- Question: What is the current national utilization of intravenous acetaminophen and is it associated with reductions in inpatient opioid prescription and opioid-related side effects in lumbar/lumbosacral spinal fusion, a procedure with relatively high opioid utilization?
- Findings: While used in 18.9% (n = 22,208) of patients, intravenous acetaminophen was associated with minimal changes in opioid prescription and opioid-related side effects.
- Meaning: As we did not show any meaningful effects of intravenous acetaminophen, its role in postoperative pain management in these patients remains to be determined.
Multimodal analgesia, the combination of different analgesic drug classes and procedures, for the treatment of postoperative pain has been promoted in recent years to improve pain control while reducing opioid consumption and subsequently opioid-related side effects.1,2 Patients undergoing spinal surgery suffer from significant procedure-related pain that is often difficult to control with opioids, making a multimodal pain management approach particularly attractive in this population. Moreover, neuraxial anesthesia is not a common choice in these procedures. While several nonopioid analgesics, including oral acetaminophen, have been used for decades on a routine basis, intravenous acetaminophen (ivAPAP) is relatively new to the US market, and its clinical efficacy and role have yet to be established.1,3 Thus far, studies on the topic have produced inconclusive results regarding the effects of ivAPAP on both opioid consumption and clinical outcomes.4–8 While some studies claim that ivAPAP reduces opioid needs and opioid-related side effects,9 others could not find any significant impact on opioid requirement.4,6 Specifically, there is a paucity of studies evaluating the use and impact of ivAPAP in everyday clinical settings for patients undergoing spinal surgeries. We therefore analyzed data from a large nationwide administrative dataset and aimed to elucidate (1) current utilization of ivAPAP, (2) whether perioperative ivAPAP reduces inpatient opioid prescription, and (3) if this coincides with decreased odds for clinically relevant opioid-related adverse outcomes.
Data Source and Study Design
This study was approved by the Mount Sinai Hospital and Hospital for Special Surgery Institutional Review Board and was exempt from the requirement to obtain informed consent from subjects, given the retrospective and observational nature of this analysis (project #14-00647, HSS#2012-050-CR2). The data for this retrospective cohort study were derived from the Premier Healthcare Database (Premier Healthcare Solutions Inc, Charlotte, NC), a large nationwide administrative claims database with data from approximately 20% of the nation’s hospitalizations.10,11 It contains information including International Classification of Diseases-9th revision Clinical Modification (ICD-9 CM) codes, Current Procedural Terminology codes, and billed items.
The study cohort consisted of patients who received a lumbar/lumbosacral fusion of the spine using a posterior technique on the posterior column or anterior column (ICD-9 CM codes 81.07 and 81.08, respectively) between 2011 and 2014. Patients were excluded for the following reasons: unknown gender (n = 11), unknown discharge status (n = 564), classified as nonelective procedure (n = 24,540) or outpatient procedure (n = 4112), hospitals with <30 spinal fusion procedures (to ensure sufficient sample size per cluster; n = 882), no inpatient billing for opioids (n = 6058), or opioid prescription >95th percentile (n = 6172). The latter 2 groups were excluded to avoid bias introduced by unreliable opioid claims or extremes in opioid prescription.
The main effect of interest was the use of ivAPAP, which was categorized into 0, 1 (1000 mg), or >1 dose given on the day of surgery (postoperative day [POD] 0), the day after surgery (POD 1), or any subsequent day (POD 1+); this represents 3 separate variables. The primary outcome was opioid prescription while secondary outcomes were length of hospital stay (LOS), cost of hospitalization (not charges), naloxone use, and respiratory, gastrointestinal, genitourinary, central nervous system, or “other” complications.12 Opioid prescription was extracted from billing data and was converted into oral morphine equivalents using Lexicomp “opioid agonist conversion”13 and GlobalRPH “opioid analgesic converter.”14 In general, hospitals participating in Premier submit their actual cost data. A smaller number of hospitals submit charges that are then converted into costs using Medicare cost to charge ratios.10 As previously suggested,15 naloxone use was used as a marker for potential opioid-related complications. The rationale behind the primary outcome relates to the intended primary effect of ivAPAP (and nonopioid analgesics in general): the management of mild to moderate pain or moderate to severe pain with adjunctive opioid analgesics. The secondary outcomes were selected as they represent potential secondary effects resulting from a reduction in opioid consumption.
Patient demographics included age, gender, and race (Caucasian, African American, Hispanic, other). Health care–related variables included primary insurance type (commercial, Medicaid, Medicare, uninsured, other), hospital location (urban, rural), hospital size (<300, 300–499, ≥500 beds), hospital teaching status, and annual number of lumbar spinal fusions performed per hospital. Procedure-related variables were year of procedure, use of patient-controlled analgesia (PCA), and use of nonopioid analgesics (gabapentin/pregabalin evaluated as a single variable, nonsteroidal anti-inflammatory drugs, Cox-2 inhibitors, and ketamine) on POD 0 or POD 1/1+. Comorbidity burden was assessed using the Quan adaptation of the Charlson comorbidity index.16 In addition, we separately included relevant comorbidities that are associated with increased opioid need, including substance use/abuse, chronic pain conditions, and psychiatric conditions.
First, study variables and outcomes by use of ivAPAP (yes/no) were univariably analyzed with χ2 tests for categorical variables and t tests (or Kruskal-Wallis tests) for continuous variables. Multivariable multilevel models—measuring associations between ivAPAP use and outcomes—were fitted using all covariates found significant at the P < .15 level in the univariable analyses and/or were deemed clinically relevant. Multilevel models are necessary as they account for the correlation between patients treated at the same hospital by fitting separate regression lines per hospital.17 Based on the dose–response relationship between morphine- and opioid-related adverse effects, we hypothesized a reduction of 25% in opioid prescription (primary outcome) to be clinically significant (ie, will lead to reductions in complications).18 The continuous outcome variables (opioid prescription, LOS, and cost) are skewed, that is, not normally distributed with extreme values affecting estimates of the mean. Therefore, as previously recommended for these skewed data,19,20 they were log-transformed applying the gamma distribution with a log link function within the GLIMMIX procedure in SAS statistical software (SAS Institute, Cary, NC). Effect estimates from binary outcomes are reported as adjusted odds ratios and 95% confidence intervals (CIs), and effect estimates for continuous outcomes are reported as percent change (compared to reference groups). Statistical significance was set at P < .05.
To assess the robustness of our results, we additionally performed a propensity score analysis using the matching method. Here, we calculated propensity scores from a multilevel multivariable regression model with the outcome of ivAPAP use and the same covariates used in the primary analysis. A patient who received ivAPAP at any point during hospitalization was matched with 3 patients who did not receive ivAPAP based on the calculated propensity scores. Unfortunately, propensity score matching did not allow for a similarly detailed assessment of ivAPAP use categories (by day and dose) as in the main analysis. Therefore, a discrepancy exists in detail of presentation of results between the main analysis (ivAPAP use by day and dose) and the propensity score analysis (any ivAPAP use without this categorization). Balance in covariates between groups in the unmatched (original) and matched sample was assessed using standardized differences; unbalanced variables (standardized difference >0.121) were included in the models assessing the main study outcomes. Odds ratios (or % change) and 95% CIs were reported. To match cases to controls based on the propensity score, we used the SAS macro OneToManyMTCH with 8-digit to 1-digit match without replacement.22
All analyses were performed using SAS 9.4.
The final cohort consisted of N = 117,269 patients undergoing a lumbar/lumbosacral spinal fusion from 2011 to 2014 across 440 hospitals nationwide. Table 1 shows study variables by use of ivAPAP. Of all patients, n = 22,208 (18.9%) received at least 1 perioperative dose of ivAPAP. Of these patients, n = 16,335 (73.6%) received 1 dose, and 20.0% (n = 4438) received >1 dose on POD 0, 9.2% (n = 2044) received 1 dose, and 14.6% (n = 3235) received >1 dose on POD 1, respectively. A total of 8.4% (n = 1873) received ≥1 doses on POD 1+ (totals do not add up to 100% as there is overlap between categories). With a steep increase in ivAPAP utilization after 2011, it appears to be particularly used in large (≥500 beds) and teaching hospitals and in patients receiving other nonopioid analgesics of which, interestingly, pregabalin/gabapentin was the most commonly used.
Table 2 shows outcomes by ivAPAP use. Patients receiving ivAPAP (compared to those who did not) had slightly fewer respiratory complications (2.2% vs 2.4%; P = .0321), were prescribed higher doses of opioids (448 oral morphine equivalents, interquartile range [IQR] 270–757 vs 415 oral morphine equivalents, 240–695; P < .0001), and had a higher cost of hospitalization ($26,660, IQR $19,614–37,017 vs $25,837, IQR $18,630–35,206; P < .0001). Interestingly, while statistically significant (P = .0059), there was no clinically significant difference in median length of stay between groups (3, IQR 2–4 vs 3, IQR 2–4).
Table 3 shows results from the multivariable regression models. After adjusting for relevant covariates, use of ivAPAP on POD 0 and 1 was associated with minimal changes in opioid prescription (no pattern of reduction), length and cost of hospitalization particularly favoring >1 ivAPAP dose with a modestly −5.2% (CI, −7.2% to −3.1%; P < .0001) decreased length of stay. Interestingly, ivAPAP use did not coincide with a consistent pattern of significantly reduced odds for complications. In addition, differences between 1 and >1 dose of ivAPAP on POD 0 and 1 were minimal.
Conversely, use of ivAPAP on POD 1+ was associated with increased resource utilization and complications, likely indicating confounding by indication, that is, ivAPAP use on POD 1+ may likely happen in those with more pain and higher risk for complications.
Sensitivity Analysis and Comparison to Control Group
Results from the propensity score analysis are presented in Supplemental Digital Content, Appendix, http://links.lww.com/AA/C315. First, matching did indeed result in increased covariate balance when compared to the unmatched (original) cohort (Supplemental Digital Content, Appendix, Table A1, http://links.lww.com/AA/C315). Supplemental Digital Content, Appendix, Table A2, http://links.lww.com/AA/C315, provides information on covariate distribution between ivAPAP groups while the main results are presented in Supplemental Digital Content, Appendix, Table A3, http://links.lww.com/AA/C315, which corroborated our findings from the main analysis, that is, ivAPAP is associated with minimal changes in opioid prescription (no pattern of reduction), as well as length and cost of hospitalization. Furthermore, no decreases in odds were observed for complications while increases were seen for gastrointestinal and genitourinary complications.
We additionally compared effect estimates of ivAPAP to those from the most commonly used nonopioid analgesic in this cohort: pregabalin/gabapentin (Table 4). The aim was to assess whether studying another nonopioid analgesic would demonstrate the anticipated effect of decreased opioid prescription along with decreased odds for complications. Results in Table 4 indeed show a pattern of decreased opioid prescription, as well as length and cost of hospital stay, combined with reduced odds for complications.
In this nationwide study of ivAPAP use in patients undergoing lumbar spine surgery, we found that a majority (73.6%) of the patients who received ivAPAP only received a single dose on the day of surgery. Use of ivAPAP on POD 1+ was associated with increased resource utilization, likely indicating confounding by indication. Further, patients who received ivAPAP had minimally altered opioid prescription in comparison to those who did not receive any ivAPAP. Moreover, except for a small reduction in respiratory complications in patients who received a single dose of ivAPAP on the day of surgery, there was no uniform pattern of altered odds for perioperative complications. Interestingly, when evaluating a control nonopioid analgesic, pregabalin/gabapentin, the most commonly used nonopioid analgesic in these patients, we did demonstrate reduced opioid prescription, combined with reduced resource utilization, and reduced odds for opioid-related adverse effects. Our main results were further corroborated by a sensitivity analysis where we applied a propensity score analysis.
We identified an overall increase in the use of ivAPAP since its introduction to the US market. However, a majority of the patients in this analysis received only 1 dose on the day of surgery, suggesting that ivAPAP is used in a limited way. Indeed, some randomized controlled studies did find some beneficial effects of ivAPAP on opioid consumption and opioid-related side effects when used in a fixed and regular study regimen.8,9 It is therefore possible that patient benefit could be increased if ivAPAP is administered regularly over a certain amount of time in the perioperative period, potentially including a preoperative initial administration (even though oral acetaminophen may be a possibility as well).23,24 However, opioid consumption in spine surgery patients is relatively high as regional anesthesia may not always be possible and patients may suffer from chronic pain. Moreover, most trials on the effectiveness of ivAPAP have been conducted in other (nonspine) surgeries.9 Therefore, the potential effect of ivAPAP may be more limited in patients undergoing spine surgery compared to those undergoing, for example, laparoscopic procedures or joint replacements. Intriguingly, in this analysis, opioid prescription was higher among patients who received ivAPAP after POD 1. Any explanation for this finding remains speculative. However, it is unlikely that ivAPAP leads to higher opioid consumption per se. It is possible, for example, that ivAPAP is given when pain is severe and opioid utilization already high in an attempt to add alternative medications to alleviate discomfort.
We found that patients receiving ivAPAP had minimally altered opioid prescription without any uniform pattern of altered odds for perioperative complications. Indeed, our prespecified threshold of a 25% reduction in opioid prescription (based on the dose–response relationship between morphine and opioid-related adverse effects18), which would theoretically lead to a reduction in odds for complications, did not materialize in our results and may have been too ambitious given how sparingly ivAPAP appears to be used. However, our sensitivity analysis suggests that a lower threshold may already lead to decreased odds for complications as the use of pregabalin/gabapentin on POD 0 was associated with a 10.6% decrease in opioid prescription combined with an up to 29% reduction in odds for opioid-related adverse effects. Future trials should focus on not only the effectiveness of ivAPAP in patients undergoing lumbar spinal fusions but also the most effective dosing strategy, which may be different from other orthopedic surgeries as these patients often already suffer from preexisting chronic pain treated with conventional analgesics or opioids that may alter pain perception in these patients, thus complicating pain management.25 In addition, spinal surgeries are known to have high opioid utilization, further complicating the threshold for the minimum opioid consumption reduction and resulting in altered odds for opioid-related adverse effects.
ivAPAP has been suggested as an appropriate alternative to oral acetaminophen when oral administration is impossible or impractical, or when rapid onset is desired. However, patients undergoing spinal surgery usually could receive oral acetaminophen both pre- and postoperatively. Indeed, in a small single-institutional retrospective study on patients undergoing spinal surgery, Smith and Hoefling26 describe an average of 4 ivAPAP doses, with the majority also receiving a preoperative dose of 1000 mg infused over 15 minutes. They acknowledge that “preoperative and postoperative oral dosages may prove just as efficacious but at a significantly lower price and less risk in terms of manipulation of the patient’s IV site.” Given the relatively high cost27 of ivAPAP (compared to the oral formulation) and only slightly faster onset in comparison to oral administration,28 there seems to be no immediate indication to give IV when oral acetaminophen administration is possible based on the evidence currently available.5
This study is limited by a number of factors that are inherent to database analyses. All the data are observational, and therefore it is not possible to establish a definitive causal relationship. Furthermore, these data were not collected for the purpose of research but for administrative use and therefore lack clinically detailed variables of interest such as surgery time and anesthetic drug doses. Importantly, information on drivers of ivAPAP prescription was not available. These include a patient’s propensity for increased postoperative pain, subjective pain thresholds or hospital policies, and personal preferences of the practitioner. Moreover, the high cost of ivAPAP relative to other nonopioid analgesics may have further influenced prescribing patterns. While we could not account for this in the current study, future studies focusing on decision making in perioperative analgesia practices will add to the discussion on protocolized versus individualized care. Given the limitations of observational data, it is crucial to perform sensitivity analyses that corroborated our main results, further highlighting the robustness of our null results. Even though the provider of the Premier database performs thorough quality checks10 on a regular basis to identify and correct coding mistakes or falsely entered data, we cannot fully exclude data entry errors. As in other observational databases, the use of ICD-9 codes and billing data may not yield the most accurate estimates of actual prevalence where complications may be underestimated and preexisting comorbidities may be overestimated (as they increase reimbursements) or underestimated (as there is a limited number of fields for ICD-9 codes). Importantly, these potential biases will be equally divided between our study groups (those receiving ivAPAP and those who did not), therefore minimizing their effect. In addition, given the nature of the data entry that relies on billing for pharmaceutical units, we do not have information on actual opioid consumption or utilization, rather opioid prescription. It is possible that, especially on the day of surgery, when intravenous formulations (either in PCA form or intraoperatively) of opioids are more likely to be used, whole vials are accounted for and not just fractions consumed. Therefore, we cannot determine with certainty if the opioids prescribed were actually consumed by the patient. However, because this includes only in-hospital records, the risk of severe overprescription is not as high as in an outpatient setting, and to some extent prescription is likely to be linked to consumption, especially after POD 0. Moreover, our multivariable models adjust for the use of PCA.
In conclusion, our analysis did not support the assumption that perioperative ivAPAP reduces inpatient opioid prescription or clinically relevant opioid-related adverse events. Further, no clinically significant benefit regarding LOS and cost was found. While no causal relationships can be established and prescription bias may exist, further studies are warranted to identify if ivAPAP might have the potential to reduce opioid consumption and lower odds for opioid-related adverse effects under specific conditions and settings. While the use of pregabalin/gabapentin seems to convey a benefit in many regards as shown here, the routine use of ivAPAP in spine surgery cannot be supported by our current data.
Name: Eva E. Mörwald, MD.
Contribution: This author helped with research conceptualization and design, data interpretation from a clinical perspective, literature search and putting results in perspective, drafting of the manuscript, and critical revisions.
Name: Jashvant Poeran, MD, PhD.
Contribution: This author helped with research conceptualization and design of statistical analysis plan, data interpretation, tables, multivariable models, figures, drafting of the manuscript, and more importantly critical revisions.
Name: Nicole Zubizarreta, MPH.
Contribution: This author helped with research conceptualization and design, data management, creating usable datasets, data analysis, data interpretation, drafting of the manuscript, and critical revisions.
Name: Crispiana Cozowicz, MD.
Contribution: This author helped with research conceptualization and design, data interpretation from a clinical perspective, literature search and putting results in perspective, drafting of the manuscript, and critical revisions.
Name: Madhu Mazumdar, PhD.
Contribution: This author helped with obtaining/purchasing the data, research conceptualization and design, data interpretation from a statistical methods perspective, drafting of the manuscript and critical revisions, and statistical input regarding revisions of analyses.
Name: Stavros G. Memtsoudis, MD, PhD, FCCP.
Contribution: This author helped with obtaining/purchasing the data, research conceptualization and design, data interpretation from a clinical perspective, drafting of the manuscript, and critical revisions.
This manuscript was handled by: Honorio T. Benzon, MD.
1. White PF. The changing role of non-opioid analgesic techniques in the management of postoperative pain. Anesth Analg. 2005;101:S5–S22.
2. Ashburn MA, Caplan RA, Carr DB, et al. Practice guidelines for acute pain management in the perioperative setting: an updated report by the American Society of Anesthesiologists Task Force on Acute Pain Management. Anesthesiology. 2004;100:1573.
3. Buvanendran A, Kroin JS. Multimodal analgesia for controlling acute postoperative pain. Curr Opin Anaesthesiol. 2009;22:588–593.
4. Hiller A, Helenius I, Nurmi E, et al. Acetaminophen improves analgesia but does not reduce opioid requirement after major spine surgery in children and adolescents. Spine (Phila Pa 1976). 2012;37:E1225–E1231.
5. Jibril F, Sharaby S, Mohamed A, Wilby KJ. Intravenous versus oral acetaminophen for pain: systematic review of current evidence to support clinical decision-making. Can J Hosp Pharm. 2015;68:238–247.
6. Kelly JS, Opsha Y, Costello J, Schiller D, Hola ET. Opioid use in knee arthroplasty after receiving intravenous acetaminophen. Pharmacotherapy. 2014;34suppl 122S–26S.
7. Remy C, Marret E, Bonnet F. Effects of acetaminophen on morphine side-effects and consumption after major surgery: meta-analysis of randomized controlled trials. Br J Anaesth. 2005;94:505–513.
8. Wininger SJ, Miller H, Minkowitz HS, et al. A randomized, double-blind, placebo-controlled, multicenter, repeat-dose study of two intravenous acetaminophen dosing regimens for the treatment of pain after abdominal laparoscopic surgery. Clin Ther. 2010;32:2348–2369.
9. Apfel CC, Turan A, Souza K, Pergolizzi J, Hornuss C. Intravenous acetaminophen reduces postoperative nausea and vomiting: a systematic review and meta-analysis. Pain. 2013;154:677–689.
11. Makadia R, Ryan PB. Transforming the Premier Perspective Hospital Database into the Observational Medical Outcomes Partnership (OMOP) Common Data Model. EGEMS (Wash DC). 2014;2:1110.
12. Kessler ER, Shah M, Gruschkus SK, Raju A. Cost and quality implications of opioid-based postsurgical pain control using administrative claims data from a large health system: opioid-related adverse events and their impact on clinical and economic outcomes. Pharmacotherapy. 2013;33:383–391.
13. Lexicomp Online®. Clinical Drug Information. Opioid Agonist Conversion. Available at: https://online.lexi.com/lco/action/calc/calculator/70050
. Accessed January 4, 2018.
14. GlobalRPH. Opioid Analgesic Converter.. Available at: http://www.globalrph.com/narcoticonv.htm
. Accessed January 4, 2018.
15. Khelemsky Y, Kothari R, Campbell N, Farnad S. Incidence and demographics of post-operative naloxone administration: a 13-year experience at a major tertiary teaching institution. Pain Physician. 2015;18:E827–E829.
16. Quan H, Sundararajan V, Halfon P, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43:1130–1139.
17. Witte JS, Greenland S, Kim LL, Arab L. Multilevel modeling in epidemiology with GLIMMIX. Epidemiology. 2000;11:684–688.
18. Cherny N, Ripamonti C, Pereira J, et al.; Expert Working Group of the European Association of Palliative Care Network. Strategies to manage the adverse effects of oral morphine: an evidence-based report. J Clin Oncol. 2001;19:2542–2554.
19. Moran JL, Solomon PJ; ANZICS Centre for Outcome and Resource Evaluation (CORE) of the Australian and New Zealand Intensive Care Society (ANZICS). A review of statistical estimators for risk-adjusted length of stay: analysis of the Australian and new Zealand Intensive Care Adult Patient Data-Base, 2008-2009. BMC Med Res Methodol. 2012;12:68.
20. Rascati KL, Smith MJ, Neilands T. Dealing with skewed data: an example using asthma-related costs of medicaid clients. Clin Ther. 2001;23:481–498.
21. Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46:399–424.
22. Parsons LS; SAS User Group International 29. Paper 165-29. Performing a 1:N Case-Control Match on Propensity Score. 2005. Available at: http://www2.sas.com/proceedings/sugi29/165-29.pdf
. Accessed November 15, 2017.
23. Moon YE, Lee YK, Lee J, Moon DE. The effects of preoperative intravenous acetaminophen in patients undergoing abdominal hysterectomy. Arch Gynecol Obstet. 2011;284:1455–1460.
24. Sinatra RS, Jahr JS, Reynolds LW, Viscusi ER, Groudine SB, Payen-Champenois C. Efficacy and safety of single and repeated administration of 1 gram intravenous acetaminophen injection (paracetamol) for pain management after major orthopedic surgery. Anesthesiology. 2005;102:822–831.
25. Bajwa SJ, Haldar R. Pain management following spinal surgeries: an appraisal of the available options. J Craniovertebr Junction Spine. 2015;6:105–110.
26. Smith AN, Hoefling VC. A retrospective analysis of intravenous acetaminophen use in spinal surgery patients. Pharm Pract (Granada). 2014;12:417.
27. Poeran J, Babby J, Rasul R, Mazumdar M, Memtsoudis SG, Reich DL. Tales from the wild west of US drug pricing: the case of intravenous acetaminophen. Reg Anesth Pain Med. 2015;40:284–286.
28. Smith HS. Perioperative intravenous acetaminophen and NSAIDs. Pain Med. 2011;12:961–981.