This study was conducted to i) compare lipid parameters between diabetic children and controls; ii) determine factors influencing lipid parameters in diabetic children; and iii) examine effect of lowering of glycosylated hemoglobin on lipids in diabetic children at one year.

This cross-sectional study with one-year follow-up was done after institutional ethical committee clearance. Considering variability in LDL-C reported in past studies [6], a sample size of 80 diabetics and 54 controls was found to have a power of 0.8 at 5% level of significance and 3% margin of error. Families with diabetic children aged 5 to 17 years attending the diabetes clinic at our institution were approached, and a random sample of 80 (39 boys) were enrolled prospectively (May 2013 to October 2014). Age- and gender-matched healthy controls were recruited from private schools. Patients with known chronic disorders were excluded. Data on medications, age at onset and duration of diabetes and insulin regimen were collected in a standardized form and verified from hospital records.

Anthropometric data were collected using standard methods and were converted to Z scores [7]. Fasting blood sample was collected to measure HbA1c by HPLC (BIO-RAD, Germany) and lipid profile (enzymatic method). The GE-Lunar DPX Pro (GE Healthcare, Wisconsin, USA) was used to measure body composition. Dietary intakes were assessed by 24-hour recall on three random days (non-consecutive)/week, including one holiday. Nutrient intakes were calculated (CDIET, Version 2, Xenios Technologies, Pune, India). Physical activity data were collected using validated activity questionnaires [8,9] adapted for Indian children.

Statistical analyses were carried out using SPSS for Windows 12 (SPSS, Chicago, IL, USA). Anthropometry, biochemical parameters and body composition were recorded in 45 (20 girls) children who regularly attended the clinic at the end of one year.

Table I summarizes the anthropometric, body composition and biochemical parameters. There were no significant differences between sexes for anthropometric and biochemical parameters (P>0.05). Hence, further analysis was performed on pooled data.

TABLE I	Anthropometry, Body Composition and Biochemical Parameters in Study Participants

35% diabetic children had high LDL-C, 18% had low HDL-C and 2% had high triglycerides [10]. HDL-C was significantly lower and LDL-C was significantly higher in diabetic children. Mean (SD) HbA1c of diabetic children was 10 (2), signifying suboptimal blood glucose control.

Using linear regression, metabolic control (HbA1c) (P=0.002) was identified as positive predictor and age at diagnosis (P=0.012) as a negative predictor of total cholesterol. For LDL-C, metabolic control (P=0.001) and age at diagnosis (P=0.027) were predictors. Total body fat % or android fat % did not affect lipid profile. When diabetic children were divided into two groups according to median HbA1c (9.7%) and then compared to controls, diabetic children with HbA1c above median had significantly higher LDL-C and lower HDL-C as compared to controls (Fig. 1).

Fig. 1 Comparison of lipid profile parameters between groups of diabetic children according to HbA1C and controls.

To study impact of lifestyle factors on lipid profile, linear regression models were utilized. Mean (SD) daily intake of energy was 1632 (43) Kcal, protein was 44 (12) g, fat was 45 (12) g and carbohydrate was 267 (75) g. Mean (SD) time spent in moderate activities by children was 37 (25) min/d. Moderate physical activity (P=0.014) was identified as protective factor against LDL-C. Dietary intakes did not affect any of the lipid profile parameters.

Contributors: LP, AK, VK: conceptualized the study, and analyzed data; SC: performed statistical analyses. All authors contributed to the manuscript and approved its contents.

Funding: Mr. Pancharatnam; Ms Lavanya Parthasarathy was funded by a fellowship grant from the University Grants Commission, Government of India.

Competing interests: None stated.