Currently Viewing:
The American Journal of Accountable Care December 2017
Enhancing Dementia Care Through Digital Health
Shannen Kim, BA; Omid B. Toloui, MPH, MBA; and Sachin Jain, MD, MBA, FACP
Transitioning Community Hospitals to Value-Based Care: Lessons From Massachusetts
Christopher J. Louis, PhD, MHA, Sara S. Bachman, PhD, MS; Dylan H. Roby, PhD; Lauren Melby, MBA, MPP; and David L. Rosenbloom, PhD
Factors Associated With Timeliness in Academic General Surgery Clinics: A Prospective Quality Assessment
Katelyn A. Young, BS; D. Priyantha Devapriya, PhD; James T. Dove, BA; Marcus Fluck, BS; Kristy A. Yohey, MHS; Marie A. Hunsinger, RN, BSHS; John E. Widger, MD; Joseph A. Blansfield, MD; and Mohsen M. Shabahang
When Goals Are Not Met in Diabetes Care
Currently Reading
Implications of DRG Classification in a Bundled Payment Initiative for COPD
Trisha M. Parekh, DO; Surya P. Bhatt, MD; Andrew O. Westfall, MS; James M. Wells, MD; deNay Kirkpatrick, DNP, APN-BC; Anand S. Iyer, MD; Michael Mugavero, MD; James H. Willig, MD; and Mark T. Dransfield, MD
Tools to Gauge Progress During Patient-Centered Medical Home Transformation
Denise D. Quigley, PhD; Zachary S. Predmore, AB; and Ron D. Hays, PhD

Implications of DRG Classification in a Bundled Payment Initiative for COPD

Trisha M. Parekh, DO; Surya P. Bhatt, MD; Andrew O. Westfall, MS; James M. Wells, MD; deNay Kirkpatrick, DNP, APN-BC; Anand S. Iyer, MD; Michael Mugavero, MD; James H. Willig, MD; and Mark T. Dransfield, MD
Diagnosis-related group coding determines eligibility for many Medicare bundled payment initiatives. This approach excluded many patients with chronic obstructive pulmonary disease likely to benefit while including others without the disease.

Objectives: Institutions participating in the Medicare Bundled Payments for Care Improvement (BPCI) initiative invest significantly in efforts to reduce readmissions and costs for patients who are included in the program. Eligibility for the BPCI initiative is determined by diagnosis-related group (DRG) classification. The implications of this methodology for chronic diseases are not known. We hypothesized that patients included in a BPCI initiative for chronic obstructive pulmonary disease (COPD) would have less severe illness and decreased hospital utilization compared with those excluded from the bundled payment initiative. 

Study Design: Retrospective observational study.

Methods: We sought to determine the clinical characteristics and outcomes of Medicare patients admitted to the University of Alabama at Birmingham Hospital with acute exacerbations of COPD between 2012 and 2014 who were included and excluded in a BPCI initiative. Patients were included in the analysis if they were discharged with a COPD DRG or with a non-COPD DRG but with an International Classification of Diseases, Ninth Revision code for COPD exacerbation.

Results: Six hundred and ninety-eight unique patients were discharged for an acute exacerbation of COPD; 239 (34.2%) were not classified into a COPD DRG and thus were excluded from the BPCI initiative. These patients were more likely to have intensive care unit (ICU) admissions (63.2% vs 4.4%, respectively; P <.001) and require noninvasive (46.9% vs 6.5%; P <.001) and invasive mechanical ventilation (41.4% vs 0.7%; P <.001) during their hospitalization than those in the initiative. They also had a longer ICU length of stay (5.2 vs 1.8 days; P = .011), longer hospital length of stay (10.3 days vs 3.9 days; P <.001), higher in-hospital mortality (14.6% vs 0.7%; P <.001), and greater hospitalization costs (median = $13,677 [interquartile range = $7489-$23,054] vs $4281 [$2718-$6537]; P <.001).

Conclusions: The use of DRGs to identify patients with COPD for inclusion in the BPCI initiative led to the exclusion of more than one-third of patients with acute exacerbations who had more severe illness and worse outcomes and who may benefit most from the additional interventions provided by
the initiative. 

The American Journal of Accountable Care. 2017;5(4):e1-e7
Rising costs have led to a number of federal initiatives to reform the US healthcare payment system and reimburse providers and hospitals based on outcomes rather than volume. As part of a multipronged strategy to control costs and address quality concerns, CMS hopes to tie more than 50% of payments to alternative value-based models by the end of 2018.1 One such program is the Bundled Payments for Care Improvement (BPCI) initiative, which reimburses hospitals and providers based on episodes of care over time rather than individual fee-for-service (FFS) billing. 

The BPCI initiative includes 4 distinct models. In Model 2, reimbursements cover the cost of an index admission, professional fees, and all Medicare part A and B costs, including postacute care and all-cause readmissions within 30, 60, or 90 days of an index hospitalization discharge.2 Reimbursements for an episode of care within this model are based on an inpatient classification system developed in the 1980s that divides diagnoses into categories, known as diagnosis-related groups (DRGs), to determine payment. The DRG assignment is given by the hospital coders with the use of a semi-automated “grouper” computer coding system. Hospital coders input specific information (eg, primary and secondary International Classification of Diseases, Ninth Edition [ICD-9] codes, complications, procedures, sex, gender) into the grouper, which connects clinical attributes to resource utilization in order to assign a specific DRG.3 At the University of Alabama at Birmingham (UAB), our medical coders use the grouper system Optum CAC (“computer assisted coding”). The planned reimbursement, or “target price,” for a given DRG in each BPCI initiative is based on historical data and adjusted for case mix and region. Hospitals may either owe Medicare for overages or gain shared savings with providers depending on whether total BPCI charges exceed or fall below the target price. 

As a participant in Model 2 of the BPCI initiative for patients hospitalized with acute exacerbations (AEs) of chronic obstructive pulmonary disease (COPD), our institution has invested significant financial resources into the development of a multidisciplinary program to deliver specialized interventions to patients included in the BPCI initiative, including expedited follow-up visits in a COPD-focused clinic, home calls, medication assistance, and tobacco cessation counseling.4 Knowledge of the characteristics and outcomes of patients who were ultimately included and excluded from the BPCI can guide the maturation of programs similar to ours. 

Our study sought to: 1) determine the clinical characteristics of patients admitted with AEs of COPD who were included and excluded from the BPCI (based on DRG coding) and 2) evaluate differences in outcomes, hospital length of stay (LOS), and cost utilization between these groups. Based on our experiences participating in the BPCI, we hypothesized that patients excluded from the BPCI initiative would have a higher rate of intensive care unit (ICU) admission and use of mechanical ventilation, longer hospital LOS, and increased index admission costs than those who were included in the initiative. 



Establishment of the Study Cohort

We included Medicare beneficiaries who were admitted to UAB Hospital between January 1, 2012, and December 31, 2014, for an AE of COPD as defined by administrative data.2 As the BPCI initiative currently stands, only Medicare FFS patients are included. However, we included all Medicare patients (FFS, managed Medicare, dual-eligible) in our analysis to increase the power of the study and because payment to private insurers may model the bundled payment approach in the future. Patients were included if they received a COPD DRG (190-192) upon discharge or an ICD-9 code that had traditionally been used to identify an AE of COPD (primary code 491.21 or 491.22; or primary code 518.81, 518.82, 518.84 with 491.21, 491.22, or 496 as secondary).5,6 In order to focus on COPD-specific DRGs, we excluded asthma DRGs (202-203) and therefore did not include ICD-9 codes related to asthma, nonspecific lung disease, or unspecified bronchitis.7 

We identified 990 patient encounters with a discharge diagnosis of AE of COPD based on COPD DRG assignment or ICD-9 coding. Of these, 698 unique patients in 2 mutually exclusive groups were included for analysis: 1) those discharged with a COPD DRG (DRG group) and 2) those discharged with a COPD ICD-9 code and a non-COPD DRG (ICD-9 group). UAB’s Institutional Review Board approved the study protocol (X121221005). 


Data were extracted from our hospital’s clinical data warehouse (Cerner PowerInsight; Cerner Corporation World Headquarters; North Kansas City, Missouri). Demographic information and comorbidities were obtained from the time of index hospitalization. Encounter information obtained included clinical data from hospitalization (vital signs, arterial blood gas values, body mass index [BMI], smoking status, use of noninvasive positive pressure ventilation [NIPPV], invasive mechanical ventilation) and administrative data (hospital and ICU LOS, hospital disposition). All-cause readmissions were evaluated at 30 and 90 days from index hospital discharge; the latter was the length of a BPCI episode for our institution. Costs were obtained from the UAB Health Services Foundation.

Study Outcomes

The co-primary outcomes were the need for ICU admission and the use of mechanical ventilation in patients who were included and excluded from the BPCI initiative based on DRG classification. Secondary outcomes included index hospital LOS, ICU LOS, in-hospital mortality, readmission rate, and costs. 

Statistical Analysis

We used descriptive statistics to compare the demographic and clinical characteristics between the DRG and ICD-9 groups. Each categorical and continuous variable was compared by χ2 and independent sample t tests, respectively. Wilcoxon-rank sum test was used to compare costs to account for skewness. Aggregate costs of all-cause 30-day and 90-day readmissions per patient were calculated by summing the costs each patient incurred within 30 and 90 days from index hospitalization discharge. In similar secondary analyses, we compared the DRG group with the ICD-9 group, excluding patients with a primary ICD-9 code of respiratory failure (518.81, 518.82, 518.84) from the latter, as these patients would be more likely to have a higher severity of illness. Finally, we separated the DRG cohort into patients who had an AE of COPD ICD-9 code and those who did not (dual coded vs DRG only) and compared these 2 groups. This analysis was performed to provide information on the subgroup of patients who may not truly have COPD, yet received a COPD DRG and were therefore part of the BPCI initiative. All hypothesis testing was 2-sided with significance set at P <.05. All analyses were performed using SPSS statistical software (version 22).


DRG Classification and Baseline Characteristics

Of the 698 unique patients included for analysis, 459 were discharged with a COPD DRG (DRG group) and 239 were discharged with a non-COPD DRG (ICD-9 group) (Table 1). Patients in the ICD-9 group more often were male, were white, and had a higher BMI (P <.05 for all comparisons) compared with patients in the DRG group. ICD-9 patients also had a lower rate of depression, osteoporosis, and coronary artery disease. The most common DRG classifications for the ICD-9 group were pulmonary edema and respiratory failure (DRG 189; n = 91) and respiratory failure with ventilator support less than 96 hours (DRG 208; n = 71). 

Physiology and Severity of Illness

As shown in Table 2, oxygen saturation and pH were lower and partial pressure of carbon dioxide and respiratory rate were higher in the ICD-9–only group compared with the DRG group (P <.05 for all), suggesting greater disease severity in the former. Likewise, patients in the ICD-9 group had higher rates of NIPPV (46.9% vs 6.5%; P <.001) and invasive mechanical ventilation (41.4% vs 0.7%; P <.001) during their index hospitalization. Patients in the ICD-9 group who did not have a code for respiratory failure and had a primary ICD-9 code of 491.21 or 491.22 (n = 34) also exhibited a higher rate of use of NIPPV (35.3% vs 6.5%) and invasive mechanical ventilation (32.4% vs 0.7%) compared with those in the DRG group (P <.001 for both). When we restricted our analysis to fee-for-service Medicare patients, we found similar differences in the use of NIPPV and mechanical ventilation between the DRG (n = 276) and ICD-9 (n = 166) groups. 

Hospital LOS and Discharge Disposition 

Patients in the ICD-9 group had a higher rate of ICU admission (63.2% vs 4.4%; P <.001), longer ICU LOS (mean = 5.2 [SD = 5.8] days vs 1.8 [SD = 1.1] days; P = .011), and longer hospital LOS (10.3 [SD = 15.6] days vs 3.9 [SD = 2.8] days; P <.001) compared with the DRG group (Table 2). Thirty-five (14.6%) patients in the ICD-9 group died during their index hospitalization compared with 3 patients (0.7%) in the DRG group (P <.001). ICD-9 patients were less likely to be discharged home than those in the DRG group (37.7% vs 69.1%; P <.001). Those in the subcohort of ICD-9 patients without respiratory failure also had a longer hospital LOS (8.9 [SD = 6.4] days vs 3.9 [SD = 2.8] days; P <.001), higher rate of ICU admission (52.9% vs 4.4%; P <.001), and longer ICU LOS (4.0 [SD = 3.6] days vs 1.8 [SD = 1.1] days;

P = .008) compared with the DRG group. ICD-9 Medicare FFS-only patients also had higher rates of ICU admission and in-hospital mortality and longer hospital LOS. 

Readmission Patterns and Cost Utilization 

There were no statistically significant differences in 30- or 90-day all-cause readmission rates between the 2 groups (Table 3). The ICD-9 group had a higher total median cost of index hospitalization than the DRG group (median = $13,677 [interquartile range = $7489-$23,054] vs $4281 [$2718-$6537]; P <.001). Total costs of index admission in the ICD-9 group, excluding respiratory failure codes, were also significantly higher than costs in the DRG group ($15,793 [$10,890-$23,590]; P <.001). Aggregate costs per patient incurred in the 30 and 90 days after index hospitalization discharge were higher in the ICD-9 group compared with the DRG group (mean = $3122 [SD = $12,564] vs $1667 [SD = $5872] in 30 days; $5376 [SD = $14,882] vs $4116 [SD = $10,493] in 90 days) (Table 3). 

Of the 459 patients who were included in the COPD DRG group, 115 did not have a COPD ICD-9 code. Their clinical characteristics, severity of illness, and readmissions were similar to others included in the COPD DRG. Patients in the DRG-only group, however, did have a longer hospital LOS (mean = 4.5 [SD = 3.2] days vs 3.7 [SD = 2.8] days; P = .013) and a costlier index admission ($5172 [$3220-$7129] vs $4013 [$2651-$6204]; P = .004) (Tables 4 and 5). 


Copyright AJMC 2006-2020 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
Welcome the the new and improved, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up