A quantitative method for assessment of prescribing patterns using electronic health records

PLoS One. 2013 Oct 10;8(10):e75214. doi: 10.1371/journal.pone.0075214. eCollection 2013.


Background: Most available quality indicators for hospitals are represented by simple ratios or proportions, and are limited to specific events. A generalized method that can be applied to diverse clinical events has not been developed. The aim of this study was to develop a simple method of evaluating physicians' prescription patterns for diverse events and their level of awareness of clinical practice guidelines.

Methods and findings: We developed a quantitative method called Prescription pattern Around Clinical Event (PACE), which is applicable to electronic health records (EHRs). Three discrete prescription patterns (intervention, maintenance, and discontinuation) were determined based on the prescription change index (PCI), which was calculated by means of the increase or decrease in the prescription rate after a clinical event. Hyperkalemia and Clostridium difficile-associated diarrhea (CDAD) were used as example cases. We calculated the PCIs of 10 drugs related to hyperkalemia, categorized them into prescription patterns, and then compared the resulting prescription patterns with the known standards for hyperkalemia treatment. The hyperkalemia knowledge of physicians was estimated using a questionnaire and compared to the prescription pattern. Prescriptions for CDAD were also determined and compared to clinical knowledge. Clinical data of 1698, 348, and 1288 patients were collected from EHR data. The physicians prescribing behaviors for hyperkalemia and CDAD were concordant with the standard knowledge. Prescription patterns were well correlated with individual physicians' knowledge of hyperkalemia (κ = 0.714). Prescribing behaviors according to event severity or clinical condition were plotted as a simple summary graph.

Conclusion: The algorithm successfully assessed the prescribing patterns from the EHR data. The prescription patterns were well correlated with physicians' knowledge. We expect that this algorithm will enable quantification of prescribers' adherence to clinical guidelines and be used to facilitate improved prescribing practices.

Publication types

  • Research Support, Non-U.S. Gov't

MeSH terms

  • Adult
  • Aged
  • Algorithms
  • Electronic Health Records*
  • Female
  • Humans
  • Male
  • Middle Aged
  • Practice Patterns, Physicians'*
  • Young Adult

Grant support

This work was supported by a National Research Foundation of Korea grant funded by the Korean government (MSIP) (2010-0028631 to R.W.P.). The study was also supported by grants of the Korea Health Technology R&D Project, Ministry of Health and Welfare, Republic of Korea (A112022 to R.W.P.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.