The departments of Mono and Couffo are located in the South-West of Benin. Their populations in 2015 were estimated at 525,025 and 786,409 inhabitants respectively [14]. The department of Mono covers an area of 1605 km² and is composed of six municipalities and 276 villages. The branches of activity in the department of Mono are agriculture, fishing, hunting, trade, catering, accommodation and manufacturing industry. The department of Couffo covers an area of 2404 km² and is made up of six municipalities with a total of 367 villages. Its populations are mainly involved in agriculture, livestock and fishing. They also carry out commercial activities and processing of agricultural products and handicrafts. Industry is almost non-existent [14].

The type of study was a cross-sectional descriptive study with an analytical focus. The study population consisted of all adults aged 18 to 69 years in the Mono and Couffo departments corresponding to the WHO age group for STEPS surveys. As inclusion criteria, subjects residing for at least 6 months in the intervention area and who gave their consent were included in the study. As exclusion criteria, subjects with major mental disorders who could not answer the questions were excluded from the study.

With respect to sampling, the sample size was estimated at 2499 participants in both departments. It was calculated using the formula designed by WHO for the calculation of the sample size in STEPS surveys. The source population was subdivided into 8 age/sex brackets (4 brackets for males and 4 for females). The size calculated according to the Schwartz formula was multiplied by 8 for the age/sex brackets and divided by 0.8 (80% response rate).

Schwartz’s formula: n = k ∗ Z α 2 ∗ p ∗ ( 1 − p ) i 2 and N = n ∗ Ω T

(n (Schwartz formula sample size), α = 0.05 (first species risk), Zα = 1.96, k = 1 (cluster effect),p = 50% (multiple RFs), i = 6.2% (desired precision), N (expected sample size), Ω = 8 (number of age/sex brackets), T = 80% (response rate).

n = 1 ∗ 1.96 2 ∗ 0.5 ∗ 0.5 0.062 2 = 249.9     N = n ∗ Ω T = 249.9 ∗ 8 0.8 = 2499

participants rounded up to 2500.

Data were collected using a three-stage random sampling technique. The sampling frame for the two departments involved was provided by the National Institute of Statistics and Economic Analysis. The first stage randomly selected 125 Enumeration Areas (EAs). In the second stage, 20 households per EA were drawn by systematic random sampling after enumerating all households in the EA. In the third stage, one individual per household drawn was selected using the WHO KISH method, either directly by the Personal Digital Assistant (PDA) or by manual KISH table. A structured individual interview was conducted with the participant. After the interview, each drawn participant was given physical measurements (weight, height, waist circumference, hip circumference, and blood pressure) on the same day at home and biological measurements (blood glucose and cholesterol) the next morning on an empty stomach at a health center or a place set up for this purpose in the EA. All these data were entered into the PDAs and on the participants’ results sheet. At the end of the day, the completeness and consistency of the data were checked by the supervisors.

The collection tools consisted of : WHO STEPS instrument (last French electronic version V.3.1) with detailed information on socio-demographic factors, behavioral and biological risk factors and history of NCDs, BOSO MEDICUS UNO brand automatic digital blood pressure monitors with medium and wide cuff meeting WHO standards, SECA brand mobile blood pressure scales, EKS Professional brand adult mechanical scales, flexible graduated non-stretchable tapes, Cardiocheck^®allowing measurement of both blood glucose and total cholesterol with blood glucose/cholesterol test strips. Physical and biological measurements were integrated and installed in the PDAs. These tools were tested, corrected and adapted prior to data collection (Annex 1).

The collection team consisted of 20 trinomials of investigators, each comprising two junior epidemiology technicians and a paramedic (nurse or laboratory technician), all of whom were trained in STEPS. The epidemiologists were responsible for enumerating households and mapping the EAs and administering the questionnaire (STEPS 1) using the PDA. The paramedics were responsible for taking physical and biological measurements (STEPS 2 and 3). The teams were supervised by senior epidemiologists and the National Statistics Institute managers. Data collection lasted two months.

The dependent variable was high cholesterol. The independent variables were exposures to socio-demographic (age, gender, marital status), cultural and economic factors (ethnicity, occupation, income, education, residence (urban/rural)), behavioural factors (tobacco use, harmful alcohol use, low fruit and vegetable (FV) consumption, physical inactivity), clinical factors (obesity, diabetes, high blood pressure (HBP)).

Total cholesterol was measured from a drop of capillary blood taken from the finger with a Cardiocheck^® measuring device, a threshold ≥ 6.2 mmol/L (≥ 240 mg/dL) was used to define total hypercholesterolemia [15] [16]. Tobacco consumption was relative to the reported amount of tobacco consumed in the 12 months prior to the survey and the mode of consumption (smoked, chewed). Alcohol consumption was quantified in standard glass containing 10 g of ethanol. It was considered harmful if at least 4 standard glasses were taken by women and 6 standard glasses by men per day. It was abusive when it was greater than or equal to 2 standard glasses for women and 4 standard glasses for men per day [15]. The information on FV consumption was obtained from the declarations following the explanations given to the respondents in relation to the self-quantification using portions of fruit and/or vegetables. The portion was defined as a mass of 80 g of fruit and/or vegetables. The interviewers brought with them equivalent bowls to facilitate comparison [2]. Thus, insufficient consumption of FV was defined as consumption of less than five servings of FV per day. Low physical activity (PA) was indicated by a series of questions on the collection tool. Inadequate PA was defined as a low level of PA corresponding to less than 30 consecutive minutes of moderate exercise or equivalent to 20 consecutive minutes of intense exercise for at least 5 days per week [15]. Obesity was defined by a body mass index (BMI) ≥ 30 kg/m². Abdominal obesity was defined by a waist circumference ≥ 80 cm in women and 94 cm in men [17]. A capillary blood glucose test was performed on the participant’s finger with Cardiocheck^®. A conversion factor of 1.1 (plasma to capillary blood glucose ratio) was applied to the random plasma capillary glucose threshold of 11.1 mmol/L [18]. Moderate fasting plasma glucose was defined by fasting plasma glucose ≥ 111 mg/dL and <126 mg/dL and type 2 diabetes by fasting plasma glucose ≥ 126 mg/dL. HBP was defined by systolic blood pressure (SBP) ≥ 140 mmHg and/or diastolic blood pressure (DBP) ≥ 90 mmHg or the presence of medical antihypertensive treatment within 15 days prior to the survey [15].

Data analysis was done according to STEPS recommendations using Epi info7.1.5.0 and SPSS20. The Chi-square test was used to compare frequencies and the Student test to compare means (if the quantitative variables have a normal distribution). The difference was statistically significant for a p-value less than 0.05. Results were presented according to the STROBE recommendations [15].

For ethical and deontological considerations, the protocol was submitted to the National Committee for Ethics in Health Research for ethical advice several months prior to data collection. The free and written informed consent of each participant submitted to the study was obtained as well as the authorization of the local authorities. The personal data of the participants were made anonymous.

The survey was conducted with 1499 female subjects versus 991 male subjects, for a sex ratio of 0.66. The mean age was 36.1 ± 12.8 years. The age range most represented was 30 to 44 years (39.3%). Of the various ethnic groups found, Adja and related were the most represented (79.3%). Married people (79%) far exceeded other marital status. The self-employed were the most numerous (76.9%). The majority of respondents (68.2%) lived in rural areas. Subjects with no formal education were in the majority (61.2%). Respondents with an annual income of ?13 or more were predominantly represented (49.1%) (Table 1).

The mean cholesterol level observed in the population was estimated at 153.20 mg/dL. Of the 2490 subjects surveyed, 116 had hypercholesterolemia, i.e. an overall prevalence of 4.70% (95% CI: [3.88 - 5.54]) with 3.4% for men and 5.5% for women (p = 0.018).

The prevalence of hypercholesterolemia was three times higher in the Mono (8.5%) than in the Couffo (2.9%). Figure 1 and Table 2 show that the municipalities of Grand-Popo (12.5%) and Comey (8.1%) in the Mono department had more individuals with hypercholesterolemia.

Socio-demographic factors

Hypercholesterolemia was more prevalent in females (5.5%) (p = 0.018) and in the 60 - 69 age group (8.4%) compared to 2.7% in the 18 - 29 age group (p = 0.002). Hypercholesterolemia, which appeared high in certain categories of occupations in particular, the unemployed (14.3%) and volunteers (9.1%) is only apparent (p = 0.094).

Behavioural factors

Hypercholesterolemia was more prevalent among subjects who reported harmful alcohol consumption (15.4%) than their counterparts who did not drink (4.5%), with a statistically significant association with the Fisher Test (p = 0.047). The prevalence of hypercholesterolemia among FV consumers was 5.1% with a statistically significant association (p = 0.028).

Biological/clinical factors

We observed a 6.8% prevalence of hypercholesterolemia in HBP subjects with a statistically significant association (p < 0.001). We found a statistically significant association between overweight, obesity and hypercholesterolemia (p < 0.001).

Multivariate analysis

In multivariate analysis, after adjustment, harmful alcohol consumption (p = 0.023), HBP (p < 0.002), overweight (p < 0.023), obesity (p = 0.043) and type 2 diabetes (p = 0.033) were the factors significantly associated with the presence of hypercholesterolemia. Those who drank harmful amounts of alcohol were 1.5 times more likely to have hypercholesterolemia. Similarly, HBP increased the risk of high cholesterol by 1.12 times. In addition, the risk of developing high cholesterol was increased by 1.89 in overweight subjects. Obesity increased the risk of hypercholesterolemia by 1.66 times. This risk was 2.77 times greater in subjects with type 2 diabetes (Table 3).

¹Odds Ratios; ²Odds Ratios adjusted; ³Fruit and Vegetable; ⁴High Blood Pressure.

The Adja and related people, the most represented (79.3%) in the study, constitute the majority ethnic group in the Couffo department. They are predominantly from rural areas (68.2%). The Couffo department tends to have a less modern profile than that of Mono when considering at least the range of occupations in the two departments as mentioned in the study setting, which could explain the lower levels of hypercholesterolemia in the latter department as reported by some authors [8] [19]. The average age of 36.1 ± 12.8 years and the predominance of women who are mostly farmers in this department are characteristics that differentiate them from the populations of the Mono department [12]. The municipalities of Mono with the highest levels of hypercholesterolemia, Grand-Popo (12.5%) and Comey (8.1%), have the characteristics of being the most developed compared to the Couffo department, crossed by an interstate road with the Atlantic Ocean as a southern border and which could be in favor of such results [5] [8] [20].

When comparing the hypercholesterolemia observed in both departments in 2008 (previous STEPS survey) [12] of 1.2% (Mono) and 1.8% (Couffo) with that obtained by the present STEPS survey (2015), which is 8.5% and 2.9% respectively, an upward trend can be observed in both departments. This progression is slower in the department of Couffo than in the department of Mono for which the characteristics of the two departments in terms of development profile could be an explanatory factor according to several authors [5] [8] [20].

Univariate analysis showed a greater hypercholesterolemia in females (5.5%) and less in males (3.4%), in urban (6.2%) than in rural (4.0%) areas with a statistically significant association (p = 0.018; p = 0.013 respectively). Education level was significantly associated with hypercholesterolemia (p = 0.029). These same findings were made in the 2008 STEPS from Benin and 2010 STEPS from Togo, with the difference that both sexes were reached in the same proportions (0.9%) in 2010 [12] [21].

The results of the national nutrition and health study conducted in 2006-2007 in France and published in 2013 showed that 19.6% of adults had hypercholesterolemia. The most affected were the 45/65 year olds (30%) [22]. In Canada, more than 40% of subjects aged 20 to 79 years have a cholesterol level that is harmful to health without any significant gender differences [23]. The prevalence of elevated total cholesterol levels increased with age. It was also higher in urban (6.2%), common-law union (33.3%), separated (9.1%) and widowed (7.4%) individuals, but less so in married, post-graduate and other less educated individuals. This may be indicative of lifestyles with similar risk factors in these groups. In the study by Song et al. [20], education was not associated with hypercholesterolemia.

In multivariate analysis in our study, none of these factors were found to be significantly associated with hypercholesterolemia (Table 3). A detailed analysis of these factors will provide more precise answers for appropriate actions on the determinants of hypercholesterolemia in this region.

In univariate analysis, hypercholesterolemia was higher in subjects reporting harmful alcohol consumption (15.4%) and those with low FV consumption (5.1%) with a statistically significant association (p = 0.047; p = 0.028 respectively). Several other authors reported the same association [9] [24]. We were unable to find a relationship between hypercholesterolemia and physical inactivity or hypercholesterolemia and smoking. In their study, Song et al. [20] reported that high consumption of protein, pork and alcohol was positively associated with hypercholesterolemia, but high consumption of FV was not associated with hypercholesterolemia. In multivariate analysis, FV consumption was also not significantly associated with hypercholesterolemia in our study (Table 3).

Biological/clinical factors significantly associated with hypercholesterolemia in both univariate and multivariate analyses were HBP (p < 0.001), overweight (p < 0.001), obesity (p < 0.001) and type 2 diabetes (p < 0.001). These results have been reported by other authors [9] [24]. In their study, Song et al. [20] reported a statistically significant association of overweight and obesity with hypercholesterolemia.

Multivariate analysis identified a group of mutually reinforcing risk factors for high cholesterol composed of harmful alcohol consumption, HBP, overweight, obesity and type 2 diabetes [19]. These factors seem to be driven by a particular environment [9] [25] as reported by Marmot et al. [5] in the study of Japanese people living in Japan, Hawaii and California showing the importance of the living environment on the health of the individual.

This progression observed in our study compared to the 2008 study [12] would merit an intervention at the risk of seeing the populations of this region under a greater burden of hypercholesterolemia with other NCDs that could be caused by the factors associated with it [9] [26]. Such an intervention must take into account the more important trends observed in different living environments in order to contextualize actions. The experience reported by Kastarinen et al. [27] in the provinces of North Karelia and Kuopio (eastern Finland), and the Turku-Loimaa region (southwestern Finland) are examples to be followed and are likely to reduce the progression of hypercholesterolemia observed. A health monitoring strategy taking into account very precise indicators and depending on the context should accompany this intervention in order to know how to adapt to the evolution of the trends that will be observed.

Our study had the advantage of using the national protocol of the NCDs survey combining the results of STEPS 1, 2 and 3 from Benin. As for the external validity of the results, our sampling method following the WHO model in this matter allowed us to have a representative sample of the region studied. However, the results obtained cannot be generalized to all of Benin. As for the internal validity, many of the variables studied were evaluated on the basis of statements made by the respondents and could suffer from measurement bias so that they do not always reflect the reality. In spite of these limitations, the results we have achieved have made it possible to identify the prevalence of hypercholesterolemia and the factors favoring its appearance in the Mono and Couffo departments in 2015 with the role that the local context should play in the control measures.