All Patient Refined-Diagnosis Related Group Classification for Hospitalized Patients With COVID-19

Abstract

The pandemic caused by COVID-19 has had an extraordinary impact throughout the United States and the world. The disease presentation ranges from asymptomatic infections to severe acute respiratory failure requiring mechanical ventilation. The mortality rate of patients admitted to ICUs requiring mechanical ventilation reaches 50% in some studies.1Bhatraju P.K. Ghassemieh B.J. Nichols M. et al.Covid-19 in critically ill patients in the Seattle Region - Case Series.N. Engl. J. Med. 2020; 382: 2012-2022Crossref PubMed Scopus (1588) Google Scholar Hospitals and clinicians involved in clinical studies need a uniform system for scoring the severity of illness in these patients to compare outcomes. The severity of illness (SOI) score and the risk of mortality (ROM) score based on All Patient Refined-Diagnosis Related Group (APR-DRG) (3M™ APR DRG Classification System and 3M™ APR DRG Software, Health Information Systems, Salt Lake City, UT) coding provides a uniform method available in most hospitals that could improve the analysis and comparison of outcomes in diverse clinical settings.2Fontaine P. Using severity adjustment classification for hospital internal and external- IFHRO congress & AHIMA convention proceedings. https://library.ahima.org/doc?oid=59268#.Xy2ZQihKiUk2004. Accessed May 5, 2020Google Scholar,33M™ APR DRG classification system and 3M™ APR DRG software, health information systems.https://multimedia.3m.com/mws/media/478415O/3m-apr-drg-fact-sheet.pdf. Accessed May 8, 2020.Google Scholar We collected information, including the APR-DRG SOI and ROM, on the initial group of patients hospitalized in a tertiary care center in West Texas during the first phase of COVID-19 in this region. A list of 64 patients with COVID-19 infections established by PCR tests was obtained from the Infection Control and Prevention Office at University Medical Center in Lubbock, Texas. The timeframe for hospitalization for these patients ranged from March 1 through a May 15 discharge date. Medical records were reviewed to determine demographic characteristics including age and sex, length of stay, and mortality. The discharge SOI and ROM scores were available for 57 patients; 7 patients had an observational status and had no final code. Results were summarized using means and standard deviations, medians and interquartile ranges, and numbers with percentages. Differences between the expected length of stay and the observed length of stay were analyzed using a paired t-test. This study was approved by the Institutional Review Board (L20–172) at Texas Tech University Health Sciences Center in Lubbock, Texas, and by administrative review at University Medical Center in Lubbock, Texas. The entire cohort included 57 patients; basic characteristics of these patients are reported in the Table. The observed length of stay (11.53 ± 11.46 days) was significantly longer than the expected length of stay (6.92 ± 3.90 days, P = 0.014). In addition, almost all deaths occurred in patients with a ROM score of 4 (14 of 20 deaths, 70.0%).TableSeverity of illness and risk of mortality scores in hospitalized COVID- 19 patients.Basic CharacteristicsEntire cohortN = 57Age, years mean ± SD63.45 ± 13.51Sex, male N (%)31 (54.4%)LOS, days mean ± SD11.53 ± 11.46Mortality, N (%)20 (35.1%)SOINumber patientsObserved LOS, days*median.100233, range: 2–103297, range: 6–942514, range: 8–20Median SOI311.53Q136Q3414Expected LOS, daysObserved LOS, daysMean ± SD6.92 ± 3.9011.53 ± 11.460.0137⁎⁎paired t-test.Median5.188Q15.186Q36.1214Minimum3.021Maximum24.5175ROMNumber patientsDeaths per subgroupTotal percent mortality100 (0%)0%270 (0%)0%3246 (25%)10.53%42614 (54%)24.56%Abbreviations: SD, standard deviation; N, number; LOS, length of stay; SOI, severity of illness (1,2,3,4 represent subgroups); ROM, risk of mortality (1,2,3,4 represent subgroups); Q1, first quartile; Q3, third quartile. median. paired t-test. Open table in a new tab Abbreviations: SD, standard deviation; N, number; LOS, length of stay; SOI, severity of illness (1,2,3,4 represent subgroups); ROM, risk of mortality (1,2,3,4 represent subgroups); Q1, first quartile; Q3, third quartile. The severity of illness (SOI, physiological derangement or loss of organ system function) score and the risk of mortality (ROM) score are calculated by a propriety algorithm owned by 3 M Corp. The medical record is reviewed using language processing software which then suggests an ICD-10 code. Each ICD-10 code has its own SOI and ROM, but it depends on the principal diagnosis. These codes are modified by secondary diagnoses and age to calculate a final SOI and ROM. The SOI and ROM each have subclasses, i.e., minor (1), moderate (2), major (3), extreme (4). These numbers are ordinal numbers which provide a medical classification. These 3 M codes can be modified by the hospitals billing and coding department, if necessary after review. The 3 M APR-DRGs have become the standard across the United States for classifying in-patients in non-Medicare populations, and as of January 2019, 27 state Medicare programs and approximately a dozen commercial payers and Medicaid managed care organizations use this system. Our study indicates that the SOI and ROM scores can help classify patients with COVID-19 and provide an estimate of severity of illness and predict the expected length of stay and the mortality risk. Preliminary scores can be determined with the initial documentation on hospital admission by the hospital coding staff; the final scores are based on the entire medial record. They do not require clinician input or a significant collection of vital signs or laboratory tests. This information can provide a method for comparisons of groups undergoing different treatment strategies. For example, patients appearing to improve with convalescent plasma might differ substantially from patients who did not receive plasma, and this could be determined by calculating average SOI and ROM scores. In addition, these scores would provide information about the characteristics of patients admitted to various units in a hospital or patients admitted to different hospitals. These classification scores would be particularly useful in patients admitted to inpatient services but not to intensive care units since these patients ordinarily would not have APACHE scores calculated. McCormick et al. analyzed the APR–DRG risk of mortality and severity of illness modifiers as a measure of perioperative risk in all adult patients undergoing non-cardiac surgery at their hospital between December 2, 200, and July 2, 2013.4McCormick P.J. Lin H.M. Deiner S.G. et al.Validation of the All Patient Refined Diagnosis Related Group (APR-DRG) risk of mortality and severity of Illness modifiers as a measure of perioperative risk.J Med Syst. 2018; 42: 81https://doi.org/10.1007/s10916-018-0936-3Crossref PubMed Scopus (55) Google Scholar They compared Charlson Comorbidity Index with ROM only, SOI only, and ROM plus SOI. This study included 63,681 patients; the overall mortality rate was 1.3%. The c-statistic for ROM was 0.974, for SOI 0.965, for SOI + ROM 0.977, and for the Charlson 0.865; they concluded to the APR–DRG subclassifications were better than the Charlson Comorbidity Index for predicting inpatient postoperative mortality. Baram and colleagues analyzed the APR–DRG risk of mortality score as a severity adjuster in MICU patients.5Baram D. Daroowalla F. Garcia R. et al.Use of the All Patient Refined-Diagnosis Related Group (APR-DRG) Risk of Mortality Score as a severity adjustor in the medical ICU.Clin Med Circ Respirat Pulm Med. 2008; 2: 19-25PubMed Google Scholar This study included 1213 patients admitted between February 2004 and March 2006. The mortality rate correlated significantly with an increase in the ROM score. A model using multiple logistic regression analysis adjusted for age and disease group demonstrated that the ROM was significantly associated with mortality risk in these patients and that a 1-unit increase in ROM was associated with a 3-fold increase in mortality. These two studies indicate that the SOI and ROM can help predict outcomes in patients with the broad spectrum of acuity levels. Our analysis suggests that these scores help predict outcomes in patients with COVID-19 but the observed the length of scale stay is longer than expected in these patients. This result suggests that the SOI may need adjustment for this particular diagnosis.