PATIENT SATISFACTION WITH AND RECOMMENDATION OF A HOSPITAL: EFFECTS OF INTERPERSONAL AND TECHNICAL ASPECTS OF HOSPITAL CARE

全文

(1)International Journal for Quality in Health Care 2003; Volume 15, Number 4: pp. 345±355. 10.1093/intqhc/mzg045. Patient satisfaction with and recommendation of a hospital: effects of interpersonal and technical aspects of hospital care SHOU-HSIA CHENG1, MING-CHIN YANG2 AND TUNG-LIANG CHIANG1 1. Institute of Health Policy & Management and Center for Health Policy Research, National Taiwan University, College of Public Health, Taipei, Taiwan, 15th Floor, No. 1, Section 1, Jen-Ai Road, Taipei, 2Institute of Health Care Organization Management and Center for Health Policy Research, National Taiwan University, College of Public Health, Taipei, Taiwan, Room 1518, No. 1, Section 1, Jen-Ai Road, Taipei, Taiwan, 100. Abstract Objectives. To examine patient satisfaction with and recommendation of a hospital, with a special focus on the correlation of these measures to patient ratings of interpersonal and technical performance of the hospital. Design. Telephone survey of patients with four specific conditions after their discharge from hospitals. Setting. Accredited district teaching hospitals and above, nationwide in Taiwan. Participants. A total of 4945 patients from 126 hospitals diagnosed with or undergoing procedures related to stroke, diabetes mellitus, Caesarean section, or appendectomy were interviewed by telephone. Main outcome measures. Overall patient satisfaction and recommendation were measured by single-item questions. Interpersonal skills were measured by three items: doctors' explanation, attitude, and caring. Technical skills were measured by another three items: hospital equipment, clinical competence, and outcome of treatment. Results. Interpersonal skills were as influential or more influential than clinical competence on patient satisfaction for three of the four disease categories. In contrast, technical competence was a more influential predictor for recommendation for patients in all four disease categories. Conclusion. The preliminary results imply that a hospital with high percentage of patient satisfaction does not necessarily receive a high level of recommendation. This finding provides new insights for researchers and for hospital managers who devote resources exclusively for achieving the highest possible levels of patient satisfaction. Keywords: patient judgment of hospital quality, patient ratings, patient satisfaction, quality of care, recommendation, Taiwan. Introduction Patient satisfaction has been an important issue for health care managers. Many previous studies have developed and applied patient satisfaction as a quality improvement tool for health care providers [1----3]. Following increased levels of competition and the emphasis on consumerism, patient satisfaction has become an important measurement for monitoring health care performance of health plans [4]. This measurement has developed along with a new feature: the patient's perspective of quality of care [5----8].. Various dimensions of patient satisfaction have been identified, ranging from admission to discharge services, as well as from medical care to interpersonal communication. Wellrecognized criteria include responsiveness, communication, attitude, clinical skill, comforting skill, amenities, food services, etc. [9----12]. It has also been reported that the interpersonal and technical skills of health care provider are two unique dimensions involved in patient assessment of hospital care [13,14]. Determinants of patient satisfaction have been reported extensively. According to previous studies, patient. Address reprint requests to Shou-Hsia Cheng, Associate Professor, Institute of Health Policy and Management, NTU, Room 1515, No. 1, Section 1, Jen-Ai Road, Taipei, Taiwan, 100. E-mail: shcheng@ha.mc.ntu.edu.tw ã 2003 International Society for Quality in Health Care and Oxford University Press. 345.

(2) S.-H. Cheng et al.. characteristics such as age and education may influence a patient's assessment of hospital performance [8,15,16]. A patient's health status and the severity of illness are also important predictors of the patient's overall satisfaction level [7,11,15,17,18]. Hospital features such as hospital size have been reported to be associated with consumer assessment of hospital quality [1,8,10,11,19]. The relationship between health care providers and patients (i.e. interpersonal skill) has been reported to be the most influential factor for patient satisfaction [8,17]; however, the findings were not totally conclusive [6]. A lack of available information for consumers to choose a preferred health care provider is commonplace in many countries. We also know that personal channels of communication with relatives and friends are a major source of information for people wishing to obtain details concerning hospital performance [20,21]. Therefore, recommendations from family or friends become an important source of information for selecting health care providers. Recommendation as well as satisfaction is based on personal experience concerning the services that one has received from health providers [9]. A number of previous reports included recommendation and satisfaction items to form an overall assessment of patient satisfaction, but no differentiation between these two measures was discussed [14,22,23]. Recent studies suggest that patient recommendation of a hospital should be analyzed separately (see below). Burroughs and colleagues [3] reported that, of the different settings of health care services (i.e. inpatient care, outpatient surgery, outpatient test/treatment, and emergency care), compassionate patient care, including personal attention, respect, comfort, and the like, were the most important factors influencing a patient's intention to recommend a provider or for that patient to return to the same provider in the future. Their findings are in agreement with previous reports indicating that patient satisfaction is determined mainly by a provider's attitude and caring rather than technical skills [8,17]. Nevertheless, they also found that physician care delivery (such as skill, frequency of checking, and explanation) was not associated with patient recommendation or return, which is not a persuasive finding. Recently, Boudreaux and colleagues [24] analyzed a small sample of 437 emergency department patients and reported that overall patient satisfaction and the likelihood of recommendation were influenced by somewhat different factors. They found that respect, safety, and understandable instructions were common factors associated with satisfaction and recommendation. On the other hand, the technical skills of nurses and waiting time were associated with recommendation only, and patient's age and insurance status were associated with satisfaction only. These findings may not be generalized to other patients due to its specific setting and small number of participants. To investigate the factors associated with patient recommendation and satisfaction respectively, the study is based on a sample of 4945 people who were recently discharged with specific diagnoses or conditions in order to control possible confounding effects due to variation of disease combinations. 346. [15]. Factors associated with hospital care were intentionally stratified into two important dimensions------interpersonal skills and clinical competence------to facilitate easier understanding and interpretation. Data were collected in Taiwan, where the hospital care market is highly competitive and patients can choose any hospital to receive care under a universal health insurance scheme. With a high degree of freedom for patients to choose among hospitals, personal channels of recommendation are very influential and thus provide an excellent opportunity to examine any potential differences between patient satisfaction and recommendation of a hospital. Health care services in Taiwan The health care service market in Taiwan is highly competitive. After implementation of the National Health Insurance (NHI) program in 1995, the situation of competition became more intense. Small-scale hospitals were driven out of the market. The number of hospitals decreased from 828 in 1994 to 669 in 2000; however, the number of hospital beds nationwide increased 21.9% from 103 733 to 126 476 during the same period [25]. Approximately 96% of all citizens in Taiwan (about 23 million) are covered under the NHI, and more than 95% of the hospitals nationwide are NHI-contracted providers. Without a formal referral requirement, people are free to choose any contracted hospitals they wish to receive health care services. Since the Bureau of the National Health Insurance (Bureau of NHI) is the only buyer (or insurer) of health care services, it sets payment schemes for all contracted providers for various kinds of services. Therefore, health care providers have to compete for patients under a mostly fee-for-service reimbursement system with nearly fixed payment schemes. Detailed descriptions of the implementation and influence of the nationwide health insurance program are available elsewhere [26----28]. The quality of health care varies significantly among hospitals in Taiwan. Almost all of the hospitals are closed systems in that hospitals hire physicians to provide outpatient and inpatient care within the hospitals. While a medical-specialist training program was established in 1988, approximately 30% of practicing physicians have not as yet undergone any form of specialist training [29]. The hospital accreditation system, which was started in 1978 by Taiwan's Ministry of Education and Department of Health, issues levels of the accreditation on the basis of a hospital's capabilities and quality of performance as determined by an expert team's on-site inspection. There are four levels to this accreditation system which, in descending order, are medical center hospital, regional hospital, district teaching hospital, and district hospital [30]. Generally speaking, the level of accreditation is positively correlated with the bed size (which also represents comprehensive ability in subspecialties) of a hospital. Among the 669 hospitals nationwide in 2000, there were 20 medical centers and 63 regional hospitals, with most of these hospitals being large-scale with 500----2000 beds and possessing advanced clinical diagnostic and therapeutic equipment. In spite of the accreditation system, in 2000 there were 163 of the 669 hospitals without accreditation, with most of these being.

(3) Patient satisfaction and recommendation. small-sized hospitals with fewer than 50 beds [31]. In Taiwan, a hospital's reputation (or level of people's recommendation) or its bed size was nearly the only information available to prospective patients or their family members in order to select a preferred hospital.. Methods Sources of subjects Since the diagnostic and treatment procedures might vary significantly among different diseases, we decided to select patients with certain diagnoses/procedures for the study. After consulting medical professionals and considering the prevalence of the diagnoses for case collection, we chose two medical diagnoses, stroke and diabetes mellitus (DM), and two surgical procedures, Caesarean section (C/S) and appendectomy, as our study conditions. This study only included patients discharged from the hospitals with accredited district teaching hospital status or above (approximately 130 hospitals account for more than 90% of all inpatient services nationwide). The aim within this project was to gather 5000 patients for analysis with no less than 1000 cases for each condition. Sample selection was conducted with the help of the Bureau of NHI. All of the NHI-contracted hospitals file claims to one of the six regional NHI branches each month. The branch offices selected discharged patients claim data for routine posterior utilization review every month. For those diagnoses paid by fee-for-services system (i.e. stroke and DM in our study), the method of selecting subjects for routine review is random-based, with a predetermined proportion to oversample smaller hospitals, with a range between 2 and 10%. For those case-payment diagnoses, 10% of the C/S patients and 15% of the appendectomy patient were randomly selected. Accordingly, during the study period, the Bureau of NHI identified candidate subjects with the abovementioned conditions and then sent the list of patients to our research team as our sampling frame. We performed a simple random sampling with appropriate proportions for each of the four disease categories (ranging between 33 and 100%) to obtain the required number of cases for the survey. The study period of the project extended from June 1999 to March 2000. Patients who expired after discharge from the hospitals or patients with more than one diagnosis of the four disease categories were excluded from the study. Questionnaire design A structured questionnaire was designed to examine several aspects of hospital care including out-of-pocket payment, medical and nursing care, and post-acute care utilization, etc. The questionnaire was standardized via a small-scaled pilot test on 80 discharged patients. A mass telephone survey was carried out by employees from the six branches of the Bureau of NHI. Explanation about the design of the questionnaire and instruction on standard rules in conducting the interviews were given to. Discharged patients or their primary family caregivers were asked questions concerning their. experience during that specific hospitalization period. The beginning of the questionnaire identified the patient's diagnosis for the study. Patients were contacted and interviewed approximately 3 months after their discharge. Two dependent variables were used in this study to indicate the patients' perception of overall hospital quality: satisfaction and recommendation. The question measuring patient satisfaction was stated as `Generally speaking, were you satisfied with the inpatient care?' Response categories were `satisfied', `fair', or `not satisfied'. When analyzing this variable, we dichotomized the responses by treating `satisfied' as one group and the remaining two responses as the other in a logistic regression model. The question for patient recommendation was Ìf someone asks you about that hospital, would you recommend it?' Response categories were `strongly recommend', `recommend', `conservatively recommend', `not recommend', and `definitely not recommend'. In a logistic regression model, the first two responses (strongly recommend and recommend) were defined as one group and the remaining three responses as the other. Using a 3-point scale for `satisfaction' and a 5-point scale for `recommendation' was a limitation of the study (see Appendix). Multiple dimensions of hospital quality of care have been identified in previous studies [10----12]. However, due to the constraints of a multi-purpose questionnaire administered by telephone survey, only a limited number of questions can be employed for measuring the quality of care. After reviewing several measurement tools developed by previous researchers and considering patients' ability to answer the questions, we selected six items to measure two major dimensions of quality of care: the clinical competence of hospital and the interpersonal skills of physician. Clinical competence included three items: hospital equipment, physician competence, and treatment outcome or recovery. Inter-personal skills also included three items: doctor's attitude, doctor's caring manner, and doctor's explanation of illness or condition. Five-level response categories were designed respectively for each of the six questions, such as `definitely enough' to òbviously insufficient', `very good' to `very poor', or `much better than expected' to `much worse than expected'. Scores ranging from 5 to 1 were assigned to the five categories of answers, the three items for each dimension summed to form the clinical competence and inter-personal skills scores, respectively, which ranged from 15 down to 3. Higher scores were representative of better-perceived quality of hospital performance. Missing values in each of the six items (from 0.4 to 13.3% of the total number of people surveyed in each category) were replaced by the mean scores for each of the four diagnoses in order to maximize the amount of useful information. Patients with different conditions were analyzed separately, as suggested by Hargraves et al. [15] to account for possible clinical variations. Statistical analysis One-way ANOVA was used in the study to examine the associations between categorical variables, such as levels of satisfaction or recommendation, and continuous scores such. 347.

(4) S.-H. Cheng et al.. Table 1 Basic information for the discharged patients interviewed grouped by the four conditions Caesarean section Appendectomy DM Stroke Total ............................................................................................................................................................................................................................. n 1704 787 1163 1291 4945 Gender (% female) 100 45.2 54.4 42.8 65.6 Age (mean sd) 30.5 14.3 31.1 13.0 63.7 4.9 68.0 16.5 48.2 21.4 Age group, years (%) 14 n.a. 14.8 0.6 0.2 2.6 15----24 11.6 25.7 1.1 1.2 8.4 25----34 67.7 22.9 1.6 0.7 27.6 35----44 20.6 16.7 5.6 3.3 12.0 45----54 0.2 11.3 13.9 11.6 8.2 55----64 n.a. 4.2 23.4 18.6 11.0 65----74 n.a. 2.9 30.2 31.8 15.8 75 n.a. 1.4 23.2 33.7 14.5 Education (%) Illiteracy 0.5 7.5 40.1 41.7 21.7 1----6 years (primary) 3.0 17.9 36.5 36.6 22.0 7----9 years ( junior high) 9.7 20.8 9.9 7.4 10.9 10----12 years (high school) 49.2 34.6 9.5 9.2 27.1 13 years (college) 37.5 18.8 3.7 4.7 18.0 Hospital accreditation level1 (%) Medical center 15.7 16.8 17.3 19.1 17.1 Regional hospital 44.0 48.5 49.2 49.1 47.3 District teaching hospital 40.3 34.7 33.5 31.8 35.6 Interviewee status2 85.8 52.9 44.4 32.9 57.0 DM, diabetes mellitus; n.a., not applicable. 1 Hospital accreditation level is issued by Taiwan's Department of Health based on expert evaluation. Advanced equipment and comprehensiveness in medical specialties are important factors. Medical center is the top level and district teaching hospital is the lowest in this study. 2 Interviewee status shows the percentage of sampled patients who answered the questionnaire by themselves instead of by their caregivers or relatives.. as patient ratings of hospital performances. Spearman correlation analysis was used to test the association between rank scores. Multiple logistic regression models were employed to examine the effects of interpersonal and technical aspects of patient ratings on dichotomized categories of satisfaction and recommendation. Standardized regression coefficients were used to compare the magnitudes of the effects. The analyses were performed by using the SAS system version 8.0.. Results Background information on study population A total of 4945 patients or primary caregivers were interviewed successfully with a response rate of 67.0%. Basic information concerning the sample population is shown in Table 1. Owing to the differences of disease prevalence and the differences in the six Bureau of NHI branches' sampling mechanism, case numbers for the four disease categories varied largely. C/S was the major procedure performed, with 1704 patients in this category, while appendectomy was 348. the least performed procedure with 787 patients. There were fewer females with appendectomy and stroke (45.2 and 42.8%, respectively). Age distribution was significantly different between patients with medical or surgical diagnoses. The mean age was around 31 years (sd 13----14) for patients with C/S or appendectomy, while the figures were approximately 64 (sd 4.9) and 68 (sd 16.5) for DM and stroke patients, respectively. There were also differences in levels of patient education. The majority of patients with C/S were high school or college graduates. The educational level of patients with appendectomy was more evenly distributed, with 34.6% of them being high school graduates. In contrast, the majority (about 76----78%) of patients with the two medical conditions were illiterate or received less than 6 years of schooling. The difference in education was due to the cohort effect along with the economic development in Taiwan after World War II. The essential differences in age and education between medical and surgical patients suggested that further analyses should take these variations into consideration. Patients were discharged from 126 hospitals. According to the hospital accreditation levels, approximately 47.3% of all.

(5) Patient satisfaction and recommendation. Table 2 Study subjects' satisfaction ratings and recommendations of hospital quality by procedure/condition group Items Caesarean section Appendectomy DM Stroke Total ............................................................................................................................................................................................................................. n 1704 787 1163 1291 4945 Satisfaction with care (%) Satisfied 74.1 63.3 58.1 57.4 64.2 Fair 23.2 31.8 34.9 36.2 30.7 Not satisfied 2.1 3.7 5.1 3.9 3.5 No answer 0.7 1.3 1.9 2.6 1.6 Recommend the hospital (%) Strongly recommend 11.7 8.3 6.8 8.3 9.1 Recommend 69.4 60.6 59.1 57.2 62.4 Conservatively recommend 5.6 5.8 7.5 7.5 6.6 Not recommend 3.4 5.8 4.6 3.9 4.2 Definitely not recommend 0.6 1.9 1.7 1.7 1.4 No answer 9.3 17.5 20.3 21.5 16.4 DM, diabetes mellitus.. Table 3 Patients' ratings (mean sd) for summarized and individual items of hospital quality Caesarean section Appendectomy DM Stroke Total Items1 ............................................................................................................................................................................................................................. n 1704 787 1163 1291 4945 Interpersonal skills 12.9 1.8 12.4 1.9 12.3 1.9 12.3 1.9 12.5 1.9 Doctor's attitude 4.2 1.0 4.2 1.0 4.0 0.9 4.1 1.0 4.1 1.0 Explanation of condition 4.5 0.6 4.2 0.7 4.2 0.7 4.2 0.7 4.3 0.7 Doctor's level of care 4.2 0.7 4.1 0.7 4.1 0.7 4.1 0.7 4.1 0.7 Technical skills 12.3 1.6 11.8 1.8 11.4 1.8 11.5 1.9 11.8 1.8 Doctor's competence 4.2 0.6 4.0 0.8 3.9 0.8 3.9 0.7 4.0 0.7 Hospital's equipment 4.1 0.9 3.9 1.0 3.7 0.9 3.7 0.9 3.9 1.0 Outcome/recovery 4.0 0.7 3.9 0.8 3.8 0.7 3.8 0.8 3.9 0.7 DM, diabetes mellitus. 1 Patient ratings for the individual items ranged from 1 to 5 and for these items grouped as summarized items (i.e. ìnterpersonal skills' and `technical skills' ) ranged from 3 to 15, with higher scores representing better performance.. patients were discharged from regional hospitals, 35.6% from district teaching hospitals, and 17.1% from medical centers. The proportion of patients discharged from regional hospitals was higher than that of the national average (47.3 versus 33.7%), while the proportions of the two other levels showed the opposite situation. Patient distribution among the three levels of hospitals also varied slightly for the four diagnoses. A total of 57.0% of the patients responded to the interview by themselves; the status of interviewees differed significantly among the four diagnoses. Univariate analysis for satisfaction and recommendation Results of patient perceptions of hospital quality are listed in Table 2. Satisfaction measurement showed that 64.2% of patients were satisfied with the hospital care they received, with C/S patients having the highest percentage (74.1%) of satisfaction. The overall proportion of hospital. recommendation (strongly recommend and recommend) was 71.5%, with C/S patients having the highest rate of recommendation (81.1%) and patients with stroke or DM the lowest rates (65.5----65.9%). We also noted that there were more missing values to the recommendation question than to the satisfaction question. The overall proportion of `no answer' response for the recommendation question was 16.4%, while only 1.6% responded with `no answer' for the satisfaction question. We tested the association between the `no answer' response to the recommendation question and patient characteristics such as age, sex, and education according to the four diagnoses and found no significant differences. Patient ratings of hospital quality The mean scores of the summary and individual items for hospital quality measurement are presented in Table 3 according to the four diagnoses. We conducted Spearman correlation analysis to examine the internal consistency within 349.

(6) S.-H. Cheng et al.. the two performance dimensions. The correlation coefficients for the three interpersonal skill items ranged between 0.41 and 0.63; while that for the technical skill items ranged between 0.28 and 0.39. All the correlation coefficients reached a statistical significance level of P 5 0.001. Generally speaking, the scores for physician's interpersonal skills (12.5 1.9 in total) were higher than that for the hospital technical skills (11.8 1.8 in total). For the six individual items, physicians' explanations of conditions had the highest scores, which ranged from 4.2 0.7 to 4.5 0.6, with an average of 4.3 0.7. The perceived recovery had the lowest scores, which ranged from 3.8 0.7 to 4.0 0.7, with an average of 3.9 0.7. Variations between groups of patients with different diagnoses were seen in the scoring of perceived performance. Scores for the six items as well as the two summary skills were consistantly higher in the C/S group and lower in the DM and stroke groups. Association between recommendation/ satisfaction and hospital ratings The association between recommendation/satisfaction and the two scores is presented in Table 4. Patient satisfaction was classified into three categories: `satisfied', `fair', and `not satisfied'. Patients who did not answer the question (1.6%) were excluded from the analysis. Recommendation of a hospital was categorized in a different way. Since 16.4% of the respondents did not answer the question, we specified this group as a unique category. These subjects might represent the opinion between `recommend' and `not recommend'. Therefore, the three groups of the recommendation question were: strongly recommend and recommend (71.4%), no answer (16.4%), and conservatively recommend or not/ definitely not recommend (12.2%). Constant associations were observed between satisfaction/ recommendation and the two sets of interpersonal and technical scores. Patients who were satisfied reported higher values for interpersonal and technical scores (13.0 1.6 and 12.3 1.6), while the lowest scores (10.1 2.8 and 9.3 2.7) were associated with dissatisfied patients. The trend was consistent for each of the four disease categories, with all of the associations having a statistical significance of P 5 0.05 level with one-way ANOVA. Recommending a hospital or not showed similar results in relation to interpersonal and technical scores. Patients who would recommend a hospital had the highest ratings (12.9 1.6 and 12.3 1.5, respectively), while patients who would not recommend a hospital had the lowest ratings (11.1 2.3 and 9.9 2.1). Patients who gave no answer to the recommendation questions had in between ratings (12.1 1.8 and 11.2 1.7). All differences were significant at P 5 0.05 level (one-way ANOVA). Association between satisfaction and recommendation The mean scores of interpersonal and technical skills were slightly different in the three categories of satisfaction and recommendation. Therefore, we constructed a 3 3. 350. cross-table to examine the degree of concordance between satisfaction and recommendation. We found that, by combining the four disease categories, 52.4% of the patients were in the satisfied----recommend cell, 6.3% of the patients were in the fair----no answer cell, and only 0.8% of the respondents were in the dissatisfied----not recommend cell. Approximately 40% of the patients were in the off-diagonal cells, indicating that some satisfied patients would not recommend a hospital or vice versa. For example, 20.8% of the `not satisfied' patients still recommended the hospital. This is an interesting finding. Findings from two logistic regression models We subsequently constructed logistic regression models for the four diagnoses respectively, to investigate the effects of patients' perceived hospital quality on their satisfaction and recommendation. Results of the regression models for the two surgical conditions, C/S and appendectomy, are presented in Table 5. In order to compare the magnitude of the effects of various independent variables, we presented the standardized regression coefficients from the models. Interpersonal skills and technical skills were the two major influential factors for both patient satisfaction and recommendation. In the C/S group, technical skills appeared to be more influential than inter-personal skills for either satisfaction or recommendation, with standardized coefficients of 0.35 versus 0.24 and 0.45 versus 0.29, respectively. However, in the appendectomy group, inter-personal skills and technical skills had similar magnitude of effects on patient satisfaction (0.29 versus 0.31); while technical skills accounted for a heavier weighting (0.68 versus 0.32) on patient recommendation of a hospital. No significant difference was found on the association between satisfaction/recommendation and different levels of hospital, and that higher education levels in the C/S group were negatively associated with the recommendation of hospitals. Table 6 shows the results of logistic regression models for the two medical diagnoses. Similar to the findings for the two surgical patients, inter-personal and technical skills were the two major determinants for patient satisfaction and recommendation. In the DM group, interpersonal and technical skills had similar magnitudes of effect on patient satisfaction, with standardized regression coefficients of 0.32 and 0.27, respectively. However, technical skills had a greater effect on patient recommendation, with coefficients of 0.77 and 0.29, respectively. Findings in the stroke group were slightly different. For patient satisfaction, inter-personal skill was more influential (0.41 versus 0.19); however, for patient recommendation, technical skills were apparently more important (0.71 versus 0.39). We also found that DM patients were more satisfied with higher-level accredited hospitals and were more likely to recommend medical centers. Compared with the illiterate, any level of education was negatively associated with satisfaction and recommendation in the DM group. No personal characteristics were found to be associated with satisfaction or recommendation among the stroke patients..

(7) Table 4 Mean scores for interpersonal and technical skills by levels of patient satisfaction and recommendation as rated by the study subjects in the four groups1 Caesarean section (n 1524) Appendectomy (n 623) Diabetes mellitus (n 908) Stroke (n 977) Total (n 4032)2 ...................................................... ................................................ ................................................... ............................................... ............................................... IS TS IS TS IS TS IS TS IS TS [mean (SD)] [mean (SD)] [mean (SD)] [mean (SD)] [mean (SD)] [mean (SD)] [mean (SD)] [mean (SD)] [mean (SD)] [mean (SD)] ................................................................................................................................................................................................................................................................................................................ Level of satisfaction reported Satisfied 13.22 (1.58) 12.65 (1.44) 12.92 (1.76) 12.24 (1.57) 12.88 (1.49) 11.88 (1.51) 12.93 (1.44) 11.99 (1.60) 13.04 (1.57) 12.28 (1.55) Fair 12.20 (2.09) 11.47 (1.60) 11.84 (1.91) 11.20 (1.76) 11.88 (1.80) 11.03 (1.71) 11.82 (1.76) 10.99 (1.86) 11.94 (1.90) 11.17 (1.74) Not Satisfied 10.93 (2.87) 10.74 (2.40) 10.37 (2.24) 9.63 (2.79) 9.44 (2.64) 8.12 (2.35) 9.94 (3.03) 9.29 (2.70) 10.10 (2.78) 9.34 (2.69) Decision to recommend hospital Recommend 13.12 (1.68) Not Recommend 11.54 (2.45) No answer 12.65 (1.60). 12.56 (1.47) 10.78 (1.91) 11.88 (1.45). 12.85 (1.71) 11.08 (2.26) 12.09 (1.96). 12.25 (1.46) 9.99 (2.06) 11.44 (1.85). 12.75 (1.57) 11.10 (2.26) 11.98 (1.81). 11.97 (1.40) 9.44 (2.13) 10.86 (1.61). 12.87 (1.47) 10.82 (2.28) 11.75 (1.69). 12.07 (1.50) 9.45 (2.13) 10.80 (1.66). 12.94 (1.62) 11.13 (2.33) 12.07 (1.78). 12.27 (1.48) 9.92 (2.13) 11.16 (1.68). 351. Patient satisfaction and recommendation. IS, interpersonal skills; TS, technical skills. 1 The mean scores of the interpersonal skills and technical skills varied significantly among the three categories of patient satisfaction in all the four condition/disease groups at P 5 0.05 level using ANOVAs, while similar differences were observed among the three categories of patients recommendation of hospital. 2 Study subjects with missing values to either satisfaction or recommendation questions were not included in the analysis..

(8) S.-H. Cheng et al.. Table 5 Determinants of patient satisfaction and recommendation of hospitals from logistic regression models of subjects with surgical conditions1 (standardized b coefficients) (P values in parentheses) Caesarean section Appendectomy ........................................................................... ......................................................................... Satisfaction Recommendation Satisfaction Recommendation ............................................................................................................................................................................................................................. n 1691 1545 774 647 0.24 (0.001) 0.29 (0.001) 0.31 (0.001) 0.32 (0.001) Interpersonal skill2 Technical skills2 0.35 (0.001) 0.45 (0.001) 0.29 (0.001) 0.68 (0.001) Hospital accreditation Medical center 0.06 (0.132) 0.15 (0.011) ----0.02 (0.655) 0.09 (0.256) Regional hospital 0.06 (0.130) 0.19 (0.001) 0.01 (0.922) 0.24 (0.003) Gender (Male 1) ----------0.06 (0.210) 0.03 (0.682) Age3 Dummy range1 ÿ0.04 (0.732) ÿ3.25 (0.950) ÿ0.02 (0.704) ÿ0.08 (0.331) Dummy range2 ÿ0.01 (0.950) ÿ3.27 (0.949) ÿ0.09 (0.087) ÿ0.06 (0.506) Dummy range3 0.01 (0.778) ÿ0.91 (0.949) 0.04 (0.510) 0.13 (0.187) Education, years (ref. 57 years) 7----9 ÿ0.12 (0.069) ÿ0.19 (0.080) ÿ0.03 (0.619) 0.09 (0.318) 10----12 ÿ0.13 (0.189) ÿ0.26 (0.161) 0.01 (0.929) 0.02 (0.853) 13 ÿ0.07 (0.471) ÿ0.35 (0.055) 0.08 (0.216) 0.04 (0.704) ÿ2 Log L c2 238.6 (0.001) 218.9 (0.001) 132.9 (0.001) 169.7 (0.001) Concordant (%) 73.6 80.1 73.8 83.5 Interviewee status4 0.04 (0.230) ÿ0.15 (0.031) 0.17 (0.001) 0.06 (0.411) Variables. 1 Results from logistic regression models presenting standardized b coefficients for patients with Caesarean section or appendectomy. Dependent variables were `satisfied 1, fair or not satisfied 0' and `strongly recommend and recommend 1, conservatively recommend and not recommend 0'. 2 Missing values to the inter-personal or technical skills variables were replaced by subgroup mean values listed in Table 3. 3 Appropriate dummy variables were created for age groups for Caesarean section and appendectomy respectively. The four categories for Caesarean section were: 24, 25----29, 30----34, and 35. The four dummies for appendectomy were: 20, 21----30, 31----40, and 41. The youngest groups were treated as references in the models. 4 Interviewee status indicates whether the sampled patients answered the questionnaire by themselves or by their caregivers or relatives, with sampled patient 1, others 0.. Discussion Patient satisfaction and recommendation are commonly used and highly correlated measures for perceived quality of care. However, we found that subjects were more hesitant to respond to `recommendation' questions (16.4% `no answer') compared with the `satisfaction' questions (1.6% `no answer'). One possible explanation for this finding is that people may feel responsible for recommending a hospital; therefore, when they are not certain about the quality of a hospital, they tend to skip the question. Besides, we found that a certain proportion (20.8%) of the `not satisfied' patients still recommended the hospital. This means that a hospital with high percentage of patient satisfaction does not necessarily receive a similar level of recommendation. Patient ratings concerning hospitals' technical competence and physicians' interpersonal skills seemed to be good predictors for patient satisfaction and patient recommendation of a hospital. Yet, perceived technical competence and interpersonal skills had different magnitudes of effect on patient. 352. satisfaction and recommendation. Notably, we found that technical aspect of care is more important than interpersonal skills in determining patient recommendation of a hospital. According to previous reports, patient satisfaction mainly reflected a hospital's service quality rather than clinical quality [3,8,17]. Our findings largely support their conclusions. Interpersonal skills were as influential or more influential than clinical competence on patient satisfaction for three of the four disease categories. In contrast, technical competence was a more influential predictor for recommendation for patients in all four disease categories (Tables 5 and 6). The personal characteristics of patients have been reported to be associated with patient satisfaction [1,8,24], while some studies have reached different conclusions [19]. When analyzing the data here for the four diagnoses combined, we found that patient age, gender and education were significantly associated with their satisfaction response. After controlling for the four disease categories in separate models, however, personal characteristics had only a slight influence on satisfaction with, or recommendation of, a hospital. Our findings also.

(9) Patient satisfaction and recommendation. Table 6 Determinants of patient satisfaction and recommendation of hospitals from logistic regression models of subjects with medical conditions1 (standardized b coefficients) (P values in parentheses) Diabetes mellitus Stroke ....................................................................... ....................................................................... Satisfaction Recommendation Satisfaction Recommendation ............................................................................................................................................................................................................................. n 1136 924 1251 1012 Interpersonal skills2 0.32 (0.001) 0.29 (0.001) 0.41 (0.001) 0.39 (0.001) Technical skills2 0.27 (0.001) 0.77 (0.001) 0.19 (0.001) 0.71 (0.001) Hospital accreditation Medical center 0.11 (0.006) 0.27 (0.001) ÿ0.02 (0.679) 0.09 (0.235) Regional hospital 0.10 (0.015) 0.09 (0.182) 0.00 (0.902) ÿ0.05 (0.448) Gender (male 1) ÿ0.01 (0.858) 0.01 (0.872) ÿ0.03 (0.418) ÿ0.07 (0.246) Age, years (ref. 555 years) 55----64 0.03 (0.564) 0.00 (0.984) ÿ0.02 (0.658) ÿ0.04 (0.603) 65----74 0.03 (0.561) 0.03 (0.754) 0.05 (0.323) 0.06 (0.486) 75 ÿ0.02 (0.649) ÿ0.03 (0.725) 0.06 (0.249) 0.02 (0.864) Education (ref. 0 years) 1----6 ÿ0.10 (0.024) ÿ0.01 (0.857) 0.06 (0.181) ÿ0.04 (0.573) 7 ÿ0.05 (0.346) ÿ0.19 (0.019) 0.04 (0.376) ÿ0.03 (0.643) ÿ2 Log L c2 190.6 (0.001) 282.7 (0.001) 226.3 (0.001) 301.2 (0.001) Concordant (%) 72.4 86.5 72.8 85.4 3 Interviewee status 0.07 (0.085) ÿ0.02 (0.735) 0.10 (0.009) ÿ0.04 (0.521) Variables. 1. Results from logistic regression models presenting standardized b coefficients for patients with stroke or diebetes mellitus. Dependent variables were `satisfied 1, fair or not satisfied 0' and `strongly recommend and recommend 1, conservatively recommend and not recommend 0'. 2 Missing values to the inter-personal or technical skills variables were replaced by sub-group mean values listed in Table 3. 3 Interviewee status indicates whether the sampled patients answered the questionnaire by themselves or by their caregivers or relatives, with sampled patient 1, others 0.. suggest that the influence of personal characteristics on patient satisfaction tended to be less conclusive. Limitations of the study should also be noted. Compared with well-developed measurement tools such as Patient Judgment of Hospital Quality (PJHQ), which has a total of 106 items [9], the six items used in our study may be arbitrary and too limited. The three items measuring the inter-personal skills of physicians might be more acceptable for some researchers than the three items for technical competence. One may argue that patients do not have the ability to judge the excellence of a hospital's equipment or a doctor's competence; however, patient perception is an influential factor for their selection of hospitals. This is the key point of our study------to show the effects of perception on patient satisfaction and recommendation of a hospital. We agree that there is room for improvement and that more items may be needed; however, this might be a good starting point to investigate patient perception of the clinical performance of hospitals and their recommendation of a hospital. Findings from the logistic regression models can be refined in advance with proper sampling weights considered in the models. Reputation, or consumer recommendation in this study, is an important source of information for patients or families when choosing a hospital. Our preliminary results indicate that, in Taiwan, a hospital's reputation is determined mainly by its clinical competence rather than interpersonal skills.. A hospital with a good reputation, i.e. being highly recommended, does not necessarily have a similar level of patient satisfaction. Our findings might reveal new insights for researchers dealing with quality of health care and for hospital managers who devote resources exclusively to achieving highest possible levels of patient satisfaction.. Acknowledgements The authors wish to thank the two reviewers for their valuable suggestions, and Mr I-Chan Huang, Miss Chin-Yu Lin, and Miss Shu-Hui Chen for their professional assistance. The authors appreciate the co-operation of the Bureau of National Health Insurance and its six branches. This study was supported by grants from the National Science Council (NSC882621-B-002-027, NSC89-2320-B-002-095) and the National Health Research Institutes (NHRI-EX90-8801PP) in Taiwan.. References 1. Young GJ, Meterko M, Desai KR. Patient satisfaction with hospital care: Effects of demographic and institutional characteristics. Med Care 2000; 38: 325----334.. 353.

(10) S.-H. Cheng et al.. 2. Jackson JL, Kroenke K. Patient satisfaction and quality of care. Mil Med 1997; 162: 273----277. 3. Burroughs TE, Davies AR, Cira JC, Dunagan WC. Understanding patient willingness to recommend and return: A strategy for prioritizing improvement opportunities. Jt Comm J Qual Improv 1999; 25: 271----287. 4. Jatulis DE, Bundek NI, Legorreta AP. Identifying predictors of satisfaction with access to medical care and quality of care. Am J Med Qual 1997; 12: 11----18. 5. Ross CK, Steward CA, Sinacore JM. A comparative study of seven measures of patient satisfaction. Med Care 1995; 33: 392----406.. 20. Hibbard JH, Jewett JJ, Legnini MW, Tusler M. Choosing a health plan: do large employers use the data? Health Aff 1997; 16: 172----180. 21. Hsieh HS, Cheng SH, Lew-Ting CY. Demand for provider selection information of inpatients: Findings from a preliminary study (in Chinese). Chin J Public Health 2000; 19: 437----445. 22. Doering ER. Factors influencing inpatient satisfaction with care. QRB Qual Rev Bull 1983; 9: 291----299. 23. Abramowitz S, Cote AA, Berry E. Analyzing patient satisfaction: A multianalytic approach. QRB Qual Rev Bull 1987; 13: 122----130.. 6. Lee K, Kasper JD. Assessment of medical care by elderly people: general satisfaction and physician quality. Health Serv Res 1998; 32: 741----758.. 24. Boudreaux ED, Ary RD, Mandry CV, McCabe B. Determinants of patient satisfaction in a large, municipal ED: the role of demographic variables, visit characteristics, and patient perceptions. Am J Emerg Med 2000; 18: 394----400.. 7. Kane RL, Maciejewski M, Finch M. The relationship of patient satisfaction with care and clinical outcomes. Med Care 1997; 35: 714----730.. 25. Department of Health. Health and vital statistics, I: General Health Statistics, 2000: Republic of China. Taipei, Taiwan: Department of Health, the Executive Yuan, 2001: 110----124.. 8. Hall JA, Dornan MC. Patient socio-demographic characteristics as predictors of satisfaction with medical care: a meta-analysis. Soc Sci Med 1990; 30: 811----818.. 26. Peabody JW, Yu JC-I, Wang Y-R, Bickel SR. Health system reform in the Republic of China: formulating policy in a market-based health system. J Am Med Assoc 1995; 273: 777----781.. 9. Rubin HR. Can patients evaluate the quality of hospital care? Med Care Rev 1990; 47: 267----326.. 27. Chiang TL. Taiwan's 1995 health care reform. Health Policy 1997; 39: 225----239.. 10. Carey RG, Seibert JH. A patient survey system to measure quality improvement: Questionnaire reliability and validity. Med Care 1993; 31: 834----845.. 28. Cheng SH, Chiang TL. The effect of universal health insurance on health care utilization in Taiwan: Results from a natural experiment. J Am Med Assoc 1997; 278: 89----93.. 11. Cleary PD, Edgman-Levitan S, Roberts M et al. Patients evaluate their hospital care: A national survey. Health Aff 1991; 10: 254----267.. 29. Chang MS. The status and planning of specialty manpower. In Proceedings of the Conference on Health Manpower Planning and Forecasting in Taiwan, 2000. Taipei: National Health Research Institutes (Supplement).. 12. Rubin HR, Ware JE, Hays RD. The Patient Judgments of Hospital Quality (PJHQ) questionnaire. Med Care 1990; 28 (suppl): S22----S29. 13. Donabedian A. The quality of care: How can it be assessed. J Am Med Assoc 1988; 260: 1743----1748. 14. Tokunaga J, Imanaka Y, Nobutomo K. Effects of patient demands on satisfaction with Japanese hospital care. Int J Qual Health Care 2000; 12: 395----401. 15. Hargraves JL, Wilson IB, Zaslavsky A et al. Adjusting for patient characteristics when analyzing reports from patients about hospital care. Med Care 2001; 39: 635----641. 16. Larsen DE, Rootman I. Physician role performance and patient satisfaction. Soc Sci Med 1976; 10: 29----32. 17. Cleary PD, McNeil BJ. Patient satisfaction as an indicator of quality care. Inquiry 1988; 25: 25----36. 18. Covinsky KE, Rosenthal GE, Chren MM et al. The relation between health status changes and patient satisfaction in older hospitalized medical patients. J Gen Intern Med 1998; 13: 223----229. 19. Finkelstein BS, Singh J, Silvers JB, Neuhauser D, Rosenthal GE. Patient and hospital characteristics associated with patient assessments of hospital obstetrical care. Med Care 1998; 36: AS68----AS78.. 354. 30. Huang P, Hsu Y-HE, Tan K-Y, Hsueh Y-S(A). Can European external peer review techniques be introduced and adopted into Taiwan's hospital accreditation system? Int J Qual Health Care 2000; 12: 251----254. 31. Department of Health. Health and vital statistics, I: General Health Statistics, 2000: Republic of China. Taipei, Taiwan: Department of Health, the Executive Yuan, 2001: 156.. Appendix. The eight questions used for patient ratings of hospital quality in the study (1) Satisfaction: `Generally speaking, were you satisfied with the inpatient care?' Response options: `satisfied', `fair', and `not satisfied'. (2) Recommendation: Ìf someone asks you about that hospital, would you recommend it?' Response options: `strongly recommend', `recommend', `conservatively recommend', `not recommend', and `definitely not recommend'. (3) Doctor's competence: `Was the doctor's competence good enough to take care of your need?' Response.

(11) Patient satisfaction and recommendation. options: `definitely enough', ènough', `fairly enough', `not enough', and `definitely not enough'. (4) Hospital equipment: `Was the hospital's equipment sufficient enough to take care of your need?' Response options: `definitely sufficient', `sufficient', `fairly sufficient', `not sufficient', and `definitely not sufficient'. (5) Outcome or recovery: `Was the treatment outcome/ recovery better than you expected or not?' (Using `recovery' for C/S patients only.) Response options: `much better than expected', `better than expected', `same as expected', `worse than expected', and `much worse than expected'.. (6) Doctor's attitude: `Was the doctor's attitude good or not?' Response options: `very good', `good', `fair', `poor', and `very poor'. (7) Doctor's explanation: `Did the doctor explain your diagnosis/situation clearly?' Response options: `very clear', `clear', `fairly clear', ùnclear', and `very unclear'. (8) Doctor's caring: `Did the doctor care about you?' Response options: `cares a lot', `cares', `fair', `doesn't care', and `doesn't care at all'.. Accepted for publication 28 March 2003. 355.

(12)