The Prevalence, Predictors and Associations of Hypertension in Sri Lanka: A Cross-Sectional Population Based National Survey
The Prevalence, Predictors and Associations of Hypertension in Sri Lanka: A Cross-Sectional Population Based National Survey
The Prevalence, Predictors and Associations of Hypertension in Sri Lanka: A Cross-Sectional Population Based National Survey
com/ceh
ISSN: 1064-1963 (print), 1525-6006 (electronic)
ORIGINAL PAPER
1
Diabetes Research Unit, Department of Clinical Medicine, Faculty of Medicine, University of Colombo, Sri Lanka, 2Oxford Centre for Diabetes,
Endocrinology and Metabolism, University of Oxford, United Kingdom, 3Department of Pharmacology, Faculty of Medicine, University of Colombo,
Sri Lanka, and 4Institute of Health and Biomedical Innovation, Queensland University of Technology, Brisbane, Queensland, Australia
Abstract Keywords
We studied the community prevalence, patterns and predictors of hypertension in a large sub- Adults, blood pressure, hypertension,
population of South Asian adults with a view of identifying differential risk factors. Data were prevalence, Sri Lanka
collected between years 2005–2006 and 5000 adults were invited for the study. The sample size
was 4485, and about 39.5% were males. Mean systolic and diastolic blood pressures were History
127.1 19.8 mmHg and 75.4 11.3 mmHg, respectively. Age-adjusted prevalence in all adults,
males and females was 23.7%, 23.4% and 23.8%, respectively. Urban adults had a significantly Received 6 June 2013
higher prevalence of hypertension than rural adults. In the binary logistic–regression analysis, Revised 1 August 2013
male gender (OR: 1.2), increasing age, Sri Lankan Moor ethnicity (OR: 1.6), physical inactivity Accepted 19 September 2013
(OR: 1.7), presence of diabetes (OR: 2.2) and central obesity (OR: 2.3) all were significantly Published online 16 January 2014
For personal use only.
associated with hypertension. In conclusion, nearly one-third of the Sri Lankan adult population
is hypertensive. Hence, public health initiatives should encourage healthier lifestyles with
emphasis on preventing obesity and increasing physical activity.
regarding the community prevalence of hypertension based Height was measured using Harpenden pocket stadi-
on a national-survey from Sri Lanka or other regional South ometers (Chasmors Ltd, London, UK) to the nearest 0.1 cm
Asian countries. The present study aims to describe the according to standard methods. Body weight was measured
community prevalence, patterns and predictors of hyperten- in indoor light clothing to the nearest 0.1 kg using a
sion in a large sub-population of South Asian adults with a SALTER 920 digital weighing scale (Salter Ltd, Tonbridge,
view of identifying differential risk factors, which may lead to UK). Waist circumference was measured at midway between
improved preventive measures and care for hypertensive iliac crest and lower rib margin at the end of normal
patients in the South Asia region. expiration using a plastic flexible tape to the nearest 0.1 cm.
Similarly, the hip circumference was also measured at the
Materials and methods widest part of the buttocks at inter-trochantric level to the
Study population and sampling nearest 0.1 cm. Body Mass Index (BMI) was calculated as
weight in kilograms divided by height in meters squared
Data were collected as a part of a wider cross-sectional (kg/m2). An interviewer-administrated questionnaire was
national study conducted in seven of the nine provinces in used to obtain socio-demographic details such as age,
Sri Lanka between August 2005 and September 2006 gender, area of residence, ethnicity, level of education and
Clin Exp Hypertens Downloaded from informahealthcare.com by QUT Queensland University of Tech on 05/27/14
(Sri Lanka Diabetes and Cardiovascular Study, SLDCS). household income. Level of education was classified in to
The Northern and Eastern provinces of the country affected four categories: no formal education, primary education
by the war during the study period were not included. (year 1–5), secondary education (year 6–13) and tertiary
Detailed sampling has been previously reported (12). In brief, education (under-/post-graduate). Similarly for the purpose
the total sampling frame was approximately 14 million adults of analysis monthly household income in Sri Lankan Rupees
18 years of age living in 12 018 ‘‘village officer’’ units (the (LKR; 1 US$ ¼ 125 LKR) was classified as follows: LKR
smallest government administrative unit). The total sample 6999, LKR 7000–24 999, LKR 25 000–49 999 and LKR
size was 5000 adults. A cluster was defined as 50 households 50 000. Physical activity during the past week was assessed
and the total sample of 5000 adults were recruited from using, the short version of the International Physical Activity
100 such clusters. The number of clusters for individual Questionnaire (IPAQ) administered by an interviewer. Urban
province was determined using a probability-proportional-to- and rural sectors were defined according to the classification
size (PPS) technique according to the total provincial of the Sri Lankan government. Blood samples were
For personal use only.
population. In each province, the clusters to be studied were collected from participants for the estimation of overnight
selected by a computer-generated random number list from fasting blood glucose, 2-h post prandial blood glucose, total/
the ‘‘Village Office Units’’ in the particular province. LDL/HDL cholesterol and triglycerides. Details of blood
Subsequently the voters’ registers in each cluster (updated sample collection and biochemical analysis are described
annually for those aged 18 years) were used to randomly elsewhere (12).
select the first household and a uniform criterion was used to
select the remaining 49 households in a cluster. The selected
households were visited by the study team and one subject Statistical analyses
was randomly selected. Relevant data of 5000 non-institutio- All data were double-entered and cross checked for consist-
nalized adults are presented here. Ethical approval was ency. Data were analysed using SPSS version 14 (SPSS Inc.,
obtained from the Ethical Review Committee, Faculty of Chicago, IL) and Stata/SE 10.0 (Stata Corporation, College
Medicine, University of Colombo, Sri Lanka. Station, TX) statistical software packages. The significance
of the differences between proportions (%) and means were
tested using z-test and Student’s t-test or ANOVA, respect-
Definitions, measurements and data collection
ively. Direct standardization of prevalence was performed
Data collection was carried out by a field team of trained according to the population data published by the United
medical graduates and nurses. Seated BP was measured after Nations population division and for the ‘‘WHO new world
a 10-min rest with Omron IA2 digital BP monitors (Omron population’’ (16). A binary logistic regression analysis was
Healthcare, Alexandra Technopark, Singapore). Subjects were performed using the dichotomous variable hypertension
considered to have ‘‘diagnosed hypertension’’ if they had (0 ¼ absent and 1 ¼ present). The independent co-variants
been previously diagnosed at a government hospital or by a (reference category) included in the binary logistic regres-
registered medical practitioner or if they were on anti- sion analysis were the following: gender (‘‘female’’), age
hypertensive treatment. New cases (‘‘undiagnosed hyperten- category (‘‘530’’ years), area of residence (‘‘rural’’),
sion’’) were diagnosed when the average of two resting seated ethnicity (‘‘Sinhalese’’), physical activity (‘‘active’’), pres-
BP readings, separated by 5 min were 140/90 mmHg. Pre- ence of diabetes and central obesity. A stepwise forward
hypertension was defined as a SBP of 120–139 mmHg and/or entry method was used to enter the variables. For each
a DBP of 80–89 mmHg A standard bladder (12–13 cm long independent variable with more than two categories, dummy
and 35 wide) was used but a larger and a smaller bladder were variables were created. The first category was taken as the
available for fat and thin arms, respectively. Diabetes was reference category. A similar binary logistic regression
diagnosed according to the American Diabetes Association analysis with above dependant and independent variables
(13) and WHO criteria (14). Central obesity was defined as was also performed separately for both males and females.
waist circumference 490 cm for males and 480 cm for In all statistical analyses, p values 50.05 were considered
females according to Asian cut-off values (15). significant.
DOI: 10.3109/10641963.2013.863321 Prevalence and predictors of hypertension in Sri Lanka 3
Results was significantly higher than the mean SBP and DBP of
Sinhalese and Tamils when considering all adults and
The original sample size was 5000, of which 4485 adults
females; however, this difference was not significant in
consented to take part in the study (response rate – 89.7%),
males (Tables 1 and 2). When considering the level of
39.5% (n ¼ 1772) were males. Mean age was 46.1 15.1
education, the highest and the lowest mean SBP in all adults
years. Socio-demographic data of the study population is
and females was observed in those with no formal education
described elsewhere (12). In summary, 21.3% (n ¼ 955) were
and tertiary education, respectively (Table 1). However,
urban residents, 86.4% (n ¼ 3877) were Sinhalese by ethni-
such a distinct pattern was not observed between the mean
city, while 6.7% (n ¼ 299) were Tamils and 6.6% (n ¼ 298)
SBP and level of education in males. The mean DBP did not
were Moors. Data from 4450 subjects aged 18 years were
demonstrate a significant variation with the level of education
used for prevalence estimations, and 35 subjects in whom
(Table 2). The relationship between mean SBP and DBP with
BP readings were not available were excluded. Mean SBP
monthly household income did not demonstrate a distinct
in all adults was 127.1 19.8 mmHg and the mean
pattern in all adults, males and female (Tables 1 and 2).
DBP was 75.4 11.3 mmHg. The mean SBP of males
The mean SBP/DBP of those who were physically active
(129.0 19.4 mmHg) was significantly higher than in females
(125.1 18.2/74.6 11.0) was significantly lower than those
Clin Exp Hypertens Downloaded from informahealthcare.com by QUT Queensland University of Tech on 05/27/14
Table 1. Mean SBP and socio-demographic characteristics in all adults, males and females.
Table 2. Mean DBP and socio-demographic characteristics in all adults, males and females.
Table 3. Correlation of systolic/diastolic blood pressures and age, anthropometric and biochemical parameters in all adults, males and females.
PPBS, post prandial blood sugar; LDL, low-density lipoprotein; HDL, high-density lipoprotein.
*Significant correlations (p50.05).
prevalence of hypertension in all adults was 23.7% (95% CI: hypertension was observed in Sri Lankan Moors, followed
22.4–25.0). The age-adjusted prevalence in males and females by Sinhalese and Tamils (Table 4). A similar pattern was also
were 23.4% (95% CI: 21.4–25.4) and 23.8% (95% CI: 22.2– observed in both males and females (Table 4). In all adults,
25.4), respectively, and while there was no significant the highest prevalence of hypertension was observed in those
difference between genders. Urban adults had a significantly with the highest level of education (‘‘tertiary education’’), this
higher prevalence of hypertension than their rural counter- was also observed independently in males, but not in females
parts and this was also observed independently in both males (Table 4). Pre-hypertension was present in 12% of males and
and females (Table 4). The mean age of those with 11.4% of females.
hypertension (56.0 13.5 years) was significantly higher Those with hypertension had a significantly higher weight,
than those without hypertension (42.4 14.0 years) BMI, waist circumference, hip circumference, waist–hip ratio,
(p50.001). The prevalence of hypertension increased with SBP and diastolic BP, fasting blood glucose, 2-h post prandial
increasing age in all adults, males and females (Table 4). In blood glucose, total and LDL cholesterol and triglycerides
the different ethnic groups, the highest prevalence of (Table 5). This was observed in both males and females
DOI: 10.3109/10641963.2013.863321 Prevalence and predictors of hypertension in Sri Lanka 5
Table 4. Age-standardized prevalence of hypertension in all adults, males and females.
*p Values – males versus females; NS, not significant; LKR, Sri Lankan Rupees.
a,b
Mean prevalence values with same superscripts in a column within the same group are significantly different (p50.05).
For personal use only.
Table 5. Mean values of clinical and biochemical parameters in those with and without hypertension.
Mean (SD)
Hypertensive Non-hypertensive p
Weight (kg) 55.5 (12.5) 52.3 (11.1) 50.001
Body mass index (kg m2) 23.1 (4.5) 21.2 (4.0) 50.001
Waist circumference (cm) 81.8 (11.8) 75.6 (11.3) 50.001
Hip circumference (cm) 91.1 (9.4) 88.2 (8.7) NS
Waist–hip ratio 0.9 (0.1) 0.8 (0.1) 50.001
Fasting blood glucose (mg/dl) 99.0 (33.2) 88.4 (24.9) 50.001
2-h post prandial blood glucose (mg/dl) 127.8 (62.6) 104.9 (46.3) 50.001
Total cholesterol (mg/dl) 220.2 (43.7) 201.7 (42.4) 50.001
LDL cholesterol (mg/dl) 144.6 (38.1) 132.0 (36.9) 50.001
HDL cholesterol (mg/dl) 46.8 (10.9) 46.8 (10.0) NS
Triglycerides (mg/dl) 143.5 (74.6) 113.6 (61.9) 50.001
independently (data not shown). The mean weekly total MET absent and 1 ¼ present) as the dependant factor and other
minutes of those with hypertension (3626.7 min/week) was independent variables are shown in Table 6. The overall
significantly lower than those without hypertension model was statistically significant as determined by the
(5107.1 min/week). A similar pattern was observed in both likelihood ratio test (2 ¼ 957.99, p50.001). The Cox &
males and females separately (data not shown). The preva- Snell R-Square and Nagelkerke R Square values were 0.194
lence of hypertension in the different physical activity and 0.280, respectively. The results indicate that male gender
categories of the IPAQ was as follows: ‘‘inactive’’ – 41.3% (OR: 1.2), increasing age, Sri Lankan Moor ethnicity (OR:
(95% CI: 37.0–45.8), ‘‘moderately active’’ – 31.9% (95% CI: 1.6), physical inactivity (OR: 1.7), presence of diabetes (OR:
29.4–34.6) and ‘‘active’’ – 22.7% (95% CI: 21.1–24.4). In 2.2) and central obesity (OR: 2.3) all were associated
patients with diabetes, the prevalence of hypertension was with significantly increased risk of developing hypertension
55.4% (95% CI: 51.0–59.6), which was significantly higher (Table 6). Similar results were seen independently in males
than in those without diabetes (23.6%, 95% CI: 22.3–24.9) and females (Table 6).
(p50.001). Hypertension was present in 40.1% (95% CI:
37.5–42.8) of participants with central obesity, while only
Discussion
21.8% (95% CI: 20.4–23.3) adults without central obesity
were hypertensive (p50.001). In this first community-based national survey on hypertension
The results of the binary logistic regression analysis in all from Sri Lanka, the age-adjusted prevalence in all adults
adults using the dichotomous variable ‘‘hypertension’’ (0 ¼ was 23.7% (males – 23.4%; females – 23.8%), while
6 P. Katulanda et al. Clin Exp Hypertens, Early Online: 1–8
Table 6. Binary logistic regression analysis in all adults, males and females.
*p50.001; #p50.01.
pre-hypertension was present in 12% of males and 11.4% of We also observed a very high prevalence of previously
females. The World Health Organization (WHO) STEPS undiagnosed hypertension; of the 1219 adults with hyperten-
survey conducted in 5 of the 21 districts of Sri Lanka during sion, 640 were newly diagnosed (52.5%) at the time of study,
the same time period as the present study reported a which is a considerably higher proportion compared to data
prevalence of 21.8% in all adults (17). A community based from other countries. A large population based study in India
survey in year 2000–2001 conducted in four out of nine reported that 22.2% of those with hypertension were previ-
provinces of Sri Lanka reported a hypertension prevalence of ously undiagnosed (21). In Mexico 40% of hypertensive men
18.8% in men and 19.3% in women. Hence, there has been a and 23% of women were undiagnosed and untreated, and in
24% increase in the prevalence over a period of 5–6 years. China, it was 15.5% (22,23). However, in developed countries
For personal use only.
The projected prevalence of hypertension in Sri Lanka adults such as USA, the presence of undiagnosed hypertension is
in year 2015 and 2020 based on the projected changes in only 8% (24). Undiagnosed hypertension is an important
population size (UN population forecasts) and estimated public health problem as it is associated with target organ
changes in incidence of hypertension (based on above data) damage, cardiovascular risk and increased mortality.
was 30.2% and 34.1%, respectively. The WHO STEPS Korhonen et al. (25) demonstrated that one in five patients
surveys conducted in regional countries have reported on with undiagnosed hypertension had evidence of subclinical
the prevalence of hypertension; Bangladesh – 17.9% (year target organ damage, and three in four patients may be
2009–2010), Bhutan – 17.1% (year 2007), India – 14.5% (year regarded as having high cardiovascular risk. The only way to
2007–2008) and Nepal – 9.3% (year 2007–2008) (17). The know whether a person has high BP is to have it measured,
National Health Survey of Pakistan conducted in 2005–2006 since majority of those with elevated BPs are asymptomatic.
estimated that hypertension affects 18% of all adults (18). The reasons for the presence of such high level of undiag-
Hence, the national prevalence of hypertension in Sri Lanka is nosed disease could be multi-factorial. Lack of awareness,
higher than in regional countries with populations sharing poor access to health-care facilities, language-barriers in
similar genetic backgrounds and life styles, a finding that health care and therapeutic inertia are all probable reasons
requires further evaluation. However, it is also important to (26). However, Sri Lanka has achieved a unique status in the
note that the Sri Lankan prevalence of hypertension is much world with health indicators that are comparable with those of
lower than the estimated global prevalence in 2008, which developed countries and its population has a very literacy
according to the WHO is around 40% (19). level (490%) (27). Hence, there is a need to further explore
The prevalence of hypertension has been shown to differ in other factors that are contributory to such high levels of non-
different countries and across geographical regions. The diagnosis.
reasons for such differences are still a matter of debate, the In the logistic regression analysis, male gender, physical
relative contributions given by environmental and genetic inactivity and Moor (Muslim) ethnicity were all significantly
factors being still poorly defined (20). A study conducted in associated with hypertension. Previously we have demon-
Italy concluded that genetic factors involved in the expression strated that the prevalence of Metabolic Syndrome and
of BP traits account for only 30% of the phenotypic variance, diabetes is higher amongst the people of Moor ethnicity
but seem to play a larger role in men; co-morbidities and from Sri Lanka (28,29). This could be due to differential
environmental factors remain of predominant importance in cultural beliefs in healthy seeking behavior, dietary habits and
determining BP (20). These findings on the differential physical activity patterns among the different ethnicities.
contribution of environmental and genetic factors on the Indeed, those of Moor ethnicity from Sri Lanka are known to
prevalence of hypertension need to be replicated in South have lower levels of physical activity than other ethnicities
Asian populations. Such understating will have implication for (30). Hence, it is important to consider ethnic differences
planning public health interventions aimed at reducing the when instituting screening programmes and preventive meas-
disease burden. ures. Physical activity is another important determinant of BP,
DOI: 10.3109/10641963.2013.863321 Prevalence and predictors of hypertension in Sri Lanka 7
and lifestyle modifications, such as physical activity, are healthier lifestyles with emphasis on preventing obesity and
recommended first-line/adjunctive therapies for hypertension. increasing level of activity.
Regular physical activity is also known to reduce mortality in
patients with high BP (31). Hence, numerous governing
Acknowledgements
bodies including the US Preventative Services Task force
(USPSTF) recommends counseling by health-care providers The National Science Foundation of Sri Lanka was the
to promote regular physical activity. However, very few primary source of funding for the study. The additional
patients receive exercise counseling from health-care pro- support provided from the Oxford Centre for Diabetes
viders to help lower BP and improve health outcomes (32). Endocrinology and Metabolism, UK, and the NIHR
In the present study, diabetes and central obesity were also Biomedical Research Centre Programme is gratefully
significantly associated with hypertension. Based on popula- acknowledged.
tion studies, risk estimates indicate that at least two-thirds of
the prevalence of hypertension can be directly attributed to Declaration of interest
obesity (33). Furthermore, abdominal adiposity has also been
implicated in the pathogenesis of coronary artery disease, The authors report no conflicts of interest. The
Clin Exp Hypertens Downloaded from informahealthcare.com by QUT Queensland University of Tech on 05/27/14
sleep apnoea, stroke and congestive heart failure (34). In authors alone are responsible for the content and writing of
overweight and obese subjects, the cardiovascular risk is not the paper.
significantly increased unless hypertension is present (35).
Hence, hypertension could be one of the mediators through
which obesity causes CVD. While obese subjects are prone to References
hypertension, the Framingham and Tecumseh studies showed
1. Carretero OA, Oparil S. Essential hypertension. Part I: definition
that hypertensive subjects are also appear to be prone to
and etiology. Circulation 2000;101:329–35.
weight gain (36). In hypertensive diabetic patients, lowering 2. Kearney PM, Whelton M, Reynolds K, et al. Global burden of
BP to less than 130/80 mmHg is a primary goal in manage- hypertension: analysis of worldwide data. Lancet 2005;365:217–23.
ment. Both hypertension and diabetes mellitus affect the same 3. Williams B. Recent hypertension trials: implications and con-
troversies. J Am College Cardiol 2005;45:813–27.
major target organs. The combined presence of hypertension 4. Lawes CM, Vander Hoorn S, Rodgers A, International Society of H.
and diabetes concomitantly accelerates the decrease in renal Global burden of blood-pressure-related disease, 2001. Lancet
For personal use only.
16. Ahmad OB, Boschi-Pinto C, Lopez AD, et al. Age standardization 25. Korhonen PE, Kautiainen H, Jarvenpaa S, Kantola I. Target organ
of rates: a new world standard. Geneva: World Health Organisation; damage and cardiovascular risk factors among subjects with
2001. previously undiagnosed hypertension. Eur J Prevent Cardiol 2013.
17. World Health Organization. Chronic diseases and health [Epub ahead of print]. doi: 10.1177/2047487312474530.
promotion: STEPS country reports; 2013. Available from: http:// 26. Eamranond PP, Patel KV, Legedza AT, et al. The association of
www.who.int/chp/steps/reports/en/index.html [last accessed language with prevalence of undiagnosed hypertension among
25 April 2013]. older Mexican Americans. Ethnicity Dis 2007;17:699–706.
18. Saleem F, Hassali AA, Shafie AA. Hypertension in Pakistan: time 27. Karunathilake IM. Health changes in Sri Lanka: benefits of primary
to take some serious action. Br J Gen Pract: J R College Gen Pract health care and public health. Asia-Pacific J Public Health/Asia-
2010;60:449–50. Pacific Acad Consortium for Public Health 2012;24:663–71.
19. World Health Organization. Raised blood pressure: situation and 28. Katulanda P, Ranasinghe P, Jayawardana R, et al. Metabolic
trends; 2013. Available from: http://www.who.int/gho/ncd/risk_- syndrome among Sri Lankan adults: prevalence, patterns and
factors/blood_pressure_prevalence_text/en/ [last accessed 25 April correlates. Diabetol Metab Syndr 2012;4:24.
2013]. 29. Katulanda P, Rathnapala DAV, Sheriff R, Matthews DR. Province
20. Biino G, Parati G, Concas MP, et al. Environmental and genetic and ethnic specific prevalence of diabetes among Sri Lankan adults.
contribution to hypertension prevalence: data from an epidemio- Sri Lanka J Diabetes Endocrinol Metabol 2011;1:2–7.
logical survey on Sardinian genetic isolates. PLoS One 2013;8: 30. Katulanda P, Jayawardana R, Ranasinghe P, et al. Physical activity
e59612. patterns and correlates among adults from a developing country: the
Sri Lanka Diabetes and Cardiovascular Study. Pub Health Nutr
Clin Exp Hypertens Downloaded from informahealthcare.com by QUT Queensland University of Tech on 05/27/14