Type of insulin | Cumulative exposure time (person-years) | N, proportion of male gender (%) | Numbers of cancers | Crude incidence rate (n/1000 person-years) | Age (±SE) (years) | Average Charlson comorbidity score | |
Actrapid (rapid acting) | Human insulin | 20 147 | 7606 (58%) | 250 | 12.4 (11.6–13.2) | 54.4 (±16.1) | 0.49 (±0.94) |
Mixtard (middle-term acting) | Human insulin | 19 827 | 7288 (54%) | 432 | 21.8 (20.7–22.8) | 60.6 (±12.7) | 0.52 (±0.91) |
Insulatard (long-term acting) | Human insulin | 70 970 | 22 078 (58%) | 1205 | 17.0 (16.5–17.5) | 58.6 (±14.7) | 0.40 (±0.81) |
Novomix (middle-term acting) | Insulin aspart | 24 286 | 11 740 (60%) | 461 | 19.0 (18.1–19.9) | 61.0 (±12.4) | 0.42 (±0.84) |
Novorapid (rapid acting) | Insulin aspart | 14 874 | 5133 (60%) | 137 | 9.2 (8.4–10.0) | 51.0 (±12.0) | 0.25 (±0.67) |
Humalog Mix (middle-term acting) | Insulin lispro | 465 | 154 (61%) | 5 | 10.8 (5.9–15.6) | 58.8 (±12.2) | 0.38 (±0.72) |
Humalog (long-term acting) | Insulin lispro | 1127 | 230 (58%) | 9 | 8.0 (5.3–10.6) | 44.6 (±12.1) | 0.09 (±0.34) |
Lantus (long-term acting) | Insulin glargine | 4483 | 2994 (57%) | 67 | 14.9 (13.1–16.8) | 56.4 (±12.7) | 0.32 (±0.78) |
Levemir (long-term acting) | Insulin detemir | 6378 | 4498 (59%) | 75 | 11.8 (10.4–13.1) | 57.0 (±13.2) | 0.27 (±0.67) |
Sulfonylureas | – | 336 713 | 89 122 (58%) | 6188 | 18.4 (18.1–18.6) | 63.60 (±13.2) | 0.27 (±0.61) |
Metformin | – | 381 229 | 117 359 (56%) | 5450 | 14.3 (14.1–14.5) | 60.4 (±12.7) | 0.20 (±0.52) |
Thiazolidinediones | – | 15 666 | 6377 (57%) | 229 | 14.6 (13.7–15.6) | 60.6 (±12.2) | 0.20 (±0.51) |
Unexposed population | – | 37 409 158 | 3 447 904 (49%) | 293 878 | 7.9 (7.8–7.9) | 48.5 (±18.7) | 0.08 (±0.35) |
Days between initiation of treatment and occurrence of cancer (median (IQR)) | Gastrointestinal cancer, n (proportion of total cancers) | Lung cancer, n (proportion of total cancers) | Urological cancer, n (proportion of total cancers) | Breast cancer, n (proportion of total cancers in women) | Gynaecological cancer, n (proportion of total cancers in women) | Prostate cancer, n (proportion of total cancers in men) | |
Actrapid (rapid acting) | 495 (49–1344) | 73 (29%) | 24 (10%) | 6 (2%) | 28 (28%) | 8 (8%) | 27 (18%) |
Mixtard (middle-term acting) | 603 (129–1270) | 151 (35%) | 55 (13%) | 17 (4%) | 35 (20%) | 10 (6%) | 33 (13%) |
Insulatard (long-term acting) | 658 (170–1447) | 376 (31%) | 132 (11%) | 31 (3%) | 104 (21%) | 54 (11%) | 100 (14%) |
Novomix (middle-term acting) | 392 (143–769) | 130 (28%) | 57 (12%) | 18 (4%) | 43 (22%) | 24 (12%) | 32 (12%) |
Novorapid (rapid acting) | 636 (120–1239) | 32 (24%) | 17 (12%) | 3 (2%) | 17 (23%) | 2 (3%) | 8 (13%) |
Humalog Mix (middle-term acting) | 557 (383–1343) | 3 (60%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) | 1 (25%) |
Humalog (long-term acting) | 1819 (1224–2211) | 2 (22%) | 4 (44%) | 0 (0%) | 0 (0%) | 0 (0%) | 0 (0%) |
Lantus (long-term acting) | 200 (76–591) | 20 (29%) | 4 (6%) | 3 (4%) | 6 (20%) | 4 (13%) | 6 (16%) |
Levemir (long-term acting) | 265 (91–541) | 21 (28%) | 9 (12%) | 0 (0%) | 5 (18%) | 2 (7%) | 7 (15%) |
Sulfonylureas | 910 (330–1759) | 1681 (27%) | 666 (11%) | 307 (5%) | 542 (22%) | 265 (11%) | 757 (20%) |
Metformin | 836 (323–1606) | 1376 (25%) | 602 (11%) | 219 (4%) | 602 (26%) | 306 (13%) | 626 (20%) |
Thiazolidinediones | 727 (296–1502) | 63 (28%) | 27 (12%) | 15 (7%) | 23 (22%) | 8 (8%) | 27 (21%) |
Unexposed population | – | 61 645 (21%) | 35 595 (12%) | 11 589 (4%) | 45 431 (29%) | 19 187 (12%) | 28 349 (20%) |
Groups are non-exclusive, that is, individuals may contribute to more than one group (except for ‘unexposed population’, which includes only those who did not initiate treatment with glucose-lowering agents or were censored prior to first claimed prescription; individuals from the other groups, however, contributed with exposure time prior to initiation with their first glucose-lowering agent). Mean age and Charlson comorbidity score were calculated as per the date of treatment initiation. For the unexposed group, numbers were calculated as per the calendar time corresponding to half of the length of follow-up.