Tramadol (with or without acetaminophen) efficacy and harm

Abstract

Objective

To evaluate the efficacy and safety of tramadol (with or without acetaminophen) in adult pain management compared with nonsteroidal anti-inflammatory drugs (NSAIDs), opioids, and acetaminophen alone.

Data sources

Studies were identified through a search of the entire Ovid MEDLINE database, the Ovid MEDLINE Epub Ahead of Print database, Embase, and Evidence-Based Medicine Reviews (Cochrane Central Register of Controlled Trials) from 1946 to August 19, 2022.

Study selection

Included studies comprised double-blind, randomized controlled trials (RCTs) of oral tramadol (with or without acetaminophen) compared with opioids, NSAIDs, or acetaminophen for treatment of acute or chronic pain that reported at least 1 of 11 outcomes: total adverse events (AEs), fatal serious AEs (SAEs), nonfatal SAEs, patient withdrawals from the study due to AEs (WDAEs), total study withdrawals, measures of dependence or addiction, quality of life, functional improvement, 30% or more pain reduction, 50% or more pain reduction, and rescue medication use. Overall, 3184 articles were screened and 37 RCTs (21 opioid, 15 NSAID, 1 acetaminophen; N=7156; length 1 hour to 12 weeks) were included. Trials were critically appraised and assessed using the Cochrane risk of bias tool for randomized trials 1.0. Meta-analyzed outcomes were reported as risk ratios (with 95% confidence intervals [CIs]).

Synthesis

Compared with opioids, tramadol with or without acetaminophen did not significantly differ in efficacy based on 50% or more or 30% or more reduction in pain or in any harm outcomes (nonfatal SAEs [risk ratio (RR)=1.35, 95% CI 0.43 to 4.20]; WDAEs [RR=0.99, 95% CI 0.80 to 1.22]; total withdrawals [RR=0.93, 95% CI 0.78 to 1.11]; or total AEs [RR=0.97, 95% CI 0.89 to 1.05]). Compared with NSAIDs, tramadol with or without acetaminophen was less likely to achieve a 30% or more reduction in pain (RR=0.82, 95% CI 0.76 to 0.90), and was more likely to result in WDAEs (RR=2.86, 95% CI 2.23 to 3.66), total withdrawals (RR=1.68, 95% CI 1.47 to 1.93), and total AEs (RR=1.37, 95% CI 1.28 to 1.47). Evidence was insufficient for meta-analysis of fatal SAEs, drug dependence or addiction, quality of life, functional improvement, or use of rescue medications.

Conclusion

Tramadol with or without acetaminophen did not differ significantly from other opioids, and was less effective and more poorly tolerated than NSAIDs. Short study duration, small sample size, extensive patient exclusion, and inconsistency in outcome reporting limit the scientific validity of conclusions.

Family physicians from Canada have indicated that pain management is challenging due to competing messages regarding undertreatment and overtreatment and a lack of education or expert guidance.¹

Tramadol (with or without acetaminophen) is an opioid analgesic licensed for treating moderate to severe pain. In 2007, following the cyclooxygenase-2 inhibitor scandals and increased awareness of nonsteroidal anti-inflammatory drug (NSAID) organotoxicity, articles and guidelines positioned tramadol as a safer alternative to NSAIDs and other opioid analgesics.² For example, in accordance with the World Health Organization (WHO) pain ladder, 2011 Canadian pain guidelines stated tramadol was effective for chronic pain and encouraged its use as preferable to more potent opioids because of lower rates of overdose, misuse, and addiction.^3,4

Although initially marketed and popularized as a non-opioid analgesic, tramadol results in a similar amount of opioid agonism as codeine, with the WHO Expert Committee on Drug Dependence concluding it has a risk of addiction similar to opioids like morphine.⁵ Consequently, tramadol predominates the opioid crisis in Africa and the Middle East.^6,7

In the early 2000s, the product monograph for tramadol was updated to include a serious warning for potential addiction, abuse, misuse, overdose, and death.⁸ A 2021 population-based cohort study determined that, compared with codeine, new prescriptions for tramadol were associated with a 2.3-fold increase in all-cause mortality.⁹ In 2022, Health Canada added tramadol to the existing list of opioids in schedule I of the Controlled Drugs and Substances Act and to the Narcotic Control Regulations.¹⁰

From 2007 to 2017, total opioid prescriptions declined across Canada, but prescriptions for tramadol were steady or increasing.¹¹ The most recent data available show a 4% increase in tramadol prescriptions filled in Canada from 2015 to 2019.¹² Given both its popularity as an analgesic and growing concerns regarding safety, we sought evidence comparing tramadol with opioids, NSAIDs, and acetaminophen.

METHODS

Data sources

This systematic review was conducted according to the Cochrane Handbook for Systematic Reviews of Interventions (Appendix 1, available from CFPlus^*).¹³ We applied the Preferred Reporting Items for Systematic Reviews and Meta-Analyses protocol for quality and publication standards.

The electronic databases searched included the entire Ovid MEDLINE database, the Ovid MEDLINE Epub Ahead of Print database, Embase, and Evidence-Based Medicine Reviews (Cochrane Central Register of Controlled Trials) from 1946 to August 19, 2022. Detailed search strategies are provided in Appendix 2.^* Additional manual searches to ensure completeness yielded 3 published Cochrane systematic reviews^14-16 and 1 Canadian Agency for Drugs and Technologies in Health common drug review report from 2007 (personal communication with Dr Ken Bassett, Co-managing Director of the Therapeutics Initiative, March 11, 2020).

Eligibility criteria

Included studies met the following criteria: a population of adults with acute or chronic pain; an intervention of oral tramadol with or without acetaminophen (any dose or formulation); a comparator of oral or transdermal opioids with or without acetaminophen, oral NSAIDs, or acetaminophen (Appendix 1^*); 1 or more outcomes of total adverse events (AEs), fatal serious AEs (SAEs), nonfatal SAEs, patient withdrawals from the study due to AEs (WDAEs), total withdrawals, drug dependence or withdrawal symptoms or measure of addiction, quality of life, functional improvement, clinically meaningful efficacy (proportion of patients with ≥30% or ≥50% pain reduction from baseline), and rescue medication use; and a double-blind randomized controlled trial (RCT) study design.

Study selection

Titles and abstracts of records identified through database searches were screened by 3 reviewers (J.A.O. and G.C., with contributions from Jano Klimas). Potentially relevant abstracts proceeded to full-text review and were each screened by 2 reviewers (J.A.O., G.C.). Any discrepancies in inclusion or exclusion reasons were resolved through discussion and consensus.

Data extraction

Two authors (J.A.O., G.C.) extracted data independently. All relevant data were extracted using Covidence software (https://www.covidence.org) and exported to Review Manager 5.4 for analysis.

Risk of bias assessment

Risk of bias of included studies was assessed by 2 reviewers (J.A.O., G.C.) independently and in duplicate using the Cochrane risk-of-bias tool for randomized trials 1.0.

Data synthesis and analysis

Outcomes with studies using the same definitions and units of measurement were pooled for meta-analysis using Review Manager 5.4. Differences between interventions were reported using summary risk ratios (RRs) with 95% confidence intervals (CIs), with statistically significant differences designated by P<.05, and heterogeneity between studies measured by I². The Mantel-Haenszel method and a fixed-effects model were used to combine dichotomous outcomes data. When study data could not be pooled (eg, heterogeneity between studies in outcome definitions, measurement, or reporting), claims were narratively summarized. No sensitivity or subgroup analyses were undertaken.

RESULTS

Searches of electronic databases yielded 3184 records published from 1946 to August 19, 2022. Following removal of duplicates, the titles and abstracts of 3182 records were screened. The full texts of 119 records were assessed for eligibility. Overall, 36 texts (37 studies)^17-52 met inclusion criteria (Figure 1).

Figure 1.

PRISMA flow diagram

Study characteristics

Twenty-one studies involving 3205 participants compared tramadol with or without acetaminophen to opioids^{21-23,27-29,31,33-36,42-44,46-52}; 15 studies (n=3931) compared it to NSAIDs^{17-19,24-26,30,32,37-41,45}; and 1 study (n=20) compared tramadol to acetaminophen.²⁰ Thirteen studies using an opioid comparator,^{21,23,27-29,31,34,46-48,50-52} and 8 using NSAIDs^{17,18,25,26,30,32,41,45} assessed acute pain; most studied immediate postprocedure pain (eg, dental extractions, acute trauma, and gynecologic procedures) but 1 continued to 2 weeks after rotator cuff repair.¹⁸ Eight opioid studies,^{22,33,35,36,42-44,49} 1 acetaminophen study,²⁰ and 7 NSAID studies^19,24,37-40 assessed chronic pain, which was usually musculoskeletal, oncologic, or unspecified in nature. Trial durations ranged from 1 hour to 12 weeks. For a summary of pain types, contexts of use, and study durations, see Appendix 3^*; study characteristics are listed in Appendix 4.^*

Risk of bias

Figure 2 shows the consensus ratings for each of 8 risk of bias domains in all 36 included papers.^17-52 All studies had low or unclear risk of bias for random sequence generation, allocation concealment, and blinding of outcome assessment. All but 1 study⁴⁵ had unclear or low risk of bias from blinding of participants and personnel. Most studies had uncertain or high risk of bias due to incomplete and selective reporting of outcomes, incomplete outcome data, funding sources, and other biases. More detail can be found in Appendix 5.^*

Figure 2.

Risk of bias: Green represents low risk, yellow unclear risk, and red high risk of bias.

Outcome summaries and results of meta-analysis

The study outcomes of interest are described below and presented in detail in Appendix 6.^* Tables 1 and 2 summarize measures that were meta-analyzed; corresponding forest plots are in Appendix 7.^* Most studies did not report on efficacy outcomes of interest to this systematic review or prespecify thresholds for clinical relevance; however, many reported on changes in pain intensity, and a narrative summary of those claims is available from Appendix 8.^*

Table 1.

Tramadol compared with NSAIDs: Outcome results of meta-analyses.

OUTCOME	TRAMADOL, OR TRAMADOL WITH ACETAMINOPHEN, n		NSAID COMPARATOR, n		RR (95% CI)	P VALUE	I2 VALUE (NO. OF STUDIES)
	TOTAL EVENTS	TOTAL PATIENTS	TOTAL EVENTS	TOTAL PATIENTS
Nonfatal SAEs	5	1792	4	1415	1.24 (0.38-4.06)	.72	45 (9)
WDAEs	293	2018	70	1688	2.86 (2.23-3.66)	<.00001	37 (12)
Total withdrawals	552	2081	225	1813	1.68 (1.47-1.93)	<.00001	28 (13)
Total AEs	1142	2035	647	1754	1.37 (1.28-1.47)	<.00001	74 (14)
No. of patients with ≥30% reduction in pain from baseline	412	785	508	798	0.82 (0.76-0.90)	<.00001	0 (2)

AE—adverse event, CI—confidence interval, NSAID—nonsteroidal anti-inflammatory drug, RR—risk ratio, SAE—serious adverse event, WDAE—withdrawal due to AE.

Table 2.

Tramadol compared with opioids: Outcome results of meta-analyses.

OUTCOME	TRAMADOL, OR TRAMADOL WITH ACETAMINOPHEN, n		OPIOID COMPARATOR, n		RR (95% CI)	P VALUE	I2 VALUE (NO. OF STUDIES)
	TOTAL EVENTS	TOTAL PATIENTS	TOTAL EVENTS	TOTAL PATIENTS
Nonfatal SAEs	6	1218	3	919	1.35 (0.43-4.20)	.61	0 (10)
WDAEs	162	1550	134	1285	0.99 (0.80-1.22)	.94	23 (16)
Total withdrawals	233	1528	190	1226	0.93 (0.78-1.11)	.41	0 (17)
Total AEs	585	1318	450	1020	0.97 (0.89-1.05)	.48	72 (12)
No. of patients with ≥30% reduction in pain from baseline	117	190	128	201	0.97 (0.83-1.13)	.67	NA (1)

AE—adverse event, CI—confidence interval, NA—not applicable, NSAID—nonsteroidal anti-inflammatory drug, RR—risk ratio, SAE—serious adverse event, WDAE—withdrawal due to AE.

Outcomes versus acetaminophen. A single study²⁰ compared tramadol with acetaminophen, but the study was too small (10 patients in each arm) to generate reliable conclusions.

Outcomes versus NSAIDs. Four NSAID comparator studies reported on mortality, and none reported any deaths.^24,25,37 The pooled nonfatal SAEs did not differ between tramadol with or without acetaminophen and NSAIDs (9 studies, RR=1.24, 95% CI 0.38 to 4.06). There were more WDAEs among patients taking tramadol with or without acetaminophen than among patients taking NSAIDs (12 studies, RR=2.86, 95% CI 2.23 to 3.66), total patient withdrawals (13 studies, RR=1.68, 95% CI 1.47 to 1.93), and AEs (14 studies, RR=1.37, 95% CI 1.28 to 1.47). No NSAID studies reported on 50% or more pain reduction. Both studies from O’Donnell et al³⁷ reported the proportion of patients with 30% or more pain reduction from baseline; fewer patients in the tramadol group compared with those taking celecoxib experienced that improvement (2 studies, RR=0.82, 95% CI 0.76 to 0.90).

Measures of drug dependence or addiction were assessed in 2 studies comparing tramadol and NSAIDs: Beaulieu et al¹⁹ found no significant difference in drug-liking index scores. DeLemos et al²⁴ employed both an addiction scale and a physical dependence scale but only reported comparisons with placebo, not with NSAIDs. Quality of life and functional outcomes could not be meta-analyzed due to inconsistency of measurement and absence of standard, validated scales, and most studies reported no significant difference or made no claims regarding these metrics. Peters et al⁴¹ reported a significant difference favouring naproxen according to patient ratings of ability to proceed with normal activities. Rescue medication use was not meta-analyzed due to significant heterogeneity in definitions and reporting. In many studies, patients using rescue medications were withdrawn. Of the NSAID comparator studies reporting on rescue medication use, 5 studies^{17,19,25,32,45} found no difference and 1 made no claim.⁴¹

Outcomes versus opioids. There were no deaths in the 3 opioid studies reporting mortality.^27,28,33 The pooled nonfatal SAEs did not differ between tramadol with or without acetaminophen and opioids (10 studies, RR=1.35, 95% CI 0.43 to 4.20). There were similar rates of total study withdrawals between tramadol with or without acetaminophen and opioid groups (17 studies, RR=0.93, 95% CI 0.78 to 1.11), WDAEs (16 studies, RR=0.99, 95% CI 0.80 to 1.22), and AEs (12 studies, RR=0.97, 95% CI 0.89 to 1.05).

Physical dependence or withdrawal symptoms were reported using various nonvalidated scales and could not be pooled for analysis. Two studies reported no difference: Leng et al³³ found equivalent subjective opioid withdrawal symptoms 1 week after tramadol or transdermal buprenorphine was started for musculoskeletal pain. Mullican et al³⁶ reported no signs of abuse or dependence on tramadol with acetaminophen or codeine with acetaminophen in patients with chronic lower back pain, but this outcome was not predefined, and it is not clear how it was measured. Most studies excluded patients with a history of substance use disorder.

Quality-of-life metrics were reported inconsistently and function was not assessed with any validated scale. Studies that reported some measure of quality of life reported no difference^23,33 or mixed results.⁵⁰ One study found tramadol resulted in a lower median stiffness duration but no change in stiffness score when compared with pentazocine in osteoarthritis.²² Two found no significant difference between study groups in mobility or activity impairment.^23,28

Hewitt et al is the only opioid study to report the proportions of patients with 50% or more and 30% or more pain reductions and found no statistically significant differences between tramadol and hydrocodone.²⁸

Rescue medication use was variably defined and reported and not able to be meta-analyzed. Of the 17 opioid comparator studies reporting rescue medication use, 8 claimed no difference,^{22,27,29,33,34,36,49,51} and 6 made no claim or did not permit rescue medication use.^{28,31,42,46,48,50} Zavareh et al⁵² found patients taking 50 mg of tramadol required less morphine as rescue medication in the 24 hours after cholecystectomy than did those in the combination 325 mg of acetaminophen and 10 mg of codeine group, but this tramadol dose is far higher in oral morphine equivalents than the codeine dose is.⁵³ Two studies favoured the opioid comparator: Stubhaug et al⁴⁷ stated that rescue medicine was needed less often in the acetaminophen-codeine group compared with 2 different tramadol dose groups after total hip replacement; Bindal et al²¹ found that for bladder discomfort, 1 patient in the tapentadol premedication group and 14 in the tramadol group required rescue medication.

DISCUSSION

Comparative data on analgesics facilitates informed discussions with patients about anticipated efficacy and harm. This review includes comparative studies from a range of settings, incorporating patients with both acute and chronic pain, and found no significant differences between tramadol with or without acetaminophen and opioid comparators for any safety or efficacy outcomes. When compared with NSAIDs, tramadol with or without acetaminophen was less effective at pain reduction of 30% or more and resulted in significantly higher WDAEs, total withdrawals, and total AEs.

Our systematic review differs slightly from existing Cochrane systematic reviews. Toupin April et al¹⁴ looked at osteoarthritis pain and separated tramadol from tramadol with acetaminophen, permitted open-label RCTs, and included both placebo and active comparators. They reported similar findings to our study, but unlike our study that found no difference, they did find low-quality evidence that tramadol alone had a greater risk of WDAEs compared with opioids.¹⁴ Duehmke et al found no clear therapeutic benefit of tramadol compared with active or placebo comparators for neuropathic pain.⁵⁴ In the Wiffen et al study,⁵⁵ very low-quality evidence suggested tramadol may offer some advantage compared with placebo, but it is not as effective as morphine for cancer-related pain.

We sought data on clinically meaningful effectiveness (a reduction in pain of ≥30% or ≥50%) in accordance with the Initiative on Methods, Measurement, and Pain Assessment in Clinical Trials consensus.⁵⁶ Most trials did not report on these outcomes, which greatly limited meaningful efficacy comparisons. Confidence in study findings was limited by small sample sizes, noncomparable doses, imbalanced patient selection, and funding by manufacturers. In addition, dropout rates were high and just 15 studies did intention-to-treat analysis^{18,19,24,27,28,33,35-38,42,46-48}; of those that did, 6 did not analyze all participants,^{19, 28,35,42,46,47} and 6 imputed missing data,^{24,27,33,36,42,48} increasing reporting biases, usually favouring tramadol.

Short durations and strict exclusion criteria limit generalizability. This is particularly concerning for real-world decision making, as opioid analgesics are often used for patients with chronic pain or with a history of substance use disorders. In addition, short studies may miss serious or uncommon adverse events and symptoms of dependence or addiction, as these tend to develop over months or years. A review focusing on tramadol for chronic pain is planned by Danish investigators, and their results may reinforce or refute our conclusions and decision to combine acute and chronic indications.⁵⁷

Recent retrospective observational studies reveal a higher likelihood of prolonged opioid use in patients receiving a prescription of tramadol compared with those receiving other short-acting opioids.^58,59 Given the persistent and overwhelming opioid crisis in Canada, the association is compelling enough to prompt concern about the addiction potential from tramadol exposure.

Conclusion

Compared to patients taking NSAIDs, patients taking tramadol with or without acetaminophen are less likely to achieve a 30% or more pain reduction from baseline and more likely to experience WDAEs, total withdrawals, and total AEs. Compared with opioids, tramadol with or without acetaminophen result in similar efficacy and harm. The use of tramadol should be considered carefully and monitored in the same ways as other opioid analgesics.