Data Collection

We conducted a retrospective cohort study of adult patients who were seen for initial consultations for back pain in our spine clinic between 2011 and 2013. Patients who had not been seen previously in the spine clinic and those who were not seen by a spine surgeon at the University of Utah in the previous 3 years were included if they received a recommendation for or against surgical intervention. Each patient was evaluated by one of 5 neurosurgeons. Of these patients, we included in our analysis all patients who completed an Oswestry Disability Index (ODI) questionnaire before their consultation and a patient satisfaction survey after their clinic appointment. Patients were mailed the ODI questionnaire 1 week prior to their clinic appointment and were asked to bring the completed questionnaire to their appointment. Patients who forgot their ODI questionnaires were asked to complete one when they checked in on the day of their clinic appointment. Satisfaction surveys were sent by mail approximately 1 week after the clinic appointment, and these were returned via postage-paid envelopes. This is the standard method of survey administration used at our institution for assessment of patient satisfaction. No additional reminders or incentives were used to encourage survey participation.

A Press Ganey medical practice survey (Press Ganey Associates, Inc.) was used as the instrument to measure outpatient patient satisfaction. This survey is divided into 6 sections, each comprising 2–10 items: access, moving through your visit, nurse/assistant, care provider, personal issues, and overall assessment. Each item corresponds to one of several components of an outpatient health care encounter such as ease of scheduling appointments and courtesy of clerical staff (access), wait times and delays (moving through your visit), interactions with the nurse or medical assistant, the patient-provider encounter, staff attentiveness to privacy and patient safety (personal issues), the ability of the staff to work together, and the likelihood of recommending the practice to others (overall assessment). Scores were transformed from the 5-point Likert scale used in the medical practice survey to a 100-point satisfaction scale graded from 0 (very poor) to 100 (very good). The mean score of each section was used in the data analysis. A composite average of each section score was also calculated. Sections with incomplete responses were excluded from analysis.

Disability was self-assessed using the Oswestry Disability Index (ODI), a 10-question, low back–specific instrument used to measure self-rated pain and disability.⁷ Scores are based on patient responses to 10 sections designed to assess pain, function, and limitations of various activities of daily living. The ODI is deliberately focused on physical activities and not the psychological consequences of acute or chronic pain.⁷ We used the ODI version 2.0 with some minor modifications for brevity.

Analysis

Our first objective was to determine whether high self-assessment of disability, as measured by the ODI, was associated with lower satisfaction scores in patients referred to a spine surgeon for back pain. The ODI scoring results were transformed to percentage scores to account for questions that were unanswered. Patients were grouped by severity of disability according to the ODI percentage score, in a method used by others:^7,36 mild to moderate disability (ODI score < 40%) and severe disability (≥ 40%). Satisfaction scores were compared between groups. Our second objective was to determine whether the neurosurgeon's recommendation for or against surgical intervention affected satisfaction scores. Patients were then stratified according to the surgeon's treatment recommendation, and satisfaction scores were compared. Nonparametric tests were used for statistical analysis because satisfaction scores had a skewed distribution. The Mann-Whitney- Wilcoxon test was used to determine whether disability severity (mild to moderate vs severe) or the surgeon's recommendation for or against surgery were associated with satisfaction scores. Multivariate linear regression analysis was performed to identify potential confounders, such as age, sex, surgeon, and previous spine surgery. A p value < 0.05 was considered statistically significant. Calculations were performed with R software (version 3.0.2, R Foundation for Statistical Computing, http://www.R-project.org).

This study was conducted in accordance with the policies of the institutional review board at our institution.

Self-Assessed Disability and Satisfaction

We found that among patients presenting for spine consultations, high self-assessment of disability was associated with lower patient satisfaction. Patients with severe disability (ODI ≥ 40%) reported lower satisfaction scores than those with mild to moderate disability (ODI < 40%). These findings are supported by multiple studies identifying that patients who have worse physical health are less satisfied with their medical care than patients in better health.^{13,17,20,26,32} Sicker patients may develop a negative affect that becomes attached to the care provider and to their encounters with the health care system, which may translate to lower satisfaction scores.^15,26 Dissatisfaction tends to persist in patients with poor health across a variety of health care environments.¹⁸ Consequently, investigators using satisfaction to evaluate the process, content, or structure of medical care should control for severity of disability, along with health status, age, sociodemographics, and other correlates of satisfaction to avoid incorrectly attributing problems to the health care system that may actually reside in the patient.^{5,6,11,17,18,26,37}

Because surveys of patients are increasingly used to measure health care quality, verifying that patients' responses serve as a reliable indicator for quality is crucial.^5,29,35,40 Satisfaction surveys seek to measure the consumer experience as it relates to the administrative processes needed to deliver care; however, patients can describe high levels of satisfaction while also describing experiences that are suboptimal.²² The satisfaction survey we used in this study attempted to capture the experience both as a patient and as a consumer. The physician encounter encompasses only one of several components of the health care experience that the patient is asked to rate. Other survey items addressed ease of scheduling appointments and courtesy of clerical staff, wait times and delays, interactions with the nurse or medical assistant, attentiveness to privacy, needs, and patient safety, staff cohesiveness and ability to work together, and the likelihood of recommending the practice to others. Knowledge of how these surveys are conducted, the items measured in the surveys, and the factors associated with positive and negative scores are essential for physicians to compete more effectively as performance evaluations and reimbursements become linked to survey results. This policy can be expected to continue as studies report that satisfaction is highly correlated with willingness to recommend the physician or practice to others.^22,34

In our study population, severity of disability was associated with lower patient satisfaction for the following sections: “Personal issues,” “Moving through your visit,” “Care provider,” “Overall assessment,” and for the composite average satisfaction score. Patients who received a recommendation against surgical intervention also had lower satisfaction scores for 4 of 6 survey items. Although the absolute value of the differences between groups may be small, in surveys of patient satisfaction even small differences can have a large impact. Assessments of satisfaction tend not to have normal distributions, and the results are often skewed.^9,22 For example, because provider scores tend to be very positive, even slightly lower scores may translate to large differences in percentile ranking among peers. As health care systems move toward linking reimbursement with survey results, physicians should be aware of the factors that may negatively affect their scores.

Timeliness

The “Moving through your visit” section consists of 2 items that address wait time and delays. We found that patients with higher self-assessment of pain and disability reported lower scores for these survey items. There is a strong inverse relationship with wait times and patient satisfaction in physician offices.^23,31 This relationship applies to the amount of time patients spend in both the waiting room and the examination room.²³ In our study population, the negative impact of poor health status appears to magnify this relationship further. Patients who spend less time waiting are more likely to recommend the practice to others.³³ The Institute of Medicine (IOM) identified timeliness as one of the 6 aims for improving the quality of health care.¹⁹ Practices that decrease wait times have seen improvements in the overall satisfaction of their patients,³¹ but despite its relevance to patient satisfaction, timeliness of care in the clinic setting is among the least studied IOM aims.²³

A key component of the wait time issue is perception. Patients who have a scheduled appointment at a certain time may perceive wait times to be longer and become dissatisfied because of a perception that the provider is breaking a promise. Conversely, patients may tolerate waiting as long as the feel they get adequate time with their physician,¹ although this will negatively affect how long other patients have to wait to see that physician. The literature on marketing suggests that the perception of waiting time can be altered. Some propose that changes that make the service environment more pleasing, such as music, lighting, temperature, color, and furnishings, may shorten the perceived wait time or at least make it more bearable.^3,28,41 Informing the patient with frequent updates and keeping their anxieties in mind can also increase satisfaction.

Patient Expectations

Patient expectations about treatment have a complex relationship with satisfaction.^12,24,25,38 Some researchers have found that satisfaction is closely related to expectations patients have about the care to be provided and that the outcome can be less important.^5,16,39 Patients generally have high expectations for improvement with spine surgery, sometimes to an unrealistic extent.³⁰ Patients may be dissatisfied when the recommended treatments, or lack thereof, do not meet their expectations. In an era in which assessment tools have been developed to measure expectation fulfillment as one measure of quality care,¹⁰ more precisely understanding the relationship between patient expectations and satisfaction for spinal disorders is important.

We found that among patients referred to neurosurgeons for a surgical opinion, patients who received a recommendation against surgical intervention reported lower satisfaction scores than those who received a recommendation for surgery. Patients who are referred to a surgeon for a consultation may have the expectation that a surgical intervention is necessary to help manage their pain and disability. When such patients are not offered surgery, dissatisfaction may result from unmet expectations. The Press Ganey medical practice survey does not include items that address patient expectations explicitly, so the underlying reasons why patients who were not offered surgery had lower satisfaction scores are not apparent. Moreover, lower satisfaction scores cannot be explained by unmet expectations alone. We surveyed patients presenting for new spine consultations. Prior research has shown that people tend not to have fully developed expectations when encountering a system for the first time.² Expectations emerge during the interaction with the physician rather than beforehand.³⁸ Therefore, counseling these patients on nonoperative treatment strategies and educating them on why surgery is not recommended may help improve their experience. Nonetheless, future studies evaluating patient satisfaction should determine the underlying reasons for high or low satisfaction scores and whether there is an association between satisfaction and met or unmet expectations.

Limitations

One of the limitations to this study is the use of the ODI to quantitate disability. Although many investigators have assumed that the ODI correlates linearly with disability, it is not possible to define the mathematical behavior of this scale.⁷ For example, a patient with a percentage score of 40% is unlikely to be twice as disabled as one with a score of 20% because the structure of several items on the ODI is not linear. Therefore, while we found a significant difference between satisfaction in patients with severe pain and disability and satisfaction in those with milder symptoms, this relationship is most likely complex and nonlinear. By aggregating the ODI into two categories of disability severity, we were able to include analysis of the affect of disability on satisfaction without making mathematical assumptions about the behavior of the ODI scale. Other groups have used an ODI of 40% as a criterion for dividing a study population into groups according to severity.^7,36 We consider an ODI percentage threshold of 40% to separate mild to moderate self-assessed disability from severe disability to be appropriate given that one study reported a weighted ODI of 27 after aggregating data from 2166 patients with primary back pain.⁷ Thus, our criterion of ≥ 40% would indicate disability that is more severe than the average patient with back pain.

The timing of survey completion in relation to the clinic visit may also influence the satisfaction score. Surveys administered outside the context of an immediate visit may be a proxy for symptom or functional improvement rather than a measure of satisfaction during the health care encounter.^20,21 If the intention is to measure satisfaction with specific physician behavior or patient experiences, the survey ideally should be administered immediately after the visit. Delays may introduce lower response rates and nonresponse bias. In our study, patients were mailed surveys approximately 1 week after their clinic visit, and the response time was not tracked, but because our study population consisted of patients seen electively in an outpatient clinic for an initial surgical consultation for back pain, the patients should not have had an expectation for improvement in functional status at the time of the survey since no surgical intervention was performed at the clinic visit.

This questionnaire-based study is also subject to recall bias. Study results are derived from voluntary responses. Patients who are either very satisfied or very dissatisfied with their outpatient health care experience may be more likely to respond than those with neutral experiences. Extreme satisfaction scores may be related to whether the surgeon's treatment recommendation matched patients' expectations, but this hypothesis needs to be tested directly in a prospective trial. Whether the demographic characteristics of survey nonresponders were similar to those of responders is unclear. Such information will help ensure that our findings are generalizable. We were not able to review demographic and clinical information on patients who declined to complete satisfaction surveys. Future prospective studies should track relevant information for both responders and nonresponders alike.