gms | German Medical Science

GMDS 2012: 57. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

16. - 20.09.2012, Braunschweig

Propensity score analysis to compare basal insulin treatment groups regarding the risk of acute myocardial infarction in type 2 diabetes mellitus

Meeting Abstract

  • Dirk Enders - BIPS – Institut für Epidemiologie und Präventionsforschung, Bremen, Deutschland
  • Sigrid Behr - BIPS – Institut für Epidemiologie und Präventionsforschung, Bremen, Deutschland
  • Bianca Kollhorst - BIPS – Institut für Epidemiologie und Präventionsforschung, Bremen, Deutschland
  • Ulrich Dittmann - BIPS – Institut für Epidemiologie und Präventionsforschung, Bremen, Deutschland
  • Antje Timmer - BIPS – Institut für Epidemiologie und Präventionsforschung, Bremen, Deutschland
  • Franz-Werner Dippel - Sanofi Deutschland GmbH, Berlin, Deutschland
  • Karlheinz Theobald - Sanofi Deutschland GmbH, Berlin, Deutschland
  • Edeltraut Garbe - BIPS – Institut für Epidemiologie und Präventionsforschung, Bremen, Deutschland

GMDS 2012. 57. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e.V. (GMDS). Braunschweig, 16.-20.09.2012. Düsseldorf: German Medical Science GMS Publishing House; 2012. Doc12gmds182

DOI: 10.3205/12gmds182, URN: urn:nbn:de:0183-12gmds1825

Published: September 13, 2012

© 2012 Enders et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by-nc-nd/3.0/deed.en). You are free: to Share – to copy, distribute and transmit the work, provided the original author and source are credited.


Outline

Text

Background: Definite long-term studies comparing the risk of cardiovascular diseases in patients with type 2 diabetes mellitus (DM2) treated with different basal insulin treatments are scarce [1]. Using observational data is promising but requires appropriate methodology to account for the differences between the basal insulin treatment groups.

Objective: To investigate the risk of acute myocardial infarction (AMI) in patients with DM2 treated with long-acting analog insulin compared to neutral protamine Hagedorn (NPH) insulin and premixed insulin using propensity score (PS) matching.

Methods: Administrative data of three German statutory health insurances were used covering the years 2004–2009 to conduct a study in a cohort of patients with DM2 who for the first time initiated insulin treatment. Naïve insulin users were defined as patients who were solely on oral antidiabetic treatment for at least 9 months (baseline period) before the first long-acting insulin analog, NPH or premixed insulin was prescribed. In these three groups, patients were eligible if they received only analog, NPH or premixed insulin during the whole follow-up period. Patients taking any insulin during the baseline period were excluded. The primary analysis of the study was a gender-stratified multivariable Cox regression comparing analog insulin with NPH insulin. PS-matched analyses were conducted to estimate adjusted Hazard Ratios (HRs) of AMI and corresponding 95%-confidence intervals (CI) for analog insulin compared to NPH and premixed insulin. Thereby, separate PS models were defined for the comparison of analog insulin with NPH insulin and premixed insulin, respectively. The covariate balance was assessed for each PS model with graphical methods and by using the standardized difference.

Results: Among 9 million insurants, 21,501 new insulin users with a median follow-up of 771 days in the cohort were identified. Throughout the study, 6,139 patients received NPH insulin, 7,050 patients received analog insulin and 4,334 patients received premixed insulin. Patients treated with premixed insulin were older than patients treated with analog or NPH insulin (mean age 70.7 vs. 64.1 and 61.6 years) and had more comorbidities. The PS-matched cohorts included 83% of the NPH insulin patients (N=5,102) and 72% of the premixed insulin patients (N=3,125). Baseline characteristics were balanced in both PS-matched cohorts. The crude incidence rate was 10.86 AMI/1,000 person-years. Adjusted HRs revealed no difference between NPH and analog insulin (HR: 0.95, 95%-CI: 0.75–1.20) but a higher risk for premixed insulin compared to analog insulin (HR: 1.27, 95%-CI: 1.02–1.59) in the unmatched analysis. The PS analysis matching NPH to analog insulin patients resulted in similar results (HR: 0.92, 95%-CI: 0.71–1.21). The PS analysis matching premixed to analog insulin patients resulted in a non-significant risk estimate of 1.05 (95%-CI: 0.81–1.37) indicating that residual confounding caused bias in the unmatched analysis.

Conclusions: PS-matching further improved confounder control compared to the confounder-adjusted Cox regression analysis. This could be a result of excluding high risk patients receiving premixed insulin from the PS-matched cohort. Regarding the risk of AMI, no difference was observed between the basal insulin treatment groups in this observational study when adequately controlling for baseline characteristics.


References

1.
Rhoads GG, Kosiborod M, Nesto RW, Fonseca VA, Lu S, Zhang Q, et al. MD Comparison of Incidence of Acute Myocardial Infarction in Patients With Type 2 Diabetes Mellitus Following Initiation of Neutral Protamine Hagedorn Insulin Versus Insulin Glargine. Am J Cardiol. 2009;104:910-6.