Incidence of Hospitalization for Heart Failure and Case-Fatality Among 3.25 Million People With and Without Diabetes Mellitus

Abstract

Background: Recent clinical trials of new glucose-lowering treatments have drawn attention to the importance of hospitalization for heart failure as a complication of diabetes mellitus. However, the epidemiology is not well described, particularly for type 1 diabetes mellitus. We examined the incidence and case-fatality of heart failure hospitalizations in the entire population aged ≥30 years resident in Scotland during 2004 to 2013.

Methods: Date and type of diabetes mellitus diagnosis were linked to heart failure hospitalizations and deaths using the national Scottish registers. Incidence rates and case-fatality were estimated in regression models (quasi-Poisson and logistic regression respectively). All estimates are adjusted for age, sex, socioeconomic status, and calendar-year.

Results: Over the 10-year period of the study, among 3.25 million people there were 91, 429, 22 959, and 1313 incident heart failure events among those without diabetes mellitus, with type 2, and type 1 diabetes mellitus, respectively. The crude incidence rates of heart failure hospitalization were therefore 2.4, 12.4, and 5.6 per 1000 person-years for these 3 groups. Heart failure hospitalization incidence was higher in people with diabetes mellitus, regardless of type, than in people without. Relative differences were smallest for older men, in whom the difference was nonetheless large (men aged 80, rate ratio 1.78; 95% CI, 1.45-2.19). Rates declined similarly, by 0.2% per calendar-year, in people with type 2 diabetes mellitus and without diabetes mellitus. Rates fell faster, however, in those with type 1 diabetes mellitus (2.2% per calendar-year, rate ratio for type 1/calendar-year interaction 0.978; 95% CI, 0.959-0.998). Thirty-day case-fatality was similar among people with type 2 diabetes mellitus and without diabetes mellitus, but was higher in type 1 diabetes mellitus for men (odds ratio, 0.96; 95% CI, 0.95-0.96) and women (odds ratio, 0.98; 95% CI, 0.97-0.98). Case-fatality declined over time for all groups (3.3% per calendar-year, odds ratio per calendar-year 0.967; 95% CI, 0.961-0.973).

Conclusions: Despite falling incidence, particularly in type 1 diabetes mellitus, heart failure remains ≈2-fold higher than in people without diabetes mellitus, with higher case-fatality in those with type 1 diabetes mellitus. These findings support the view that heart failure is an under-recognized and important complication in diabetes mellitus, particularly for type 1 disease.

Keywords: diabetes mellitus; electronic health records; epidemiology; heart failure; incidence; mortality; registries.

Figure 1.

Age-, sex-, and deprivation-adjusted incidence of heart failure hospitalization by diabetes mellitus type, age, and sex. The lines represent the predicted rates obtained from quasi-Poisson regression models of incident heart failure events. The ribbons are 95% confidence intervals. Covariates included in the model were age, sex, deprivation, and diabetes mellitus type, with interaction terms included where these improved model fit. Predictions were made at the median deprivation score. Points represent event rates stratified by age (in years), sex, and diabetes mellitus type. Models are given in full in the Appendix in the online-only Data Supplement.

Figure 2.

Age-, sex-, and deprivation-adjusted trends in incident heart failure hospitalization by diabetes mellitus type, sex, and calendar-year. The lines represent the predicted rates obtained from generalized additive models of incident heart failure events. The ribbons are 95% confidence intervals. Covariates included in the model were age, sex, deprivation, diabetes mellitus type, and calendar year, with interaction terms included where these improved model fit. The model was fit with a log-link and Poisson likelihood, with correction of the standard errors for overdispersion. Penalized thin plate regression splines were used to model nonlinear associations for calendar year by diagnosis type. Predictions were made for men and women aged 50 (as this was the closest decade to the mean age in the general population). Models are given in full in the Appendix in the online-only Data Supplement. See https://ihwph-hehta.shinyapps.io/dm_hf_fig2/ for an interactive version of this plot.

Figure 3.

Age-, sex-, and deprivation-adjusted 30-day case-fatality of incident heart failure hospitalization by age, sex, and diabetes mellitus type. The lines represent the predicted case-fatality proportions obtained from logistic regression models of death. Covariates included in the model were age, sex, deprivation, and diabetes mellitus type, with interaction terms included where these improved model fit. Predictions were made at the median deprivation score. Points represent case-fatality proportions stratified by age, sex, and diabetes mellitus type, with the point size being proportional to the number in the denominator. Models are given in full in the Appendix in the online-only Data Supplement.

Figure 4.

Age-, sex-, and deprivation-adjusted trends in 30-day case-fatality of incident heart failure hospitalization. The lines represent the predicted rates obtained from generalized additive models of heart failure 30-day case-fatality on age, sex, deprivation, diabetes mellitus type, and calendar year, with interaction terms included where these improved model fit, using a logit-link and binomial likelihood. Predictions were made for men and women aged 50 (as this was the closest decade to the mean age in the general population). Models are given in full in the Appendix in the online-only Data Supplement.

Figure 5.

Age-, sex-, and deprivation-adjusted incidence of heart failure hospitalization by diabetes mellitus type, age, and sex in people without previous ischemic heart disease. The lines represent the predicted rates obtained from quasi-Poisson regression models of incident heart failure events. The ribbons are 95% confidence intervals. Covariates included in the model were age, sex, deprivation, and diabetes mellitus type, with interaction terms included where these improved model fit. Predictions were made at the median deprivation score. Points represent event rates stratified by age (in years), sex, and diabetes mellitus type. Models are given in full in the Appendix in the online-only Data Supplement.