Perioperative Opioid Use in Urogynecologic Mesh Removal.

This was a secondary analysis of a longitudinal, prospective cohort study of patients undergoing urogynecologic mesh removal. Institutional Review Board approval was obtained. Patients were included in this analysis if they had urogynecologic mesh removal at our institution and if either they responded “yes” to the Pelvic Floor Distress Inventory-20 question: “Do you have pelvic pain?”, they had a baseline pelvic pain visual analog scale (VAS) score >10, or pain was the indication for removal. Patients were excluded if they had mesh removal >7 years ago due to PDMP data availability. Demographics, medical and surgical history, medications, and mesh characteristics were documented upon enrollment, and questionnaires including Pelvic Floor Distress Inventory-20 and VAS pain scores were collected at baseline and at 6, 12, and 24 months. Presence or absence of a priori defined chronic pain disorders listed on the patient’s problem list or medical history was abstracted and verified from the EMR (EPIC, Verona, WI). Pain diagnoses included joint pain (including back and neck pain), muscle pain, nerve pain, fibromyalgia, headaches, interstitial cystitis, chronic pain syndrome, endometriosis, and active cancer. To confirm opioid use, the PDMP was used, which lists all controlled substance prescriptions filled within Pennsylvania, Connecticut, Delaware, District of Columbia, Florida, Illinois, Maine, Maryland, Massachusetts, Military Health System, New York, North Carolina, Ohio, South Dakota, Texas, and Virginia. Data were verified by R.G. and A.M.A. Complete vaginal mesh excisions were performed as described in the American Urogynecologic Society/International Urogynecological Association “Joint” Statement on the Management of Mesh-Related Complications for the FPMRS Specialist.” 11 All available opioid prescription data were extracted from the PDMP and classified into preoperative use (within 3 months before surgery), immediate postoperative use (filled within the first 3 months of surgery), and prolonged postoperative use (≥3 months after surgery). 12 , 13 New opioid prescriptions >6 months after surgery were not considered chronic use unless they were a continuation of preoperative use, and when possible, these prescriptions were linked to new diagnoses or surgical procedures to distinguish new indications for opioid use. For patients with preoperative and prolonged use, preoperative and postoperative prescription quantity was quantified using oral morphine milligram equivalents (MMEs) 14 when available to determine whether an increase or decrease in the amount prescribed occurred. Characteristics of patients with and without preoperative narcotic use were compared using t tests and χ 2 tests; logistic regression with use of the likelihood ratio test was used to determine a best fit model using forward selection for variables with P < 0.05 on univariate analysis. Using this convenience sample of 139 patients and an α of 0.05, we had 95% power to detect the observed 35% difference in preoperative narcotic use based on the presence of a chronic pain condition. To further explore the effect of age on preoperative narcotic use, we grouped patients into age <51 years and age ≥51 years based on the average age of menopause, which is a natural marker of aging, and performed multivariable logistic regression. Of those who had preoperative narcotic use, we compared characteristics of patients who did or did not have prolonged postoperative use using t tests and χ 2 tests. We compared median change in VAS scores in those with and without narcotic use and chronic pain conditions using both the earliest available follow-up score and the last value carried forward. Because a sensitivity analysis demonstrated no differences between those who were and were not missing baseline VAS scores, complete case analysis was used for modeling. The PDMP prescriptions were compared with opioids in the EMR medication list, with PDMP considered the gold standard. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated.

One hundred thirty-nine patients were included in the study ( Supplemental Digital Content 1 , figure of participant flow diagram, http://links.lww.com/FPMRS/A501 ): 81 (58.3%) with slings removed, 56 (40.3%) with transvaginal mesh removed, and 2 (1.4%) with both removed. Of the 139 patients, 30 (21.6%) patients were taking narcotics within 3 months preceding mesh removal. More than half of all patients had chronic pain conditions. Patients using narcotics preoperatively were younger and more likely to have a chronic pain condition (87% [26/30] compared with 51% [56/109] of those not taking preoperative narcotics, P < 0.05; Table 1 ). The most common chronic pain conditions were joint or muscle pain and back pain. Joint or muscle pain, chronic pain syndrome, and endometriosis were higher in patients with preoperative narcotic use compared with those without preoperative narcotic use ( Table 2 ). Preoperative narcotic use did not differ by body mass index, mesh type, or presence of an exposure. Those taking narcotics preoperatively had higher median baseline VAS pelvic pain scores by 27 points ( P = 0.01). On logistic regression, the presence of a pain comorbidity increased the odds of preoperative narcotic use 6.5 times, with younger age and higher baseline VAS pelvic pain score remaining weaker predictors ( Table 1 ). The addition of smoking to the model did not improve model fit (likelihood ratio test, P > 0.05). We found that in addition to higher rates of preoperative narcotic use, patients aged <51 years were more likely to have had a sling procedure and had a higher average BMI, worse baseline dyspareunia, and less improvement in pelvic pain scores than older patients at first available follow-up ( P = 0.047) but not last follow-up ( P = 0.22, Table 3 ; Appendix 2 , Supplemental Table 5 , http://links.lww.com/FPMRS/A502 ) with similar improvements in dyspareunia scores ( P = 0.8). Importantly, when controlling for baseline differences in these 2 age groups, we found that preoperative use of narcotics remained strongly and independently predictive of younger age group ( Table 4 ). Seventy-one percent (98/139) of patients filled immediate postoperative prescriptions, and 6% (9/139) filled additional narcotic prescriptions within the first 2 months. Eighteen percent (25/139) had prolonged postoperative use. Over the 7 years of data for each patient, 71% (99/139) had at least 1 unrelated narcotic prescription as determined by timing (>3 months before surgery and >6 months after surgery without continuous use) and an alternative diagnosis at the time of the prescription. For those patients taking narcotics preoperatively, only 27% (8/30) discontinued narcotic use postoperatively. These patients had similar characteristics as those who continued narcotics, including baseline pain scores and rates of comorbid pain conditions ( Table 5 ). Eighty-seven percent of patients who continued taking narcotics postoperatively had a previous pain diagnosis, most commonly joint or muscle pain. Of the 22 patients who continued taking prolonged postoperative narcotics, MME was able available for 9. Five of these 9 (55.6%) remained at their baseline MME, 2 (22.2%) had an increased MME, and 2 (22.2%) had a decreased MME. Due to low numbers at each time point, changes in VAS scores were analyzed both with earliest available follow-up score and last value carried forward. Median follow-up time was 6 months (range 0–24 months). Both analyses demonstrated that patients taking preoperative narcotics had a statistically similar improvement in pelvic pain after removal ( P = 0.3; Table 1 ; Supplemental Table 2 , http://links.lww.com/FPMRS/A502 ). However, patients with chronic pain showed a differential response in pelvic pain over time, which was not seen when grouped by preoperative narcotic use. Patients with and without chronic pain conditions initially had similar rates of improvement in pain scores; however, when looking at a patient’s first and last follow-up time points available, patients with a chronic pain diagnosis listed in their EMRs were much less likely (37%) to have a clinically meaningful (≥13 mm) 15 and persistent (still present at last follow-up time point) change in pain compared with those without a chronic pain condition (61%, Appendix 2 , http://links.lww.com/FPMRS/A502 ). Improvements in dyspareunia were consistent across groups and across time, with approximately 50% of patients having clinically meaningful improvement ( Tables 1 , 3 ; Appendix 2 , Supplemental Tables 3 and 4 , http://links.lww.com/FPMRS/A502 ). Of the 109 patients not taking narcotics preoperatively, only 3 had prolonged postoperative use at ≥3 months without another acute diagnosis with no significant predictors of prolonged use (n = 106, P > 0.2; Appendix 2 , Supplemental Table 1 , http://links.lww.com/FPMRS/A502 ). Of the 139 patients, 12 had filled narcotic prescriptions preoperatively that were not listed in their EMRs. Twenty-one patients had narcotics prescriptions listed in their EMRs that were not current (no prescription filled within 3 months). One hundred patients had concordant EMR and PDMP data. The PPV of a narcotic listed in the EMR medication list was 46%, and the NPV was 88%.

Among patients undergoing mesh removal for pain in this large cohort, we found that (1) 1 in 5 patients will be using narcotics at baseline; (2) these patients are more likely to be younger and to have preexisting joint or muscle pain with higher reported baseline pelvic pain; and (3) mesh removal did not eliminate narcotic use in most of these patients or improve their pain scores. Finally, EMR medication lists were poorly predictive of filled prescriptions. Preoperative narcotic use in our study might be due to mesh pain or chronic comorbidities or both. Many patients in this study had mesh-related pain for years before removal, making the indication for narcotic use difficult to track. In this study, 59% of patients undergoing mesh removal for pain had a comorbid pain condition, compared with 22% of patients nationally. 16 This may indicate that patients with chronic comorbidities are at risk of pain-related mesh complications due to central pain and pain sensitization; however, studies of all patients with mesh implanted are needed to answer this question. We also determined that 71% of patients in this study had an unrelated narcotic prescription over the 7 years of data available for each patient, suggesting either high general narcotic prescription rates or perhaps an underlying predisposition to pain. Younger age was predictive of preoperative narcotic use, even when controlling for other baseline differences between ages. This is different than the general population, in which patients aged ≥65 years were 2.6 times as likely to be prescribed an opioid compared with patients in their early 20s. 17 This could be because younger patients experience mesh pain differently than older patients or health care providers are more likely to believe their pain is an acute process. In many types of surgery for pain, narcotic use and comorbid pain conditions can complicate recovery. 4 – 8 Here we saw statistically similar rates of improvement in VAS pelvic pain and dyspareunia scores in those using and not using preoperative narcotics. However, other studies found that the risk for ongoing postoperative pain after mesh removal is 7 times higher for those taking preoperative narcotics. 18 Our study does suggest that those with comorbid pain conditions were less likely to have a durable improvement in pain. Interestingly, younger patients had less, if any, improvement in generalized pelvic pain after mesh removal with a median improvement of 2 VAS points compared with 20 points of improvement in older patients. This finding is concerning and may be related to differences in type, distribution, and function of pain receptors in young versus older patients. Increasing age is associated with decreased narcotic use after urogynecologic surgery, 19 further supporting that older women may have decreased pelvic pain sensitivity or perception. 20 Younger patients did see more significant relief in dyspareunia, similar to older patients. Although this can be affected by frequency of intercourse, vaginal atrophy, or other factors not measured, it is reassuring that the majority of patients experienced improvement. These data can enhance preoperative counseling and planning. 21 – 23 Studies suggest that patients taking narcotics who undergo surgery to relieve pain are more likely to discontinue narcotics than those who do not undergo surgery. 24 In orthopedic surgery, this number can range from 66% after hip surgery to 22% after lumbar fusion. 25 , 26 In this study, less than a third discontinued them postoperatively, and most remained at their baseline dose, likely reflecting the significance of preexisting pain diagnoses in this population. Opioid-induced hyperalgesia or pain centralization could explain those who did not stop narcotics postoperatively in the absence of other chronic pain conditions. Although not statistically significant, a much higher percentage of those with midurethral sling as the type of mesh removed as compared with prolapse mesh discontinued narcotics. This could be due to an increased mesh burden and heightened inflammatory response in prolapse mesh complications 27 and deserves further research. Finally, 3% of patients not taking preoperative narcotics had prolonged postoperative opioid use, consistent with 3–6% in prior studies. 12 , 25 As we work toward decreasing unneeded opioid prescriptions after surgery, 28 , 29 consideration could be given toward few or no narcotic pills to avoid this risk. We also demonstrated a reasonably high NPV of opioid usage listed in the EMR compared with PDMP but a low PPV. Automatic updating of EMR medication lists by PDMP and automatically adding the associated diagnosis into the PDMP would improve accuracy and health care provider communication regarding opioid prescribing. This study was a large, prospective cohort with standardized mesh removal techniques, robust objective and patient-reported outcomes, PDMP verification of perioperative opioid prescriptions, and 6- to 24-month follow-up. Limitations include small numbers in subgroup analyses, including too low of numbers to group by individual mesh types, which have different complication risk profiles. There were no statistically significant differences in patient characteristics between those with and without missing baseline VAS data, suggesting that at least baseline data are missing at random; however, this can still result in bias. In particular, pain may influence the ability or desire to follow up, which could lead to an overestimation of improvement. Other treatments for pelvic pain during the study period such as multimodal pain prescriptions or pelvic floor physical therapy were not examined in this study. However, pelvic floor physical therapy is routinely prescribed in our practice when myofascial involvement is suspected. 11 Finally, although we do not know the indication for preoperative narcotic use, all patients included in this study reported pelvic pain using validated patient-reported outcomes, or pain was the indication for mesh removal. It should also be noted that this population was primarily White, and this limits generalizability.

Despite improvement in pelvic pain scores in some patients with mesh-related pain, many chronic opioid users continued to take narcotics postoperatively, suggesting the importance of other chronic comorbid pain conditions or perhaps hyperalgesia from opioid use. Younger patients are at higher risk of preoperative narcotic use and experienced less improvement in generalized pelvic pain after removal. Additional counseling is warranted in patients with baseline pain conditions undergoing mesh removal that pain and narcotic use may persist. More data are needed to understand whether chronic comorbid pain conditions or opioid use is a risk factor for developing mesh-related pain.