Body fat percentage and factors associated with cholesterol levels among patients at healthy choice clinic in Kathmandu, Nepal: a cross-sectional study Authors: Ayush Adhikari, Rubina Karki, George Chikondi Samu Published: 2026 ABSTRACT AbstractBackgroundThe study of body composition and its correlation with lipid profiles are significant markers for the cardiovascular risk factors of the urban Nepali population. This study is an attempt to evaluate the percentage of body fat and determine the parameters that influence the cholesterol levels in the patients attending at Healthy Choice Clinic, Kathmandu, Nepal.MethodsA quantitative cross-sectional study was conducted among 106 adult patients of Healthy Choice Clinic aged 18 years and above. Their socio-demographic profile, lifestyle and dietary practice were sourced through structured interview questionnaire. Body Mass Index (BMI), waist-to-hip ratio (WHR), and percentage of body fat were measured through Bioelectrical Impedance Analysis (BIA). For total cholesterol, triglyceride, HDL, LDL and VLDL assessment, blood sample was taken through laboratory tests. Chi-square and fisher’s exact test were used for correlation.ResultsThe statistical prevalence showed that the overweight and obese respondent was 51.9%, and 21.7% obese. In addition, about 86.8% respondents had significant abdominal obesity with high waist to hip ratio (WHR). Of which, alcohol intake significantly correlated with high total cholesterol level among other respondents (p = 0.044). The body mass index (BMl) was significant with total cholesterol level (p = 0.022). The body mass index (BMI), WHR, and body fat percentage were all significantly correlated with high triglycerides (p = 0.019, p = 0.004, p = 0.011; respectively). Regular exercise significantly correlates with reduced very-low-density lipoprotein (VLDL). No significant association was reported between cholesterol and fat percentage and WHR and most of the dietary variables.ConclusionsAlthough we could not find a strong correlation between body fat percentage (BFP) and cholesterol levels, BMI and central obesity indicators can provide valuable correlation with CV risk relating to dyslipidemia. Physical activity is shown to have a protective effect against dyslipidemia. These results can support the significance of a detailed clinical assessment with anthropometric measurements and lifestyle history to identify and manage patients with risk factors for CV disease.Keywords: Body fat percentage, Cholesterol, Bioelectrical impedance analysis, Lipid profile, Obesity, Cardiovascular risk, Nepal Background The impact of body composition assessment on health prediction is crucial to health risk screening. Due to the increasing number of obesity and obesity-related diseases around the globe, the application of body composition assessment in clinical and research activities is becoming highly significant [1]. Body fat percentage (BF%) means one’s health and it directly correlates to an individuals’ conditions. High BF% makes an individual susceptible to chronic health diseases such as cardiovascular diseases, diabetes mellitus, and metabolic syndrome as compared to minimum BF% [2]. Obesity is an excess of adipose tissue. The mechanisms of obesity are associated with a positive energy balance [3]. In 2016, there were more than 1.9 billion adults aged 18 years and older with overweight, of which more than 650 million were obese [4]. In Nepal NCDs are responsible for 66% of total deaths (2019) which was 60% in 2014 [5]. Cardiovascular diseases in particular account for 22% of all deaths, ischemic heart disease still one of the two leading causes of death with chronic obstructive pulmonary disease. Bioelectrical Impedance Analysis (BIA) is a simple, non-invasive, fast, inexpensive, and widely used technique that estimates body composition (fat mass, fat-free mass, and body water) by measuring the impedance and reactivity of the body to a small harmless electrical current. It is clinically used to determine hydration and muscle composition, as well as cell health [6]. Bioelectrical impedance analysis (BIA) is a method that applies a low-level current to the study subject whose bioimpedance is to be measured. The reactance and resistance of the subject is then measured which gets the electrical conductivity of the study subject. This measure is then used to estimate body composition including the estimation of total body water, extracellular water, intracellular water, FM, fat-free mass, body fat percentage and bone mineral density [7]. In addition to this, it is stated that the risk of metabolic syndrome in individuals with normal Body Mass Index (BMI), but with high body fat is four times higher than in individuals with normal BMI and low body fat [8]. It was recently proposed that health outcomes may be better evaluated by using a measure of body composition, rather than just BMI [9]. The determination of the body fatness accurately and analysis of the factors related to the body fat percentage could provide significant clinical significance to the health professionals. It could enhance the risk of determining the disease at a higher priority level especially in people with obesity and could help formulate more preventive measures to combat morbidity and mortality [10]. The objective of this study was to determine the percentage of body fat and the factors associated with the cholesterol levels in clients of Healthy Choice Clinic, Kathmandu, Nepal. The accurate assessment of body fat along with the evaluation of determinants of body fat percentage could carry potential clinical value to the health professionals by refining disease risk assessment, particularly in individuals with obesity and enabling the practitioners to devise preventive strategies for mitigating morbidity and mortality [11]. The present study was specifically designed to evaluate the percentage of body fat and determinants od cholesterol levels in the patients of Healthy choice clinic, Kathmandu, Nepal. The secondary objective was to conduct an exploratory assessment in an attempt to relate various anthropometric indices (BMI, WHR, Body Fat percentage) with the lipid profile aspects in consideration of the feasibility. Methods Study design and setting The study design was a cross-sectional, conducted in Healthy Choice Clinic, located in Kathmandu metropolitan city of the country Nepal. The duration of the study was six month which includes two month of data collection. Study population and sampling The study was conducted on patients of age group of 18 years and above visiting at Healthy Choice Clinic for diet consultation. Through purposive sampling 106 respondents were recruited for the study. Sample size was calculated through the formula for infinite population n = (z²pq)/d² where z = 1.96 (confidence interval of 95%), p = 0.50 (prevalence) and d = 0.10 (margin of error). Sample size of study calculated is 96, considering 10% as non-response rate the final sample size is 106. Inclusion criteria were Inclusion criteria were adult patients (≥ 18 years) who came for diet counseling, were communicated with in the interview and agreed to provide blood sample for cholesterol analysis. Exclusion criteria were pregnant women and lactating mothers, patients unable to stand and amputated patients. Data collection Data were collected using a pre-coded semi-structured questionnaire partially adopted from the Nepal Demographic and Health Survey (NDHS) 2016. The questionnaire included sections on socio-demographic characteristics, lifestyle behaviors (exercise habits, smoking, alcohol consumption), and dietary patterns. A semi structured questionnaire was used which was pre-coded and adapted from the course of Nepal Demographic and Health Survey (NDHS) 2016. The questionnaire consists of socio-demographic characteristics, lifestyle pattern including physical activity, smoking, alcohol drinking etc. and eating habit. Anthropometric data like height (made using stadiometer), weight, BMI, body fat percentage, waist to hip ratio was obtained by Bioelectrical Impedance Analysis (BIA) using New Angie GS6.7 (2019, manufactured in China). BIA machine produces electrical current of low-level and sends it through body to obtain electrical impedance from which body fat percentage and other characteristics can be calculated. Due to more water content, lean tissues have higher conductivity compared to fat tissues and these parameters can be calculated from the machine to obtain total body water, fat free body mass and fat mass. Questionnaire was translated into Nepali language and back translated into English by two bilingual experts and pre-test was conducted among 15 individuals (excluded in the final study). Internal Consistency Reliability was calculated and showed Cronbach alpha coefficient of 0.72 for dietary and 0.68 for lifestyle items. Body composition analysis was conducted using Bioelectrical Impedance Analysis New Angie GS6.7 (2019, made in China). For measurement process, participants fasted (at least) four hours, avoided alcohol intake and caffeine for twelve hours, did not do strenuous activity least for eight hours and urine secretion was inhibited before the measurement. Measurement was done at room temperature (22–25 degrees C) in the morning (8:00 to 11:00 h) and participants were instructed to remove footwear and metallic materials (if any). Percentage of body fat was classified based on recommended cut-off criteria for each sex: for males < 25 non-obese and > = 25 obese; for females < 32 non-obese and > = 32 obese according to the ACE committee [12]. The blood samples analysis included estimation of lipid profile such as Total cholesterol, Triglycerides, high density lipoprotein (HDL), low density lipoprotein (LDL) and very low density lipoprotein (VLDL). From the results, VLDL was calculated by Friedewald equation (VLDL = Triglycerides / 5), when triglyceride level < 400 mg/dL [13].Calculation of LDL was as follows: LDL = Total Cholesterol – HDL – VLDL. Normal ranges of lipid profile are: Total cholesterol (< 200 mg/dL); Triglycerides (< 150 mg/dL); HDL (> 35 mg/dL); LDL (< 100 mg/dL); VLDL (< 30 mg/dL). Statistical analysis Data was entered in Excel and analyzed using SPSS statistical software. Descriptive statistics were performed for socio-demographic and other characteristics. Bivariate analysis was conducted using Chi-square test and Fisher’s exact tests and Simple linear regression was conducted for multivariate analysis. A p-value < 0.05 was set to be statistically significant. Ethical considerations Permission was obtained from The CAFODAT college Research Committee for carrying out the study. Verbal and written approval was obtained from the clinic’s administration for conducting this study. Informed consent was taken from the study participants after explaining the aim and significance about the study. The confidentiality of the participants was maintained and secured. Results Socio-demographic characteristics The study consisted of 106 subjects. The age of the subjects ranged from 18 year to 60 years with a mean age of 29.91 years (SD = 8.40). The maximum age group surveyed was 18–29 (56.6%) followed by 30–39 years (20.8%), 40–49 years (9.4%) and 50–60 years (3.8%). The subjects comprised of 52.8% females and 47.2% males. The caste/ethnic composition of the subjects was Brahmin/Chhetri (50%), Janajati (36.8%), Madheshi (6.6%) and Others (4.7%). By educational status, 98.1% had superior secondary level education. In case of employment, majority of the participants were working in the private sector (62.3%), government (17.9%), unemployed (14.2%), and housewives (5.7%) respectively. Among the responders, majority 88 (83.0%) belonged to nuclear families as compared to 18 (17.0%) of the joint family (Table 1). Table 1Socio-demographic characteristics of the respondents (n = 106)CharacteristicsFrequencyPercent (%)Age (years) 18–296056.6 30–392220.8 40–49109.4 50–6043.8Sex Male5047.2 Female5652.8Ethnicity Brahmin/Chhetri5350.0 Janajati3936.8 Madheshi76.6 Others54.7Education Higher secondary10498.1Occupation Private job6662.3 Government job1917.9 Unemployed1514.2 Homemaker65.7 Family type Nuclear8883.0 Joint1817.0 Socio-demographic characteristics of the respondents (n = 106) Lifestyle characteristics In terms of physical activity, the participants were 79.2% non-exercisers and 20.8% exercisers. With regards to smoking status, the subjects were 84% non-smokers and 16% smokers. According to alcohol drinking status, the participants were 64.2% non-drinkers and 35.8% alcohol-drinkers. Most of the participants were non-vegetarians (88.7%), while the vegetarians constituted minority (11.3%) (Table 2). Table 2Lifestyle characteristics of the respondents (n = 106)CharacteristicsFrequencyPercent (%)Exercise habits Yes2220.8 No8479.2Smoking habit Yes1716.0 No8984.0Alcohol consumption Yes3835.8 No6864.2Dietary habit Non-vegetarian9488.7 Vegetarian1211.3 Lifestyle characteristics of the respondents (n = 106) Anthropometric measurements In terms of BMI, 51.9% of participants were overweight, followed by 21.7% with Obese I classification. Only 16% had normal BMI, while 6.6%, 2.8%, and 0.9% of respondents were underweight, Obese II, and Obese III respectively. Most of the respondents had high waist-hip ratio (WHR) at 86.8%, which corresponds with increased risk of ailments associated with central obesity. Only 13.2% had normal WHR (Table 3). Table 3Anthropometric measurements of the respondents (n = 106)MeasurementCategoryFrequencyPercent (%)BMIUnderweight76.6Normal1716.0Overweight5551.9Obese I2321.7Obese II32.8Obese III10.9WHRNormal1413.2High9286.8Body fat percentageNon-obese4542.5Obese6157.5 Anthropometric measurements of the respondents (n = 106) Associations with total cholesterol Among total of 106 respondents, 103 (97.2) had normal cholesterol level and 3 (2.8%) had high cholesterol. There is a significant association between alcohol intake and their total cholesterol level (p-value = 0.044). Among 38 alcohol drinkers, 3 (7.9%) had high cholesterol and none of the 68 respondents are alcohol non-drinkers. There is a significant relation between BMI and Total Cholesterol (p = 0.022). All the underweight, normal weight and overweight individuals had normal Total Cholesterol. However, the elevated level of Cholesterol was found in 4.3% of Obese I, 14.3% of Obese II and 33.3% of Obese III, with significant increasing trend (Table 4). Table 4Association of anthropometric measurements with total cholesterol (n = 106)VariableNormal n (%)High n (%)χ²p-valueBMI15.8090.022* Underweight7 (100)0 (0) Normal17 (100)0 (0) Overweight55 (100)0 (0) Obese I22 (95.7)1 (4.3) Obese II6 (85.7)1 (14.3) Obese III2 (66.7)1 (33.3)WHR0.4701.000* Normal14 (100)0 (0) High89 (96.7)3 (3.3)Body fat percentage2.2780.260* Non-obese45 (100)0 (0) Obese58 (95.1)3 (4.9)*Fisher's Exact Test Association of anthropometric measurements with total cholesterol (n = 106) *Fisher's Exact Test No significant relationship was found between WHR and total cholesterol (p = 1.000) and percentage body fat and total cholesterol (p = 0.260). None of the dietary variables had significant associations with total cholesterol levels. Associations with triglycerides Family type had a significant association with triglycerides levels and the p-value is 0.010. The joint family members demonstrated greater percentages of high triglycerides (p = 0.010). However, significantly higher proportion of participants from joint families had high triglyceride level (77.8%) as compared to participants with nuclear families (44.3%). There is a significant association between anthropometric measures and triglycerides. Triglycerides had a significant association with BMI (p = 0.019); a high BMI correlated to high triglycerides. Triglycerides are significantly associated with WHR (p = 0.004) as 55.4% of participants with elevated WHR have elevated triglyceride level in comparison to 14.3% with normal WHR. Body fat percentage also shows significant association (p = 0.011), as 60.7% of obese with elevated triglycerides compared to 35.6% of non-obese (Table 5). Table 5Association of anthropometric measurements with Triglyceride levels (n = 106)VariableNormal n (%)High n (%)χ²p-valueBMI11.7670.019* Underweight1 (100)0 (0) Normal11 (64.7)6 (35.3) Overweight32 (58.2)23 (41.8) Obese I5 (21.7)18 (78.3) Obese II3 (42.9)4 (57.1) Obese III1 (33.3)2 (66.7)WHR8.2300.004 Normal12 (85.7)2 (14.3) High41 (44.6)51 (55.4) Body fat percentage6.5260.011 Non-obese29 (64.4)16 (35.6)Obese24 (39.3)37 (60.7)*Fisher's Exact Test Association of anthropometric measurements with Triglyceride levels (n = 106) *Fisher's Exact Test Associations with HDL, LDL, and VLDL Age (p = 0.033) was significantly associated with HDL level; younger individuals were more likely to have a higher HDL. Consumption of white potatoes (p = 0.020) and sugary drinks (p = 0.009) were significantly associated with HDL level. Statistically significant relation of age was found with LDL (p = 0.026). Individuals with age of 30–49 years had the highest LDL (Table 6). Table 6Association of lifestyle factors with VLDL levels (n = 106)VariableNormal n (%)High n (%)χ²p-valueExercise habit14.1480.001* Yes18 (81.8)4 (18.2) No31 (36.9)53 (63.1)Smoking habit2.3030.185* Yes5 (29.4)12 (70.6) No44 (49.4)45 (50.6)Alcohol consumption0.4050.549* Yes16 (42.1)22 (57.9) No33 (48.5)35 (51.5)BMI11.4180.024* Underweight1 (100)0 (0) Normal13 (76.5)4 (23.5) Overweight25 (45.5)30 (54.5) Obese I6 (26.1)17 (73.9) Obese II3 (42.9)4 (57.1) Obese III1 (33.3)2 (66.7) WHR10.1180.002* Normal12 (85.7)2 (14.3) High37 (40.2)55 (59.8) Body fat percentage8.0490.006* Non-obese28 (62.2)17 (37.8) Obese21 (34.4)40 (65.6)*Fisher's Exact Test Association of lifestyle factors with VLDL levels (n = 106) *Fisher's Exact Test For VLDL, the gender (p = 0.046), occupation (p = 0.034) and exercise pattern of the subjects (p = 0.001) are significantly correlated. Exercising was found to be protective as 18.2% of the subjects exercising had high VLDL levels in comparison to 63.1% of the non-exercisers. Also, amongst the three measurements, VLDL was significantly related to BMI (p = 0.024), WHR (p = 0.002) and body fat (p = 0.006) (Table 7). Table 7Association of anthropometric measurements with VLDL levels (n = 106)VariableNormal n (%)High n (%)χ²p-valueBMI11.4180.024* Underweight1 (100)0 (0) Normal13 (76.5)4 (23.5) Overweight25 (45.5)30 (54.5) Obese I6 (26.1)17 (73.9) Obese II3 (42.9)4 (57.1) Obese III1 (33.3)2 (66.7)WHR10.1180.002* Normal12 (85.7)2 (14.3) High37 (40.2)55 (59.8)Body fat percentage8.0490.006* Non-obese28 (62.2)17 (37.8) Obese21 (34.4)40 (65.6)*Fisher's Exact Test Association of anthropometric measurements with VLDL levels (n = 106) *Fisher's Exact Test Analysis of multiple linear regression was conducted to determine the independent effects of demographic, occupational, lifestyle and anthropometric factors on Total cholesterol as presented in Table 8. The overall model was significant was significant, F (21, 74) = 2.07, p = 0.012, accounting for 37.0% of the variation in total cholesterol (R² = 0.370, Adjusted R² = 0.192). Table 8Multiple linear regression for the predictors of Total CholesterolPredictorCategoryEstimateStd. Errort valuePr(>|t|)Significance(Intercept)-12.3752267569.85579386-0.1771539060.859871496Age (Years)Continuous Variable1.074017850.52091722.0617822760.042740708*SexFemaleReferenceMale-15.2408578110.2186605-1.4914731550.14008815EthnicityBrahminReferenceChhetri-4.1959516959.106095495-0.4607849430.646304266Dalit-1.97092160330.49068946-0.0646401130.948634849Janjati-4.3278862678.956324815-0.4832212270.630365338Madheshi-1.17464727715.12736839-0.0776504710.938315671Muslim-29.6659857729.83893954-0.9942037560.323363582Others-28.9342443115.38441604-1.880750250.063940511†OccupationDesk jobReferenceActive Standing Job12.096502038.6795870811.3936725240.167588053Housewife-0.0561649915.88709094-0.003535260.99718879Retired Army-75.993551433.9597-2.2377568530.028246616*Strenuous Uniformed Job30.1878144717.174853651.7576752080.082938276†Student11.595708499.876357751.1740875320.244124695Family TypeJointReferenceNuclear-6.3927219418.951488043-0.7141518720.477379914ExerciseNoReferenceYes-6.0906605629.158902867-0.6649989250.508118431SmokingNoReferenceYes14.7509260612.489554791.1810610.241360535Alcohol IntakeNoReferenceYes12.940702648.8069156461.4693796510.145968568DietNon-VegReferenceVeg8.53673934510.937858980.7804762670.437600081BMIContinuous Variable1.9728616471.0447400341.8883756560.06289512†WHRContinuous Variable96.7103050487.215247021.1088692440.271077418PBFContinuous Variable-0.3933263750.75398295-0.5216648130.603461639p < 0.05, † p < 0.10Model fit statistics: Multiple linear regression for the predictors of Total Cholesterol p < 0.05, † p < 0.10 Model fit statistics: Age was a statistically significant positive predictor (β = 1.074, SE = 0.521, p = 0.043) when all other variables were controlled for in the model, though it only increased total cholesterol by about 1.07 mg/dL per additional year. Occupationally, Retired Army resulted in significantly lower total cholesterol as compared to desk jobs (β = -75.994, SE = 33.960, p = 0.028). Several predictors were marginally significant at p < 0.10. The Others’ ethnic group was marginally significant with Brahmin as a reference group (β = -28.934, SE = 15.384, p = 0.064). Response for Strenuous Uniformed Jobs was marginally significant positive (β = 30.188, SE = 17.175, p = 0.083) with desk jobs as comparing group. BMI was marginally significant with a positive beta of 1.973(β = 1.973, SE = 1.045, p = 0.063). The multivariable model did not find any statistically significant independent associations of other demographic, lifestyle (male, ethnicity, desk job, exercise, no smoking, alcohol, good diet) or anthropometric (WHR, PBF) variables on total cholesterol. The categorical variables were referenced on the following: Female (Sex), Brahmin (Ethnicity), Desk Job (Occupation), Joint Family (Family Type) and “No” (Exercise, No Smoking, No Alcohol). Statistically significant as *p < 0.05 and †p < 0.10. The multiple linear regression results for the independent association of demographic, occupational, lifestyle, and anthropometric variables affecting Triglyceride levels is shown in Table 9. The regression model was statistically significant, F (21, 74) = 1.795, p = 0.035, with an explanatory power of nearly 33.7% in triglycerides (R² = 0.337; Adjusted R² = 0.149). Table 9Multiple linear regression for the predictors of TriglyceridesPredictorCategoryEstimateStd. Errort valuePr(>|t|)Significance(Intercept)-38.0143157.5572-0.241270.810011Age (Years)Continuous Variable1.5744851.1749091.340090.184316SexFemaleReferenceMale6.06569523.047810.2631790.793144EthnicityBrahminReferenceChhetri-21.309120.53846-1.037520.302872Dalit-158.60468.77062-2.306280.023897*Janjati-4.8558920.20066-0.240380.810698Madheshi-35.227834.11922-1.032490.305204Muslim14.3003567.300620.2124850.832313Others-46.85534.69898-1.350330.181026OccupationDesk jobReferenceActive Standing Job6.0636719.576490.3097420.757627Housewife-8.3900835.83275-0.234150.815518Retired Army-107.92476.59485-1.409020.163018Strenuous Uniformed Job-26.773838.73725-0.691160.491625Student-12.082922.27576-0.542430.589155Family TypeJointReferenceNuclear-12.421720.18975-0.615250.54028ExerciseNoReferenceYes-16.643620.65757-0.805690.423004SmokingNoReferenceYes24.5570328.169730.8717530.386163Alcohol IntakeNoReferenceYes-7.0215119.86367-0.353480.72473DietNon-VegReferenceVeg11.8440524.669940.48010.632572BMIContinuous Variable5.779382.3563732.452660.016537*WHRContinuous Variable71.45267196.71080.3632370.717463PBFContinuous Variable-1.761811.700581-1.036010.303573p < 0.05Model fit statistics: Multiple linear regression for the predictors of Triglycerides p < 0.05 Model fit statistics: The model analysis showed that when controlling for other covariates, BMI was a statistically significant positive predictor of triglyceride levels (β = 5.779, SE = 2.356, p = 0.017). A 1-unit increase in BMI is estimated to be associated with an additional 5.78 mg/dL level of triglycerides. In terms of ethnicity, the Dalit subjects showed lower levels of triglycerides when compared with the Brahmin reference (β = -158.604, SE = 68.771, p = 0.024). Other ethnicities did not show significant differences with the Brahmin group. The rest of the remaining variables namely age, gender, occupation, family type, exercise, smoking status, alcohol consumption, diet, waist to hip ratio (WHR), and percent body fat (PBF) were not significantly associated with triglycerides (TGs) in this multivariable model. The categorical levels were assigned reference groups as: Female (Sex), Brahmin (Ethnicity), Desk Job (Occupation), Joint Family (Family Type) and “No” (Exercise, Smoking, Alcohol intake). *p < 0.05 was considered significant. Multiple linear regression analysis results of HDL cholesterol is shown in the Table 10. Overall, the model was not statistically significantly (F(21,74) = 1.076, p = 0.391). The overall model fit results indicated that predictor variables minimally explained the variation of HDL cholesterol (Adjusted R2 = 0.0166). Table 10Multiple linear regression for the predictors of HDL cholesterolPredictorCategoryEstimateStd. Errort valuePr(>|t|)Significance(Intercept)62.1291818.131583.4265730.001001**Age (Years)Continuous Variable0.276910.1352082.048030.044106*SexFemaleReferenceMale-4.248292.652327-1.601720.113478EthnicityBrahminReferenceChhetri-2.090732.363553-0.884570.379252Dalit0.1270697.9140790.0160560.987233Janjati1.3235882.3246790.5693640.570834Madheshi1.8737873.9264180.4772260.634608Muslim-7.388577.744912-0.953990.343195Others-0.505263.993136-0.126530.899654OccupationDesk jobReferenceActive Standing Job-0.686522.25285-0.304730.761425Housewife-7.470264.123609-1.811580.074108†Retired Army-5.185618.814486-0.588310.558118Strenuous Uniformed Job5.8844824.4578571.3200250.190896Student3.29142.563481.2839580.203162Family TypeJointReferenceNuclear-0.554562.323423-0.238680.812011ExerciseNoReferenceYes-0.682142.377259-0.286950.774956SmokingNoReferenceYes-0.314983.241754-0.097160.922858Alcohol IntakeNoReferenceYes2.3624252.2858991.0334780.304746DietNon-VegReferenceVeg2.0871572.8390.7351730.464558BMIContinuous Variable-0.001710.27117-0.006310.994981WHRContinuous Variable-37.681722.63735-1.664580.100225PBFContinuous Variable0.1502150.1957020.767570.445185p < 0.05 (*), p < 0.01 (**), p < 0.10 (†)Model fit statistics: Multiple linear regression for the predictors of HDL cholesterol p < 0.05 (*), p < 0.01 (**), p < 0.10 (†) Model fit statistics: Among the continuous variables examined, only Age was a significant predictor (β = 0.277, SE = 0.135, p = 0.044), such that increased age was positively associated with increased HDL cholesterol levels. Age was associated with an increase in HDL cholesterol of approximately 0.28 units for each additional year. There was a marginally significant negative association between WHR and HDL (β = −37.682, p = 0.100). Other continuous variables, including BMI (p = 0.995) and PBF (p = 0.445) were not significant predictors. Among the categorical predictors, Sex was not a significant predictor of HDL, however there was a trend for males to have lower HDL cholesterol levels compared to females (β = −4.248, p = 0.113). There was no significant influence for Ethnicity for any of the classifications in comparison to the reference group, Brahmin (all p > 0.34). With regard to Occupation, the Housewife classification demonstrated a marginally significant negative correlation when compared to desk jobs (β = −7.470, p = 0.074). All other classifications related to occupations were not significant. There were no significant predictors of family type, exercise, smoking habits, alcohol consumption and diet (all p > 0.30). In short, the only significant predictor for HDL cholesterol was age, which was found to be positively associated with this cholesterol type. Finally, the model had a poor explanatory power. This shows that the study variables have poor utility for predicting population variation in HDL cholesterol level. Table 11 shows the multiple linear regression results for LDL cholesterol level. The model was not statistically significant, F(21,74) = 1.553, p = 0.086, and the model adjusted R² = 0.1089, indicating that approximately 11% of the variance can be explained by the predictors in the model. Table 11Multiple linear regression for the predictors of LDL cholesterolPredictorCategoryEstimateStd. Errort valuePr(>|t|)Significance(Intercept)-28.34248.43937-0.58510.560259Age (Years)Continuous Variable0.7040670.3612141.9491670.055066†SexFemaleReferenceMale-3.048057.085819-0.430160.668328EthnicityBrahminReferenceChhetri1.115656.3143450.1766850.860239Dalit5.65052121.142840.2672550.790017Janjati-7.078576.210491-1.139780.258056Madheshi1.73557210.489610.1654560.869036Muslim13.5862720.69090.656630.513456Others-11.039710.66785-1.034850.304107OccupationDesk jobReferenceActive Standing Job0.407436.0185950.0676950.946211Housewife3.60946711.016420.3276440.744106Retired Army-58.137223.54832-2.468850.015863*Strenuous Uniformed Job10.7615411.909380.9036190.36913Student5.4177936.848460.7910970.431416Family TypeJointReferenceNuclear-4.487226.207137-0.722910.472012ExerciseNoReferenceYes-1.033326.350962-0.16270.871196SmokingNoReferenceYes-5.435188.660501-0.627580.532209Alcohol IntakeNoReferenceYes6.8876686.1068871.1278520.263026DietNon-VegReferenceVeg1.1287937.5845250.1488280.882094BMIContinuous Variable0.604230.7244430.8340620.40693WHRContinuous Variable101.339460.476761.6756750.098023†PBFContinuous Variable-0.38570.522827-0.737730.463012p < 0.05, † p < 0.10Model fit statistics: Multiple linear regression for the predictors of LDL cholesterol p < 0.05, † p < 0.10 Model fit statistics: Among the continuous variable, age showed a marginally significant positive association with LDL cholesterol (β = 0.704, SE = 0.361, p = 0.055) such that for every year increase of age, LDL cholesterol increased on average about 0.70 unit. Waist-to-Hip Ratio (WHR) also showed positive marginally statistically significant association (β = 101.339, SE = 60.477, p = 0.098) while BMI (p = 0.407) and Percent Body Fat (p = 0.463) was not significant. The only significant categorical predictor was Occupation. Retired Army personnel had significantly lower LDL when compared to desk job holders (β = −58.137, SE = 23.548, p = 0.016). All other predictors were non-significant and included Occupation type, Sex (p = 0.668), Ethnicity (all p > 0.25), Family type (p = 0.472), Exercise (p = 0.871), Smoking (p = 0.532), Alcohol intake (p = 0.263) and Diet (p = 0.882). Conclusively, it can be said that Retired Army occupation type is a significant predictor of lower LDL cholesterol. Age and WHR were still marginally significant, this time with a positive relation with the target variables. The overall model is significant but has a very low power which tells us that these predictors can hardly predict any changes in LDL cholesterol in this population. The analysis of multiple linear regression depicting the result for VLDL cholesterol has been demonstrated in Table 12. The model was significant (F(21,74) = 2.519, p = 0.002) with the adjusted R² = 0.2514 which indicates that the predictors explain nearly 25% variance in VLDL cholesterol levels. Table 12Multiple linear regression for the predictors of VLDL cholesterolPredictorCategoryEstimateStd. Errort valuePr(>|t|)Significance(Intercept)-29.708630.47356-0.97490.332787Age (Years)Continuous Variable0.3068520.2272421.3503270.181027SexFemaleReferenceMale3.6769544.457740.8248470.412109EthnicityBrahminReferenceChhetri-5.334853.9724-1.342980.183384Dalit-34.413913.30111-2.587290.011636*Janjati-0.83123.907065-0.212740.832113Madheshi-1.424456.599092-0.215860.829694Muslim4.51690813.01680.3470060.729572Others-12.91256.711225-1.924010.058199†OccupationDesk jobReferenceActive Standing Job0.9179533.7863420.2424380.809112Housewife-3.162876.930509-0.456370.649461Retired Army-23.217514.81442-1.567220.121329Strenuous Uniformed Job-5.335917.492277-0.712190.478588Student-3.664274.308416-0.850490.397796Family TypeJointReferenceNuclear-1.709413.904955-0.437750.662839ExerciseNoReferenceYes-4.391463.995436-1.099120.275279SmokingNoReferenceYes5.7449415.4483841.054430.295117Alcohol IntakeNoReferenceYes-3.573583.841887-0.930160.355312DietNon-VegReferenceVeg1.6472514.771480.3452290.730902BMIContinuous Variable1.1099510.4557522.4354250.017283*WHRContinuous Variable39.3975938.046371.0355150.303801PBFContinuous Variable-0.354780.328914-1.078650.284246p < 0.05, † p < 0.10Model fit statistics: Multiple linear regression for the predictors of VLDL cholesterol p < 0.05, † p < 0.10 Model fit statistics: Results indicated a significant positive association of VLDL cholesterol with BMI among continuous variables (β = 1.110, SE = 0.456, p = 0.017). An increase of 1 unit in BMI would increase VLDL cholesterol by nearly 1.11 units. Age (p = 0.181), Waist to Hip Ratio (WHR) (p = 0.304) and Percent Body Fat (PBF) (p = 0.284) were non-significant.There were one categorical predictor of Ethnicity which was significant. The Dalit ethnic group had a significantly lower value of VLDL compared to the Brahmin reference group (β = −34.414, SE = 13.301, p = 0.012). Other ethnicity showed marginally significant and negative estimate coefficients (β = −12.913, p = 0.058). All other groups Chhetri, Janjati, Madheshi, Muslim with p > 0.18 were non-significant. Variables that were not statistically significant were Sex (p = 0.412), Occupation (all p > 0.12), Family type (p = 0.663), Exercise (p = 0.275), Smoking (p = 0.295), Alcohol intake (p = 0.355) and Diet (p = 0.731). BMI and caste ‘Dalit’ were statistically significant variables for VLDL cholesterol. One unit increase in BMI was associated with increase in VLDL cholesterol in all cases. The average VLDL cholesterol was significantly reduced in ‘Dalit’ caste when compared to the reference group ‘Brahmins’. Overall, the model (BMI, caste) accounted for significant variation in VLDL cholesterol levels as it seems to have moderate discriminative ability contributing to one-fourth of explained motion in VLDL cholesterol levels in present cohort. Discussion The relationship of body fat percentage and cholesterol among patients of Healthy Choice Clinic, Kathmandu was assessed through this cross-sectional study with 106 adult patients. This study provided useful correlations of anthropometric, lifestyle and lipid profile characteristics of patients in current urban Nepali setting. Overweight (51.9%) and obesity (21.7%) were prevalent among the study participants, with 86.8% having WHR above normal. The observation is consistent with existing literature corroborating urbanization, sedentary lifestyle, and diet transition as a significant marker for the alarming obesity status in the South Asian population. Also, due to its relationship with cardiovascular risk, central obesity observed as WHR above normal, is alarming [14]. Meanwhile, our results showed no significant associations between total cholesterol and WHR and BFP. There could be a few reasons for this. First, our study population had a relatively young age (29.9 years old) and a majority of participants with normal total cholesterol (97.2%). This caused a lower probability for the presence of associations. In younger populations, lipid metabolism is still optimal, and thus, adiposity has lesser impact on dyslipidemia [15]. Second, total cholesterol alone is too simple to describe lipid metabolism; our findings showed several significant associations between anthropometric measurements with triglycerides/VLDL, suggesting that fat buildup has greater impacts on these markers. Other literature reports inconsistent results for BFP associations with cholesterol indicators, suggesting that population characteristics have to be considered [16]. The positive relationship between BMI and total cholesterol (p = 0.022), which indicated higher cholesterol levels among higher obesity categories, was in line with known links between body fat and dyslipidemia [17]. Likewise, the association between total cholesterol and alcohol consumption (p = 0.044), although statistically significant, needs to be taken cautiously because all three patients with high cholesterol were alcohol consumers. BMI was an independent positive predictor of triglycerides (β = 5.779, p = 0.017) and VLDL cholesterol (β = 1.110, p = 0.017), and had borderline association with total cholesterol (β = 1.973, p = 0.063) in our multivariable linear regression. According to a study conducted in 2018 by Khanal et al., it was found that overweight and obese populations in rural Nepal had higher prevalence of hypertriglyceridemia (27.2%) and that there was a positive association of waist circumference and BMI with triglycerides [18]. These results confirm our bivariate associations, and are consistent with existing data from South Asian populations that show associations between generalized adiposity and inflammatory lipid fractions, since this ethnicity appears to have an excess of dyslipidemia associated with insulin resistance and enhancement of hepatic VLDL secretion. The waist-to-hip ratio was associated with marginally significant negative coefficient independent of the other predictors, for HDL (β = −37.682, p = 0.100) and positive for LDL (β = 101.339, p = 0.098). Interestingly, percent body fat was not able to independently predict any of our lipid parameters. This suggests that, in this young urban cohort, the BMI may be a better predictor of metabolic consequences of excess adiposity than the percent body fat or waist-to-hip ratio itself. It is possible that BMI accounts for the effects of gain in lean and fat mass on hepatic VLDL synthesis, which explain triglyceride levels better. It is also possible that measures of central obesity correlate more closely with cardiovascular risk mediated by dyslipidemia in South Asian populations [19]. Apart from anthropometric (BMI) measures, multivariable analysis revealed that age is an independent determinant of total cholesterol (β = 1.074, p = 0.043) and HDL (β = 0.277, p = 0.044), possibly due to established age-related alterations in lipid metabolism manifesting as increased cholesterol production and modified clearance of lipoproteins. Occupation is another independent influence as retired army personnel had significantly reduced total cholesterol (β = -75.994, p = 0.028) and LDL (β = -58.137, p = 0.016) as compared to office workers, which is likely attributed to the residual cardiometabolic impact of intensely physical training, as military or veteran employees with superior fitness status report healthier lipid profiles and lesser cardiometabolic risk. Other lifestyle variables, including exercise, alcohol consumption, smoking and diet, lost statistical significance in the adjusted model, suggesting that their correlation with lipid levels is likely dependent on age, BMI, and occupation; indeed, prospective studies have shown that the cardiovascular influence of lifestyle factors on lipid levels is usually abated in the multivariable model, hinting that these characteristics exert influence partly via shared metabolic pathways. Interestingly, Dalit caste was an independent determinant of reduced triglycerides (β = -158.604, p = 0.024) and VLDL (β = -34.414, p = 0.012) compared to Brahmin upper caste, under-investigated in terms of caste-related lipid variation in Nepal. However, South Asian countries exhibit considerable ethnic diversity for dyslipidemia prevalence and lipid phenotypes due to genetic polymorphisms and socioeconomics, underscoring the need for “post-hoc” reasoning if unmeasured dietary, occupational, or environmental variables reflect underlying cause for the differences in our study population [20]. Regular physical activity is well investigated factor with protective effect against high levels of VLDL (p = 0.011). The clinical and cardiovascular benefits of habitual exercise in reduction of LDL and stimulating HDL is majorly discussed in study done by Thompson et al. [21]. Supportive to this finding, vigorous physical exercise is encouraged as major intervention in affairs of reducing cardiovascular risk. The lipid profile showed no significant correlation with most of the dietary factors. This might be due to various reasons. First, the sample size was small, which might have reduced the power of the statistical tests to detect subtle correlations. Second, the estimation of the dietary factors was based on a short query rather than using more quantitative methods. Third, the participants were young and health conscious as all were visiting a diet clinic, and therefore their dietary practices might have a narrow variance range. In general, lack of any findings does not infer lack of effect but infers lack of inference as a product of methodological shortcomings. Limitations There are methodological limitations associated with our study findings. Sample size (n = 106) is one of the most significant limitations, coupled with having sparse events for some of the outcomes recorded (There were only 3 participants who had high total cholesterol levels). Reporting conclusions on such sparse events is not methodologically sound and these findings requires replication on a larger sample. Second, the study used a purposive sampling technique with the limiting factor of single diet clinic. Hence, selection bias is inevitable and generalizability of results is badly affected. The study population was found to be predominantly young age, highly educated and from urban area of Kathmandu which is not representative of the whole population of Nepal. A high overweight/obesity prevalence as reported in the study is also likely overestimated when extrapolated to general population. Third, the dietary assessment method used a semi-structured questionnaire, which lacked proper validation against gold standard methods including the 24-hour dietary recall or FFQ. Although adapted from NDHS and pre-tested questionnaire, validation was not done. Fifth, the body fat percentage was assessed using the bioelectrical impedance analysis (BIA). While BIA is a convenient and non-invasive approach, the results may be affected by hydration status and recent food consumption. Measurements were standarized, however, BIA is less accurate compared to reference techniques such as dual-energy X-ray absorptiometry (DEXA). The cross-sectional design of the study also limits the interpretation of causal relationships and temporal order of factors. Nonetheless, this study provides preliminary exploratory data on the associations of body composition, lifestyle factors, and lipid profiles in an urban Nepali population that could guide larger future studies. Conclusions Our study offers some initial insights into the relationships between body composition, lifestyle parameters, and lipid profiles in urban Nepali adults visiting a diet clinic. Despite a lack of significant correlation of body fat percentage with total cholesterol values, we found some exploratory associative trends between BMI and indices of central obesity with cardiovascular risk factors related to their lipid profile, notably with triglycerides and VLDL. Physical activity could be protective against dyslipidemia-related CV risk, at least to some extent for VLDL levels. The multivariable adjustment analyses showed that BMI and age remained as independent correlates of atherogenic lipid fractions. In contrast, lifestyle factors had no independent contribution, further reaffirming the importance of adiposity and age-specific factors on lipid metabolism in this demographic. The observed relationships from these regressions further support the study’s major conclusion that cardiovascular risk profiling among urban Nepali adults should emphasize the global characterization of body composition, as opposed to independent lifestyle factors with lesser influence, along with the exclusion of possible factors related to occupational background and ethnicity as significant, yet under-researched, determinants of dyslipidemia risk. Acknowledgements The author would like to express gratitude to Mr. Bhupal Baniya for supervision and guidance, Prof. Kalpana Tiwari for valuable input, and the management team of Healthy Choice Clinic for providing consent and opportunity to conduct this research. Publication declaration The authors declare that this research has not been published in any other journal, therefore it is declared to comply with BMC publications policy. There are no images used, or other materials which require anonymity, the questionnaire followed informed consent and anonymity. Abbreviations Bioelectrical Impedance Analysis Body Mass Index Cardiovascular Disease High-Density Lipoprotein Low-Density Lipoprotein Non-Communicable Diseases Very Low-Density Lipoprotein Waist-to-Hip Ratio Authors’ contributions Mr.Ayush Adhikari conceived the study, acquired, analyzed, and interpreted data. Ms. Rubina drafted the manuscript and Mr. George contributed in data tabulation, presentation and final editing. Author has read and approved the final manuscript. Funding The author declares no funding received for this study. Data availability The datasets that were used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate The College of applied food and dairy technology ( CAFODAT) Research Committee approved the study protocol. A written informed consent form was provided to each participant. The study protocol was implemented in accordance with the relevant guidelines and regulations. The person from the ethics committee responsible for the approval of this research is Mr. Parbat Thapa Magar. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Rubina Karki, Email: rubukarki4@gmail.com. George Chikondi Samu, Email: georgechikondisamu@outlook.com. References Associated Data Data Availability Statement The datasets that were used and/or analyzed during the current study are available from the corresponding author on reasonable request.