CLINICAL ARTICLE

J Neurosurg Spine 35:275–283, 2021

The change in postoperative opioid prescribing after

lumbar decompression surgery following state-level opioid
prescribing reform
Robert D. Winkelman, MD, MS,1–3 Michael D. Kavanagh, MD,1,2 Joseph E. Tanenbaum, MD, PhD,1,2,4
Dominic W. Pelle, MD,1,5 Edward C. Benzel, MD,1,3 Thomas E. Mroz, MD,1,5 and
Michael P. Steinmetz, MD1,3
Center for Spine Health, Neurological Institute, and Departments of 3Neurosurgery and 5Orthopaedic Surgery, Cleveland Clinic;
1

Case Western Reserve University School of Medicine; and 4Department of Population and Quantitative Health Sciences, Case
2

Western Reserve University School of Medicine, Cleveland, Ohio

OBJECTIVE On August 31, 2017, the state of Ohio implemented legislation limiting the dosage and duration of opioid
prescriptions. Despite the widespread adoption of such restrictions, few studies have investigated the effects of these
reforms on opioid prescribing and patient outcomes. In the present study, the authors aimed to evaluate the effect of
recent state-level reform on opioid prescribing, patient-reported outcomes (PROs), and postoperative emergency depart-
ment (ED) visits and hospital readmissions after elective lumbar decompression surgery.
METHODS This study was a retrospective cohort study of patients who underwent elective lumbar laminectomy for de-
generative disease at one of 5 hospitals within a single health system in the years prior to and after the implementation
of the statewide reform (September 1, 2016–August 31, 2018). Patients were classified according to the timing of their
surgery relative to implementation of the prescribing reform: before reform (September 1, 2016–August 31, 2017) or after
reform (September 1, 2017– August 31, 2018). The outcomes of interest included total outpatient opioids prescribed in
the 90 days following discharge from surgery as measured in morphine-equivalent doses (MEDs), total number of opioid
refill prescriptions written, patient-reported pain at the first postoperative outpatient visit as measured by the Numeric
Pain Rating Scale, improvement in patient-reported health-related quality of life as measured by the Patient-Reported
Outcomes Measurement Information System–Global Health (PROMIS-GH) questionnaire, and ED visits or hospital re-
admissions within 90 days of surgery.
RESULTS A total of 1031 patients met the inclusion criteria for the study, with 469 and 562 in the before- and after-
reform groups, respectively. After-reform patients received 26% (95% CI 19%–32%) fewer MEDs in the 90 days following
discharge compared with the before-reform patients. No significant differences were observed in the overall number
of opioid prescriptions written, PROs, or postoperative ED or hospital readmissions within 90 days in the year after the
implementation of the prescribing reform.
CONCLUSIONS Patients undergoing surgery in the year after the implementation of a state-level opioid prescribing
reform received significantly fewer MEDs while reporting no change in the total number of opioid prescriptions, PROs, or
postoperative ED visits or hospital readmissions. These results demonstrate that state-level reforms placing reasonable
limits on opioid prescriptions written for acute pain may decrease patient opioid exposure without negatively impacting
patient outcomes after lumbar decompression surgery.
https://thejns.org/doi/abs/10.3171/2020.11.SPINE201046

I
KEYWORDS postoperative opioid prescribing; opioid reform; lumbar laminectomy

n 2017, 47,600 people died of opioid overdose in the Prevention (CDC) released guidelines in 2016 calling for
United States, and more than 35% of these deaths a 1-week limit on opioid medications as management for
involved a prescription opioid medication.1,2 To curb acute pain.3 More than half of all states enacted some form
the medical community’s contribution to the opioid epi- of opioid prescribing restriction in response to the CDC
demic, the United States Centers for Disease Control and recommendation.4,5 At the federal level, the Centers for

ABBREVIATIONS CDC = Centers for Disease Control and Prevention; EMR = electronic medical record; MED = morphine-equivalent dose; NPRS = Numeric Pain Rating
Scale; PDMP = prescription drug monitoring program; PRO = patient-reported outcome; PROMIS-GH = Patient-Reported Outcomes Measurement Information System–
Global Health.
SUBMITTED June 9, 2020. ACCEPTED November 2, 2020.
INCLUDE WHEN CITING Published online July 9, 2021; DOI: 10.3171/2020.11.SPINE201046.

©AANS 2021, except where prohibited by US copyright law J Neurosurg Spine Volume 35 • September 2021 275

Unauthenticated | Downloaded 05/11/24 02:12 PM UTC

Winkelman et al.

Medicare & Medicaid Services implemented a 7-day limit Primary Outcome

for all opioid prescriptions to opioid-naive Medicare Part The primary outcome measure in this study was the
D beneficiaries that went into effect on January 1, 2019.6 total morphine-equivalent doses (MEDs) of all outpatient
While these prescribing reforms aim to reduce opioid-re- opioids prescribed by the surgical team within the 90 days
lated adverse events such as chronic use, overdose, death, following discharge. The surgical team was defined as
and illicit diversion, it remains unclear how these reforms surgeons, residents, and advanced practice providers who
may impact opioid prescribing after elective surgery when assisted the primary surgeon in the postoperative manage-
outpatient opioids are a staple of postoperative pain man- ment of their patients. We believed that postoperative out-
agement. patient opioid prescribing up to 90 days after discharge
Although opioid medications are commonly used to from surgery represented a reasonable maximum for
manage postoperative pain, the amount of outpatient which a surgical team provider may have continued to pre-
opioids prescribed following surgery varies significantly scribe opioid therapy related to surgery. Opioid prescrib-
across surgical specialties, providers, and procedures.7–16 ing for pain lasting more than 3 months (i.e., > 90 days) is
Prior studies found that patients undergoing spine surgery generally considered chronic opioid prescribing according
receive larger postoperative opioid prescriptions compared to CDC guidelines, and thus prescribing beyond 90 days
with patients in other surgical subspecialties.9,10 Previous was not examined in the present study.3
research also indicated a high prevalence of chronic opi- Opioid prescribing data were obtained by querying
oid use following spine surgery, with estimates reaching the EMR database for outpatient opioid prescription or-
as high as 59%.17–20 Given the liberal utilization of opioid ders that were placed within 90 days following discharge
analgesia after spine surgery, new restrictions on opioid from surgery. The theoretical daily MEDs and duration
prescribing may negatively impact spine surgery providers for each opioid prescription were calculated using the total
and patients, leading to more prescription refills and worse pill count, dose regimen, and a standard morphine-equiv-
patient-reported outcomes (PROs). alent conversion value specific to the opioid compound
In the present study, we quantified the effect of a state- in the prescription.24 After all opioid prescriptions were
level reform in Ohio on opioid prescribing practices af- converted to MEDs, the MEDs prescribed within 90 days
ter elective lumbar decompression surgery. On March 30, following discharge were totaled for each patient. Of note,
2017, the state of Ohio announced new legislation to limit only patients who received at least one outpatient opioid
outpatient opioid prescriptions for acute pain.21 The new prescription in the postoperative period were included in
law, which went into effect on August 31, 2017, restricts the analysis of the primary outcome.
initial opioid prescriptions for acute pain to a duration of
no more than 7 days.22,23 Using electronic medical record Secondary Outcomes
(EMR) data from a large multihospital healthcare system, The MEDs and frequency of discharge and postdis-
we compared the amount of outpatient opioids prescribed charge refill prescriptions for opioids were compared be-
within 90 days of hospital discharge following elective tween the year before and the year after the Ohio opioid
lumbar decompression surgery in the year before and the prescribing reform. The types of opioid prescribed were
year after the introduction of the Ohio opioid prescrib- also compared between the two study years. Since tablets
ing reform. To assess the new law’s impact on physician containing 5 mg oxycodone or 5 mg hydrocodone were
outpatient prescribing practices and patient outcomes, we by far the most prevalent types of outpatient opioids pre-
also quantified changes in the prescribing frequency of scribed, each discharge and refill opioid prescription was
other analgesic medications, patient-reported postopera- classified into one of three groups: oxycodone 5 mg, hy-
tive pain and health-related quality of life, and the inci- drocodone 5 mg, or other (which included opioid prescrip-
dence of postoperative emergency department visits and tions for all other doses or other opioid compounds). The
readmissions in the year after the reform. prescribing frequency of other adjunct analgesic medica-
tions, such as skeletal muscle relaxants (e.g., cyclobenzap-
Methods rine, tizanidine, methocarbamol) and gabapentinoids (ga-
bapentin, pregabalin), within the 90 days following dis-
Population charge was also compared between the two study years.
This study is a retrospective cohort study of patients PRO data of interest included patient-reported pain at
who underwent elective lumbar laminectomy surgery for the first postoperative visit and improvement in health-
degenerative disease of the lumbar spine at one of 5 loca- related quality of life as measured by the Numeric Pain
tions within a single regional health system in Ohio from Rating Scale (NPRS) and the Patient-Reported Out-
September 1, 2016, through August 31, 2018. These pa- comes Measurement Information System–Global Health
tients were identified by querying the EMR database for (PROMIS-GH) questionnaire, respectively. The NPRS
all surgical cases billed using Current Procedural Termi- and PROMIS-GH questionnaires are routinely adminis-
nology (CPT) code 63047 (single-level laminectomy) with tered to all spine surgery patients during outpatient visits
or without the addition of 63048 (additional lumbar lami- prior to being seen by their provider as part of an institu-
nectomy at the adjacent vertebral interspace). The opera- tional prospective patient outcomes registry. Responses on
tive report for each case was reviewed, and patients were the NPRS questionnaire range from zero (no pain) to 10
excluded from the study for the following: the surgical (worst pain of life), whereas the mental health and physical
indication was not for degenerative disease or the surgery health subscores of the PROMIS-GH questionnaire range
was not performed on an elective basis. from approximately 16 (worst health) to 68 (best health).25

276 J Neurosurg Spine Volume 35 • September 2021

Unauthenticated | Downloaded 05/11/24 02:12 PM UTC

 Winkelman et al.

Improvement in each of the PROMIS-GH subscores was The primary study question was assessed using a log-
calculated by taking the score at the patient’s first post- linear regression model of total MEDs prescribed within
operative visit and subtracting the preoperative score (the 90 days following discharge from surgery as a function
last recorded score within 6 months prior to surgery). of timing of surgery relative to the reform. To assess how
Thus, a positive value would indicate improved mental or inferences from the matched analysis might change in the
physical health after surgery. Improvement in each of the presence of hidden bias, a Rosenbaum sensitivity test for
PROMIS-GH subscores was only calculated for patients Wilcoxon signed-rank p value was performed on the pri-
having both preoperative and postoperative values. Lastly, mary outcome.29 Secondary outcomes were analyzed uti-
the incidence of patient emergency department visits and lizing the same methods as described in analyses of base-
hospital readmissions within 90 days of the initial surgery line characteristics. Of note, only patients who had both
were also compared between the two study years. preoperative and postoperative PROMIS-GH scores were
included in the analysis of improvement in PROMIS-GH
Patient Demographics score. Additionally, bootstrapped point estimates and con-
Patient age, sex, race, medical comorbidities, location fidence intervals for differences in means or proportions/
of surgery, and hospital length of stay were obtained by percentages were calculated to estimate effect sizes for
querying the EMR database. A manual chart review was each secondary outcome.30 Study data were analyzed us-
performed to determine history of preoperative opioid ing R and RStudio and the tidyverse package.31,32
use, history of prior spine surgery, American Society of
Anesthesiologists Physical Status Classification score, and Results
number of surgical levels (1–2 vs ≥ 3 levels) for each pa- Over the 2-year study period, 1031 patients met inclu-
tient. A patient was classified as a preoperative chronic sion criteria. Of these patients, 469 patients underwent
opioid user if 1) they were actively taking opioids on ad- lumbar laminectomy surgery in the year prior to the Ohio
mission and 2) a manual review of their chart indicated opioid prescribing reform, and 562 patients underwent
daily opioid use in the 3 months prior to surgery. surgery in the year after the reform. Baseline characteris-
tics for these two groups are shown in Table 1. Notable dif-
Statistical Analysis ferences between these two groups include a significantly
Each patient was classified into one of two groups. The lower overall prevalence of preoperative chronic opioid
before-reform group included patients who underwent use and shorter lengths of stay in the patients undergo-
surgery in the year prior to the implementation of the Ohio ing surgery in the year after the reform. Additionally, pa-
opioid prescribing reform (September 1, 2016, through tients undergoing surgery in the year after the reform had
August 31, 2017), and the after-reform group included a lower prevalence of diabetes and higher preoperative
patients who underwent surgery in the year following PROMIS-GH subscores. One-to-one matching of the two
the reform (September 1, 2017, through August 31, 2018). groups yielded 373 matched patient pairs, and no signifi-
Patients with missing preoperative NPRS and PROMIS- cant differences in baseline variables were observed after
GH scores imputed via predictive mean matching from matching (Table 1).
a model specified using nonmissing variables.26 Baseline Unadjusted analysis of the primary outcome showed
patient demographics were compared between the two that patients in the year following the Ohio opioid pre-
groups using the t-test for independent samples and the scribing reform were prescribed 26% (95% CI 19%–32%)
Mann-Whitney U-test for normally and nonnormally dis- fewer MEDs within 90 days following surgery compared
tributed variables, respectively. Categorical variables were with patients having surgery in the year before the reform.
compared using Pearson’s chi-square test. This decrease corresponds to an average decrease of 138
Both unadjusted and propensity score–based matched fewer MEDs (95% CI 114–160) prescribed per patient,
analyses were used to compare the differences in outcomes which is approximately 18 fewer tablets of the most com-
between the two study groups. To account for baseline dif- monly prescribed postoperative opioid medication (5 mg
ferences in measured covariates between the two groups, oxycodone–325 mg acetaminophen, 7.5 MEDs/tablet).
one-to-one matching without replacement of subjects was Figure 1 illustrates how the decrease in MEDs developed
performed using exact matching on preoperative opioid use over the 2-year study period. The matched analysis also
and hospital length of stay and then greedy caliper match- demonstrated a significant, although less pronounced,
ing on the propensity score. The propensity score for each decrease in postoperative opioids prescribed in the year
patient was calculated using a logistic regression model following reform. After matching, patients undergoing
where timing of surgery relative to the opioid reform (i.e., surgery in the year following reform received 23% (95%
before vs after reform) was the dependent variable. Inde- CI 15%–30%) fewer MEDs within 90 days after surgery
pendent variables in the propensity model included age, compared with patients who had surgery in the year prior
sex, race, location of surgery, number of levels of surgery, to reform (121 fewer MEDs, 95% CI 93–138). The Rosen-
length of stay, preoperative opioid use, preoperative NPRS baum sensitivity analysis for the observed difference in
score, and preoperative PROMIS-GH subscores. To assess the primary outcome of the matched cohort analysis yield-
adequacy of matching, covariate balance before and after ed a design sensitivity, Γ, of 1.6.
matching was examined using the absolute standardized
mean differences of covariates and the standardized mean Secondary Opioid Outcomes
and variance differences of propensity scores.27,28 The results of the analyses of secondary opioid out-

J Neurosurg Spine Volume 35 • September 2021 277

Unauthenticated | Downloaded 05/11/24 02:12 PM UTC

Winkelman et al.

TABLE 1. Demographics of unadjusted and propensity-matched cohorts

Unadjusted Cohorts Propensity-Matched Cohorts
Before Reform After Reform p Value Before Reform After Reform p Value
No. of patients 469 562 373 373
Age in yrs, mean (SD) 66.50 (11.94) 65.85 (11.66) 0.38 65.97 (11.90) 65.92 (11.21) 0.96
Male sex 317 (67.6) 366 (65.1) 0.44 263 (70.5) 252 (67.6) 0.43
Race 0.62 0.94
White 426 (90.8) 500 (89.0) 330 (88.5) 333 (89.3)
Black 25 (5.3) 36 (6.4) 25 (6.7) 23 (6.2)
Other 18 (3.8) 26 (4.6) 18 (4.8) 17 (4.6)
Surgery at tertiary center 303 (64.6) 343 (61.0) 0.26 249 (66.8) 236 (63.3) 0.36
Chronic obstructive pulmonary disorder 51 (10.9) 62 (11.0) >0.99 37 (9.9) 38 (10.2) >0.99
Congestive heart failure 13 (2.8) 16 (2.8) >0.99 11 (2.9) 9 (2.4) 0.82
Diabetes 81 (17.3) 69 (12.3) 0.03 55 (14.7) 54 (14.5) >0.99
Chronic kidney disease 13 (2.8) 8 (1.4) 0.19 6 (1.6) 7 (1.9) >0.99
American Society of Anesthesiologists class 0.73 0.86
I or II 104 (22.2) 136 (24.2) 88 (23.6) 83 (22.3)
III 345 (73.6) 404 (71.9) 272 (72.9) 275 (73.7)
IV 20 (4.3) 22 (3.9) 13 (3.5) 15 (4.0)
Preop chronic opioid use 182 (38.8) 158 (28.1) <0.01 127 (34.0) 127 (34.0) >0.99
Length of stay in days, median [IQR] 2 [1–3] 1 [1–2] <0.01 1 [1–2] 1 [1–2] >0.99
Discharge to skilled nursing or acute rehab facility 52 (11.1) 44 (7.8) 0.09 25 (6.7) 24 (6.4) >0.99
History of prior spine surgery 115 (24.5) 138 (24.6) >0.99 86 (23.1) 87 (23.3) >0.99
1- to 2-level decompression 377 (80.4) 453 (80.6) 0.99 305 (81.8) 300 (80.4) 0.71
Preop NPRS score, mean (SD) 5.35 (2.81) 5.26 (2.81) 0.61 5.30 (2.83) 5.23 (2.76) 0.76
Missing 77 (16.4) 97 (17.3)
Preop PROMIS-GH mental subscore, mean (SD) 45.00 (8.45) 46.41 (9.01) 0.01 45.34 (8.48) 45.57 (8.58) 0.71
Missing 77 (16.4) 97 (17.3)
Preop PROMIS-GH physical subscore, mean (SD) 36.79 (6.07) 37.73 (6.65) 0.02 37.32 (6.00) 37.22 (6.35) 0.82
Missing 77 (16.4) 97 (17.3)
Values represent the number of patients (%) unless stated otherwise.

comes are presented in Table 2. The total MEDs of dis- 42 and 41 days after discharge for the before and after the
charge and refill opioid prescriptions were significantly reform, respectively. Patient-reported outcome data were
lower in the year following the reform (difference in dis- available for 75.5% and 77.9% of patients in the year be-
charge MEDs −126 [95% CI −156 to −96]), difference in fore and after the opioid reform, respectively. There was
refill MEDs −162 [95% CI −236 to −108]). The percent- also no significant difference in patient-reported pain at
age of patients receiving prescriptions for discharge opi- the first postoperative visit or the degree of improvement
oids (93.8% vs 94.5%, percent difference 0.7% [95% CI for either of the PROMIS subscores following surgery be-
−2.0% to 3.5%]), postdischarge opioid refills (24.1% vs tween the two groups (Table 2). Lastly, there was no sig-
27.6%, percent difference 3.6% [95% CI −1.6% to 9.1%]), nificant difference in the incidence of patient emergency
or adjunct pain medications (skeletal muscle relaxants, department visits and hospital readmissions within 90
42.4% vs 39.9%, percent difference −2.3% [95% CI −7.9% days of the initial surgery between the two study years
to 4.0%]; gabapentinoids, 17.3% vs 19.4%, percent differ- (Table 2).
ence 2.1% [95% CI −2.7% to 7.2%]) were not significantly
different between the year before and the year after the
reform. Additionally, the total count of opioid refill pre- Discussion
scriptions as well as the types of opioids prescribed were We conducted a retrospective review of postoperative
not significantly different between the two study years opioid prescribing in patients who underwent lumbar de-
(Table 2). compression surgery in the year before and the year after
implementation of the 2017 Ohio reform. Patients who
PROs and Other Outcomes had surgery in the year after the reform were prescribed
The results of the analyses of PROs are presented in significantly less outpatient opioid medication within the
Table 2. The median time of postoperative follow-up was 90 days following discharge when compared with patients

278 J Neurosurg Spine Volume 35 • September 2021

Unauthenticated | Downloaded 05/11/24 02:12 PM UTC

 Winkelman et al.

FIG. 1. Total MEDs prescribed in 90 days after discharge versus the month the surgery was performed. The dashed line repre-
sents a locally weighted smoothed regression line of the exponentiated mean of log(MEDs) prescribed in each month over the
2-year study period.

in the year prior to the reform. Notably, despite the large scriptions for total knee arthroplasty, breast lumpectomy,
reduction in total outpatient MEDs prescribed, there were and cholecystectomy.35 Thus, the present study adds to the
no significant differences in the proportion of patients growing body of literature on this topic by demonstrat-
who received a discharge or postdischarge refill prescrip- ing that statewide prescribing limits may reduce outpa-
tion for opioids between the two study years. Additionally, tient opioid prescriptions. Although the exact mechanism
there were no significant differences observed between of this finding is unclear, one possible explanation is that
the two groups in the types of opioid medications pre- state-level reforms change provider and patient expecta-
scribed, the prescribing rate of adjunct analgesic medica- tions regarding the appropriate duration of postoperative
tions, patient-reported pain at the first postoperative visit, opioid therapy. Additionally, shorter durations of opioid
improvement in health-related quality of life within 90 prescriptions require providers to more frequently assess
days of discharge, or the incidence of emergency depart- the indication for opioid analgesics, which may help to
ment visits or readmissions within 90 days of discharge prevent unnecessarily lengthy courses of postoperative
Our results demonstrate a decrease in total postopera- opioid therapy.
tive opioid prescribing after elective spine surgery follow- The statewide prescribing efforts also helped to cata-
ing the implementation of a state-level prescribing reform. lyze institutional change related to postoperative pain
Unlike the decrease in opioid prescribing reported in a re- management. Following the announcement of the 2017
cent study of nonoperative trauma patients following the Ohio opioid reform, the health system under study adopt-
2017 Ohio reform, the overall decrease in the quantity of ed new guidelines on methods to set expectations with
opioids prescribed in the present study was not associated patients regarding the management of their postoperative
with a difference in prescribing rates of outpatient opi- pain and updated EMR prescribing workflows to aid pro-
oids following discharge.33 Although the frequency of dis- vider compliance with the new prescribing restrictions.
charge and refill opioid prescriptions did not change sig- Prior to the introduction of statewide opioid prescribing
nificantly, the average size of discharge and refill opioid limits, state-level efforts to curb outpatient opioid pre-
prescriptions decreased in the year following the intro- scribing were limited primarily to prescription drug moni-
duction of the prescribing reform. Previous studies have toring programs (PDMPs). Although research has demon-
also found reduced postoperative opioid prescribing for strated that PDMPs may help to decrease opioid-related
several different procedures following the introduction deaths and statewide opioid prescribing,37,38 the ability of
of state-level reforms.34–36 Reid et al. found that follow- PDMPs to curb postoperative opioid prescribing has been
ing the introduction of a 2017 Rhode Island opioid pre- less encouraging. In a previous single-center study of 1057
scribing reform, lumbar spine surgery patients received general surgery patients, study investigators reported that
significantly smaller discharge opioid prescriptions (59% the state-mandated use of a PDMP prior to prescribing
reduction) as well as 30% less opioid medication overall outpatient opioids had no impact on reducing the rate or
in the 30 days after surgery.34 Similarly, Porter et al. found amount of postoperative opioid prescriptions.39 In contrast,
that discharge MEDs were reduced for 18 of 25 differ- the adoption of institutional postoperative opioid prescrib-
ent surgeries following a 2018 Florida prescribing reform, ing guidelines has demonstrated success in reducing post-
with greater than 40% decreases in discharge opioid pre- operative opioid prescriptions, with decreases ranging

J Neurosurg Spine Volume 35 • September 2021 279

Unauthenticated | Downloaded 05/11/24 02:12 PM UTC

 280
Winkelman et al.

TABLE 2. Secondary outcomes in unadjusted and propensity-matched cohorts

Unadjusted Cohorts Propensity-Matched Cohorts
Before Reform After Reform Difference (95% CI)* p Value Before Reform After Reform Difference (95% CI)* p Value
No. of patients 469 562 373 373
Opioid outcomes
Received opioids at discharge 440 (93.8) 531 (94.5) 0.7 (−2.0 to 3.5) 0.747 368 (98.7) 372 (99.7) 1.1 (0 to 2.5) 0.219
Discharge opioid type 0.334 0.825
   Oxycodone 5 mg 357 (81.1) 445 (83.8) 305 (82.9) 314 (84.4)
   Hydrocodone 5 mg 39 (8.9) 47 (8.9) 33 (9.0) 29 (7.8)
  Other 39 (7.3) 44 (10.0) 30 (8.2) 29 (7.8)

J Neurosurg Spine Volume 35 • September 2021

MEDs prescribed at discharge, median [IQR] 450 [375–675] 315 [210–450] −126 (−156 to −96) <0.001 450 [375–675] 338 [225–450] −126 (−156 to −96) <0.001
Received opioid refill 113 (24.1) 155 (27.6) 3.6 (−1.6 to 9.1) 0.23 93 (24.9) 102 (27.3) 2.5 (−4 to 9) 0.505
Refill opioid type 0.258 0.578
   Oxycodone 5 mg 64 (56.6) 101 (65.2) 58 (62.4) 69 (67.6)
   Hydrocodone 5 mg 11 (9.7) 16 (10.3) 7 (7.5) 9 (8.8)
  Other 38 (33.6) 38 (24.5) 28 (30.1) 24 (23.5)
   No. of refills, median [IQR] 0.00 [0.00–0.00] 0.00 [0.00–1.00] 0.1 (0–0.3) 0.172 0.00 [0.00–0.00] 0.00 [0.00–1.00] 0.2 (0–0.3) 0.322
MEDs of each refill prescriptions, median [IQR] 450 [300–475] 210 [158–315] −162 (−236 to −108) <0.001 450 [300–450] 210 [158–315] −175 (−268 to −113) <0.001
Adjunct analgesic medications
Received skeletal muscle relaxant (%) 199 (42.4) 224 (39.9) −2.3 (−7.9 to 4.0) 0.44 160 (42.9) 162 (43.4) 0.6 (−6.7 to 7.3) 0.941
Received gabapentinoid 81 (17.3) 109 (19.4) 2.1 (−2.7 to 7.2) 0.426 67 (18.0) 68 (18.2) 0.4 (−5.3 to 5.6) >0.99
PROs
Timing of 1st postop visit in days, median [IQR] 42 [19–48] 41.00 [19–48] 0 (−2.5 to 2.6) 0.711 42.00 [18–48] 42 [22–48] 0.7 (−2.3 to 3.5) 0.925
Had PRO follow-up 354 (75.5) 438 (77.9) 2.5 (−2.9 to 7.7) 0.392 282 (75.6) 290 (77.7) 2.4 (−3.6 to 8.4) 0.544
Change in PROMIS-GH physical health subscore, 5.39 (7.46) 4.58 (7.28) −0.8 (−2.1 to 0.4) 0.212 5.21 (7.66) 5.02 (6.78) −0.2 (−1.6 to 1.3) 0.79
mean (SD)
Change in PROMIS-GH mental health subscore, 2.50 (6.77) 1.32 (7.18) −1.2 (−2.5 to 0) 0.055 2.11 (6.77) 1.90 (7.19) −0.2 (−1.6 to 1.2) 0.769
mean (SD)
NPRS score at 1st postop visit, median [IQR] 3.0 [0–5.0] 2.5 [0–5.0] 0 (−0.5 to 0.4) 0.765 2.0 [0–5.0] 2.0 [0–5.0] −0.1 (−0.6 to 0.3) 0.531
Other patient outcomes
Emergency department visits w/in 90 days 31 (6.6) 31 (5.5) −1.1 (−4.1 to 2.2) 0.546 24 (6.4) 18 (4.8) −1.4 (−4.5 to 1.8) 0.427
Readmission w/in 90 days 34 (7.2) 31 (5.5) −0.2 (−2.7 to 2.6) 0.312 28 (7.5) 18 (4.8) −1.6 (−4.3 to 1.1) 0.171
* Difference columns represent bootstrapped difference in means when the row corresponds to a continuous variable and bootstrapped difference in percentages when the row corresponds to a binary variable.

Unauthenticated | Downloaded 05/11/24 02:12 PM UTC

 Winkelman et al.

from 15% to 63% in overall opioids prescribed following a ing the reform in this study likely represents a conservative
variety of different surgeries.40–43 While PDMPs may en- estimate of the true decrease in opioid prescribing for elec-
able the individual provider to change how they prescribe tive lumbar decompression patients following the opioid
opioids, state-level limits on opioid prescriptions may be reform. The sensitivity analysis of the propensity-matched
more effective for catalyzing a collective effort to reduce analysis also suggests that any unidentified confounding
overall opioid prescribing. variable would need to be at least 60% more prevalent in
Critics of state-level opioid prescribing reforms wor- either cohort and strongly associated with postoperative
ried that these laws would increase provider workload and opioid use in order to explain the 118 MED average de-
negatively impact patient outcomes after elective surgery. crease following the prescribing reform. Second, the pres-
The present study found that the reduction in postoper- ent analysis only assesses changes in postoperative opi-
ative opioids prescribed following the 2017 Ohio reform oid prescribing following elective lumbar decompression
occurred without a significant change in the prescribing surgery within a single health system, and the impact of
frequency of analgesic medications (opioid and nonopioid state-level reforms on postoperative opioid prescribing for
medications), PROs, or postoperative emergency depart- other providers and/or surgical procedures may be highly
ment visits/readmissions. Similarly, Reid et al. did not variable. Third, the use of EMR data for opioid prescrip-
observe a significant increase in postoperative emergency tions written by the surgical team may not accurately rep-
department visits/readmissions or a clinically meaningful resent all opioid prescriptions that were filled or ultimately
increase in prescription refills for lumbar spine surgery taken by the patients, since patients may not fully utilize
patients after the 2017 Rhode Island prescribing reform.34 their prescriptions from surgical providers or they may ob-
Additionally, Vu et al. found that the adoption of statewide tain opioid prescriptions from providers other than their
postoperative opioid prescribing guidelines was not asso- surgeon (e.g., primary care physicians). However, given
ciated with a significant change in patient satisfaction or that the vast majority of opioid prescriptions are printed
pain scores after general surgery, vascular, and gyneco- after an EMR workflow and that surgeons are the primary
logical procedures.36 These findings suggest that the state- managers of postoperative pain, we believe that the opioid
level prescribing reform is not excessively burdensome to order data of the surgical team are an appropriate metric
providers and patients after elective lumbar decompression for assessing changes in postoperative prescribing. Further
surgery. Additionally, the lack of a compensatory change work is needed to see if decreased postoperative opioid
in the frequency of refill prescriptions and the lack of a prescribing from the surgical team has led to increases
statistically significant change in PROs suggest that the in postoperative opioid prescribing by other providers.
reduction in postoperative opioids may represent opioids Fourth, although no significant change in PRO metrics
that were previously prescribed in excess. was detected after the implementation of the state-level
State-level prescribing reforms may help to mitigate iat- prescribing reform, this finding does not necessarily prove
rogenic opioid-related harm in two important ways. First, that there was no change in patient outcomes as a result of
smaller postoperative opioid prescriptions may reduce the state-level prescribing reform. Different PRO metrics
the substantial pool of opioids available for diversion to from those examined here or PROs collected earlier on
nonmedical use or abuse, as research has shown that up in the postoperative course (i.e., closer to date of surgery/
to two-thirds of postoperative opioid prescriptions may hospital discharge) may be more sensitive for detecting
have some unused portion.44 Second, prior studies found a changes in patient experience following changes in overall
positive correlation between the size of initial opioid pre- postoperative opioid prescribing. Lastly, the incomplete
scriptions and the risk of chronic opioid use.45 Therefore, PRO follow-up may represent potential selection bias as
decreasing postoperative opioid prescriptions may help the PROs of nonresponders may be substantially different
to decrease patient exposure to opioids and lower rates from those who responded. Therefore, continued investi-
of subsequent chronic postoperative opioid use. Further gation assessing the impact of state-level opioid prescrib-
research is needed to determine if reduced postoperative ing reforms on patient experience is warranted.
opioid prescribing following state-level prescribing limits
is successful in mitigating these two important factors that
may contribute to opioid-related harm. Conclusions
We acknowledge several limitations in attempting to Patients who underwent lumbar decompression surgery
quantify the impact of a state-level prescribing reform within a single large healthcare system were prescribed
using retrospective data. First, the observed decrease in significantly less postoperative opioid medication follow-
postoperative opioid prescribing after the reform may be ing the introduction of a state-level prescribing reform. No
due to preexisting trends toward reduced opioid prescrib- significant changes in PROs, the incidence of postopera-
ing or unidentified confounding variables. However, the tive emergency department visits or readmissions, or the
decrease in opioid prescribing, as shown in Fig. 1, appears prescribing frequency of adjunct analgesic medications
temporally related to the timing of the announcement and were observed after the reform. Although further research
implementation of the Ohio opioid reform. Furthermore, is needed to determine if state-level prescribing reforms
given that reductions in opioid prescribing seem to have are effective at reducing opioid-related harm, these results
occurred in anticipation of the reform’s implementation suggest that state-level prescribing limits are effective at
date (i.e., decreases in opioid prescribing were also seen in reducing postoperative opioid prescriptions without nega-
the months leading up to the August 2017 implementation tively impacting PROs after elective lumbar decompres-
date), the estimated decrease in opioid prescribing follow- sion surgery.
J Neurosurg Spine Volume 35 • September 2021 281

Unauthenticated | Downloaded 05/11/24 02:12 PM UTC

Winkelman et al.

Acknowledgments use and its association with perioperative opioid demand and
postoperative opioid independence in patients undergoing
We thank John Fredieu for his assistance with proofreading spine surgery. Spine (Phila Pa 1976). 2014;​39(25):​E1524–
the manuscript. E1530.
20. Pugely AJ, Bedard NA, Kalakoti P, et al. Opioid use follow-
References ing cervical spine surgery:​trends and factors associated with
long-term use. Spine J. 2018;​18(11):​1974–1981.
1. Drug overdose deaths. Centers for Disease Control and Pre- 21. Borchardt J. New Ohio rules limit some opioid prescriptions
vention. Accessed January 21, 2021. https:​//www.cdc.gov/ to 7-day supplies. Cleveland.com. March 30, 2017. Accessed
drugoverdose/data/statedeaths.html January 21, 2021. https:​//www.cleveland.com/metro/2017/03/
2. Prescription opioid data. Centers for Disease Control and pain_prescription_limits_set.html
Prevention. Accessed January 21, 2021. https:​//www.cdc.gov/ 22. For prescribers - New limits on prescription opioids for acute
drugoverdose/data/prescribing.html pain. State Medical Board of Ohio. August 18, 2017. Ac-
3. Dowell D, Haegerich TM, Chou R. CDC guideline for pre- cessed January 21, 2021. http:​//med.ohio.gov/Publications/
scribing opioids for chronic pain — United States, 2016. RecentNews/TabId/246/ArticleId/50/new-limits-on-
MMWR Recomm Rep. 2016;​65(No RR-1):​1–49.
prescription-opioids-for-acute-pain.aspx
4. Lowenstein M, Grande D, Delgado MK. Opioid prescribing
23. Rule 4731-11-13:​Prescribing of opioid analgesics for acute
limits for acute pain - striking the right balance. N Engl J
pain. State Medical Board of Ohio. Accessed January 21,
Med. 2018;​379(6):​504–506.
2021. https:​//med.ohio.gov/Portals/0/DNN/PDF-FOLDERS/
5. Bulloch M. Opioid prescribing limits across the states.
Laws-Rules/Newly-Adopted-Rules/4731-11-13%2C eff 8-31-
Pharmacy Times. February 5, 2021. Accessed January 21,
17.pdf
2021. https:​//www.pharmacytimes.com/contributor/marilyn-
bulloch-pharmd-bcps/2019/02/opioid-prescribing-limits- 24. McPherson ML. Demystifying Opioid Conversion Calcula-
across-the-states tions:​A Guide for Effective Dosing. ASHP;​2009.
6. A prescriber’s guide to the new Medicare Part D opioid 25. A brief guide to the PROMIS Global Health Instruments.
overutilization policies for 2019. Centers for Medicare and HealthMeasures. Accessed January 21, 2021. http:​//www.
Medicaid Services. November 1, 2018. Accessed January healthmeasures.net/images/PROMIS/manuals/PROMIS_
21, 2021. https:​//www.cms.gov/Outreach-and-Education/ Global_Scoring_Manual.pdf
Medicare-Learning-Network-MLN/MLNMattersArticles/ 26. van der Loo M. simputation:​Simple imputation. Ac-
downloads/SE18016.pdf cessed January 21, 2021. https:​//cran.r-project.org/
7. Clarke H, Soneji N, Ko DT, et al. Rates and risk factors for package=simputation
prolonged opioid use after major surgery:​population based 27. Ahmed A, Husain A, Love TE, et al. Heart failure, chronic
cohort study. BMJ. 2014;​348(February):​g1251. diuretic use, and increase in mortality and hospitalization:​
8. Hill MV, McMahon ML, Stucke RS, Barth RJ Jr. Wide varia- an observational study using propensity score methods. Eur
tion and excessive dosage of opioid prescriptions for common Heart J. 2006;​27(12):​1431–1439.
general surgical procedures. Ann Surg. 2017;​265(4):​709–714. 28. D’Agostino RB Jr. Propensity score methods for bias reduc-
9. Scully RE, Schoenfeld AJ, Jiang W, et al. Defining optimal tion in the comparison of a treatment to a non-randomized
length of opioid pain medication prescription after common control group. Stat Med. 1998;​17(19):​2265–2281.
surgical procedures. JAMA Surg. 2018;​153(1):​37–43. 29. Rosenbaum PR. Observation and Experiment:​An Introduc-
10. Thiels CA, Anderson SS, Ubl DS, et al. Wide variation and tion to Causal Inference. Harvard University Press;​2017.
overprescription of opioids after elective surgery. Ann Surg. 30. Efron B, Tibshirani R. Bootstrap methods for standard errors,
2017;​266(4):​564–573. confidence intervals, and other measures of statistical accu-
11. Nobel TB, Zaveri S, Khetan P, Divino CM. Temporal trends racy. Stat Sci. 1986;​1(1):​54–75.
in opioid prescribing for common general surgical proce- 31. R:​A language and environment for statistical computing.
dures in the opioid crisis era. Am J Surg. 2019;​217(4):​613– R Foundation;​2020. Accessed January 21, 2021. https:​//
617. www.r-project.org/
12. Eid AI, DePesa C, Nordestgaard AT, et al. Variation of opi- 32. RStudio:​Integrated Development for R. RStudio, Inc;​2020.
oid prescribing patterns among patients undergoing similar Accessed January 21, 2021. http:​//www.rstudio.com/
surgery on the same acute care surgery service of the same 33. Zolin SJ, Ho VP, Young BT, et al. Opioid prescribing in mini-
institution:​time for standardization? Surgery. 2018;​164(5):​ mally injured trauma patients:​effect of a state prescribing
926–930. limit. Surgery. 2019;​166(4):​593–600.
13. McDonald DC, Carlson K, Izrael D. Geographic variation in 34. Reid DBC, Shah KN, Ruddell JH, et al. Effect of narcotic
opioid prescribing in the U.S. J Pain. 2012;​13(10):​988–996. prescription limiting legislation on opioid utilization follow-
14. Nooromid MJ, Blay E Jr, Holl JL, et al. Discharge prescrip- ing lumbar spine surgery. Spine J. 2019;​19(4):​717–725.
tion patterns of opioid and nonopioid analgesics after com- 35. Porter SB, Glasgow AE, Yao X, Habermann EB. Association
mon surgical procedures. Pain Rep. 2018;​3(1):​e637. of Florida House Bill 21 with postoperative opioid prescrib-
15. Makary MA, Overton HN, Wang P. Overprescribing is major ing for acute pain at a single institution. JAMA Surg. 2020;​
contributor to opioid crisis. BMJ. 2017;​359:​j4792. 155(3):​263–264.
16. Wunsch H, Wijeysundera DN, Passarella MA, Neuman MD. 36. Vu JV, Howard RA, Gunaseelan V, et al. Statewide imple-
Opioids prescribed after low-risk surgical procedures in the mentation of postoperative opioid prescribing guidelines. N
United States, 2004-2012. JAMA. 2016;​315(15):​1654–1657. Engl J Med. 2019;​381(7):​680–682.
17. Dunn LK, Yerra S, Fang S, et al. Incidence and risk fac- 37. Lin HC, Wang Z, Boyd C, et al. Associations between state-
tors for chronic postoperative opioid use after major spine wide prescription drug monitoring program (PDMP) require-
surgery:​a cross-sectional study with longitudinal outcome. ment and physician patterns of prescribing opioid analgesics
Anesth Analg. 2018;​127(1):​247–254. for patients with non-cancer chronic pain. Addict Behav.
18. Schoenfeld AJ, Nwosu K, Jiang W, et al. Risk factors for 2018;​76(76):​348–354.
prolonged opioid use following spine surgery, and the asso- 38. Patrick SW, Fry CE, Jones TF, Buntin MB. Implementation
ciation with surgical intensity, among opioid-naive patients. J of prescription drug monitoring programs associated with re-
Bone Joint Surg Am. 2017;​99(15):​1247–1252. ductions in opioid-related death rates. Health Aff (Millwood).
19. Armaghani SJ, Lee DS, Bible JE, et al. Preoperative opioid 2016;​35(7):​1324–1332.

282 J Neurosurg Spine Volume 35 • September 2021

Unauthenticated | Downloaded 05/11/24 02:12 PM UTC

 Winkelman et al.

39. Stucke RS, Kelly JL, Mathis KA, et al. Association of the use Author Contributions
of a mandatory prescription drug monitoring program with Conception and design: Winkelman, Pelle, Benzel, Mroz, Stein-
prescribing practices for patients undergoing elective surgery. metz. Acquisition of data: Winkelman. Analysis and interpreta-
JAMA Surg. 2018;​153(12):​1105–1110. tion of data: Winkelman, Kavanagh, Tanenbaum, Benzel. Drafting
40. Hill MV, Stucke RS, McMahon ML, et al. An educational the article: Winkelman, Kavanagh, Pelle. Critically revising the
intervention decreases opioid prescribing after general surgi- article: all authors. Reviewed submitted version of manuscript: all
cal operations. Ann Surg. 2018;​267(3):​468–472. authors. Approved the final version of the manuscript on behalf of
41. Stanek JJ, Renslow MA, Kalliainen LK. The effect of an all authors: Winkelman. Statistical analysis: Winkelman.
educational program on opioid prescription patterns in hand
surgery:​a quality improvement program. J Hand Surg Am. Supplemental Information
2015;​40(2):​341–346.
42. Chiu AS, Jean RA, Hoag JR, et al. Association of lowering Previous Presentations
default pill counts in electronic medical record systems with Portions of this paper were previously presented in oral form at
postoperative opioid prescribing. JAMA Surg. 2018;​153(11):​ the 2019 AANS Annual Scientific Meeting, San Diego, Califor-
1012–1019. nia, April 13–17, 2019; Lumbar Spine Research Society Annual
43. Delgado MK, Shofer FS, Patel MS, et al. Association be- Meeting, Chicago, Illinois, April 4–5, 2019; and Midwest Spine
tween electronic medical record implementation of default Symposium 2019, Pittsburgh, Pennsylvania, September 6–7, 2019.
opioid prescription quantities and prescribing behavior in
two emergency departments. J Gen Intern Med. 2018;​33(4):​ Correspondence
409–411. Robert D. Winkelman: Cleveland Clinic Foundation, Cleveland,
44. Bicket MC, Long JJ, Pronovost PJ, et al. Prescription opioid OH. winkelr@ccf.org.
analgesics commonly unused after surgery:​a systematic re-
view. JAMA Surg. 2017;​152(11):​1066–1071.
45. Shah A, Hayes CJMB, Martin BC. Characteristics of initial
prescription episodes and likelihood of long-term opioid
use - United States, 2006-2015. MMWR Morb Mortal Wkly
Rep. 2017;​66(10):​265–269.

Disclosures
Dr. Mroz: royalties from Stryker. Dr. Steinmetz: royalties from
Zimmer Biomet and Elsevier; consultant for Globus; and hono-
raria from Stryker and Globus.

J Neurosurg Spine Volume 35 • September 2021 283

Unauthenticated | Downloaded 05/11/24 02:12 PM UTC