Does Ketorolac reduce the intensity of postoperative pain after impacted third molars surgery in adults compared to the use of tramadol? A systematic review and meta-analysis

This systematic review and meta-analysis evaluated if ketorolac reduces the intensity of postoperative pain after impacted third molars surgery in adults compared to the use of tramadol. A comprehensive search was performed in the MEDLINE/PubMed, Scopus, Web of Science, LILACS, BBO, EMBASE, Cochrane Library, SIGLE and grey literature, in accordance with the PRISMA guidelines. The quality of the evidence was evaluated using the GRADE approach. Meta-analysis was performed on studies considered at low risk of bias. After duplicates removal, 4526 articles were identified, but only nine studies were included for qualitative analysis. After updating in 2021, four studies were added, totaling 13 studies included for qualitative analysis. Only two studies, classified at “low” risk of bias, were included in the meta-analysis of the primary outcome. The difference in means for pain intensity (moderate quality of evidence due to imprecision) was – 0.27 (95% CI = – 0.82 to 0.28; p = 0.34). Data from adverse effects (low quality of evidence due to very serious issues in imprecision) was just reported in one study at “low” risk of bias. Data was not heterogeneous (Chi2 test p = 0.14; I2 = 55%). It was not possible to evaluate any secondary outcomes (time to first rescue analgesic drug in h, total amount of analgesics consumed and adverse effects) due to low number of studies included. There is a lack of strong evidence to assure the superiority of ketorolac or tramadol in reducing the postoperative pain after extraction of impacted third molars. Research, Society and Development, v. 10, n. 3, e19410313137, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i3.13137 2

These conflicting findings prevent clinicians from drawing a conclusion about which drug is more effective in controlling pain with fewer side effects. Therefore, the aim of this systematic review and meta-analysis was to answer the following PICO question (P=Population, I=Intervention, C=Comparison, O=Outcome): Is ketorolac more effective than tramadol in reducing the intensity of postoperative pain after surgery of impacted third molars in adults?

Methodology
The description of the article followed the recommendations of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Moher, Liberati, Tetzlaff, & Altman, 2009).

Protocol and registration
This study protocol was registered at the International Prospective Register of Systematic Reviews (#CRD42016036410). This study was accomplished from March to December of 2016 and updated July of 2018 and January of 2021.

Information sources and search strategy
The controlled vocabulary (MeSH terms) and free keywords in the search strategy were defined based on the following PICOS question, reported in the end of the introduction section. For each one of the concepts (population and intervention) medical subheadings (MESH) and free keyword were combined with the Boolean operator "OR" and the two Research, Society andDevelopment, v. 10, n. 3, e19410313137, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v10i3.13137 concepts combined with the Boolean operator "AND".
Pain intensity within the first 24 h was the primary outcome. Time to take the first rescue analgesic drug in hours, total amount of analgesics consumed in mg or in number of capsules and the total number of adverse effects were the secondary outcomes.
The electronic databases MEDLINE via PubMed, Scopus, Web of Science, Latin American and Caribbean Health Sciences Literature database (LILACS), Brazilian Library in Dentistry (BBO), EMBASE and Cochrane Library (Table 1) were searched for the primary studies. Their reference lists were hand searched for additional publications. The related articles links of each primary study in the PubMed database were another source of search. We did not implement restrictions on publication date or languages.  Inhibitors" ) OR TITLE-ABS-KEY ( "Cox 2" ) nOR TITLE-ABS-KEY ( "Cyclooxygenase 2" ) ) OR ( TITLE-ABS-KEY ( "Analgesic drug" ) OR TITLE-ABS-KEY ( analgesic* ) OR TITLE-ABS-KEY ( narcotic* ) OR TITLE-ABS-KEY ( "opioid combination" ) OR TITLE-ABS-KEY ( tramadol ) OR TITLE-ABS-KEY ( "Analgesic opioid" ) ) ) )

Eligibility criteria
We included randomized controlled trials that compared ketorolac vs. tramadol for treatment and prevention of postoperative pain after surgery of impacted third molars. Studies with parallel or crossover designs in humans were included.
RCTs were excluded if they (1) evaluated only one of the drugs (ketorolac or tramadol), (2) associated the ketorolac with tramadol for treatment and prevention of postoperative pain or (3) compared ketorolac vs. tramadol for treatment and prevention of postoperative pain for maxillofacial surgery or extraction of all four third molars in a single session.

Study selection and data collection process
Initially, the articles were selected by title and abstracts. Duplicate articles were removed. Full-text articles were obtained when the title and abstract presented insufficient information to make a clear decision. Subsequently, two reviewers Research, Society and Development, v. 10, n. 3, e19410313137, 2021 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org /10.33448/rsd-v10i3.13137 8 (L.D.M. and M.R.) classified those that met the inclusion criteria. Each study received a study ID combining the first author and year of publication. Any disagreement between the reviewers over the eligibility of particular studies was resolved through discussion with a third reviewer (A.R.).
Relevant information about the study design, participants, interventions and outcomes was extracted using customized extraction forms. Two reviewers (L.D.M. and M.R.), independently and in duplicate, extracted the data. If there were any disagreements between the reviewers, a third reviewer was consulted (A.R.). When data were not reported in the studies, the authors were contacted by email at least twice to request the missing information.

Data extraction and management
Two review authors, independently and in duplicate, extracted data using a data extraction form. Any disagreement was discussed, and a third review author was consulted when necessary.

Risk of bias in individual studies
The quality of the selected trials was assessed using the Cochrane Collaboration tool for assessing the risk of bias. For each aspect of the quality assessment, the risk of bias was scored following recommendations of the Cochrane Handbook for Systematic Reviews of Interventions 5.1.0 (http://handbook.cochrane.org).
The assessment criteria included six items: sequence generation, allocation concealment, blinding of the outcome assessors, incomplete outcome data, selective outcome reporting and other possible sources of bias. For this systematic review, with the patient-centered outcome of pain intensity, the key domains were sequence generation, allocation concealment, and patient blinding.
For each aspect of the quality assessment, the risk of bias was scored following recommendations of the Cochrane Handbook for Systematic Reviews of Interventions 5.1.0 (http://handbook.cochrane.org). At the domain level, the judgment for each entry involved recording "yes" to indicate a low risk of bias, "no" for a high risk of bias or "unclear," indicating either lack of information or uncertainty over the potential for bias.
At the study level, the paper was considered to be at "low" risk of bias if all key domains for each outcome were at low risk of bias. If one or more key domains were judged as "unclear" or at "high" risk of bias, the study was considered at "high" risk of bias. When the study was judged as "unclear" in its key domains, the authors were contacted to obtain more information, which allowed a definitive "yes" or "no" judgment. The whole process of quality assessment was done by two independent reviewers (L.D.M. and M.R.) and disagreements solved through discussion and by consulting a third reviewer (A.R.).

Summary measures and synthesis of the results
Data were analyzed using RevMan 5 (Review Manager Version 5.3, the Cochrane Collaboration, Copenhagen, Denmark). Only studies classified at "low" risk of bias in the key domains were entered into the meta-analysis. The calculation of the standardized mean difference for the continuous data (pain intensity) and the risk ratio for dichotomous data allowed us to summarize the outcomes.
Depending on the number of included studies, subgroup analysis would be performed to evaluate (1) whether or not the outcomes differ if the drug was administered pre-or postoperatively, (2) whether or not the outcomes differ depending on the type of administration of ketorolac (sublingual, intramuscular or intravenous) and (3) whether or not the outcomes differ depending on the dosage of the medicines (low and high dosages).
Random-effects models were used. Heterogeneity was assessed using the Cochran Q test and I2 statistics. Sensitivity analyses were also conducted to investigate the reasons for high heterogeneity whenever detected.

Assessment of the quality of evidence using GRADE
We graded the quality of the evidence for each outcome across studies (body of evidence) using the Grading of Recommendations: Assessment, Development and Evaluation (GRADE) (http://www.gradeworkinggroup.org/). This technique allows one to determine the overall strength of evidence for each meta-analysis (Guyatt, Oxman, Schünemann, Tugwell, & Knottnerus, 2011). The GRADE approach grades the evidence in four levels: very low, low, moderate, high. The "high quality" suggests that we are very confident that the true effect lies close to the estimate of the effect. On the other extreme "very low quality" suggests that we have very little confidence in the effect estimate and the estimate reported can be substantially different from what it was measured.
For randomized clinical trials, the GRADE approach addresses five reasons (risk of bias, imprecision, inconsistency, indirectness of evidence and publication bias) to possibly rate down the quality of the evidence in 1 or 2 levels (Guyatt et al., 2011). Each domain was assessed as "no limitation" (0); "serious limitations" (1 level downgraded) and "very serious limitations" (2 levels downgraded). The GRADEpro Guideline Development Tool, available online (www.gradepro.org), was used to create Summary of findings table as suggested in the Cochrane Handbook for Systematic Reviews of Interventions (Schünemann et al., 2008).

Study selection
The search strategy was realized on 03/06/2016 and updated twice (in 01/07/2018 and 15/01/2021). After the database screening and removal of duplicates, 4526 studies were identified (Figure 1). After title screening, 110 studies remained. This number was reduced to 12 after the abstracts were read and their full texts were assessed to check eligibility. Of these, three were excluded. The reasons for exclusion were (1) two studies evaluated postoperative pain after maxillofacial surgery (Shankariah, Mishra, & Kamath, 2012;Zackova, Taddei, Calò, Bellocchio, & Zanello, 2001) and (2)
The dose of the drugs was also variable. Oral ketorolac was used in doses of 10 mg to 20 mg; intravenous and intramuscular doses were 30 mg. The doses of oral tramadol ranged from 50 to 100 mg; intravenous and intramuscular doses were 50 mg (  Shah et al., 2013;.

Assessment of the risk of bias
The assessment of the risk of bias of the selected studies is presented in Figure 2. In the sequence generation domain, five studies were considered to be at "low" risk of bias (Isiordia-Espinoza et al., 2016;Mishra & Khan, 2012;Ong & Tan, 2004;Pathi et al., 2020;Shah et al., 2013), seven studies were judged as at "unclear" risk (Chethan et al., 2015;Gopalraju et al., 2014;Mangalgi et al., 2018;Nasyam, 2020;Passi et al., 2018; and one study was judged at a "high" risk of bias (Dayashankara Rao, 2010).
However, the study of Pathi et al. (2020) was not included in the meta-analysis as the outcome was the sum of hourly pain intensity scores, making it impossible mix with the other two studies, which measured the average intensity of pain. Important to say that the authors described the outcome in a very unusual way, which put the study at high risk of bias in selective outcome reporting.

Meta-analysis
The meta-analysis was performed on studies classified as a "low" risk of bias in the key domains and from which the information could be extracted. In the study protocol registered at the Prospero database, we planned to extract other secondary outcomes, which were described earlier in the materials and methods section. However, in the two studies at "low" risk of bias, the secondary outcomes (time to take the first rescue analgesic drug in hours, total amount of analgesics consumed in mg or in number of capsules and the total number of adverse effects) were not described, which prevented us from running their metaanalyses (Table 3). Only data from the intensity of pain could be meta-analyzed.

Intensity of pain using VAS pain scales
This analysis was based on two studies that totalized 138 patients. Data on postoperative pain intensity between 6 h and 12 h post-surgery was extracted, depending on the information provided by the authors. The standardized difference in means was -0.27 (95% CI -0.82 to 0.28; p = 0.34), and therefore we do not have enough evidence to show superiority of one drug over the other (Figure 3). Data was not heterogeneous (chi2 test p = 0.14; I2 = 55%). The statistical heterogeneity of only 2 studies does not provide reliable information and should only be seen as an exploratory analysis.
The standard deviations of the study by Mishra and Khan (2012) was not described in the text. In face of that, we arbitrarily imputed a standard deviation corresponding to half of the mean. This decision was based on the coefficient of variation of the other primary included studies, which was around 50%. A sensitivity analysis was run using more extreme standard deviations, and no deviation from the findings reported herein was observed. Source: Authors.

Assessment of the quality of evidence using GRADE
For the main outcome pain intensity, the GRADE quality of evidence was moderate due to "serious" limitations in imprecision ( Table 4). The large confidence interval of the standardized mean difference included potential improvement as well as inferiority of ketorolac compared to tramadol. For the outcome total number of adverse effects, the GRADE quality of evidence was low due to "very serious" limitations for imprecision, where only a single study was included with very low sample size.

Discussion
The qualitative evaluation of the studies included in this systematic review showed that there is great diversity in the protocols used for both drugs, in either the route of administration or dosage.
The absorption is rapid for all three forms of dosing. The time to reach peak serum concentration is shortest after intravenous use, which takes approximately 5 min. After oral use, peak concentration takes 30-53 min (Catapano, 1996), while in intramuscular use, it takes 45-50 min (Catapano, 1996). The serum half-life ranges from 5 to 6 h for all three forms of administration (Catapano, 1996). It is also reported that there is no difference between ketorolac 10 and 20 mg taken orally for pain control (Brown et al., 1990;Buckley & Brogden, 1990;Forbes, Butterworth, Burchfield, & Beaver, 1990;Forbes, Kehm, Grodin, & Beaver, 1990;Olmedo, Gálvez, & Vallecillo, 2001), suggesting that there is a plateau in this agent's analgesic efficacy at the 10 mg level (Olmedo et al., 2001). On the other hand, an increase in analgesic efficacy can be obtained with 30 mg of ketorolac taken intramuscularly when compared with 10 mg taken per the oral rote for surgery models (Fricke et al., 1992).
Another factor that may influence the effect of the drugs is the moment they were given to patients, which may be before or after surgery. This may play a significant role, as pain resulting from surgery is most severe between 6 and 8 h after the surgery (Ong & Tan, 2004;Seymour et al., 1985). In theory, the strategy of pre-surgical analgesic administration is to establish effective blood levels for maximum analgesic effect at the time pain is most severe. The analgesic effect of tramadol begins within 1 h and reaches a peak in approximately 2 to 3 h (Mishra & Khan, 2012), while ketorolac's effect begins 10-20 min after intramuscular administration and 30-60 min after oral administration (Vadivelu et al., 2015) and reaches a peak in 1-3 min (intravenous), 30-60 min (intramuscular) and approximately 1 h (orally) (Flores-Murrieta & Granados-Soto, 1996).
To the extent of our knowledge, this systematic review and meta-analysis is the first to compare ketorolac and tramadol in the management of pain intensity after surgery of impacted third molars. Unfortunately, we could not find evidence to support ketorolac being superior to tramadol for reduction of pain intensity. However, this conclusion was only based on the findings of two studies classified at "low" risk of bias. This leads to a conclusion with moderate quality of evidence due to the low number of studies and participants included in the meta-analysis, leading to a high imprecision of the data.
Bias is a systematic error that leads to deviation from the truth in the results. It can underestimate or overestimate the true intervention effect (Higgins et al., 2019). Meta-analysis of results from biased studies can result in false positive or false negative results. The most conservative approach is to meta-analyze only data from studies with a "low" risk of bias, as performed in this study.
Among the thirteen studies that compared both drugs, ten were judged to be at "high" risk of bias. The adequate management of the domains of sequence generation and allocation concealment allows minimization of selection bias. It was already demonstrated that odds ratios were exaggerated by 41% for inadequately concealed trials and by 30% for unclearly concealed trials (adjusted for other aspects of quality) (Schulz, Chalmers, Hayes, & Altman, 1995).
The judgment of the risk of bias of these two domains was not straightforward and the great majority of the studies included in the qualitative synthesis, required contact with the authors. However, the response rate from the authors was very low. This highlights the fact that the quality of the RCTs that compared both drugs is still far from the ideal, preventing us from being confident in their findings.
For the same reason, we could not compare the total number of adverse effects of the drugs or the most frequent adverse effect for both drugs, as only two studies remained for quantitative evaluation. Among these two, only Mishra and Khan (2012) reported the presence of adverse effects of both medicines. For tramadol, weakness, tiredness, sleepiness, dizziness and giddiness were the most frequent adverse effects, while for ketorolac sleepiness, dizziness, giddiness, nausea and vomiting were reported. The quality of evidence for this outcome was judeged as low in the GRADE approach, due to very serious issues in imprecision (very low number of participants).
Contrary to our findings, another systematic review and meta-analysis that evaluated the analgesic efficacy of a single dose of tramadol in comparison with an NSAID (Isiordia-Espinoza et al., 2014) concluded that the analgesic efficacy of tramadol was lower than that of the NSAID for the management of pain after surgery on the third molars. However, this systematic review (Isiordia-Espinoza et al., 2014) meta-analyzed all included studies without taking into consideration their risk of bias. This may have resulted in a biased meta-analysis.
Although the present meta-analysis found no evidence of better analgesic efficacy for management of postoperative pain after impacted third molar surgery, the imprecision of the effect size for the intensity of pain, represented by a large confidence interval, does not allow us to conclude that a difference between the two drugs does not exist. More RCTs with a high standard and "low" risk of bias should be conducted to help clinicians in their choices of prescribing a drug for postoperative pain control after impacted third molar surgery.
Finally, the limitations of this systematic review should be reported. The two studies included in the meta-analysis used a single dose of the drugs, and the pain was evaluated at different time periods. Ideally, a study to evaluate the analgesic efficacy of drugs should focus on post-extraction pain, which should include the highest pain peak (6-8 h) (K. S. Ong & Tan, 2004). In the study of K. S. Ong and Tan (2004), the pain was assessed at 12 h, while Mishra and Khan (2012) evaluated the pain only during the first 6 h, which prevented the authors from evaluating pain at its highest level. Lastly, we should mention that the low sample size of the studies is also a limitation, as it does not allow the detection of clinically important differences.

Conclusion
This systematic review and meta-analysis of randomized clinical trials showed that there is a lack of evidence to conclude that ketorolac is better than tramadol for reduction of postoperative pain after extraction of impacted third molars, although only two randomized clinical trials were included.