br Methods Study population Our
Methods Study population: Our study was based on electronic records from the largest health maintenance organization in Israel, Clalit Health Services, insuring 53% (4.3 million) of the nation's population. All persons 21–89 years old on January 1, 2002 were included in a closed cohort that was followed until December 31, 2012. Individuals free of diabetes at study-entry but diagnosed with diabetes during follow-up were followed for cancer-incidence from the time of diagnosis. Individuals diagnosed with diabetes prior to study entry (diabetes-prevalent patients) were followed for cancer incidence throughout. For cancer diagnoses, the cohort data file, comprising detailed high-quality demographic, clinical, and pharmaceutical information, was linked to the Israel National Cancer Registry, established in 1960. The registry benefits from the 1982 national law mandating registration of cancer, with 97% coverage of solid tumors, and approximately 88% coverage of hematologic cancers . Diabetes definition: Incident diabetes was defined as the fulfilment of at least one of the following 6 criteria during the period January 1 2002 to December 31 2010: (1) a record of diabetes in the Clalit Chronic Disease Registry; (2) a physician's diagnosis of diabetes together with a plasma LY3009120 test > 126 mg/dl within 12 months; (3) HbA1c of 6.5% (47.5 mmol/mol) or higher; (4) a two-hour plasma glucose during an oral glucose tolerance test ≥ 200 mg/dl (although this criterion accounted for only 0.3% of the patients classified as having incident diabetes); (5) two plasma glucose measurements >126 mg/dl within 12 months; (6) three or more purchases of glucose lowering medications within 12 months. Prevalent diabetes was defined by a record in the Clalit Chronic Disease Registry as having diabetes on January 1st 2002, or fulfilling criterion #6 above (since computerized information on medication use was available from 1998) by the same date.
Results At the start of follow-up 159 104 individuals were classified as having diabetes (prevalent-diabetes) and an additional 386 982 were diagnosed with diabetes during 2002–2012 (incident-diabetes). However, our analyses included only the prevalent-diabetes patients who were alive and free of the cancer of interest on Jan 1 2004, and the incident-diabetes patients who were alive and free of the cancer of interest 2 years after diagnosis. Table 1 shows baseline characteristics of this study population for patients entering the analysis of all sites cancer, according to prevalent and incident-diabetes. The numbers for the analyses of specific types of cancer were similar. The incident-diabetes group was characterized by younger age, having more children, and comprising more past and current smokers. Both diabetes groups had similar distributions of sex, ethnic origin and socioeconomic status. In the total group, during the study period, there were 48 812 cancer events in the incident-diabetes group and 16 785 in the prevalent-diabetes group; however, after exclusions of cancers diagnosed before 2004 or within 2 years of diabetes diagnosis, the numbers were 14 066 in the incident group and 12 821 in the prevalent group (Table 2). There was no apparent association between HbA1c and all-sites cancer in the incident or prevalent-diabetes groups (Table 2 and Fig. 1). However, blood glucose showed a weak positive association (Table 2) with all-sites cancer (HR for a 30 mg/dl increase in mean glucose-level 1.04, 95%CI: 1.02–1.06 among incident diabetes patients, and 1.01, 95%CI 1.00–1.02 among prevalent patients). Null or weak associations with poorer glycemic control were also seen for colorectal, breast, lung and liver cancers (Table 2). In some cases, these associations reached conventional statistical significance: there was a positive association for colorectal cancer in incident cases of diabetes; a negative association for breast cancer with HbA1c level in prevalent cases; and a positive association for lung cancer in prevalent cases. However, all these hazard ratio estimates were close to null (between 0.94 and 1.09).