http://orcid.org/0000-0002-8329-8927
Figures
Abstract
Background
Many comparability problems appear in the process of the performance assessment of medical service. When comparing medical evaluation indicators across hospitals, or even within the same hospital, over time, the differences in the population composition such as types of diseases, comorbidities, demographic characteristics should be taken into account. This study aims to introduce a standardization technique for medical service indicators and provide a new insight on the comparability of medical data.
Methods
The medical records of 142592 inpatient from three hospitals in 2017 were included in this study. Chi-square and Kruskal-Wallis tests were used to explore the compositions of confounding factors among populations. The procedure of stratified standardization technique was applied to compare the differences of the average length of stay and the average hospitalization expense among three hospitals.
Results
Age, gender, comorbidity, and principal diagnoses category were considered as confounding factors. After correcting all factors, the average length of stay of hospital A and C were increased by 0.21 and 1.20 days, respectively, while that of hospital B was reduced by 1.54 days. The average hospitalization expenses of hospital A and C were increased by 1494 and 660 Yuan, whilst that of hospital B was decreased by 810 Yuan.
Citation: Wu L, Ji C, Lu H, Hong X, Liu S, Zhang Y, et al. (2018) Standardization of medical service indicators: A useful technique for hospital administration. PLoS ONE 13(11): e0207214. https://doi.org/10.1371/journal.pone.0207214
Editor: Sharon Mary Brownie, Aga Khan University, KENYA
Received: April 13, 2018; Accepted: October 26, 2018; Published: November 28, 2018
Copyright: © 2018 Wu et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the paper and its Supporting Information files.
Funding: This study was supported by the Administration of Traditional Chinese Medicine of Zhejiang Province (NO.2017ZZ007) to CJ. The website of the program is http://zgj.zjtcm.net/page/jc/article/notice.html?id=99 The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Healthcare systems across the world are facing the challenges of meeting growing demand, as well as increasing productivity, reducing costs and improving outcomes[1–3]. How to fairly allocate the scarce medical resources in an efficient and effective manner to meet the medical needs of the population while at the same time curb the excessive growth of medical costs is one of the major challenges for governments at all levels[4–6]. To solve this dilemma, various reimbursement mechanisms and medical quality evaluation indicators were introduced[7–9], for instance, diagnosis-related group (DRG), a patient classification system that standardizes prospective payment to hospitals and encourages cost containment initiatives which firstly adopted by the US Medicare Programme as the currency for reimbursing hospitals on a prospective, per-case basis[10]. As a hospital reimbursement and performance monitoring tool, DRG now has been introduced and indigenized in several countries [8, 11, 12]. Chinese health regulator has also actively explored the feasibility of DRG in China and piloted in some areas. Besides, in order to alleviate the problems of biased resource allocation and high patient flows to large hospitals, China implemented a hierarchical medical system[13]. It’s a two-way referral system that enables the basic hospitals to treat common diseases, and patients with intractable diseases are transferred to higher-level hospitals. Optimizing the average length of stay (ALOS) and controlling the average hospitalization expense(AHE) were cited as high priorities for health service providers, behind these policies, and were considered as two important efficiency indicators to assess the medical quality and management level of many health systems[14–16].
However, there are differences in rates of some phenomena between populations. They are usually confounded by the population compositions which cannot be directly compared [17, 18]. Similarly, comparability problems also exist in the assessing of medical services performance. For instance, the costs of surgical patients are higher than those of non-operative patients who have the same disease. Meanwhile, the length of hospital stays and costs for critical patients are usually higher than those of mild patients. When evaluating the medical service indicators among hospitals, disease interference is inevitable as long as there exist attribute and severity differences, which will eventually result in the medical variance. And this difference can even be caused by unreasonable medical expenses. Simply comparing the values without considering the actual condition of the patients is unfair to those hospitals with more critical patients and will dampen their enthusiasm.
Therefore, to improve the evaluation quality and make the medical service indicators results more comparable among hospitals, and among different time periods of the same hospital, the details of population composition such as types of diseases, comorbidities, demographic characteristics, etc. should be taken into account[4, 19]. DRG-based payment approach can control the costs, reduce the care intensity and shorten the hospital stays by grouping similar patients. However, it's a composite indicator which is not applicable in the assessment of single medical service indicators. Thus, it is imperative for us to find a more appropriate and objective method for the comparation of medical indicators. Standardization method is a commonly used technique for adjusting the confounding effects of population composition to enhance the comparability of indicators among multiple populations [20, 21]. The purpose of this study is to introduce a specific standardization technique for medical service indicators in hospital management by using the first-hand clinical data from three general hospitals. ALOS and AHE, the most commonly used indicators, were taken as example for presentation. Altogether, this study identified the existing needs for the assessment of medical service utilization and provided a new insight for financial reimbursement.
Materials and methods
Study design and data source
The study began on February 1, 2018. A total of 160164 inpatient medical records in 2017 were collected from three tertiary general hospitals, retrospectively. Patients who discharged from the emergency medical department or less than 18 years old were excluded. To protect patients’ privacy, their identities were concealed, and only medical record numbers were used. This study was in conformity with the “Ethics review methods for biomedical research involving human” promulgated by the Ministry of Health of The People's Republic of China and was performed in according to the Helsinki Declaration. The protocol has been approved by the Ethics Committee of the First Affiliated Hospital of Zhejiang Chinese Medical University.
Measurement of variables
The data set used in this research included the individual patient variables of age, sex, date of admission and discharge, the principal diagnosis and disease code (first listed diagnosis), the number of comorbidities, admission type, length of stay, and total expenses. The principal diagnoses were coded at discharge according to the International Classification of Disease, Tenth Revision (ICD-10). There was no missing data in this analysis.
In order to facilitate the comparison of internal diseases among three hospitals, the principal diagnoses were divided into 22 categories according to ICD10 codes. The categories are as shown in Table 1.
Statistical analysis
Standardized indicators are more comparable, which means, it’s important to distinguish the differences between observed indicators into true differences and the differences caused by the component effects of the confounding factors[20, 22].
Suppose there is a medical service indicator (T) that needs to compare among three hospitals, recorded as T1, T2 and T3. The differences originated from confounding factors (e.g. age, disease). The algebraic expression for computing a standardized indicator with two confounding factors is shown as follows.
Here, the sum of discharges from three hospitals was used as the standard population. Nij denotes the standard population in the ijth category of confounding factor (i = 1,2,3…,i; j = 1,2,3,…,j); Tij is the crude value in the ijth category; T′ denotes standardized value. The calculation process is shown in Table 2.
In this study, the ALOS and AHE were standardized according to confounding factors stratified and compared among three hospitals. According to literatures, ALOS and AHE were associated with age, females, patients with more comorbidities, patients with a higher DRG weight and the incentives of the financing system [23–25]. Hence, age, gender, comorbidity and principal diagnoses category were considered as confounding factors. T test, Chi-square and Kruskal-Wallis tests were used to verify these factors. Statistical analyses were performed using the Statistical Package for Social Sciences (SPSS for windows, version 18.0; Chicago, Illinois, USA). Statistical significance was set at P<0.05 (two-tailed).
Results
Clinical and demographic characteristics
Totally 160164 discharged patients were collected, 17572 cases were excluded according to exclusion criteria, and 142592 cases were analyzed. As shown in Table 3, the ALOS and AHE in total were 11.75 days and 16341 Yuan, respectively. The ALOS of three hospitals from low to high was hospital C (10.55 days) < hospital A (11.50 days) < and hospital B (16.13 days). Whereas, the AHE was hospital B (11028 Yuan) < hospital C (16663 Yuan) < hospital A (17299 Yuan).
Table 4 provided the baseline characteristics of patients from each hospital. The results indicated that the population compositions, e.g. gender, age, comorbidities and disease classification, were significantly different among three hospitals. The proportion of female were higher than that of male in all three hospitals. The median age of patients in hospital A, hospital B, and hospital C were 57 years (IQR 41~72), 62 years (IQR 45~78), and 53 years (IQR 39~67), respectively. Patients in hospital B were older than hospital A and hospital C. The percentage of comorbidities in hospital B (79.1%) was also higher than the other hospitals. Meanwhile, there were some differences in the disease composition among three hospitals. The highest percentage of diseases diagnoses category in hospital A, B, and C was I00-I99 (17.9%), I00-I99 (17.4%), and C00-D48 (30.0%), respectively.
Standardization of ALOS
The total number of discharged patients in all three hospitals was taken as a common standard to facilitate comparisons. Assume that the principal diagnoses category played a crucial role in hospitalization days and expense among all confounding factors. Firstly, the standard population were stratified by the principal diagnoses category, and the standardized ALOS was calculated. Then, the ALOS was adjusted by comorbidities and the principal diagnoses category. The standardized processes were shown in Tables 5–7. Similarly, the remaining confounding factors were adjusted in the same way. Table 7 presented the adjusted ALOS in each step. It changed every time after adjusting each of the confoundings. After correcting all factors, the ALOS of three hospitals from low to high was hospital A (11.71 days) < hospital C (11.75 days) < and hospital B (14.59 days). In other words, the ALOS of hospital A and C were increased by 0.21 and 1.20 days, respectively, whilst that of hospital B was reduced by 1.54 days.
Standardization of the average hospitalization expense
The calculation method of standardized AHE was the same as standardized ALOS. Its calculation process was shown in Tables 8–10. Table 10 summarized the results of AHE in adjusting confounding factors. If the compositions of all confounding factors (i.e. gender, ethnicity, age, and education in this study) were the same, the differences in standardized AHE of hospital A and C were increased by 1494 and 660 Yuan, while that of hospital B was reduced by 810 Yuan. The outcome is in accordance with ALOS.
Discussion
A total of 142592 discharged patients from three hospitals were analyzed in the current study. Our results implied that the differences in compositions of demographic characteristics, comorbidity and principal diagnoses category might impose a substantial impact on comparing observed outcomes among three hospitals. When comparing with the crude value, the ALOS of hospital A and C were increased by 0.21 and 1.20 days, while that of hospital B was decreased by 1.54 days. The AHE displays the same trend. That is, the standardized AHE of hospital A and C were increased but that of hospital B was reduced when compared with before.
As far as we know, standardization technique is commonly used for comparing rates, such as cure rates, death rates and birth rates, between different groups or populations[17, 18, 26]. However, there currently has been no report on the application of standardization method in hospital administration. Our findings indicated that the idea of stratified standardization can also be applied to the evaluation of medical services. Our study presented a detailed analysis and discussion of the standardization method, including the determination of common standard, identification of confounding factors, and hierarchical standardization of ALOS and AHE. The results demonstrated that after adjusting confounding factors the real differences in ALOS and AHE among three hospitals were much smaller than the original values, although the order did not change dramatically. For instance, the difference between the maximum and the minimum values of ALOS was reduced from 5.58 days to 2.88 days after adjusting disease category, comorbidity, age and gender. And the trend of AHE is consistent with that of ALOS which suggested that the adjusted medical service indicators are more reflective of the quality of care between hospitals.
Under the reform of public hospitals in China, the evaluation mechanisms of public hospitals are becoming more and more competitive. The standardization method could effectively increase the comparability of medical service indicators and has positive significance for the formulation of public hospital policy. Standardized indicators could improve the fairness of hospital assessment and reduce speculation. In order to control the average hospital expense, some medical institutions adopt unreasonable ways to reduce the expense, such as re-admission of long-term inpatients and admission of mild patients who do not require hospitalization. The impact of these opportunistic behaviors could be adjusted through standardization during assessment. On the other hand, it is conducive to promoting the implementation of hierarchical medical system, reducing the burden of large hospitals and enhancing the capacity of primary medical services. China has vigorously promoted the implementation of the hierarchical medical system to provide different levels of medical services according to the patients' conditions and to realize rational allocation of medical resources[13]. Although the hierarchical medical system has many advantages, its impact is still limited. Large hospitals are still overcrowded, while primary medical institutions are to some extent unwanted. Through the standardization of medical data, it is possible to make the problem more obvious for health department and make it easy to identify those high-level hospitals that treated a large number of patients with mild illnesses. Based on this, the government and health department can better supervise these hospitals and eventually optimize the system model to achieve the rational allocation of medical resources.
The application and continuous evaluation of clinical pathways (CP) in health-care settings benefit the institutionalization of culture of quality in hospitals[27]. The standardized method can be used to adjust the assessment indexes in each stage of CP. The process of disease diagnosis and treatment will be more normative after standardized indicators were applied. Moreover, the symptoms of inpatients are complex and diverse, the adoption of a “one-size-fits-all” approach will inevitably dampen doctors' enthusiasm. The hierarchical standardization of medical indicators is likely to promote the classification management of disease and provide direction for continuous improvement of medical quality.
The allocation of funds and health resources as well as the control of deficits of the national health system are the major and long-standing problems, which are also at the heart of health care reform. [28]. Through the standardization of the composition of medical expenses, it will be possible to find and solve the core problem of “expensive medical treatment”. Currently, DRG approach has been recognized and our standardization method could be a complement to it. It provides a standard for more precise grouping of DRGS and an objective basis for differentiated financial subsidies.
In drawing meaningful conclusions from this study, it is important to be cognizant of its limitations. Not all the confounding factors have been taken in to consideration. Only age, sex, and disease category were selected in this analysis. Some specific details of disease diagnosis, such as tumor stage, were not included as confounding factors. Notwithstanding these limitations, this study highlights an approach and some suggestions in the comparison of medical indicator evaluation. In addition, these standardized indicators no longer reflect reality, they are only a reference level for comparisons between hospitals and departments within hospitals.
Conclusions
There are many comparability problems in the assessment of medical service performance. In this study, by taking ALOS and AHE as examples, we introduced a specific technique for standardization, which will be helpful to improve the comparability of medical service indicators. In addition to the confounding factors described in this paper, other potential confounding factors may also contribute to the standardization. Our findings showed that our standardization method could be a practical technique and worthy of promotion.
Acknowledgments
This study was supported by the Administration of Traditional Chinese Medicine of Zhejiang Province (NO.2017ZZ007). The authors wish to express their sincere thanks to Dr. Kan Zhu from University of California Davis for critical reading and editing of the manuscript.
References
- 1. Tucker S, Hargreaves C, Wilberforce M, Brand C, Challis D. What becomes of people admitted to acute old age psychiatry wards? An exploration of factors affecting length of stay, delayed discharge and discharge destination. International journal of geriatric psychiatry. 2016. pmid:27515770.
- 2. Pirozek P, Komarkova L, Leseticky O, Hajdikova T. Corporate governance in Czech hospitals after the transformation. Health policy (Amsterdam, Netherlands). 2015;119(8):1086–95. Epub 2015/05/24. pmid:26001299.
- 3. Kuhlmann E, Larsen C. Why we need multi-level health workforce governance: Case studies from nursing and medicine in Germany. Health policy (Amsterdam, Netherlands). 2015;119(12):1636–44. http://dx.doi.org/10.1016/j.healthpol.2015.08.004.
- 4. Shi L. Patient and hospital characteristics associated with average length of stay. Health care management review. 1996;21(2):46–61. Epub 1996/01/01. pmid:8860041.
- 5. Yin J, Luras H, Hagen TP, Dahl FA. The effect of activity-based financing on hospital length of stay for elderly patients suffering from heart diseases in Norway. BMC health services research. 2013;13:172. Epub 2013/05/09. pmid:23651910; PubMed Central PMCID: PMCPMC3651263.
- 6. Wolff J, McCrone P, Patel A, Normann C. Determinants of per diem Hospital Costs in Mental Health. PloS one. 2016;11(3):e0152669. pmid:27031955; PubMed Central PMCID: PMC4816317.
- 7. Adirim T, Meade K, Mistry K, Council On Quality I, Patient S, Committee On P, et al. A New Era in Quality Measurement: The Development and Application of Quality Measures. Pediatrics. 2017;139(1). pmid:28025242.
- 8. Endrei D, Zemplenyi A, Molics B, Agoston I, Boncz I. The effect of performance-volume limit on the DRG based acute care hospital financing in Hungary. Health policy (Amsterdam, Netherlands). 2014;115(2–3):152–6. Epub 2014/01/11. pmid:24406058.
- 9. Or Z. Implementation of DRG Payment in France: issues and recent developments. Health policy (Amsterdam, Netherlands). 2014;117(2):146–50. pmid:24962536.
- 10. Fetter RB. Diagnosis related groups: understanding hospital performance. Interfaces. 1991;21(1):6–26.
- 11. Serden L, O'Reilly J. Patient classification and hospital reimbursement for inguinal hernia repair: a comparison across 11 European countries. Hernia: the journal of hernias and abdominal wall surgery. 2014;18(2):273–81. pmid:24077862.
- 12. Busse R, Geissler A, Aaviksoo A, Cots F, Hakkinen U, Kobel C, et al. Diagnosis related groups in Europe: moving towards transparency, efficiency, and quality in hospitals? BMJ (Clinical research ed). 2013;346:f3197. pmid:23747967.
- 13. Wang Y, Sun L, Hou J. Hierarchical Medical System Based on Big Data and Mobile Internet: A New Strategic Choice in Health Care. JMIR Medical Informatics. 2017;5(3):e22. PubMed PMID: PMC5566626. pmid:28790024
- 14. Taylor NF, Brusco NK, Watts JJ, Shields N, Peiris C, Sullivan N, et al. A study protocol of a randomised controlled trial incorporating a health economic analysis to investigate if additional allied health services for rehabilitation reduce length of stay without compromising patient outcomes. BMC health services research. 2010;10:308. Epub 2010/11/16. pmid:21073703; PubMed Central PMCID: PMCPMC2998505.
- 15. Borghans I, Heijink R, Kool T, Lagoe RJ, Westert GP. Benchmarking and reducing length of stay in Dutch hospitals. BMC health services research. 2008;8:220. pmid:18950476; PubMed Central PMCID: PMC2582034.
- 16. Clarke A, Rosen R. Length of stay. How short should hospital care be? European journal of public health. 2001;11(2):166–70. Epub 2001/06/26. pmid:11420803.
- 17. Hertz-Picciotto I, Din-Dzietham R. Comparisons of infant mortality using a percentile-based method of standardization for birthweight or gestational age. Epidemiology. 1998;9(1):61–7. pmid:9430270.
- 18. Jie X, Qingyan M, Zheng WY. Comparable results between standardization methods and regression analysis in predicting mortality rate among samples with burns. Burns: journal of the International Society for Burn Injuries. 2003;29(3):247–55. pmid:12706618.
- 19. Kim SJ, Park EC, Jang SI, Lee M, Kim TH. An analysis of the inpatient charge and length of stay for patients with joint diseases in Korea: specialty versus small general hospitals. Health policy (Amsterdam, Netherlands). 2013;113(1–2):93–9. Epub 2013/10/22. pmid:24139937.
- 20. Wang JC, Carlson RG, Falck RS, Leukefeld C, Booth BM. Multi-sample standardization and decomposition analysis: An application to comparisons of methamphetamine use among rural drug users in three American states. Stat Med. 2007;26(19):3612–23. PubMed PMID: WOS:000248549200007. pmid:17243192
- 21. Wang J, Rahman A, Siegal HA, Fisher JH. Standardization and decomposition of rates: useful analytic techniques for behavior and health studies. Behavior research methods, instruments, & computers: a journal of the Psychonomic Society, Inc. 2000;32(2):357–66. pmid:10875185.
- 22. Yuan C, Wei C, Wang J, Qian H, Ye X, Liu Y, et al. Self-efficacy difference among patients with cancer with different socioeconomic status: application of latent class analysis and standardization and decomposition analysis. Cancer epidemiology. 2014;38(3):298–306. pmid:24656649.
- 23. Kuo YF, Goodwin JS. Effect of hospitalists on length of stay in the medicare population: variation according to hospital and patient characteristics. Journal of the American Geriatrics Society. 2010;58(9):1649–57. pmid:20863324; PubMed Central PMCID: PMC2946246.
- 24. Wong EL, Cheung AW, Leung MC, Yam CH, Chan FW, Wong FY, et al. Unplanned readmission rates, length of hospital stay, mortality, and medical costs of ten common medical conditions: a retrospective analysis of Hong Kong hospital data. BMC health services research. 2011;11:149. Epub 2011/06/18. pmid:21679471; PubMed Central PMCID: PMCPMC3146405.
- 25. Theurl E, Winner H. The impact of hospital financing on the length of stay: evidence from Austria. Health policy (Amsterdam, Netherlands). 2007;82(3):375–89. pmid:17166618.
- 26. Doak CM, Wijnhoven TM, Schokker DF, Visscher TL, Seidell JC. Age standardization in mapping adult overweight and obesity trends in the WHO European Region. Obesity reviews: an official journal of the International Association for the Study of Obesity. 2012;13(2):174–91. pmid:22060052.
- 27. Hilario AL, Oruga JDH, Turqueza MPB, Hilario DV. Utilization of clinical pathway on open appendectomy: A quality improvement initiative in a private hospital in the Philippines. International journal of health sciences. 2018;12(2):43–9. pmid:29599694; PubMed Central PMCID: PMC5870321.
- 28. Baine SO, Kakama A, Mugume M. Development of the Kisiizi hospital health insurance scheme: lessons learned and implications for universal health coverage. BMC health services research. 2018;18(1):455. pmid:29903016.