This pooled analysis of three phase 3 randomized controlled trials assessed the safety and tolerability of CTC 200 mg BID versus tramadol 50 mg QID and 100 mg QID, celecoxib 100 mg BID, and placebo in patients with acute moderate-to-severe pain. The results demonstrated that the incidence of AEs related to study drug within the first 48 h of treatment was numerically lower with CTC 200 mg BID compared with tramadol 50 mg QID and 100 mg QID but greater than the incidence observed with celecoxib 100 mg BID or placebo. The safety profiles observed in our analysis are consistent with those reported for the study treatments in the individual ESTEVE-SUSA-301, STARDOM1, and STARDOM2 trials (Tables S5, S6, and S7; Fig. S1) [25,26,27] and also with those reported from the early phase trials with CTC [21, 22, 24].

Direct comparison with AE rates for tramadol in other studies is complicated for a number of reasons, namely, differences in patient populations, doses, formulations, treatment durations, and trial durations; the evaluation of patient-reported data using questionnaires that can lead to data inaccuracies inherent with this form of data collection; previous use of patient-controlled analgesia, which has been shown to increase tolerance to some analgesic-related AEs [29, 30]; and differences in allowed use of concomitant rescue medications [31]. Opioids are often associated with side effects such as somnolence, nausea, vomiting, and constipation [8,9,10, 16]. In keeping with these observations, we noted that the incidence of somnolence, nausea, vomiting, and constipation related to study drug was greater in patients receiving tramadol 100 mg QID than in those receiving CTC 200 mg BID, celecoxib 100 mg BID, or placebo. The incidence of somnolence, dizziness, and vomiting related to study drug was lower with tramadol 50 mg QID than with tramadol 100 mg QID, suggesting that these events were dose dependent. As may be expected, the incidence of these AEs was also numerically higher with the CTC 200 mg BID treatment group compared with the celecoxib 100 mg BID or placebo treatment groups, which would suggest that the safety profile of CTC is driven primarily by the effects of tramadol. Indeed, the safety profile of CTC was consistent with the expected safety profiles of the individual tramadol and celecoxib components, based on their labels [19, 32, 33], and CTC did not appear to have an additive effect. The US prescribing information for CTC [19] does contain a boxed warning reflective of its components. Due to the tramadol component of CTC, the US prescribing information for CTC contains warnings about the potential for addiction, abuse, and misuse, as well as the potential harm from respiratory depression and accidental ingestion, and the risks of concomitant use with benzodiazepines or other central nervous system depressants, interactions with drugs affecting cytochrome P450, neonatal opioid withdrawal syndrome, and the ultra-rapid metabolism of tramadol and other risk factors for life-threatening respiratory depression in children. The label highlights the need for an opioid analgesic Risk Evaluation and Management Strategy to ensure that the benefits of CTC treatment will outweigh the risks of addiction, abuse, and misuse. The risks of gastrointestinal bleeding and cardiovascular thrombotic events associated with nonsteroidal anti-inflammatory drugs and cyclooxygenase-2 inhibitors are also highlighted in the boxed warning [19].

Multi-modal analgesia is now a widely accepted approach for the management of acute pain and the concurrent limiting of opioid consumption and opioid-related AEs [2]. CTC provides a multi-modal approach to analgesia, targeting both central and peripheral pain pathways [18]. Indeed, CTC 200 mg BID has been reported to provide more effective analgesia than both tramadol 50 mg QID [25] and 100 mg QID [26] and celecoxib 100 mg BID [25] in patients with acute pain while also reducing the need for rescue medication [25,26,27]. Since rescue medications may also cause AEs, a reduction in rescue medication with CTC could help contribute to its overall tolerability profile. In addition, there may be other ways in which the tolerability of CTC may be improved compared with tramadol. For example, the multi-modal approach and the pharmacokinetics of the components in CTC allow patients to achieve adequate analgesia while also receiving lower equivalent doses of tramadol (176 mg daily) than patients receiving tramadol alone (200 or 400 mg daily). In addition to providing multi-modal analgesia, the co-crystallization of tramadol and celecoxib in CTC optimizes the bioavailability and pharmacokinetics of the individual components [21, 22, 24]. Results from pharmacokinetic studies have demonstrated that the maximum plasma concentrations of tramadol and celecoxib achieved with CTC 200 mg are lower than those observed with tramadol 100 mg and celecoxib 100 mg when administered alone [21, 22, 24]. Furthermore, the time to achieve maximum plasma concentration of tramadol with CTC is delayed compared with tramadol alone [21, 22, 24]. Slower administration of tramadol has been suggested to help improve tolerability [21, 22, 24]; the lower maximum plasma concentration of tramadol with CTC and the delay to achieve this maximum concentration may both therefore help contribute to the improved tolerability profile of CTC versus tramadol alone. This may also contribute to the observation of fewer AEs with CTC versus tramadol 100 mg QID. Indeed, we observed that the study drug-related AE rate with CTC was comparable to that seen with tramadol 50 mg QID (200 mg daily) and approximately half that seen with tramadol 100 mg QID (400 mg daily). Thus, the clinical efficacy of CTC is not accompanied by an increase in the number or severity of AEs compared with tramadol alone. CTC appears to possess a benefit-risk profile more favorable than that of either tramadol or celecoxib alone, providing more effective pain management than celecoxib and tramadol 50 mg QID [25] or better tolerability than tramadol 100 mg QID.

CTC may represent a valuable addition to the options available to physicians treating patients with acute pain, particularly after surgery. When treating postoperative pain, physicians may face numerous challenges, such as encouraging mobility during postoperative pain, discharge delays due to uncontrolled pain, and hospital readmissions due to pain [7]. CTC is administered as a regular dose, which may offer advantages versus regimens that are administered as needed and therefore require more consistent medical input. The positive benefit-risk profile and regular dosing schedule with CTC may help to reduce the occurrence of analgesia gaps and the need for rescue medication. This, in turn, may help patients achieve a quicker discharge without the need for direct nursing or other medical input to manage as-needed treatments. However, further studies may be required to establish whether these benefits of CTC will translate into reduced readmissions for pain.

Our pooled analysis has strengths and limitations that may need to be considered when interpreting the findings. Data were pooled from several trials; this increased number of patients may help to strengthen conclusions regarding the safety and tolerability of treatments. The trials employed different models of acute postoperative pain; these models were considered standard, validated, and sensitive for assessing the treatment of acute pain [34, 35]. Moreover, the models were complementary, because they comprised a complex pathophysiology involving tissue damage and inflammation resulting from surgery. Another consideration for our pooled analysis is that the treatment duration differed for each of the trials, although the overall safety follow-up time was similar (~ 7 days). Our analysis focused on safety data for up to 48 h after randomization, as most AEs were expected to appear in the first few hours of treatment and the AE profiles at follow-up were expected to have only minimal differences from those at 48 h. We also assessed safety and tolerability data up to the last follow-up, and both the pattern of adverse effects and the overall conclusions were comparable to the data after 48 h.

No statistical analyses were performed between treatment groups for AEs. Indeed, AEs are reported spontaneously and can differ greatly in their nature, severity, and duration; consequently, they differ from pre-determined endpoints and represent challenges for performing statistical analyses. Moreover, no statistical analysis was conducted to determine the contribution of rescue medication to AE rates.

A greater use of rescue medication was reported for placebo versus CTC in these studies [25,26,27], and AEs from these rescue medications may contribute more to the AE rate for placebo than for other treatments. Among patients who did not receive opioid rescue medication in ESTEVE-SUSA-301 (the only trial in this pooled analysis in which patients were permitted opioid-containing rescue medication), nausea and vomiting occurred less frequently with CTC than with tramadol alone; for patients who did receive oxycodone rescue medication, the incidence of nausea was similar for CTC and tramadol [25, 36].

In the present analysis, data for tramadol 50 mg QID were derived from only one trial. Consequently, there were some differences in patient demographics at baseline and treatment exposure for tramadol 50 mg QID compared with groups where data were pooled from two or more studies. These differences were not apparent when comparing baseline demographics in just the ESTEVE-SUSA-301 trial [25]. Thus, although we report data for tramadol 50 mg QID data for completeness, the differences in baseline demographics and treatment exposure would indicate that caution should be used when comparing with data pooled across multiple studies. Indeed, this may explain why some AEs, such as somnolence and fatigue, occurred at greater rates with pooled placebo data than with tramadol 50 mg QID (Table S2), but this was not observed when comparing these treatment groups directly in the ESTEVE-SUSA-301 trial (Table S5) [25].

Care should also be taken when generalizing our analysis, as the population may not be fully representative of a broader acute pain patient population; the majority of patients in our pooled analysis were female because of the inclusion of two trials in patients undergoing a bunionectomy or an elective abdominal hysterectomy. Moreover, patients were predominantly < 65 years of age, and care should therefore be taken when generalizing safety to older patients.