Ethnic differences in the association between waist-to-height ratio and albumin-creatinine ratio: the observational SUNSET study

The study population consisted of participants in the SUNSET study that is based on a random sample of 2975 Surinamese (i.e. with at least one parent born in Surinam) and Dutch individuals, aged 35 to 60 years of age, drawn from the Amsterdam population register [6, 21]. In 1975, almost half of the population of the former Dutch colony Surinam migrated to the Netherlands. It is estimated that approximately 36% of these immigrants were Hindustani-Surinamese (originally from the Indian subcontinent) and 41% African-Surinamese (predominantly of African origin) [22].

Ethnicity was classified according to self-identification. If information on the self-identified ethnicity of the individual was lacking, information on the ethnic group of the mother, the father and the mother’s ancestors was used to classify participants.

Between 2001 and 2003, all subjects in the sample were approached for a face-to-face, structured interview by interviewers who had been matched by sex and origin. The interview included questions on self-identified ethnicity, migration history, demographic variables, lifestyle, and health status.

Immediately after the interview, participants were invited for a medical examination. Before the physical examination, trained physicians and research assistants verified that participants were fasting. During the examination, staff recorded: weight in light clothing on a SECA mechanical scale (SECA gmbh & co, Hamburg, Germany) to the nearest 200 grams; height to the nearest 0.01 meter by tape measure with participants standing against a wall, heels together, feet at a 45 degree angle and head positioned in the Frankfurt plane; waist circumference midway between the lower rib margin and the iliac crest and hip circumference at the maximum point over the greater trochanters to the nearest 0.01 meter by tape measure. After the subjects had emptied their bladder and had been seated for at least 5 minutes, blood pressure and resting heart rate measurements were obtained from each subject’s arm at heart level using an OMRON-M4 semi-automatic sphygmomanometer (Omron Healthcare Europe BV, Hoofddorp, the Netherlands) with an appropriate-sized cuff. All anthropometric measurements and blood pressure measurements were obtained twice and the means were used for analysis.

Fasting glucose concentrations (mmol/l; HK/Glucose-6-P dehydrogenase test; P800 analyser, Roche Diagnostics, Indianapolis, IN) were determined in samples obtained at the time of the physical examination. Finally, urinary albumin (mg/l; Jaffé method; P800 analyser, Roche Diagnostics) and creatinine concentration (mmol/l; Immune-turbidimetry; P800 analyser, Roche Diagnostics) were measured in an early morning urine sample, collected on the day of the physical examination. These measurements were carried out at the Laboratory of Clinical Chemistry of the Academic Medical Center of the University of Amsterdam.

The overall response to the interview was 60% (Figure 1). Participants in the interview were more likely to be female, married and living with a partner and/or children, and to live in an area with a density < 2500 addresses/km². However, the differences between participants and non-participants were small and trends were similar across ethnic groups (data not shown). Of those who were eligible (n = 1626), 89% participated in the physical examination (Figure 1).

In the present analysis, we included subjects who had participated in the medical examination and had complete data on waist circumference, height, urine albumin and creatinine: 334 Hindustani-Surinamese, 589 African-Surinamese and 493 Dutch. As compared to those whose data were analyzed, those who did not undergo a physical examination or had incomplete data were similar with regard to sex, self-reported morbidity and self-rated health (data not shown).

First, we calculated the ACR by dividing the urine albumin concentration by the creatinine concentration (mg/mmol). Due to the skewed distribution, the resulting values were log-transformed before analysis (logACR). Because of the occurrence of albumin concentrations of 0 mg/l (n = 41), a constant of 0.005 was added prior to transformation and, therefore, no back-transformations could be performed. Microalbuminuria was defined as an ACR 2.5-30 mg/mmol for men and 3.5-30 mg/mmol for women and macroalbuminuria as an ACR ≥ 30 mg/mmol [23] Due to the relatively small numbers of persons with macroalbuminuria, these categories were combined for analysis.

Then, WHtR was determined by dividing the waist circumference (m) by the measured height (m). High WHtR was defined as a value greater than 0.50 [15].

Finally, we defined the body mass index by weight (kg) divided by height (m²); type 2 diabetes by fasting glucose ≥ 7.0 mmol/l and/or self-reported type 2 diabetes, excluding the self-reported diagnosis of gestational diabetes; hypertension by a SBP ≥ 140 mm Hg, or DBP ≥ 90 mm Hg, or being on anti-hypertensive therapy.

Overall differences in characteristics between groups were assessed using Chi-square tests for categorical data and analysis of variance or Kruskal-Wallis tests for continuous measures. Then, post hoc tests were used to determine whether the characteristics of the Hindustani Surinamese and the African Surinamese differed from the characteristics of the Dutch.

Next, we used linear regression analysis to estimate the association between the WHtR and logACR in the total population. We adjusted for ethnicity and known determinants of ACR: sex, age, BMI, type 2 diabetes and hypertension. Moreover, we considered interaction by sex and age, as the association between body composition and chronic kidney disease has been reported to be age- and sex-dependent [24]. In addition, to verify the consistency of the associations found among subjects with and without pre-existent morbidity, we considered interaction due to the presence of hypertension or type 2 diabetes.

Subsequently, we examined the association of WHtR and logACR, stratified by ethnicity. We also tested formally for interaction between WHtR and ethnicity by adding a multiplicative interaction term to the fully adjusted model.

All analyses were performed using the SAS package, version 9.1 (SAS Institute Inc., Cary, NC). P-values ≤0.05 for the likelihood ratio test (≤0.10 for interaction), were considered statistically significant.

The SUNSET-study was approved by the Institutional Review Board of the Academic Medical Center, and carried out in accordance with the Helsinki Declaration. All participants provided a written informed consent.

Dutch participants were older and had a higher level of education than Hindustani-Surinamese and African-Surinamese participants (Table 1). Among men, the African Surinamese smoked more frequently than the Dutch, while among women the Hindustani Surinamese and the African Surinamese smoked less frequently than the Dutch. Among women, but not men, the Dutch more frequently complied with the guideline for physical activity than the Hindustani Surinamese. Dutch women, but not Dutch men, had a lower mean BMI than the Hindustani-Surinamese and African-Surinamese. Type 2 diabetes (in men and women) and hypertension (women only) were more prevalent among Hindustani-Surinamese and African-Surinamese than Dutch participants.

The mean WHtR among Hindustani Surinamese men and women and among African Surinamese women was higher than the mean WHtR among Dutch men and women (Table 1). Hindustani-Surinamese men had the highest median ACR of 0.36 (0.20-0.91), followed by the African-Surinamese 0.26 (0.16-0.47) and the Dutch 0.24 (0.16-0.45) mg/mmol (Table 1). The prevalence of albuminuria did not differ across the groups among men. Among women, the Hindustani-Surinamese and African-Surinamese women had a higher median ACR than the Dutch (median ACR of 0.48 (0.30-0.97), 0.40 (0.24-0.76) and 0.34 (0.20-0.59) mg/mmol, respectively). A similar pattern was observed for the prevalence of albuminuria.

In the total population, the logACR increased by 0.359 (0.278-0.440) for every 0.1 point increase in the WHtR (Table 2). The association remained similar after adjustment for sex, age, and smoking, BMI and the presence of type 2 diabetes or HT; the association was 0.368 (0.174-0.562). No significant interaction was observed for age and sex, and type 2 diabetes or hypertension (Table 3).

Among the Hindustani-Surinamese, the adjusted association between WHtR and logACR appeared stronger than among the other ethnic groups (Table 4). For every unit increase in the WHtR, the logACR increased by 0.522 (0.096-0.949) log mg/mmol among the Hindustani-Surinamese, by 0.334 (0.047-0.622) among the African-Surinamese and by 0.356 (−0.010-0.721) among the Dutch. However, the formal test for interaction provided no evidence for a differential association across ethnic groups (Table 5).

Central obesity, defined by the WHtR, was associated with a higher ACR among a multiethnic population of men and women aged 35–60 years. The association was consistent across sex and age groups and was not accounted for by the potentially mediating effect of pre-existent morbidity.

The association seemed somewhat stronger among subjects of Hindustani-Surinamese origin than among subjects of African-Surinamese or Dutch origin, although formal testing of interaction provided no further evidence for a differential association across ethnic groups.

Our study also has some limitations that should be discussed. Firstly, any associations found between WHtR and ACR have to be interpreted with caution. Due to the cross-sectional nature of the study, no causal inferences can be made.

Second, we used WHtR based on a double measurement of the waist circumference and height of participants as a marker of central obesity. This type of measure is often chosen in epidemiological studies, although several more complex methods are available to quantify central obesity, such as magnetic resonance images, computed tomography, and dual-energy X-ray absorptiometry [40‐42]. However, these are relatively time-consuming and costly and thereby less suitable for epidemiological studies. Nevertheless, an imperfect marker may have led to underestimation of the associations. However, as the bias is expected to be similar across ethnic groups, the differences in the associations between groups are expected to remain.

Moreover, we used the ACR based on a single urine sample. It has been argued that the use of a single urine sample may be less accurate than a 24-hour urine collection. Nevertheless, the ACR from a single untimed specimen has been shown to correlate well with a 24-hour collection, and is more practical and less expensive than 24-hour collections for epidemiological studies [43‐45].