Abstract
Study Design:
Retrospective administrative database review
Objective:
Analyze patterns of opioid use in patients undergoing lumbar surgery and determine associated risk factors in a Medicaid population.
Summary of Background Data:
Opioid use in patients undergoing surgery for degenerative lumbar spine conditions is prevalent and impacts outcomes. There is limited information defining the scope of this problem in Medicaid patients.
Methods:
Longitudinal cohort study of adult South Carolina (SC) Medicaid patients undergoing lumbar surgery from 2014–2017. All patients had continuous SC Medicaid coverage for 15 consecutive months, including 6 months prior to and 9 months following surgery. The primary outcome was a longitudinal assessment of post-operative opioid use to determine trajectories and group-based membership using latent modeling. Univariate and multivariable modeling was conducted to assess risk factors for group-based trajectory modeling (GBTM) and chronic opioid use (COU).
Results:
A total of 1,455 surgeries met inclusion criteria. GBTM demonstrated patients fit into 5 groups; very low use (23.4%), rapid wean following surgery (18.8%), increasing use following surgery (12.9%), slow wean following surgery (12.6%) and sustained high use (32.2%). Variables predicting membership in high opioid use included pre-operative opioid use, younger age, longer length of stay, concomitant medications, and readmissions. More than three-quarters of patients were deemed COUs (76.4%). On bivariate analysis, patients with degenerative disc disease (DDD) were more likely to be COUs (24.8% vs. 18.6%; p=0.0168), more likely to take opioids prior to surgery (88.5% vs. 61.9%; p<0.001) and received higher amounts of opioids during the 30-days following surgery (mean MME 59.6 vs 25.1; p<0.001).
Conclusions:
Most SC Medicaid patients undergoing lumbar elective lumbar spine surgery were using opioids pre-operatively and continued long-term use post-operatively at a higher rate than previously reported databases. Pre-operative and perioperative intake, DDD, multiple prescribers, depression and concomitant medications were significant risk factors.
Keywords: Lumbar spine, opioid use, spine surgery, risk factors, Medicaid, Administrative Database
Brief Abstract
A Medicaid database was analyzed to determine opioid use patterns in patients undergoing lumbar spine surgery. Approximately 75% of all patients were using opioids prior to as well as over 3 months after surgery. Pre-operative use, 30-day post-operative use, surgery for degenerative disc disease, and depression were significant risk factors.
Introduction
Opioids are a cornerstone therapy used to manage back pain. Opioid expenditure for spine problems increased nearly 7-fold from 1997 to 2006.1. Patients who ultimately undergo surgery for back pain often have a protracted course of nonoperative treatment and are therefore vulnerable to chronic opioid consumption prior to surgery. Medicaid services were estimated to increase to 25.3 million individuals in the US in 2019 2, yet there is a paucity of data describing opioid use in this patient population
The cost and morbidity of chronic opioid use (COU) are substantial and are associated with poor patient-reported outcomes following surgery 3,4,5. These data were derived from the non-Medicaid population. Patients undergoing spinal surgery represent a high-risk population for perioperative opioid use 6,7. Improved understanding of risk factors for opioid use following back surgery is a critical step towards reducing opioid utilization. The purpose of this study was to analyze the patterns of opioid use in Medicaid patients undergoing lower back surgery and determine the associated primary risk factors in the Medicaid population of South Carolina (SC).
Methods
Study Design and Population.
This was a retrospective cohort study of SC Medicaid patients. Using methods previously reported 8, we formed an inception cohort of SC Medicaid patients who had undergone one or more elective lumbar spine surgeries. We used de-identified medical and pharmacy outpatient claims data for patients older than 18 years on the admission date and were enrolled in Medicaid between January 2014 and December 2017 A lumbar spine surgery phenotype was developed based on ICD-9 and −10 procedure codes and current procedural terminology (CPT) codes following a thorough literature review (Supplementary Table 1) 9,10,11 Clinical focus was on degenerative disorders. Surgeries were excluded if codes indicated trauma, spinal cord injury, hardware removal, spina bifida, infection, cancer, or an autoimmune condition. Data for each surgery was analyzed from 6 months before the surgery date to 9 months after the surgery. Patients without continuous Medicaid coverage or pharmacy claims data during the 15-month study period were excluded. Outpatient pharmacy data were used to determine patient receipt of opioids, but actual consumption data were not available. We determined the morphine equivalent daily dose (MEDD) 12. Mean morphine milligram equivalents (MME) per day were determined by averaging the total MME dispensed over 30-day intervals. Due to irregularities in opioid data for medications provided in patches or liquid form, we excluded prescriptions for these forms. The study was approved by The Medical University of South Carolina Institutional Review Board.
Timeline, Outcome and Exposure.
Figure 1 provides an overview of the study timeline, where the day of discharge was defined as day 0. Comorbidities were assessed for 6 months prior to surgery, but since 90 days is often considered to be opioid naïve, the pre-surgical period for opioid use was 3 months. The exposure period was defined as the 30 days after the date of discharge and is the time after which we anticipated discontinuation of post-surgery opioids. The follow-up period was 31–270 days where continued opioid use was assessed. The outcome was chronic use of opioids, defined as any opioid prescription filled 90 or more days after discharge through 270 days after discharge. The exposure was opioid use characterized in two ways:1) daily opioid use categorized into five MME levels (no opioids, >0 to < 20, ≥20 to < 50, ≥50 to <90, and ≥90 MME/day); and 2) opioid days categorized by four levels (0, 1–5, 6–14, and > 14 days with opioids). Group based trajectory modeling was used to determine patterns of chronic opioid use
Figure 1:
Study timeline
Covariates.
Demographic variables included age at the date of surgery, sex, and race/ethnicity. Chronic pre-operative comorbid conditions were identified using ICD-9-CM or ICD-10-CM codes. 13 Baseline opioid use prior to surgery was characterized by two variables:1) opioid-naïve status over the 90 days prior to surgery and 2) daily opioid use categorized into:0–49, 50–89, and ≥90 MME/day groups. The exposure period of opioid use was characterized by 1) opioid type (long- and short-acting); 2) single opioid versus combination medications that contained an opioid; 3) number of opioid prescriptions; 4) prescriber specialty (surgeon, primary care, other, none); 5) concomitant medications that might influence COU (antidepressants, antipsychotics, gabapentin, pregabalin, benzodiazepines, non-benzodiazepine sedative/hypnotics, anxiolytics, muscle relaxants, and duloxetine); and 6) alternative pain treatments (NSAID use, physical therapy, and occupational therapy). Surgery and process of care variables included: 1) hospital length of stay (0, 1–3, 4–7 and > 7 days); 2) re-hospitalization within 30 days or 31–270 days of discharge; 3) discharge to skilled nursing or rehabilitation facility; 4) days following discharge until the first post-operative visit; 5) total outpatient visits, total unique opioid prescribers, and total unique non-prescribers visited (e.g., physical therapists).
Statistical Analyses.
Using the methods described previously (1) and briefly summarized here, summary statistics and related tests were conducted using t-tests, ANOVA, chi-square, or Fisher’s exact tests, as appropriate. We first used group-based trajectory models (GBTM) to identify clusters of lumbar spine surgeries where patients had similar patterns of monthly MME averages post-operatively. These models are based on finite mixture model techniques, where we used a zero-inflated Poisson distribution for the MME values 14,15,16. The estimation was performed using SAS PROC TRAJ with up to fourth-order polynomial functions. The best fit based on the number of groups was determined by comparing the Bayesian information criterion (BIC) statistics. Within each group, the parameter estimates for the highest order polynomial (p <0.05) were used to determine the best-fitting curve. Group assignments were then used as the dependent variable in a multinomial logistic model to predict the variables that were most influential in predicting group assignments. We next estimated the risk ratios (RR) and 95% CI for all measured risk factors using generalized linear mixed models (GLMM) with a Poisson distribution and log link 17. We estimated the relative risk instead of odds ratios because COU was common in this population, occurring in 76% of surgeries 18. We used separate GLMM models to assess each outcome. All analyses were performed using SAS 9.4 (SAS Institute, Inc., Cary, NC, USA).
RESULTS
After exclusion, there were 1,455 surgeries in 1,348 patients in the final cohort for analysis (Figure 2).
Figure 2:
Patient screening flow chart
Group Based Trajectory Modeling (GBTM) for Opioid Utilization Following Lumbar Surgery
Figure 3 displays five distinct opioid-use phenotypes based on the GBTM analysis. These included low use throughout (23.4% of the cohort; group 1), rapid wean following surgery (18.8%; group 2), increasing use following surgery (12.9%; group 3), slow wean following surgery (12.6%; group 4) and sustained high use (32.2%; group 5). Table 1 shows univariate comparisons of characteristics based on group membership demonstrating several characteristics associated with opioid use. Group 1 patients were more likely to be slightly younger (18–34) or older (>65) compared to the other groups, much more likely to be naïve, had fewer opioid days during pre-surgical exposure, follow-up periods less likely to have some other multi-modal medications, less likely to have degenerative disc disease, more likely to have herniation, less likely to be rehospitalized 30–270 days after discharge, and have fewer distinct prescribers than other groups. Those using opioids prior to surgery were more likely to be in groups 2 to 5. Opioid use in the 30-days following surgery also strongly predicted group membership as did diagnostic category. DDD patients were most likely to be sustained high users (group 5). Concomitant medications, time to post-operative visit, hospital length of stay, rehospitalization and provider characteristics (type and number of unique prescribers) were also significantly associated with group membership. In the group-based trajectory multinomial models predicting group membership (Table 2), significant variables consistently included pre-operative opioid naivety and 30-day post-operative opioid use (>14-day exposure or ≥50 MME/day).
Figure 3:
Opioid MME trajectory groups
Table 1:
Demographic and clinical characteristics for lumbar trajectory groups. SC Medicaid, 2014–2017.
|
| ||||||||
|---|---|---|---|---|---|---|---|---|
| Measure: all values N(%) unless otherwise shown | Level | TRAJECTORY GROUP | Total | p-value | ||||
|
| ||||||||
| 1 | 2 | 3 | 4 | 5 | ||||
|
| ||||||||
| Number of surgeries per group | 341 (23.4) | 274 (18.8) | 189 (12.9) | 182 (12.6) | 469 (32.2) | 1455 (100) | <.0001 | |
|
|
||||||||
| Surgeries with chronic outcomea | 91 (26.7) | 180 (65.7) | 189 (100) | 182 (100) | 469 (100) | 1111 (76.4) | <.0001 | |
|
|
||||||||
| Age | mean (std) | 46.8 (15.3) | 48.9 (12.7) | 46.6 (12.2) | 47.2 (11) | 49.1 (11.2) | 47.9 (12.7) | 0.0259 |
|
| ||||||||
| Age category N(%) | 18–34 years | 82 (24) | 40 (14.6) | 35 (18.5) | 29 (15.9) | 57 (12.2) | 243 (16.7) | <.0001 |
|
| ||||||||
| 35–54 years | 146 (42.8) | 124 (45.3) | 99 (52.4) | 103 (56.6) | 244 (52) | 716 (49.2) | <.0001 | |
|
| ||||||||
| 55–64 years | 68 (19.9) | 85 (31) | 45 (23.8) | 41 (22.5) | 135 (28.8) | 374 (25.7) | <.0001 | |
|
| ||||||||
| ≥ 65 years | 45 (13.2) | 25 (9.1) | 10 (5.3) | 9 (4.9) | 33 (7) | 122 (8.4) | <.0001 | |
|
| ||||||||
| Sex | % male | 112 (32.8) | 83 (30.3) | 47 (24.9) | 66 (36.3) | 178 (38) | 486 (33.4) | 0.0142 |
|
| ||||||||
| Race_ethnicity | White | 189 (55.4) | 138 (50.4) | 103 (54.5) | 103 (56.6) | 283 (60.3) | 816 (56.1) | 0.11 |
|
| ||||||||
| Black | 97 (28.4) | 89 (32.5) | 58 (30.7) | 39 (21.4) | 108 (23) | 391 (26.9) | ||
|
| ||||||||
| Hispanic | 4 (1.2) | 1 (0.4) | 1 (0.5) | 2 (1.1) | 2 (0.4) | 10 (0.7) | ||
|
| ||||||||
| Other/Unknown | 51 (15) | 46 (16.8) | 27 (14.3) | 38 (20.9) | 76 (16.2) | 238 (16.4) | ||
|
| ||||||||
| Baseline measures (90 days pre-surgery period) | ||||||||
|
| ||||||||
| Opioid naïveb | 0 days | 151 (44.3) | 50 (18.2) | 30 (15.9) | 12 (6.6) | 16 (3.4) | 259 (17.8) | <.0001 |
|
| ||||||||
| Opioid use during 30 days before surgery | >0–49 MME/day | 185 (54.3) | 214 (78.1) | 154 (81.5) | 145 (79.7) | 266 (56.7) | 964 (66.3) | |
|
|
||||||||
| 50–89 MME/day | 4 (1.2) | 8 (2.9) | 4 (2.1) | 20 (11) | 136 (29) | 172 (11.8) | ||
|
|
||||||||
| ≥90MME/day | 1 (0.3) | 2 (0.7) | 1 (0.5) | 5 (2.7) | 51 (10.9) | 60 (4.1) | ||
|
| ||||||||
| Opioid use among all patients | ||||||||
|
| ||||||||
| Mean opioid days [mean(std)] | 90 days pre surgeryc | 15.74 (22.8) | 34.73 (29.96) | 43.82 (31.56) | 51.87 (29.48) | 67.76 (26.54) | 44.25 (33.87) | <.0001 |
|
|
||||||||
| exposure periodd | 10.01 (7.93) | 18.98 (8.6) | 17.04 (9.21) | 23.04 (7.79) | 24.69 (7.22) | 18.98 (9.79) | <.0001 | |
|
|
||||||||
| ≥ 90 days post surgery | 4.89 (14.02) | 18.21 (24.05) | 96.9 (45.48) | 80.04 (41.4) | 155.15 (27.17) | 77.18 (69.04) | <.0001 | |
|
| ||||||||
| Mean MME [mean(std)] | 90 days pre surgery | 6.54 (16.78) | 13.95 (21.88) | 16.31 (16.24) | 27.15 (24.12) | 49.84 (41.07) | 25.74 (33.42) | <.0001 |
|
|
||||||||
| exposure period | 19.09 (16.74) | 43.27 (32.73) | 33.81 (23.53) | 61.39 (36.85) | 83.08 (53.41) | 51.47 (45.38) | <.0001 | |
|
|
||||||||
| ≥ 90 days post surgery | 0.48 (1.09) | 2.19 (2.47) | 18.82 (10.93) | 17.44 (9.22) | 64.44 (42.32) | 25.92 (36.82) | <.0001 | |
|
| ||||||||
| Percent with opioid use | 90 days pre surgery | 55.7 | 81.8 | 84.1 | 93.4 | 96.6 | 82.2 | <.0001 |
|
|
||||||||
| exposure period | 79.8 | 96.0 | 91.5 | 98.4 | 98.3 | 92.7 | <.0001 | |
|
|
||||||||
| ≥90 days post surgery | 26.4 | 65.7 | 100.0 | 100.0 | 100.0 | 76.3 | <.0001 | |
|
|
||||||||
| Exposure period among patients dispensed opioids | ||||||||
|
|
||||||||
| mean opioid days [mean(std)] | mean(std) | 12.55 (6.85) | 19.77 (7.83) | 18.62 (7.95) | 23.43 (7.25) | 25.12 (6.5) | 20.48 (8.52) | <.0001 |
|
| ||||||||
| Opiod type | long act | 1 (0.4) | 0 (0) | 0 (0) | 0 (0) | 3 (0.7) | 4 (0.3) | <.0001 |
|
|
||||||||
| short act | 267 (98.2) | 246 (93.5) | 164 (94.8) | 158 (88.3) | 366 (79.4) | 1201 (89.1) | <.0001 | |
|
|
||||||||
| both | 4 (1.5) | 17 (6.5) | 9 (5.2) | 21 (11.7) | 92 (20) | 143 (10.6) | <.0001 | |
|
| ||||||||
| Single vs combo opioid | single | 54 (19.9) | 32 (12.2) | 26 (15) | 27 (15.1) | 69 (15) | 208 (15.4) | <.0001 |
|
|
||||||||
| combo | 188 (69.1) | 149 (56.7) | 104 (60.1) | 102 (57) | 212 (46) | 755 (56) | ||
|
|
||||||||
| both | 30 (11) | 82 (31.2) | 43 (24.9) | 50 (27.9) | 180 (39) | 385 (28.6) | ||
|
| ||||||||
| MME/day | mean(std) over 30 days | 23.94 (15.33) | 45.08 (32.16) | 36.94 (22.11) | 62.42 (36.28) | 84.53 (52.73) | 55.56 (44.67) | <.0001 |
|
|
||||||||
| Opioid prescription count | mean (std) | 1.5 (0.68) | 2.47 (1.01) | 2.31 (1.09) | 2.96 (1.43) | 3.2 (1.42) | 2.57 (1.34) | <.0001 |
|
| ||||||||
| Provider practice specialty | surgeon | 207 (76.1) | 225 (85.6) | 136 (78.6) | 139 (77.7) | 325 (70.5) | 1032 (76.6) | <.0001 |
|
|
||||||||
| other | 38 (14) | 20 (7.6) | 19 (11) | 14 (7.8) | 54 (11.7) | 145 (10.8) | ||
|
|
||||||||
| primary care provider | 6 (2.2) | 15 (5.7) | 11 (6.4) | 21 (11.7) | 58 (12.6) | 111 (8.2) | ||
|
|
||||||||
| Other medications during exposure period (all patients) | ||||||||
|
|
||||||||
| NSAIDS, APAP | 28 (8.2) | 42 (15.3) | 25 (13.2) | 28 (15.4) | 56 (11.9) | 179 (12.3) | 0.0513 | |
|
|
||||||||
| Antidepressants | 46 (13.5) | 77 (28.1) | 38 (20.1) | 57 (31.3) | 118 (25.2) | 336 (23.1) | <.0001 | |
|
|
||||||||
| Antipsychotics | 15 (4.4) | 20 (7.3) | 10 (5.3) | 9 (4.9) | 28 (6) | 82 (5.6) | 0.6073 | |
|
|
||||||||
| Gabapentin | 70 (20.5) | 76 (27.7) | 47 (24.9) | 54 (29.7) | 155 (33) | 402 (27.6) | 0.0023 | |
|
|
||||||||
| Pregabalin | 10 (2.9) | 9 (3.3) | 12 (6.3) | 10 (5.5) | 31 (6.6) | 72 (4.9) | 0.084 | |
|
|
||||||||
| Benzodiazepines | 53 (15.5) | 66 (24.1) | 42 (22.2) | 51 (28) | 187 (39.9) | 399 (27.4) | <.0001 | |
|
|
||||||||
| Selected sedatives/hypnoticse | 15 (4.4) | 19 (6.9) | 15 (7.9) | 18 (9.9) | 38 (8.1) | 105 (7.2) | 0.1517 | |
|
|
||||||||
| Muscle relaxants | 97 (28.4) | 121 (44.2) | 71 (37.6) | 84 (46.2) | 241 (51.4) | 614 (42.2) | <.0001 | |
|
|
||||||||
| Duloxetine | 14 (4.1) | 15 (5.5) | 14 (7.4) | 5 (2.7) | 35 (7.5) | 83 (5.7) | 0.0805 | |
|
| ||||||||
| Alternative pain treatments | Physical therapy | 65 (19.1) | 46 (16.8) | 39 (20.6) | 36 (19.8) | 76 (16.2) | 262 (18) | 0.5856 |
|
|
||||||||
| Occupational therapy | 28 (8.2) | 24 (8.8) | 21 (11.1) | 19 (10.4) | 28 (6) | 120 (8.2) | 0.1639 | |
|
| ||||||||
| Surgery and process of care variables | ||||||||
|
| ||||||||
| Surgery typeg | degenerative_disc | 50 (14.7) | 71 (20.9) | 42 (12.4) | 50 (14.7) | 127 (37.4) | 340 (23.4) | 0.0003 |
|
|
||||||||
| herniation | 119 (28.5) | 76 (18.2) | 61 (14.6) | 46 (11) | 116 (27.8) | 418 (28.7) | 0.015 | |
|
|
||||||||
| stenosis | 166 (25.6) | 122 (18.8) | 78 (12) | 72 (11.1) | 210 (32.4) | 648 (44.5) | 0.285 | |
|
|
||||||||
| spondylolisthesis | 30 (16.4) | 40 (21.9) | 27 (14.8) | 26 (14.2) | 60 (32.8) | 183 (12.6) | 0.1678 | |
|
|
||||||||
| pseudoarthrosis | 1 (10) | 3 (30) | 1 (10) | 4 (40) | 1 (10) | 10 (0.7) | 0.0577 | |
|
|
||||||||
| post_Laminectomy | 5 (17.9) | 6 (21.4) | 2 (7.1) | 3 (10.7) | 12 (42.9) | 28 (1.9) | 0.6797 | |
|
|
||||||||
| other | 23 (29.9) | 15 (19.5) | 4 (5.2) | 14 (18.2) | 21 (27.3) | 77 (5.3) | 0.0927 | |
|
| ||||||||
| Discharge to skilled nursing facility N(%) | 9 (2.6) | 6 (2.2) | 4 (2.1) | 5 (2.7) | 6 (1.3) | 30 (2.1) | 0.6565 | |
|
| ||||||||
| Rehospitalization N(%) | within 30 days of discharge | 10 (2.93) | 3 (1.09) | 7 (3.7) | 11 (6.04) | 18 (3.84) | 49 (3.37) | 0.0627 |
|
|
||||||||
| 30–270 days after discharge | 26 (7.62) | 27 (9.85) | 51 (26.98) | 25 (13.74) | 84 (17.91) | 213 (14.64) | <.0001 | |
|
| ||||||||
| Days to post-discharge surgeon visit | 14 days or less | 63 (18.5) | 48 (17.5) | 34 (18) | 41 (22.5) | 88 (18.8) | 274 (18.8) | 0.03 |
|
|
||||||||
| More than 14 days | 33 (9.7) | 45 (16.4) | 25 (13.2) | 28 (15.4) | 41 (8.7) | 172 (11.8) | ||
|
|
||||||||
| NO visits | 245 (71.8) | 181 (66.1) | 130 (68.8) | 113 (62.1) | 340 (72.5) | 1009 (69.3) | ||
|
| ||||||||
| PCM provider days to first visit | 14 days or less | 50 (14.7) | 39 (14.2) | 37 (19.6) | 34 (18.7) | 96 (20.5) | 256 (17.6) | 0.04 |
|
|
||||||||
| More than 14 days | 39 (11.4) | 51 (18.6) | 32 (16.9) | 28 (15.4) | 78 (16.6) | 228 (15.7) | ||
|
|
||||||||
| NO visits | 252 (73.9) | 184 (67.2) | 120 (63.5) | 120 (65.9) | 295 (62.9) | 971 (66.7) | ||
|
| ||||||||
| Total visits during exposure period [mean(std)] | Primary care provider | 0.8 (2.51) | 0.76 (1.91) | 0.7 (1.48) | 0.87 (2.28) | 0.8 (1.68) | 0.79 (2) | 0.9166 |
|
|
||||||||
| Surgeon | 0.41 (0.79) | 0.53 (0.89) | 0.4 (0.67) | 0.54 (0.82) | 0.42 (0.82) | 0.45 (0.81) | 0.1393 | |
|
|
||||||||
| Other prescribers | 0.57 (2.61) | 0.33 (0.78) | 0.83 (2.46) | 0.62 (1.44) | 0.49 (1.38) | 0.54 (1.84) | 0.0067 | |
|
|
||||||||
| Non prescribers (other HC prof)f | 0.05 (0.41) | 0.02 (0.21) | 0.04 (0.34) | 0.11 (0.58) | 0.03 (0.3) | 0.04 (0.37) | 0.3437 | |
|
|
||||||||
| total visits | 1.83 (4.29) | 1.64 (2.47) | 1.97 (3.34) | 2.14 (3.47) | 1.73 (2.78) | 1.82 (3.3) | 0.4629 | |
|
| ||||||||
| Hospital length of stay | single day | 117 (34.3) | 67 (24.5) | 63 (33.3) | 65 (35.7) | 107 (22.8) | 419 (28.8) | <.0001 |
|
| ||||||||
| 1–3 days | 135 (39.6) | 125 (45.6) | 70 (37) | 67 (36.8) | 239 (51) | 636 (43.7) | ||
|
| ||||||||
| 4–7 days | 57 (16.7) | 66 (24.1) | 35 (18.5) | 44 (24.2) | 95 (20.3) | 297 (20.4) | ||
|
| ||||||||
| > 7 days | 32 (9.4) | 16 (5.8) | 21 (11.1) | 6 (3.3) | 28 (6) | 103 (7.1) | ||
|
| ||||||||
| Number of unique providers visited (claims data) | mean(std) | 1.14 (1.47) | 1.22 (1.3) | 1.42 (1.71) | 1.58 (1.9) | 1.32 (1.63) | 1.31 (1.59) | 0.0478 |
|
|
||||||||
| Count of distinct prescribers (pharmacy data) | mean (std) | 0.97 (0.63) | 1.59 (0.74) | 1.46 (0.75) | 1.74 (0.88) | 1.8 (0.83) | 1.51 (0.83) | <.0001 |
|
|
||||||||
| Number of unique non-prescribers visited | mean (std) | 0.02 (0.17) | 0.01 (0.12) | 0.03 (0.19) | 0.06 (0.32) | 0.01 (0.15) | 0.02 (0.19) | 0.3686 |
Use of opioids ≥ 90 days after discharge from procedure is the measure of “chronic use”
Opioid-naïve for analyses defined as 0 opioid days during 90 days prior to surgery
Presurgical period = 90 days prior to procedure
Exposure period = 30 days after discharge from procedure
This category included non-benzodiazepine sedative/hypnotics and selected anxiolytics
Includes other health professions who can submit Medicaid claims, such as therapists
Some surgeries involved more than 1 type
Table 2:
Group Trajectory Model results. Model adjusted for age, sex, LOS, PT/OT/SNF. Values displayed as odds ratios with 95% CI. Each odds ratio is referenced to group 1
| Odds Ratio (95% CI) |
|||||||||
|---|---|---|---|---|---|---|---|---|---|
| Opioid days exposure | MME/day exposure | ||||||||
|
| |||||||||
| Variable | Level | 2: slow wean to 0 | 3: gradual increase | 4: high use decreasing over time | 5: stable high use | 2: slow wean to 0 | 3: gradual increase | 4: high use decreasing over time | 5: stable high use |
|
| |||||||||
| Opioid naïve | No vs. Yes | 1.55 (1.25, 1.92) | 1.78 (1.39, 2.28) | 2.67 (1.89, 3.76) | 3.67 (2.71, 4.98) | 1.51 (1.22, 1.87) | 1.79 (1.39, 2.29) | 2.47 (1.74, 3.49) | 3.27 (2.38, 4.49) |
|
|
|
||||||||
| Opioid days in 30 days following discharge | 1–5 days vs. 0 days | 0.46 (0.25, 0.83) | 0.43 (0.23, 0.80) | 0.43 (0.18, 1.03) | 0.25 (0.12, 0.55) | ||||
| 6–14 days vs. 0 days | 0.83 (0.57, 1.21) | 0.76 (0.51, 1.11) | 0.61 (0.34, 1.08) | 0.53 (0.34, 0.84) | |||||
| > 14 days vs. 0 days | 2.84 (1.92, 4.21) | 2.04 (1.35, 3.09) | 5.49 (3.21, 9.38) | 6.26 (4.06, 9.66) | |||||
|
| |||||||||
| Mean daily MME in 30 days following discharge | 0<MME<20 vs. 0 MME | 0.26 (0.15, 0.44) | 0.41 (0.23, 0.73) | 0.12 (0.06, 0.25) | 0.08 (0.04, 0.15) | ||||
| 20≤ MME<50 vs. 0 MME | 0.57 (0.35, 0.95) | 0.54 (0.31, 0.94) | 0.41 (0.23, 0.72) | 0.29 (0.17, 0.49) | |||||
| 50≤ MME<90 vs. 0 MME | 1.82 (0.95, 3.51) | 1.83 (0.91, 3.68) | 3.92 (1.95, 7.87) | 3.50 (1.84, 6.66) | |||||
| MME ≥ 90 | 8.35 (1.63, 42.8) | 2.33 (0.39, 13.9) | 24.7 (4.78, 128) | 43.6 (8.74, 218) | |||||
|
| |||||||||
| Patient age | 35–54 vs < 35 | 1.14 (0.85, 1.54) | 1.56 (1.12, 2.18) | 1.55 (1.08, 2.21) | 1.39 (1.03, 1.86) | 1.19 (0.88, 1.59) | 1.62 (1.17, 2.26) | 1.66 (1.16, 2.39) | 1.46 (1.07, 1.99) |
| 55–64 vs < 35 | 1.47 (1.05, 2.07) | 1.25 (0.84, 1.86) | 1.07 (0.70, 1.64) | 1.36 (0.96, 1.92) | 1.56 (1.11, 2.19) | 1.27 (0.86, 1.89) | 1.19 (0.77, 1.83) | 1.50 (1.05, 2.16) | |
| ≥ 65 vs < 35 | 0.92 (0.57, 1.50) | 0.52 (0.28, 0.96) | 0.71 (0.36, 1.38) | 0.96 (0.58, 1.60) | 0.75 (0.45, 1.23) | 0.44 (0.24, 0.82) | 0.52 (0.26, 1.04) | 0.72 (0.42, 1.23) | |
|
| |||||||||
| Lumbar surgery type | degenerative disc | 1.36 (1.02, 1.82) | 1.17 (0.85, 1.61) | 1.37 (0.98, 1.91) | 1.21 (0.90, 1.62) | 1.33 (0.99, 1.77) | 1.18 (0.86, 1.62) | 1.32 (0.94, 1.85) | 1.15 (0.85, 1.56) |
| herniation | 0.95 (0.72, 1.24) | 0.82 (0.61, 1.10) | 0.78 (0.57, 1.08) | 0.72 (0.55, 0.95) | 0.98 (0.75, 1.28) | 0.85 (0.63, 1.14) | 0.88 (0.64, 1.22) | 0.87 (0.65, 1.15) | |
| stenosis | 0.97 (0.75, 1.24) | 0.84 (0.64, 1.10) | 0.87 (0.65, 1.17) | 0.84 (0.65, 1.08) | 0.99 (0.77, 1.26) | 0.86 (0.66, 1.13) | 0.92 (0.68, 1.23) | 0.91 (0.70, 1.17) | |
| spondylolisthesis | 1.22 (0.87, 1.73) | 1.16 (0.80, 1.70) | 1.27 (0.85, 1.88) | 1.07 (0.76, 1.52) | 1.18 (0.84, 1.66) | 1.15 (0.79, 1.67) | 1.28 (0.86, 1.91) | 1.15 (0.80, 1.64) | |
| other lumbar surgery | 0.97 (0.63, 1.49) | 0.47 (0.26, 0.88) | 1.16 (0.72, 1.89) | 0.76 (0.48, 1.20) | 0.95 (0.62, 1.46) | 0.46 (0.25, 0.85) | 1.19 (0.73, 1.93) | 0.78 (0.49, 1.25) | |
|
| |||||||||
| Number of unique prescribers during exposure period | per additional prescriber | 2.44 (1.72, 3.46) | 2.14 (1.46, 3.13) | 2.48 (1.71, 3.61) | 2.51 (1.79, 3.54) | 2.79 (1.96, 3.98) | 2.59 (1.77, 3.80) | 2.60 (1.77, 3.81) | 2.47 (1.73, 3.53) |
|
| |||||||||
| Other medications during exposure period | Any vs. none | 1.30 (1.05, 1.61) | 1.12 (0.90, 1.41) | 1.26 (0.98, 1.63) | 1.58 (1.27, 1.97) | 1.30 (1.05, 1.61) | 1.15 (0.92, 1.44) | 1.27 (0.98, 1.64) | 1.56 (1.24, 1.96) |
|
| |||||||||
| Re-hospitalization | within 30 days of discharge | 0.65 (0.32, 1.33) | 1.01 (0.56, 1.81) | 1.75 (1.00, 3.06) | 1.40 (0.82, 2.39) | 0.68 (0.33, 1.39) | 1.06 (0.59, 1.89) | 1.82 (1.04, 3.19) | 1.45 (0.84, 2.50) |
| > 30 days and < 240 days of discharge | 1.30 (0.94, 1.79) | 2.37 (1.76, 3.19) | 1.41 (0.99, 2.00) | 1.84 (1.36, 2.49) | 1.31 (0.95, 1.80) | 2.43 (1.80, 3.26) | 1.46 (1.02, 2.07) | 1.89 (1.39, 2.58) | |
|
| |||||||||
| Number of comorbidities | 1 – 2 vs none | 0.85 (0.64, 1.13) | 0.87 (0.64, 1.19) | 0.71 (0.50, 0.99) | 0.93 (0.70, 1.24) | 0.90 (0.68, 1.19) | 0.90 (0.66, 1.23) | 0.80 (0.56, 1.12) | 1.12 (0.83, 1.51) |
|
|
|||||||||
| > 2 vs. none | 1.04 (0.80, 1.36) | 1.04 (0.77, 1.39) | 1.15 (0.84, 1.56) | 0.96 (0.73, 1.25) | 1.00 (0.77, 1.31) | 1.00 (0.75, 1.34) | 1.12 (0.82, 1.54) | 0.97 (0.73, 1.29) | |
Younger patients tended to cluster into groups 3, 4, and 5, while race and sex did not predict group membership. Concomitant medications and the number of unique prescribers all independently predicted the likelihood of membership in groups 2 to 5 compared to group 1. There were a total of 107 surgeries or 7.4% where there were no opioids found during the exposure period. Since only outpatient data was evaluated, these patients either took no opioids post op or consumption was limited to in patient prescribing only. 53 (49.5%) were opioid naïve preoperatively.
Risk Factors for COU Following Lumbar Surgery
Table 3 displays the results of multivariable modelling using GLMM regression to assess COU (yes/no; at least one opioid prescription filled ≥90 days after surgery). Pre-surgery opioid users had a 33% higher risk (RR 1.33, 95% CI 1.17–1.51) of COU. 30-day post-operative opioid use was significantly associated with chronic opioid use in a dose-dependent manner. Patients receiving >14 days of opioids in the 30-day post-operative period had 55% higher risk of COU (RR 1.55, 95% CI 1.24–1.94), as compared to 0 days. Sex, race, length of hospital stay, indication for lumbar surgery, concomitant medications, and comorbidity burden were not significantly associated with COU. COU risk was not directly determined based on procedure.
Table 3:
Generalized linear mixed models predicting chronic outcome. Model adjusted for age, sex, LOS, PT/OT/SNF, comorbidity burden. Each odds ratio from the multinomial model is an estimate of the likelihood of group membership referenced to group1
| Variable | Level | Risk Ratio (95% CI) |
|
|---|---|---|---|
| Opioid days exposure | MME exposure | ||
|
| |||
| Opioid naïve | No vs. Yes | 1.34 (1.18, 1.52) | 1.33 (1.17, 1.51) |
|
| |||
| Opioid days in 30 days following discharge | 1–5 days vs. 0 days | 1.08 (0.80, 1.46) | |
| 6–14 days vs. 0 days | 1.22 (0.96, 1.54) | ||
| > 14 days vs. 0 days | 1.55 (1.24, 1.94) | ||
|
| |||
| Mean daily MME in 30 days following discharge | 0<MME<20 vs. 0 MME | 1.21 (0.95, 1.53) | |
| 20≤ MME<50 vs. 0 MME | 1.36 (1.09, 1.71) | ||
| 50≤ MME<90 vs. 0 MME | 1.64 (1.30, 2.05) | ||
| MME ≥ 90 | 1.63 (1.29, 2.05) | ||
|
| |||
| Number opioid prescriptions during exposure period | per additional prescription | 1.05 (1.03, 1.07) | 1.04 (1.02, 1.06) |
| Number unique prescribers during exposure period | per additional prescriber | 1.01 (0.98, 1.04) | 1.02 (0.98, 1.05) |
|
| |||
| Patient age | 35–54 vs < 35 | 1.17 (1.07, 1.27) | 1.16 (1.06, 1.27) |
| 55–64 vs < 35 | 1.21 (1.10, 1.33) | 1.21 (1.10, 1.33) | |
| ≥ 65 vs < 35 | 1.13 (0.98, 1.31) | 1.09 (0.94, 1.26) | |
|
| |||
| Re-hospitalization | within 30 days of discharge | 1.06 (0.93, 1.20) | 1.05 (0.93, 1.20) |
| > 30 days & < 240 days of discharge | 1.20 (1.13, 1.26) | 1.20 (1.13, 1.27) | |
|
| |||
| Lumbar surgery type | degenerative disc | 1.00 (0.93, 1.08) | 1.00 (0.93, 1.07) |
| herniation | 0.96 (0.89, 1.04) | 0.98 (0.90, 1.06) | |
| stenosis | 0.94 (0.87, 1.01) | 0.95 (0.88, 1.02) | |
| spondylolisthesis | 1.03 (0.94, 1.12) | 1.03 (0.94, 1.13) | |
| other lumbar surgery | 0.92 (0.80, 1.05) | 0.91 (0.80, 1.04) | |
|
| |||
| Other medications during exposure period | Any vs. none | 1.05 (0.97, 1.12) | 1.05 (0.97, 1.13) |
Finally, we conducted a subgroup analysis of 101 patients undergoing repeat lumbar surgery during the study timeframe, of which 95 underwent two surgeries and 6 underwent three surgeries. Of those who underwent two surgeries, 69 (73%) were using opioids prior to the second lumbar surgery and remained chronic users and 19 (20%) were using opioids before and ceased afterwards.
Patients and Baseline Characteristics
Supplementary table 2 displays the baseline demographic characteristics and compares them based on the outcomes of COU. The most common indications for lumbar surgery included stenosis (45%), disc herniation (29%), degenerative disc disease (DDD) (23%), and spondylolisthesis (13%). Following lumbar surgery, more than three-quarters of the patients were deemed to be chronic opioid users (or COUs), that is, 76.4% filled at least one opioid prescription ≥90 days after lumbar surgery. In the bivariate analysis, patients with DDD were more likely to be COUs (24.8% vs. 18.6%; p=0.02). COUs were more likely to take opioids prior to surgery (88.5% vs. 61.9%; p<0.001), were dispensed higher amounts of opioids during the 30-days following surgery (mean MME 59.6 vs 25.1; p<0.001), received more concomitant medications and had more unique opioid prescribers (1.6 vs. 1.1; p<0.0001). Comorbid conditions including chronic pulmonary disease, substance use disorder, depression, and opioid use disorder were significantly more common in the chronic opioid cohort (Supplementary Table 3).
Discussion
There is a high prevalence of back pain in the US, and with historical treatment paradigms, there is also a high use of opioids in this population 19. The results of our study support these data, with three-quarters of those undergoing lower back surgery defined as chronic opioid users following surgery. Patients undergoing spinal surgery are more susceptible to higher opioid utilization 6. In a study assessing 90-day post-operative opioid use in orthopedic trauma patients, spine surgery patients had the longest length of stay, highest pain scores, and received the most opioid prescriptions compared with all other groups 20. Furthermore, opioid use among patients with low back pain is associated with worse surgical and non-surgical outcomes21–25. Based on this information, there is an opportunity for improvement in patient care. To that end, this study offers insights into risk factors for COU.
78% of all patients were taking opioids prior to surgery. This is slightly higher but comparable to a study by Dunn et al. where pre-operative intake was 70% 26. Due to the nature of retrospective studies, it is not possible to determine if some of these opioid prescriptions were solely for pre-operative use or for other conditions. Regardless, it is clear that a very high proportion were taking opioids preceding surgery, and piques concern about this particular patient population. Our findings also demonstrate a concerning troubling trend: conversion from opioid-naïve to chronic opioid users following lumbar surgery, which occurred in 50% of those not taking opioids prior to lumbar surgery. Although this was only 9% of the total cohort, it represents a potential failure of the care system and warrants further investigation.
Patient demographics predictive of COU were psychological in nature and included depression and substance abuse disorders. Multiple studies have observed increased risk of COU in patients with depression, anxiety, and substance abuse disorders 5,27–33Better treatment pathways including appropriate psychological optimization may help reduce COU 34.
Four cohorts of surgical diagnoses were evaluated: spondylolisthesis, stenosis, disc herniation, and DDD. Patients with DDD were 33% more likely to have COU. Prior studies suggest surgical outcomes for DDD are less predictable 35–39. This reemphasizes the importance of comprehensive treatment for patients with DDD in particular, including strategies to manage opioid use and addressing cognitive behavioral contributions to pain 40,41,42. This is also consistent with the guideline recommendations of the American College of Physicians 43.
The strongest associations were opioid-naïve to predict lack of COU development and opioid consumption before and 30 days after surgery to predict chronic use. Furthermore, the risk of COU increased substantially as opioid use increased during the 30-days after from lumbar surgery, consistent with several previous studies 28,44–47. The results of this analysis also question the potential benefits of preoperative weaning from opioids. There is minimal data to support its long-term benefit. In a recent publication, Jain et al. concluded that the two predictors of successful weaning following spinal fusion were last preoperative prescription greater than 2 months prior to surgery, and less than 14-day supply of medication preoperatively 48. This study has several weaknesses; however, it provides some initial data. There is a large opportunity to better understand the benefits of pre-operative weaning. Post-operative management strategies and addressing other pre-operative factors such as depression are also necessary. Studies assessing the usefulness of predictive analytics suggest that these factors are important for minimizing the COU following lumbar surgery 49,50.
Phenotype analysis yielded several novel and important findings. Three phenotypes were reassuring, but the remaining two were worrisome: one had a high COU over time (32.2%), and the other increased opioid use over time (12.9%). Factors that strongly predicted membership in these groups included opioid use prior to surgery, younger age, opioid use the 30-days after discharge (in a dose-dependent manner), multiple prescribers, concomitant medications and rehospitalization 30–270 days post-operatively. These factors also predicted COU outcomes in the logistic model. This points towards the benefit of opioid stewardship to reduce the likelihood of patients becoming sustained high users 51,52,53. These trajectories are also important in validating opioid use. Since opioid use was defined by filling a prescription and not by actual consumption, the analysis was subject to spurious error in defining COU. However, these trajectories clearly show longitudinal patterns over a 9 month period, and substantially mitigate against concern for high rates of false positives.
Finally, we assessed patients who underwent multiple lumbar surgeries during the study period, as there was significant interest in the expected benefit of revision surgery. The sample size was very small, but these data largely mirror primary surgery data.
Direct comparisons with commercially insured patients are difficult. However, this study demonstrates that there is likely some disparity in this population. In a study of Tricare patients, there was also a high percentage of patients taking preoperative opioids before spinal surgery 54. Unlike the population of patients in this study, most of those patients were weaned from opioids within a year of surgery. Opioid misuse in patients with spinal disorders often appears multifactorial. Opportunities for improvement will likely require access to comprehensive and multidisciplinary care both before and after surgery.
This study has several important limitations worthy of discussion. This is an administrative dataset and dependent on the fidelity of the data input. Opioid prescriptions were surrogates for opioid use. There is no way to know how much medication was consumed or if they were used for comorbid conditions. The group-based trajectories do demonstrate longitudinal utilization and provide more certainty regarding true positive COU. Although the overall sample size was large, some of the cohorts were relatively small, thus limiting our ability to draw robust conclusions. Although included in the model, postoperative provider visit data are difficult to interpret due to the global period associated with surgical charges, which may not capture all post-operative visits in the first 3 months following surgery. There may be a concern regarding the collection period ending in 2017. Based on time to for database availability, acquisition and analysis, this information is relatively current for this type of database. The methods grouped cohorts by diagnosis and not by procedure. Thus, influence of procedure type such as decompression versus fusion on COU cannot be accurately determined. The question regarding the value of pre-operative weaning looms large. These data were analyzed extensively, but unfortunately, there was not enough pre-operative information to accurately answer this question. Finally, this population was from South Carolina. It is an assumption to extrapolate this to Medicaid populations to other areas of the country, and there is likely to be some variation across regions. Despite these limitations, this is convincing evidence that there is substantial opioid use in this population.
Summary
Over 75% of SC Medicaid patients undergoing lumbar spine surgery for chronic lower back pain used opioids preoperatively and continued long-term use post-operatively. Furthermore, approximately 50% of opioid-naïve patients became chronic opioid users following surgery. This is higher than reported for other populations of cohorts. The most important predictor of postoperatively opioid use was preoperative and perioperative intake, which had an expected dose-response risk. DDD, multiple prescribers, depression, and concomitant medications are important factors associated with COU. This population appears more susceptible to chronic use than other US patient populations. Many risk factors are modifiable. Deliberate strategies to reduce long-term opioid intake postoperatively for those undergoing lumbar surgery are needed to successfully address the opioid epidemic in the US.
Supplementary Material
Key Points.
There is an approximate 75% prevalence of opioid use in Medicaid patients undergoing elective lumbar spine surgery.
9 months following surgery, about 75% of post-operative patients still use opioids
The primary predictors of opioid use post operatively are pre-operative naiveté, and dose dependent initial 30 day post-operative consumption
Benefits of pre-operative weaning are unknown at this time
Other risk factors include surgery for degenerative disc disease, multiple prescribers, depression, and concomitant medications.
Acknowledgments
Data used in this project was received through a contract (A201912450A) with the SC Department of Health and Human Services (SCDHHS) for Drug Utilization Review services and supported in part by the NIH National Institute on Drug Abuse (NIDA) grant DA0036566 NIH National Center for Advancing Translational Sciences (NCATS) through Grant Number UL1 TR001450. Neither SCDHHS nor NCATS played any role in the preparation, approval, or decision to submit the manuscript for publication.
This study was approved by the MUSC Office of the Research IRB (Pro00082280). This project was deemed not to be a human research project and was not subject to oversight by the MUSC IRB.
Contributor Information
Charles A. Reitman, Department of Orthopaedics and Physical Medicine, Medical University of South Carolina, 96 Jonathan Lucas St., Suite 708, MSC622, Charleston, SC 29425.
Ralph Ward, Department of Public Health Sciences, Medical University of South Carolina, 135 Cannon St., Charleston SC 29425.
David J. Taber, Department of Surgery, Medical University of South Carolina, 96 Jonathan Lucas Street, CSB 409, Charleston, SC 29425.
William P. Moran, Department of Medicine, Medical University of South Carolina, 75 Jonathan Lucas Street, Charleston, SC 29425.
Jenna McCauley, Department of Psychiatry and Behavioral Sciences, Medical University of South Carolina, 67 President Street, MSC861, Charleston, SC 29425.
William T. Basco, Jr., Department of Pediatrics, Medical University of South Carolina, 135 Rutledge Avenue, Room RT 280Q, Charleston, SC 29425.
Mulugeta Gebregziabher, Department of Public Health Sciences, Medical University of South Carolina, 135 cannon St., Suite 303, MSC 835, Charleston, SC 29425.
Mark Lockett, Department of Surgery, Medical University of South Carolina, 30 Courtenay Drive, BM 235 Thurmond Gazes, Charleston, SC 29425.
Sarah J. Ball, Department of Medicine, Medical University of South Carolina, 171 Ashley Avenue, Charleston, SC 29425.
References
- 1.Deyo R, Von Korff M, Duhrkoop D. Opioids for lower back pain. BMJ 2015;350:g6380. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Decker SL, Abdus S, Lipton B: Eligibility for and enrollment in Medicaid among nonelderly adults after implementation of the Affordable Care Act. Medical Care Research Review 2021; 1077558721996851. [DOI] [PubMed] [Google Scholar]
- 3.Brummett C, Evans-Shields J, England C, et al. Increased healthcare costs associated with persistent opioid use after major surgery in opioid-naïve patients. J Manag Care Spec Pharm 2021;27:760–71 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lee J, Vu J, Edelman A, et al. Healthcare spending and new persistent opioid use after surgery. Ann Surg 2020:272:99–104 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lee D, Armaghani S, Archer K, et al. Preoperative opioid use as a predictor of adverse postoperative self-reported outcomes in patients undergoing spine surgery. J Bone Joint Surg Am 2014;96:e89(1–8). [DOI] [PubMed] [Google Scholar]
- 6.Lo Y, Lim-Watson M, Seo Y, et al. Long-term opioid prescriptions after spine surgery: a meta-analysis of the prevalence and risk factors. World Neurosurgery 2020;141:e894–e920. [DOI] [PubMed] [Google Scholar]
- 7.Stratton A, Wai E, Kingwell S, Phan P, Roffey D, E Koussy M, et al. Opioid use trends in patients undergoing elective thoracic and lumbar spine surgery Can J Surg 2020:63:e306–e312 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Basco WT Jr., Ward RC, Taber DJ et al. Patterns of dispensed opioids after tonsillectomy in children and adolescents in South Carolina, United States, 2010–2017. Int J Pediatr Otorhinolaryngol. 2021;143:110636. [DOI] [PubMed] [Google Scholar]
- 9.Lurie JD, Tosteson AN, Deyo RA, et al. Indications for spine surgery: Validation of an administrative coding algorithm to classify degenerative diagnoses. Spine 2014;39(9):769. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Doermann A, Masse DH, Mayo BC, et al. ICD-10 and its relevance to spinal surgeons. Contemporary Spine Surgery 2016;17(9):1–5. [Google Scholar]
- 11.Kazberouk A, Martin BI, Stevens JP, McGuire KJ. Validation of an administrative coding algorithm for classifying surgical indications and operative features of spinal surgery. Spine 2015;40(2):114–120. [DOI] [PubMed] [Google Scholar]
- 12.U.S. Centers for Disease Control and Prevention (2019). Opioid overdoses data resources. https://www.cdc.gov/drugoverdose/resources/data.html. Accessed June 20, 2019.
- 13.Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Medical Care 2005. 43(11), 1130–1139. [DOI] [PubMed] [Google Scholar]
- 14.Jones BL, Nagin DS. Advances in Group-Based Trajectory Modeling and SAS Procedure for Estimating Thems. Sociological methods and research 2007;35(4):542–571. [Google Scholar]
- 15.Nagin D Group-based modelling of development. Cambridge: Harvard University Press, 2005. [Google Scholar]
- 16.Nagin DS, Odgers CL. Group-based trajectory modelling in clinical research. Annual Review of Clinical Psychology 2010;6(1):109–138. [DOI] [PubMed] [Google Scholar]
- 17.Zou G Modified Poisson regression approach to prospective studies with binary data. Am J Epidemiol 2004; 159(7):702–6. [DOI] [PubMed] [Google Scholar]
- 18.Lacolio AR, Margolis DJ, Berlin JA. Relative risks and confidence intervals were easily computed indirectly using multivariable logistic regression. J Clin Epidemiol 2007;60(9):874–882. [DOI] [PubMed] [Google Scholar]
- 19.Deweerdt S Natural history of the epidemic. Nature 2019;573:S10 to S12. [DOI] [PubMed] [Google Scholar]
- 20.Fisher N, Hooper J, Bess S, Leucht P, et al. Ninety-day postoperative narcotic use after hospitalization for orthopaedic trauma. J Am Acad Orthop Surg 2020;28:e560–565. [DOI] [PubMed] [Google Scholar]
- 21.Mendoza-Elias N, Dunbar M, Ghogawala Z, Whitmore R: Opioid use, risk factors, and outcome in lumbar fusion surgery. World Neurosurgery 2020;135:e580–87. [DOI] [PubMed] [Google Scholar]
- 22.Frazer K, Stevermer J. More is not better with acute low back pain treatment. The Journal of Family Practice 2016:65:404–406 [PMC free article] [PubMed] [Google Scholar]
- 23.Jain N, Phillips F, Weaver T, Khan S. Preoperative Chronic Opioid Therapy. Risk factors for complications, readmission, continued opioid use, and increased costs after one- and two-level posterior lumbar fusions. Spine 2018:43:1331–38 [DOI] [PubMed] [Google Scholar]
- 24.Villavicencio A, Nelson E, Kantha V, Burneikiene S. Prediction based on preoperative opioid use of clinical outcomes after transforaminal lumbar interbody fusion. J Neurosurg Spine 2017;26:144–49. [DOI] [PubMed] [Google Scholar]
- 25.White A, Arnold P, Norvell D, et al. Pharmacological management of chronic low back pain. Spine 2011;36:S131–S143 [DOI] [PubMed] [Google Scholar]
- 26.Dunn L, Yerra S, Fang S, et al. Incidence and risk factors for chronic postoperative opioid use after major spine surgery: A cross-sectional study with longitudinal outcomes. Anesth Analg 2018:127:247–54 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Brummett C, Waljee J, Goesling J, et al. New persistent opioid use after minor and major surgical procedures in adults in the US. JAMA Surgery 2017:152(6):e170504. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Costelloe C, Burns S, et al. , Urman R. Analysis of predictors of persistent postoperative pain in spinal surgery. Current Pain and Headache Reports 2020;24:11. [DOI] [PubMed] [Google Scholar]
- 29.Feingold D, Brill S, Goor-Aryeh I, et al. Association between severity of depression and prescription opioid misuse among chronic pain patients with and without anxiety: A cross-sectional study. Journal of Affective Disorders. 2018:235:293–302 [DOI] [PubMed] [Google Scholar]
- 30.Feingold D, Brill S, Goor-Aryeh I, et al. Misuse of prescription opioids among chronic pain patients suffering from anxiety: a cross-sectional analysis. General Hospital Psychiatry 2017;47:36–42 [DOI] [PubMed] [Google Scholar]
- 31.Martel M, Bruneau A, Edwards R: Mind-body approaches targeting the psychological aspects of opioid use problems in patients with chronic pain: Evidence and opportunities. Transl Res 2021;234:114–128 [DOI] [PubMed] [Google Scholar]
- 32.Sun E, Darnall B, Baker L, Mackey S. Incidence of and Risk Factors for Chronic Opioid Use among Opioid-Naive Patients in the Postoperative Period. JAMA Intern Med 2016;176:1286–93 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Webster L Risk Factors for Opioid-Use Disorder and Overdose. Anesth Analg 2017;125:1741–1748 [DOI] [PubMed] [Google Scholar]
- 34.Hruschak V, Cochran G, Wasan AD. Psychosocial interventions for chronic pain and comorbid prescription opioid use disorders: A narrative review of the literature. J Opioid Manag 2018, Vol.14(5), p.345–358 [DOI] [PubMed] [Google Scholar]
- 35.Fritzell P, Hagg O, Wessberg P, Nordwall A. 2001 Volvo Award Winner in Clinical Studies: lumbar fusion versus nonsurgical treatment for chronic low back pain. Spine 2001;26:2521–34 [DOI] [PubMed] [Google Scholar]
- 36.Hedlund R, Johansson C, Hägg O, et al. The long-term outcome of lumbar fusion in the Swedish lumbar spine study. Spine J. 2016;16:579–87 [DOI] [PubMed] [Google Scholar]
- 37.Weinstein J, Tosteson T, Lurie J, et al. Surgical vs nonoperative treatment for lumbar disk herniation: the Spine Patient Outcomes Research Trial (SPORT): a randomized trial. N Engl J Med 2008;358:794–810. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Abdu W, Lurie J, Spratt K, et al. Degenerative spondylolisthesis: does fusion method influence outcome? Four-year results of the spine patient outcomes research trial. Spine 2009;34(21):2351–60 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Weinstein J, Tosteson T, Lurie J, et al. Surgical versus nonoperative treatment for lumbar spinal stenosis four-year results of the Spine Patient Outcomes Research Trial. Spine 2010;35:1329–38 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Brox J, Sorensen R, Friis A, et al. Randomized clinical trial of lumbar instrumented fusion and cognitive intervention and exercises in patients with chronic low back pain and disc degeneration. Spine 2003;28:1913–21 [DOI] [PubMed] [Google Scholar]
- 41.Fairbank J, Frost H, Wilson-MacDonald J, et al. Randomised controlled trial to compare surgical stabilisation of the lumbar spine with an intensive rehabilitation programme for patients with chronic low back pain: the MRC spine stabilisation trial. BMJ 2005;330(7502):1233. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.O’Sullivan P, Caneiro J, O’Keeffe M, et al. Cognitive functional therapy: an integrated behavioral approach for the targeted management of disabling low back pain. Physical Therapy 2018;98:408–23. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Qaseem A, Wilt T, McLean R, Forciea M. Noninvasive treatments for acute, subacute, and chronic low back pain: A clinical practice guideline from the American College of Physicians. Ann Intern Med 2017;166:514–30 [DOI] [PubMed] [Google Scholar]
- 44.Deyo R, Hallvik S, Hildebran C, et al. Use of prescription opioids before and after an operation for chronic pain (lumbar fusion surgery). Pain 2018;159:1147–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Connoly J, Javed Z, Raji M, et al. Predictors of lont=term opioid use following lumbar fusion surgery. Spine 2017;42:1405–11 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46.Rosenthal B, Suleiman L, Kannan A, et al. Risk factors for prolonged postoperative opioid use after spine surgery: A review of dispensation trends from a state=run prescription monitoring program. J Am Acad Orthop Surg 2019;27:32–38. [DOI] [PubMed] [Google Scholar]
- 47.Schoenfeld A, Belmont P, Blucher J, et al. Sustained preoperative opioid use is a predictor of continued use following spine surgery. J Bone Joint Surg Am 2018;100:914–21. [DOI] [PubMed] [Google Scholar]
- 48.Jain N, Phillips F, Malik A, Khan S. Preoperative opioid weaning before major spinal fusion. Spine 2020;46:80–86. [DOI] [PubMed] [Google Scholar]
- 49.Karhade A, Ogink P, Thio Q, et al. Development of machine learning algorithms for prediction of prolonged opioid prescription after surgery for lumbar disc herniation. Spine J. 2019;19:1764–71 [DOI] [PubMed] [Google Scholar]
- 50.Karhade A, Cha T, Fogel H, et al. Predicting prolonged opioid prescriptions in opioid-naïve lumbar spine surgery patients. The Spine J 2020:20:88–95 [DOI] [PubMed] [Google Scholar]
- 51.Varley PR, Zuckerbraun BS. Opioid stewardship and the surgeon. JAMA Surgery 2018;153(2):e174875–e174875. [DOI] [PubMed] [Google Scholar]
- 52.Hyland SJ, Brockhaus KK, Vincent WR, et al. (2021, March). Perioperative pain management and opioid stewardship: A practical guide. In Healthcare (Vol. 9, No. 3, p. 333). Multidisciplinary Digital Publishing Institute. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Anderson I, Alger J. The tightrope walk: pain management and opioid stewardship. Orthopaedic Nursing 2019;38(2), 111–115. [DOI] [PubMed] [Google Scholar]
- 54.Schoenfeld AJ, Jiang W, Chaudhary MA, et al. Sustained Prescription Opioid Use Among Previously Opioid-Naive Patients Insured Through TRICARE (2006–2014). JAMA Surg 2017;152(12):1175–76. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.