The Inter99 study is a Danish intervention study aimed at reducing the incidence of ischaemic heart disease and type 2 diabetes  (www.Inter99.dk, accessed 7 September 2010). From a population-based sample (n = 61,301) of 30- to 60-year-old individuals, a random sample of 13,016 was invited. The participation rate was 52.5% (n = 6,784). We used data from the baseline examinations, which were performed from 1999 to 2001. Participants gave written informed consent prior to inclusion. The protocol was in accordance with the Helsinki Declaration, approved by the local ethics committee (KA98155) and registered with ClinicalTrials.gov (NCT00289237).
Participants completed a self-administered questionnaire on health, socioeconomic factors, lifestyle and maternal/paternal diabetes status (yes/no/unknown). Lifestyle variables included physical activity, diet, alcohol consumption and smoking. After an overnight fast, a 75 g OGTT was performed. Venous samples for measurement of plasma glucose and serum insulin concentrations were taken at 0, 30 and 120 min and analysed as previously described . Type 2 diabetes was diagnosed according to the WHO 1999 criteria .
Midwife records of 4,744 participants were identified as previously described . Ponderal index, i.e. birthweight(kg)/length(m3), was calculated as a measure of infant body composition. We included 4,644 participants born as singletons and without type 1 diabetes in this study.
Participants born prematurely (before week 37) were compared with a group of singletons born at term (weeks 37–42) and matched for birthweight; this group was referred to as small for gestational age (SGA). The remaining participants born at term were considered appropriate for gestational age. Detailed information on gestational age was available in 363 of the 443 participants born prematurely and fetal growth rates were assessed using birthweight z-scores, calculated as: (birthweight-normal birthweight for gestational age)/SD of normal birthweight. A Swedish reference curve for normal weight was used .
Insulin sensitivity and beta-cell function
Insulin sensitivity was assessed by HOMA of insulin sensitivity (1/HOMA of insulin resistance ) and a modified Matsuda index . Areas under the curve for plasma glucose and serum insulin were calculated using a trapezoidal method. Beta cell function (disposition index) was assessed as the product of insulin secretion (AUCinsulin/AUCglucose) and insulin sensitivity (modified Matsuda index). Participants with known type 2 diabetes were excluded from the analysis of insulin sensitivity, insulin secretion and beta cell function.
The association between birthweight and measures of glucose regulation and type 2 diabetes was investigated in four steps. Model 1 adjusted for age and sex. Model 2 additionally adjusted for adult BMI. Model 3 additionally adjusted for maternal parity and parental diabetes status, with missing or ‘unknown’ information on parental diabetes treated as ‘no diabetes’. Model 4 further adjusted for socioeconomic factors and lifestyle variables (physical activity, diet, alcohol consumption and smoking).
Risk of type 2 diabetes was assessed by logistic regression analysis. Multivariate linear regression analyses were performed to study the relationship between birthweight or prematurity and OGTT-derived measures of glucose regulation. The outcome variables presented in Table 1 were loge-transformed to meet the assumption of normality. Results are in percentage change per 1,000 g increase in birthweight. We did not use correction for multiple testing, as our work was based on the predefined rationale of a potential association between low birthweight and prematurity, and (1) type 2 diabetes, (2) reduced insulin sensitivity and (3) beta cell dysfunction as previously described. SAS version 9.1 (SAS Institute, Cary, NC, USA) was used for statistical analysis.