Gaming hospital-level pneumonia 30-day mortality and readmission measures by legitimate changes to diagnostic coding
Michael W. Sjoding1, MD, Theodore J. Iwashyna, MD, PhD1-3, Justin B. Dimick, MD, MPH4,5, Colin R.
Cooke, MD, MSc, MS1,5 Affiliations:
1The Division of Pulmonary & Critical Care Medicine, Department of Internal Medicine, University of Michigan, Ann Arbor, MI
2VA Center for Clinical Management Research, Ann Arbor, MI
3Institute for Social Research, Ann Arbor, MI
4Department of Surgery, University of Michigan, Ann Arbor, MI
5Center for Healthcare Outcomes & Policy, Institute for Healthcare Innovation & Policy, University of Michigan, Ann Arbor, MI
Corresponding Author: Michael Sjoding, MD, University of Michigan, 3916 Taubman Center, 1500 E.
Medical Center Dr., SPC 5360, Ann Arbor, MI 48109-5360. Phone: 734-763-9077. Fax: 734-936-5048.
Email: [email protected]
Supplemental digital content Cohort generation details
eFigure 1 - Mortality and readmission cohort generation eTable 1 – ICD-9-CM codes for acute organ dysfunction
eTable 2 - Improved pneumonia 30-day mortality and hospital readmission rates when 500 hospitals above the 50th percentile recoded patients to a primary diagnosis of sepsis or respiratory failure
eFigure 3 - Ability of hospitals to improve mortality rates to below the 50th percentile when varying percentages of eligible patients are recoded and varying numbers of other hospitals are also recoding patients
Cohort Generation Details
Following CMS methods for calculating mortality and readmission rates, we limited analysis to hospitals with more than 25 pneumonia admissions in 2009. In the mortality cohort, we excluded patients with a length of stay of less than two days or greater than one year, those who left against medical advice or were enrolled in hospice prior to the index admission. We randomly selected one admission among patients with multiple pneumonia admissions to ensure independence of observations. Hospitalizations involving transfers of care between two acute-care hospitals were linked as a single episode of care if the principal diagnosis for both hospitalizations was pneumonia, and mortality was attributed to the transferring hospital. Thirty- day mortality was defined as death due to any cause within 30 days of admission.
In the readmission cohort, we limited analysis to patients surviving the initial hospitalization and excluded patients who left against medical advice. Readmission was defined as any admission to an acute care hospital within 30-days of discharge from an index hospitalization. Because certain readmissions within 30 days of hospital discharge are planned, we used the clinical classification software developed by AHRQ to identify planned
readmissions and excluded these from analysis. Readmissions among patients transferred to another acute-care hospital were attributed to the hospital that discharged the patient to a non- acute care setting. A single patient could have multiple index hospitalizations included as long as each index hospitalization was at least 30 days from the previous index hospitalization.
eFigure 1 A. Mortality cohort derivation B. Readmission cohort derivation
eTable 1 – ICD-9-CM codes for acute organ dysfunction
Cardiovascular 785.5X, 458.X Respiratory 96.7X
Renal 584.X
Liver 570, 573.4
Hematologic 287.4X, 287.5X, 286.9X, 286.6X Neurologic 248.3X, 293.X, 348.1X
eTable 2 Improved pneumonia 30-day mortality and hospital readmission rates when 500 hospitals above the 50th percentile recoded patients to a primary diagnosis of sepsis or respiratory failure
When 500 hospitals with mortality rates above the 50th percentile recoded patients
After half of eligible patients recoded
After all eligible patients recoded Average percent decrease in mortality rate
among the hospitals (95% CI) 0.74 (0.68-0.82) 1.35 (1.45-1.53) Number who improved per 100 hospitals
(95% CI) 82 (76-85) 90 (87-93)
Number who dropped below the 50th
percentile per 100 hospitals (95% CI) 18 (15-20) 35 (33-38) When 500 hospitals with readmission rates
above the 50th percentile recoded patients
After recoding half of eligible patients
After recoding all eligible patients Average percent decrease in readmission
rate among the hospitals (95% CI) 0.26 (0.2-0.32) 0.54 (0.47-0.61) Number who improved per 100 hospitals
(95% CI) 60 (55-65) 67 (63-71)
Number who dropped below the 50th
percentile per 100 hospitals (95% CI) 10 (7-11) 14 (11-16) Results presented as means and 95% confidence interval estimates from Monte-Carlo simulations unless otherwise stated. Confidence intervals are percentiles of the simulated results. Results are per 100 hospitals for easier comparison with table 2. In each simulation: (1) 100 hospitals are selected to recode patients (2) among selected hospitals, 50% or 100% of patients with pneumonia and organ failure are dropped (recoded) and the mortality or readmission rate is re-calculated.
Patients eligible for recoding are those with a primary ICD-9-CM code for pneumonia and secondary code for acute organ failure
eFigure 2. Ability of hospitals to improve mortality rates to below the 50th percentile when varying percentages of eligible patients are recoded and varying numbers of other hospitals are also recoding patients
Each set of simulations was performed 25 times, parameters adjusted between each set of simulations included the number of hospitals recoding patients, and the percentage of eligible patients recoded.
Results are presented as the mean probability of a positive result (dropping below the 50th percentile) for each set of simulations.
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45
0 25 50 75 100
Probability of a hospital with mortality above the 50th percentile dropping below after recoding eligible patients
Percentage of eligible patients recoded
20% of hospitals
simultaneously recoding patients (N=585)
50% of hospitals
simultaneously recoding patients (N=1462)
100% of hospitals simultaneously recoding patients (N=2922)