Pain that lingers beyond the early weeks after the acute postoperative period is an important risk factor for chronic postsurgical pain. This study examined the hypothesis that patients’ expectations about their postsurgical pain would be independently associated with lingering postsurgical pain.
The study included 3,628 patients who underwent diverse surgeries between February 2015 and October 2016 in a single U.S. tertiary hospital and participated in the Systematic Assessment and Targeted Improvement of Services Following Yearlong Surgical Outcomes Surveys (SATISFY-SOS) observational study. Preoperatively, patients were asked about their expectations about pain 1 month after surgery. Patients were considered to have lingering postsurgical pain if they endorsed having pain in the area related to their surgeries during a follow-up survey obtained 1 to 3 months postoperatively. The independent associations between preselected perioperative variables and lingering postsurgical pain were evaluated.
Of the cohort, 36% (1,308 of 3,628) experienced lingering postsurgical pain. Overall, two thirds (2,414 of 3,628) expected their postsurgical pain to be absent or improved from baseline, and 73% of these had their positive expectations fulfilled. A total of 19% (686 of 3,628) expected new, unabated, or worsened pain, and only 39% (257 of 661) of these had their negative expectations fulfilled. Negative expectations were most common in patients with presurgical pain unrelated to the reason for surgery, undergoing surgeries not typically performed to help alleviate pain. Endorsing negative expectations was independently associated with lingering postsurgical pain (odds ratio, 1.56; 95% CI, 1.23 to 1.98; P < 0.001). Additional major factors associated with lingering postsurgical pain included recollection of severe acute postoperative pain (odds ratio, 3.13; 95% CI, 2.58 to 3.78; P < 0.001), undergoing a procedure typically performed to help alleviate pain (odds ratio, 2.18; 95% CI, 1.73 to 2.75; P < 0.001), and preoperative pain related to surgery (odds ratio, 1.91; 95% CI, 1.52 to 2.40; P < 0.001).
Lingering postsurgical pain is relatively common after diverse surgeries and is associated with both fixed surgical characteristics and potentially modifiable factors like pain expectations and severe acute postoperative pain.
Pain that continues between 1 and 3 months after surgery, also known as lingering pain, is associated with long-term chronic pain after surgery
Factors associated with lingering pain remain poorly understood
In a single-center cohort of 3,628 patients undergoing a broad range of surgeries, 36% experienced lingering pain
Factors associated with lingering pain include negative expectations of pain, preoperative pain related to the surgery, recollection of severe acute postoperative pain, and undergoing a procedure typically performed to help alleviate pain
IT is estimated that 10 to 50% of surgical patients suffer from chronic postsurgical pain, which is pain in the surgical area that persists after the surgical insult has healed.1,2 These patients require more healthcare resources, suffer impaired quality of life, and often become chronic opioid users in the months after their surgeries.1,3–6 Subacute pain that lingers throughout the first weeks to 3 months after surgery is a strong predictor of chronic postsurgical pain, which is typically assessed at least 3 to 6 months after surgery.7–9 Identifying reliable risk factors for lingering postsurgical pain is important because it may help refine potential interventions (educational, pharmacologic, behavioral, and procedural) before pain becomes chronic in those most vulnerable.
A growing body of evidence suggests that patients’ attitudes and expectations about their pain may be associated with their general health outcomes.10–16 For example, optimistic pain expectations have been associated with better quality of life, increased functionality, and overall superior clinical outcomes.16–19 Patients who believed their chronic pain was due to an injury or medical mismanagement had lower pain thresholds, poorer response to treatment, and increased deconditioning.20,21 In addition, fear of surgery has been associated with increased pain, poorer recovery, and worse quality of life 6 months after surgery.17 However, the association between patients’ expectations about their postoperative pain and their actual lingering postsurgical pain outcomes has not been evaluated.
The primary aim of this study was to investigate the independent associations between certain perioperative characteristics (preoperative baseline pain, patients’ expectations about their postoperative pain, and other proposed factors) and lingering postsurgical pain in a diverse surgical cohort. We hypothesized that patients’ expectations about their postsurgical pain would be independently associated with lingering postsurgical pain.
Materials and Methods
Study Population and Design
This observational cohort study included subjects from the Systematic Assessment and Targeted Improvement of Services Following Yearlong Surgical Outcomes Surveys (SATISFY-SOS) project (NCT02032030, ClinicalTrials.gov), which was designed to track surgical patients’ postoperative health and well-being. SATISFY-SOS includes, upon obtaining written informed consent, unselected adult patients receiving anesthesia services for scheduled surgical and interventional procedures at Barnes-Jewish Hospital in St. Louis, Missouri. Patients who were unable to provide informed consent, less than 18 years old, or suffering from dementia were excluded from SATISFY-SOS. All patients enrolled in SATISFY-SOS who completed pre- and postoperative surveys and underwent a single surgical procedure were included in this analysis. Patients were enrolled and completed baseline surveys during their anesthetic preoperative assessment and planning clinic visits, typically 1 to 21 days before their procedures. The preoperative survey included the Veterans RAND 12-item health survey and additional questions about baseline pain, motivations for surgery, and expectations for postoperative pain.22 Starting approximately 30 days after surgery, the patients were emailed a postoperative survey that included questions about their postoperative pain. Nonresponders were offered further opportunities to respond by mail and then by telephone. Additional SATISFY-SOS methodologic details have been published separately.23 This is the second SATISFY-SOS analysis regarding lingering postsurgical pain and the first and principal analysis of the SATISFY-SOS subcohort with pain expectations data.24 The Washington University Institutional Review Board (St. Louis, Missouri; Human Research Protection Office approval No. 201602024) approved this study, and its reporting is compliant with the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.25
The primary outcome of this study was lingering postsurgical pain. Lingering postsurgical pain was defined by a positive endorsement to the follow-up survey question, “Currently, do you have any pain in your surgical incision or area related to your surgery?” Two secondary outcomes were included to assess for clinically meaningful pain and to better understand lingering postsurgical pain in the context of surgeries intended to improve preexisting pain. These secondary outcomes included (1) moderate-to-severe lingering postsurgical pain rated 4 or higher on a 0 to 10 numeric rating scale and (2) lingering postsurgical pain that worsened at least 2 points on a 0 to 10 numeric rating scale compared to the preoperative pain severity.
Baseline pain was determined by asking patients whether they had pain at the time of enrollment and whether or not it was “Related to [their] need for surgery” (Supplemental Digital Content 1, http://links.lww.com/ALN/C573). The primary exposure variable, patients’ preoperative expectations about their postoperative pain, was created by asking patients, “What is your expectation about pain a month after your surgery?” This question was created for the survey because no validated measure existed at the time of survey generation. Survey responses were characterized as being positive, negative, or “Do not know.” Patients who expected no pain or pain improvement were categorized as having positive expectations, and those who expected new, unabated, or worsened pain were categorized as having negative expectations. Patients were also asked whether their motivation for surgery was to “Decrease pain and/or symptoms,” among other options. Baseline emotional health was estimated by the Veterans RAND 12-item mental component score. This score ranges from 0 to 100 and is normalized so that a threshold score of 50 represents the U.S. population mean.22 Lower scores are associated with increased likelihood of affective disorders such as depression.26 Patients’ recollection of severe acute postoperative pain was ascertained from the SATISFY-SOS follow-up survey question, “While still in the hospital after your recent procedure, did you suffer from severe pain that required treatment?” Other covariables of interest, including patient comorbidities and characteristics, were collected from the patients’ electronic medical records.
To account for the intraoperative nociceptive severity of the diverse surgical procedures, the estimated surgical pain of each procedure was categorized as mild, moderate, or severe based on surgical service and procedure (Supplemental Digital Content 2, http://links.lww.com/ALN/C574). Procedures were also manually categorized by whether or not they are typically performed to alleviate pain. These categorizations were formulated from a healthcare provider perspective and may differ from patients’ own motivations. Procedures considered to be typically performed to help alleviate pain included major arthroplasty (shoulder, elbow, hip, knee, and ankle), other joint surgeries (e.g., arthroscopies), spine surgeries, other orthopedic surgeries (e.g., fracture repairs, tendon repairs, osteotomies, carpal/tarsal/cubital tunnel releases), abdominal and inguinal hernia surgeries, peripheral nerve surgeries, and thoracic outlet surgeries. Procedures considered to be not typically performed to alleviate pain included organ transplantation, cancer resections, breast, cardiac, thoracic, intracranial, otolaryngologic, ophthalmologic, and most major vascular surgeries (e.g., abdominal aortic aneurism repair, carotid endarterectomy), endoscopies, bronchoscopies, and interventional cardiology procedures. The remainder were considered on a case-by-case basis.
This study’s data analysis and statistical plan was defined after the data were accessed to determine the sample size with the available data but before any analyses were performed. Continuous variables were assessed for normality using the Kolmogorov–Smirnov test, by assessing skewness and kurtosis, and by inspection of the frequency and quantile–quantile plots. Continuous variables are reported as mean ± SD and/or median [first and third quartiles]. Differences in characteristics were compared with independent (two-sided) t tests or chi-square tests, as appropriate. We evaluated the unadjusted associations between candidate risk factors and lingering postsurgical pain using univariable logistic regressions. A single-step multivariable logistic regression was used to evaluate the independent associations between lingering postsurgical pain and baseline pain, pain expectations, and other preselected factors. These factors were selected by the senior author before the statistical analysis was performed and were included based on suspected clinical relevance or statistical significance in previous studies. Specifically, a set of literature-supported chronic postsurgical pain risk factors were collected and considered to be covariables of interest. Finally, we created a second set of models that included all available covariables based on reviewer feedback. Multicollinearity was determined to be present when variance inflation factor is greater than 5, tolerance is less than 0.2, or the Pearson correlation coefficient is 0.5 or more. In the event of a collinearity, the senior author decided which of the variables to include based on their use in published literature. The covariables included are elderly age (at least 65 yr old), female sex, obesity (body mass index of at least 30 kg/m2), below-average Veterans RAND 12-item mental component scores, American Society of Anesthesiologists (Schaumburg, Illinois) Physical Status scores III or higher, diabetes, current cancer, current tobacco use, preoperative pain related to surgery, severe preoperative pain, negative pain expectations, unknown pain expectations, surgical procedure typically performed to help alleviate pain, estimated severe surgical pain, use of regional anesthesia, and patient’s recollection of severe acute postoperative pain. The Hosmer–Lemeshow test and C-statistic were performed to evaluate the overall goodness of fit (calibration) and to provide a measure of the each model’s ability to discriminate between patients who experienced lingering postsurgical pain and those who did not, respectively. A two-tailed P < 0.05 was considered statistically significant for all statistical tests. No a priori statistical power calculation was performed; the sample size was based on the available data. Missing data in this study were estimated to be missing not at random. Per reviewer request, four sensitivity analyses were performed to evaluate the robustness of our findings to missing data, nonprespecified covariables, and exclusion of the covariable measuring recollection of severe acute postoperative pain: one with missing data imputed by median/zero values, a second using Bayesian bootstrap hot deck imputation, a third that included all available covariables, and a fourth with all prespecified covariables except for recollection of severe acute postoperative pain. All primary analyses relied on complete case analysis because of the simplicity of this technique and a lack of difference in findings when analyzed by either replacement with the median value or hot-deck imputation based on the Bayesian bootstrap method for survey data. All statistical analyses were performed with SAS version 9.4 (SAS Institute Inc., USA).
This analysis included 3,628 patients who underwent surgery between February 2015 and October 2016 (fig. 1). Patients tended to be older-middle-age (60 ± 14 yr), female (58%, n = 2,105), and white (86%, n = 3,109), with substantial comorbidities (45% American Society of Anesthesiologists Physical Status of III or higher, n = 1,647; table 1). Surgeries were most frequently performed by orthopedic (19%), general (15%), and gynecologic (7%) surgical services. Overall, 67% (n = 2,414 of 3,628) expressed positive expectations for their postoperative pain, and 19% (n = 686 of 3628) expressed negative expectations (fig. 2). Of patients with positive expectations, 73% (1,721 of 2,354) had their positive expectations fulfilled. Of patients endorsing positive expectations, 44% (1,042/2,386) underwent surgeries that are typically performed to alleviate pain, whereas only 26% (174 of 679) of patients with negative expectations underwent such procedures (table 2).
Approximately 36% (n = 1,308) of patients reported lingering postsurgical pain. Patients who experienced lingering postsurgical pain rated its severity 3.8 ± 2.1 (mean ± SD) on a 0 to 10 scale. Patients with preoperative pain related to their surgeries reported a much higher incidence of lingering postsurgical pain (55% vs. 23% if no pain, P < 0.001). These patients with preoperative pain related to their surgeries were half as likely to express negative expectations about their postoperative pain compared to patients who had no preoperative pain (9% vs. 19%, P < 0.001, based on data presented in table 2). Patients frequently mispredicted their lingering pain: 26% (623 of 2,354) of patients with positive expectations actually experienced new lingering pain or worsening of their preexisting pain, and 61% (404 of 661) of patients with negative expectations experienced no lingering pain or an improvement in their preexisting pain (table 3). However, patients with negative expectations were also twice as likely to report new or worsening lingering postsurgical pain compared to those who had positive expectations (23% vs. 11%, P < 0.001; table 3). Overall, far more patients experienced new, unabated, or worsened lingering postsurgical pain about a month after their surgeries (n = 1,029) than the number of patients who expected such an undesired outcome preoperatively (n = 661; table 3).
The unadjusted odds ratios for baseline and perioperative factors associated with lingering postsurgical pain are presented in tables 4 and 5. After adjusting for other covariables, the strongest associations with lingering postsurgical pain in this diverse surgical cohort were (1) recollection of severe acute postoperative pain (odds ratio, 3.13; 95% CI, 2.58 to 3.78; P < 0.001); (2) undergoing a procedure typically performed to alleviate pain (odds ratio, 2.18; 95% CI, 1.73 to 2.75; P < 0.001); (3) preoperative pain related to surgery (odds ratio, 1.91; 95% CI, 1.52 to 2.40; P < 0.001); and (4) negative expectations about lingering postsurgical pain (odds ratio, 1.56; 95% CI, 1.23 to 1.98; P < 0.001; fig. 3). Younger patients also reported more lingering postsurgical pain (if age is 65 yr or older; odds ratio, 0.63; 95% CI, 0.52 to 0.76; P < 0.001). The secondary analyses indicate that these associations were generally similar when the definition of lingering postsurgical pain was limited to moderate-severe pain (4 out of 10 or higher) or pain that worsened at least 2 points on a 0 to 10 numeric rating system compared to the preoperative pain severity (fig. 3). There were no major differences in these findings after adjusting for all available covariables or after imputing missing data (Supplemental Digital Content 3, http://links.lww.com/ALN/C575, and Supplemental Digital Content 4, http://links.lww.com/ALN/C576). Removal of the covariable measuring recollection of severe acute postoperative pain did not lead to inflation of the standard errors of any remaining covariable, but having a below-average Mental Component Score of the Veterans RAND 12-item questionnaire became significantly associated with the outcome of lingering pain (odds ratio, 1.21; 95% CI, 1.01 to 1.45; P = 0.043; Supplemental Digital Content 5, http://links.lww.com/ALN/C577).
This observational study of 3,628 patients found that more than a third of patients reported lingering postsurgical pain 30 or more days after a variety of surgical procedures. As expected, preexisting pain was strongly associated with lingering postsurgical pain. However, preoperative expectations for pain outcomes were also influential. Despite being a high-risk group for experiencing lingering postsurgical pain, patients with preexisting pain related to the reason for surgery had higher expectations for postoperative pain improvement or resolution. Negative expectations for postoperative pain were associated with increased lingering postsurgical pain, and this was independent of whether the surgical procedure is typically performed to help alleviate pain.
Several additional candidate risk factors demonstrated statistically significant associations with lingering postsurgical pain in our multivariable model. Some of these, including preoperative pain related to surgery, severe acute postoperative pain, and female sex, have been previously reported.24,27–31 As with many previous studies, older age was associated with less postoperative pain.27,31 Surgical procedure—specifically undergoing a procedure typically performed to help alleviate a painful condition—was significantly associated with lingering postsurgical pain. These findings persisted in secondary analyses evaluating outcomes of moderate-to-severe lingering pain and new/worsened lingering postsurgical pain. Overall, the data from multivariable logistic regression analyses suggest that negative patient expectations about their pain and severe acute postoperative pain may be two of the largest modifiable factors associated with lingering postsurgical pain.
Our results demonstrate that both preoperative pain (particularly if severe and if related to the reason for surgery) and expectations for pain worsening are independently associated with lingering postsurgical pain. Although some risk factors such as preoperative chronic pain have been extensively studied, other factors such as negative pain expectations have not been previously described in detail and warrant further exploration. Patients who expected more painful stimuli in the laboratory setting went on to experience higher pain severity via modulated pain processing activity in the brain.32 Our findings suggest that a similar phenomenon might also exist quite broadly in the clinical perioperative environment. Additionally, patients undergoing breast surgeries who endorsed high preoperative psychologic distress in addition to negative expectations tended to experience more severe acute postsurgical pain.33 It is conceivable that optimistic patients with better psychologic resilience have more positive pain expectations and go on to experience better outcomes. The fact that fewer than 40% of patients with negative expectations went on to experience new or unabated pain suggests that there is a significant gap in which educational initiatives might help patients form pain expectations that are both optimistic and realistic. Several studies have shown that satisfaction with postoperative outcomes is substantially influenced by preoperative expectations.34,35 However, the overwhelming majority of patients undergoing surgery are unaware of the risk of lingering and chronic postsurgical pain.36 Therefore, understanding the relationship between preoperative pain expectations and the patient-reported lingering postsurgical pain can provide valuable information to inform education and interventions to mitigate lingering pain and minimize patients’ dissatisfaction with surgical outcomes.
Among the strengths of this study are its large sample size and diverse surgical population, which increase its precision and likelihood of external validity. Furthermore, this study includes several patient-reported outcomes collected from a validated survey instrument. We specifically asked questions about postoperative pain expectations and analyzed patient outcomes based on dichotomous report of lingering postsurgical pain, clinically meaningful pain, and worsening of pain from baseline. By including questions from the Veterans RAND 12-item health survey, we could reliably adjust for patients’ baseline physical and mental well-being.
This study also has limitations. First, the questions about patients’ expectations cannot distinguish the psychologic, or wishful, component of their expectations versus their innate predictive capability to know what will happen to themselves. It is probable that both interpretations were mixed together in patient responses to this survey question. Although patients with negative expectations ultimately experienced more negative outcomes, many patients who were at highest risk for poor outcomes—those with preexisting pain related to their surgeries—reported unrealistically positive expectations. This complicates the identification of a group that might maximally benefit from increased expert coaching of their expectations. Second, patient-reported factors may have been affected by recall bias and inaccuracy. For example, patients with lingering postsurgical pain might recall their acute postsurgical pain as more severe because they are reviewing past experiences from the vantage of their current situation.37 However, information about patient expectations was collected preoperatively, and the association between acute postsurgical pain and lingering postsurgical pain that we identified has also been supported by other studies.28,38 In addition, methodologic limitations, such as selection bias toward patients who consented to participate in this observational cohort and provided baseline and follow-up data, should be considered. Furthermore, survey length and wording could limit this study’s interpretability. Because of survey constraints, we did not gather preoperative opioid-consumption data, which is known to be associated with chronic postsurgical pain. Additionally, the primary exposure survey question—patient expectations—includes information about patients’ baseline pain. Although postoperative expectations intuitively depend on presurgical pain, the entwined nature of this topic could lead to questions about this study’s interpretability. Finally, this study focuses on “lingering postsurgical pain,” or pain that lingered until the time of the postsurgical assessments 30 or more days postoperatively. There is temporal variability in this outcome because although the questionnaires were sent to patients 30 days after surgery, some participants responded immediately, whereas others required multiple reminders or telephone calls. Lingering postsurgical pain differs from the corpus of literature that primarily focuses on “chronic” or “persistent” postsurgical pain, which is typically assessed more than 3 to 6 months postoperatively.17,28,39–41 International Association for the Study of Pain (Washington, D.C.) classification guidelines suggest that postsurgical pain is considered “chronic” if it lasts at least 3 months after surgery but also note that the transition to persistence frequently begins much earlier.2 This caveat highlights the continued gaps in knowledge regarding the impact of acute and subacute postoperative pain, which may have overarching implications for patient recovery, satisfaction, and long-term outcomes. The period of subacute pain, approximately 1 to 3 months after surgery, may be an important and frequently missed opportunity for interventions to prevent the development of chronic postsurgical pain. Incorporation of longer-term longitudinal data will be critical to our understanding of pain chronification and potential interventions.
In summary, lingering postsurgical pain was common in this large, diverse surgical cohort and was associated with patients’ preoperative expectations about their postoperative pain. Despite being a high-risk group for chronic postsurgical pain, patients with preexisting pain related to the reason for surgery were less likely to expect continued or worsening pain compared to patients without presurgical pain. These findings highlight the complexity in preoperatively assessing the risk of transitioning from acute to chronic postsurgical pain and may help inform future studies of risk prediction and educational interventions for mitigating chronic postsurgical pain.
The authors acknowledge the contributions of Anshuman Sharma, M.D., Troy Wildes, M.D., Alexander Kronzer, B.A., and Dan Helsten, M.D., Department of Anesthesiology, Washington University School of Medicine, St. Louis, St. Louis, Missouri, in the useful discussions to the study design and assistance with data retrieval.
Support was provided solely from institutional and departmental sources, including grant No. BJHF#7937-77 from the Barnes-Jewish Hospital Foundation, St. Louis, Missouri, toward the parent SATISFY-SOS study.
Dr. Haroutounian reports research support from Disarm Therapeutics (Boston, Massachusetts) and consultancy/advisory board fees from Medoc Ltd. (Ramat Yishay, Israel) and Rafa Laboratories (Jerusalem, Israel), outside the scope of this project. The other authors declare no competing interests.