Interplay of gender, environmental sustainability, and cognitive impairment among older adults in Ghana: insights from the 2021 population and housing census

Abstract

Background

Cognitive impairment, which includes difficulties with memory and concentration, poses a growing challenge to healthy ageing, particularly in low- and middle-income countries. In Ghana, the prevalence of cognitive impairment is significant, yet the combined associations of key social and environmental determinants remain under-researched. This study investigated how sex and environmental sustainability factors are associated with cognitive impairment among older adults in Ghana.

Methods

This study analysed pooled data from the 2021 Ghana Population and Housing Census (GPHC), focusing on a sample of 32,486 adults aged 60 years and older. Cognitive impairment was assessed based on self-reported difficulties in remembering or concentrating. The primary independent variables were sex and environmental sustainability factors, including household sources of drinking water, toilet facilities, methods of solid waste disposal, cooking fuel, and cooking space. Descriptive statistics, chi-squared tests, and hierarchical multivariate logistic regression were employed to examine associations, with results presented as adjusted odds ratios (AORs) with 95% confidence intervals (CIs). The outcome was coded as “no cognitive impairment” to align with a health assets/strengths-based perspective, facilitating interpretation of protective factors.

Results

The analysis revealed that older women were significantly less likely to be free from cognitive impairment compared to their male counterparts (AOR = 0.61, 95% CI = 0.55, 0.69). Regarding environmental factors, the use of clean cooking fuel (LPG, biogas, or electricity) was independently and significantly associated with a higher likelihood of no cognitive impairment (AOR = 1.41, 95% CI = 1.23, 1.61). While access to improved water (AOR = 1.16, 95% CI = 0.87, 1.55), improved toilet facilities (AOR = 1.08, 95% CI = 0.97, 1.20), improved waste disposal (AOR = 1.14, 95% CI = 0.98, 1.32), and improved cooking space (AOR = 1.01, 95% CI = 0.91, 1.12) showed positive point estimates, none of these associations reached statistical significance. These null findings should not be interpreted as evidence of no effect; rather, they highlight the need for longitudinal studies with more detailed environmental exposure measures to fully understand the relationship between WASH factors and cognitive health in this population.

Conclusions

This study suggests that cognitive health in later life is shaped by modifiable structural factors. The pronounced sex disparity, with older women at higher risk of cognitive impairment, underscores the need for gender-sensitive health strategies that address cumulative socioeconomic disadvantages. Furthermore, the significant protective association found for clean cooking fuel highlights household energy transitions as a promising intervention for cognitive preservation. These findings advocate for integrating healthy ageing objectives into Ghana’s gender equality and environmental sustainability agendas.

Clinical trial number

Not applicable.

Background

Cognitive impairment in older adults refers to a decline in memory, thinking, and decision-making abilities, ranging from mild cognitive impairment (MCI) to severe dementia [1, 2]. Around 55 million individuals worldwide had dementia in 2021, accounting for 5–8% of those over 60 years [3]. By 2050, that number is expected to rise to 139 million. Dementia prevalence among older persons in Africa is 2–5%, and it is predicted to increase dramatically with population ageing [3]. In Ghana, the dementia rate among persons over 60 years is 6.3%, and the country’s ageing population and poor healthcare access are anticipated to contribute to an increase in cognitive impairment [4]. Of older persons worldwide, 10–20% experience mild cognitive impairment (MCI); comparable developments are anticipated in Africa and Ghana [4, 5].

As cognitive impairment worsens, it significantly reduces quality of life and increases dependence on long-term care, but information about the subject is limited [5, 6]. Understanding cognitive impairment among older adults, especially in low- and middle-income countries like Ghana, is vital for developing effective strategies to address these challenges within resource-limited healthcare systems.

Access to healthcare is a significant challenge, particularly in rural areas with poor road networks [4]. Many older adults experience social isolation, neglect, and barriers to healthcare, such as expired health insurance or long wait times, further undermining their well-being [7]. Around 58% of the population resides in urban areas, where rapid growth in cities like Accra and Kumasi has strained infrastructure and environmental quality [8, 9].

Previous studies reveal that environmental sustainability is crucial for older adults vulnerable to poor air quality, unimproved sanitation, and waterborne diseases [10–14]. Transitioning to improved/renewable energy and waste management can reduce the health risks associated with pollution and poor sanitation [14–17]. Older adults in Ghana often live in inadequate housing with poor sanitation, lack of clean water, and unsafe cooking conditions, which worsen health problems and increase the need for care [18–20].

Gender also plays a critical role in mental health outcomes among older adults [21]. Studies have shown that gender disparities in access to healthcare, resources, and social support have significant effects on cognitive health [22–24]. Women have greater risk of depression and anxiety due to caregiving burdens and hormonal changes during menopause [25]. However, they tend to have stronger social networks, which can mitigate cognitive decline, though caregiving responsibilities may add stress [26]. Men, in contrast, often face greater social isolation, negatively affecting their mental health [27]. At the same time, men face higher risks of substance abuse and suicide, aggravated by societal norms discouraging help-seeking [27]. Social and economic factors are associated with cognitive health: women often experience financial instability due to lower lifetime earnings, while men struggle with social isolation, particularly post-retirement [28–30].

Despite growing evidence linking environmental exposures to cognitive health, most studies originate from high-income countries with different exposure profiles and healthcare systems, limiting generalizability to sub-Saharan African settings. Also, existing research has typically examined environmental factors in isolation, without considering their combined effects or interactions with social determinants such as sex. Further, the specific mechanisms through which household environmental conditions—particularly cooking fuel, water quality, and sanitation are associated with cognitive outcomes in later life remain poorly specified in the African context. This study advances beyond prior work by: (1) using nationally representative data from Ghana, (2) simultaneously examining multiple environmental sustainability factors, and (3) testing sex-specific associations while accounting for a comprehensive range of confounders.

In this study, environmental sustainability factors are operationalized as household-level environmental conditions that are modifiable through policy and infrastructure investments. Specifically, these include; drinking water sources (improved vs. unimproved, based on WHO/UNICEF JMP classifications), toilet facilities (improved sanitation vs. unimproved), solid waste disposal methods (improved collection vs. open dumping/burning), cooking fuel type (clean: LPG/biogas/electricity vs. polluting: charcoal/kerosene/biomass); and cooking space (separate room vs. veranda/open space/bedroom). These indicators reflect the immediate household environment rather than broader community or regional environmental conditions.

To address these gaps, this study tested two primary hypotheses: (1) older women would have significantly lower odds of being free from cognitive impairment compared to older men, after accounting for socioeconomic and demographic confounders; and (2) improved environmental sustainability factors, specifically clean cooking fuel, improved water, sanitation, waste disposal, and cooking space, would each be independently associated with higher odds of no cognitive impairment. The study also examined whether these environmental associations persisted after controlling for potential confounders including age, education, marital status, health insurance, employment, and rural-urban residence.

Throughout this paper, I distinguish between “sex” (biological category, male/female as recorded in the census) and “gender” (socially constructed roles, norms, and relations). The census variable captures biological sex, but we interpret findings through a gender lens, recognizing that observed differences between men and women likely reflect cumulative social, economic, and structural disadvantages rather than purely biological mechanisms. Where the literature discusses social mechanisms, “gender” is used; where referring to the measured variable or biological pathways, used “sex.

Conceptual framework for the study

In Fig. 1, this study conceptualises how environmental sustainability factors and gender are associated with cognitive impairment in older adults, accounting for key confounders. Cognitive impairment is measured by memory and concentration. Memory decline involves difficulties retaining and recalling information, while concentration issues affect focus and task management. Both are critical indicators of cognitive decline, which can increase reliance, reduce quality of life, and lead to other health complications.

Fig. 1.

Conceptual framework of environmental sustainability, gender cognitive impairment in older adults

Environmental factors such as access to clean water, sanitation, improved waste management, and cooking conditions directly affect older adults’ cognitive conditions. Poor access to these resources can lead to health issues that worsen cognitive decline, while harmful cooking practices (e.g., using charcoal or kerosene) can cause respiratory problems and indoor pollution, further impacting cognitive health.

Gender is also a significant factor, as men and women experience cognitive impairment differently due to associated factors like social, biological, and economic. Women may face additional stress from socioeconomic issues, while men often experience social isolation and suicidal ideation. These gender-specific challenges help explain the varied experiences of cognitive impairment between older men and women.

Also, confounders such as age, marital status, education, health insurance, employment, and residence may be associated with the environmental conditions, gender, and cognitive health. For example, higher education and stable employment can enhance cognitive resilience, while social support and access to resources vary between urban and rural settings, shaping cognitive outcomes among older adults.

The pathways linking environmental factors to cognitive outcomes operate through multiple mechanisms. These include first, the exposure to household air pollution from solid cooking fuel combustion generates systemic inflammation and oxidative stress, which are established contributors to neuropathology and cognitive decline [31, 32]. Second, unimproved water and sanitation increase risk of enteric infections and environmental enteropathy, conditions that drive chronic inflammation and nutrient malabsorption, indirectly affecting brain health [12, 33, 34]. Third, poor waste disposal creates breeding grounds for disease vectors and exposes residents to neurotoxicants, with cumulative effects over the life course [34]. Fourth, these environmental conditions may operate through intermediary pathways including cardiovascular disease, diabetes, and nutritional status, which are themselves risk factors for cognitive impairment.

Methods

Study area

Ghana has 30.8 million residents as of 2021 [4]. There are 16 administrative regions in total, with Greater Accra acting as the regional hub. Ghana’s borders are to the north with Burkina Faso, to the east with Togo, to the south with the Gulf of Guinea, and the west with Cote d’Ivoire [4]. Over half of the population (56.7%) lives in urban areas. In addition, the majority of Ghanaians (58.1%) are in the 15–59 age range, 35.4% are in the 0–14 age range, and 6.5% are 60 years of age and beyond [4].

Study aim and design

This study examined the gender-specific effects on cognitive health in the context of environmental challenges and investigated the association between environmental sustainability factors and cognitive impairment in older adults. Pooled data from the 10% of the Ghana Population and Housing Census (GPHC) 2021 were used in this study. The Ghana Statistical Service is in charge of the GPHC survey program, which is a nationally representative survey carried out in Ghana [4]. To improve and provide nationally updated data on the demographic, social, economic, migration, fertility, mortality, sanitation, housing and dwelling characteristics of the population, as well as information on the environmental sustainability of housing options for older adults in Ghana, the GPHC is a comprehensive enumeration of all individuals and households in Ghana [4].

The purpose of the census is to gather data from every person and household at one particular point in time through a cross-sectional study design. The 2021 GPHC employed a de jure method, in which people are counted at the place of their usual abode, guaranteeing that the data nearly matches the population distribution as feasible. Despite being Ghana’s sixth census since independence, this one is the first to fully utilise the Computer Assisted Personal Interview (CAPI) and Batch Program for Data Entry and Cleaning [35]. A total of 30.8 million respondents were recorded. Only older persons with complete data who were 60 years or older (n = 32486) were included in the analysis for this study. Older respondents under the age of 60 were, therefore, disqualified. Respondents were divided into three age groups using the functional age brackets: 60–69 (young-old), 70–79 (old-old), and 80+ (oldest-old) [36].

Missing data were minimal for the analytic sample. Of the 32,486 older adults included, complete data were available for all primary independent variables (sex and environmental sustainability factors) and the outcome variable (cognitive impairment), as these were core census questions with mandatory completion. For confounding variables, missingness ranged from 0% (age, sex, residence) to 2.1% (education level). Given the low proportion of missing data (< 5% for all variables), complete-case analysis was employed, retaining all 32,486 participants. This approach is unlikely to introduce significant bias when missingness is below 5% [37].

Study variables

Outcome variables

The primary outcome variable was cognitive impairment. In this study, it is conceptualised as the condition where older persons experience difficulty remembering information or concentrating on tasks due to challenges with memory recall, focus, or the ability to think clearly and make decisions [1]. It was captured as an intellectual disability (remembering or concentrating) in the 2021 GPHC. During the census, older adults were asked to indicate their difficulties in performing intellectual activities [4]. The responses were as follows: (1) No difficulty, (2) Yes, some difficulty, (3) Yes, a lot of difficulties, and (4) Cannot do it at all. These were categorised as 1= “No cognitive impairment”, and “2,3, and 4 = 0 “Cognitive impairment”. The study focussed on “No cognitive impairment” in the analysis.

Dichotomization of cognitive impairment into a binary variable was undertaken for three reasons. First, the census instrument used a single question on functional difficulty, which does not support graded distinctions beyond the presence or absence of reported difficulty. Second, the very low proportion of respondents reporting “a lot of difficulties” or “cannot do at all” (combined < 1%) precluded meaningful multi-category analysis. Third, binary classification aligns with the analytic approach used in comparable population-based studies of cognitive impairment in sub-Saharan Africa [35, 38]. We acknowledge that dichotomization loses information about severity, but this limitation is inherent to the available data.

The reliance on self-reported cognitive difficulties warrants careful consideration of potential misclassification bias. Self-report may lead to either underreporting (due to stigma, lack of awareness, or cultural norms that minimize cognitive problems) or overreporting (due to concurrent depression or anxiety). In the Ghanaian context, where mental health stigma remains significant, underreporting is more likely [39]. This would bias estimates toward the null, meaning that the true associations between environmental factors and cognitive impairment may be stronger than observed. Furthermore, the single-item measure may not fully capture clinical cognitive impairment, which typically requires comprehensive neuropsychological assessment. However, population-level screening using self-reported functional difficulties is consistent with WHO’s International Classification of Functioning framework and has demonstrated acceptable concordance with clinical assessments in large-scale surveys [35].

Independent variables

The independent variables considered were gender and environmental sustainability factors. Gender was captured as sex with responses such as (1) “Male” and (2) “Female”. Environmental sustainability variables include subcategories like household sources of drinking water, toilet facilities, methods of solid waste disposal, cooking fuel, and cooking space. The question used to assess the source of drinking water was “What is the main source of drinking water for the household?” These were categorised as “Improved water sources”, including pipe-borne inside dwelling, pipe-borne outside dwelling but on compound, pipe-borne outside dwelling but in neighbour’s house/compound, public tap/standpipe, borehole/tube well, protected well, rainwater, protected spring, and bottled water. “Unimproved water sources” comprised sachet water, tanker supplied/vendor provided, unprotected well, unprotected spring, river/stream, and dugout/pond/lake/dam/canal [40].

The type of toilet facility used by older adults was assessed by the question, “What type of toilet facility is mainly used by the household?” categorised as “improved sanitation” (septic tank (manhole), KVIP/VIP, Enviro Loo, Bio-digester (e.g., bio fill), Biogas), and “unimproved sanitation” (pit latrine, Bucket/Pan, Portable toilet (e.g., Water potti), Sewer, Public toilet, and none) [40]. The method of waste management was assessed by the question, “How does the household mainly dispose of rubbish (refuse)?” Categorisation included compaction trucks and central containers as “improved waste disposal”. “Unimproved waste disposal” included other vehicles, tricycles, and central containers buried in the ground, burned, public dump/open spaces, pushcarts/walk-in attendant/bicycle/wheelbarrow, and dumped indiscriminately [40].

Two questions were asked in the 2021 GPHC to assess the cooking condition of older adults. The responses to the question “What is the source of cooking fuel for this household?” were categorised as “improved cooking fuel” (LPG, Bio Gas, and Electricity) and “unimproved cooking fuel” (Kerosene, Charcoal, Crop residue, sawdust, Animal waste, and cooking gel, none). “What type of cooking space does this household use?” was used to measure improved and unimproved cooking space. Improved cooking space constituted a separate room for exclusive use of household and separate room shared with other household(s). Unimproved cooking space included enclosure without a roof, structure with a roof but without walls, bedroom/hall/living room, veranda/porch, and open space in the compound [40, 41].

Confounding factors

Confounders were selected a priori based on established theoretical frameworks and empirical evidence from prior studies of cognitive ageing in low- and middle-income countries [22, 23, 42]. Age was included because neuropathological changes accumulate with advancing years. Education was included based on the cognitive reserve hypothesis, which posits that intellectual stimulation builds neural resilience. Marital status and employment were included as indicators of social and economic integration. Health insurance was included as a proxy for healthcare access. Rural-urban residence was included to capture differential access to services and environmental exposures. All confounders were measured using standard census questions as previously validated by the Ghana Statistical Service.

Data analysis

Data analysis was performed using SPSS v27.0. First, frequencies and percentages described the background characteristics of respondents. Second, cross-tabulations with chi-squared tests examined the percentage distribution of older adults with no cognitive impairment by sex, environmental sustainability factors, and confounders. Statistical significance for chi-squared tests was set at p < 0.05. Third, three hierarchical multivariate logistic regression models were performed. Model 1 included sex only. Model 2 added environmental sustainability variables (drinking water, toilet facility, waste disposal, cooking fuel, cooking space). Model 3 added all confounders (age, marital status, education, health insurance, employment status, sector of employment, residence). Variables were retained in the model based on theoretical importance regardless of statistical significance, as recommended to ensure complete adjustment for known determinants of cognitive health [43].

Model diagnostics were assessed as follows: The Hosmer-Lemeshow test indicated good model fit (χ² = 8.23, df = 8, p = 0.41). Classification accuracy for the final model was 94.8%, with sensitivity of 98.2% and specificity of 32.1% (reflecting the outcome imbalance). The area under the ROC curve (AUC) was 0.71 (95% CI = 0.68, 0.74), indicating acceptable discriminative ability. Multicollinearity was assessed using variance inflation factors (VIF); all VIF values were below 1.5 (mean VIF = 1.12), indicating no problematic multicollinearity. Odds ratios (ORs) with 95% confidence intervals (CIs) are presented. Statistical significance is indicated when the 95% CI does not include 1.0 (p < 0.05). Where CIs cross 1.0, findings are described as “not statistically significant” regardless of the magnitude of the point estimate.

Sensitivity analysis: To assess the robustness of findings to outcome coding, all models were re-ran with the outcome coded as “cognitive impairment” (rather than “no cognitive impairment”). The direction and significance of associations were reversed but symmetrical, confirming that findings are not artifacts of coding choice. While interaction terms (sex × environmental factors) were not included in the primary models due to sample size considerations for the oldest-old subgroups, we recommend future research with larger samples to formally test whether environmental associations differ by sex.

In Model 3, older adults who used improved toilet facilities showed higher odds of no cognitive impairment (OR = 1.08), but this association was not statistically significant as the confidence interval crossed 1.0 (95% CI = 0.97, 1.20). Similarly, improved waste disposal (OR = 1.14, 95% CI = 0.98, 1.32) and improved cooking space (OR = 1.01, 95% CI = 0.91, 1.12) were not significantly associated with the outcome. Improved water source showed a point estimate of 1.16, but the wide confidence interval (0.87, 1.55) indicates imprecision and lack of statistical significance. Interpretation focuses on the fully adjusted model (Model 3).

Results

Demographic characteristics of respondents

Of the 32,486 respondents sampled, results in Table 1 revealed that more than half (59.8%) were male, and the majority (92.3%) were young-olds (60–74 years). About two-thirds (64.9%) were married, with 55.7% having completed middle/JHS level of education. Most older adults have health insurance (74.3%), are self-employed (80.4%), work in the private sector (93.0%), and reside in urban areas (58.4%). For environmental sustainability, the results show that a little more than half of older adults have improved sources of household drinking water (58.3%), toilet facilities (68.4%), sources of cooking fuel (60.6%), improved cooking space (68.3%), and the majority (79.6%) use improved methods of waste disposal. The prevalence of cognitive impairment was 5.4%, with 94.6% reporting no cognitive impairment.

Table 1.

Univariate analysis of background characteristics of respondents (n = 32,486)

Variables	n	(%)
Sex
Male	19,417	59.8
Female	13,069	40.2
Age
60–74 years	29,998	92.3
75–84 years	2208	6.8
85 + years	280	0.9
Marital status
Never Married	722	2.2
Married	21,073	64.9
Separated/Divorced	4923	15.2
Widowed	5768	17.8
Level of education
No Education	99	0.3
Primary	4509	13.9
Middle/JHS	18,089	55.7
Secondary/SHS	3741	11.5
Vocational	2029	6.2
University graduate	4019	12.4
Health insurance
Yes	24,146	74.3
No	8340	25.7
Employment status
Not employed	196	0.6
Employee	5403	16.6
Self employed	26,103	80.4
Casually employed/Apprentice	784	2.4
Sector of employment
Public	1439	4.4
Private	30,223	93.0
NGO/Religious Org/International Org	824	2.5
Location
Urban	18,985	58.4
Rural	13,501	41.6
Environmental sustainability
Household source of drinking water
Improved	18,950	58.3
Unimproved	13,536	41.7
Toilet facility
Improved	22,210	68.4
Unimproved	10,276	31.6
Method of solid waste disposal
Improved	25,875	79.6
Unimproved	6611	20.4
Cooking fuel
Improved	19,676	60.6
Unimproved	12,810	39.4
Cooking space
Improved	22,196	68.3
Unimproved	10,290	31.7
Cognitive impairment
Cognitive impairment	1741	5.4
No cognitive impairment	30,745	94.6

Environmental sustainability, gender and cognitive impairment

Apart from health insurance, all the variables were statistically significant. Among older adults with no cognitive impairment, a higher proportion were female (53.6%), while 46.4% were male. This distribution reflects the sex composition of the cognitively healthy subgroup, not the proportion of females who were cognitively healthy. The results in Table 2 show that older adults who drank from improved sources of water (58.0%), used improved toilet facilities (68.8%), cooking space (68.5%), as well as those who used improved methods of solid waste disposal (79.3%) and cooking fuel (59.8%) had no cognitive impairment.

Table 2.

Bivariate analysis of gender, environmental sustainability by no cognitive impairment (n = 32,486)

Variables	No Cognitive Impairment
	%	(n) (95% CI)	p-value
Gender			< 0.001
Male	46.4	19,417 (0.46, 0.47)
Female	53.6	13,069 (0.53, 0.54)
Environmental sustainability
Household source of drinking water			< 0.001
Improved	58.0	17,847 (0.57, 0.59)
Unimproved	42.0	12,898 (0.41, 0.43)
Toilet facility			P = 0.05
Improved	68.8	21,139 (0.68, 0.69)
Unimproved	31.2	9606 (0.30, 0.32)
Method of solid waste disposal			< 0.001
Improved	79.3	24,383 (0.79, 0.80)
Unimproved	20.7	6362 (0.20, 0.22)
Cooking fuel			< 0.001
Improved	59.8	18,381 (0.59, 0.61)
Unimproved	40.2	12,361 (0.39, 0.41)
Cooking space			< 0.001
Improved	68.5	21,058 (0.68, 0.69)
Unimproved	31.5	9687 (0.31, 0.32)
Confounders
Age			< 0.001
60–74 years	92.9	28,559 (0.93, 0.93)
75–84 years	6.3	1952 (0.61, 0.65)
85 + years	0.8	234 (0.75, 0.85)
Marital status			< 0.001
Never Married	2.2	686 (0.19, 0.25)
Married	65.7	20,195 (0.65, 0.66)
Separated/Divorced	14.9	4580 (0.14, 0.16)
Widowed	17.2	5284 (0.16, 0.18)
Level of education			< 0.001
No Education	0.3	90 (0.21, 0.39)
Primary	13.4	4125 (0.12, 0.14)
Middle/JHS	55.8	17,151 (0.55, 0.57)
Secondary/SHS	11.6	3556 (0.11, 0.13)
Vocational	6.3	1946 (0.61, 0.65)
University graduate	12.6	3877 (0.12, 0.14)
Health insurance			0.339
Yes	74.3	22,835 (0.74, 0.75)
No	25.7	7910 (0.25, 0.27)
Employment status			< 0.001
Not employed	0.6	182 (0.53, 0.67)
Employee	16.9	5199 (0.16, 0.18)
Self employed	80.1	24,640 (0.80, 0.81)
Casually employed/Apprentice	2.4	724 (0.21, 0.27)
Sector of employment			< 0.001
Public	4.5	1388 (0.42, 0.48)
Private	93.0	28,595 (0.93, 0.93)
NGO/Religious Org/International Org	2.5	762 (0.22, 0.28)
Location			< 0.001
Urban	59.2	18,216 (0.74, 0.75)
Rural	40.8	12,529 (0.25, 0.26)

Age, marital status, education, employment, sector of employment and place of resident, were statistically and significantly associated with no cognitive impairment. Majority of those in the young-old age group (60–74 years) had no cognitive impairment. A higher percentage who were married (65.7%), completed middle/JHS (55.8%), with health insurance (74.3%), self-employed (80.1%), working in the private sector (93.0%), and living in urban areas (59.2%), had no cognitive impairment.

Gender, environmental sustainability, and no cognitive impairment of older adults

Results are presented with the outcome coded as “no cognitive impairment” (i.e., absence of reported difficulty remembering or concentrating) (Table 3). Thus, higher odds ratios indicate protective factors associated with better cognitive health. Readers should interpret all associations in this direction.

Table 3.

Associations between gender, environmental sustainability and no cognitive impairment (n = 32,486)

Housing Options	Model 1	Model 2	Model 3
	OR [95% CI]	OR [95% CI]	OR [95% CI]
Gender dimension
Sex
Male	1.00	1.00	1.00
Female	0.56(0.51, 0.62)**	0.54(0.49, 0.60) **	0.61(0.55, 0.69) **
Environmental Sustainability
Drinking Water
Unimproved		1.00	1.00
Improved		1.02(0.83, 1.26)	1.16(0.87, 1.55)
Toilet facility
Unimproved		1.00	1.00
Improved		1.17(1.05, 1.30)	1.08(0.97, 1.20)
Waste disposal
Unimproved		1.00	1.00
Improved		1.28(1.11, 1.48)	1.14(0.98, 1.32)
Cooking fuel
Unimproved		1.00	1.00
Improved		1.72(1.52, 1.94)	1.41(1.23, 1.61)
Cooking space
Unimproved		1.00	1.00
Improved		1.01(0.91, 1.12)	1.01(0.91, 1.12)
Confounders
Age
60–74 years			1.00
75–84 years			0.44(0.38, 0.51)**
85 + years			0.31(0.22, 0.44) **
Marital status
Never Married			1.00
Married			1.27(0.90, 1.79)
Separated/Divorced			0.85(0.60, 1.22)
Widowed			0.86(0.60, 1.22)
Level of education
No Education			1.00
Primary			1.44(0.70, 2.95)
Middle/JHS			1.84(0.90, 3.75)
Secondary/SHS			1.63(0.79, 3.36)
Vocational			1.92(0.92, 4.04)
University graduate			1.90(0.92, 3.95)
Employment status
Not employed			1.00
Employee			1.17(0.66, 2.09)
Self employed			1.16(0.67, 2.03)
Casually employed			0.91(0.49, 1.68)
Sector of employment
Public			1.00
Private			1.06(0.76, 1.48)
NGO/Religious Org/International Org			0.64(0.43, 0.94)**
Place of residence
Urban			1.00
Rural			0.67(0.59, 0.75)**

Note: Outcome = no cognitive impairment. Thus OR > 1 indicates protective association (lower odds of cognitive impairment); OR < 1 indicates higher odds of cognitive impairment. **p < 0.01; *p < 0.05

Females had 39% lower odds of being free from cognitive impairment compared to males (AOR = 0.61, 95% CI = 0.55, 0.69). This indicates that females had significantly higher odds of cognitive impairment relative to males. The magnitude of this association was partially attenuated after adjustment for confounders (from OR = 0.56 in Model 1 to AOR = 0.61 in Model 3), suggesting that some of the sex disparity is explained by socioeconomic and demographic factors. Among the environmental sustainability variables, the use of clean cooking fuel was associated with significantly higher odds of being free from cognitive impairment (AOR = 1.41, 95% CI = 1.23, 1.61). This association remained robust after adjustment for all confounders, representing a 41% increase in the odds of no cognitive impairment among users of clean cooking fuel compared to users of polluting fuels.

In contrast, the remaining environmental factors showed positive point estimates that did not reach statistical significance. Improved water source (AOR = 1.16, 95% CI = 0.87, 1.55), improved toilet facilities (AOR = 1.08, 95% CI = 0.97, 1.20), improved waste disposal (AOR = 1.14, 95% CI = 0.98, 1.32), and improved cooking space (AOR = 1.01, 95% CI = 0.91, 1.12) each had confidence intervals that crossed 1.0. These findings do not provide statistical evidence for independent associations with cognitive health in the fully adjusted model, though the direction of effects is consistent with the study hypotheses.

Old-olds (75–84 years) and oldest-olds (85 + years) were less likely to have no cognitive impairment (OR = 0.44, 95% CI = 0.38, 0.51) and (OR = 0.31, 95% CI = 0.22, 0.44) when compared with young olds (60–74 years). Being married was more likely to be associated with no cognitive impairment (OR = 1.27, 95% CI = 0.90, 1.79) when compared with the never married. Older adults who completed primary (OR = 1.27, 95% CI = 0.70, 2.95), middle/JHS (OR = 1.84, 95% CI = 0.90, 3.75), secondary/SHS (OR = 1.44, 95% CI = 0.79, 3.36), vocational (OR = 1.92, 95% CI = 0.92, 4.04), and university (OR = 1.90, 95% CI = 0.92, 3.95) levels of education had higher odds of no cognitive impairment when compared with their peers with no education. Compared to the non-employed participants, employees (OR = 1.17, 95% CI = 0.66, 2.09) and self-employed (OR = 1.16, 95% CI = 0.67, 2.03) were more likely to have no cognitive impairment. Also, older adults in the private sector had higher odds of no cognitive impairment (OR = 1.06, 95% CI = 0.76, 1.48) compared with those in the public sector. Residing in rural areas was less likely to be associated with no cognitive impairment (OR = 0.67, 95% CI = 0.59, 0.75) when compared with older adults in urban settings. The results of the hierarchical multivariate logistic regression models are presented in Table 3.

Discussion

This study examined the gender-specific effects on cognitive health in the context of environmental challenges and the relationship between factors of environmental sustainability and cognitive impairment in adults 60 years and older in Ghana. Older women had significantly higher odds of cognitive impairment than older men. While several environmental sustainability factors were associated with cognitive health in the expected direction, only the use of clean cooking fuel suggested a statistically significant protective association. Other factors, such as access to improved water, sanitation, and waste disposal, showed positive trends but did not reach statistical significance in the fully adjusted models.

Gender and cognitive impairment

Consistent with our hypothesis, being female was associated with significantly lower odds of being free from cognitive impairment. This finding aligns with a substantial body of evidence indicating that women are disproportionately affected by dementia and cognitive decline globally. Epidemiological studies consistently suggests that while men may have higher mortality rates from cardiovascular disease, women have a higher age-standardized prevalence of dementia, which may be partially attributed to their longer life expectancy and neurobiological factors [23].

However, the conceptual framework posited that socioeconomic stressors, rather than biological sex alone, drive much of this disparity. This interpretation aligns with recent evidence. A systematic review by Dreyer et al. [16] examining sex differences in cognitive performance among people living with HIV in South Africa found that while unadjusted models suggested women performed worse, these differences were fully explained by psychosocial and structural factors. Specifically, fewer years of education, illiteracy, and greater food insecurity—markers of broader gender inequality—predicted lower cognitive performance in women but not in men. This mirrors the situation in Ghana, where older women often have lower lifetime earnings, reduced access to formal education, and greater caregiving burdens. Although women in this study sample reported stronger social networks, the cumulative stress from economic instability and caregiving responsibilities appears to outweigh these protective benefits, contributing to the higher likelihood of cognitive impairment observed [23, 26].

The observed sex difference, with women significantly less likely to be free from cognitive impairment, requires distinction between biological and social mechanisms. From a biological perspective, oestrogen has neuroprotective effects, and its decline after menopause may accelerate cognitive ageing [44, 45]. However, this biological pathway alone cannot explain the magnitude of the disparity observed in this study, particularly given that most women in the sample were post-menopausal. From a social perspective, the cumulative disadvantages women face over the life course in Ghana—including lower educational attainment, reduced labour force participation, lower lifetime earnings, and greater caregiving burdens—are more plausible drivers of the observed disparity [46]. The conceptual framework posited that socioeconomic stressors, rather than biological sex alone, drive much of this disparity. This interpretation is supported by evidence that adjusting for education and employment substantially attenuated the sex coefficient (from OR = 0.56 in Model 1 to OR = 0.61 in Model 3, a 9% attenuation), though the persistent residual disparity suggests that unmeasured factors, including differential healthcare access, health-seeking behaviour, and lifetime stress exposure, may also contribute.

Environmental sustainability and cognitive health

The non-significant findings for water, sanitation, and waste disposal require careful interpretation. Several factors may explain why these environmental factors did not reach statistical significance in the fully adjusted model. First, the pathway between WASH infrastructure and cognitive function is likely indirect and mediated through physical health, operating over the life course [47]. Cross-sectional studies are inherently limited in detecting cumulative effects that may manifest only after prolonged exposure. Second, measurement constraints inherent to census data—including binary classifications that obscure heterogeneity within categories—may have attenuated effect estimates. Third, survivorship bias may attenuate observed associations if lifelong exposure to unimproved conditions increases premature mortality, leaving a selectively resilient cohort. Importantly, the null findings should not be interpreted as evidence of no effect, but rather as an indication that the cognitive returns on WASH infrastructure may be indirect, cumulative, and difficult to quantify without longitudinal data. These findings underscore the need for longitudinal studies with more detailed exposure assessments.

Second, measurement constraints inherent to census data may have attenuated effect estimates. The binary classification of water and sanitation using WHO/UNICEF JMP standards, while useful for global monitoring, obscures substantial heterogeneity within categories. Improved water sources may nonetheless be intermittently contaminated, seasonally unavailable, or located at distances that are prohibitive for older adults with limited mobility [48, 49]. Similarly, the presence of an improved toilet facility does not guarantee consistent use by all household members, nor does it reflect the broader sanitation environment of the community. Such non-differential exposure misclassification would bias estimates toward the insignificant findings and may explain the attenuation of odds ratios towards unity after covariate adjustment.

Third, the single-item outcome measure—difficulty remembering or concentrating—may lack the granularity to detect domain-specific cognitive effects associated with environmental exposures. Emerging evidence suggests that specific neurotoxic contaminants sometimes found in groundwater, such as lead, arsenic, and manganese, are differentially associated with executive function and processing speed rather than global memory impairment [50, 51]. The census measure, while pragmatic for population-level surveillance, may be insufficiently sensitive to capture such subtle or domain-limited deficits. Future research employing comprehensive neuropsychological batteries is required to delineate whether WASH factors selectively are associated with particular cognitive abilities in older African populations.

Fourth, survivorship bias may attenuate observed associations. If lifelong exposure to unimproved water and sanitation increases premature mortality from infectious or cardiovascular causes, the surviving cohort of older adults in such conditions may represent a selectively resilient group, leading to underestimation of the true population-level burden [52]. Despite this attenuation, the directionally consistent (albeit non-significant) effects across multiple WASH indicators suggest these factors do contribute to cognitive health, even if independent effects are modest or difficult to isolate using cross-sectional methods. Similar patterns have been reported in neighbourhood deprivation studies, where individual environmental features often fail to reach significance despite overall trends favouring higher-resource environments [53].

Crucially, the null findings for WASH variables should not be interpreted as evidence of no effect, nor should they diminish policy commitment to universal water and sanitation coverage. The present findings simply highlight that the cognitive returns on environmental infrastructure may be indirect, cumulative, and challenging to quantify in the absence of longitudinal data.

Confounding factors

The significant effects of the control variables were consistent with established literature. The strong inverse association between advanced age (75 + years) and cognitive health for 85 + years reflects the neuropathological changes associated with ageing [42]. Higher education showed a graded association with better cognitive health, consistent with the cognitive reserve hypothesis. For marital status, while the point estimate suggested a protective association for married individuals (AOR = 1.27), this did not reach statistical significance in the fully adjusted model (95% CI = 0.90, 1.79) [54]. This lack of significance may reflect the influence of other socioeconomic factors that were controlled for in the analysis.

Furthermore, the finding that rural residents were significantly less likely to be free from cognitive impairment (AOR = 0.67, 95% CI = 0.59, 0.75) highlights the structural health inequalities in Ghana. This disparity persisted even after adjustment for socioeconomic and environmental factors, suggesting that unmeasured aspects of rural life—including reduced access to specialised healthcare, lower quality of health services, greater social isolation, and cumulative exposure to environmental risks—may contribute to poorer cognitive outcomes among rural older adults. Rural older adults in Ghana face particular vulnerabilities: they are more likely to rely on biomass fuel for cooking, have limited access to healthcare facilities, and may experience greater social isolation due to out-migration of younger family members. This spatial disparity in cognitive health underscores the need for decentralised mental health services, improved rural healthcare infrastructure, and targeted environmental upgrades in rural communities [55]. However, the cross-sectional design precludes determination of whether rural residence causes cognitive decline or whether individuals with cognitive impairment are more likely to reside in rural areas (reverse causality).

Strengths and limitations

The study provides important insights into cognitive impairment among older adults in Ghana, focusing on the impact of environmental factors and gender. It is notable for its comprehensive approach and use of robust, nationally representative data, which helps fill a research gap in low- and middle-income countries. Additionally, it offers valuable gender-specific findings, showing how men and women experience cognitive decline differently.

However, the study’s cross-sectional design limits the ability to determine causality, and its reliance on self-reported data may introduce bias, particularly in the assessment of cognitive health. The cognitive impairment measure was derived from self-reported difficulties in remembering or concentrating, which may not fully align with clinical diagnoses of cognitive impairment and could be associated with social desirability or lack of awareness. Furthermore, the survey item specifically referred to “intellectual disability (remembering or concentrating),” which captures aspects of intellectual functioning rather than a broader clinical definition of cognitive impairment.

While this operationalization provides a practical proxy within a large-scale census, it may not fully encompass the complexity of cognitive disorders such as mild cognitive impairment (MCI) or dementia, which involve additional domains like executive function, language, and visuospatial skills. While the study covers a broad range of environmental factors, further research is needed to better understand their direct association with cognitive health. The absence of longitudinal data and deeper cultural analysis also limits the study’s long-term insights.

Cognitive impairment was assessed using a single self-reported item about difficulty remembering or concentrating. While pragmatic for a national census and concordant with clinical assessments in population-based research, this approach does not substitute for comprehensive neuropsychological evaluation. Self-report may be influenced by mood, stigma, or lack of awareness, and the item does not distinguish between mild cognitive impairment and dementia nor capture domain-specific deficits (e.g., executive function, language, visuospatial skills). Future research should employ validated cognitive screening instruments (e.g., the Community Screening Instrument for Dementia) or clinical assessments. The cross-sectional design precludes causal inference. Reverse causality is possible: older adults with cognitive impairment may be less able to maintain improved environmental conditions (e.g., using polluting fuels because they cannot safely operate LPG cylinders, or having poorer sanitation due to reduced mobility or financial constraints). Similarly, the observed rural-urban disparity could reflect selective migration, with healthier older adults moving to urban areas for better services. Longitudinal studies are needed to establish temporal ordering and direction of effects.

Unmeasured confounders: Despite adjusting for a comprehensive set of confounders, residual confounding remains possible. Important unmeasured variables include: lifetime occupational exposures, diet and nutritional status, physical activity, social engagement, genetic factors (e.g., APOE ε4 status), and early-life conditions (e.g., childhood nutrition, infectious disease history). These factors may confound the observed associations between environmental conditions and cognitive health.

Conclusions

This study provides evidence that cognitive health among older adults in Ghana is significantly associated with two determinants: sex (with women at higher risk) and household cooking fuel type. While sex is not itself modifiable, the socioeconomic inequalities that likely contribute to the observed sex disparity—including differences in education, lifetime earnings, and caregiving burden—are potentially addressable through policy interventions. Household cooking fuel, in contrast, is a modifiable environmental factor. The associations for water, sanitation, and waste disposal, while directionally consistent with hypotheses, did not reach statistical significance in this cross-sectional analysis.

Whether these associations reflect causal relationships cannot be determined from this cross-sectional study. The observed association between clean cooking fuel and cognitive health is consistent with mechanisms hypothesized in prior research, including reduced exposure to household air pollution and its downstream effects on systemic inflammation, oxidative stress, and cardiovascular health. However, these mechanisms were not directly tested in this study. Future research should incorporate biomarkers and neuroimaging to investigate the pathways through which cooking fuel type may influence cognitive health. Similarly, whether interventions to improve environmental conditions would improve cognitive outcomes requires testing through experimental or quasi-experimental designs.

Despite these uncertainties, the findings have immediate policy relevance. First, cognitive health strategies should be gender-sensitive, addressing women’s cumulative socioeconomic disadvantage through improved educational access, formal employment opportunities, and social protection programmes. Second, the association between clean cooking fuel use and better cognitive health suggests that Ghana’s clean energy agenda may have potential cognitive co-benefits. Future research should investigate whether LPG subsidies and rural electrification interventions specifically improve cognitive outcomes among older adults. Cross-sectoral collaboration between health, energy, and environmental agencies could facilitate monitoring of these potential co-benefits. From a public health perspective, WASH and clean energy interventions deliver multiple established benefits—reducing diarrhoeal disease, improving respiratory health, enhancing dignity, and supporting overall well-being—which justify continued investment irrespective of their specific effects on cognition. However, the cognitive benefits demonstrated here require confirmation through longitudinal and intervention studies before they can be considered a primary rationale for environmental policy.

Abbreviations

MCI

Mild Cognitive Impairment

GPHC

Ghana Population and Housing Census

CAPI

Computer Assisted Personal Interview

SPSS

Statistical Package for the Social Sciences

OR

Odds Ratio

CI

Confidence Interval

KVIP

Kumasi Ventilated Improved Pit

VIP

Ventilated Improved Pit

LPG

Liquefied Petroleum Gas

JHS

Junior High School

SHS

Senior High School

NGO

Non-Governmental Organization

Author contributions

The author, Joseph Kojo Oduro, conceptualised the study, analysed the data, and drafted and reviewed the first draft of the manuscript. The final manuscript was validated and approved by the author.

Funding

No funding was received for this study.

Data availability

The datasets generated and/or analysed during the current study are available at the Ghana Statistical Service database at the repository: https://statsbank.statsghana.gov.gh/.

Declarations

Ethics approval and consent to participate

This study was conducted in line with the appropriate guidelines and regulations conforming to both national legal frameworks and international statistical standards, including the principles of the Declaration of Helsinki for research involving human participants. The 2021 PHC adhered to the Statistical Service Act, 2019: Act 1003, and the United Nations Principles and Recommendations for Population and Housing Censuses. The Ghana Health Service Ethical Review Committee (GHS-ERC) approved the ethics framework for the census (Approval number: GHS-ERC: 05/03/2021). Participation in the 2021 PHC was based on voluntary consent. Participants provided consent to participate in the survey in line with international standards for public data collection. The author of this manuscript was not directly involved in the data collection processes but rather obtained access by requesting the data.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

General section

The reporting of this study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist for cross-sectional studies.