The Accuracy of Identification of Patients with Pneumonia Through Administrative Data and Impact of Clinical Reclassification on Readmission Rates

1Duke University Medical Center, Durham, NC

Meeting: Hospital Medicine 2014, March 24-27, Las Vegas, Nev.

Abstract number: 134

Background:

In October 2012, the Hospital Readmissions Reduction Program allowed CMS to reduce payments to hospitals with excess 30‐day re‐hospitalization rates for patients discharged after a hospitalization for pneumonia. As such, the accuracy of identification of patients with the clinical diagnosis of true pneumonia during an index admission is critical. We sought to determine if billing data accurately classified pneumonia diagnoses used in calculation of readmission rates, and to examine the effects of data reclassification using clinical and chart adjudication.

Methods:

All Duke University Hospital discharges from 12/1/2012 to 11/21/2013 with a principal diagnosis ICD‐9 code of pneumonia as defined by CMS methodology were reviewed. Once identified, each case was evaluated by 1 of 3 independent MDs to determine if the patient met the clinical diagnosis for pneumonia in accordance with a reference standard (adapted from the CDC algorithm). Using this algorithm, the diagnosis was determined as one of 4 categories: Definite, Probable, Probably Not or Not pneumonia. If a case was determined to be Probably Not or Not pneumonia, it was sent for review by the discharging provider to make a final clinical diagnosis. If the discharging provider agreed with the independent review, the discharge documentation was amended to reflect the diagnosis. Rates for re‐hospitalization for the true pneumonia group (Definite & Probable) were compared with the original population and the subset deemed to not be pneumonia (Probably Not & Not). The Duke Institutional Review Board approved this study.

Results:

From 12/1/2012 to 11/21/2013, 218 pneumonia discharges underwent review with a readmission rate of 18.9%. Independent review found 72% were Definite or Probable; 28% wereProbably Not (20%) or Not Pneumonia (8%). 65 cases were sent for re‐review by the discharging provider and 43 cases (19.7 % of total discharges reviewed) were changed to an alternate diagnosis other than pneumonia. Common alternative diagnosis were aspiration, bronchitis, heart failure, and acute chest syndrome. The 30‐day re‐hospitalization rate for the 175 meeting the clinical reference standard was 18.2% and for the 43 recoded cases was 20.9%. It took on average 7.8 days to receive notification of a pneumonia discharge and 3.4 days for review and amended discharge documentation.

Conclusions:

This study illustrates that identification of pneumonia via claims data has inherent inaccuracies. Re‐classification of discharges to accurately reflect pneumonia diagnosis could alter readmission rates and impact the extent of readmission penalties. Furthermore, accurate identification of this population is important to improve pneumonia care. Using a similar review process, hospitals can ensure the accuracy of patient identification for inclusion in readmission penalties and improve pneumonia care in general.

To cite this abstract:

Ruopp M, Bae J, Holland T, Govert J, Shah B, Stillwagon M, Velazquez M. The Accuracy of Identification of Patients with Pneumonia Through Administrative Data and Impact of Clinical Reclassification on Readmission Rates. Abstract published at Hospital Medicine 2014, March 24-27, Las Vegas, Nev. Abstract 134. Journal of Hospital Medicine. 2014; 9 (suppl 2). https://www.shmabstracts.com/abstract/the-accuracy-of-identification-of-patients-with-pneumonia-through-administrative-data-and-impact-of-clinical-reclassification-on-readmission-rates/. Accessed May 26, 2019.

« Back to Hospital Medicine 2014, March 24-27, Las Vegas, Nev.