To assess the prevalence, awareness, treatment and control of diabetes and to examine the relationship of obesity with raised blood glucose in the slums of Nairobi, Kenya.
To assess the prevalence, awareness, treatment and control of diabetes and to examine the relationship of obesity with raised blood glucose in the slums of Nairobi, Kenya.
We used data from a cross-sectional population-based survey, conducted in 2008–2009, involving a random sample of 5190 (2794 men and 2396 women) adults aged ≥18 years living in two slums – Korogocho and Viwandani – in Nairobi.
The prevalence (weighted by sampling and response rates) of diabetes was 4.8% (95%CI 4.0–5.7) in women and 4.0% (95%CI 3.3–4.7) in men. Less than a quarter of those found to have diabetes were aware of their condition among which just over half of men and three-quarters of women reported being on any treatment in the 12 months preceding the survey. Overall, fewer than 5% of all people with diabetes had their blood sugar under control. Obesity and overweight were significantly associated with increased odds (1.7, 95%CI 1.1–2.6) of raised blood glucose only among women while adjusting for important covariates.
The prevalence of diabetes in this impoverished population is moderately high, while the levels of awareness, treatment and control are quite low. In this population, obesity is an important risk factor for raised blood glucose particularly among women. Prevention and control strategies that target modifiable risk factors for diabetes and increase access to treatment and control in such disadvantaged settings are urgently needed.
Détermine la prévalence, la sensibilisation, le traitement et le contrôle du diabète, et examiner la relation entre l'obésité et la glycémie élevée dans les bidonvilles de Nairobi, au Kenya.
Nous avons utilisé les données d'une enquête transversale auprès de la population, menée en 2008–2009, auprès d'un échantillon aléatoire de 5190 (2794 hommes et 2396 femmes) adultes âgés de ≥ 18 ans et vivant dans deux bidonvilles de Nairobi, Korogocho et Viwandani.
La prévalence du diabète (pondérée par les taux d’échantillonnage et de réponse) était de 4.8% (IC 95%: 4.0 à 5.7) chez les femmes et de 4.0% (IC 95%: 3.3 à 4.7) chez les hommes. Moins d'un quart des personnes dépistées avec le diabète étaient au courant de leur statut. Parmi ces derniers, un peu plus de la moitié des hommes et trois quart des femmes ont déclaré avoir été sous un traitement durant les 12 mois précédant l'enquête. Dans l'ensemble, moins de 5% de toutes les personnes atteintes de diabète avaient leur glycémie sous contrôle. L'obésité et la surcharge pondérale sont significativement associées à une probabilité accrue (1.7; IC 95%: 1.1–2.6) de glycémie élevée chez les femmes, lorsque les variables importantes sont ajustées.
La prévalence du diabète dans cette population pauvre est modérément élevée, alors que les niveaux de sensibilisation, de traitement et de contrôle sont très faibles. Dans cette population, l'obésité est un facteur de risque important pour la glycémie élevée en particulier chez les femmes. Les stratégies de prévention et de contrôle qui ciblent les facteurs de risque modifiables pour le diabète et l'augmentation de l'accès au traitement et au contrôle dans les milieux défavorisés sont urgemment nécessaires.
Evaluar la prevalencia, los conocimientos, el tratamiento y el control de la diabetes, y examinar la relación entre la obesidad y niveles aumentados de glucosa en sangre, en las barriadas de Nairobi, Kenia.
Hemos utilizado datos de un estudio croseccional basado en la población, realizado entre el 2008–2009, con una muestra aleatoria de 5190 (2794 hombres y 2396 mujeres) adultos con ≥18 años viviendo en dos barriadas –Korogocho y Viwandani – en Nairobi.
La prevalencia (ponderada por el muestreo y la tasa de respuesta) de diabetes era un 4.8% (IC 95% 4.0–5.7) en mujeres y un 4.0% (IC 95% 3.3–4.7) en hombres. Menos de una cuarta parte de aquellos hallados con diabetes eran conscientes de su condición, entre los que un poco más de la mitad de los hombres y tres cuartas partes de las mujeres reportaron haber recibido tratamiento en los 12 meses anteriores a la encuesta. En total, menos del 5% de todas las personas con diabetes tenían la glucosa en sangre bajo control. La obesidad y el sobrepeso estaban significativamente asociados con una mayor probabilidad (1.7, IC 95% 1.1–2.6) de tener un nivel alto de glucosa en sangre solo entre las mujeres, cuando se ajustaba para covariables importantes.
La prevalencia de diabetes en esta población empobrecida es moderadamente alta, mientras que los niveles de conocimiento, tratamiento y control son bastante bajos. En esta población, la obesidad es un factor de riesgo importante para tener aumentado el nivel de glucosa en sangre, particularmente entre las mujeres. Se requiere urgentemente de estrategias de prevención y de control que se centren en factores de riesgo modificables de la diabetes y que aumentan el acceso al tratamiento y al control en emplazamientos desaventajados como estos.
In many low- and middle- income countries (LMICs), especially those in sub-Saharan Africa (SSA), there is an upsurge in the burden of non-communicable diseases (NCDs) such as diabetes, stroke and cancers(World Health Organization 2011). This upsurge is believed to be largely driven by rapid urbanisation and the attendant adoption of so-called western lifestyles –such as consumption of high-calorie diets and reduced physical activity – especially in the urban centres of LMICs (World Health Organization 2011). As a result of their already high burden of infectious diseases, such as HIV/AIDS and malaria, and weak healthcare systems, many countries are facing the prospect of the so-called double burden of disease (Oti 2012). At least 80% of deaths from NCDs occur in LMIC (World Health Organization 2011). NCDs tend to occur among younger and more economically active populations in LMICs than in high-income countries (World Health Organization 2011). This implies that economic development in such countries will be impeded by an uncontrolled NCD epidemic.
Available evidence suggests that diabetes mellitus (mostly type 2) will be a key contributor to the rise of NCDs in LMICs (World Health Organization 2011). For instance, the number of people affected by diabetes mellitus (DM) in SSA alone is projected to double from 12 to 24 million within the next two decades (Mbanya 2010). This increase will be driven in part by the rise in the prevalence of risk factors for DM in these settings. Specifically, several studies from SSA have shown that excessive body weight and obesity are independent risk factors for DM (Tuei et al. 2010). Obesity is reaching epidemic proportions globally – hence the coining of the term globesity, and the SSA region has not been spared. Although data on trends in obesity in the SSA region are limited, a study that reviewed Demographic and Health Survey (DHS) data from seven African countries over 10 years revealed rising trends in overweight and obesity among urban women (Ziraba et al. 2009). The prevalence of being overweight or obese had increased by about 35% between 1992 and 2005, when the DHS surveys were conducted. Even more worrying, the increase in overweight or obesity among the poorest urban women was about seven times higher than among the richest urban women (Ziraba et al. 2009).
Thus, upward trends in the levels of excessive body weight and obesity in SSA could contribute to an increasing prevalence of DM. Consequently, an increasing burden of DM will place further strain on the already overburdened health systems of LMICs. Patients with DM in SSA are faced with huge challenges in accessing basic health care such as steady access to diabetes medication, particularly insulin, at an affordable cost (Mbanya 2010; Hall et al. 2011). In SSA, patients with diabetes are often undertreated, resulting in poor glycaemic control and being at risk of developing avoidable complications and premature death (Hall et al. 2011). At the population level, DM remains under-diagnosed as screening/detection opportunities are few, and awareness about the disease is poor (Mbanya 2010; Hall et al. 2011). Increasing levels of an important risk factor for DM such as obesity combined with poor awareness, treatment and control present an ominous scenario in low resource settings.
We set out to determine if such a scenario exists in the urban slums of Kenya. A recent study in Nairobi, the capital city of Kenya, suggests that DM is a major problem in one slum (Ayah et al. 2013) and that obesity and overweight are important correlates for DM. Our study sought to assess the prevalence, awareness, treatment and control of diabetes, in addition to examining the relationship of obesity with DM in a large population-based cohort in two urban slums in Nairobi, Kenya, established in 2002 as part of a demographic and health surveillance system (Emina et al. 2011).
The study was a population-based cross-sectional survey of two slums – Viwandani and Korogocho in Nairobi, Kenya. Large sections of these slums are covered by the Nairobi Urban Health and Demographic Surveillance System (NUHDSS). The NUHDSS is operated by the African Population and Health Research Center (APHRC), a regional research institution headquartered in Nairobi. Details of the operation of the NUHDSS have been published elsewhere (Emina et al. 2011). In brief, the NUHDSS covers approximately 71,000 individuals residing in about 28,500 households in both slums. Every 4 months, demographic data including birth, death and migration status of every resident of the NUHDSS are updated. The NUHDSS therefore provides an up to date sampling frame for in-depth studies on various health and other social outcomes such as ours.
Our study used the sampling frame of adults (≥18 years) known to be resident in the NUHDSS area at the preceding round of data collection about 3 months prior to the beginning of the survey. We conducted stratified random sampling based on the WHO STEPwise protocol for chronic disease risk factor surveillance (World Health Organization 2013a). The STEPwise approach focuses on collecting core data on the established behavioural and physiological risk factors for NCDs that determine the major chronic disease burden. These risk factors include diabetes, elevated blood glucose, measures of excessive body weight and obesity. Based on the STEPwise protocol, a sample of 250 respondents in each of the following strata: sex (male and female), age group (18–29, 30–39, 40–49, 50–59, 60 years and over), slum of residence (Korogocho and Viwandani), was required. Therefore, in each stratum, a sampling frame was generated from the NUHDSS database and a computer-based program (STATA® statistical software) used to randomly select eligible individuals.
Details of the data collection methods have been described in detail elsewhere(van de Vijver et al. 2013). In summary, trained interviewers conducted interviews with study participants using a structured questionnaire that was translated into Kiswahili – the main local language in the study area. The field interviewers were trained in interviewing techniques, basic research ethics and in taking anthropometric and clinical measurements in accordance with the WHO STEPwise protocol. Measurements were taken using WHO recommended and validated equipment (Table 1). Overall, we considered several socio-demographic, behavioural and physiological risk factors for DM based on the framework developed by Wong et al. (Wong et al. 2005). Diabetes, our main outcome, is defined as random capillary blood glucose ≥11.1 mm or previously diagnosed by a health professional or confirmed by oral glucose tolerance test (OGTT) in accordance with WHO criteria (World Health Organization/International Diabetes Federation 2006). Pre-diabetes is defined as two-h (post-75 g oral glucose load ingestion) blood glucose of between 7 and 11.0 mm (World Health Organization/International Diabetes Federation 2006) (Table 2).
|SECA 201 circumference measurement tape||Waist and hip circumference in cm|
|SECA 874 flat scale electronic||Weight in kg|
|SECA 214 stadiometer transportable||Height in cm|
|OMRON M6 blood pressure machine||Systolic and diastolic blood pressure in mmHg|
|ACCUCHECK glucometer and test strips||Blood glucose in mm|
|Age||18–29; 30–39; 40–49; 50–59; 60 years and over|
|Education||No formal schooling; did not complete primary school; completed primary school; and completed secondary school or higher|
|Ethnicity||Luo; Luhya; Kamba; Kikuyu; and ‘Others’|
|Current daily smoker||Yes; No|
|Current daily consumer of alcohol||Yes; No|
|Adequate physical activity||Yes – engaging in ≥3 days of vigorous activity of at least 20 minutes per day or ≥5 days of moderate intensity activity or walking of at least 30 minutes per day (World Health Organization 2010); No|
|Insufficient fruits and vegetable intake||Yes – consuming <5 servings of fruits and/or vegetables per day (World Health Organization 2003); No|
|Body mass index||Normal(<25 kg/m2); overweight (25–29.9 kg/m2); and obese (≥30 kg/m2)|
|Waist circumference||Normal; high (>80 cm in women and > 94 cm in men)|
|Waist–Hip ratio||Normal; high (>0·80 in women and > 0·95 in men)|
|Hypertensive||Yes – systolic blood pressure >=140 mmHg and/or diastolic blood pressure >=90 mmHg or previously diagnosed; No|
|aRaised blood glucose||Normal (RBG<7 mm), pre-diabetes (RBG 7–11·0 mm), diabetes mellitus (RBG ≥11·1 mm or previously diagnosed or confirmed by OGTT.|
|Awareness||Self-reporting of any prior diagnosis of DM by a healthcare professional|
|Treatment||Treatment of DM was defined as receiving prescribed diabetes medication for management of DM at any time in the 12 months preceding the survey|
|Control||Proportion of patients reporting diabetes therapy with random blood glucose of <7 mm.|
Data were analysed using stata 12 (StataCorp. 2011. College Station, TX: StataCorp LP). All estimates were weighted for sampling probability (using the size of the stratum in the NUHDSS database as denominator) and for response probability (using the total number sampled per stratum as denominator). A composite weight, taking both weights into account, was then applied to all estimates. Study participants' socio-demographic, behavioural and physiological risk factors by BMI (normal versus overweight/BMI) were presented descriptively using chi-square and anova tests for significance at P < 0.05. The awareness, treatment and control levels for DM by gender are also presented (Table 2).
To examine the relationship between diabetes and obesity, we performed univariate random effects logistic regression analysis of the outcome variable over BMI stratified by gender and controlling for key covariates. To improve the power of our analysis, we collapsed certain subcategories within selected variables. Within the BMI variable, overweight and obesity are combined into one subcategory, while pre-diabetes and diabetes are combined into one category: raised blood glucose. Factors significant at P < 0.20 were then included in an adjusted analysis. All analyses were stratified by gender as it is expected that some associations might differ between women and men. Also, all analyses considered the outcome variable in its weighted form.
The study protocol was approved by the Kenya Medical Research Institute (KEMRI)/National Ethical Review Committee.
A total of 5190 adults aged 18 years and older were successfully interviewed. This constitutes overall response rates of 94% in Korogocho and 95% in Viwandani. Of those interviewed, 2396 (46%) were women and 2794 (54%) were men.
Table 3 shows the background characteristics of the study population by gender and diabetes status. Among both women and men, raised fasting glucose varied significantly by age (see also Figure 1), ethnicity, waist–hip ratio, waist circumference and hypertension status. Among men only, raised fasting glucose varied significantly by study site.
|Females (N = 2396)||P-value||Males (N = 2794)||P-value|
|Normal n (%)||Pre-diabetes n (%)||Diabetes n (%)||Normal n (%)||Prediabetes n (%)||Diabetes n (%)|
|Korogocho||1121 (91.7)||42 (3.5)||60 (4.9)||0.406||1126 (94.3)||41 (3.4)||27 (2.3)||<0.001|
|Viwandani||1057 (90.7)||54 (4.7)||54 (4.6)||1412 (88.4)||101 (6.3)||85 (5.3)|
|Age (in years)|
|18–29||1171 (93.3)||36 (2.9)||47 (3.8)||<0.001||1026 (91.7)||52 (4.7)||41 (3.7)||0.009|
|30–39||573 (90.6)||32 (5)||28 (4.4)||791 (93.2)||34 (4.1)||23 (2.7)|
|40–49||267 (88.6)||17 (5.6)||18 (5.8)||429 (88.3)||34 (7)||23 (4.7)|
|50–59||100 (86.6)||7 (6.3)||8 (7.1)||208 (86.7)||16 (6.5)||16 (6.8)|
|60 or older||71 (76.5)||7 (7.6)||15 (15.9)||86 (85.2)||7 (6.8)||8 (7.9)|
|No school||196 (90.6)||8 (3.6)||13 (5.8)||0.734||96 (91.4)||6 (5.5)||3 (3.1)||0.642|
|Not finished primary school||454 (91.2)||17 (3.4)||27 (5.4)||379 (91)||22 (5.3)||15 (3.7)|
|Primary school||1044 (90.6)||54 (4.6)||55 (4.8)||1228 (91.7)||58 (4.3)||53 (4)|
|Secondary school + higher||488 (92.2)||21 (4)||20 (3.9)||837 (89.6)||57 (6.1)||40 (4.3)|
|Kamba||434 (87.8)||27 (5.5)||33 (6.7)||0.027||665 (91.2)||24 (3.3)||40 (5.5)||0.005|
|Kikuyu||813 (89.3)||47 (5.1)||51 (5.6)||755 (89)||62 (7.3)||32 (3.7)|
|Luhya||262 (94.1)||9 (3.1)||8 (2.8)||363 (90)||23 (5.7)||17 (4.3)|
|Luo||266 (93.5)||8 (2.8)||11 (3.7)||385 (95.8)||10 (2.4)||7 (1.8)|
|Others||407 (95)||9 (2.1)||12 (2.9)||372 (90.4)||24 (5.9)||15 (3.7)|
|No||2161 (91)||98 (4.1)||115 (4.8)||0.701||2037 (90.7)||119 (5.3)||89 (4)||0.71|
|Yes||20 (93.2)||1 (4.2)||1 (2.6)||502 (91.6)||24 (4.3)||22 (4.1)|
|No||2121 (91.1)||95 (4.1)||113 (4.8)||0.668||2152 (91.6)||119 (5.1)||79 (3.4)||0.003|
|Yes||61 (90.5)||4 (6.1)||2 (3.4)||387 (87.3)||24 (5.3)||33 (7.4)|
|Insufficient fruit and vegetable intake|
|No||842 (90)||37 (3.9)||57 (6)||0.144||1318 (91)||76 (5.2)||55 (3.8)||0.865|
|Yes||1339 (91.7)||63 (4.3)||59 (4)||1222 (90.8)||67 (5)||57 (4.2)|
|Adequate physical activity|
|No||565 (89.6)||27 (4.2)||39 (6.2)||0.2766||174 (89.3)||10 (5.1)||11 (5.6)||0.215|
|Yes||1616 (91.6)||73 (4.1)||76 (4.3)||2364 (91)||133 (5.1)||100 (3.8)|
|Normal||485 (94.1)||23 (4.4)||8 (1.5)||<0.001||2198 (92.8)||122 (5.1)||48 (2)||<0.001|
|≥0.95 (men)/≥0.80 (women)||1679 (92.4)||76 (4.2)||63 (3.4)||339 (90.8)||21 (5.6)||14 (3.6)|
|Normal||915 (94.4)||35 (3.6)||19 (2)||<0.001||2393 (93.1)||126 (4.9)||53 (2)||<0.001|
|≥94 cm (men)/≥80 cm (women)||1267 (88.8)||64 (4.5)||96 (6.7)||146 (65.9)||17 (7.7)||59 (26.4)|
|Body mass index (kg/m2)|
|Normal||1440 (94.7)||44 (2.9)||37 (2.4)||<0.001||2210 (92.9)||118 (4.9)||51 (2.2)||<0.001|
|Overweight||535 (92.3)||25 (4.3)||20 (3.5)||293 (91.1)||21 (6.6)||7 (2.3)|
|Obese||205 (82.4)||30 (12.1)||14 (5.4)||34 (84.4)||3 (7.9)||3 (7.7)|
|No||1940 (92.8)||69 (3.3)||82 (3.9)||<0.001||2257 (91.8)||107 (4.4)||94 (3.8)||<0.001|
|Yes||242 (79.3)||30 (9.8)||33 (10.9)||283 (84.1)||36 (10.6)||18 (5.3)|
Table 4 shows the distribution of prevalence, awareness, treatment and control of DM by gender. A quarter of women with DM were aware of their condition as compared to 16% of men. Two-thirds of women and about half of men who were aware of having DM were on treatment. Less than a third of those on treatment, of either gender, had their blood sugar levels under control.
|Has diabetes (a)||Awarea(b)||Treatedb(c)||Controlledc|
|Females (n)||n = 162||n = 64||n = 48||n = 14|
|As a % of N (5190)||4.81 (3.95–5.67)||1.18 (0.75–1.62)||NA||NA|
|As a % of (a)||NA||24.6 (17.9–31.3)||19.1 (12.9–25.2)||5.01 (1.62–8.40)|
|As a % of (b)||NA||NA||77.5 (67.0–88.0)||20.4 (10.2–30.5)|
|As a % of (c)||NA||NA||NA||26.3 (13.4–39.2)|
|Males (n)||n = 136||n = 37||n = 24||n = 7|
|As a % of N (5190)||3.99 (3.27–4.72)||0.63 (0.33–0.92)||NA||NA|
|As a % of (a)||NA||15.7 (9.51–21.9)||8.44 (3.71–13.2)||3.98 (0.65–7.31)|
|As a % of (b)||NA||NA||53.8 (36.9–70.6)||25.3 (10.64–40.1)|
|As a % of (c)||NA||NA||NA||28.2 (8.79–47.6)|
Table 5 shows the distribution of respondent characteristics and prevalence of raised blood glucose stratified by BMI categories. The prevalence of raised blood glucose did not differ significantly by gender, education, smoking status, fruit/vegetable intake, physical activity or waist circumference across either BMI category. However, the prevalence of raised blood glucose increased significantly with age in both BMI categories. The prevalence of raised blood glucose was significantly higher among current smokers if they had a normal BMI. Hypertensives had a significantly higher prevalence of raised blood glucose whether they had normal weight or were overweight/obese. In the logistic regression analyses for women (Table 6), the odds of raised blood glucose were significantly increased in the overweight/obese, oldest age group (60 years and older) and among hypertensives while controlling for other factors. Women belonging to ‘other’ ethnicities had lower odds of raised blood glucose than the reference group (Kamba). Among men, BMI and all other factors were not significantly associated with raised blood glucose except hypertension. Male hypertensives had twice the odds of raised blood glucose as non-hypertensives after controlling for all other factors.
|BMI <25||BMI ≥25|
|n (%)||Prevalence (95%CI)||n (%)||Prevalence (95%CI)|
|Gender||P = 0.071||P = 0.538|
|Females||1217 (35.8)||5.3 (4.1–6.9)||855 (66.0)||10.7 (8.6–13.2)|
|Males||2179 (64.2)||7.1 (6.0–8.4)||441 (34.0)||9.6 (7.3–12.5)|
|Age (in years)||P = 0.006||P < 0.0001|
|18–29||1645 (48.5)||5.9 (4.5–7.6)||483 (37.3)||4.2 (2.1–8.1)|
|30–39||964 (28.4)||5.7 (4.2–7.6)||410 (31.6)||9.7 (6.8–13.6)|
|40–49||476 (14.0)||7.3 (5.6–9.5)||237 (18.3)||15.3 (11.9–19.5)|
|50–59||209 (6.2)||9.0 (6.9–11.5)||109 (8.4)||18.1 (14.0–23.0)|
|60 or older||101 (3.0)||13.4 (10.6–16.9)||57 (4.4)||23.1 (18.1–29.1)|
|Education||P = 0.284||P = 0.051|
|No formal schooling||195 (5.7)||6.1 (4.2–8.9)||87 (6.7)||13.5 (9.2–19.3)|
|Did not finish primary school||568 (16.7)||4.7 (3.4–6.5)||256 (19.7)||14.2 (10.7–18.7)|
|Completed Primary school||1662 (49.0)||6.8 (5.5–8.3)||590 (45.5)||8.5 (6.3–11.4)|
|Secondary school + higher||971 (28.6)||7.0 (5.3–9.2)||364 (28.1)||9.5 (6.6–13.6)|
|Ethnicity||P = 0.018||P = 0.002|
|Kamba||845 (24.9)||6.1 (4.5–8.3)||243 (18.8)||12.1 (8.3–17.4)|
|Kikuyu||1062 (31.3)||8.1 (6.5–10.1)||503 (38.8)||13.4 (10.6–16.8)|
|Luhya||436 (12.9)||7.3 (4.9–10.8)||192 (14.8)||5.5 (3.2–9.1)|
|Luo||490 (14.4)||2.9 (1.6–5.2)||157 (12.1)||9.4 (5.5–15.7)|
|Others||562 (16.6)||6.0 (4.1–8.7)||201 (15.5)||4.8 (2.7–8.5)|
|Current smoker||P = 0.491||P = 0.869|
|No||2933 (86.4)||6.3 (5.4–7.4)||1230 (94.9)||10.4 (8.7–12.3)|
|Yes||461 (13.6)||7.3 (5.1–10.3)||66 (5.1)||9.7 (4.7–19.0)|
|Current daily drinking||P = 0.018||P = 0.919|
|No||3039 (89.5)||6.1 (5.2–7.1)||1192 (92.0)||10.3 (8.6–12.2)|
|Yes||355 (10.5)||9.8 (6.8–13.9)||104 (8.0)||10.7 (5.3–20.1)|
|Insufficient fruit and vegetable intake||P = 0.983||P = 0.429|
|No||1618 (47.7)||6.5 (5.2–8.0)||533 (41.1)||11.1 (8.7–14.2)|
|Yes||1777 (52.3)||6.5 (5.3–7.9)||763 (58.9)||9.7 (7.7–12.2)|
|Adequate physical activity||P = 0.626||P = 0.3293|
|No||463 (13.7)||6.5 (4.5–9.3)||251 (19.4)||12.1 (8.4–17.0)|
|Yes||2930 (86.3)||6.5 (5.5–7.6)||1045 (80.6)||9.9 (8.2–11.9)|
|Waist–hip ratio||P = 0.351||P = 0.037|
|Normal||2250 (66.3)||6.8 (5.7–8.2)||446 (34.4)||7.7 (5.4–10.9)|
|≥0.95 (males)/≥0.80 (females)||1136 (33.5)||5.7 (4.4–7.3)||841 (64.9)||11.7 (9.6–14.1)|
|Waist circumference||P = 0.2749||P = 0.0039|
|Normal||2851 (84.0)||6.7 (5.7–7.8)||381 (29.4)||6.1 (4.0–9.4)|
|≥94 cm (males)/≥80 cm (females)||544 (16.0)||5.4 (3.9–7.6)||915 (70.6)||12 (10.0–14.3)|
|Hypertension||P < 0.0001||P < 0.0001|
|No||3076 (90.6)||5.6 (4.7–6.6)||1071 (82.7)||6.9 (5.3–8.8)|
|Yes||318 (9.4)||14.3 (11.0–18.3)||225 (17.3)||23.8 (19.2–29.1)|
|Body mass index (kg/m2)|
|Age (in years)|
|60 or older||4.30***||(2.82–6.55)||2.10*||(1.14–3.87)||1.90**||(1.26–2.88)||1.34||(0.81–2.21)|
|No formal schooling||1||1|
|Did not finish primary school||0.94||(0.62–1.42)||1.05||(0.56–1.97)|
|Completed Primary school||1.01||(0.68–1.48)||0.96||(0.53–1.73)|
|Secondary school† higher||0.82||(0.49–1.36)||1.22||(0.67–2.23)|
|Current daily drinking|
|Insufficient fruit and vegetable intake|
|Adequate physical activity|
|≥95 cm (males)/≥80 cm (females)||1.32||(0.79–2.22)||1.12||(0.62–2.03)||1.31†||(0.91–1.88)||0.84||(0.55–1.27)|
|≥94 cm (males)/≥80 cm (females)||3.61***||(1.94–6.71)||0.84||(0.49–1.44)||6.95***||(4.90–9.86)||2.00**||(1.28–3.14)|
We set out to determine the magnitude (prevalence, awareness, treatment and control) of diabetes and to examine the relationship of obesity with raised blood glucose in a random population of adult slum dwellers. Overall, the weighted prevalence of diabetes was about 4% and this did not differ significantly by gender. Overall levels of awareness, treatment and control for diabetes were quite low among men and women, although women fared slightly better than men. A total of 40% of women in our study were either overweight or obese, whereas fewer than 20% men were. The adjusted analysis shows that for women, being overweight/obese, 60 years or older, and being hypertensive were associated with significantly increased odds of raised blood glucose. For men, only being hypertensive was significantly associated with increased odds of raised blood glucose.
The overall prevalence of DM in our study is comparable to a previous study in Kenya that showed a prevalence of 4.2–5.3% (Christensen 2009). Specifically, a recent study in another major slum in Nairobi showed a prevalence of 5.3% (95% CI 4.2–6.4) (Ayah et al. 2013). In comparison with studies from other settings in SSA, the prevalence of diabetes in our study was higher than in a number of recent studies (Sobngwi 2002; Motala 2008; Maher 2010; Oladapo 2010) and similar or lower than in others (Aspray 2000; Erasmus 2001; Amoah et al. 2002; Nyenwe 2003; Alberts 2005; Balde 2007; Nshisso et al. 2012). The prevalence of diabetes across the subcontinent ranged from 3% in Benin to 14.5% in urban Democratic Republic of Congo (Mbanya 2010; Hall et al. 2011; World Health Organization 2013b). The prevalence of diabetes in SSA has been increasing over the past three decades (Mbanya 2010). However, with the exception of a recent study in one slum in Nairobi (Ayah et al. 2013), the evidence does not typically reflect the situation of diabetes in the slums of SSA, as these studies focused on rural and or urban populations in general rather than slum populations.
It is noteworthy that the prevalence of pre-diabetes in our study was 5.3% in women and 5.7% among in men. This is particularly significant because evidence suggests that up to 5–10% per year of people with pre-diabetes will develop diabetes (Tabak et al. 2012). This situation presents an opportunity for preventive interventions in the slum population, barring which a further rise in the prevalence of diabetes will be unavoidable. Lifestyle modification is the cornerstone of diabetes prevention and could provide up to 70% relative risk reduction among people with pre-diabetes (Tabak et al. 2012). However, with the low levels of awareness, treatment and control in the study population, chances are that the majority of those with pre-diabetes will remain unidentified and untreated, thus further aggravating the burden of diabetes in this population.
In terms of awareness, our findings are comparable to other studies in SSA. Generally, the level of awareness of diabetes is usually less than 50% (Hall et al. 2011) – ours was only about 20%. One study in rural Guinea found that 100% of those surveyed and diagnosed with diabetes were previously unaware of their diagnosis (Balde 2007). An exception was a study in urban South Africa where more than 50% were aware of having diabetes, comparable to findings in high-income countries (Levitt et al. 1993). This has been attributed to better health care and availability of opportunistic screening in urban South Africa. On the other hand, evidence from rural South Africa shows awareness levels as low as other countries in SSA (Motala 2008). As regards treatment and control, evidence from SSA is scanty. Our study found that less than a third of those on treatment had their blood glucose under control. This is comparable to a study in Cameroon, which found that 27% of those on treatment were under control (Ministry of Health 2004).
Overall, however, it is noteworthy that fewer than 5% of all people with diabetes in our study had their blood glucose under control. Poor control of blood sugar is associated with high risk of complications that in turn lead to high morbidity and mortality. Evidence suggests that chronic vascular complications such as retinopathy, nephropathy and neuropathy are common in SSA, and macrovascular complications such as stroke are also on the increase (Hall et al. 2011). Reasons for low levels of awareness, treatment and control are lack of screening opportunities, high cost of treatment, medication stock-outs, late diagnoses, poor adherence and preference for alternative medicines for the management of diabetes (Mbanya 2010; Tuei et al. 2010; Hall et al. 2011). These reasons usually reflect a general lack of prioritisation and investment of care for chronic NCDs and unresponsive healthcare systems across SSA (Mbanya 2010; Tuei et al. 2010; Hall et al. 2011). Slum populations remain particularly disadvantaged, and therefore, extra attention is needed to prevent and control the diabetes epidemic in these populations.
Our study found that a high BMI indicative of overweight/obesity was significantly associated with increased odds of raised blood glucose among women but not among men. There is evidence that obesity is increasing in SSA especially among women (Abubakari et al. 2008). Studies from Mali, Tanzania and Nigeria also show that obesity is associated with increasing risk of diabetes (Tuei et al. 2010). The reasons for the high rates of obesity among women in the slums may need further examination. Sociocultural perceptions of ideal body size may be a contributory factor. In African populations, large body size is perceived positively as a sign of wealth and good health, and attractive to the opposite sex (Siervo et al. 2006). As women come in contact with health service more frequently than men due to maternal and child health demands, there is an opportunity for health professionals to educate women about the health risks of obesity and overweight. Reproductive (family planning) and maternal health services should be integrated with primary preventive strategies (such as screening and lifestyle counselling) for diabetes and related conditions.
Our study found that the prevalence of diabetes increased with age among both men and women and was highest in the 60+ age group. This is consistent with other studies from SSA (Mbanya 2010; Hall et al. 2011). There were no significant gender differences in the prevalence of diabetes in our study population. This is in keeping with a few studies in SSA although other studies do show gender differences (Mbanya 2010; Hall et al. 2011). We also found that women in our study belonging to the broad group of ‘other’ ethnicities had a significantly lower risk of diabetes than the reference ethnic group. It is hard to explain this finding as this group is not homogeneous and comprises more than 30 other tribes in Kenya. One previous study in Kenya found that the ethnic group of Luos had the highest risk of diabetes owing to the higher total dietary energy intake (mostly from cereal grains) in that community (Christensen 2009). However, in our study, being Luo was not associated with increased odds of diabetes. Finally, we found that being hypertensive was significantly associated with increased odds of raised fasting glucose among both men and women. This finding is well established from previous studies across the world (Sowers et al. 2001). Hypertension is one of the components of metabolic syndrome, and three main factors have been proposed to contribute to hypertension in diabetes: hyperinsulinaemia, extracellular fluid volume expansion and increased arterial stiffness. Early treatment of hypertension is particularly important in patients with DM to prevent CVD and to minimise progression of renal disease and diabetic retinopathy. In patients with DM, the benefits of tight BP control may be as great or greater than the benefit of strict glycemic control (UK Prospective Diabetes Study Group 1998).
Our study is limited by the fact that some of our explanatory variables (such as smoking and physical inactivity) were based on self-reports which could have been misreported. However, self-reports remain useful and validated components of large surveys such as ours across the world. The accuracy of BMI as a measure of obesity is limited although its use remains recommended (Romero-Corral et al. 2008; Flegal et al. 2013). Our study was also limited by the fact that we did not measure or assess other important known risk factors for increased blood glucose such as underlying infection (HIV, TB) and medication (steroids, antiretrovirals) (Brown et al. 2005; Dooley & Chaisson 2009; Simmons et al. 2012). Neither did we collect objective information on type of or adherence to diabetes medication (e.g. HbA1c measurement) nor examine the respondents for manifestations of microvascular and macrovascular complications of diabetes. The main strength of our study is its large sample size and representativeness of the slum population, for whom data are often unavailable.
In conclusion, we found that diabetes was moderately prevalent in the slum population in Nairobi. Levels of awareness, treatment and control of diabetes are dismally low in the study setting. Obesity was significantly associated with increased odds of raised blood glucose especially among women. The presence of comorbidities such as hypertension also increased the odds of having raised blood glucose. Thus, it is imperative that health planners and policy makers in SSA pay attention to chronic NCDs such as diabetes especially in the often neglected but significant population of slum dwellers.
We acknowledge the contribution of APHRC's dedicated field and data management teams and the study participants who shared their time with us. We thank Prof. Joep M.A. Lange and Dr. Gabriela G. Gomez for academic support. We also acknowledge the Wellcome Trust UK for grant supports to the project and the Academic Medical Center Foundation (through the Amsterdam Institute for Global Health and Development) for staff-time to write this manuscript.