Machine learning for prediction of sustained opioid prescription after anterior cervical discectomy and fusion.

The severity of the opioid epidemic has increased scrutiny of opioid prescribing practices. Spine surgery is a high-risk episode for sustained postoperative opioid prescription. To develop machine learning algorithms for preoperative prediction of sustained opioid prescription after anterior cervical discectomy and fusion (ACDF). Retrospective, case-control study at two academic medical centers and three community hospitals. Electronic health records were queried for adult patients undergoing ACDF for degenerative disorders between January 1, 2000 and March 1, 2018. Sustained postoperative opioid prescription was defined as uninterrupted filing of prescription opioid extending to at least 90-180 days after surgery. Five machine learning models were developed to predict postoperative opioid prescription and assessed for overall performance. Of 2,737 patients undergoing ACDF, 270 (9.9%) demonstrated sustained opioid prescription. Variables identified for prediction of sustained opioid prescription were male sex, multilevel surgery, myelopathy, tobacco use, insurance status (Medicaid, Medicare), duration of preoperative opioid use, and medications (antidepressants, benzodiazepines, beta-2-agonist, angiotensin-converting enzyme-inhibitors, gabapentin). The stochastic gradient boosting algorithm achieved the best performance with c-statistic=0.81 and good calibration. Global explanations of the model demonstrated that preoperative opioid duration, antidepressant use, tobacco use, and Medicaid insurance were the most important predictors of sustained postoperative opioid prescription. One-tenth of patients undergoing ACDF demonstrated sustained opioid prescription following surgery. Machine learning algorithms could be used to preoperatively stratify risk these patients, possibly enabling early intervention to reduce the potential for long-term opioid use in this population.

Karhade AV ，Ogink PT ，Thio QCBS ，Broekman MLD ，Cha TD ，Hershman SH ，Mao J ，Peul WC ，Schoenfeld AJ ，Bono CM ，Schwab JH ... -  《-》

Chronic opioid use following anterior cervical discectomy and fusion surgery for degenerative cervical pathology.

Although prescribing opioid medication on a limited basis for postoperative pain control is common practice, few studies have focused on chronic opioid use following anterior cervical discectomy and fusion (ACDF). To determine the prevalence of and risk factors for chronic opioid use following one and two-level ACDF for degenerative cervical pathology. Retrospective cohort. Using an insurance claims database, we identified patients aged 18-64 who underwent one or two-level primary ACDF from 2010 to 2015 for degenerative cervical pathology. Opioid prescription strength at various timepoints pre- and postoperatively and development of chronic postoperative opioid use. Prescription opioid use was examined during the following periods: 90 days before 7 days preceding surgery (preoperative), 6 days preceding surgery to 90 days following surgery (perioperative) and from 91 to 365 days following surgery (postoperative). The primary outcome was chronic postoperative opioid use, defined as ≥120 days' supply of opioid prescriptions filled or ≥10 opioid prescriptions between 3 and 12 months postoperatively. Secondary outcomes were high-dose (>90 morphine milligram equivalents [MME]/day) and very high-dose (>200 MME/day) opioid prescriptions. A multivariate logistic model (area under the ROC curve 0.75, p<.001) was built to predict long-term opioid use. Among 28,813 patients who underwent ACDF, most were female (55%) and underwent single-level ACDF (68%), with mean age of 50±8.0 years. Fifty-two percent of patients filled an opioid prescription in the preoperative period, 95% of patients filled a prescription in the perioperative period, and 39% of patients filled a prescription in the postoperative period. High-dose and very high-dose opioid prescriptions in the perioperative period were identified in 45% and 24% of patients, respectively, whereas 17% met criteria for chronic postoperative opioid use. The odds of chronic opioid use were highest in the Western US (odds ratio [OR] 1.5, 95% confidence interval [CI] 1.3, 1.6). Duration of opioids prescribed was also highest in the Western US (median 111 days, interquartile range 11-336), p<.001. Factors associated with the highest risk for chronic opioid use were preoperative opioid use (OR 5.7, 95% CI 5.3, 56.2), drug abuse (OR 3.5, 95% CI 2.6, 4.5), depression (OR 1.7, 95% CI 1.6, 1.9), anxiety (OR 1.5, 95% CI 1.4, 1.6), and surgery in the western region of the United States (OR 1.5, 95% CI 1.3, 1.6). Patients undergoing ACDF commonly receive high-dose opioid prescriptions after surgery, and certain patient factors increase risk for chronic opioid use following ACDF. Intervention focusing on patients with these factors is essential to reduce long-term use of prescription opioids and postoperative care.

Harris AB ，Marrache M ，Jami M ，Raad M ，Puvanesarajah V ，Hassanzadeh H ，Lee SH ，Skolasky R ，Bicket M ，Jain A ... -  《-》

Validating the Stopping Opioids after Surgery (SOS) score for sustained postoperative prescription opioid use in spine surgical patients.

The opioid epidemic has increased scrutiny of health-care practices and care episodes, such as surgery, that increase the risk of opioid dependence. The Stopping Opioids after Surgery (SOS) score to predict sustained prescription opioid use was previously developed within a population of patients receiving general surgery, orthopedic, and urologic procedures. Notably, the performance for this score has not been assessed in a spine surgical cohort. We sought to validate the SOS score in a series of patients undergoing cervical and lumbar spine surgery, including inpatient and outpatient cohorts. Retrospective review at two academic medical centers and three community hospitals. Sustained prescription opioid use was defined as opioid prescription without interruption for 90 days or longer following surgery. The performance of the SOS score was assessed in the study population by calculating the c-statistic, receiver-operating curve, and observed rates of sustained prescription opioid use. Among 7,027 patients included in this study, 2,374 (33.8%) underwent anterior cervical discectomy and fusion and 4,653 (66.2%) underwent surgery for lumbar disc herniation. The median age was 46 (interquartile range=38.0-53.5). Overall, 604 patients (8.6%) had prolonged opioid prescription. The c-statistic of the risk score was 0.764. The sensitivity of the score at the low risk cutoff of 30 was 0.72. At the high-risk cutoff of 60, the specificity was 0.99. The observed risk (95% confidence interval) of prolonged opioid prescription was 3.6% (3.1-4.2) in the low-risk group (scores <30), 17.2% (15.6-18.7) in the intermediate-risk group (scores 30-60), and 46.0% (36.2-55.9) in the high-risk group (scores >60). We have validated the use of a clinically relevant bedside risk score for sustained prescription opioid use after spine surgery. The score's ease of use, combined with its exceptional performance, renders it a valuable tool for spine care providers in counseling patients and determining appropriate postdischarge management to prevent sustained opioid use.

Karhade AV ，Chaudhary MA ，Bono CM ，Kang JD ，Schwab JH ，Schoenfeld AJ ... -  《-》

Postoperative opioid consumption patterns diverge between propensity matched patients undergoing traumatic and elective cervical spine fusion.

Prolonged opioid therapy following spine surgery is an ongoing postoperative concern. While prior studies have investigated postoperative opioid use patterns in the elective cervical surgery patient population, to our knowledge, opioid use patterns in patients undergoing surgery for traumatic cervical spine injuries have not been elucidated. The purpose of this study was to compare opioid use and prescription patterns in the postoperative pain management of patients undergoing traumatic and elective cervical spine fusion surgery. Retrospective cohort study. Adult patients with traumatic cervical injuries who underwent primary anterior cervical discectomy and fusion (ACDF) or posterior cervical decompression and fusion (PCDF) during their initial hospital admission. The propensity matched, control group consisted of adult elective cervical fusion patients who underwent primary ACDF or PCDF. Demographic data, surgical characteristics, spinal disease diagnosis, location of cervical injury, procedure type, operative levels fused, and Prescription Drug Monitoring Program (PDMP) data. PDMP data included the number of opioid prescriptions filled, preoperative opioid use, postoperative opioid use, and use of perioperative benzodiazepines, muscle relaxants, or gabapentin. Opioid consumption data was collected in morphine milligram equivalents (MME) and standardized per day. A 1:1 propensity match was performed to match traumatic injury patients undergoing cervical fusion surgery with elective cervical fusion patients. Traumatic injury patients were matched based on age, sex, CCI, procedure type, and cervical levels fused. Pre- and postoperative opioid, benzodiazepine, muscle relaxant, and gabapentin use were assessed for the traumatic injury and elective patients. T- or Mann-Whitney U tests were used to compare continuous data and Chi-Squared or Fisher's Exact were used to compare categorical data. Multivariate stepwise regression using MME per day 0 - 30 days following surgery as the dependent outcome was performed to further evaluate associations with postoperative opioid use. A total of 48 patients underwent fusion surgery for a traumatic cervical spine injury and 48 elective cervical fusion with complete PDMP data were assessed. Elective patients were found to fill more prescriptions (3.19 vs 0.65, p=.023) and take more morphine milligram equivalents (MME) per day (0.60 vs 0.04, p=.014) within 1 year prior to surgery in comparison to traumatic patients. Elective patients were also more likely to use opioids (29.2% vs 10.4%, p=.040) and take more MMEs per day (0.70 vs 0.05, p=.004) within 30 days prior to surgery. Within 30 days postoperatively, elective patients used opioids more frequently (89.6% vs 52.1%, p<.001) and took more MMEs per day (3.73 vs 1.71, p<.001) than traumatic injury patients. Multivariate stepwise regression demonstrated preoperative opioid use (Estimate: 1.87, p=.013) to be correlated with higher postoperative MME per day within 30 days of surgery. Surgery after traumatic injury was correlated with lower postoperative MME use per day within 30 days of surgery (Estimate: -1.63 p=.022). Cervical fusion patients with a history of traumatic spine injury consume fewer opioids in the early postoperative period in comparison to elective cervical fusion patients, however both cohorts consumed a similar amount after the initial 30-day postoperative period. Preoperative opioid use was also a risk factor for higher consumption in the short-term postoperative period. These results may aid physicians in further understanding patients' postoperative care needs based on presenting injury characteristics and highlights the need for enhanced follow-up care for traumatic cervical spine injury patients after fusion surgery.

Pohl NB ，Narayanan R ，Lee Y ，McCurdy MA ，Carter MV ，Hoffman E ，Fras SI ，Vo M ，Kaye ID ，Mangan JJ ，Kurd MF ，Canseco JA ，Hilibrand AS ，Vaccaro AR ，Schroeder GD ，Kepler CK ... -  《-》

Development of machine learning algorithms for prediction of prolonged opioid prescription after surgery for lumbar disc herniation.

Spine surgery has been identified as a risk factor for prolonged postoperative opioid use. Preoperative prediction of opioid use could improve risk stratification, shared decision-making, and patient counseling before surgery. The primary purpose of this study was to develop algorithms for prediction of prolonged opioid prescription after surgery for lumbar disc herniation. Retrospective, case-control study at five medical centers. Chart review was conducted for patients undergoing surgery for lumbar disc herniation between January 1, 2000 and March 1, 2018. The primary outcome of interest was sustained opioid prescription after surgery to at least 90 to 180 days postoperatively. Five models (elastic-net penalized logistic regression, random forest, stochastic gradient boosting, neural network, and support vector machine) were developed to predict prolonged opioid prescription. Explanations of predictions were provided globally (averaged across all patients) and locally (for individual patients). Overall, 5,413 patients were identified, with sustained postoperative opioid prescription of 416 (7.7%) at 90 to 180 days after surgery. The elastic-net penalized logistic regression model had the best discrimination (c-statistic 0.81) and good calibration and overall performance; the three most important predictors were: instrumentation, duration of preoperative opioid prescription, and comorbidity of depression. The final models were incorporated into an open access web application able to provide predictions as well as patient-specific explanations of the results generated by the algorithms. The application can be found here: https://sorg-apps.shinyapps.io/lumbardiscopioid/ CONCLUSION: Preoperative prediction of prolonged postoperative opioid prescription can help identify candidates for increased surveillance after surgery. Patient-centered explanations of predictions can enhance both shared decision-making and quality of care.

Karhade AV ，Ogink PT ，Thio QCBS ，Cha TD ，Gormley WB ，Hershman SH ，Smith TR ，Mao J ，Schoenfeld AJ ，Bono CM ，Schwab JH ... -  《-》