ABSTRACT

Objective: To explore the effect of improving the quality of medical coding on the comprehensive performance evaluation of disease diagnosis related groups (DRGs) on the first page of medical records. Methods: The defect details of disease diagnosis coding and DRG specific indicators on the first page of hospital medical records were collected and analyzed. The number of defects between 2019 and 2020 was compared by using test or Fisher’s exact test, and technique for order preference by similarity to ideal solution (TOPSIS) method was used to rank the comprehensive evaluation indexes of hospital DRG performance in the same period. Results: The number of defects on discharged medical records in 2020 (20 cases, 0.033%) was significantly lower than that in 2019 (406 cases, 0.603%), and the difference was statistically significant (P < 0.001). From the evaluation results of TOPSIS method, the hospital’s corresponding period value increased from 0.488 to 0.512, and the comprehensive DRGs performance of the hospital kept improving, which was consistent with the quality improvement trend of the hospital’s medical record home page. Conclusion: The effective quality improvement of filling in the disease diagnosis coding on the first page of medical records lays a solid foundation for the comprehensive performance evaluation of DRGs and the improvement of medical quality.

Key words: medical records, disease diagnosis coding, plan-do-study-act cycle, TOPSIS method, diagnosis related group

INTRDUCTION

The data quality on the home page of inpatient medical records is an important aspect of healthcare quality management and a main indicator to measure healthcare quality and the standard of hospital management.[1] In recent years, China has emphasized using quality data as a starting point for the management of hospital healthcare quality, and the disease diagnostic codes on the medical record home page (MRHP) are a crucial data source. The rapid progress of pilot projects concerning the performance evaluation and diagnosis-related group (DRG)-based payment system of tertiary public hospitals has highlighted the significance of assessing quality control of the disease diagnostic codes in MRHPs. Since April 2019, a grade A tertiary general hospital in Shandong Province has incorporated the quality improvement of MRHPs into the Deming Cycle, a project management process based on continuous improvement, which includes a problem-solving four stages plan-do-study-act (PDSA). Processes such as organizational improvement, institutional guarantee, personnel training, and informatization management have greatly assisted the hospital in enhancing the quality of its MRHPs, while also laying a solid foundation for the comprehensive performance evaluation of DRGs and improvement of healthcare quality.

DATA SOURCE AND METHODS

Data source

Information of 128,447 patients discharged from a grade A tertiary general hospital in Shandong Province between January 1, 2019 and December 31, 2020 were collected from the MRHPs, of which 67,286 were from 2019 and 61,161 were from 2020. The number of MRHPs was consistent with the data in China Health Statistics Yearbook, national public hospital performance evaluation MRHP reporting system, and Shandong Province quality management and performance evaluation platform. The classification of MRHP disease diagnosis and coding defects (Table 1) and DRG-related evaluation indicators were acquired from the Shandong Province quality management and performance evaluation platform and the data were credible. The DRG-related evaluation indicators included the number of disease groups, enrollment rate, casemix index (CMI), proportion of grade-3 and -4 surgery, mortality rate in low-risk groups (total mortality rate of DRGs with a risk of mortality score of 1), time index, and cost index.

Table 1: Classification of disease diagnosis and coding defects on Shandong Province Quality management and performance evaluation platform
No. Defect type Remarks
1 Unclassifiable diagnosis No specific cause, unable to convey the purpose of hospital admission
2 Non-standard coding Failure to meet the requirements of the National Health Commission ICD-10 disease coding database
3 Invalid primary diagnosis Delivery outcomes, personal history, allergy history, postoperative status, etc.*
4 Irregular diagnosis Erroneous use of category and subcategory codes
5 Primary diagnostic errors in non-neonates Erroneous use of neonatal disease coding for patients aged over 28 days at admission
6 Gender errors in diagnostic coding Mismatch of disease coding with patient gender
7 Diagnostic coding errors in children aged 0–18 years Erroneous use of adult disease coding for children
8 Other coding errors Other coding errors that do not fall under any of the categories above
*Invalid primary diagnostic codes (partial) refer to B95–B97: bacterial, viral, and other infectious pathogens; T31: burns classified according to the extent of body surface involved; Z37–Z38: outcome of delivery, Z85: personal history of malignant neoplasm, Z86–Z87: personal history of other diseases, Z88: personal history of allergy to drugs, medicaments, and biological substances, Z89: acquired absence of limbs, Z90: acquired absence of organs, not elsewhere classified, Z91: personal risk factors, not elsewhere classified, Z92: personal history of medical treatment, Z93: artificial opening status, Z94: transplanted organ and tissue status, Z95: presence of cardiac and vascular implants and grafts, Z96–Z97: presence of other functional implants and other devices, Z98 other postprocedural states, and Z99: dependence on enabling machines and devices, not elsewhere classified.

Methods

The number of MRHP disease diagnoses and coding defects were compared in 2019–2020 using a chi-square test or Fisher’s exact test. The significance level was set at P = 0.05, and P < 0.05 indicated a statistically significant difference. A comprehensive DRG evaluation of the hospital during the same period was conducted using the technique for order preference by similarity to an ideal solution (TOPSIS). This method was originally proposed by Hwang CL and Yoon K in 1981, which involves ranking a finite set of alternatives by comparing their degree of similarity to an ideal target. It is used for multi-objective decision-making for finite alternatives in systems engineering[2] and is employed in the evaluation of hospital healthcare quality.[3]

RESULTS

MRHP disease diagnosis and coding defects

The number of MRHP defects among discharged cases in 2020 (20 cases, 0.033%) was lower compared to that in 2019 (406 cases, 0.603%) and the difference was statistically significant (P < 0.001).

Comparison of specific defect types revealed that the four subtypes: invalid primary diagnosis, gender errors in diagnostic coding, irregular diagnosis, and unclassifiable diagnosis showed significant differences (P < 0.001), whereas primary diagnostic errors in non-neonates did not (P = 0.685, Table 2).

Table 2: MRHP disease diagnosis and coding defects in 2019–2020
Item 2019 2020 χ 2 value P value
Total number of discharged patients (n) 67,286 61,161
Total number of disease diagnosis and coding defects [n (%)] 406 (0.603) 20 (0.033) 315.674 < 0.001
Specific defect type [n (%)]
    Invalid primary diagnosis 170 (0.253) 1 (0.002) < 0.001
    Primary diagnostic errors in non-neonates 3 (0.004) 1 (0.002) 0.685
    Gender errors in diagnostic coding 53 (0.079) 3 (0.005) < 0.001
    Irregular diagnosis 74 (0.110) 1 (0.002) < 0.001
    Unclassifiable diagnosis 106 (0.158) 14 (0.023) < 0.001

Comprehensive evaluation of DRG performance based on TOPSIS

To examine the impact of improving the quality of disease diagnosis and coding on the comprehensive evaluation of DRG system of the hospital in 2019–2020, TOPSIS method was employed. A raw data matrix of I rows and j columns was constructed using data from two years and the seven evaluation indicators, among which the indicators of high quality were X1 (number of disease groups), X2 (enrollment rate), X3 (CMI), X4 (proportion of grade-3 and -4 surgery), X5 (mortality rate in low-risk groups), X6 (time index), and X7 (cost index) (Table 3).

Table 3: Indicators for the comprehensive evaluation of hospital DRG performance
Year Number of disease groups Enrollment rate (%) Casemix index Proportion of grade-3 and -4 surgery (%) Mortality rate in low-risk groups (%) Time index Cost index
2019 639.00 97.11 0.874 56.86 0.00 1.00 0.82
2020 649.00 99.96 0.914 58.05 0.00 1.03 0.86

To ensure that all indicators changed in the same direction, CMI was multiplied by 100 (Xi × 100) and low-quality indicators were converted to high-quality indicators. Particularly, mortality rate in low-risk groups underwent difference transformation (1–Xi), while the time and cost indices underwent reciprocal transformation (1/Xi × 100) to obtain a homogenized indicator matrix (Table 4).

Table 4: Homogenized indicator matrix for the comprehensive evaluation of hospital DRG performance
Year X 1 X 2 X 3 X 4 X 5 X 6 X 7
2019 639.00 97.11 87.42 56.86 100.00 100.00 121.95
2020 649.00 99.96 91.37 58.05 100.0 97.09 116.28

Normalization was then performed on the data according to equation (1). See Table 5.

Table 5: Normalized indicator matrix for the comprehensive evaluation of hospital DRG performance
Year X 1 X 2 X 3 X 4 X 5 X 6 X 7
2019 0.702 0.697 0.691 0.700 0.707 0.717 0.724
2020 0.713 0.717 0.723 0.714 0.707 0.697 0.690

The positive-ideal solution and negative-ideal solution were as follows:

Z + = (0.713, 0.717, 0.723, 0.714, 0.707, 0.718, 0.724)

Z – = (0.713, 0.717, 0.723, 0.714, 0.707, 0.718, 0.724)

Using equation (2) and equation (3), the Euclidean distances Di+ and Di− of the indicators from the positive-ideal solution Z+ and negative-ideal solution Z– were calculated, respectively. Thereafter, equation (4) was used to calculate the closeness of the evaluation indicators for each alternative to the ideal solution Ci, where Ci takes values within the range (0, 1) and values closer to 1 indicated that the alternative was closer to the optimal level. The evaluation and ranking results are shown in Table 6. The TOPSIS evaluation results showed that the Ci value of the hospital had increased from 0.488 to 0.512 from 2019 to 2020. Thus, the DRG evaluation results indicated a significant upward trend, which was consistent with the trend of improvement found in the MRHP quality of the hospital.

Table 6: D i +, Di−, closeness and ranking for the comprehensive evaluation of hospital DRG performance
Year Di + Di − Ci Rank
2019 0.042 0.040 0.488 2
2020 0.040 0.042 0.512 1

DISCUSSION

Status of disease diagnosis and coding defects

Based on the feedback data on the Shandong Province quality management and performance evaluation platform, the MRHP disease diagnosis and coding defects in 2019–2020 primarily involved three aspects: (1) Invalid primary diagnosis: It is stipulated in a document issued by the National Health Commission that in principle, the primary diagnosis for a given hospital admission should be one with the greatest threat to the patient’s health, consumes the most healthcare resources, and involves the longest hospital stay.[4] Invalid primary diagnosis refers to diagnoses that cannot be classified as the primary diagnosis of the disease, as they cannot be included in the DRG grouping. (2) Unclassifiable diagnosis: The most common cases of unclassifiable diagnoses were those with “multiple burns” as the primary diagnosis and failed to indicate the location, degree, and area of burns according to the coding principle; and (3) Gender errors in diagnostic coding involved two cases: (i) Erroneous use of the code “N94.806” for the diagnosis of male patients with pelvic effusion, as “N94.8” refers to other specified conditions associated with female genital organs and the menstrual cycle.[5] Male patients with pelvic effusion in their diagnosis should be classified as peritoneal effusion; and (ii) Erroneous use of the code “O” for neonatal intrauterine infection instead of “P.” These errors indicate that coders tend to rely heavily on the code database in day-to-day practice, without the application of professional knowledge.[6]

Outcomes of comprehensive DRG performance evaluation

Following the reform of China’s health insurance payment methods, DRG-based payment system has received attention; results of DRG enrollment can have a direct impact on the economic benefits and performance management of hospitals.[7] Based on the TOPSIS analysis, it was seen that as the quality of the MRHP disease diagnosis improved and the number of disease groups increased from 639 in 2019 to 649 in 2020, while the enrollment rate increased from 97.11% to 99.96%, which is an accurate reflection of the wider disease spectrum treated at the hospital in its capacity as a regional medical center. Furthermore, the CMI increased from 0.874 to 0.914 and the proportion of grade 3 and 4 surgery increased from 56.86% to 58.05%, which suggests the reinforcement of departmental construction contributed to increased difficulty in diagnosis and treatment and number of patients admitted. The mortality rate of low-risk groups remained at a low level, indicating effective safeguarding of healthcare quality and patient safety; however, the time and cost indices showed clear upward trends. These trends are related to the hospital’s development of new technologies and new projects, and promotion of rehabilitation medicine, and also indicate that the hospital should attach greater importance to controlling the patients’ average length of hospital stay and average hospitalization costs.

Systematic quality improvement of disease diagnostic coding

The PDSA concept is derived from the Deming (plan-do-check-act, PDCA) Cycle, and the primary difference from PDCA is the replacement of passive checking (check) with active learning and evidence-based research (study), emphasizing both short-term continuous improvement and long-term organizational learning.[8–9] In accordance with the spirit of the “Notice on Initiating the 2019 National Tertiary Public Hospital Performance Evaluation” (National Health Commission Letter [2019] No. 371), the hospital used problems within MRHP disease diagnostic coding in the DRG performance evaluation feedback as the first step to incorporate MRHP quality improvement into the PDSA continuous-improvement project management. MRHP quality improvement was achieved primarily through strengthening: (1) the organizational and institutional guarantees, emphasizing the standardized establishment of the medical records department and cultivation of coding professionals, reinforcing knowledge and skill training of coders and clinicians in disease diagnostic coding, and an innovative implementation of medical record coders working alongside the clinical departments; (2) the establishment of mechanisms for multi-departmental coordination and incorporating MRHP disease coding defects into the closed-loop management of adverse events in medical safety; (3) the informatization and data-driven management by developing a MRHP logic verification system, continuous improvement of quality control rules over the course of the disease, and the advancement of medical record management from pre- to post-event control and process quality management; and (4) the monitoring and evaluation mechanisms of MRHP disease diagnosis and coding, incorporating the quality of MRHP disease coding in the management of the departmental comprehensive target responsibility system, providing regular feedback on relevant data and problems to clinical departments in the form of quality briefings, and prompting the departments to execute continuous improvement assessments.

Study limitations and prospects

This study only described the MRHP disease diagnosis coding and comprehensive DRG performance evaluation at the hospital level and did not present a comparison among different departments and attending physician teams. The requirement for a higher quality in hospital development and acceleration towards DRG-based and diagnosis-intervention packet payment systems require the development of intelligent MRHP coding systems based on artificial intelligence in the future, using technologies such as natural language processing, machine learning, self-learning algorithms, and strong error tolerance. Such systems will enhance the work efficiency, clinician and coder quality,[10] and will form a key component in the informatized construction of smart hospitals.

DECLARATIONS

Secondary publication declaration

This article was translated with permission from the Chinese language version first published by Modern Hospital Management.

Conflicts of interest

There is no conflict of interest among the authors.

Data sharing statement

No additional data is available.

REFERENCES

  1. Xue MQ, Sun RR. Analysis of the quality improvement in filling the home page of inpatient medical records. Zhongguo Bingan. 2018, 19(4):12–14.
  2. Liu GF. Medical Statistics. 2nd edition. Beijing: Peking Union Medical College Press, 2007: 336–337.
  3. Wang JW, Wang XQ, Wang SZ, et al. Comprehensive evaluation of hospital healthcare quality based on the TOPSIS and comprehensive index methods. Zhongguo Bingan. 2020, 21(11):75–78.
  4. General Office of the National Health and Family Planning Commission. Notice of the General Office of the National Health and Family Planning Commission on Printing and Distributing the Quality Standards for Filling the Home Page of Inpatient Medical Records (Provisional) and the Quality Management and Control Indicators for the Home Page Data of Inpatient Medical Records (2016 version). General Office of the National Health Commission (2016) No. 24, 2016.
  5. Dong JW. ICD-10: The International Statistical Classification of Diseases and Related Health Problems, 10th Revision. 2nd edition. Beijing: People’s Medical Publishing House, 2016: 565.
  6. Luo Z, You Y. Case analysis of coding ovarian sex cord-stromal tumors. Zhongguo Bingan. 2018, 19(7):21–23.
  7. ZhK. Impact of disease coding defects on DRG enrollment outcomes in traditional Chinese medicine hospitals under DRG-based payments. Zhongguo Weisheng Jingji. 2021(4):80–82.
  8. Jennifer C, Jennifer LS, Rivelli JS, Petrik AF, Seibel E, D'Agostini B, et al. Applying the plan-do-study-act (PDSA) approach to a large pragmatic study involving safety net clinics. BMC Health Serv Res. 2017, 17(1):411.    DOI: 10.1186/s12913-017-2364-3    PMID: 28629348
  9. Tang QL, Tao YL. Improving the compliance rate of nurses for implementing patient identification procedures based on the PDSA cycle. Zhongguo Yiyao Kexue. 2019, 9(15):176–180.
  10. Zhu MY. Research on the intelligent coding of the medical records home page based on medical artificial intelligence technology. Zhongguo Shuzi Yixue. 2018, 13(4):34–36.