Clinical prediction rule to identify high-risk inpatients for adverse drug events: the JADE Study

Authors

  • Mio Sakuma,

    1. Center for General Internal Medicine and Emergency Care, Kinki University School of Medicine, Osaka-sayama, Japan
    Search for more papers by this author
  • David W. Bates,

    1. Division of General Internal Medicine and Primary Care, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA
    2. Department of Health Policy and Management, Harvard School of Public Health, Boston, MA, USA
    Search for more papers by this author
  • Takeshi Morimoto

    Corresponding author
    • Center for General Internal Medicine and Emergency Care, Kinki University School of Medicine, Osaka-sayama, Japan
    Search for more papers by this author

  • Presented in part at the 28th International Conference of the International Society for Quality in Health Care, Hong Kong, China, 16 September 2011.

T. Morimoto, Center for General Internal Medicine and Emergency Care, Kinki University School of Medicine, 377-2 Ohno-higashi, Osaka-sayama, Osaka 589-8511, Japan. E-mail: morimoto@kuhp.kyoto-u.ac.jp

ABSTRACT

Purpose

Adverse drug events (ADEs) are common health problems worldwide. Developing a prediction rule to identify patients at high risk for ADEs to prevent or ameliorate ADEs could be one attractive strategy.

Methods

The Japan Adverse Drug Events (JADE) study is a prospective cohort study including 3459 participants. We randomly divided the JADE study cohort into the derivation and the validation sets, using an automated random digit generator. We calculated the probabilities of ADE in each patient in the validation set after applying the prediction rule developed in the derivation set. The actual incidence and area under the receiver operating characteristic curve (AUC) in the validation set were compared with those in the derivation set to evaluate the prognostic ability of our developed prediction rule.

Results

The developed prediction rule included eight independent risk factors. Each patient in the validation set was classified into three categories of risk for the ADEs according to the probability of ADEs calculated by the developed prediction rule. Eight percent (137/1730) of patients in the validation set fell into the high-risk group, and 35% of this group (48/137) had at least one ADE. The AUC in the validation set was 0.63 (95%CI 0.60–0.66), and the performance to discriminate the probability of ADE was similar (p = 0.08) compared with that in the derivation set.

Conclusions

This prediction rule had the modest predictive ability and could help physicians and other healthcare professionals to make an estimation of patients at high risk for ADEs. Copyright © 2012 John Wiley & Sons, Ltd.

Ancillary