Beruflich Dokumente
Kultur Dokumente
1093/intqhc/mzl022
Advance Access Publication: 19 July 2006
Abstract
Objectives. To develop and validate a concise scale for measuring outpatient satisfaction suitable across specialties and cultures.
Design & Setting. Item generation adopted a concept-driven approach, and 10 candidate items were administered together
with a battery of validation items and scales in a cross-sectional survey at a government-aided Chinese medicine specialized
In recent years, there has been an increasing demand of Among the existing ones, some scales have yet to complete
accountability and productivity by consumers. It is now a the validation process [9]. Some reported satisfactory or good
global trend in healthcare development toward integrating levels of psychometric properties, but very often they contain
subjective user satisfaction into the evaluation of medical many items (typically 20–30 items) [10–12]. On the one hand,
service quality [1,2]. There is evidence that patient satisfac- looking at individual aspects of service (at least the most
tion is associated with treatment outcomes [3,4] though the important ones) is necessary, as most patients are not uni-
cause–effect relationship is not clear. It is possible that ill- formly satisfied with all aspects even though they are gener-
ness behavior plays a role in the mechanism through which ally satisfied with the service [13]. Also, it is conceivable that a
satisfaction affects clinical outcomes. Ferris and the Health longer scale more likely yields more reliable results. On the
Services Research Group [5] reported that satisfied patients are other hand, a user satisfaction measure with numerous items
more likely than their dissatisfied counterparts to show positive has its own drawbacks. A long scale can lead to low motiva-
illness behaviors, e.g. complying with their medical regimens tion of patients to complete the questions and subsequently a
and disclosing important medical information to their physi- low response rate and biased sample as well as more missing
cians [6–8]. Therefore, knowledge of users’ satisfaction with data. It is also difficult for many outpatient settings to incor-
the service can serve not only as a performance indicator but porate the administration of a long questionnaire into their
also to identify areas of improvement to provide better care busy daily operation, needless to mention that very often
and services for the betterment of the users’ health. other questionnaires or assessments are administered
There are very few user satisfaction scales that are suitable together at the same time. Seeing this limitation, Perneger
for use in outpatient settings across different specialties. et al. [14] derived a 16-item outpatient satisfaction survey
Address reprint requests to S. M. Ng, Assistant Professor, Centre on Behavioral Health, The University of Hong Kong, G/F
Pauline Chan Building, 10 Sassoon Road, Pokfulam, Hong Kong. E-mail: ngsiuman@hku.hk
from previously published instruments. The scale was and interviewer’s independency were emphasized to avoid or
administered to a large sample (n = 1027) in Switzerland and minimize biased responses due to worries of loss of anonymity
the seven dimensions of the scale showed satisfactory to and negative impact on service to be received.
good internal consistency. An even briefer instrument (one
item of overall satisfaction and eight items of different satis-
faction aspects) was developed by Rubin et al. [15] to com- The Concise Outpatient Department User
pare outpatient satisfaction with different kinds of practices Satisfaction Scale
in the United States. The authors successfully collected eval- A scale was constructed to measure user satisfaction. The
uation data from 17 671 outpatients and found that overall investigators adopted a concept-driven approach for genera-
satisfaction was associated with patients’ staying with or tion of items that are considered generally applicable to out-
leaving the physicians or systems in 6 months. However, the patient settings of different specialties. The construct was
relationship between different satisfaction aspects and out- conceptualized as having three domains, namely physical
come was not examined. It would also have been helpful if setup, clinic operation, and case physician. Specific items
the authors could provide detailed psychometric statistics of were then generated under each domain. Because the scale
the questionnaire. intends to be applicable across medical specialties, and hope-
Another important issue of outpatient satisfaction assess- fully across culture as well, great caution had been taken in
ment is that most scales are developed in the West. These avoiding items that are restricted to a certain disease group or
scales often include items that are specific or applicable to cultural related lifestyle. The initial item pool was reviewed
276
Concise Outpatient Department User Satisfaction Scale
each of whom was a retired registered nurse with over 20 years systematic bias in the successfully recruited sample. Among
of experience in public hospitals in Hong Kong and had been the 92 non-responders/dropouts, 34 (37.0%) were male and
trained by the investigators (S.M.N. and I.F.T.) on administer- 58 (63.0%) were female. Their mean age was 45.2 years (SD =
ing the questionnaires of the study, including the Concise Out- 18.1). The responders and the non-responders/dropouts
patient Department User Satisfaction Scale. There were two were not statistically different in terms of age (P = 0.593) and
interviewers standing by at the clinic at any time during the sex (P = 0.116).
recruitment period to recruit and interview subjects.
After receiving information of the study and giving informed
Factor structure
consent, each subject was interviewed by an interviewer. Inter-
view instead of self-completion was employed to minimize Inter-item correlations were examined before proceeding
invalid or missing data [12] and to obtain more reliable to factor analysis. Item 9 (case physician’s attitude and
responses by providing instant explanations of the instructions manner) was removed from the scale because it was too
and items of the questionnaire to subjects when necessary. The highly correlated with items 7, 8, and 10 (r = 0.91, 0.83,
subjects were followed up on telephone by their corresponding and 0.88, respectively) and thus considered a redundant
interviewers 1 month after first interview (i.e. from October to item.
November 2004). If the subject could not be reached at the The number of valid cases (i.e. those free of missing data)
first attempt, then at least two more attempts were made before for exploratory factor analysis was 305. Using principal com-
he/she was considered unreachable and dropout. ponent analysis with criterion of eigenvalue > 1, two factors
277
I. F. Tso et al.
Table 2 Means and standard deviations for the Concise Outpatient Department User Satisfaction Scale
Total satisfaction 7.00 (1.15) 7.11 (1.00) 6.95 (1.20) 6.75 (1.03) 7.02 (1.12) 7.13 (1.15) 6.80 (1.31) 7.06 (1.10)
General service 7.03 (1.21) 7.14 (1.06) 6.99 (1.27) 6.72 (1.15) 7.09 (1.17) 7.16 (1.29) 6.90 (1.35) 7.07 (1.17)
Case physician 6.94 (1.45) 7.07 (1.21) 6.89 (1.53) 6.80 (1.09) 6.90 (1.47) 7.09 (1.62) 6.58 (1.76) 7.04 (1.33)
Discussion
Internal consistency
By odd–even split of the nine items, the Guttman split-half We successfully created a scale with a parsimonious number of
reliability was estimated to be 0.93 and the unequal-length items that is suitable for outpatient settings across specialties.
Spearman–Brown reliability was 0.94. The coefficient alpha The scale had high levels of internal consistency and validities
of the scale, which represents the average of all possible as suggested by alpha coefficient and the scale’s correlations
split-half reliability coefficients, was 0.91, falling into the with other related measures. Factor analysis suggested that
range of ‘good’. The factor alphas were also in the good about 75% of the variance of the items could be explained by
range: 0.90 for general service and 0.93 for case medical two latent variables, namely general service and case physician.
practitioner. Contrary to expectation, physical setup and clinic operation of
an outpatient department could not be differentiated from one
another; they contributed to user satisfaction as a single factor.
In this study, a 10-point response format in which the two
Criterion-related validity
ends are labeled respectively ‘not satisfied at all’ and ‘extremely
Correlations between the satisfaction scores and validation satisfied’ was adopted. The decision was based on the rationale
items/scales are summarized in Table 3. Almost all individual that giving ratings in a 10-point scale is intuitively more direct
items were significantly correlated with the three anchor and thus easier than, say, 4-, 5-, or 6-point scales. This
items: overall satisfaction (r = 0.51–0.77), intended future response format was also preferable to Likert scale for two
reutilization (r = 0.13–0.50), and likelihood of recommending reasons: respondents can give a rating without needing to read
the clinic to others (r = 0.10–0.50). The total satisfaction the description of each response choice as in Likert response
score (averaged sum of all items) was also significantly corre- format and it offers a more sensitive measure and avoids
lated with the three anchor items (r = 0.85, 0.38, and 0.38, skewedness potentially caused by the choice of the number of
respectively, all P < 0.001), supporting the concurrent valid- response choices (e.g. 5- versus 6-point Likert scale) [20]. The
ity. The two subscale scores were also significantly correlated high level of internal consistency (alpha > 0.90) and low rate
with the three anchor items. For general service, the correla- of missing data (<2%) support that the scale was easy to
tion coefficients were 0.76, 0.26, and 0.23, respectively (all understand and effective in obtaining reliable measures.
P < 0.001). For case physician, the correlation coefficients Interestingly, although the total satisfaction score was corre-
were 0.77, 0.49, and 0.51, respectively (all P < 0.001). lated with the single-item overall satisfaction more strongly than
278
Concise Outpatient Department User Satisfaction Scale
Subscores
General service 0.941
Case physician 0.811 0.561
Anchor items
Overall satisfaction 0.851 0.761 0.771
Future reutilization 0.381 0.261 0.491
Recommend to others 0.381 0.231 0.511
Construct validity items
Compliance (follow up) 0.231 0.151 0.311
Compliance (medical regimen) 0.181 0.11 0.251
Chinese Affect Scale—positive affect 0.05 –0.05 0.201
Chinese Affect Scale—negative affect –0.261 –0.221 –0.241
1
Significant at the 0.001 level.
279
I. F. Tso et al.
9. Dawn AG, Lee PP, Hall-Stone T, Gable W. Development of 15. Rubin HR, Gandek B, Rogers WH, Kosinski M, McHorney CA,
a patient satisfaction survey for outpatient care: a brief report. Ware JE Jr. Patients’ ratings of outpatient visits in different prac-
J Med Pract Manage 2003; 19: 166–169. tice settings: results from the Medical Outcome Study. JAMA
1993; 270: 835–840.
10. Gasquet I, Villeminot S, Estaquio C, Durieux P, Ravaud P,
Falissard B. Construction of a questionnaire measuring outpa- 16. Cho WH, Lee H, Kim C, Lee S, Choi KS. The impact of visit fre-
tients’ opinion of quality of hospital consultation departments. quency on the relationship between service quality and outpatient sat-
Health Qual Life Outcomes 2004; 2: 43. isfaction: a South Korean study. Health Serv Res 2004; 39 (1): 13–33.
11. Lubeck DP, Litwin MS, Henning JM, Mathias SD, Bloor L, 17. Fitzpatrick R, Hopkins A. Problems in the conceptual frame-
Carroll PR. An instrument to measure patient satisfaction with work of patient satisfaction research: an empirical exploration.
healthcare in an observational database: results of a validation Sociol Health Illn 1983; 5: 297–311.
study using data from CaPSURE. Am J Manag Care 2000; 6: 70–76.
18. Hamid PN, Cheng ST. The development and validation of an
12. Labarere J, Francois P, Bertrand D, Peyrin JC, Robert C, Fourny M. index of emotional disposition and mood state: the Chinese
Outpatient satisfaction: validation of a French-language ques- Affect Scale. Educ Psychol Meas 1996; 56: 995–1014.
tionnaire: data quality and identification of associated factors.
19. Cronbach LJ. Coefficient alpha and the internal structure of
Clin Perform Qual Health Care 1999; 7: 63–69.
tests. Psychometrika 1951; 16: 297–334.
13. Coulter A, Cleary PD. Patients’ experiences with hospital care in
20. Ware JE, Hays RD. Methods for measuring patient satisfac-
five countries. Health Aff (Millwood) 2001; 20: 244–252.
tion with specific medical encounters. Med Care Rev 1988; 26:
Appendix
Concise Outpatient Department User Satisfaction Scale
Please indicate your satisfaction with each of the following aspects of this clinic by circling a number from 1 to 10, where ‘1’
means not satisfied at all and ‘10’ means extremely satisfied.
1. Physical environment 1 2 3 4 5 6 7 8 9 10
2. Equipment and facilities 1 2 3 4 5 6 7 8 9 10
3. Appointment arrangement 1 2 3 4 5 6 7 8 9 10
4. Waiting time 1 2 3 4 5 6 7 8 9 10
5. Service of the dispensary 1 2 3 4 5 6 7 8 9 10
6. Support staff 1 2 3 4 5 6 7 8 9 10
7. Your case physician’s professionalism 1 2 3 4 5 6 7 8 9 10
8. Explanations given by your case physician 1 2 3 4 5 6 7 8 9 10
9. Consultation time 1 2 3 4 5 6 7 8 9 10
280