International Journal for Quality in Health Care 15:433-440 (2003)
© 2003 International Society for Quality in Health Care
Patient-reported quality of life before, compared with after a DRG intervention
Department of Public Health and Caring Sciences, Uppsala University, Uppsala, Sweden
Objective. To evaluate the effects of the introduction of a payment system based on Diagnosis-Related Groups (DRG) on orthopedic patients’ perceptions of quality of life after discharge.
Design. A prospective longitudinal study performed between 1992 and 1996.
Setting. Two surgical clinics in the County of Gävleborg, Sweden.
Study participants. The selection of patients was defined by surgical treatments/diagnoses, and by geographical area. Out of the 40 consecutively recruited patients per year from each hospital, >85% (n = 145 + 140 = 285) completed a questionnaire 1 week post-discharge, and >75% (n = 128 + 127 = 255) did so at a 6-week follow-up.
Main outcome measure. The Swedish Health-Related Quality of Life Survey (SWED-QUAL) was used to assess quality of life. Straight factoral analyses of variance (ANOVAs) were performed to compare years of assessment in the same hospital/district.
Results. The results demonstrated no pattern of significant pre–post differences in patient-reported quality of life. In all years, most quality of life ratings were lower at 1 week compared with 6 weeks after discharge. These differences were probably the consequence of recently undergone surgery.
Conclusion. The overall conclusion is that changes implemented in health care, including the DRGs, had no demonstrable impact on patient ratings of quality of life.
Keywords: Diagnosis-Related Groups, orthopedic patients, patient satisfaction, quality of care, quality of life
Address reprint requests to Birgitta Ljunggren, Department of Caring Science and Sociology, University of Gävle, S-80176 Gävle, Sweden. E-mail: birgitta.ljunggren{at}hig.se
Accepted for publication June 3, 2003.
By the early 90s, some Swedish county councils initiated market-oriented steering mechanisms to meet new demands for cost efficiency [1]. Among these was the County Council of Gävleborg. The roles of purchasers and providers of health care were separated when a new organization, Steer93, with a population-based resource allocation system (in Swedish the ‘Styr93’) was established in 1994 [2,3]. The relationship between the two players was contract-regulated, and the US Diagnosis-Related Groups (DRG) system was used as an output measure for delivered in-patient care.
The DRG system utilizes the International Classification of Diseases, 9th revision, Clinical Modification (ICD 9 CM) in the recording of patient diagnoses. Besides diagnoses (principal and secondary), the DRG classification includes surgical procedures, age, sex, and discharge status [4–6]. Principal diagnoses represent illnesses that would bring patients to hospital care. Secondary diagnoses represent problems that could influence the treatment process, for example comorbidities or post-operative complications [7]. The main difference between the DRG and traditional diagnostic systems is that the DRGs result in only one code per hospital stay [4,8].
When the DRG system was introduced in Sweden as a basis for payment, both hopeful expectations and fears were expressed concerning the use of this system [9]. Potential conflicts were pointed out between the requirement of high quality care, as stated in the Health and Medical Care Act [10], and the increasing awareness of costs among the medical staff [8]. The DRG-based prospective payment system includes the idea of risk and reward. Consequently, hospitals could choose more ‘profitable’ patients and transfer elderly or ‘unprofitable’ patients to other levels of care [11]. Thus, in a system without a comprehensive patient view, the care of the elderly, for example, may deteriorate to a lower level than that of other patient groups [4]. Thus, there is good reason to examine this issue.
A literature search of Medline yielded no study concerning patient experiences of care and quality of life after discharge, performed when a DRG-based payment system was introduced in health care. However, the quality of care given to US Medicare patients before and after the implementation of the DRG-based prospective payment system was evaluated by Kahn et al. [12]. These authors demonstrated increased sickness at admission, increased instability at discharge, and improved nursing care after the implementation of DRG.
In Sweden, effects of the so-called Stockholm model were evaluated by Svensson and Garelius [13]. The aim was to examine how economic incentives (DRGs) had influenced activities in the departments of urology and surgery. However, patient experiences of care and perceived quality of life after discharge were not investigated. The results showed that physicians’ decision making had been affected by the new system. Patients had been discharged after shorter stays, and admitted more frequently to hospital departments than before the introduction of the model. The DRGs also created incentives for physicians to reach a higher degree of precision in their recording of diagnoses and to include secondary diagnoses. Thus, both studies showed negative as well as positive impacts on medical care as an effect of the DRG intervention. However, from the patient perspective, effects in terms of increased sickness at admission, increased instability at discharge, and discharge after shorter stay may have influenced patient satisfaction with care received as well as quality of life at different time-points post-discharge.
The bases for conclusions about quality of care can be classified into three categories: structure, process, and outcome [14–16]. In the present study, the chosen category is the outcome of care in terms of health-related quality of life.
Quality of life has no uniform definition or unique meaning, and is considered both complex and multidimensional [17–19]. During the past two decades, quality of life has become an important attribute often related to patient care [20,21]. Quality of life instruments have been used to study various groups of patients, to predict patient outcome, and to evaluate therapeutic interventions and care [20–22]. Psychological well-being and physical functioning are the dimensions most frequently used to assess quality of life, both in healthy adults and in patients undergoing treatment [22]. In the health care context, the more restricted term ‘health-related quality of life’ is often used [23]. A good ‘health-related quality of life’ implies that the patient is as free as possible from physical or psychological inconveniences, and possesses as good physical, psychological, and social functions as possible in daily living, taking disease and secondary effects of treatment into consideration [23].
The present study is part of a larger project concerned with the effects of a DRG intervention on the quality and outcome of care at two surgical clinics in the County of Gävleborg. In accordance with the above line of reasoning, patient experiences of care received and their quality of life after discharge may change as a result of the DRG intervention. Possibly, such changes could be negative as well as positive, depending upon goals, efforts, and decisions connected to the intervention. The larger project also evaluates which factors, beyond the intervention, may have influenced the health care situation, independently or in interaction with the DRG system.
The aim of the present study was to investigate whether any changes in patient quality of life at 1 and 6 weeks post-discharge could be identified as a result of the DRG intervention, and to evaluate to what extent factors beyond the DRG intervention may have contributed to changes in patient quality of life.
 |
Subjects and methods |
|---|
 Top Subjects and methods ResultsDiscussionConclusionsReferences |
|---|
The design is prospective and longitudinal in that changes in patient quality of life were evaluated for the period 1992–1996. The surgical clinic in one of the hospitals (Hudiksvall) that participated in the project ‘Swedish DRG weights’ during 1994 [24]. The project was expected to start in 1993 and was considered to be the only intervention in the original plan of the present study. Another hospital (Söderhamn) was selected to serve as a control condition. A decision at the central administrative level [25] to use the DRG system and to involve all surgical clinics, starting in January 1995, made it impossible to maintain the original quasi-experimental design. Therefore, it was not considered meaningful to make statistical comparisons between hospitals/districts. In addition, from the autumn of 1995, a hospital merger was initiated between the control hospital and another hospital (Bollnäs). However, the delay in DRG implementation made it possible to perform more than one pre-test. Furthermore, as the study became longitudinal with a duration of 4–5 years, the opportunity to examine trends and changes over time in each hospital increased. The hospital managers and the head physicians at the participating hospitals were informed about the purposes of the study and gave their permission. The study was also approved by the Research Ethics Committee of the Faculty of Medicine at Uppsala University.
The design of the study is presented in detail in Figure 1.
|
In Figure 1, O1 and O2 represent the two pre-tests. In the North Hälsingland district (Hudiksvall hospital), O3 represents a test after the hospital’s participation in the ‘Swedish DRG weights’ project in the spring of 1994 (x). ‘X’ denotes the DRG intervention in both hospitals in 1995 and 1996. In the Söderhamn district, O3 represents an additional pre-test. During 1992–1994, each data collection period (including the 6-week follow-up) started in the autumn and lasted until February in North Hälsingland and until April the following year in the Söderhamn district. When the hospital merger between the Söderhamn and Bollnäs hospitals was initiated in 1995, data collection became impracticable until the activities were settled in the new hospital organization. In addition, a nurses’ strike at the end of the year prevented further assessment. For these reasons, the post-test (O4) was delayed until February 1996.
Instrument
The questionnaire comprised two parts: part I examined patient satisfaction with care received [26] and part II concerned patient quality of life. The ‘Swedish Health-Related Quality of Life Survey’ (SWED-QUAL) [27] was used to assess quality of life as perceived by the patients.
The SWED-QUAL contains 62 Likert scale items (11 multi-item scales and two single items), covering the following aspects of health-related quality of life (HRQoL): physical functioning, mobility, satisfaction with physical health, role limitations due to physical and emotional health problems, pain, emotional well-being (positive and negative effect), sleep, satisfaction with family life, marital functioning, sexual functioning, and general health perception. It also includes sociodemographic items concerned with age, sex, marital status, education, and income.
All items except the sociodemographic ones employ a ‘now’ or ‘during the last week’ time window. The item scores are transformed to scale scores in two steps: (1) the item scores are transformed linearly to a 0–100 possible range (0 = lowest, 100 = highest possible score); and (2) items measuring the same HRQoL dimension are summed and an average score is calculated for each dimension. In all dimensions (negative as well as positive aspects of HRQoL), high scores indicate a more favourable state of health. The response alternative ‘not applicable’ is coded as missing data.
The SWED-QUAL has been adapted from measures used in the Medical Outcome Study (MOS) [28]. Validity and reliability of the original version have been extensively tested, and the measures cover a spectrum of health states [28]. The SWED-QUAL has been tested in a random sample of individuals (n = 2366) from the Swedish population. The tests conducted support the construct validity of the SWED-QUAL, and Cronbach’s alpha exceeded 0.78 for all scales [27].
Patients
The purpose of the study was to recruit, at each assessment, new patient samples that were as homogeneous as possible in order to facilitate comparisons. The inclusion criteria were: (1) patients aged 18–85 years, cared for in the surgical clinic at any of the selected hospitals; (2) patients suffering from hip, femur, or knee fracture, or admitted for hip or knee replacement; (3) patients living in the North Hälsingland or Söderhamn districts; and (4) patients discharged to live at home. An information letter and a consent form were mailed together with a self-administered questionnaire to the patients’ homes 3–5 days after discharge. Of the 40 consecutively recruited patients per year discharged from each hospital, >85% completed the first questionnaire, and >75% completed the questionnaire at the 6-week follow-up. Medical data concerning patients’ diagnoses/treatments were collected from the medical records. Sample characteristics are presented in Table 1. There were no significant differences (degrees of freedom (df) = 3, chi-square tests [29]) between the years of assessment within districts with regard to sex, marital status, or diagnoses/treatments. However, analyses of variance (ANOVA) with Fisher’s Protected Least Significant Difference (PLSD) post hoc tests (P < 0.05) [29] demonstrated that patients in the North Hälsingland district in 1992 and 1993 were significantly older than patients in 1996 (1 week: F = 2.85, df = 3/141, P < 0.05), and patients in the Söderhamn district in 1992 and 1993 were significantly younger than patients in 1996 (6 weeks: F = 2.93, df = 3/123, P < 0.05).
|
Ten females and five males from the North Hälsingland district dropped out of the study, as did 11 females and nine males from the Söderhamn district. This included patients suffering from both hip, femur, and knee fractures, and knee and hip replacements. Their mean ages were 70 and 71 years, respectively.
Statistical methods
Straight factorial ANOVAs with PLSD post hoc tests (P < 0.05) were performed to compare years of assessment in the same hospital/district [29]. Analyses were performed separately for the 1- and 6-week assessments. Paired (two-tailed) student’s t-tests were used to compare the 1- and 6-week assessments summed over years within groups.
In the text below, patients discharged from the Söderhamn and the Bollnäs-Söderhamn hospitals will be denoted ‘Söderhamn district’, and patients discharged from the Hudiksvall hospital will be denoted ‘North Hälsingland district’.
|
Results |
|---|
|
TopSubjects and methodsResultsDiscussionConclusionsReferences |
|---|
Health-related quality of life
Table 2 presents the quality of life data for each year and point of assessment for patients in both hospitals/districts.
|
ANOVAs demonstrated significant differences only in the Söderhamn district. Patients in 1992 (1 week) scored significantly higher with respect to emotional well-being, negative affect, than patients in 1993 and 1996, and 1994 patients scored higher than 1993 patients (F = 3.02, df = 3/122, P < 0.03). Patients in 1994 scored significantly higher than 1993 and 1996 patients with respect to role limitations due to emotional health (1 week: F = 2.90, df = 3/133, P < 0.04). Patients in 1996 scored significantly higher than patients in 1993 and 1994 with respect to mobility (6 weeks: F = 2.94, df = 3/122, P < 0.04). Thus, ANOVAs revealed no consistent pattern of results during the study period.
It should be noted that in 1993, patients in the Söderhamn district obtained the lowest mean values of all years in that district on seven out of 13 dimensions at 1 week, and on 10 out of 13 dimensions at 6 weeks. In the North Hälsingland district, 1992 patients obtained the lowest mean values of all years in that district with respect to 10 out of 13 quality of life dimensions (1 week). The demographic and medical characteristics (Table 1) did not differ sufficiently between 1993 (in Söderhamn) and 1992 (in North Hälsingland) and the remaining years to explain the low 1993 and 1992 quality of life values. The only significant difference between the years of assessment was found for age. Patients in the North Hälsingland district in 1992 and 1993 were significantly older than patients in 1996, and patients in the Söderhamn district in 1992 and 1993 were significantly younger than patients in 1996. Thus, sample characteristics seem to have no relation to patient perception of quality of life dimensions.
The low week 1 mean values (Table 2, all years) were presumably the consequence of recently undergone surgery, and of the restrictions imposed in connection to discharge with respect to the appropriate movements and exercise. Thus, the 6-week assessments showed consistently higher scores, especially on the subscales pain, sleep problems, physical functioning, mobility, satisfaction with physical health, and role limitation due to physical health.
Paired (two-tailed) t-tests were performed to compare quality of life at 1 and 6 weeks post-discharge (Table 3). The t-tests demonstrated significantly higher scores in week 6 than week 1 for seven out of 13 dimensions in the North Hälsingland district, and for 10 out of 13 in the Söderhamn district. The only exception was noted for patients in the Söderhamn district with respect to marital functioning, which was rated higher in week 1 than in week 6.
|
|
Discussion |
|---|
|
TopSubjects and methodsResultsDiscussionConclusionsReferences |
|---|
The main focus of this study concerns the effects of the DRG intervention on patient quality of life 1 and 6 weeks after discharge at two surgical clinics in the County of Gävleborg in Sweden. ANOVAs demonstrated a few significant differences between the years of assessment, all of them in the Söderhamn district, but there was no consistent pattern related to the DRG intervention. We have previously reported patient experiences of received care in the same patient group [26]. The results demonstrated a pattern of successively decreasing mean values during the study period, with significant pre–post differences, especially with respect to treatment by staff.
A notable change in patient quality of life after discharge is indicated by the low initial mean values (all years at 1 week) compared with those of the 6-week assessment, especially in dimensions related to physical functioning. These low initial scores were almost equal in all years, and probably a consequence of recently undergone surgery and restrictions imposed by the physician at discharge. Thus, there were significant differences between the 1- and 6-week assessments for seven out of 13 dimensions in the North Hälsingland district, and 10 out of 13 in Söderhamn, with higher mean values in the 6-week than in the 1-week assessments. This pattern of results failed to parallel the quality of care data reported earlier [26], where most means were higher at 1 week compared with those at 6 weeks post-discharge. In the quality of care assessment, paired t-tests demonstrated significant differences for 11 of the 18 items in the North Hälsingland district, and for six in Söderhamn. Most of the significant differences (seven out of 11 in North Hälsingland, and five out of six in Söderhamn) concerned treatment by staff [26].
Thus, patient experiences of received care and their quality of life after discharge did not show the same patterns. Changes implemented in health care in the County of Gävleborg between 1992 and 1996 seem to have negatively influenced patient experiences of received care, but not quality of life.
These negative as well as positive findings could not be attributed entirely to the DRG intervention, however. According to documents, the goals of using DRG on local levels were: method of measurement, basis for payment, and quality and follow-up system [25]. The explanatory analyses performed demonstrated that previously stated goals of using DRGs were only partially fulfilled, and that the DRG intervention was implemented gradually. A DRG pricing was employed, and the DRG system was used as a measure of in-hospital somatic care. The effects caused by the DRG intervention, besides those constituting DRG goal fulfilment, were difficult to separate from the possible influence of co-operative factors, such as organizational and structural changes within the hospitals, quality assurance activities, and the new market-oriented steering organization Steer93 [30].
Steer93 was seen as a way to meet new demands for higher effectiveness necessitated by limited resources in the county [25]. The impact of the Steer93 organization on health care performance was assumed to be mostly of a negative character in terms of increasing demands for efficiency and cost-awareness among staff [30]. However, the split between purchasers and providers of health care resulted in a more customer-oriented concentration of activities. The local boards were made responsible for provision of the best possible care to their patients and citizens, and quality assurance activities were emphasized [30].
In the present study, patient experiences of perceived quality of life were monitored over a 5-year period. In 1997, a new proposal and later a political decision were made in the County Council of Gävleborg with respect to an additional hospital merger in Hälsingland. These circumstances prevented further measurement and continuation of the present study. Future studies of the effects of health care reforms would yield more distinct conclusions if it was possible to separate effects of a single intervention from the impact of external factors, as well as to evaluate the intervention before full-scale implementation in the health care arena.
No a priori power calculations were performed to determine appropriate group sizes. Retrospective calculations show, however, that power was quite sufficient for the week 1 versus week 6 comparisons, and for the comparisons of patients’ quality of care ratings between the years of assessment in the previous report on the same study [26]. However, with regard to the between-year comparisons of quality of life, two to four times more patients would have been required to achieve statistical significance. This is mainly a reflection of the fact that changes in quality of life over the years were quite small.
The extent to which the findings in the present study can be generalized to other contexts is debatable. The aim was to find comparable groups and settings for each year of assessment. This was done with the purpose of investigating whether any observed effect was caused by the DRG intervention in the specific hospital or district investigated. Thus, quite homogeneous groups and settings were studied. Besides the DRGs, most of the external factors evaluated concern a number of changes in health care, for example organizational changes, quality assurance activities, and restricted resources that have also occurred in other health care organizations lately. For this reason, it may be assumed that the findings are also relevant to other subjects and settings in contemporary health care sectors.
|
Conclusions |
|---|
|
TopSubjects and methodsResultsDiscussionConclusionsReferences |
|---|
In all study years, most quality of life ratings were lower at 1 week compared with those at 6 weeks after discharge. These differences were presumably the consequence of recently undergone surgery and restrictions imposed in connection with discharge at 1 week. Furthermore, patient ratings of quality of life did not parallel their ratings of quality of care. There was no pattern of successively decreasing mean values with significant pre–post differences. Thus, the changes implemented in health care including the DRG intervention had no demonstrable impact on patient quality of life.
This work was supported by grants from the Faculty of Social Sciences, Uppsala University, Sweden, and from the Research and Development Department, County Council of Gävleborg, Sweden.
|
References |
|---|
|
TopSubjects and methodsResultsDiscussionConclusionsReferences |
|---|
- Anell A, Rosén P, Svarvar P. Health Care Reforms in Sweden—Striving for Equity, Efficiency and Containment. Working Paper 8. Lund: The Swedish Institute for Health Economic IHE, 1996, pp. 9–23.
- County Council of Gävleborg. Annual Report 1993 (in Swedish). Gävle: Landstinget Gävleborg, 1994, pp. 13–15.
- County Council of Gävleborg. Annual Report 1994 (in Swedish). Gävle: Landstinget Gävleborg, 1995, pp. 6–7.
- Andrén-Sandberg
. Diagnosis Related Groups, Background, Practical Problems, Quality Controls, and Ethics (in Swedish). Stockholm: Liber Utbildning AB, 1994, pp. 7–33. - Averill R. Development. In: Fetter R, ed. DRG. Their Design and Development. Ann Arbor, MI: Health Administration Press, 1991, pp. 28–42.
- Swedish Institute for Development of Health Services (Spri). DRG in Sweden. Adaptation and Registration Studies (in Swedish). Spri-report 281. Stockholm: Spri, 1990, pp. 8–19.
- Fetter R. Background. In: Fetter R, ed. DRG. Their Design and Development. Ann Arbor, MI: Health Administration Press, 1991, pp. 3–7.
- Swedish Institute for Development of Health Services (Spri). What Does the Patient Cost? (in Swedish). Spri-report 262. Stockholm: Spri, 1989, pp. 8–25.
- Berleen G, Rehnberg C, Wennström G. The Reforms of Health Care in Sweden. Stockholm: Spri, 1992, p. 59.
- Swedish Statute-book. The Health and Medical Care Act (in Swedish). Stockholm: The Ministry for Social Affairs, SFS, 1982, p. 763.
- Davis C, Rhodes D. The impact of DRGs on the cost and quality of health care in the United States. Health Policy 1988; 9: 117–131.[CrossRef][Web of Science][Medline]
- Kahn K, Keeler E, Sherwood M et al. Comparing outcomes of care before and after implementation of the DRG-based Prospective Payment System. J Am Med Assoc 1990; 264: 1984–1988.
[Abstract/Free Full Text] - Svensson H, Garelius L. Have Economic Incentives Affected Physicians in their Decision Making. An Evaluation of the Stockholm model (in Swedish). Stockholm: Spri, 1994, pp. 5–110.
- Donabedian A. Some issues in evaluating the quality of nursing care. Am J Public Health 1969; 59: 1833–1836.
- Donabedian A. The quality of medical care. Science 1978; 200: 856–864.
[Abstract/Free Full Text] - Donabedian A. The quality of medical care. How can it be assessed? J Am Med Assoc 1988; 260: 1743–1748.
[Abstract/Free Full Text] - Naess S. The concept of quality of life. In: Björk S, Vang J, eds. Assessing Quality of Life. Linköping: Linköping University, 1989, pp. 9–16.
- Sullivan M. Quality of life measurement in clinical research—superior than traditional measures of the outcome of care (in Swedish). Läkartidningen 1990; 87: 1369–1372.[Medline]
- Zhan L. Quality of life: conceptual and measurement issues. J Adv Nurs 1992; 17: 795–800.[Web of Science][Medline]
- Gill T, Feinstein A. A critical appraisal of the quality of quality of life measurement. J Am Med Assoc 1994; 272: 619–625.
[Abstract/Free Full Text] - Sjödén PO, Bates S. Is it Possible to Assess Quality of Life? (in Swedish). Uppsala: Department of Clinical Psychology, Uppsala University, 1988, pp. 1–2.
- Goodinson S, Singleton J. Quality of life: a critical review of current concepts, measures and their clinical implications. Int J Nurs Stud 1989; 26: 327–341.[CrossRef][Web of Science][Medline]
- Sjödén PO. Quality of life measurement in cancer research and care (in Swedish). In: Ploug Hansen H, Tishelman C, eds. Complex and Varied. An Anthology about Swedish-Danish Psychosocial Cancer Care. Lund: Studentlitteratur, 1997, pp. 20–35.
- Weibull H, Andrée R, Eklund A, Fält B, Schnell CR, Zetterberg C. The CPP (Cost PER Patient) project (in Swedish). Stockholm: Spri, 1996, pp. 7–63.
- County Council of Gävleborg. Office of the County Council of Gävleborg 1994-07-13 (in Swedish). Gävle: Landstinget Gävleborg, 1994, pp. 1–5.
- Ljunggren B, Sjödén PO. Patient reported quality of care before vs after the implementation of a Diagnosis Related Groups (DRG) classification and payment system in one Swedish county. Scan J Caring Sci 2001; 15: 283–294.[CrossRef]
- Brorson B, Ifver J, Hays RD. The Swedish health-related quality of life survey (SWED-QUAL). Qual life Res 1993; 2: 33–45.[CrossRef][Web of Science][Medline]
- Stewart AL, Ware JE. Measuring Functioning and Well-being: The Medical Outcomes Study Approach. Durham, NC: Duke University Press, 1992.
- Howell D. Statistical Methods for Psychology. Third edition. Belmont: Duxbury Press, 1992, pp. 123–356.
- Ljunggren B. Quality of Health Care: Patient and Staff Perspectives. A Study of Changes 1992–1996 with Special Reference to Diagnosis Related Groups. Uppsala: Doctoral Dissertation, Department of Public Health and Caring Sciences, Uppsala University, 1998, pp. 169–184.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||