Setting
Jiading District, located in the northwest of Shanghai, covers an area of 463.55 square kilometers. There were 1.58 million inhabitants in Jiading District (2017). The per capita GDP in Jiading District was 20,508 USD (2017), exceeding the high-income country threshold (12,235 USD) [17]. The life expectancy in Shanghai was estimated to be 83.85 years (2017), similar to some high-income countries [18].
Several advantages in Jiading district ensure the effective operation of DRGs-based ISM in the region. First, the healthcare information system infrastructure in Jiading District is well developed and unified coding system has been implemented. Second, Jiading district has a strong staffing, equipped with a quality control team responsible for the integrity of the front page of medical records (FPMR) data and a government department specialized in healthcare service management.
There is a total of 5 public hospitals in Jiading District including Central Hospital, Nanxiang hospital, Anting Hospital, Traditional Chinese Medicine Hospital and Maternal and Child Health Care Hospital. As the largest healthcare provider, public hospitals account for appropriately 95% of outpatient and inpatient services [18].
In this study, we included all the 5 public hospitals covered in the ISMS. No private hospitals were included since they were not covered in ISMS. Therefore, the performance of regional inpatient service mentioned in this study can only represent the situation of these 5 public hospitals.
Policy Intervention
In 2015, Jiading District Health Commission released policy documents planning to introduce and use DRGs-based inpatient service management. Then Jiading District entered the stage of policy preparation, and gradually carried out the construction of electronic medical records (EMR), standardization of filling requirements for FPMR, standardization of disease classification system. Specifically, the FPMR applied the 2012 national standard version, and the coding system adopted ICD-10 and ICD-9 Shanghai version. In 2016, a district level FPMR quality control group was established to conduct supervision and training on the integrity of FPMR. In 2017, Jiading District completed the construction of DRGs-based ISMS and entered the policy implementation period.
There are five aspects for implementing DRGs-based ISM policy: (i) DRGs-based Budget: including the budget for number of cases and cost. (ii) Supervision of Inpatient Service Quality: including the quality of EMR, disease classification and performance evaluation of inpatient service based on three dimensions of capacity, efficiency and quality. (iii) Incentive Mechanism: linking the performance evaluation results with government's investment in hospitals. (iv) Publicity of Inpatient Service Information: opening the supervision information of all DRGs to the hospital. (v) Discipline Evaluation: including the evaluation of the balanced development of different disciplines, as well as the evaluation of key disciplines of the hospital.
DRGs-based ISM policy differs from traditional approach of hospital management mainly in the following five points: (i) In the past, hospital management by government was relatively rough. The monitoring of service quantity and cost was mainly at hospital level rather than by different types of diseases. This may due to the high management cost and asymmetric of information. While through implementing ISMS, the budget management can be carried out based on DRGs, and the change in service quantity and average cost of diseases can be mastered at the disease level. (ii) In China, the health sector evaluates hospital performance on an annual basis, which induced deficiencies in three aspects compared with the performance evaluation in ISM policy. The first aspect is the data authenticity. In traditional approach of hospital management, the performance evaluation data was filled in by the hospital, rather than being retrieved through the backstage data in real time like ISMS did, which could avoid falsification of hospitals. The second aspect is the timeliness of the data. In the past, the performance of hospital was assessed once or twice a year while ISMS can achieve monthly performance evaluation, which greatly enhance the timeliness of data, and reflect the real-time performance of the hospital. The third aspect is the comparability of data. We mentioned above that traditional performance indicators such as average length of stay and average cost may have poor comparability without taking into consideration of different types of diseases. The performance indicators of ISMS can effectively solve these problems. (iii) The incentive mechanism of ISM policy can better encourage hospitals to achieve better medical output. In the past, government investment was not evidence-based. It mainly depended on the scale or losses of hospitals, instead of the output of hospitals. Through ISM, we can reward hospitals with high efficiency and productivity and maximize the use of national finance. (iv) Hospital performance can be shared with hospital in real time through ISMS. Hospitals can then obtain its own performance and compare it with other hospitals to assess whether its performance is excellent or whether there is room for improvement. (v) Finally, through the evaluation of hospital disciplines, it helps government to determine the advantages of each hospital in the region, and help the hospital to improve departments with poor performance.
Data sources and outcome indicators
Original data came from the FPMR of all five public hospitals covered in the ISMS in Jiading District from 2013 to 2019, including more than 510,000 discharged cases and involving a total cost of 589.15 million USD. ISMS groups the cases based on diagnosis name and treatment data extracted from FPMR. In addition, the medical expenses, length of stay, mortality and other data of each case were collected to calculate performance indicators which could be reported monthly, quarterly and annually. So through the ISMS, we collected quarterly data, from 2013 to 2019, of 7 DRGs-based performance measures [19, 20] including the capacity dimension (DRGs number, case-mix index (CMI), total weight), efficiency dimension (time efficiency index (TEI), cost efficiency index (CEI)), quality dimension (inpatient mortality of low-risk group cases (IMLRG), inpatient mortality of medium-to-low risk group cases (IMMLRG)).
Statistical analysis
2015-2016 is the preparation period of the policy, and 2017 is the actual implementation period of the policy. Therefore, this study uses the interrupted time series (ITS) design to evaluate the policy effect of DRGs-based ISM policy on the performance of regional inpatient service after its implementation in Jiading District in 2017. ITS design is considered as a strong quasi-experimental methodology in effect evaluation, which can be used to evaluate the long-term effect of a policy intervention without control group[21].
We used the segmented linear regression model to detect the change of level and trend (slope) of 7 performance measures before and after the implementation of ISM policy. The change of level indicated the change of performance measures at the time of intervention, while the change of trend indicated the long-term effect of policy. The typical segmented regression model of ITS is as follows [22, 23]:
In order to verify that the data meet the requirements of ITS design, we first plotted the quarterly data of 7 performance indicators of regional DRGs-based inpatient service in Jiading District from 2013 to 2019 and visually compared the trend of each quarter of performance indicators before and after the intervention[24]. We assumed linearity of the trend lines within each segment. In addition, we used Dickey–Fuller test to examine the stationarity of time series data [25]. Durbin-Watson test was used to check the serial autocorrelation. If there is serial autocorrelation, Prais-Winsten estimation was used to correct the first-order serial correlation error [18]. All data analysis was performed using R 3.5.1.