed to determine socio-economic, infrastructural and health-related risk factors for heat-related adverse health effects. Methods This was a cross-sectional study. Data of interest were collected using a pretested and validated questionnaire administered to parents of children attending schools participating in a school heat study. Information related to demographic, socio-economic and household-level determinants of health, which has an impact on the individual prevalence of adverse heat-health effects associated with hot weather, was collected for 136 households and 580 individuals. Results Sweating (n = 208 individuals; 35%), headache and nausea (n = 111; 19%) and weakness, fatigue and dizziness (n = 87; 15%) were the most common heat-health effects reportedly experienced by individuals (n = 580) during hot weather. Individuals who suffered from hypertension (OR = 2.32, 95% CI: 1.34 - 4.05, p = 0.003) and individuals older than 60 years (OR = 1.81, 95% CI: 1.27-1.99, p < 0.001) compared to other age groups were more likely to experience 'any heat-health effects'. Living in government-sponsored detached housing and in houses with asbestos roofs were associated with an increase in reported experience of 'any heat-health effects' compared to living in other housing types. Conclusion Heat-health awareness campaigns should target people suffering from pre-existing diseases and the elderly, as these groups are especially vulnerable to heat. Focus should also be given to appropriate roofing and insulation in government-sponsored housing since summertime temperatures are projected to increase.


Introduction
In a changing climate, many parts of the world are experiencing steadily increasing temperatures [1]. While much uncertainty about climate change remains, the few studies of large, low-and middleincome country populations (specifically in the tropics) suggest possible effects of heat, and less so effects of cold, on mortality [2].
In some parts of the world, climate change is exacerbating health problems by increasing risks of morbidity and mortality caused by intense heat episodes [3]. Those groups most vulnerable to these impacts tend to be indigent communities living in high-density urban settings [4], as well as children [5], the elderly [6], people with disabilities [7] and pre-existing diseases [4] and families from lower socio-economic groups [8]. Africa is projected to see significant increases in the number of days when health may be adversely affected by increasing maximum temperatures due to climate change [9]. These increases in Africa may occur at a rate faster than elsewhere [10,11]. The ability of people to acclimatise to relatively rapid and sustained change in temperature, such as those predicted for climate change, is mostly unknown [12,13]. People are likely to experience heat-related health impacts including discomfort, fatigue and heat stroke [14]. The association between high temperature and increased mortality from all-cause mortality, as well as cardiovascular, respiratory and cerebrovascular diseases, has been demonstrated internationally and in parts of Africa [15][16][17]. In northern Ghana, for example, all-cause mortality was statistically significantly associated with mean daily temperatures above the 75th percentile (the threshold applied in that study) [18]. Very little is known about heatrelated morbidity in Africa. Some work to understand possible heathealth symptoms and impacts has been done in South Africa. Indoor temperatures measured in waiting rooms in rural clinics indicated that maximum daily temperatures exceeded 38°C [19]. The difference between indoor and outdoor temperatures was between 2-4°C on average with warmer termperatures experienced in the clinic waiting rooms. 'Real-feel' temperatures, that include humidity measurements, were 4°C higher than measured indoor temperature. This suggests that patients waiting for treament may have experienced a feeling of 'stuffiness' and discomfort [20]. A study conducted in the City of Johannesburg (CoJ) found that students (n = 252) aged ~14-18 years felt tired and found it hard to breathe when classroom temperatures ('real-feel') were greater than 32°C [21]. Higher indoor classroom temperatures were consistently observed in classrooms constructed of prefabricated asbestos sheeting with corrugated iron roofs and converted shipping containers, compared to brick and other materials used in classroom construction. In South Africa, about 11% of the urban population live in informal settlements [22] where levels of poverty and overcrowding are usually high, and housing comprises mainly shacks constructed using metal and plastic sheeting. Other settings of urban poverty may comprise areas of inner-city degradation and government-built brick dwellings with basic amenities (but not necessarily ceilings). These residents are extremely vulnerable to variations in ambient temperature. Insight is needed into heat-health effects and associated risk factors, including those related to dwelling characteristics, to help plan for public health interventions to reduce heat-related morbidity and mortality during high temperatures and extreme heat events in Africa. Therefore, this study aimed to determine socio-economic, infrastructural and healthrelated risk factors associated with heat-health effects reportedly experienced during hot weather in an African urban environment.

Methods
Study site and data collection: this was a cross sectional study. usually eats with the remaining household members. Failure to complete section 3 was also considered an exclusion criterion. We assumed that all individuals living in the same dwelling were exposed to the same household risk factors.
Statistical analyses: statistical analyses were performed using STATA version 14.0 [24]. Bivariate logistic regression was conducted to assess associations between 'any heat-health effects' and risk factors [25]. These were (a) existing chronic conditions and (b) age group and gender. Bivariate analysis was used to examine associations between individual reported 'any heat-health effects' and living conditions / dwelling characteristics as risk factors [26]. Crude Odds Ratios (ORs) and their respective 95% confidence intervals (CIs) and p-values were calculated by means of Chi-squared tests and Wald Chi-Square (one-tailed and two-sided) tests. Variables were considered significant at the p ≤ 0.05 level.

Results
Individual heat-health effects: four hundred and eighty household questionnaires were distributed, and the overall parent / caregiver response rate was 45% (n=217). After applying the inclusion criteria, i.e. fully completed questionnaire sections 2 and 3 (see methods section above for full explanation), a total of 136 household questionnaires (28%) were eligible for inclusion, and data for 136 households and 580 individuals were evaluated. Primary household caregivers were asked about heat-health symptoms experienced by household individuals when the weather is hot (Table   1). These heat-health symptoms included sweating, weakness, fatigue and dizziness, and muscle cramps. For doctor-diagnosed chronic health outcomes experienced by household individuals as reported by the primary household caregiver, the highest prevalences were reported for hypertension and stress, followed by skin rash / dermatitis and 'any chronic injuries' ( Table 2).
Chronic health outcomes in relation to individual 'any heathealth effects': using the doctor-diagnosed health outcomes in Table 2, we combined any / all heat-health effects reportedly experienced by an individual to create a new variable 'any heat-health effects' reportedly experienced by an individual and considered chronic health outcomes and socio-demographic variables as potential risk factors.
Associations between 'any heat-health effects' and risk factors: age was a significant risk factor for 'any heat-health effects' (Table 3). Individuals older than 60 years were almost two times more likely to experience 'any heat-health effects' compared to individuals in other age groups (OR = 1.81, 95% CI 1.27 -1.99, p < 0.001).
Females were reportedly 1.15 times more likely to suffer from 'any heat-health effects' compared to males although this finding was not statistically significant (OR = 1.15, 95% CI 0.82 -1.62, p = 0.412).
The results of the bivariate analyses showed that hypertension was the only pre-existing chronic health condition that had a statistically significant association with 'any heat-health effects'. Individuals who suffered from hypertension were about two times more likely to experience 'any heat-health effects' (OR = 2.32, 95% CI 1.34 -4.05, p = 0.003).

Household living and dwelling characteristics in relation to
individual 'any heat-health effects': most families lived in  (Table 3). Most dwellings had a tiled roof followed by houses with a corrugated iron roof. Fourfifths of dwellings had a dedicated, separate kitchen for cooking.
Households mainly had electricity as their primary energy carrier for cooking and lighting followed by paraffin. Less than 2% of households used solid fuel, i.e. coal or wood, for cooking. Most households had in-house electricity and piped water. Almost all dwellings had a toilet inside the house or yard (asked together as one question). Less than 10% of individuals lived in households with air-conditioners. Living in a government-sponsored house was associated with an increase in 'any heat-health effects' (Table 3). Individuals living in dwellings with roofs made from asbestos sheets had the highest odds of reporting heat-health effects compared with people living in dwellings with other roof types, although this finding was not statistically significant. Hence, we focused on an urban African setting where special attention is needed to prevent adverse health impacts in a changing climate. We found that, for non-modifiable risk factors, individuals over the age of 60 years were reportedly most likely to experience 'any heat-health effects'. Several studies have shown that people above 60 years are at increased risk of heat-related illness because of physiological impairments in the regulation of core body temperature in hot conditions [17,28,29]. Females were reported to be more likely to experience any heat-health symptoms than men although this finding was not statistically significant. This was also found during the Euro HEAT project that considered the impact of heat waves on mortality in nine European cities [30]. Over the summers of 1990 to 2004 and in the year 2003, separately, the highest effect from heat on mortality was observed among women aged 75 to 84 years. Women may be more vulnerable to the effects of heat in African urban settings where they spend considerable time indoors caring for children, the elderly and looking after their dwellings. Regarding pre-existing diseases in relation to heat-health effects, our findings agreed with previous studies [28,31,32] where hypertension was a statistically significant risk factor for an individual to experience 'any heat-health effects'. Adaptation strategies need to be implemented to prevent heat-related morbidity and may include cooling the room in which the person is located, and possibly also changing medical treatment doses during hot weather [32]. African cities are renowned for having a mix of different housing types in terms of nature, design, multi-use (e.g. cottage industries) [33] and building materials, among other factors. Here, we found that living in RDP housing was associated with an increase in reported experience of 'any heat-health effects' compared to living in other housing types, e.g. flat / apartment and informal dwelling or shack. RDP homes built by contractors are often very small and of poor quality [34]. In a recent study, also in the CoJ, low-cost RDP houses had the greatest variation in indoor temperature and experienced indoor temperatures 4 to 5°C higher compared to ambient outdoor temperatures [35].

Discussion
These types of houses are typically built from bricks (either cement or a combination of sand and cement) with high thermal conductivity and an exterior wall of single bricks no more than 15 cm thick. They also do not usually have ceilings or thermal insulation in ceilings [36] thereby reducing their thermal comfort.
Despite not being statistically significant, the strongest effect for a modifiable risk factor in relation to 'any heat-health effects' was found for living in a dwelling with asbestos sheet roofing. Asbestos roofs can increase indoor temperatures due to an intense transfer of heat into the internal environment during hot weather [37,38]. In Sri Lanka, excessive heat transferred through the asbestos roof was one of the main causes of thermal discomfort in warm, humid climate conditions [39]. There, a cement-asbestos fibre sheeting roof compared to clay tile roofing resulted in warmer indoor temperatures.
While asbestos roof sheeting is banned in South Africa [40], it was widely installed in low-cost housing in the past and remains in place in some dwellings throughout the country. In a survey of 1 488 households in Soweto in the CoJ, 52% of dwellings had a corrugated asbestos roof [41]. In addition to the heat-health induced temperature effects of asbestos leading to warmer indoor environments, asbestos fibres when disturbed become airborne and are associated with asbestosis and various cancers [42].
Rehabilitation of asbestos roofing therefore needs to be addressed professionally and with care for residents' health. The study was constrained in design by using a sample made during another study and by asking the parent / primary caregiver to provide information about the house as well as all demographic and health information for every individual living in the house. We also assumed that the primary household caregiver understood the terminology used for heat-health effects and chronic health outcomes. We did not collect all potential confounders, nor did we detail severity of heat-health effects contained in the questionnaire. We also did not measure temperature inside any of the respondents' houses, nor did we specify hot weather events / hot weather referred to in the questionnaire when we asked respondents to explain which symptoms they themselves as well as members of their family experienced during such hot weather. We also asked one question about whether the toilet was inside the house or in the yard -this question should have asked about 'inside' and 'in yard' toilets separately given the public health consequences of sanitation locations. The small sample and poor response, likely due to the request for detailed information about each individual living in the household and a self-completed questionnaire format from the primary caregiver, resulted in relatively low disease prevalence and subsequent wide CIs despite statistically significant odd ratios, hence these risk factors should be tested in a larger sample.

Conclusion
We have highlighted specific risk factors that should be considered when trying to reduce adverse heat-health effects among vulnerable groups living in urban African settings. We have identified ways in which individuals and policy makers can act. Specifically, due to their vulnerability to heat, people with pre-existing diseases (especially hypertension) and the elderly should be targeted for heat-health information on ways to cope during hot weather. Our findings provide information that can inform the development of community heat and health adaptation awareness campaigns and prevention of adverse heat-related health impacts in African urban settings.
What is known about this topic  Climate change is occurring and causing temperatures to increase especially in southern Africa;  Heat-related health impacts are well understood in individuals working in occupational settings such as in underground mines. Less has been done in community settings and even less in the African context.  These findings build on the body of evidence needed in Africa to provide for evidence-informed policy advice towards heat-health adaptation plans.

Competing interests
The authors declare no competing interests.  Tables   Table 1: heat-health effects experienced by household individuals as reported by the primary caregiver for the question 'is there any member of your family who experiences the following symptoms / outcomes when it is hot?' Table 2: chronic health outcomes experienced by household members as reported by the primary caregiver for the question 'Has any member of your family ever been told by a doctor that they have any of the following illnesses?'