ABSTRACT
Background: Opioids are commonly used for postoperative pain management in spine surgery. However, few guidelines exist for appropriate prescribing in the acute postoperative phase of care. We identify risk factors for inpatient (IP) opioid use and examine relationships between IP requirements and discharge (DC) opioid prescriptions.
Methods: Retrospective review of elective spine surgeries between January 2014 and May 2018 identified cases of lumbar decompression (LD), LD with fusion (LDF), and cervical decompression with fusion (CDF) at our high-volume spine center. Multiple regression examining potential risk factors for opioid use was performed. Opioid use was normalized into daily morphine milligram equivalents (MME).
Results: A total of 2281 patients who underwent 1251 LD, 384 LDF, and 648 CDF procedures were identified (54.1% male, mean age = 57.9 years, mean body mass index = 30.3 kg/m2, median American Society of Anesthesiologists [ASA] score = 2). Mean IP opioid use was 44.4 MME/day and average DC prescriptions totaled 496.5 MME. Multiple regression models identified younger age and increased ASA score as predictive of increased daily IP opioid consumption (βAGE = −0.36, P < .001, βASA = 10.1, P < .001; R2 = 0.308) and increased DC opioid amounts (βAGE = −4.62, P < .001, βASA = 72.1, P < .001; R2 = 0.097). Highest IP and DC opioid use was observed among LDF followed by CDF and LD patients. Significant positive correlations were found between IP opioid usage and DC opioid prescriptions by IP opioid quartiles (r = 0.99 LD, 0.98 LDF, 0.96 CDF).
Conclusions: Younger patients and higher ASA scores correlated with increased IP opioid use and DC opioid prescriptions. DC prescriptions appropriately reflect IP use.
Level of Evidence: 3.
Clinical Relevance: Adequate pain management is an integral component to successful outcomes in spine surgery. Awareness of candidates likely to require higher levels of opioid analgesia will be beneficial in guiding surgeon prescribing practices.
INTRODUCTION
The use of opioid analgesia has increased following the introduction of the Pain as the Fifth Vital Sign campaign of the mid-1990s.1,2 Concurrently, international collaborations such as those published in the Lancet as part of the 2013 Global Burden of Disease Study have identified chronic back and neck pain as a leading cause of disability-adjusted life-years lost worldwide.3 Patients suffering from such complex pain syndromes often experience impairment in function and quality of life, usually requiring multimodal treatment options which may include opioid medication management. Formerly reserved for intractable pain control among the terminally ill, opioid medications have since become a common tool in managing painful spine-related conditions including postoperative pain.
In parallel, the volume of spine surgeries has grown substantially in recent years and surgical treatment has been shown to have favorable outcomes across a wide range of spinal pathologies.4,5 However, as the frequency of spine surgeries continue to increase, it is important to understand the role of pharmacological pain management in the restoration of normal function to these patients.6 This is especially true in light of the current opioid epidemic as policy makers act to throttle prescriptions while patient perceptions of pain control have become incorporated into satisfaction scores used as a surrogate measure of hospital quality.7
Strategies for postoperative pain management have evolved over time. Previous attempts relied on patient-controlled intravenous analgesia (PCA), but anecdotal evidence and published reports suggest that the use of PCA increases hospital length of stay.8 More recent approaches include a variety of multimodal oral analgesic agents.9,10 When ordering postoperative medications, surgeons and care teams consider several factors including patient age, medical comorbidities, surgical invasiveness, and prior medication use. However, there is no agreed upon standard for managing pain following specific spine-related procedures. Furthermore, little is published regarding risk factors for inpatient (IP) opioid use or prescription pain control needs at discharge (DC) for these patients. The goal of our study was to identify risk factors for IP opioid use following 3 common spine surgeries and examine the relationship between IP opioid use and DC prescriptions.
MATERIALS AND METHODS
Data Collection
This study was deemed institutional review board exempt by the institutional clinical research committee. This investigation used internal data analyzed retrospectively from spine surgery cases performed at a single institution with a dedicated spine surgery IP unit.
Data of Interest
Patient demographics included age, sex, and body mass index (BMI). Indices of patient comorbidities included the American Society of Anesthesiologists (ASA) score and the Centers for Medicare and Medicaid Services Hierarchical Conditions Category (CMS-HCC) score, previously used to stratify financial risk in Medicare Advantage patients.11 Opioid data included postoperative IP opioid use and amount of opioid prescribed at DC based on available electronic medical record data and normalized into morphine milligram equivalents (MMEs).
Inclusion and Exclusion Criteria
Our database included information on over 5000 spinal surgery cases based on an internal query of patients undergoing spine surgery between January 2014 and May 2018. Patients were categorized by procedure type and included lumbar decompression (LD), LD and fusion (LDF), or cervical discectomy with fusion (CDF). The database includes both primary and revision surgical procedures.
Opioid Use
IP opioid consumption was calculated from all narcotic orders placed during the duration of individual patients' length of stay (LOS) and subsequently normalized into daily usage as a function of recorded LOS. Opioid medications prescribed at time of DC were quantified as those prescriptions written at DC for patients undergoing the identified spinal surgery procedure. IP opioid use and opioids prescribed at DC were normalized into MMEs in accordance with the Centers for Disease Control and Prevention's established guidelines (Table 1).12 All opioid use data were stratified into quartiles by MME for subsequent analysis.
Conversion chart of morphine milligram equivalents (MMEs). Conversions are those presented by the Centers for Disease Control and Prevention.12
CMS-HCC Score Calculation
The CMS-HCC score was used in this study as a possible predictor of opioid use. The publicly available scoring methodology, originally designed to predict financial risk in Medicare Advantage patients, was replicated and used to generate an objective risk score for our series of patients. In brief, data are scored using health plan enrollee diagnoses and demographic data in order to predict medical expenditure risk, from which a risk adjustment model is able to identify high- and low-cost individuals based on established hierarchical categories.11 Our previous work and forthcoming studies suggest that this score may have clinical use for care teams.
STATISTICAL METHODS
Statistical Analysis
All statistical analyses were performed using MathWorks MATLAB version R2017b. Patient cohorts were compared using 1-way analysis of variance (ANOVA) and Pearson χ2 analysis to establish significant differences between groups. Inherent difference between the 3 procedure groups of study are controlled through including a procedure variable in further statistical analyses.
Multiple linear regression analysis was performed to examine predictors of both IP daily opioid consumption and amount of opioid medication prescribed at DC. Predictor variables including age, BMI, ASA score, and CMS-HCC score were assessed for relationship with both outcome variables. IP daily opioid consumption was included in the multiple linear regression model as a potential predictor of DC opioid amounts.
Quartile Analysis
Patients were stratified into quartiles based on IP daily opioid consumption and opioid medication amounts prescribed at DC. One-way ANOVA was performed on IP opioid consumption quartiles against DC opioid prescriptions controlling for procedure type. Correlation analysis was also performed between opioids dispensed during hospitalization and those prescribed at time of DC.
RESULTS
Patient Demographics
We identified 2281 patients during a 53-month interval between January 2014 and May 2018 who underwent elective LD, LDF, or CDF at our center (Table 2). Patients exhibited similar BMI values across all procedures with an average of 30.3 kg/m2. Significant differences were found across cohorts by age, sex, ASA score, and CMS-HCC score (P < .001). Patients undergoing LDF tended to be older (LDF = 61.2 years, LD = 57.9 years, CDF = 56.0 years) and were determined to have higher ASA and CMS-HCC scores (LDF = 2.5/0.93, LD = 2.3/0.77, CDF = 2.4/0.80) compared with the other surgical cohorts. A larger proportion of patients undergoing LD at our center were male (LD = 54%, LDF = 49%, CDF = 48%).
Summary of descriptive statistics for patients undergoing elective spine surgery at our center. Data reported as mean ± standard error.
Opioid Use
LD patients consumed lower average daily amounts of opioid medications during IP hospitalization (LD = 19.7 MME, LDF = 105.3 MME, CDF = 56.0 MME, P < .001) and average total amounts at time of DC (LD = 406 MME, LDF = 595 MME, CDF = 611 MME, P < .001) as compared with the LDF or CDF cohorts (Table 2). Opioid use varied across procedures for both IP daily opioid consumption and DC prescriptions (Figure 1). Among the IP cohorts, average daily opioid consumption ranged from 0 to 754.6 MME. By procedure, LDIP demonstrated median (interquartile range) of 10 MME (5–23 MME) compared with LDFIP = 85.7 MME (60–127.4 MME) and CDFIP = 44.2 MME (23.9–73.7 MME; Table 3). At DC, prescription totals included amounts from 0 to 5400 MME, with LDDC = 300 MME (225–450 MME), LDFDC = 450 MME (375–675 MME), and CDFDC = 480 MME (337.5–750 MME). From these data, observable trends illustrated strong linearity, with coefficients of correlation for relatedness between daily IP opioid amounts and prescriptions at time of DC greater than 90% across all quartiles and procedures. (LD R = 0.93, LDF R = 0.96, CDF R = 0.96). Statistically significant relatedness was demonstrated between daily IP opioid consumption and outpatient DC prescriptions for LDF and CDF (LDF P = .043, CDF P = .003) but not for LD (P = .068).
Discharge opioid prescriptions for lumbar decompression, lumbar decompression with fusion, and cervical decompression with fusion stratified by inpatient daily opioid consumption (quartiles). MME, morphine milligram equivalent.
Quartile analysis of opioids dispensed at inpatient and discharge phases of care. Data are reported in average opioid amount (MME) per quartile with accompanying low and high values.
Statistical Modeling
Multiple linear regression analysis was performed to assess demographic and preoperative variables to postoperative opioid use in both the IP and outpatient setting while controlling for procedure type (Table 4). Age and ASA score were found to be statistically predictive of opioid use during hospitalization. Increased IP daily opioid use was predicted in younger patients, with a decrease of 0.36 MME per additional year of age (β = −0.36, 95% confidence interval [CI] = −0.52 to −0.19, P < .001) and an increase of 10.1 MME per point increase in ASA score (β = 10.1, 95% CI = 5.95 to 14.17, P < .001, R2 = 0.308). With regard to amounts of opioid medications prescribed at time of DC, modeling demonstrated age, ASA score, and IP daily opioid consumption to be significant predictors of DC opioid amounts. Increases in amount of prescribed DC opioid medication were associated with younger patients at a rate of 3.88 MME per year of age (β = −3.88, 95% CI = −5.04 to −2.72, P < .001), and increase of 51.11 MME per 1 point increase in ASA score (β = 51.11, 95% CI = 21.62 to 80.60, P ≤ .001). Relatedness between daily amounts of IP opioid medications while hospitalized and DC prescriptions demonstrated a 2:1 relationship, with patients receiving approximately twice the amount of opioid medication at time of DC for each additional MME received during an average day of hospitalization (β = 2.08, 95% CI = 1.79 to 2.38, P < .001, R2 = 0.168).
Multiple regressions for patient demographic data controlling for spine surgery by procedure. Inpatient opioid consumption reported in MME/d (R2 = 0.308). Prescribed opioid amounts at discharge reported in MME (R2 = 0.168).
DISCUSSION
Previous literature has investigated the relationship between baseline opioid use and sustained long-term opioid usage. Work from the Spine Patient Outcomes Research Trial (SPORT) cohort of 2,110 patients demonstrated significant risk factors associated with baseline opioid use including young age, lack of college education, unemployment, disability status, smoking history, and concurrent mental disorders.13 The relationship between prior opioid usage and long-term continuation following spinal surgery has also been reported.14–17 This is in agreement with the well-studied potentiation of analgesia through μ-opioid receptor agonism which has been clearly and repeatedly demonstrated to possess substantial addictive potential.18,19
Preoperative chronic opioid use itself presents substantial risks to postoperative spine surgery outcomes. Specifically, heavy opioid users are at significant risk for increased 90-day wound complications and pain disorders, culminating in increased returns to the emergency department and readmission to the hospital as well as continued postoperative use.20 Similar trends have been observed in instances of worker's compensation cases, where opioid use prior to cervical fusion was correlated to inferior return to work, reoperation rate, and permanent disability outcomes.21 Survey instruments designed to identify chronic baseline opioid use prior to surgery have been effective at predicting suboptimal patient reported outcomes and may play a role in helping to augment how surgeons plan postoperative follow up or counseling.22 Identifying these risks prior to surgical treatment may allow treatment teams to allocate resources designed to optimize patient outcomes in the acute perioperative period.
Our findings suggest that ASA score and age are associated with anticipated increases in opioid medication both during the hospital encounter and at time of DC. Increasing ASA score was linked with higher IP and DC opioids, while increasing age is inversely related to their opioid use. This observation may be explained by age-related differences in pain perception or caution in prescribing higher doses of opioids in older patients. However, independent of underlying comorbidities, there remains risk that younger patients will consume greater amounts of opioid medications throughout their postoperative course. These risk factors may potentially serve as useful adjunctive markers to known related comorbidities linked to long-term opioid usage following spine surgery such as migraine disorders, diabetes, chronic depression, and smoking.23
Our data also serve as a first step in developing standardized, appropriate opioid prescriptions at time of DC from specific and common spine procedures. DC prescriptions appear to reflect actual IP opioid use rather than arbitrary and often large amounts of opioids. Based on our findings, we suggest that IP opioid use may be an accurate means of assessing potential outpatient needs during the immediate postoperative period.
Our study was limited to data in our electronic medical records. While this accurately depicts IP use and DC prescriptions, capturing additional refills or actual outpatient consumption was beyond the scope of this study, although additional work is currently underway in order to track individual patient use and refills beyond DC.
CONCLUSIONS
Our analysis demonstrates younger patients and higher ASA scores are associated with greater IP and DC opioid use. In addition, DC prescriptions reflect IP use. Further study is underway to track postoperative prescription refills and use beyond the initial DC prescription. Ultimately, this will help surgeons develop refined, procedure-specific prescribing protocols.
Footnotes
Disclosures and COI: The authors declare that there are no relevant conflicts of interest to disclose.
- ©International Society for the Advancement of Spine Surgery
- This manuscript is generously published free of charge by ISASS, the International Society for the Advancement of Spine Surgery. Copyright © 2020 ISASS.
REFERENCES
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.