Diagnosis-Related Group (DRG)-Based Prospective Hospital Payment System can be well adopted for Acute Care Surgery: Taiwanese Experience with Acute Cholecystitis

Laparoscopic cholecystectomy (LC) is a common procedure for cholelithiasis paid by diagnostic-related groups (DRGs) systems. However, acute cholecystitis (AC) patients usually have heterogeneous conditions that compromise the successful implementation of DRGs. We evaluated the quality/efficiency of treating AC patients under the DRG system in Taiwan. All AC patients who underwent LC between October 2015 and December 2016 were included. Patient demographics, treatment outcomes, and financial results were analyzed. Patients were reimbursed by one of the two DRG schemes based on their comorbidities/complications (CC): DRG-1, LC without CC; and DRG-2, LC with CC. Hospitals were reimbursed the costs incurred if they were below the lower threshold (balanced sector); with the outlier threshold if costs were between the lower and outlier thresholds (profitable sector); and with the outlier threshold plus 80% of the exceeding cost if costs were higher than the outlier threshold (profit-losing sector). Among 246 patients, 114 were paid by DRG-1, and 132 were by DRG-2. In total, 195 of 246 patients underwent LC within 1 day after admission, and patients with mild AC had shorter hospital stays than those with moderate or severe AC. The complication rate was 7.3% with only one mortality. In total, 92.1% of patients in DRG-1 and 90.9% of patients in DRG-2 were profitable. The average margin per patient was 11,032 TWD for DRG-1 and 24,993 TWD for DRG-2. DRGs can be well adopted for acute care surgery, and hospitals can still provide satisfactory services without losing profit.


Introduction
Since the introduction of the first diagnosis-related group (DRG)-based payment system in the 1980s, DRGs have gradually become the basis for paying hospitals in many countries [1,2]. By providing a clear definition for each group, DRGs enable comparisons among healthcare systems, hospitals, and individual practitioners [3][4][5]. DRGs provide incentives for hospitals to pursue optimal care for patients by utilizing limited resources per patient [6].
DRGs make good sense in settings where patients have relatively homogenous characteristics, take routine tests, undergo routine treatments, and have a predictable length of hospital stay. Accordingly, elective laparoscopic cholecystectomy (LC) for symptomatic cholelithiasis is one of the most common procedures implemented in the DRG payment system [7]. On the other hand, an acute inflammatory gallbladder usually significantly increases surgical difficulty [8]. Furthermore, acute cholecystitis (AC) patients are typically admitted via the emergency department and undergo various diagnostic studies on an urgent basis. They may be operated on at midnight, with significantly longer operation times, and consume various amounts of resources based on the complexity of the disease [9]. These factors have negatively impacted the implementation of DRGs in AC patients.
In Taiwan, National Health Insurance (NHI) is the principal payer of the country's healthcare system [10,11]. Regarding DRGs for LC, NHI reimburses hospitals with incremental payments for 00 complications or comorbidities 00 (CC) without taking the elective/emergent nature and complexity of surgery into consideration. Although a similar DRG payment design has been criticized to raise potential financial risks for medical centers where the most expensive and riskiest patients are being cared for [9,12], in the current study, we reappraised the quality and efficiency for treating AC patients under the same DRG payment scheme in Taiwan.

Material and methods
This was a retrospective study that all of the patients who underwent LC due to AC in Chang Gung Memorial Hospital between October 2015 and December 2016 were identified. Patient demographics, preexisting diseases, Charlson Comorbidity Index (CCI) score, American Society of Anesthesiologists (ASA) classification for physical status, body mass index (BMI), routine laboratory tests, imaging studies, length of hospital stays, surgical outcomes, morbidities, and in-hospital mortalities were recorded. The severity of AC was classified according to Tokyo guidelines 2018 (TG18) as mild, moderate, and severe [13]. Patient comorbidities were defined according to the CCI score for their preexisting diseases. In addition, fees for medical services were provided by the hospital's financial system. The hospital cost was calculated as the sum of all the fees incurred during the index admission.
Since 2010, the Taiwan-version DRG (TW-DRG) payment system has been implemented by NHI. Each TW-DRG payment rate has a lower threshold and an outlier threshold. Patients who claim a fee lower than the lower threshold are reimbursed the same amount they claimed (sector A, the balanced sector); those who claim a fee between the lower and outlier thresholds are reimbursed with the outlier threshold (sector B, the profitable sector); and those who claim a fee more than the outlier threshold are reimbursed with the outlier threshold as in sector B plus an additional outlier payment (80% of the exceeding cost) to marginally compensate for the exceeding cost (sector C, the profit-losing sector). All the patients fell into one of the following two groups: DRG-1, 00 LC, without major complications or comorbidities 00 ; and DRG-2, 00 LC, with major complications or comorbidities. 00 The lower threshold and outlier threshold of the payment rate for DRG-1 and DRG-2 were 38,716 TWD and 64,146 TWD and 39,997 TWD and 81,843 TWD, respectively (TWD: Taiwan dollars, one US dollar is approximately 30 TWD). These DRGs reimbursed the hospital claim for the index admission during which LC was performed. According to the TW-DRG, there were add-on payment adjustments, such as for higher hospital levels, a higher case mix index (CMI), and for hospitals located in remote areas. As these adjustments were subject to change over time, only base rates were applied in the current study to avoid confusion.
The patients in each DRG were further divided into three subgroups according to the severity of AC: group 1, mild AC; group 2, moderate AC; and group 3, severe AC. AC was managed according to the well-recognized principle: early LC was the procedure of choice whenever possible. Patients who were contraindicated for early LC were managed by percutaneous gallbladder drainage followed by delayed LC 6-8 weeks later. Patient demographics, results of laboratory tests, operative details, preoperative and postoperative hospital stay, ICU stay, total length of hospital stay (LOS), treatment outcome, and financial aspects of the DRG, including medical costs, reimbursement and margin, were compared between the groups.
Descriptive statistics were calculated for the cohort. The financial variables, including cost, revenue, and margin, are shown in TWD. Frequency tables were generated for categorical variables, and continuous variables were summarized by the means and standard deviations (SDs). Continuous data were analyzed using Student's t-test or one-way ANOVA to compare the means of two or more independent groups, respectively. Tukey's post hoc test was used following one-way ANOVA to test for differences between the groups. All statistical analyses were performed using SPSS (version 21.0, Chicago, IL, USA). A value of p \ 0.05 was considered statistically significant.

Results
In total, 246 patients (142 males and 104 females) were included in this study. Their mean age was 56.7 ± 15.8 years. Most of the patients had mild or moderate AC, and only 14 patients had severe AC. Patients with severe AC were significantly older than those with mild or moderate AC (67.6 ± 12.7 vs. 56.1 ± 15.7 years, p = 0.008). According to the financial records, 114 patients were paid by DRG-1, and the remaining 132 patients were paid by DRG-2. The male/female sex distribution, BMI, white blood cell (WBC) count, and percentage of neutrophils were similar between groups; however, patients in DRG-2 were older, more likely to have multiple comorbidities (CCI C 3), and had worse physical status (ASA classification C 3) compared to those in DRG-1. Moreover, the CCI and ASA scores were more likely to be higher in patients with severe AC than in patients with mild or moderate AC. The mean plasma CRP levels were significantly higher in the patients with severe AC than in the patients with mild and moderate AC; the mean CRP level for the DRG-2 patients was also higher than that for the DRG-1 patients (Table 1).
Overall, 195 of the 246 patients (79.3%) underwent LC within 1 day after admission, and the average time from admission to surgery was longer for patients in DRG-2 than for those in DRG-1 (1.2 ± 1.4 vs. 0.7 ± 1.3 days, p = 0.015). The mean operation time and amount of blood loss were similar between the two groups. Eighteen of the 246 patients had a surgical complication that was classified as Clavien-Dindo grade 2 or higher. The complication rate was similar between the two groups. Among the 246 patients, the mean total LOS was significantly shorter for patients with mild AC than for patients with moderate and severe AC (3.2 ± 1.4 vs. 5.2 ± 7.5 vs. 5.3 ± 4.2 days, respectively, p = 0.025). Six of the 14 patients with severe AC (42.9%) required postoperative care in the ICU, and the mean ICU stay was 2.3 ± 3.5 days. For DRG-1, the mean postoperative LOS and total LOS were significantly different between patients with ( Fig.1) mild, moderate and severe AC. In contrast, such differences were not observed in DRG-2 ( Table 2).
As shown in Table 3, the hospital cost for most of the patients fell into sector B. Only 3 patients were in sector A, and 18 patients were in sector C. Although those patients who fell into sector C resulted in negative margins (i.e., the hospital was reimbursed less than the actual hospital cost), the overall margin for the hospital was still positive, as it was compensated by the net income from sector B. The average margin per patient was higher in DRG-2 than in DRG-1 (24,993 TWD for DRG-2 and 11,032 TWD for DRG-1). As shown in Tables 1 and 2, the costs associated with each patient were regrouped in 5000 TWD scales starting from 35,000 TWD (below the lower threshold) to 95,000 TWD and above (higher than the outlier threshold). A large proportion of hospital claims fell into the lower half of sector B of DRG-1 and DRG-2; however, 10 patients in DRG-2 had extraordinarily high hospital costs that fell into in sector C (Fig. 2). By comparing the 18 negative-margin patients with the other 128 patients, several demographic and clinical factors that were related to the consumption of more medical resources are identified. As shown in Table 4, these factors include older age, higher CCI, severe AC, prolonged operation time, presence of surgical complications, prolonged postoperative, ICU, and total length of hospital stay.   Fig. 1 Distribution of hospital claims for mild, moderate, and severe acute cholecystitis in DRG-1. Each DRG has a lower threshold and an outlier threshold. Hospitals were reimbursed with the incurred costs if the costs claimed were below the lower threshold (sector A); with the outlier threshold if the costs claimed were between the lower and outlier thresholds (sector B); and with the outlier threshold plus 80% of the exceeding cost if the costs were higher than the outlier threshold (sector C). The costs of each patient were regrouped in 5000 TWD scales starting from 35,000 TWD (below the lower threshold) to 95,000 TWD and above (higher than the outlier threshold). The values are shown in Taiwan dollars (TWD). One USD = approximately 30 TWD  In general, patients with a poor preoperative physical condition or severe AC, or sustained a surgical complication often had a slow recovery that led to negative margins.

Discussion
The original purposes of DRGs were to facilitate hospital management by improving the quality of health care [1,14]. It is now a principal type of hospital payment system in many countries [1,15]. The successful implementation of the DRG payment system has been reported elsewhere [16][17][18]. LC has already become the treatment of choice for AC [19,20]. LC for AC has proven efficacious and safe [13,21,22]. A number of randomized trials have shown consistent results: the mean hospital stay is between 3 and 5 days, and the incidence of serious complications is approximately 6%, with almost no mortality [8,23,24]. Such consistency provided a solid rationale to incorporate AC into the DRG payment system. When each AC grade within a DRG scheme was compared, patients with severe AC in DRG-1 had a longer postoperative stay and ICU stay than those with mild and moderate AC. On the other hand, such differences were less significant in DRG-2: only ICU stay, not postoperative stay, was longer in patients with severe AC. This could in part be explained by the fact that, regardless of the severity of AC, patients with more complex comorbidities required a certain amount of time to recover from surgery. Therefore, these patients should be reimbursed by a DRG scheme different from those without major comorbidities.
Similar to DRGs of many countries, the incremental payment for the LC DRG scheme was based on ''comorbidities and complications'' but not the urgent or emergent nature of the procedure [3,9]. However, this policy has been under debate. Schneider et al. [25] argued that surgeons are likely to be discouraged from treating emergency patients if they cannot be compensated properly for the increased risks and challenges. In an earlier investigation, Chen et al. [9] reported that the total cost for urgent cholecystectomies was approximately 90% higher than that for elective cholecystectomies. They suggested that such increased costs and resource consumption should be recognized when establishing reimbursement rates [9]. Nonetheless, following decades of DRG implementation, it appears that patient comorbidities are still the decisive factor that accounts for the costs and resource consumption. For example, a recent study by Boehme et al. [26] showed that patient comorbidities increased postoperative resource utilization after LC. They suggested that comorbidities play a major role in clinical risk stratification and should be considered in the bundled reimbursement package [26]. Similar conclusions can be found in the literature for other diseases [27]. In one study regarding the treatment of intertrochanteric femur fractures, there was a net profit for treating more comorbid patients but a net deficit for treating healthier patients [28]. Likewise, our results also showed that patients in DRG-2 (with comorbidities) had higher margins per patient than those in DRG-1 (without comorbidities). This finding is most likely due to the design of the weight-adjusted reimbursement system which provides incentives for hospitals to treat patients with more comorbidities. Along with the DRG payment system, so-called ''cherry picking'' and ''dumping'' are potential unintended consequences that have been widely discussed. To maximize profits and avoid unnecessary financial losses, hospitals may attempt to select the less costly, more profitable cases and transfer or avoid the unprofitable ones. However, research around the world suggested that these unintended consequences are relatively rare through regular auditing and monitoring [1,29,30]. In Taiwan, add-on payment adjustments exist for higher hospital levels with higher case mix indexes to compensate for the efforts of treating complicated cases. Furthermore, our national data suggested that the best solution to earning a profit under the DRG payment scheme is to curtail the costs of radiological images and medication and perform fewer additional operations, which is more likely to be achieved in larger centers with experienced staff [10]. In this study, the 18 patients who led to negative margins appeared to either be old with poor physical status or had sustained a surgical complication. Both conditions tended to consume more medical resources and required longer hospitalization. Medical centers find it impossible to give up treatment for these patients. The best strategy to reduce costs and limit hospital financial losses is a meticulous evaluation of the patient, treatment of acute cholecystitis with an appropriate approach, and avoidance of complications that may lead to a prolonged hospital stay. Therefore, cherry picking and dumping are not unintended consequences of DRG but are ways to improve the global results for the treatment of acute cholecystitis.
For high-income countries, the average profit per patient in this study seems to be very low (24,993 TWD, approximately USD 850, Table 3). This is probably due to the design of Taiwan's NHI system that provides universal healthcare coverage for virtually every Taiwanese individual [31]. The universal NHI system has gained worldwide attention due to its low rates and high quality [11,32]. The national health expenditures have been held to less than 7% of GDP, which compares very favorably to the United States' 17.8% and approximately 10% of other high-income countries as of 2016 [31,[33][34][35][36]. Even with such a constrained budget, hospitals have still managed to provide efficient and quality medical services, as demonstrated by the short period of admission to operation and a short LOS. Furthermore, as shown in Figs. 1 and 2, the claims for most of the cases managed to fall into the lower half of sector B, which is the 00 most profitable location. 00 There are several limitations to this study. First, the current study was a retrospective review of a single-center experience, not necessarily reflecting the financial results of other hospitals. However, as our institution routinely receives difficult and complicated cases from around the country, our results showed that DRGs can be well adopted under such settings for acute care surgery. Second, the costing methodologies are also different across countries to address the specific financial structure of each country [3,37]. DRG authorities in each country may exchange experiences for new perspectives to optimize their existing DRG systems.
In conclusion, hospitals have been triggered to enhance their management to maintain profit from the DRG payment system. Our experience shows that, using AC as an example, DRGs can be well adopted for acute care surgery.

Compliance with Ethical Standards
Conflict of interest The authors declare that they have no conflict of interest.
Informed Consent This is a retrospective chart-review study, no identifying information of any individual is included.