Patient satisfaction with outpatient physical therapy: instrument validation. (Research Report).

In the current health care environment, physical therapists are faced with the challenge of practicing in an increasingly competitive marketplace. The rapidly growing number of practicing physical therapists and the widespread use of "care extenders" has allowed patients more options when choosing providers. Physicians are often under pressure to reduce referrals for physical therapy, and people from other disciplines, such as personal trainers, are aggressively marketing their services to prospective patients. As marketplace competition continues to grow, patient satisfaction with physical therapy is emerging as an outcome variable of critical importance. (1-4) Patients who report high satisfaction with care are more likely to continue the relationship with the health care practitioner, to seek additional medical care when needed, and to adhere to recommended treatment plans. (5-8)

Patient satisfaction is often considered to be an abstract, multidimensional phenomenon. (6,7,9,10) Because it usually is not observable directly, patient satisfaction must often be measured in what we would consider an indirect manner (ie, from self-report measures). (1-4,6,11-14) A simple self-report method for assessing satisfaction is to ask global questions such as, "Overall, I am completely satisfied with my care." (6) These questions, although easy to administer, do not provide information about why a person is or is not satisfied; therefore, many authors (1-4,6,9,10) recommend the use of multidimensional measures. Consequently, the question arises about which variables are needed to assess patient satisfaction adequately. If a measure does not include all relevant variables, important information may be missed, whereas sampling too many variables may provide irrelevant or misleading information. For example, an instrument may erroneously identify a lack of parking and poor location as sources of patient dissatisfaction. This could result in a clinic undergoing an expensive relocation when the actual source of dissatisfaction was insufficient therapist time with the patient.

Numerous aspects of patient satisfaction have been described, and the most common factors are: the patient-practitioner relationship (competence, personality of the practitioner, communication), location and accessibility of services, continuity of care, cost and payment issues, and the facility (eg, cleanliness, noise, equipment). (1,4,6,14) Although several researchers (5,8-10,12-22) have described the development or use of measures of patient satisfaction with overall medical care, the applicability of these measures to patient satisfaction with physical therapy is uncertain. In our opinion, the unique aspects of care related to outpatient physical therapy--such as the need for frequent visits over a short period of time as well as the need for patients to stay in the clinic for sessions that are longer than those of a typical physician's visit--may require a different, "specialty-specific" scale.

Recently, 2 groups (1,4) have described the development of instruments for assessing patient satisfaction in outpatient physical therapy settings. Both instruments had what we would consider good psychometric properties; however, we believe they reflect different aspects of patient satisfaction. Roush and Sonstroem (4) developed an outpatient satisfaction survey by sequentially testing 3 patient samples totaling 607 people. The authors proposed using a 34-item survey questionnaire to measure 4 dimensions: enhancers, detractors, location, and cost. Location and cost were found to be the greatest influences to satisfaction. Goldstein et al (1) reported measurement properties on a different instrument that was tested on 289 subjects. The authors proposed a 15-item questionnaire that indicated that a single dimension representing patient-therapist interaction was most important. Cost of care was not correlated with overall satisfaction.

The variation in the content of these instruments suggests a need to investigate the variables associated with patient satisfaction. The purpose of our study was to develop and test an instrument that would provide adequate measurement properties to assess which variables were most closely associated with the overall satisfaction of patients receiving outpatient physical therapy for occupation-related musculoskeletal conditions.

Method

General Description

Our study was divided into 2 phases. The first phase consisted of instrument development and pilot testing. In the second phase, the instrument was revised and administered to a large sample of people who were from several geographic areas and who had a variety of diagnoses. Using these data, further instrument revision using factor analysis and reliability testing was performed.

Phase 1: Initial Development of the Instrument and Pilot Testing

Initial development of the instrument. The first step in the development of the survey instrument was to create items that we believed could reflect the potential variables that influence patient satisfaction in an outpatient physical therapy setting. Taking into account the disagreement in the previous published work concerning the dimensions of satisfaction (1,4) and based on discussions with physical therapists and patients and our own views, chose items reflecting what we considered 2 broad variables: personal aspects of the therapist and system/external aspects. The personal aspects consisted of a series of items regarding the patient's interaction with the physical therapist and associated staff (ie, physical therapist assistant, receptionist, and other office staff). The systems/external aspects related to issues such as clinic accessibility and location, waiting time, registration, and cleanliness of facilities. Research has suggested that cost of care is an important variable in patient satisfaction (4); however, because we studied people receiving workers' compensation, we chose not to include this item.

A list of 25 questions reflecting personal aspects and system/external variables was initially generated through informal discussions with patients and physical therapists as well as through a review of the current literature. (2,3) To test face validity, all items were reviewed by 2 physical therapists who assessed the items for logical consistency and content. These therapists were both doctorally trained and had an average of 35 years of academic clinical experience. They followed a protocol for analyzing the scale items and indicated whether, in their opinion, there were any awkward or inappropriate words or misleading phrases. Based on their recommendations, a revised 18-item questionnaire was created for initial pilot testing. The sequence of the items was randomized (ie, by pulling numbers out of a hat), and each item was rated on a 5-point scale (1=strongly disagree and 5=strongly agree). To control for response bias, positively and negatively worded items were included. In addition, we initially included 4 global measures of patient satisfaction to act as dependent variables: (1) "Overall, I am completely satisfied with the services I receive from my therapist," (2) "I would return to this office for future care," (3) "I would not recommend this therapist to a friend," and (4) "This therapy has helped me as much as expected." We expected that the summary score on the final instrument would be highly correlated with one or more of these global measures of satisfaction.

Pilot testing. Pilot testing was performed to assess the relationship of the 4 global items with one another and with each of the remaining variables. We used this information to determine the need to retain each of the global measures (ie, if a global measure is not strongly related to other items, it may reflect a different construct and may not be appropriate for this instrument).

The sample of respondents for this pilot phase included 191 people who were receiving outpatient physical therapy at any one of several physical therapy clinics throughout the United States. These clinics were privately or corporately owned, specialized in the evaluation and treatment of adults with musculoskeletal problems, and employed an average of 4.5 full-time physical therapists. All facilities were participating members of MedRisk Inc,* which is a preferred provider organization and an expert provider organization that contracts with physical therapists and other health care practitioners to provide care for people who are covered by workers' compensation. Thus, all subjects were receiving workers' compensation. The subjects were receiving treatment for one or more musculoskeletal conditions commonly encountered by physical therapists. All subjects had to be able to read English. Subjects were asked to complete the survey after at least 3 visits.

All rights of the subjects were protected. All responses were obtained anonymously. Patients who were asked to fill out a questionnaire were assured in its written instructions that their responses would be aggregated so that no one person's response could be identified, either through survey code markings or any other method.

Data analysis and results. An inter-item correlation matrix was calculated to determine the correlation (r) of each of the items (survey questions) with each of the global measures. All 18 items were correlated with each of the 4 global measures (ie, all of the questions were related to the patient's overall sense of satisfaction). A multiple regression analysis was used to estimate the correlation of variables with the global measures. The overall r values ranged from .35 to .74. Two of the 4 global measures were not strongly related to other items and were dropped, leaving a revised instrument of 18 items and the 2 remaining global measures: "Overall, I am completely satisfied with the services I receive from my therapist" and "I would return to this office for future care." The regression model had an adjusted [r.sup.2] value of .76; that is, the full model predicted approximately 76% of the causes of patient satisfaction with a physical therapy clinic. The Cronbach alpha, an indicator of internal consistency or the degree to which similar questions yield similar answers, (1) was calculated as .93 for the 18 items, thus indicating a high degree of internal consistency.

Phase 2: Content Validation and Reliability Testing

The revised 20-item instrument (18 specific questions and 2 global measures) was completed by patients following their course of physical therapy. Once again, all patients were receiving treatment under workers' compensation regulations. Outpatient clinics in 17 states participated in this phase of the study. Representatives of MedRisk Inc mailed the survey instrument to the patient's home approximately 4 weeks after completion of physical therapy. Of the 9,315 survey questionnaires that were sent out, 1,868 (20%) were returned and used in the analysis. The 20% return rate, in our opinion, represents a satisfactory return, but others may disagree,. (23) The mean age of the respondents was 46.9 years (SD=11.9); 676 respondents were female, and 1,192 respondents were male. Table 1 lists the number of respondents by geographic region. The majority of respondents received care in the Mid-Atlantic region. Table 2 summarizes the anatomic location of symptoms for which the patient sought treatment. There was a difference by gender for symptom location ([chi square]=24.12, df=5, P=.00). Observed counts exceeded those expected for male patients with thoracolumbar or lower-extremity problems and for female patients with hand problems or "other" symptoms.

To determine the interrelationships of the various items, a series of tests were used. Items with reverse coding were recoded so that all variables were positively coded (ie, 1="strongly disagree," 5="strongly agree"). An item-correlation matrix was generated to determine the correlation of items with the global measures. Following this, we conducted a principal components analysis (PCA). (4,24) This test reduces the number of variables into a small number of components (ie, groups of questions thought to represent similar concepts). Eigenvalues derived by PCA indicate the variance accounted for by each of the potential components and provide fundamental information regarding the total number of different components in the data set (ie, how many different groups of similar questions are included). Eigenvalues greater than 1.0 obtained during the PCA were retained. (24)

To determine how many components the final questionnaire should include and to determine whether a 2- or 3-component solution was superior, a factor analysis was performed. Factor analysis acts to maximize apparent differences among groups of questions, and it provides a coefficient of strength of association for each question to groups of questions (now called "factors"). (4,24,25) "Rotating factors" is a procedure used to clarify the differences among the factors. We used "Oblimin rotation" because this procedure is recommended to clarify relationships between attitude or belief components. (4) A .60/.30 criterion for factor retention was used to indicate the degree of correlation (loading) of each item to each factor (ie, an item must load greater than or equal to .60 on one factor and less than .30 on all other factors), representing a conservative inclusion criterion. (26)

Reliability. Reliability of the measurements obtained was tested by calculating the standard error of measurement (SEM). (27-30) Using the method described by Roddey et al (27) and others, (28,29) the SEM was calculated using the following equation:

SEM=SD V 1-[alpha]

where SD is the standard deviation of the observed scale or subscale and [alpha] is the Cronbach alpha for that scale or subscale. The lower the SEM, the less error is associated with the measure (ie, the closer the observed score is likely to be to the "true score"). The SEM provides a range or interval of scores about which the true measure is likely to fall. By adding and subtracting 1.96 SEMs to an observed score, one can estimate (with 95% certainty) the range in which the true score for a person will lie. For example, if the observed mean score on a scale is 4.5 and the SEM is 0.25, one may be 95% certain that the true score lies between 4.01 and 4.99 (4.5-0.49 and 4.5+0.49).

Results

Data Reduction and Reliability Testing

The item-correlation matrix revealed that all items were positively correlated with the 2 global measures (Tab. 3). The values ranged from .095 ("The clinic location is convenient") to .722 ("My therapist explains the treatment") for global measure 1 ("Overall, I am completely satisfied with the services I receive from my therapist") and from .106 ("The clinic location is convenient") to .681 ("My therapist answers my questions") for global measure 2 ("I would return to this office for future care").

The PCA demonstrated the existence of g eigenvalues greater than 1.0, explaining 57.03% of the variance (Tab. 4). The Scree test (4) provides a graphic representation of eigenvalues and was used to further clarify the number of components to rotate. Using the plot created by the Scree test, a cutoff point at which the slope of the decreasing eigenvalues begins to level off can be found. Based on the eigenvalues and the plot created by the Scree test, we examined a 2- and 3-component solution. Both analyses were followed by Oblimin rotation.

For the 2-component solution, we used PCA to identify a total of 10 items, whereas the 3-component solution retained 12 items (Tab. 5). Table 5 reports the factor loadings for each item. In both solutions, the first component consisted of 7 items related to the patient-therapist interaction (ie, personal or internal elements). Similarly, in both solutions, the second component consisted of g items related to the patient's perception of the receptionist, registration process, and waiting room (ie, systems/external aspects). In the 3-component solution, the third component contained 2 items similar to the "convenience" of receiving care (location of clinic and time waiting for the therapist). The 2 items in the convenience component, however, had the lowest correlations of any of the variables with the 2 global measures of satisfaction (Tab. 3).

The internal consistency (Cronbach alpha) of each scale and subscale for the 2- and 3-component solutions is shown in Table 6. The 2-component solution yielded an overall alpha of .90, whereas the 3-component solution had an overall alpha of .85. The alpha levels of the subscales ranged from .44 (convenience) to .92 (personal or internal elements). Reliability testing showed a similar trend, with the 2-component solution overall representing the least error variation and the convenience subscale yielding the highest error.

Testing of Concurrent Validity

Concurrent validity is a measurement property describing relationships between 2 measurements obtained relatively close in time. (30) Concurrent validity may be investigated by correlating the measurements from a new, untested instrument with a measurement that is believed to be a valid measure of the construct under investigation (ie, overall satisfaction). (1) The 2 global measures of satisfaction ("Overall, I am completely satisfied with the services I receive from my therapist" and "I would return to this office for future care") were included as correlates for the satisfaction items as a way to assess concurrent validity. Because there is no absolute "gold standard" by which to assess satisfaction, we felt that the development and evaluation of these global measures was appropriate. This procedure is similar to one taken by Goldstein et al. (1) These 2 measures were among the original 4 global measures suggested by the panel that developed our original questionnaire, and they appear to make sense (ie, have face validity). The 10 items retained in the 2-component solution and the 12 items retained in the 3-component solution were each used to form 2 separate mean summary scores. These scores were correlated with each of the scores of 2 global measures ("Overall, I am completely satisfied with the services I receive from my therapist" and "I would return to this office for future care") and with the mean value of both global measures. This provided an estimate of the degree to which the overall scales correlated with the criterion variables. Table 7 illustrates the correlations of the 2- and 3-component solutions and the associated subscales with these criterion variables (global measures). In all cases, the 2-component solution demonstrated higher correlations.

Discussion

Our findings suggest that reliability, as well as content and concurrent validity, can be obtained from a subscale that directly relates to a patient's interaction with his or her physical therapist, whereas measures of perceptions not directly related to patient care (ie, environment and location) have lower reliability. Our observation is in agreement with previous work by Goldstein et al, (1) who identified a single-factor solution related to patient-physical therapist interaction. These authors speculated that satisfaction was not strongly influenced by "ancillary aspects of care such as courtesy of the support staff and parking." The findings of other authors (6,31) support this idea. They found that the most important issues for patients receiving medical care are being treated with respect and being involved in treatment decisions. (6,31) Non-patient care issues such as parking and cleanliness of the facility are less important. (6,31)

In contrast to our findings, Roush and Sonstroem (4) contended that patient satisfaction is strongly influenced by factors such as location and cost. In the instrument proposed by Roush and Sonstroem, 15 of 34 questions sampled the patient's satisfaction with location, accessibility, and cost of care. In our sample, however, we found very low correlations among location and the global measures of satisfaction. In our analysis, the item addressing "location" factor-loaded with time spent waiting for the physical therapist; however, this component had low internal consistency, and, when it was deleted, the overall internal consistency of the measurements obtained from the remaining items improved. Because the population we studied was receiving workers' compensation, no items relating to cost of care were included. Recent data, however, suggest that less than 1% of patients receiving physical therapy pay the complete charges out of pocket (Advisory Committee on Practice, American Physical Therapy Association, personal communication, November 1998). Goldstein et al (1) also found a low correlation between cost of service and satisfaction. We question whether an instrument that is heavily weighted toward questions relating to cost and location is appropriate. Further study using concurrent comparisons of these instruments should be performed.

Maximizing patient satisfaction is a sound philosophy from both a clinical perspective and a business perspective. Satisfied patients are more likely to adhere to treatment and to continue to seek health care at a given facility. (32-39) Our findings indicate that adequate time spent in patient care and the professionalism of the therapist and clinic staff are more important for patient satisfaction than are the location of the facility, the quality of equipment, and the availability of parking. We believe that, in the current health care environment, the emphasis on cost-cutting, high patient volume, and the use of "care extenders" can reduce the time for the patient-therapist interactions that appeared to contribute to satisfaction. We believe that this trend will have a negative effect on patient satisfaction, which, in turn, will have a negative effect on the marketplace and on job satisfaction.

Hudak and Wright (6) provided an excellent review of the characteristics and use of a patient satisfaction measure. They noted that it is important to differentiate between patient satisfaction with outcome and patient satisfaction with care. Patient satisfaction with outcome relates to the results of treatment, whereas satisfaction with care reflects the service the patient received during a course of care. Arguably, these represent separate entities and should be assessed with different instruments. The instrument developed in our study was designed to be a self-administered, forced-choice format for assessing a person's satisfaction with the process of receiving physical therapy in an outpatient environment. We did not examine satisfaction with outcome. Based on our data, we believe that the 10-item instrument (which also includes 2 global questions and thus contains 12 total questions) provides a tool that is complete and easy to administer. We believe the instrument has sound measurement properties for assessing current patient satisfaction within the population of people receiving outpatient physical therapy related to workers' compensation (Appendix).

We believe instrument validation is an ongoing process. Our data suggest that reliable measurements may be obtained that describe various aspects of patient satisfaction with care at a given time. Although a repeated-measures design was not used in this study, test-retest reliability using the SEM has been described previously. (27) The low values of SEM for the 10-item instrument and its 2 subscales indicate a low degree of measurement error and, therefore, a high degree of test-retest reliability. Further study is needed to assess the criterion-referenced validity of the measurements obtained from this instrument.

We believe that a satisfaction measure must be viewed in the context in which it will be used. (6) We argue that, in many cases, standardized measures may not provide adequate data. (6) By including both closed- and open-ended survey questions, we contend that a richer understanding of satisfaction can be obtained. In addition to a standardized instrument, we suggest that clinicians may want to include a small number of open-ended questions to target issues unique to a given facility. We have no data, however, to suggest that this would be beneficial.

The data we obtained, in our opinion, are generalizable to adults with occupation-related musculoskeletal impairments from a large geographic area. The appropriateness of the proposed instrument to people who are not receiving workers' compensation is not known. Our instrument was constructed to assess patient satisfaction with care and is not adequate to measure satisfaction with outcomes. The overall assessment of outcome is a multidimensional task that, in addition to patient satisfaction with care, should also include other relevant measures such as health status, functional capacity, and quality of life.

Conclusions

The instrument developed during the process described here has been demonstrated to yield measurements that are reasonably reliable and have some content and concurrent validity. The results of our study show that patient satisfaction with care is most strongly correlated with the quality of patient-therapist interactions. This includes the therapist spending adequate time with the patient, demonstrating strong listening and communication skills, and offering a clear explanation of treatment. Non-patient care issues such as clinic location, equipment, and parking are less important in determining patient satisfaction. We did not study patient satisfaction with the outcome of care.

Table 1.
Numbers and Percentages of Respondents by Region

Region                      n       Percentage

Mid-Atlantic (NJ, Pa, NY,
  Md, Del, DC, WVa)         1,260    67.45
South (Fla, Tenn, Ga, NC,
  SC, Tex, Ky)                373    19.97
West (Calif)                  160     8.57
New England (Conn, RI)         75     4.01
Total                       1,868   100.00
Table 2.
Patient Sex by Primary Location of Symptoms

                              Male      Female     Total

Cervical spine                113         66         179
Wrist and hand                 88         88 (a)     176
Lower extremity               280 (a)    149         429
Lumbar and thoracic spine     360 (a)    164         524
Upper extremity               331        188         519
Other                          20         21 (a)      41
Total                       1,192        676       1,868

(a) Observed exceeded expected count.
Table 3.
Inter-item Correlation Matrix Representing the Correlation (r) of the
Items With Each of the 2 Global Measures ("Overall, I Am
Completely Satisfied With the Services I Receive From My Therapist~
[Overall Satisfaction] and "I Would Return to This Office for Future
Care" [Would Return]) and With the Mean of These 2 Measures (a)

                                 Overall        Would
Item                             Satisfaction   Return     Mean

PT answers my questions          .703           .681       .692
PT explains treatment            .722           .625       .674
PT listens                       .712           .612       .662
PT is courteous                  .695           .605       .650
PT spends enough time            .682           .597       .640
PT gives detailed instructions   .653           .548       .600
PT advises me                    .579           .522       .551
Office staff is courteous        .527           .503       .515
The office is clean              .516           .482       .499
Up-to-date equipment             .520           .456       .488
The receptionist is courteous    .428           .413       .421
The office hours are
  convenient for me              .427           .410       .419
Registration is simple           .398           .391       .395
Waiting room is comfortable      .387           .386       .387
Parking is convenient            .272           .261       .267
PTA is courteous                 .269           .236       .253
Time waiting for PT              .175           .149       .162
Location is convenient           .095           .106       .101

(a) PT=physical therapist, PTA=physical therapist assistant.
Table 4.
Total Variance Explained

                             %          Cumulative
Component (a)   Eigenvalue   Variance   %

1               7.61         42.26      42.26
2               1.42          7.86      50.12
3               1.24          6.91      57.03

(a) Component 1 = personal aspects (7 items), component 2=system/
external aspects (3 items), component 3=convenience of receiving
care (2 items).
Table 5.
Factor Loadings by Component for the 3-Component Solution (a)

                                             Factor
Item                             Component   Loading

PT explains treatment            1           .874
PT answers my questions          1           .866
PT listens                       1           .858
PT is courteous                  1           .829
Time waiting for PT              3           .792
PT spends enough time            1           .797
Location is convenient           3           .779
PT gives detailed instructions   1           .779
Registration is simple           2           .776
The receptionist is courteous    2           .763
PT advises me                    1           .759
The office is clean              *           .708 (b)
Office staff is courteous        *           .700 (b)
Waiting room is comfortable      2           .732
Up-to-date equipment             *           .620 (b)
The office hours are
  convenient for me              *           .536
Parking is convenient            *           .534
PTA is courteous                 *           .410

(a) PT=physical therapist, PTA=physical therapist assistant.
Asterisk indicates items that failed to load on any component.

(b) Cross-loaded on 2 or more components.
Table 6.
Internal Consistency (Alpha) Reliability Characteristics of 2- and 3-
Component Solutions and Their Associated Subscales (a)

Scale                   Alpha  [bar]X  SD      SEM

2-component overall
  (10 items)            .9044    4.43   0.611  .189
3-component overall
  (12 items)            .8461    4.42   0.559  .219
PT-patient interaction
  subscale (7 items)    .9163    4.44   0.693  .201
Clinic environment
  subscale (3 items)    .7381    4.41   0.645  .331
Convenience
  subscale (2 items)    .4368    4.36   1.020  .765

(a) SEM=standard error of measurement, PT=physical therapist.
Table 7.
Pearson Correlation Coefficient (4 and Spearman Rank Correlation
([rho]) Between 2- and 3-Component Solutions and the Global Measures
("Overall, I Am Completely Satisfied With the Services I Receive From
My Therapist" [Overall Satisfaction] and "I Would Return to This
Office for Future Care" [Would Return])

              Overall       Would
              Satisfaction  Return        Mean

Solution        r    [rho]    r    [rho]    r    [rho]

2-component   .801   .760   .711   .716   .756   .738
3-component   .781   .737   .695   .695   .738   .716

* MedRisk Inc, 640 Freedom Business Center, Ste 300, King of Prussia, PA 19406-1332

References

(1) Goldstein MS, Elliott SD, Guccione AA. The development of an instrument to measure satisfaction with physical therapy. Phys Ther. 2000;80:853-863.

(2) Patient Satisfaction Instruments: A Compendium. Alexandria, Va: American Physical Therapy Association; 1995.

(3) Pinto MB. Outcome measures in assessing physical therapy. Orthop Clin North Am. 1995;4:269-281.

(4) Roush SE, Sonstroem RJ. Development of the Physical Therapy Outpatient Satisfaction Survey (PTOPS). Phys Ther. 1999;79:159-170.

(5) Bush T, Cherkin D, Barlow W. The impact of physician attitudes on patient satisfaction with care for low back pain. Arch Fam Med. 1993;2:301-305.

(6) Hudak PL, Wright JG. The characteristics of patient satisfaction measures. Spine. 2000;25:3167-3177.

(7) Keith RA. Patient satisfaction and rehabilitation services. Arch Phys Med Rehabil. 1998;79:1122-1128.

(8) Nitse PS, Rushing V. Patient satisfaction: the new area of focus for the physician's office. Health Mark Q. 1996;14(2):73-84.

(9) Linder-Pelz S, Struening EL. The multidimensionality of patient satisfaction with a clinical visit. J Community Health. 1985;10:42-54.

(10) Hall JA, Dornan MC. Meta-analysis of satisfaction with medical care: description of research domain and analysis of overall satisfaction levels. Soc Sci Med. 1988;27:637-644.

(11) Olejnik S, McKinely CO, Ellis RA, et al. Construct validation of customer satisfaction inventories. Journal of Rehabilitation Outcomes Measurement. 1998;2 (5):30-38.

(12) Baker R. Development of a questionnaire to assess patients' satisfaction with consultations in general practice. Br J Gen Pract. 1990;40: 487-490.

(13) DiTomasso RA, Willard M. The development of a patient satisfaction questionnaire in the ambulatory setting. Fam Med. 1991;23: 127-131.

(14) Ware JE Jr, Hays RD. Methods for measuring patient satisfaction with specific medical encounters. Med Care. 1988;26:393-402.

(15) Cherkin DC, MacCornack FA. Patient evaluations of low back pain care from family physicians and chiropractors. West J Med. 1989;150: 351-355.

(16) Deyo RA, Diehl AK. Patient satisfaction with medical care for low-back pain. Spine. 1986;11:28-30.

(17) Elliott-Burke TL, Pothast L. Measuring patient satisfaction in an outpatient orthopedic setting, part 1: key drivers and results. Journal of Rehabilitation Outcomes Measurement. 1997; 1:18-25.

(18) Hall JA, Dornan MC. What patients like about their medical are and how often they are asked: a meta-analysis of the satisfaction literature. Soc Sci Med. 1988;27:935-939.

(19) Pinto MB, Leonidas L. The impact of office characteristics on satisfaction with medical care: a "before and after" analysis. Health Mark Q. 1994;12:43-54.

(20) Safran DG, Taira DA, Rogers WH, et al. Linking primary care performance to outcomes of care. J Fam Pract. 1998;47:213-220.

(21) Sixma HJ, Spreeuwenberg PM, van der Pasch MA. Patient satisfaction with the general practitioner: a two-level analysis. Med Care. 1998;36:212-229.

(22) Smucker DR, Konrad TR, Curtis P, Carey TS. Practitioner self-confidence and patient outcomes in acute low back pain. Arch Fam Med. 1998;7:223-228.

(23) Dilman DA. Mail and other self-administered questionnaires. In: Rossi PH, Wright JD, Anderson AB, eds. Handbook of Survey Research. New York, NY: Academic Press; 1982:68-95.

(24) Nunnally JC. Psychometric Theory. New York, NY: McGraw-Hill Inc; 1978.

(25) Domholdt E. Physical Therapy Research: Principles and Applications. Philadelphia, Pa: WB Saunders Co; 1993:290-294.

(26) Hair JF Jr, Anderson RE, Tatham RL, Black WC. Multivariate Data Analysis With Readings. 3rd ed. New York, NY: Macmillan Publishing Co; 1987.

(27) Roddey TS, Olson SL, Cook KF, et al. Comparison of the University of California-Los Angeles Shoulder Scale and the Simple Shoulder Test with the Shoulder Pain and Disability Index: single-administration reliability and validity. Phys Ther. 2000;80:759-768.

(28) McHorney CA, Tarlov AR. Individual-patient monitoring in clinical practice: are available health status surveys adequate? Qual Life Res. 1995;4:292-307.

(29) Diamond JJ. A practical application of reliability theory to family practice research. Fam Pract Res J. 1991;11:357-362.

(30) Rothstein JM, Echternach JL. Primer on Measurement: An Introductory Guide to Measurement Issues. Alexandria, Va: American Physical Therapy Association; 1993:64.

(31) Cleary PD, Edgman-Levitan S. Health care quality: incorporating customer perspectives. JAMA. 1997;278:1608-1612.

(32) Baker J, Lamb CW Jr. Physical environment as a hospital marketing tool. J Hosp Mark. 1992;6:25-35.

(33) Bitner MJ. Evaluating service encounters: the effects of physical surroundings and employee responses. Journal of Marketing. 1990; 54(April):69-82.

(34) Brand RR, Cronin JJ, Routledge JB. Marketing to older patients: perceptions of service quality. Health Mark Q. 1997;15:1-31.

(35) Gardner K. Patients decision to sue often hinges on therapeutic relationship. American Academy Pediatric News. 1998;14(10):13.

(36) Hawes JM, Prough GE. Analyzing the market for dental services. Health Mark Q. 1987-88;5(1-2):171-182.

(37) Rao CP, Rosenberg LJ. Consumer behavior analysis for improved dental services marketing. Health Mark Q. 1986;3(Summer):83-96.

(38) Rubin HR. Patient evaluations of hospital care: a review of the literature. Med Care. 1990;28:S3-S9.

(39) Simon SE, Patrick A. Understanding and assessing consumer satisfaction in rehabilitation. Journal of Rehabilitation Outcomes Measurement. 1997;1(5):1-14.

PF Beattie, PT, PhD, OCS, is Clinical Associate Professor, Program in Physical Therapy, Department of Exercise Science, School of Public Health, University of South Carolina, Columbia, SC 29208 (USA) ([e-mail]). Address all correspondence to Dr Beattie.

MB Pinto, PhD, is Assistant Professor of Marketing, School of Business, Penn State Erie, Erie, Pa.

MK Nelson, PT, PhD, CPA, is Associate Professor, School of Business Administration Franklin & Marshall College, Lancaster, Pa.

R Nelson, PT, PhD, FAPTA, is Vice President, Expert Clinical Benchmarks, MedRisk Inc, King of Prussia, Pa.

All authors provided concept/research design and consultation (including review of manuscript before submission). Dr Beattie, Dr Pinto, and Dr Martha Nelson provided writing. Dr Roger Nelson provided data collection, fund procurement, and facilities/equipment. Dr Beattie and Dr Roger Nelson provided data analysis. Dr Beattie and Dr Martha Nelson provided project management. Dr Roger Nelson and Claire Coyne provided clerical/secretarial support. The authors thank Ann Wilhelm, Rosselle Perrucci, PT, MBA, Shelly Boyce, Claire Coyne, and Paul Stratford, PT, for their valuable assistance on this project. Dr Beattie and Dr Pinto were paid consultants for this project.

This study was approved by the Human Subject Policy Review Board of MedRisk Inc.

This study was supported by MedRisk Inc, 640 Freedom Business Center, Ste 300, King of Prussia, PA 19406-1332.

This article was submitted May 2, 2001, and was accepted November 12, 2001.