Error bars represent SD (0.74 in period A and 0.66 in period B).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Identify all potential conflicts of interest that might be relevant to your comment.
Conflicts of interest comprise financial interests, activities, and relationships within the past 3 years including but not limited to employment, affiliation, grants or funding, consultancies, honoraria or payment, speaker's bureaus, stock ownership or options, expert testimony, royalties, donation of medical equipment, or patents planned, pending, or issued.
Err on the side of full disclosure.
If you have no conflicts of interest, check "No potential conflicts of interest" in the box below. The information will be posted with your response.
Not all submitted comments are published. Please see our commenting policy for details.
Louie CE, Kelly JL, Barth RJ. Association of Decreased Postsurgical Opioid Prescribing With Patients’ Satisfaction With Surgeons. JAMA Surg. 2019;154(11):1049–1054. doi:10.1001/jamasurg.2019.2875
Is decreased opioid prescribing by surgeons associated with lower patient-reported satisfaction ratings of their surgeon?
In this survey study, surgeons could greatly decrease the proportion of their patients receiving opioids and the number of pills prescribed with no significant change in their satisfaction ratings.
Concern regarding clinician satisfaction scores should not be a barrier to appropriate opioid prescribing by surgeons.
Opioid overdose is the leading cause of injury-related death in the United States. Several studies have shown that surgeons overprescribe opioids, and guidelines for appropriate opioid prescribing are available. Concern about patient-reported satisfaction scores may be a barrier to surgeons adopting guideline-directed prescribing.
To determine whether decreased opioid prescribing is associated with a decrease in patient-reported satisfaction with their surgeon.
Design, Setting, and Participants
Retrospective analysis of clinician satisfaction scores at Dartmouth-Hitchcock Medical Center obtained in 2 periods: 1 before (period A) and 1 after (period B) an educational intervention that resulted in decreased opioid prescribing. The analysis included 11 surgeons who performed 5 common outpatient general surgical operations on 996 patients. Data were analyzed between March and August 2018.
Main Outcomes and Measures
Patient-reported overall satisfaction rating of the surgeon (scale, 0-10). This was collected by a nonstudy-related, routine general institutional survey of approximately 40% of all outpatient encounters.
Of the total number of patients, 67% were women (667 of 996), and the mean patient age was 58 years. Comparing period A with B, the proportion of patients prescribed opioids decreased from 90.2% (n = 367 of 407) to 72.8% (n = 429 of 589) (P < .001). The mean number of opioid pills per prescription decreased from 28.3 to 13.3 (P < .001) and significantly decreased for each of the 11 surgeons. One hundred five of 996 patients (10.5%) undergoing index operations responded to the survey. There was no difference in the mean clinician satisfaction ratings from period A vs B (9.70 vs 9.65; P = .69). During the study periods, 640 total surveys were collected referencing these surgeons (including outpatient encounters associated with operations other than the 5 index cases). There was no difference in the mean satisfaction ratings from period A vs period B (9.55 vs 9.59; P = .62). When individual clinicians were analyzed, none had a significant difference in overall satisfaction rating from period A vs period B.
Conclusions and Relevance
Despite a marked decrease in the proportion of patients receiving opioids and in the number of pills prescribed, there was no significant change in clinician satisfaction ratings.
Opioid misuse, addiction, and overdose is an ongoing and rapidly evolving public health crisis in the United States. Opioid overdose is now the leading cause of injury-related death in the United States, surpassing deaths from motor vehicle crashes.1 Every day more than 115 people die of opioid overdose in the United States.2 Deaths secondary to prescription overdose in particular have increased by more than 400% since 1999, reaching nearly 19 000 annually.1,3 This rate has risen in parallel with that of opioid prescribing and rates of opioid addiction.4
There is a growing consensus that surgeons have been prescribing more opioids than are needed to treat postoperative pain.5-9 Overprescribing may lead to high rates of prolonged opioid use after surgery and provides a supply of pills that can be diverted to long-term users.10,11 Physicians clearly have a critical role in this epidemic, reflected in the US Food and Drug Administration statement that “until clinicians stop prescribing opioids far in excess of clinical need, this crisis will continue unabated.”12
However, surgeons may be reluctant to prescribe fewer opioids out of concerns for their patients, whose pain may not be adequately controlled. Surgeons may also be concerned that inadequate pain control will reflect negatively on themselves. It is now routine to administer surveys for monitoring patient experiences after medical encounters, resulting in hospital and physician-specific ratings. These ratings are increasingly available publicly online and may influence patients when selecting clinicians. Patient satisfaction scores may also be used in assessments of clinician performance. Some satisfaction scores, such as the Centers for Medicare & Medicaid Services Hospital Consumer Assessment of Healthcare Providers and Systems (HCAHPS) survey, are used in hospital reimbursement.13
There is a paucity of information on the association between opioid prescribing and patient-reported satisfaction measures. While 1 study showed that lower postoperative opioid prescriptions were not associated with lower pain management scores, another study did show a link between opioid prescriptions and higher patient satisfaction.14,15Quiz Ref ID The purpose of this study was to determine whether significantly reduced opioid prescribing by surgeons is associated with a decrease in clinician satisfaction scores.
Quiz Ref IDAfter obtaining approval from the Dartmouth Committee for the Protection of Human Subjects, we analyzed the 5 most commonly performed outpatient general surgery procedures at Dartmouth-Hitchcock Medical Center, Lebanon, New Hampshire: partial mastectomy (PM), partial mastectomy with sentinel lymph node biopsy (PM SLB), laparoscopic cholecystectomy (LC), laparoscopic inguinal hernia repair (LIH), and open inguinal hernia repair (IH). A total of 996 operations during these periods were analyzed. These operations were performed by 11 different surgeons: 4 surgical oncologists, 3 minimally invasive surgeons, and 4 general surgeons. These surgeons also performed other outpatient procedures and a wide array of operations requiring postoperative inpatient stays. Postoperative opioid prescriptions entered into the electronic medical record were tabulated for these index operations during the preintervention period of May 1, 2015, to December 31, 2015 (period A), and the postintervention period of July 1, 2016, to June 20, 2017 (period B).
In April and May 2016, we conducted an educational intervention for clinicians at our institution based on our findings of variability and overprescribing, and recommended opioid prescribing guidelines for each of the aforementioned index procedures.5,16 We previously reported that this intervention resulted in a greater than 50% decrease in the number of opioid pills prescribed for patients undergoing these operations during the 4-month period of June 1, 2016, to Sept 31, 2016.16
Oxycodone tablets (5 mg) were used as the standard dose to describe the number of pills prescribed. All opioid use was converted to an equianalgesic equivalent (ie, 1 pill = 5 mg oxycodone = 7.5 mg morphine equivalent).17 Our 2016 guidelines specified the following number of opioid pills: 5 for PM, 10 for PM SLB, 15 for LC, 15 for LIH, and 15 for IH.
Patient consent was waived because the study was considered to be a minimal risk research project because the research involved materials collected for routine, nonresearch purposes. Patient satisfaction surveys formulated by the Press Ganey Company were routinely sent to some Dartmouth-Hitchcock patients seen in outpatient encounters. Patients were sent a letter with the survey questions. If they did not return the questionnaire within 4 weeks, a second letter was sent. The survey data were collected independent of this research effort by the Press Ganey Company, and results from patients who responded to the survey were provided to the Dartmouth-Hitchcock Data Analytics Institute (DAI). When patients were surveyed, they were informed that their responses would be kept anonymous. Therefore, the DAI could provide this research team with only summary information. We were provided with information on the satisfaction score, the surgeon, and the operation but were not given the patient name, so we were not able to compare responders with nonresponders with regard to other specific patient characteristics. Press Ganey did not inform the DAI whether a particular patient was sent a survey, so we cannot precisely calculate the survey response rate. We do know that the Dartmouth-Hitchcock institution-wide response rate during periods A and B were 27.6% and 26.6%. Thus, there was no difference in response rates between periods A and B. Assuming that the response rate for patients who were in our study cohort were the same as the institution response rates, and the fact that 105 responses were received, we can infer that approximately 400 of 996 patients (40%) were surveyed. The DAI provided us with data from surveys collected from the periods of June 1, 2015, to January 31, 2016 (correlating with cases done in period A), and August 1, 2016, July 31, 2017 (correlating with cases done in period B).
We compared the Overall Clinician Satisfaction Rating (which is registered on a scale of 0-10) between periods A and B, stratified both by surgeon and by procedure type. On this scale, a score of 0 was the worst and 10 was the best. The Overall Clinician Satisfaction Rating question was asked in the survey in exactly the same way during periods A and B. The survey question was “rate your clinician on a 0-10 scale.”
For continuous variables, means were compared using the t test. Proportions were compared using the χ2 statistic. Two-tailed P values less than .05 were considered significant. The false discovery rate method was used to adjust for multiple comparisons. Both the raw and false discovery rate method–adjusted P values are presented in Table 1, Table 2, and Table 3.
Eleven surgeons from our department performed 996 index operations: 242 PM, 182 PM SLB, 315 LC, 144 LIH, and 113 IH. In period A (prior to the educational intervention), 407 operations were performed; 589 were performed in period B (after the educational intervention). The characteristics of the patients in periods A and B are described in Table 4. When comparing periods A vs B, there was no significant difference in mean (SD) patient age (57  years vs 58  years), proportion of patients with a history of opioid abuse (0% [0 of 407] vs 1% [4 of 589]) or history of opioid use within 30 days of surgery (5% [20 of 407] vs 6% [33 of 589]). The proportion of patients who were men was slightly higher in period B (28% [114 of 407] vs 36% [212 of 589]). There was a slight change in the relative frequency of operations performed: the frequency of PM slightly decreased (30% [121 of 407] to 21% [121 of 589]) and the frequency of LIH repair slightly increased (10% [40 of 407] to 18% [104 of 589]). There was no significant change in the frequency of the other 3 operations. There was no significant change in the percentage of patients contributed by each of the 11 surgeons in the 2 periods.
As shown in Table 1, Quiz Ref IDthe percentage of patients prescribed an opioid was significantly lower in period B than in period A (72.8% [429 of 589] vs 90.2% [367 of 407]; P < .001). Of the patients who received opioids, the mean (SD) number of pills prescribed per patient was 53% lower in period B vs period A (13.3  vs 28.3  pills; P < .001; Table 2). For patients who were prescribed opioids, fewer opioid pills were prescribed in period B for each of the 5 index procedures: PM, 21 vs 8; PM SLB, 22 vs 9; LC, 34 vs 16; LIH, 34 vs 15, IH, 31 vs 15; P < .001 for all comparisons.
The percentage of patients prescribed opioids and the mean number of pills prescribed (for those patients who had prescriptions written) for each of the 11 surgeons are detailed in Tables 1 and 2. All 11 surgeons significantly decreased the mean number of pills they prescribed, even when P values were adjusted for multiple comparisons (Table 2). Ten of 11 surgeons decreased the mean number of pills per prescription by 50% or more. Ten of 11 surgeons prescribed opioids to a lower proportion of their patients; this was statistically significant in 3 of 11 surgeons (27%) (Table 1).
Quiz Ref IDOf the 996 index case patients, 105 responded to the satisfaction survey (10.5%), 37 in period A and 68 in period B. There was no difference in the mean (SD) clinician satisfaction ratings from period A vs period B (9.70 [0.7] vs 9.65 [0.7]; P = .69, Figure). The median clinician satisfaction ratings in periods A vs B were 10 vs 10, respectively (interquartile range, 10-10 vs 9.75-10 and range, 7-10 vs 8-10, respectively).
During periods A and B, there were 640 total surveys collected referencing the 11 clinicians (including outpatient encounters associated with operations other than the 5 index cases). There was no overall difference in the mean (SD) satisfaction ratings for all surveys collected from period A vs period B (9.55 [1.3] vs 9.59 [0.9]; P = .62; Table 3). The median clinician satisfaction ratings in periods A vs B were 10 vs 10, respectively (interquartile range, 10-10 vs 9-10 and range, 0-10 vs 0-10, respectively).
The mean satisfaction scores for each of the 11 clinicians are shown in Table 3. One clinician did not have any surveys returned for period A. Quiz Ref IDWhen corrected for multiple comparisons, there was no significant change in the satisfaction scores of any of the remaining 10 clinicians.
Notably, only 6 of 996 patients (0.6%) required an opioid prescription refill: 3 of 407 in period A and 3 of 589 in period B. Of the 6 patients requiring refills, 4 underwent LC, 1 patient had a PM, and 1 patient had an IH.
In light of the nationwide emphasis on prudent opioid prescribing, there is reasonable concern among prescribers regarding appropriate pain management and patient satisfaction. We found that educating surgeons on how many opioids patients actually used after surgery led to a dramatic and durable decrease in opioid prescribing. Opioid prescriptions dropped to fewer than half of the baseline number. The mean number of pills prescribed significantly decreased for all 5 of the operations we analyzed and for all 11 surgeons in our study, indicating that this effect is generalizable across multiple clinicians and procedures. We had previously reported that this educational intervention decreased opioid prescriptions for a 4-month period; we now show that this effectively decreased opioid prescribing for at least 1 year.16
While we achieved these results at 1 institution, it is unclear whether surgeons in general will adopt new, appropriate opioid prescribing guidelines. One potential obstacle to adoption is the concern that prescribing fewer opioids may result in lower satisfaction scores. Patient satisfaction metrics have been a notable measure in quality health care improvement initiatives and are often factors in determining hospital reimbursement, physician ratings, compensation, and promotion. Results from patient satisfaction surveys, ie, “overall satisfaction rating,” as in our study or more specific questions regarding pain management, can influence health care practices.18,19 We decided to focus on the Overall Clinician Satisfaction Rating for several reasons. This question was asked in the exact same way during the entire time frame of our study, whereas other questions, including those that specifically asked about pain symptoms, were not consistently asked in the same fashion. Furthermore, the Overall Clinician Satisfaction Score was the score that was most likely to be posted on rating scales and was most important to the clinician because it serves as a summary score. Of note, owing to concerns about the potential effect of satisfaction scores on opioid prescribing, in 2018 the Centers for Medicare & Medicaid Services adjusted the HCAHPS survey dimension on pain management to instead focus on pain communication practices.20
We found that opioid prescribing can be markedly decreased (by more than 50%) and not adversely affect clinician satisfaction scores. One of the potential reasons that satisfaction scores did not decrease is that even with fewer prescriptions, we were still adequately taking care of patients’ pain. When we first instituted the guidelines, we were not sure that we were going to adequately treat all patients’ pain: we developed the guidelines to provide enough opioids to satisfy 80% of patients. However, in period B, after the educational intervention, which resulted in a decrease in opioid prescriptions by more than 50%, only 3 of 589 patients (0.51%) required a refill prescription. It is clear that we were adequately treating their pain. This is likely because we encouraged prescribers to set patients’ expectations regarding postoperative pain, discuss the dangers of opioid medications, and promoted the use of nonopioid analgesics (ie, acetaminophen and ibuprofen) in conjunction with these opioid prescribing guidelines. Surgeons should not be concerned about patient satisfaction scores if they prescribe an appropriate number of pills and set appropriate patient expectations.
When prescribing fewer opioids it is important to also manage patient expectations about postoperative pain. Research suggests that fulfilling patient expectations is often associated with with more satisfied patients, and conversely, nonfulfillment is associated with patient dissatisfaction, ultimately reflected in satisfaction scores.21 In addition to prescribing fewer opioids, we began to have discussions with patients prior to surgery, describing how much pain they could expect and how we were going to manage their pain with nonopioid analgesics, such as ibuprofen and acetaminophen, in addition to appropriate opioids. In effect, we assured patients that we cared about their postoperative pain.
One limitation of this study is that the results are from 1 institution. Another limitation is that we used survey data obtained by a company independent of our institution. The strengths of this approach were that responders knew their answers would be kept anonymous, the survey was not influenced by the research team, and it consisted of real-world survey results that are actually used to determine clinician satisfaction scores on our institution’s website. A disadvantage of using this survey data, because patient anonymity was maintained, is that we could not compare characteristics of responders vs nonresponders and could not evaluate the possible effects of certain patient characteristics on satisfaction scores. Furthermore, assuming that the response rate from our index surgery cases was similar to the overall institutional response rate, the response rate was relatively low (approximately 25%). There may be an unpredictable bias among the patients who submitted a satisfaction survey. Additionally, when we considered all satisfaction survey results (Table 3), including patients who had undergone operations other than the index cases during these periods, we assumed that clinicians had prescribed fewer opioids to these non–index case patients. At least 1 study suggests that this is a valid assumption.22 When prescribing guidelines were introduced for laparoscopic cholecystectomy at the University of Michigan, opioid prescribing decreased for 4 other surgical procedures (laparoscopic sleeve gastrectomy, laparoscopic appendectomy, LIH, and thyroidectomy).22
Our intervention of developing standardized opioid prescribing guidelines, aligning prescribing with patient use of opioids, substantially reduced opioid prescribing. These data suggest that surgeons in every discipline can markedly decrease opioid prescribing by first determining a reasonable opioid standard for the operations they perform and then disseminating the guidelines to prescribers in their specialty. Furthermore, we now show that substantial reductions in opioid prescribing were not associated with patient satisfaction scores. Concern about clinician satisfaction scores should not be a barrier to reducing opioid prescribing by surgeons and thus decreasing the pool of excess opioid pills available for misuse, diversion, and overdose.
Corresponding Author: Richard J. Barth Jr, MD, Section of General Surgery, Dartmouth-Hitchcock Medical Center, One Medical Center Dr, Lebanon, NH 03756 (email@example.com).
Accepted for Publication: May 26, 2019.
Published Online: October 16, 2019. doi:10.1001/jamasurg.2019.2875
Author Contributions: Dr Barth had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Barth.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Louie, Barth.
Critical revision of the manuscript for important intellectual content: Kelly, Barth.
Statistical analysis: Louie, Kelly.
Administrative, technical, or material support: Louie, Barth.
Conflict of Interest Disclosures: None reported.