Diagnosis Related Group (DRG) in Colon Surgery: Identifying Areas of Improvement to Drive High-Value Care

Abstract

Background:

Diagnosis Related Group (DRG) Migration is defined as the reassignment of colectomy patients from DRG 331 to 330 based exclusively on postoperative complications. Strategic and comparative application of this metric has the potential to demonstrate baseline and excessive rates of complications related directly to patient care differences across institutions. The aim of this study was to report the variability of DRG Migration across U.S. hospitals and its impact on overall cost and length of stay (LOS).

Methods:

This study investigated the variability of DRG Migration rates across U.S. hospitals polling 5% of the national Medicare data. The study endpoints were total cost, LOS, and DRG Migration rate. Hospitals were classified into tertiles for low (0.1-16.6%), moderate (16.7-23.0%) and high (23.1-83.3%) DRG Migration rates.

Results:

The study included 5,120 patients from 615 hospitals. DRG Migration rates for hospitals ranged from 0.1% to 83.3%, with 157 in the low tertile, 183 in moderate, and 364 in high. DRG Migration resulted in a progressively increased LOS and hospital costs from lowest to highest tertile. Several diagnoses were identified which are suggestive of failure to integrate evidence-based-processes of care across the tertiles.

Conclusion:

The data confirms a wide variation in DRG Migration rates from DRG 331 to 330 based only on postoperative complications. These ranges allow for the potential definition of both best practice, as well as opportunities for quality improvement with respect to postoperative complications, identification of hospital outliers, and the economics of care as part of a value based care program.

INTRODUCTION

Bending the cost curve for the U.S. healthcare system is a national priority, as healthcare expenditures are projected to consume an increasing percentage of the Gross Domestic Product (GDP) in the near future. As of 2015, healthcare costs accounted for $3.2 trillion, or 17.8% of the GDP, with an anticipated increase to approximately 20% by 2025.^{1, 2} Of 2015 expenditures, the Centers for Medicare and Medicaid (CMS) spent an estimated $646.2 billion on healthcare. Further, by 2024, approximately four of every ten dollars spent on healthcare in the United States will be spent through CMS.² Thus far, the reliance upon penalties within the CMS value-based purchasing program have not been associated with improved care based upon specific complications that drive hospital payments, possibly because of the lack of correlation between the selected outcomes and specific processes of care.^3–5 In an attempt of fill this gap, we have previously identified the concept of Diagnosis Related Group (DRG) Migration as a mechanism to identify patients receiving more costly care directly as a result of postoperative complications. A significant advantage of this approach is that each of these diagnoses results in significant additional cost to CMS, and conversely each one have a specific set of potentially mitigating processes of care which a hospital could implement.

The purpose of this study is to analyze variations in DRG Migration variation among U.S. hospitals related to the performance of colectomy which has been reported to have a greater proportion of adverse events than many other procedures.⁶ This analysis has the potential to identify a “best practice” rate of these potentially costly complications, while also identifying high cost providers based on complications tied only to postoperative care. This analysis could also be used by an individual hospital to identify opportunities for both process and outcome improvement based on actual practice.

METHODS

2.1. Data Source and Study Design

This retrospective study used 5% of national Medicare data from 2011 to 2015. Medicare claims data involves patients’ sociodemographic, enrollment, and healthcare utilization information. The present study utilized both the Medicare Provider Analysis and Review file, which captures information about the inpatient stay, as well as the Beneficiary Summary File. The study was reviewed and approved by the University’s IRB.

2.2. Study Cohort

The study cohort included patients over 65 years of age undergoing colectomy. We included colectomies with DRG 331 (major abdominal small and large bowel surgery without comorbidity/complication) and 330 (major abdominal small and large bowel surgery with comorbidity/complication). The study included patients who a) underwent a colectomy (n=22,763) and b) had no CCs present on admission (n=1,577) (Figure 1). Those patients with DRG 330 classification based upon diagnoses present upon admission (n=13,803) (i.e., before a colectomy was performed) and those with zero CCs but classified as 330 (n=353) (i.e., coding error) were excluded. In addition, all hospitals with fewer than five patients who underwent colectomies were excluded because DRG Migration rates may not be reliably calculated based on a small sample size (n=3,487). The final cohort included 5,120 patients. Hospitals totaled 615 in this study.

Figure 1.

DRG Cohort Selection.

2.3. DRG Migration Rate

Explloration of colectomy outcomes provided an opportunity to compare and determine the performance of hospitals by (a) the nature of complication rates and (b) the quality of care; both of which may be based on the Medicare Severity-Diagnosis Related Group (MS-DRG) Migration rate.^{7, 8} For colectomy, three MS-DRG codes are available: 329, 330, and 331. We excluded DRG 329 as these patients are assigned by the presence of specific major complications/comorbidities present on admission and this assignment is unrelated to the occurrence of postoperative complications. Therefore, DRG Migration is defined as a change in the “admission” DRG status of DRG 331 (no complications or comorbidities, or CCs) to a final DRG 330 (with CCs) assignment for claim submission (Figure 2). A hospital with a high rate of DRG Migration relative to the norm is more likely underperforming both in patient care and in costs incurred to CMS.⁷

Figure 2. Colectomy DRG Migration Rate in U.S. Hospitals.

Each Tertile is coded. White represents Tertile 1 [0.1-16.6%], Diagonal Stripes represent Tertile 2 [16.7-23.0%], and Grey represents Tertile 3 [23.1-83.3].

DRG Migration rate was calculated for each hospital. The numerator was defined by the sum of patients admitted and discharged as DRG 331 plus patients discharged as DRG 330 only as a result of postoperative complications as this represents the population of potential DRG 331 patients. The numerator was the group of patients assigned to DRG 330 only as a result of specific postoperative complications.⁷

2.4. Outcomes

The study outcomes were LOS and costs incurred by Medicare.

2.5. Statistical analysis

Descriptive statistics were used to describe the study cohort. Patient characteristics between DRG 331 and DRG 330 were compared using chi-square test for categorical variables and Student t-test for continuous variables. Among DRG migrated groups, we determined the top five CCs and reported the results using proportion. Hospital level DRG Migration rates were calculated, and hospitals were classified into one of three groups: low (0.1-16.6%), moderate (16.7-23.0%) and high (23.1-83.3%). We calculated LOS and Medicare costs for four groups (DRG 331 and three DRG migrated groups) and compared them using t-test. All statistical analyses were performed using SAS 9.4. A p-value of <0.05 was considered as the threshold for statistical significance.

RESULTS

Demographics characteristics

A total of 15,380 patients with DRG 330 were identified; after excluding 1) 13,803 patients with CC present on admission, 2) 353 patients with no CC (incorrectly coded), and 3) 3,487 patients from hospitals with fewer than five cases; 1,244 patients were included. Similarly, 3,896 patients coded as DRG 331 were included. Therefore, the final cohort included 5,120 patients who underwent colectomy in 615 hospitals (Figure 1). Of these patients, the majority were non-Hispanic white (87.5% for DRG 331 vs. 86.1% for migrated DRG 330) and women (61.1% for DRG 331 vs. 59.2% for migrated DRG 330). The mean (SD) age was 73.02 (±8.85) years for DRG 331 and 72.33 (±9.75) years in the migrated DRG 330 group. There were no statistically significant differences in patient demographics, including age, sex, and race across DRG 331 and DRG-migrated patients (Table 1).

Table 1.

Patient Demographics and Hospital by DRG Status Characteristics.

	DRG 331	DRG 330 migrated patients	p-value
Number of patients	4416	704
Age (yr)	73.02 (8.85)	72.33 (9.75)	0.0609
Sex
Male	1717 (38.9%)	287 (40.8%)
Female	2699 (61.1%)	417 (59.2%)
Race			0.3080
Non-Hispanic White	3863 (87.5%)	606 (86.1%)
Non-Hispanic Black	324 (7.3%)	62 (8.8%)
Hispanics	83 (1.9%)	9 (1.3%)
Others	146 (3.3%)	27 (3.8%)
Discharge Status (%)
Home	4136 (93.7%)	612 (86.9%)
Rehab	257 (5.8%)	80 (11.4%)
Non-Rehab (e.g., SNF)	13 (0.3%)	7 (1.0%)
Inpatient Death	2 (0.1%)	3 (0.4%)

DRG Discharge Status was statistically significant.

A significant difference was noted for discharge status with a higher percentage of patients in the migrated DRG 330 cohort who were discharged to a rehabilitation or non-rehabilitation (e.g., skilled nursing) facility (p<0.0001) (Table 1). Notably, LOS was shortest for the DRG 331 cohort and progressively increased by tertile for the migrated DRG 330 by almost 3 days (DRG 331: 4.63±1.17 vs tertile 1: 7.40 ±4.15; tertile 2: 7.51 ±3.46; tertile 3: 7.42 ±2.82; p<0.0001). Similar statistically significant patterns (p<0.0001) were ascertained for payment and total charge between the DRG 331 cohort and the DRG 330 Migration tertiles, as shown in Table 2.

Table 2.

Outcomes by No Migration and DRG Migration.

	No Migration	DRG Migration (DRG 331 to 330)
Number of hospitals	216	133	131	135
Number of patients	4416	157	183	364
Length of Stay (D)* (±SD)	4.63 (1.17)	7.40 (4.15)	7.51 (3.46)	7.42 (2.82)
Payment ($)* (±SD)	6,984.24 (3,292.88)	10,714.70 (8,084.78)	11,341.83 (8,025.84)	10,740.16 (5,848.65)
Total Charge ($)* (±SD)	46,278.46 (24,448.50)	58,409.78 (43,234.23)	65,782.75 (39,959.05)	64,846.35 (34,787.88)
Excess Days		434	527	1016

^*Length of stay, Payment, and Total Charges were all statistically significant, <0.001. Standard deviation in parentheses.

DRG Migration and Outcomes

DRG Migration rates among hospitals ranged from 0.1% to 83.3% (Figure 3). The top five reasons for DRG Migration are reported in Figure 3. For all migrated patients, the length of stay, total hospital cost, and CMS payment were significantly greater when compared to the reference group, DRG 331; however, the outcomes for migrated patients were similar across the tertiles (Table 2). LOS analysis for the tertiles demonstrated an excess of 93 and 582 days for tertile 2 and 3, when compared to the lowest tertile group, respectively. The most common complication after colectomy was postoperative ileus (or paralytic ileus) in each tertile, followed by other digestive system complications and acute posthemorrhagic anemia (Figure 3).

Figure 3. Top 5 Complications for DRG Migration by Tertile.

Each Tertile is coded with the associated DRG Migration Rate. White represents Tertile 1, Diagonal Stripes represent Tertile 2, and Grey represents Tertile 3.

DISCUSSION

An extraordinarily wide variability of DRG Migration rates (0.1-83.3%) for colectomy among U.S. Hospitals was ascertained and the lowest tertile group represents a possible “best practice” rate of postoperative complications. The advantage of this approach avoids the unrealistic expectation of zero tolerance for postoperative adverse outcomes. As expected, the tertiles were associated with a progressive increase in excess costs, payments, and hospital days with an increasing rate of DRG migration. This study complements current literature by confirming a high degree of outcome variability across US hospitals,^9–11 but also adds a list of potentially actionable but costly complication diagnoses which could be addressed by components of care often referenced within an enhanced recovery program.^12–14

Index cost drivers among high-cost and low-cost hospitals have been linked to procedure volume and DRG severity, including major complications and co-morbidities.^{15, 16} The present study refines this discussion by identifying clinically similar low-risk populations with no reported CC diagnoses present on admission across institutions. Barring errors in clinical documentation with respect to important diagnoses present on admission, or correctible administrative coding errors, colectomy patients who migrate from an initial classification of DRG 331 to a final classification of DRG 330 eliminates from consideration those patients whose reimbursement would have been defined by major co-morbidities or underlying physiological disease.

Our concept of DRG migration is designed to focus on diagnoses that drive assignment to reimbursement categories rather than readmission and mortality which are more heavily dependent on factors outside the clinician’s control and importantly are unrelated to reimbursement for the index hospital admission. This concept also allows the targeting of wasteful care plans, rather than simply attempting to reduce costs by failure to apply potentially beneficial processes of care without assessing the impact on the cost of care from the payer’s perspective.¹⁷ Therefore, DRG Migration provides additional granularity to relative quality of care both within and across institutions, as well as a platform for an effective cost-benefit assessment of proposed and implemented care component(s).

The implication of this study is to contain costs and enhance the quality of care for Medicare beneficiaries. The U.S. Department of Health and Human Services plans to allocate 90% of current Medicare fee-for-service payments to payments based on quality and value-based driven metrics, with 50% being tied to alternative payment models (APM).¹⁸ Approximately $500 billion in revenue resulted from operative procedures annually,¹⁹ so this shift in payment paradigm is especially important for the surgical specialties, who contribute much to the financial stability of healthcare institutions.

This paper has several limitations. The first is the utilization of administrative data, which does not include relevant clinical information and is necessarily subject to human error. Another limitation may be a heavy reliance on Medicare data, which may limit the generalizability of specific complications on outcome variation. As such, further studies should analyze this approach in broader insurance groups. Finally, this study did not explore the additional dollar impact of readmission rates and post-acute care costs because these factors are unrelated to the discharge DRG assignment, although these burdens clearly are greater for patients who migrated to DRG 330 from 331.^{15, 20} Despite this fact, it is evident that complications resulting from care during the index admission are associated with surgical readmission,^{21, 22} which are important when characterizing a hospital’s quality of care.⁸

CONCLUSIONS

DRG Migration is a metric potentially available to hospitals via the billing system to better understand the quality of care being provided to patients and to avoid financially punitive measures. The metric may also provide a better cross institution comparator of quality and a marker of value-based performance and ultimately refinement of payment policy-based on variations in outcomes.