Reporting and analyses of sex/gender and race/ethnicity in randomized controlled trials of interventions published in the highest-ranking anesthesiology journals

We analyzed 732 RCTs published during 4 years between January 2014 and December 2017. The majority of analyzed RCTs were published in the journal Anesthesia & Analgesia (Table 1). Most of the trials were conducted in the USA (23%), followed by Canada, Denmark, UK, Republic of Korea and China (Table 1). The trials addressed large variety of indications; the most common indications that were not sex-specific were procedures related to total knee arthroplasty, cardiac surgery and low back pain (Table 1).

In 70 (10%) RCTs, there were no information about either sex/gender or race/ethnicity of participants. There were 120 (17%) trials that included only men or only women. Among the 88 trials that included only women, 84 trials analyzed gynecological conditions, while four trials analyzed condition that affects both men and women (diseases of the thyroid, fibromyalgia and migraine).

Of the 32 trials that included only men, six trials analyzed male urological conditions. In the remaining 24 trials that did not cover male-specific conditions, eight trials quoted specific differences between women and men as a motive for including only men; one trial simply said they included only men to avoid sex differences without any further specifications, while the majority of those trials – 17 (71%) of 24 – did not report reason for including only men.

In this study we found that very few anesthesiology trials seek to stratify allocation of participants on the basis of sex/gender and race/ethnicity, and that it is feasible that the outcomes reported are influenced by one or both. This demographic imbalance at baseline therefore renders outcomes vulnerable to an important and avoidable source of bias.

Along with studies about proportion of women and men, and prespecified analyses by sex, in trials published in prominent journals [1,2]; in 2018, Prakash et al. reported that only 4% of trials had 100% sex matching, and concluded that sex bias is still present in current-day clinical trials [15]. Our findings were similar to findings of Prakash et al [15]. Since half of the studies that reported sex-related analyses showed that there was significant difference between women and men in at least one outcome, it appears pertinent to plan for such analyses a priori when designing RCTs in the field of anesthesiology.

A study published in 2017 by Welch et al. found that 98% of the analyzed trials described the demographics of the participants, only 6% reported subgroup analysis based on sex/gender, and none of that trials reported any comprehensive analysis about sex or gender [16]. Trial authors used varied terminology; some trials used the term gender, while others used the term sex. However, not a single RCT provided or referenced definition of sex or gender. Likewise, none of the trials explained any intention to conduct or specify gender or sex analysis [16]. In our sample of 732 trials, we also did not find any provision or reference of a definition for sex or gender.

Sex and gender are not synonymous [17], and therefore, these terms should not be used interchangeably. Our analysis shows that some authors even use both terms in one manuscript, indicating lack of understanding of this difference.

In the field of pain research, influence of sex on experimentally induced pain and pain sensitivity is the subject of an ongoing debate [3,4]. However, it has been highlighted recently that there is insufficient attention on sex-related reporting in clinical trials about pain [5]. Due to these studies that imply that there are sex differences in the context of pain and anesthesiology, we decided to conduct this study in this particular field of research.

Insufficient reporting of sex effects was also found by Duan-Porter et al. in systematic reviews [6]; most reviews that reported sex effects used meta-regression analysis, whereas nine systematic reviews used subgroup analysis or individual-patient data meta-analysis [6].

Previous studies have also warned about lack of attention to race/ethnicity in trials [1] and that race/ethnicity was associated with important pain-related differences between participants [7–9]. Therefore, race/ethnicity should be reported as relevant information about the baseline characteristics of participants. Furthermore, attempts should be made to report outcomes based on race/ethnicity so that more insight can be gained into potential differences in the effect of interventions related to race/ethnicity.

When trials reported race/ethnicity of participants, 80% of participants were described as white or Caucasian, which could be due to lack of trialists’ effort to recruit balanced sample, but also due to certain groups’ unwillingness to participate in clinical trials [18].

There are several potential reasons why sex/gender and ethnicity/race are not considered in a large proportion of studies. This could be ascribed to oversight, lack of understanding, funding issues, conducting study in a country with less ethnic diversity and recruitment difficulties. Ethnicity could exert influence over the likelihood that individuals will give consent to participate in trials. Often times, it is probably simply lack of regard for these issues, as many analyzed trials did not even report demographics of their patients, let alone some analyses. Reporting demographics would have been simple, and for this reason, awareness about these important aspects of trial design and reporting should be increased among trialists.

A limitation of our study is that we did not aim to retrieve and analyze registered trial protocols, so we cannot make any conclusions about whether our included trials perhaps planned to do sex-specific analyses, but failed to report them subsequently. We only analyzed information available in manuscripts published in the target journals. Furthermore, in this study, we did not include all clinical trials, but only RCTs because they are considered the highest level of primary evidence in the hierarchy of medicine; therefore, these findings may not be generalizable to all types of clinical trials. Earlier analysis of sex/gender reporting in the field of anesthesiology included the first issue of eight anesthesia journals published in 2017 and any type of study that included live women and men [14]. That study showed higher number of reporting by sex/gender compared with our study, which was focused on RCTs only. Our study shows that trials are doing worse in this respect, compared with any type of studies on humans [14].

The sample of included trials was not divided in any subgroups, which also might be considered a limitation, but our findings were consistent across the trials, in other words none of them had stratified allocation regarding sex or race/ethnicity, so we believe that the major findings of our trials are generalizable to various topics. The strength of our study is that we have analyzed a large sample of trials, and reached conclusions that may be very relevant to trialists for designing and reporting their trials.

In conclusion, we found that recent trials published in the highest-ranking anesthesiology journals devote suboptimal attention to issues of sex/gender and race/ethnicity, in terms of reporting baseline characteristics of participants, matching participants by sex, and conducting relevant analyses of outcomes. Higher awareness about importance of sex/gender and race/ethnicity reporting is needed among trialists in the field of anesthesiology. Furthermore, since half of the trials that did make sex-based analyses have reported significant differences between sexes for at least one outcome, it appears pertinent to make such analyses in all trials.

L Puljak focused on study design. D Begic, C Janda-Martinac, M Vrdoljak and L Puljak dedicated to data collection and analysis, performed critical revision of the manuscript, approved final version of the manuscript and agreed to be accountable for the work. L Puljak contributed in writing first draft of the manuscript.

This study was partly based on the thesis of Clemens Janda-Martinac, which was published and defended in Croatian language at the University of Split School of Medicine in Split, Croatia. For the purpose of this manuscript, additional data collection and analyses were conducted.

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.

The study did not include clinical trial, or secondary analysis of clinical trial data. All data collected and analyzed within this study are available from the corresponding author on request.