Study design

SAUDI-DM is a nationwide household population-based cross-sectional study using a multistage stratified cluster sampling technique taking into consideration the urban to rural ratio in the 13 administrative regions in Saudi Arabia during the period from 2007 to 2009. The study was conducted through primary healthcare centres by trained physicians and nurses through recruiting Saudi nationals’ family members from every third house in the selected areas. All family members who were available during the visit of the survey team were recruited regardless of their age, gender or diabetes status, excluding participants who refused to participate or were not present during the recruitment visit.

Using the Saudi national regional census survey in 2007, 87 417 participants were recruited and adjusted for age, area of residency and gender distribution excluding 34 047 participants. Of the remaining 53 370 adjusted participants, children and adolescents aged ≤18 years totalling to 23 523 participants were considered the eligible study cohort.

All demographic data which include participants’ age, gender and family history of diabetes were collected by a standard questionnaire. The area of residency was classified according to the Ministry of Municipal and Rural Affairs classification and the family’s monthly income was expressed in Saudi Riyals (SR) (1SR=US$0.267). Any participant known to have diabetes was classified based on the patient's age at diagnosis, history of ketocidosis, and management into either type 1 if the patient was diagnosed before 30 years of age and was using insulin with or without a history of diabetic ketoacidosis, or type 2 if the patients was diagnosed after the age of 30, having an elevated blood glucose which is currently being managed with diet alone or with oral hypoglycaemic agents with or without insulin and with no history of diabetic ketoacidosis. A total of 72 participants who were not sure about their diabetic status and were not on any treatment or not being followed clinically were considered non-diabetic and were subjected to screening using fasting plasma glucose (FPG).

Anthropometric measurements of weight and height were used to calculate body mass index (BMI) for age-specific and gender-specific percentile, where overweight was considered when BMI was from ≥85th to <95th centile and obesity was considered when BMI was ≥95th centile.12 At the same time, systolic and diastolic blood pressure (DBP) were measured in a sitting position using a mercury sphygmomanometer with appropriate cuff size.

Since children <6 years of age have difficulty in fasting more than 10 h overnight and may not be capable of providing a venous blood sample, a total of 6211 (26.52%) children <6 years were not included for FPG assessment. All participants with and without diabetes aged ≥6 years, in addition to children with diabetes aged <6 years, were asked to report to the allocated primary care centres for venous blood collection after 10 h of overnight fasting with the usual activity and diet. The overall response rate for FPG collection was 86.39%.

Blood glucose, serum cholesterol, high-density lipoprotein (HDL), low-density lipoprotein (LDL) and triglyceride assessments were performed at the Strategic Center for Diabetes research laboratory using the Mindray (B5–200) chemistry analyser.

This study was approved by the Institutional Review Board (IRB) at the College of Medicine, King Saud University. All children and adolescents who were participating in this study consented either through their legal guardian or directly if they were older than 14 years, which is the legal age in the Kingdom.

Case identification

According to the American Diabetes Association (ADA) criteria 2003,13 FPG is normal if the level is <110 mg/dL (<6.1 mmol/L) and is considered IFG if the level is between 110 mg/dL (6.1 mmol/L) and 125 mg/dL (6.9 mmol/L), while diabetes is considered when FPG is ≥126 mg/dL (7.0 mmol/L).

Age-specific and gender-specific 85th centile values of systolic blood pressure (SBP) and DBP were used to define high blood pressure among the study participants. Using the National Cholesterol Education programme (NCEP) guidelines,14 triglycerides (TR) were considered high if the level was ≥1.13 mmol/L for participants ≤9 years and 1.47 mmol/L for participants aged 10–18 years. LDL cholesterol was considered high if the level was ≥3.37 mmol/L, while total cholesterol was considered high if the level was ≥5.18 mmol/L. HDL was considered low if the level was ≤0.91 mmol/L.

Statistical analysis

The strengthening reporting of observational study in epidemiology (STROBE) guidelines were used to design and report this study. All data were analysed using Software Package for Social Sciences (SPSS) V.20.0. A t test was used for continuous data, and χ² test for categorical ones. Overweight and obesity were defined using sex-specific and age-specific between 85th and 95th centiles. The same was used for both systolic and DBP to define high blood pressure at the 85th centile. Imputation, based on a regression model, was used to estimate missing data. The prevalence of diabetes and IFG was estimated for the whole study population and for population subgroups taking into account the sampling weight. Pearson’s correlation (r) was used to assess the strength of association between the age and the prevalence of known type 1, type 2 and new diabetes and IFG cases. Factors that were known as being significantly associated with childhood diabetes were assessed using univariate, age-adjusted and gender-adjusted, and multivariate logistic regression models. OR and its 95% CIs were used to express different risks. A p value of <0.05 was chosen as the level of significance.