The Economic Burden of Patients With Type 2 Diabetes Mellitus Hospitalized for Heart Failure

Study conducted by: Stafkey-Mailey D, Fuldeore RM, Shetty S, et al.

Background

  • Incidence of heart failure (HF) among patients with type 2 diabetes mellitus (T2DM) has been estimated to be 2 to 30 cases per 1,000 person‑years.1‑4
  • It is important to characterize the burden of HF hospitalization, in order to understand the potential value of new oral anti‑diabetic therapies.

Study Objectives

  • Among patients with T2DM, determine the direct cost of an HF‑related hospitalization.
  • Compare healthcare resource utilization and costs of T2DM patients with an HF‑related hospitalization to T2DM patients who did not have a HF‑related hospitalization.

Methods

Data Source and Study Design

  • IMS LifeLink PharMetrics Plus Claims Database
  • Retrospective, matched cohort study design
HF identification period: All patients with HF hospitalization were matched (1:1) to patients without an HF hospitalization

Retrospective, Matched Cohort Study Design

Sample Population Selection

  • Target Population: T2DM patients with an HF hospitalization (HF cohort) during the identification period, were matched to T2DM patients without an HF hospitalization (non‑HF cohort).
  • Inclusion
    • T2DM diagnosis during the pre‑index period
    • Aged ≥18 years
    • Continuous plan enrollment during the pre‑index period
  • Exclusion
    • Had an HF hospitalization in the pre‑index period
    • Missing demographic information (age, gender, geographic region)
    • Diagnosis of pregnancy, gestational diabetes, secondary diabetes, or type 1 diabetes mellitus any time during the pre‑index or post‑index period
  • Stratification
    • Patients were stratified into low‑risk and high‑risk groups, based upon the presence of a CV‑related hospitalization in the pre‑index period.
  • Matching
    • HF and non‑HF patients were matched (1:1 ratio) separately in the high- and low‑risk cohorts using the estimated propensity score (PS).
      • Propensity scores were estimated using a logistic regression model, including: age (continuous); gender; geographic region; plan type; index year; presence of CV conditions or CV hospitalization during the pre‑index period; cancer; chronic obstructive pulmonary disease; and CV risk factors (hypertension, dyslipidemia, obesity).

Definitions

  • T2DM status: identified during the pre‑index period, and defined as having:
    • ≥1 hospitalization with a diagnosis of T2DM in any diagnosis field OR
    • ≥2 medical claims at least 30 days apart within 12 months in the outpatient setting, except lab and radiology, with a diagnosis of T2DM in any diagnosis field OR
    • ≥1 pharmacy claim indicated for T2DM, including a) a non‑insulin injectable, or b) an oral anti‑diabetic agent, except metformin, or c) metformin pharmacy claim without a diagnosis code for pre‑diabetes or polycystic ovary syndrome5
  • HF hospitalization: hospitalization with primary discharge diagnosis for HF (ICD‑9‑CM: 428.xx, 402.x1, 404.x1, 404.x3)
  • CV‑related hospitalization: hospitalization with a primary discharge diagnosis for an AMI (ICD‑9‑CM: 410.xx), unstable angina (ICD‑9‑CM: 411.1x), stroke (ICD‑9‑CM: 430.xx‑436.xx), arrhythmias (ICD‑9‑CM:427.xx, except 427.5) or cardiac arrest (ICD‑9‑CM: 427.5), or a hospitalization with a revascularization procedure with other primary discharge diagnosis

Study Outcomes

  • All patients were included for analysis of outcomes from the index HF hospitalization, but only those surviving the index HF hospitalization were analyzed for outcomes during the variable post‑index period.
  • HF hospital characteristics (HF cohort only)
    • Length of stay
    • 30‑day all‑cause readmission
    • Percentage of patients discharged as deceased
  • Healthcare resource utilization
    • Length of stay per hospitalization
    • Bed‑days during the post‑index period among patients with ≥1 hospitalization
    • Number of all‑cause visits by setting of care (reported per patient per month)
  • Healthcare costs in 2015 United States dollars (USD)
    • All‑cause costs (reported per patient per month)

Statistical Analysis

  • Baseline characteristics: paired t‑tests and McNemar’s test to account for matching.
  • Success of matching: a standardized difference of <10% was indicative of acceptable balance.
  • Healthcare resource use and costs: paired t‑tests and McNemar's test to account for matching.
  • All analyses were conducted separately for low- and high‑risk patients, using SAS® version 9.2 (SAS Institute; Cary, NC, USA).
Additional information can be found by downloading the full poster.

References: 1. Chen YT, Vaccarino V, Williams CS, Butler J, Berkman LF, Krumholz HM. Risk factors for heart failure in the elderly: a prospective community‑based study. Am J Med. 1999;106(6):605‑612. 2. Iribarren C, Karter AJ, Go AS, et al. Glycemic control and heart failure among adult patients with diabetes. Circulation. 2001;103(22):2668‑2673. 3. Kannel WB, Hjortland M, Castelli WP. Role of diabetes in congestive heart failure: the Framingham study. Am J Cardiol. 1974;34(1):29‑34. 4. Lloyd‑Jones DM, Larson MG, Leip EP, et al. Lifetime risk for developing congestive heart failure: the Framingham Heart Study. Circulation. 2002;106(24):3068‑3072. 5. Solberg LI, Engebretson KI, Sperl‑Hillen JM, et al. Are claims data accurate enough to identify patients for performance measures or quality improvement? The case of diabetes, heart disease, and depression. Am J Med Qual. 2006;21(4):238‑245.

Sample Characteristics

  • PS matching resulted in a final sample of 10,256 patients (HF=5,128; non‑HF=5,128) in the low‑risk cohort and 602 patients (HF=301; non‑HF=301) in the high‑risk cohort.
  • After matching, the post‑match standardized differences were <10% for all baseline characteristics, with the exception of Charlson Comorbidity Index for the low‑risk cohort, whereas only age, gender, and angina were <10% in the high‑risk cohort.

Study Outcomes

  • The average number of visits was higher among the HF cohort compared to the non‑HF cohort for all visit types.
Number of All‑cause Healthcare Visits, per Patient per Month by Setting of Care During the Post‑index Period

Number of All‑cause Healthcare Visits, per Patient per Month by Setting of Care During the Post‑index Period*

*All comparisons between cohorts are significant at P<0.001
Other outpatient visits may include labs, diagnostics, home health, etc.
HF - heart failure.
  • In addition to index hospitalization, post‑index mean all‑cause costs were 3.4 and 4.9 times higher in the HF cohort compared to the non‑HF among low- and high‑risk patients, respectively.
Healthcare Costs During Index Hospitalizations and per Patient per Month During the Post‑index Period

Healthcare Costs During Index Hospitalizations and per Patient per Month During the Post‑index Period*

*All comparisons were statistically significant at P<0.001
CV - cardiovascular; HF - heart failure; T2DM - type 2 diabetes mellitus; USD - United States dollars.
  • Inpatient costs accounted for the largest difference, and were 5.4 and 7.9 times higher in the HF cohort compared to the non‑HF cohort among low- and high‑risk patients, respectively.
Number of All-cause Healthcare Visits, per Patient per Month by Setting of Care During the Post-index Period

All‑cause Healthcare Costs, per Patient per Month by Settings of Care During the Post‑index Period*

*All comparisons were statistically significant at P<0.001
Other outpatient visits may include labs, diagnostics, home health, etc.
HF - heart failure; United States dollars.
Low Risk: HF (n=5,128); Non‑HF (n=5,128)
High Risk: HF (n=301); Non‑HF (n=301)
Please see the poster for demographic data.

Conclusion

  • A HF-related hospitalization among patients with T2DM costs more than $20,000, ongoing cost post‑discharge ranges from $4,000 to $8,000, per patient per month.
Additional information can be found by downloading the full poster.

Limitations

  • Results may be confounded by missing some important HF risk factors, such as smoking status and obesity.
  • Complete cost data were not available for all patients; patients were censored in the event of death, disenrollment, or the end of available data.
  • Results of the analysis are primarily generalizable to a commercially insured population.
  • The standardized difference was >10% for majority of covariates in the high‑risk cohort after matching, thus results may be confounded.
  • Only 13.7% of patients in the high‑risk cohort experiencing an HF hospitalization were matched; thus, the results may not be representative of all high‑risk patients experiencing an HF hospitalization.
Additional information can be found by downloading the full poster.