Do Non-Clinical Factors Improve Prediction of Readmission Risk?
Results From the Tele-HF Study

Krumholz HM, Chaudhry SI, Spertus JA, et al.
J Am Coll Cardiol HF 2016;4:12–20
 

Background

The risk of readmissions after hospitalisations due to heart failure is hard to predict, since most known risk models were shown to have low discrimination and predictive ability [1-3]. This can be possibly explained by the lack of socioeconomic, health status, adherence, and psychosocial characteristics in these risk models.
The objective of this study is to evaluate whether the incorporation of socioeconomic,
health status, and psychosocial characteristics into a heart failure readmission risk model can improve its predictive ability.
In the context of the Telemonitoring to Improve Heart Failure Outcomes (Tele-HF) trial [4,5], data from medical records were collected and telephone interviews were conducted with 1,004 heart failure patients who were recently discharged from the hospital. A total of 110 variables were divided into 2 groups:

• demographic and clinical variables
• socioeconomic, health status, adherence, and psychosocial variables.

The readmission rate was calculated within a 180-day follow-up period.
 

Main results

• The median time from discharge to the interview was 12 days (interquartile range: 6 to 19 days).
• The 30-day readmission rate was 17.1%
• Using the 3-level risk score derived from the restricted medical record variables, patients with a score of 0 (no risk factors) had a readmission rate of 10.9% (95% CI: 8.2% -14.2%); patients with a score of 2 (all risk factors) had a readmission rate of 32.1% (95% CI: 22.4% - 43.2%); with a C-statistic of 0.62.
• Using the 5-level risk score derived from all variables, patients with a score of 0 (no risk factors) had a readmission rate of 9.6% (95% CI: 6.1% - 14.2%); patients with a score of 4 (all risk factors) had a readmission rate of 55.0% (95% CI: 31.5% - 76.9%); with a C-statistic of 0.65.  

Conclusion

Patient-reported information on socioeconomic, health status, adherence, and psychosocial variables improved model discrimination and extended the predicted ranges of readmission rates, but the model performance remained poor. There is a need to better understand which other factors may influence substantially readmission risk, like for example health system quality of care, hospitalisation stress, and propensity to admit.
 

Editorial comment [6]

Although the enrichment of a risk score by 83 additional variables did not improve its predictive accuracy for hospital readmissions of health failure patients, according to Konstam and Upshaw ‘’the efforts of Krumholz et al. are to be applauded for seeking information directly from patients and reaching beyond traditional medical factors.’’ Reasons explaining the results of the original investigation may include other factors that are not measured like health system quality and a hospital’s propensity for admission. Other important limitations include a) the time between discharge and interview, since patients answers to the same questions may vary substantially from day to day during the first month from hospital discharge, and b) the exclusion of patients who died or were hospitalized before the interview.
Current knowledge regarding prognosis and treatment of heart failure can be used effectively to predict and prevent post-discharge heart failure related morbidity and mortality, however re-hospitalisations of these patients for all other causes is very difficult to predict. Despite that, the centers for Medicare and Medicaid Services (CMMS) are penalising hospitals on the basis of the number of all-cause hospital readmissions within 30 days following an index hospitalization for heart failure, acute myocardial infarction, and pneumonia [7]. In this context ‘’Metrics that are flawed in driving patient benefit may also be difficult to predict, rendering their assessment even more flawed. They seem to create a perfect storm for suboptimal care, both by sidestepping the best interest of the patient and by thwarting assessment of risk for both clinicians, in their care, and for CMMS in its attempt at “fair” adjudication and penalty assignment. In the larger picture, we need to move entirely away from artificial metrics and penalties and toward greater direct responsibility of health care systems for quality and efficiency, with rewards linked to long-term patient benefit, through innovative approaches to care. Otherwise, those who have wished to devote their lives to healing will instead forever be endlessly rolling a boulder toward a remote and unreachable precipice.‘’
 
Find this article online
 

References

1. Kansagara D, Englander H, Salanitro A, et al. Risk prediction models for hospital readmission: a systematic review. JAMA 2011;306:1688–98.
2. Keenan PS, Normand SL, Lin Z, et al. An administrative claims measure suitable for profiling hospital performance on the basis of 30-day all-cause readmission rates among patients with heart failure. Circ Cardiovasc Qual Outcomes 2008;1:29–37.
3. Ross JS, Mulvey GK, Stauffer B, et al. Statistical models and patient predictors of readmission for heart failure: a systematic review. Arch Intern Med 2008;168:1371–86.
4. Chaudhry SI, Barton B, Mattera J, et al. Randomized trial of Telemonitoring to Improve Heart Failure Outcomes (Tele-HF): study design. J Card Fail 2007;13:709–14.
5. Chaudhry SI, Mattera JA, Curtis JP, et al. Telemonitoring in patients with heart failure. N Engl J Med 2010;363:2301–9.
6. Konstam MA, Upshaw J. Sisyphus and 30-Day Heart Failure Readmissions. Futility in Predicting a Flawed Outcome Metric. J Am Coll Cardiol HF 2016;4:21-23.
7. Rau J. Medicare to penalize 2,217 hospitals for excess readmissions. Kaiser Health News August 13, 2012. Available at: http://khn.org/news/medicare-hospitals-readmissions-penalties.Accessed November 6, 2015.