Introduction

Elevated blood pressure (BP) is a predictor and one of the most important causal factors for premature mortality and morbidity due to myocardial infarction, stroke and other cardiovascular complications (Selmer and Tverdal 1995; Ellekjaer et al 2001; Gu et al 2008). Many studies have investigated the relationship of body fatness with BP (Gerber et al. 1995; Baumgartner 1989). There are studies that have demonstrated an association of a progressive increase in BP with an increase in adiposity (Mark et al. 1999). Overweight and obesity are important determinants of health leading to adverse metabolic changes, including an increase in BP. Being overweight was reported to be associated with a two- to six-fold increase in the risk of developing hypertension (HT) (WHO 1996).

Body mass index (BMI) is a measure of overall adiposity (Bose and Mascie-Taylor 1998; Kopelman 2000). Studies indicate that BMI has its independent role in the identification of overweight and obesity, and has been widely recommended for epidemiological surveys because of their independent association with major cardiovascular and metabolic risk factors (Bjorntorp 1987; Seidell et al. 1990; Zhou et al. 2002; Canoy et al. 2004). The health risk has been shown to increase in a graded fashion when moving from the lower to higher BMI categories (Brown et al. 2000; Must et al. 1999). BMI has been shown to be a risk factor for total and ischemic stroke in men and women, and high blood pressure is one of the major risk factors for the same (Hu et al. 2007). BMI has been shown to be associated with BP among adults in Finland (Jousilahti et al. 1995), Sweden (Henriksson et al. 2003), Argentina (Feldstein et al. 2005), the African Diaspora (Cappuccio et al. 2008) and also in Pakistan (Khan et al. 2008).

Once believed to be a problem of developed and industrialized nations, HT is now considered as a major public health concern in developing countries, like India, too (Nirmala Reddy 1998; Kusuma et al. 2004; Kearney et al. 2004). Reports as early as the early 1990s said that India had nearly 20 million hypertensive people (Shah 1992). In India, 57% of all stroke deaths and 24% of all coronary heart disease deaths are due to HT, and there are 31.5 million and 34 million hypertensive people in rural and urban India, respectively (Gupta 2004). Recent studies from Kolkata, India (Ghosh et al. 2000, Bhadra et al. 2002; Bose et al. 2003, 2005; Sadhukhan et al. 2007; Ghosh and Bandyopadhyay 2007), have indicated that HT is a major problem among Bengalee people, too.

There are studies from other populations in India that showed a positive relationship of BP levels with various anthropometric measurements, e.g., waist-hip ratio (WHR) (Gupta and Mehrishi 1997); WHR and BMI (Santhirani et al. 2003); waist circumference (WC) and BMI (Deshmukh et al. 2006); WC, WHR and BMI (Gupta et al. 2007). There are works among the Bengalee population from in and around Kolkata, West Bengal, India, that have dealt with older hypertensive and normotensive men (Ghosh et al. 2000) and women (Bose and Das Chaudhuri 2001), young adults (Bhadra et al. 2002), male jute mill workers (Bose et al. 2003), adults of 18 years and above (Das et al. 2005) and adults aged 20–61 years (Ghosh and Bandyopadhyay 2007). But very few (Bose et al. 2003) of the above studies explored these kinds of relationships among distinctively low socio-economic urban groups, in particular, among the slum dwellers.

It has been noted that, in recent years, accompanying the economic and demographic transition in developing countries, there is a shift from diseases caused by poverty to the chronic, non-communicable, lifestyle-related diseases. The progressive risk of cardiovascular disease may be attributed to a large extent to a lack of public response to the conditions and an incorrect notion of the policy makers and the public that these are the problem of the urban rich only (Ramaraj and Alpret 2008). The urban poor face the worst consequences of an urbanized lifestyle, making them at significantly high risk of non-communicable diseases (Yadav and Krishnan 2008). In a very recent study in Kolkata, the prevalence of stroke incidence was greater among slum dwellers than the non-slum subjects (Das et al 2007). A poor understanding and control of risk factors, such as hypertension, has been proposed to be responsible for this (Pandian et al. 2007).

Therefore, in this study, we attempted to investigate the relationship of BMI with systolic (SBP), diastolic (DBP) pressure (MAP) as well as the prevalence of HT across different BMI categories among male slum dwellers of Kolkata, India.

Materials and Methods

Area of study

The district north 24 Parganas lies between 21° 31′ and 22° 57′ north latitude and between 88° 2′ and 89° 6′ east latitude. The South Dum Dum area is adjacent to Kolkata city and is one of the urban centers of the district of North 24 Parganas. It is about 10 km to the north from the heart of the city of Kolkata (formerly Calcutta). The fieldwork was conducted on adult Bengalee men (aged 18–84 years) living in a slum called ‘Bidhan Colony’ approximately 15 km from Kolkata town center. Most of the subjects were engaged in jobs of low socio-economic status, as per the local standards, ranging from factory workers to rickshaw pullers or day laborers for males and domestic helpers for females. There are landlords who were the first settlers of the locality, and they have a huge number of tenants. The general hygienic condition clearly seems to be poor. The sanitation, sewerage systems and household structures are the silent but definite indications of the poverty and poor quality of life.

The sample

This study was carried out as a part of a research project on the nutritional status of the adult males of the area mentioned above. The municipal authorities and local community leaders were informed before commencement of the study. Households were selected randomly from the particular locality. The houses were situated in a linear fashion, one after another from one end to the other (east-west) of the slum. Each household was approached during field visits from one direction to another of the slum, and the available adult male member(s) were selected as participants. They were measured on the same day or another as per their agreement by fixing prior appointments. No household was visited twice for new enrollment of subjects. The whole slum was covered from one direction to the other in the same manner. Participants were interviewed and measured at their respective households. In some cases, due to logistical problems, they were taken to a common place where a number of them were examined. However, all the participants essentially resided inside the administrative boundary of the project area, i.e., Bidhan Colony. No strict statistical sampling for the inclusion of the participants could be followed due to operational compromise in the field (Khongsdier 2002). The whole slum area was covered in the total exercise of the data collection. The overall response rate was found to be around 80%. Informed consent was also obtained from each participant. Apparently healthy men who were reportedly not suffering from any acute illness and were self-satisfied with their normal day-to-day work schedule at the time of measurements were enrolled to participate in the study. The apparently and self-reportedly healthy male members of each family unit were thus sampled randomly. No subjects were taking anti-hypertensive drugs or under any other kinds of prolonged medication. A total of 475 men aged 18 years and above were interviewed and measured. Among them, five were excluded from the analyses for the present study due to missing data on hypertension. Therefore, the final sample size remained 470 for the study.

The field investigation, including anthropometric measurements, was done entirely by the first author (R.C.). Information on the ethnicity, age, number of family members, monthly family income (MFI), smoking status (smoker or non-smoker) and alcohol consumption (yes or no) was also collected from each subject with the help of a pre-tested questionnaire. Monthly per capita income (MPCI) was calculated by dividing the MFI by the number of family members. All the measurements were taken following the standard techniques (Lohman et al. 1988). Height was measured to the nearest 0.1 cm using a locally made Martin’s type anthropometer and weight to the nearest 0.5 kg using a standard bathroom scale. Technical errors of measurements were found to be within acceptable limits (Ulijaszek and Kerr 1999). BMI was computed using the following standard equation: BMI = weight (kg)/height (m2). Individuals were classified into different categories of BMI using cutoff values of BMI for the Asia-Pacific populations recommended by the International Obesity Task Force (WHO 2000) as follows:

  • Undernourished: BMI <18.5

  • Normal: BMI = 18.5–22.9

  • Overweight: BMI = 23.0–24.9

  • Obese: BMI ≥25.0

Measurement of blood pressure

An automatic digital BP monitor (Home Health, Switzerland) was used to measure BP following standard procedures. All subjects were measured between 7 p.m. and p.m. (±15 min). Resting systolic and diastolic BPs (in mmHg) were measured with the subject in a sitting position for at least 15 min prior to measurement and again at least 10 min after the first reading. The mean values of two measures were used in analyses. Hypertension was defined as systolic blood pressure (SBP) ≥140 mmHg and diastolic blood pressure (DBP) ≥90 mmHg following the US Seventh Joint National Committee on Detection, Evaluation and Treatment of Hypertension (JNC VII) guidelines (Chobanion et al. 2003).

Statistical analyses

The distributions of the anthropometric and socio-economic variables were not significantly skewed; thus, they were not transformed. Age, anthropometry and BP variables were described by their means and standard deviations. One-way analyses (Scheffe’s procedure) (Mascie-Taylor 1994a, b) were carried out to test for differences in mean SBP, DBP and MAP across the four BMI categories. Chi-square analysis was done to determine the differences in the prevalence of hypertension among nutritional categories. Odds ratios (OR) were calculated to determine the change of its values with increasing BMI. Multiple linear regression analysis was carried out to determine the influence of BMI on SBP, DBP and MAP, independent of age, MPCI, smoking and alcohol intake status. To do so, the independent variables age, MPCI, smoking and alcohol intake status were entered first in the equation en bloc and then BMI in the second block. For this analysis, smoking and alcohol intake status remained categorical (yes = 1, no = 0, in each case). Pearson’s correlation coefficient (r) was computed to test the association of all the continuous variables. Partial correlation coefficients (age effect removed) of BMI with SBP, DBP and MAP were also computed to show the independent associations. All statistical analyses were undertaken using the SPSS Statistical Package, version 7.5. Statistical significance was set at p < 0.05.

Results

The characteristics of the sample are presented in Table 1. Mean age and BMI of the subjects were 37.5 years (SD = 14.2) and 20.3 kg/m2 (SD = 3.2). Means for SBP, DBP and MAP were 121.6 mmHg (SD = 15.5), 79.6 mmHg (SD = 9.7) and 93.5 mmHg (SD = 11.1), respectively.

Table 1 Characteristics of the sample (n = 470)
Full size table

Table 2 presents the means of age, height, weight and BMI among normotensive and hypertensive subjects. Significant (p < 0.001) differences were observed for age (T = -6.29), weight (T = -3.42) and BMI (T = -3.91) between these two groups. For all the parameters, hypertensives had the higher means.

Table 2 Differences in mean characteristics between BP groups
Full size table

Correlation analyses (results not presented) demonstrated that age had significant (p < 0.001) correlations with SBP (r = 0.40), DBP (r = 0.24) and MAP (r = 0.30). Age did not have any significant association with BMI. BMI was also significantly (p < 0.001) correlated with SBP (r = 0.24), DBP (r = 0.34) and MAP (r = 0.33). Significant (p < 0.001) partial correlations (age effect removed) of BMI with SBP (r = 0.27), DBP (r = 0.36), and MAP (r = 0.36) were also observed.

The prevalences of CED, overweight and obese subjects were 32.8%, 11.3% and 7.9%, respectively. Table 3 presents the mean values of SBP, DBP and MAP by BMI categories. Significant (p < 0.001) differences in all the measures were observed among the categories. Mean values increased steadily from the undernourished through the normal to the overweight groups. Obese subjects had the highest means for all the BP variables. The magnitudes of differences were always significantly higher between the normal and the overweight than between the undernourished and the normal or between the overweight and the obese. Scheffe’s test revealed that for the SBP, differences among all groups were statistically significant except between undernourished and the normal. On the other hand, DBP differed significantly among all the groups except between overweight and the obese. A noteworthy point was that age did not differ significantly among the nutritional categories.

Table 3 Differences in blood pressures among BMI categories
Full size table

Table 4 demonstrates the percentage prevalence of individuals with HT in different BMI categories. There were significant differences in the percentage of hypertensive among the categories except between the undernourished and the normal groups. Lower prevalences of HT were observed in the undernourished and the normal groups (16.2% and 15.0%, respectively) and higher in the overweight and the obese groups (37.7 and 43.2%, respectively). The Odds ratios became significantly higher in the overweight group, and it was about four times higher in the obese group than in the undernourished and the normal groups.

Table 4 Prevalence (%) of hypertension according to the nutritional categories
Full size table

Multiple regression analyses of the BP variables on BMI, age and other behavioral parameters demonstrated (Table 5) that BMI had significant influence on these measures. BMI explained 5.8%, 11.4% and 10.6% variation in SBP, DBP and MAP, respectively, independent of age, MPCI, smoking and alcohol consumption status. In the regression analyses, all the independent variables except for the BMI were entered first en bloc and then BMI in the second block. From the first block entered in the regression model, age was the only variable retained with significant influences on SBP (Adj R2 = 0.132), DBP (Adj R2 = 0.054) and MAP (Adj R2 = 0.086) among the variables other than BMI. In the regression models for the three BP variables, the other variables were never entered, and their T values were insignificant (results not shown).

Table 5 Results of regressions showing the effect of BMI on SBP, DBP and MAP, after controlling for age, MPCI, smoking and alcohol intake status
Full size table

Discussion

Obesity and HT are two major inter-related cardiovascular risk factors (Khan et al. 2008). Adiposity is an established risk factor for cardiovascular disease, but the relationship of adiposity with the risk of cerebrovascular disease is still to some extent unclear (Hu et al. 2007). Being overweight is associated with a risk of developing HT. In general, overweight status exposes an individual to two to six times the increase in the risk of being hypertensive (WHO 1996). Studies consistently reported that both obesity and chronic diseases are equally prevalent in poor people like the urban slum dwellers (Bunnag et al 1990; Misra et al 2001).

In a recent study (Ghosh and Bandyopadhyay 2007) involving 180 middle-class adult Bengalee males (aged 20–61 years) from a Kolkata suburb, the mean (SD) age, BMI, SBP and DBP was reported to be 35.7 (9.3) years, 22.4 kg/m2, 118.77 (11.85) mmHg and 78.10 (8.74) mmHg, respectively. They had higher mean BMI, but slightly lower BP values than those of our sample. This might be due to a lower age range (20–61 years) of the study as compared to ours. Deshmukh et al. (2006) found these values to be 38.2 (10.4) years, 19.1 (4.2) kg/m2, 120.2 (17.6) mmHg and 77.7 (12.4) mmHg, respectively, in a rural adult population of low socio-economic status from western India. Male agriculturist (middle socioeconomic class) Reddies in southern India had mean values of age = 39.12 (14.8) years, BMI = 21.5 (3.7) kg/m2, SBP = 127.67 (18.5) mmHg and DBP = 83.6 (8.8) mmHg, whereas agricultural laborer (low socio-economic class) Malas were reported to have age = 36.8 (13.0), BMI = 19.0 (2.5), SBP = 113.0 (8.8) mmHg and DBP = 76.0 (7.0) mmHg (Venkataramana et al. 2001). Among male agricultural laborers (Malas and Muslims) with reportedly ‘heavy’ physical activity levels in Andhra Pradesh, the mean values were 113.7 and 70.3 for SBP and DBP, respectively (Nirmala Reddy 1998). In the same study, the service and business communities demonstrated higher mean values (127.3 and 77.5). However, none of these groups belonged to an urban underprivileged group like that in the present study. In a study on coastal urban slum fishermen, the mean SBP and DBP were found to be 124.9 (9.6) and 82.7 (8.1), respectively (Gopi Chand and Sambasiva Rao 2007).

In our study, we found significant (p < 0.001) positive associations of BMI with SBP (r = 0.24), DBP (r = 0.34) and MAP (r = 0.30), independent of age, MPCI and smoking and alcohol consumption status. In a follow-up study among middle-aged people in Finland, BMI was found to be correlated positively with SBP and DBP cross-sectionally, and it also predicted the future increase in BP (Jousilahti et al. 1995). A similar finding was reported in another study among middle-aged Swedish men (Henriksson et al. 2003). A study from India demonstrated significant positive correlation of BMI with SBP and DBP in both men and women (Gupta et al. 2007). Another study among the adult Reddies and Malas of the southern part of India also showed positive correlations of BMI with SBP and DBP in both sexes. In the male Reddies, the correlation coefficient of BMI with SBP and DBP were 0.41 and 0.31, respectively. In the Mala males, those values were 0.29 and 0.23, respectively (Venkataramana et al. 2001). In the slum study mentioned above (Gopi Chand and Sambasiva Rao 2007), BMI had a significant positive correlation with SBP (r = 0.33) and DBP (r = 0.33) in men.

The Indian study mentioned above (Gupta et al. 2007) also reported that with increasing BMI, the prevalence of HT increased significantly. In a Norwegian population, an independent effect of change in BMI on change in SBP and DBP was reported in both women and men; also, people who increased their BMI were at increased risk for HT (Droyvold et al. 2005). In adolescent Bengalees of Kolkata, SBP and DBP of both sexes increased with an increase in weight, height and BMI with a significant positive linear correlation (Saha et al. 2007). In our study SBP, DBP, and MAP also increased significantly from the undernourished group through normal to the overweight group. Moreover, the prevalences of HT were lowest in the undernourished and normal groups (16.2% and 15.0%, respectively) and increased significantly in the overweight group (37.7%), being highest in the obese group (43.2%). It is also interesting to mention that the undernourished and the normal groups did not show significant differences in the prevalence of HT as well as in the mean values of SBP. The difference in mean DBP between the undernourished and the normal group was also less than the differences in the mean between the normal and the overweight groups. Similarly, the differences between the overweight and the obese groups was less and not statistically significant for both SBP and DBP. Therefore, it seems that the increase in BMI has a significant clinical effect on BP variables only when it reaches the critical value of 23.0 kg/m2 or higher. It may be mentioned here that the value of 23.0 kg/m2 is the proposed cutoff point of BMI for overweight in Asian populations (WHO 2000). In Finnish men (Jousilahti et al. 1995), HT was 18% among the leanest (BMI <20). It is also worth mentioning here that the prevalence of hypertension even in the lower BMI categories (15 and 16%) is no less in view of public health consequences. Therefore, it may be inferred that in this population with tough challenges in life, there are other important factors in addition to body weight and adverse dietary profile that contribute substantially to the elevated level of BP.

Regression analysis also revealed that BMI had a significant positive effect on BMI independent of age, MPCI, smoking and alcohol intake status. BMI explained more variation of DBP (16.8) than of SBP (8.3%), whereas age explained relatively more of SBP than DBP. It is also noteworthy to mention that the HT group had significantly higher values of age, weight and BMI than their NT counterparts. In the study mentioned above (Gopi Chand and Sambasiva Rao 2007), BMI had an independent significant influence on SBP and DBP, and both of these elevated markedly with increasing BMI.

In conclusion, our study demonstrated that even in a slum-dwelling population with low mean BMI and a high rate of under nutrition, increased BMI, particularly above a certain range (>22.9 kg/m2), is an independent risk factor for HT. This may have serious public health implications and requires further research. More importantly, prospective studies on various ethnic groups having low BMI and a high rate of under nutrition are required to better understand the etiology of hypertension. There may be significant ethnic differences in the relationship of BMI and BP. Lastly, the relationship of body fat, as measured by skin folds and other methods of body composition, with hypertension should be investigated in populations having a low mean BMI. These studies will probably give a better insight into the relationship of adiposity, independent of BMI, with HT.