Dyslipidemia and Associated Risk Factors in the Elderly Population in Asmara, Eritrea: Results from a Community-Based Cross-Sectional Study

Background The ultimate goal of the study was to approximate the burden and patterns of dyslipidemia in a subset of the elderly population (≥60–85 years) living in Asmara, Eritrea, and to identify modifiable risk drivers. Methods A total of 319 (145 (45.5%) male vs. 174 (54.5%) female, mean age ± SD (68.06 ± 6.16 years), participants from randomly selected estates within Asmara were enrolled. Demographic and medical information was collected using a standardized questionnaire. Anthropometric, lipid panel, fasting plasma glucose (FPG), and blood pressure (BP) measurements were subsequently taken. Results The prevalence of dyslipidemia was 70.5%. The proportions of dyslipidemias were (in order of decreasing frequency) high TC (51.2%), LDL-C (43.7%), low HDL-C (28.2%), and TG (27.6%). The average (±SD) concentrations in mg/dL of TC, LDL-C, non-HDL-C, TG, HDL-C, TC/HDL-C, and TG/HDL-C were 202.2 ± 40.63, 125.95 ± 33.16, 151.72 ± 37.19, 129 ± 57.16, 50.48 ± 10.91, 4.11 ± 0.91, and 2.72 ± 1.49, respectively. Furthermore, 17.5%, 21.6%, 11.0%, and 5.0% had abnormalities in 1, 2, 3, and 4 lipid disorders with the copresence of TC+LDL-C abnormalities dominating. Regarding National Cholesterol Education Program Third Adult Treatment Panel risk strata, 18.5%, 14.5%, 28.2%, and 12.9% were in high or very high-risk categories for TC, LDL-C, TG, and HDL-C, respectively. The high burden of dyslipidemia coexisted with an equally high burden of abdominal obesity (43.1%), FPG ≥ 100 mg/dL (16%), hypertension (28.5%), and physical inactivity. Overall, dyslipidemia was associated with sex (females: aOR = 2.6, 95%CI = 1.1–6.1, p = 0.017) and daily physical activity—higher in individuals undertaking physical activity for <1 hour (aOR = 2.6, 95%CI = 1.1–6.1, p = 0.029), 1-2 hours (aOR = 3.2, 95%CI = 1.24–8.5, p = 0.016), and 2-3 hours (aOR = 2.0, 95%CI = 0.7–5.8, p = 0.192) (Ref: >3 hours). Additional associations included increasing FPG (aOR = 1.02, 95%CI = 1.0–1.04, p = 0.039), and BMI (aOR = 1.19, 95%CI = 1.09–1.3, p < 0.001). These factors, along with waist circumference (WC), consumption of traditional foods, systolic BP, and diastolic BP, were, with some variations, associated with disparate dyslipidemias. Conclusions The burden of dyslipidemia in the elderly population in Asmara is high. Modifiable risk drivers included FPG, WC, physical inactivity, and low consumption of traditional food. Overall, efforts directed at scaling up early recognition and treatment, including optimal pharmacological and nonpharmacological therapy, at all levels of care, should be instituted.


Introduction
Globally, the proportion of people > 60 years (elderly) is expected to double between 2000 and 2050 with the elderly outnumbering the young (<5 years old) in many countries in the near term [1].This represents a formidable problem for low-and middle-income countries (LMIC) since this subgroup is inherently vulnerable to cardiovascular diseases (CVDs) [2] and is susceptible to multimorbidity, malignancies, neurodegenerative diseases, polypharmacy, and long disability periods, among others.For example, the aging and growth of the populations across the world resulted in an increase in global CVD-related mortality between 1990 and 2013, despite a decrease in age-specific death rates in most regions.More recently, the 2019 global burden of disease (GBD) estimate reported that CVDs caused approximately 17.8 million deaths globally [3] with a disproportionate number of cases in LMIC-a significant proportion of these cases were in individuals > 60 years, with an even larger prevalence (>85%) in individuals > 80 years in some jurisdictions [4].Therefore, it can be argued that increases in years of life lost (YLLs), years lived with disability (YLDs), and losses in disability-adjusted life years (DALY) associated with heart failure (HF), coronary heart disease (CHD), peripheral arterial disease (PAD), and atrial fibrillation (AF) [2] epitomize the convergence of CVDs and aging in a rapidly aging world.
From an epidemiological standpoint, aging populations in Sub-Saharan Africa (SSA) will continue, at least partially, to shift the long-standing paradigm of infectious disease dominance to a multifaceted scenario where aging and NCDs, particularly CVDs, are playing an increasingly important role [5].Predictably, few countries in SSA can shoulder the disproportionate number of hospitalizations, procedures, and costs associated with the emerging burden of CVDs and other age-related disorders in their elderly populations.Besides age, which is regarded as a CVD risk equivalent (most individuals are already at (very) high risk at the age of 65 years) [6], available evidence indicates that multiple modifiable risk factors can trigger or augment the said risk [7].A 2013 GBD project estimation model indicated that the population-attributable fraction (PAF) for individual risk factors for CHD burden in LMICs was as follows: high blood pressure, 54%; high cholesterol concentrations, 32%; body mass index ≥ 25 kg/m 2 , 18%; unhealthy diet, 67%; and smoking, 18% [8].In other words, the report indicated that dyslipidemia is one of the leading contributors to CVD and mortality in elderly populations across the world.
The absolute risk associated with pathological changes in lipids rises substantially with advancing age [2].For instance, the Established Populations for Epidemiology Studies in the Elderly (EPESE) study noted that elevation in total cholesterol (TC), low-density lipoprotein cholesterol (LDL-C), and high-density lipoprotein cholesterol (HDL-C) concentrations is associated with increased morbimortality in older adults.Other investigators have demonstrated a linear, dose-dependent relationship between total cholesterol (TC) and CHD without an obvious lower threshold of TC [9].Crucially, and taking the controversies regarding the management of dyslipidemias in geriatrics (particularly the mass use of statins in centenarians) into account, the Prospective Study of Pravastatin in Elderly at Risk (PROSPER) and Intervention Trial Evaluating Rosuvastatin (JUPITER) established the efficacy and safety of statin therapy in the reduction of primary composite endpoints of coronary death, nonfatal myocardial infarction (MI), and cerebrovas-cular accident (stroke) in the elderly [10,11].More to the point, the data suggests that the proper management of dyslipidemia can substantially attenuate CVD risk or morbimortality in a significant proportion of the elderly population [12,13].
Even though dyslipidemia is regarded as one of the most important CVD risk factors in the elderly [2] and ranks highly in prominent CVD prevention guidelines, the condition is underdiagnosed and undertreated in elderly subgroups in SSA.Additionally, critical knowledge gaps on prevalence, pattern, and distribution of key risk factors remain.These descriptions describe the situation in Eritrea where a growing burden of CVD [14] (due to increasing life expectancy, among others) is compounded by suboptimal treatment, lack of knowledge and awareness, and limited understanding of the epidemiology of dyslipidemia.The present study was therefore undertaken to provide baseline data on dyslipidemia.Considering the underrepresentation of the elderly subgroups in dyslipidemia-related community-based surveys from SSA, we believe that this study provides significant information.

Study Setting and Study
Design.This was a communitybased cross-sectional study conducted among the elderly population (age ≥ 60 -85 years) between January and June 2018 in Asmara, Eritrea.

Sample Size Calculation, Participant Recruitment, and
Selection.The study targeted the elderly civilian population in Asmara, Eritrea.The choice of the setting was informed by the fact that Asmara has a relatively large population of elderly citizens.It is also regarded as the most cosmopolitan center in the country owing to its status as the capital city.The sample size was estimated using a single proportion formula.The prevalence of dyslipidemia was assumed to be 30% [15], with a margin of 5% and a 95% confidence interval (CI).After adjustments for nonresponse (10%), a total of 319 participants were recruited.
To locate participants, a stratified sampling design was employed.Based on previous experience, 13 estates (1 estate per subzone) were selected using the lottery method.Individuals ≥ 60 to ≤85 years working or residing within the selected estates were subsequently invited to participate in the study.Apart from age, additional inclusion criteria included Asmara city residents for at least 1 year and the ability to give consent.On the other hand, the exclusion criterion was based on the following considerations: mental illness/dementia, persons not willing to grant consent, and diabetes mellitus (DM) patients.

Demographic and Health History.
A predesigned data collection form (the World Health Organization (WHO) STEPS questionnaire) [16] was used to collect data.Demographics, factors associated with lifestyle, medications (antilipid medications), and existing comorbidities were collected using this document.

Participants' Lipid Profiles as per the ATP III and Adult
Treatment Panel III Risk Schema.The NCEP-ATP III guidelines were used to evaluate participants' atherosclerotic cardiovascular disease risk (ASCVD).Based on this scheme, 18.5%, 14.5%, 28.2%, and 12.9% were in high or very highrisk categories for TC, LDL-C, TG, and HDL-C, respectively.A significant difference between males and females was observed in all groupings assessed (see Table 3).

Factors
Associated with Selected Dyslipidemias.The prevalence of elevated TC, LDL, TG, non-HDL-C, TC/HDL, and TG/HDL; the presence of at least one dyslipidemia; and low HDL-C have been stratified by specific covariates.Compared to males, females had significantly higher proportions in TC, LDL-C, non-HDL-C, TG, and dyslipidemia.Education level (mostly higher in the unemployed) was associated with high TC and low HDL-C, while the frequency of physical activity (higher in individuals having <1-hour daily physical activity) was associated with high LDL, non-HDL, TG, and TC/HDL ratio.Reduced frequency of traditional food consumption was associated with elevated TC, LDL-C, and non-HDL-C.Similarly, DBP ≥ 90 mmHg was associated with high TC, LDL-C, non-HDL-C, TG, and TC/HDL-C ratio and TG/HDL.At the same time, a positive diagnosis of hypertension was associated with elevated TC and low HDL.BMI and WC were significantly associated with all lipid markers evaluated (see Table 4).

Discussion
At the end of the study, it was possible to establish the frequency of dyslipidemias and related associations.According to the data obtained, 225 (70.5%) (122 (54.2%) females vs. 103 (45.8%) in males, p value < 0.001) of the study participants had some type of dyslipidemia.These numbers are nearly 2-folds higher than the WHO 2008 estimate of 39% (40% women vs. 37% men) [21].More importantly, some studies suggest that the prevalence of dyslipidemia in adults in Africa is between 15% and 50% [15,22].Due, in part, to differences in definition criteria, published prevalence estimates of dyslipidemia often vary.In addition, the scarcity of population-based studies on the burden of dyslipidemia in elderly populations in SSA limits the comparisons of our results.These notwithstanding, data from general populations have reported a prevalence of 67.3% in South Africa [23] and 71.3% in Uganda [24].Unfortunately, the prevalence of dyslipidemia was nearly similar or even higher than what has been observed in some jurisdictions outside SSA-India (79%) [25] and China (39.9%) [26].
In terms of individual lipid and lipoprotein markers, the hierarchy of dominance was as follows: high TC (51.2%) > high LDL-C (43.7%) > low HDL-C (28.2%) > high TG (27.6%).The mean values of TC, LDL-C, non-HDL-C, TG, HDL-C, TC/HDL-C, and TG/HDL-C were also high with a significant disparity between males and females.Equally high prevalence, particularly of TC, has been observed in other jurisdictions in SSA: >70% TC levels in individuals aged >70 years in Nigeria [27].When the mean values for lipids were analyzed, equally high magnitudes were observed.Another important observation was the fact that   the high burden of dyslipidemia coexisted with a high burden of abdominal obesity (43.1%), dysglycemia (16%), hypertension (28.5%), and physical inactivity.It must be noted that the WHO threshold (≥110 mg/dL) was used for FPG; as such, the use of the American Diabetes Association (ADA) threshold of 100 mg/dL would have yielded higher proportions of dysglycemia.In a nutshell, a significant proportion of the elderly has elevated risk of CVD.Although the proportion of participants presenting with obesity was marginal, and most were in the normal range for BMI, normal BMI coexisted often with abnormal WC.The coexistence of low levels of general obesity with a very high burden of abdominal obesity (abnormality in WC) is in line with studies across broad population ranges in SSA [28].
Furthermore, controversies regarding some of the lipid markers evaluated in this study persist.For instance, questions persist about the role of TC as a general mortality risk among the elderly (≥85 years) with some investigators suggesting that the relationship may be positive, negative, or null [29].A retrospective cohort study suggested that the elderly with the lowest quartile of TC (<175 mg/dL) and LDL-C (<100.4mg/dL) were at higher risk of all-cause mortality [30].Similarly, the importance of lipid ratios in evaluating CVD risk in geriatrics is underinvestigated.A similar situation exists with regard to mixed dyslipidemia.Regardless, it is our considered position that the high burden of CVD risk factors, particularly dyslipidemia, could partly explain the excess burden of CVD in the elderly in the setting.
In terms of sociodemographic characteristics, our results suggest that dyslipidemia and low HDL were independently associated with sex.Although this study was cross-sectional by design, the results were generally in keeping with those of the widely respected Framingham study [31].In their analysis of a randomly selected population of elderly patients aged ≥65 years, the authors noted that dyslipidemia was 1.8 (95% CI: 1.1-3.1)times higher in females compared to males [32].In general, some evidence suggests that women in LMIC subregions had higher TC than their counterparts in HIC.The contributory role of this difference to the global CVD risk of women in SSA has not been quantified, in part because of a lack of available historical data.Another lipid marker that had a strong association with sex was low HDL-C [17,33].The relationship is not unique and is well recognized in treatment guidelines such as NCEP-ATP which has a different threshold for men and women.
Regarding the analysis of physical activity and its relationship to disparate lipid disorders, our results suggest that low physical activity rate (particularly less than 1 hour of exercise per day) was independently associated with LDL − C ≥ 130 mg/dL, non − HDL ≥ 130 mg/dL, and total dyslipidemia.A strong relationship between low rates of physical activity and low HDL-C was also observed in the crosstabulation analysis.The results uncovered in this analysis are broadly similar to what has been articulated in established literature [34].In general, physical inactivity is associated with higher BP or cardiovascular deconditioning, worse cholesterol levels, poorer glucose metabolism, poorer mental health, and obesity in the elderly.Consequently, low physical activity was ranked as a leading risk factor for lost daily adjusted life years (DALY) in 2010 [5].According to some investigators, the elimination of risky behaviors such as physical inactivity would make it possible to prevent at least 80% of CVD [6].
To a large extent, the data relating low physical activity to aberrant lipid concentrations is augmented by the analysis showing a strong association between type of work and multiple lipid parameters.In our analysis, manual workers were less likely to present with lipid abnormalities.Furthermore, women were more likely to be unemployed (particularly in manual work) or engage in physical activity and had worse cardiometabolic outcomes (Lipids, FPG, BMI, WC, DBP, and SBP).These associations are generally supported by pathophysiological data which suggest that physical activity has multiple physiological, immunological, and metabolic effects [35].For instance, research suggests that vigorously contracting muscle can release large quantities of myokines with auto-, para-, and endocrine effects.The best-studied myokine is Interleukin-6 (IL-6) which, among a myriad of functions, stimulates β-cell proliferation, reduces stress induced β-cell apoptosis, and increases glucose transporter (GLUT4) translocation, thus, basal glucose uptake [36].Moreover, the welldocumented benefits of physical activity supersede molecular adaptations of working skeletal muscle and include    All the presented aOR are significant at p value < 0.05.Bold letters: p value < 0.05.Proportion of the population with lipid disorders considered in the analysis.HDL: high-density lipoprotein; LDL: low-density lipoprotein; TG: triglycerides; TC: total cholesterol; non-HDL: non-high-density lipoprotein cholesterol; FPG: fasting plasma glucose; BMI: body mass index; aOR: adjusted odds ratios; CI: confidence interval.pvalue = 0.05.Note that age, FPG, BMI, and WC were modeled as continuous variables.10 Journal of Lipids improvement of coronary endothelial function, enhancement of muscle strength and power, and reduction in the aging process or sarcopenia.By and large, it is clear from our analysis that physical inactivity and sedentary behavior differ by sex with women on the high-risk side.In a way, physical inactivity and sedentarism may partially explain the observed disparities in lipid health or other metabolic abnormalities observed in this population.The possibility has far-reaching implications for intervention efforts in this setting and requires further elucidation.Therefore, and to reiterate a general guideline, improving physical activity among the elderly population represents a cheap, realistic, and cost-effective pathway for improving lipid health in this population.An analysis of the frequency of consumption of traditional food and its association with specific lipid biomarkers was also attempted.Interestingly, infrequent consumption of traditional foods was associated with aberrant TC, LDL, and non-HDL-C concentrations.The observed relationship may be linked to the nature of traditional cuisine in the country.In general, the traditional Eritrean cuisine is low in energy density and high in fiber content and deemphasizes the heavy use of oils or fats.Reduced intake of energy-dense micronutrient-poor foods and high dietary intake of fiber is recommended by WHO as lifestyle targets for reducing obesity.By and large, the observed relationships are important in multiple respects.First and foremost, investigators have attempted to formulate culturally adopted diets which may be used by high-risk segments of the population to ameliorate CVD risk.Accordingly, the data suggests that in Eritrea, a country with a minuscule number of dieticians and where such efforts have not been undertaken; the solution or starting point appears to be a renewed emphasis on traditional cuisines.Second, the data confirms the much-touted possibility that dietary transition or westernization of diets among elderly communities living in urban centers in SSA is aggravating CVD burden.Despite the preliminary nature of these results, they point to an intriguing area for future inquiry-studies on the relationship between consumption of traditional foods and impact on lipid health or health in general.
In a separate analysis, we demonstrated a strong positive cross-sectional association between BMI, WC, and abnormalities in all the lipid markers.A linear-by-linear association was also prominent.However, associations between BMI and multiple lipid disorders were mostly attenuated in the adjusted logistic regression analysis.Interestingly, a strong association between WC (a surrogate marker of visceral adiposity), FPG, elevated DBP, and multiple dyslipidemias persisted in the multivariable logistic regression analysis.These associations have been observed in younger cohorts [25,36].From a more practical perspective, computed tomography (CT) studies have demonstrated that visceral adiposity is more closely related to lipid, glucose, and blood pressure abnormalities [37].The copresence of these abnormalities, the metabolic syndrome (also known as "the deadly quartet for CVD"), has therefore been explained by a shared pathophysiological process with abnormality in WC as the trigger [38].Importantly, studies have shown that reduction of visceral fat ameliorates the "deadly quartet" simultaneously.To circle back to our original point, WC and not BMI ≥ 25 kg/m 2 is a more important driver of CVD in the elderly population in this setting.
Although the importance of WC is well documented and appears to play a prominent role in this setting, its use is not well established in Eritrea.This cheap-to-measure marker is rarely recorded by clinicians, and awareness regarding the dangers of high WC values is low.Instead, the use of BMI, even in the elderly, prevails.About the latter issue, we have to note that although BMI has a good positive correlation with total adiposity and with morbimortality from many diseases [39], the exclusive use of BMI in the elderly has been questioned owing to its poor correlation with visceral adiposity [40].Higher central adiposity at normal BMI levels is a common presentation in the elderly particularly in those with some degree of sarcopenia.However, we have to concede that the absence of WC cut-off for the region militates against the widespread use of WC.This notwithsatnding, the simultaneous use of BMI and WC, a prescription in many treatment guidelines, should be encouraged in this setting.

Study Limitations and Strength.
Understanding the burden, distribution, and factors associated with dyslipidemia in elderly populations is of paramount importance.The importance of such studies is magnified in countries in SSA where very little is known about the health of the elderly populations.This has undermined population-level efforts targeted at prevention or health education.In this regard, it is our opinion that this study, although preliminary, can contribute substantially to a better understanding of the burden of dyslipidemias and associated risk factors in this jurisdiction.This aside, some limitations must be mentioned.The sample represents a group of elderly participants working or residing within specific residential estates in Asmara, Eritrea.As such, it can be argued that the sample was not fully randomized.Nevertheless, this mode of data acquisition is superior to outpatients' data-a method that is prominent in most studies from the region.Importantly, and for the most part, the subjects presented with sociodemographic and clinical characteristics that were broadly similar to the general elderly population.Additional methodological issues that can undermine our results were the cross-sectional nature of the study and the fact that subclinical CVD events or comorbidities were not evaluated.Importantly, the use of hemoglobin A 1C to verify cases of DM or more granular characterization of physical activity and traditional food regimens could have added value to the study.For example, physical activity measurement instruments generally omit domestic activities such as cooking, cleaning, and childcare in physical activity computations.This was the case in this study.Lastly, miscoding, clerical errors, and unverifiable responses by respondents may also be limiting.Indeed, older adults are prone to forgetfulness and often overestimate their level of physical activity.

Conclusions
Several important findings emerged from this study.First, a large proportion of the study participants had some type of dyslipidemia (70.5%).All the study participants were not on any medication, and all were unaware of their condition.

Journal of Lipids
The most dominant dyslipidemia was high TC (51.2%), followed by low LDL-C (43.7%), low HDL-C (28.2%), and high TG (27.6%).Regarding NCEP-ATP III risk strata, 18.5%, 14.5%, 28.2%, and 12.9% were in high or very highrisk categories for TC, LDL-C, TG, and HDL-C, respectively.The most predominant mixed dyslipidemia was the TC +LDL-C combination.At the same time, 60 (17.5%), 103(21.6%),35 (11.0%), and 16 (5.0%)had abnormalities in 1, 2, 3, and 4 lipid abnormalities.Based on the study results, it is our opinion that a large proportion of the study is at risk of developing or aggravating preexisting CVD.Women were disproportionately affected across most of these categories.Furthermore, stepwise multivariate modeling demonstrated that the frequency of dyslipidemia was associated with sex, daily physical activity, increasing FPG, and BMI.Apart from the poor lipid health, a large proportion of the study participants had dysglycemia and elevated BP or hypertension.The proportion of the study participants presenting with abdominal obesity, a powerful surrogate marker of CVD, was also high.Taken together, these observations call for concerted effort directed at scaling up early recognition and treatment, including optimal pharmacological and nonpharmacological therapy at all levels of care.

Table 3 :
NCEP-ATP III based characterization of lipid disorders in the elderly in Asmara, Eritrea.

Table 2 :
Frequency of dyslipidemia and isolated and mixed dyslipidemia.

Table 4 :
Single and mixed dyslipidemia and some associated factors in the elderly in Asmara, Eritrea.

Table 5 :
Factors associated with dyslipidemia from a univariate logistic regression model.