Risk of cancer incidence and mortality associated with diabetes: A systematic review with trend analysis of 203 cohorts.

AIM: Whether the relative risk of cancer incidence and mortality associated with diabetes has changed over time is unknown. DATA SYNTHESIS: On August 12th, 2020, we electronically searched for observational studies reporting on the association between diabetes and cancer. We estimated temporal trends in the relative risk of cancer incidence or mortality associated with diabetes and calculated the ratio of relative risk (RRR) comparing different periods. As many as 193 eligible articles, reporting data on 203 cohorts (56,852,381 participants; 3,735,564 incident cancer cases; 185,404 cancer deaths) and covering the period 1951-2013, were included. The relative risk of all-site cancer incidence increased between 1980 and 2000 [RRR 1990 vs.1980: (1.24; 95% CI: 1.16, 1.34); 2000 vs.1990: (1.23; 1.15, 1.31)] and stabilised thereafter at a relative risk of 1.2; the relative risk of all-site cancer mortality was constant at about 1.2 from 1980 to 2010. Both magnitudes and trends in relative risk varied across cancer sites: the relative risk of colorectal, female breast, and endometrial cancer incidence and pancreatic cancer mortality was constant during the observed years; it increased for bladder, stomach, kidney, and pancreatic cancer incidence until 2000; and decreased for liver while increased for prostate, colon and gallbladder cancer incidence after 2000. CONCLUSIONS: Alongside the increasing prevalence of diabetes, the temporal patterns of the relative risk of cancer associated with diabetes may have contributed to the current burden of cancer in people with diabetes.

Statins and risk of thromboembolism: A meta-regression to disentangle the efficacy-to-effectiveness gap using observational and trial evidence.

BACKGROUND AND AIMS: Meta-analyses of randomised controlled trials (RCTs) and observational studies indicate a lower risk of venous thromboembolism (VTE) associated with statin treatment. We aimed to compare the effect of statin therapy in these two settings and to identify and quantify potential factors to explain statin efficacy and effectiveness. METHODS AND RESULTS: We electronically searched on December 11th, 2018, articles reporting on first VTE events in RCTs (statin vs placebo) and in observational studies (participants exposed vs non-exposed to statin). We performed Knapp-Hartung random-effect meta-analyses to calculate pooled relative risks (RRs) of VTE events associated with statin treatment, separately for RCTs and observational studies; and estimated the ratio of the relative risk (RRR) comparing RCTs and observational studies using meta-regressions, progressively adjusted for study-level characteristics. Twenty-one RCTs (115,107 participants; 959 events) and 8 observational studies (2,898,096 participants; 19,671 events) were included. Pooled RRs for RCTs and observational studies were 0.82 (95% confidence interval (CI): 0.67-1.00; I2 19.2%) and 0.60 (95% CI: 0.42-0.86; I2 86.3%), respectively. In meta-regressions, the unadjusted RRR indicated a nonsignificant 23% smaller benefit in RCTs (RRR 0.77; 95% CI: 0.52-1.13); accounting for age, sex, geographical region, and duration of follow-up, there was a sensible change of the RRR which resulted 0.30 (95% CI: 0.13-0.68). CONCLUSION: Differences in the characteristics between patients included in RCTs and those in observational studies may account for the differential effect of statins in preventing VTE in the two settings.

Association of ethnicity and socioeconomic status with health outcomes in women with gestational diabetes: Clinical practice research datalink cohort study.

AIMS: To investigate in women with prior gestational diabetes mellitus (GDM), differences by ethnicity and socioeconomic status in the incidence of recurrent GDM, type 2 diabetes (T2D), hypertension, and depression. METHODS: This was a retrospective cohort study including 10,868 women diagnosed with GDM in the Clinical Practice Research Datalink (CPRD GOLD) between January 01, 2000 and November 05, 2018. Linked data were obtained for Hospital Episode Statistics and the Index of Multiple Deprivation. We estimated incidence rates and hazard ratios, by ethnicity and socioeconomic status. RESULTS: During a follow-up of 58,479 person years (mean (SD): 5.38 (3.67) years), the crude incidence was 9.67 (95 % confidence interval: 9.30-10.00) per 100 person years for recurrent GDM, 3.86 (3.70-4.02) for depression, 2.15 (2.03-2.27) for T2D and 0.89 (0.81-0.97) for hypertension. South Asian ethnicity was associated with an increased risk of T2D compared to White (adjusted hazard ratio: 1.65; 1.34-2.05) and Black ethnicity was associated with a greater risk of hypertension (2.93; 1.93-4.46). Black and South Asian ethnicity were associated with a reduced risk of depression compared to White: 0.23 (0.13-0.39) and 0.37 (0.29-0.46), respectively. Incidence rates were higher for all conditions with increasing deprivation level. CONCLUSIONS: The risk of health complications in women with a prior history of GDM differs by ethnicity and socio-economic status, suggesting the opportunity for targeted assessment in the years following pregnancy. These findings may inform future guidelines on screening for health outcomes in women with GDM.

Screening for type 2 diabetes after a diagnosis of gestational diabetes by ethnicity: A retrospective cohort study.

AIMS: To estimate rates and identify determinants of post-partum glucose screening attendance in women with a history of gestational diabetes mellitus (GDM). METHODS: Retrospective cohort study using the Clinical Practice Research Datalink linked to Hospital Episode Statistics, to identify women diagnosed with GDM between 01/01/2000 and 05/11/2018. Age adjusted odds ratios (aOR) and 95% confidence intervals (CI) were estimated using multivariable logistic regression models. RESULTS: In 10,868 women with GDM, with an average follow-up of 5.38 years (95% CI 5.31,5.45), there was an average of 3.79 (95% CI 3.70,3.89) screening episodes per individual, with a mean time to first screening test of 1.22 (95% CI 1.18, 1.25) years. South Asian women had a significantly greater likelihood of being screened compared to White women within the first 5 years post-partum, aOR: 1.89 95% CI (1.20,2.98). A low proportion of women received at least one test per year of follow-up (23.87%). Older age at GDM diagnosis, polycystic ovary syndrome, prescribed medication for GDM, and living in England, were all associated with a greater likelihood of being screened. CONCLUSION: While the majority of women with previous GDM receive at least one glucose screening test within the first 5 years post-partum, fewer than a quarter of them receive on average one test per year of follow-up. Developing strategies to motivate more women to attend screening in primary care is essential.

Association of Type 2 Diabetes With Cancer: A Meta-analysis With Bias Analysis for Unmeasured Confounding in 151 Cohorts Comprising 32 Million People.

BACKGROUND AND PURPOSE: Whether the association between type 2 diabetes (T2D) and cancer is causal remains controversial. The goal of this work is to assess the robustness of the observational associations between T2D and cancer to unmeasured confounding. DATA SOURCES AND STUDY SELECTION: PubMed, Web of Science, and the Cochrane library were systematically searched on 10 January 2019 for observational studies investigating associations between T2D and cancer incidence or mortality. DATA EXTRACTION AND DATA SYNTHESIS: Cohort-level relative risk (RR) was extracted. RRs were combined in random-effects meta-analyses and pooled estimates used in bias analyses. A total of 151 cohorts (over 32 million people, 1.1 million cancer cases, and 150,000 cancer deaths) were included. In meta-analyses, T2D was associated with incidence of several cancers, from prostate (RR 0.83; 95% CI 0.79, 0.88) to liver (2.23; 1.99, 2.49), and with mortality from pancreatic cancer (1.67; 1.30, 2.14). In bias analyses, assuming an unmeasured confounding associated with both T2D and cancer with a RR of 1.5, the proportion of studies with a true effect size larger than a RR of 1.1 (i.e., 10% increased risk in individuals with T2D) was nearly 100% for liver, pancreatic, and endometrial, 86% for gallbladder, 67% for kidney, 64% for colon, 62% for colorectal, and <50% for other cancer incidences, and 92% for pancreatic cancer mortality. LIMITATIONS: Biases other than unmeasured confounding were not analytically assessed. CONCLUSIONS: Our findings strongly suggest a causal association between T2D and liver, pancreatic, and endometrial cancer incidence, and pancreatic cancer mortality. Conversely, associations with other cancers were less robust to unmeasured confounding.