Meaning: There is no evidence to support withholding perioperative dexamethasone in diabetic patients due to concern for infectious complications, though studies aimed at optimizing dose and understanding risk in poorly controlled diabetes are currently lacking.
Dexamethasone is a synthetic corticosteroid that reduces immune cell populations, cytokines, and specialized lipid mediators of inflammation resolution within the first few hours after administration. While reduced postoperative nausea and vomiting (PONV) is the most well-established reason for the perioperative use of dexamethasone, other potential benefits include reduced postoperative pain and total opioid consumption, less-frequent intubation-related throat pain, faster return of bowel function, less-frequent postoperative atrial fibrillation, and reduced hospital length of stay. The benefits of dexamethasone in neuro- or craniofacial surgery are also well-established, but tend to involve longer-course treatment.
Despite the broad benefits of dexamethasone, its use as a perioperative adjunct in diabetic patients remains controversial given concerns for hyperglycemia and subsequent postoperative infection. Indeed, dexamethasone increases blood glucose levels, particularly in diabetic patients, and elevated postoperative blood glucose is a well-documented risk factor for infection. A Cochrane systematic review and meta-analysis that was performed in 2017 failed to show an increased risk of postoperative infections among nondiabetic patients receiving dexamethasone, though controversy remains as to whether the same is true for diabetics in whom the baseline risk of infection and postoperative morbidity is increased.
Numerous trials investigating the potential benefits of perioperative dexamethasone have been published, a subset of which have explicitly included patients with diabetes. However, there have been no focused attempts to aggregate the data from randomized controlled trials investigating the use of perioperative dexamethasone in diabetics. Given that infection risk tends to be the principal reason for withholding perioperative dexamethasone in diabetic patients, this study aimed to evaluate the incidence of infections in diabetic patients receiving preoperative dexamethasone compared to controls. Secondary aims were to assess the relationship between perioperative dexamethasone utilization in diabetic patients and the occurrence of adverse events as an aggregate end point. The hypothesis of this study was that there is no difference in infectious outcomes between dexamethasone-treated patients and controls.
METHODS
This systematic review was conducted using a web-based collaboration software platform that streamlined the production of systematic reviews via optimization of import, abstract and full-text screening, and export of both published and unpublished studies (Covidence systematic review software, Veritas Health Innovation). This allowed for analysis consistent with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines and the PRISMA statements. The review was prospectively registered on the International Prospective Register of Systematic Reviews (PROSPERO) (registration ID# 380323) and adheres to the applicable PRISMA guidelines.
Search Strategy
The search strategy aimed to identify both published and unpublished studies using a 3-step search strategy. First, an initial limited search of MEDLINE (PubMed) and Google Scholar was performed to identify articles on the topic. Second, the text words contained in the titles and abstracts of relevant articles and the MeSH (Medical Subject Headings) terms used to describe them were used to develop a full search strategy in the PubMed, Embase, and Cochrane databases. Search terms included "((dexamethasone [Title/Abstract]) OR (decadron [Title/Abstract])) AND (surgery[Title/Abstract] OR surgical[Title/Abstract] OR operation [Title/Abstract] OR operative [Title/Abstract] OR 'post-op'[Title/Abstract]) AND (clinicaltrial[Filter] OR randomizedcontrolledtrial[Filter]))" for PubMed, "(Dexamethasone:ti,ab OR decadron:ti,ab) AND (surgery:ti,ab OR surgical:ti,ab OR operation:ti,ab OR operative:ti,ab OR 'post-op':ti,ab) AND ('clinical trial'/de OR 'randomized controlled trial'/de)" for Embase, and "((dexamethasone OR decadron) AND (surgery OR surgical OR operation OR operative OR 'post-op'))):ti,ab,kw" for Cochrane Databases, respectively. This search strategy was adapted for each included database and information source and performed on November 22, 2022. Finally, references to systematic reviews on the same or similar topics were examined. Results were collected and deduplicated using EndNote 20 (Clarivate). They were then uploaded to a web-based collaboration software platform that streamlined the production of systematic reviews via optimization of import, abstract and full-text screening, and export of both published and unpublished studies (Covidence systematic review software, Veritas Health Innovation).
Screening and Assessment of Quality
Details regarding the search methods for study selection and reasons for exclusion can be found in Figure 1. Studies were initially screened by 2 independent reviewers (B.G. and M.R.). Inclusion criteria included full-text articles published in English, studies that compared intravenous (IV) dexamethasone injections to "true placebo" (ie, nonactive placebo such as normal saline or sham) injections, and studies that administered the dexamethasone intervention on the day of operation. There were no date restrictions or subject age restrictions. Exclusion criteria included studies that were not published in English, lacked an accessible abstract or full-text, or included nonhuman or cadaver subjects. Additionally, studies that did not utilize a "true placebo" comparative group (ie, injection of an alternative corticosteroid, pain, or nausea medication) were excluded, as well as studies that explicitly excluded diabetic patients.
The focus of early screening was to ensure that true randomization was used for assignment of participants to treatment groups, a true (nonactive) placebo was used, and adequate treatment concealment and blinding were used. Full-text review was performed by 2 independent reviewers (I.J. and M.L.). For full-text review, studies that dosed patients for more than 72 hours (eg, extended dexamethasone tapers) were excluded to ensure the results were representative of the effects of short-term dosing. Additionally, full-text review was conducted to ensure diabetic patients were included in both treatment arms. Studies that alluded to the inclusion of diabetic patients (eg, insulin treatment was mentioned in article) but did not explicitly state that diabetic patients were included and did not provide information about diabetic patients in their population demographics were excluded. All potentially eligible clinical trial registries and protocols were cross-referenced using Google Scholar and PubMed (ie, by searching these databases for the listed trial registry number) to identify shared or published data meeting study inclusion criteria. Finally, in keeping with the aim of understanding the clinical effects of perioperative dexamethasone, studies that did not include at least 1 clinical outcome (eg, those looking exclusively at blood glucose response) were excluded. The finalized plan was well-aligned with our original PROSPERO submission, although with further refined inclusion/exclusion criteria and specific primary and secondary aims based on the available data. A third reviewer (N.H.) was available to serve as a tiebreaker to resolve discrepancies, though was not needed.
All studies, regardless of their methodological quality, underwent data extraction and synthesis when possible. Methodological quality was performed by 2 independent authors (I.J. and M.L.) using an National Institutes of Health quality assessment tool, which was developed to help highlight concepts that are key to a study's internal validity. Details regarding the methodological quality assessment can be found in Supplemental Digital Content 1, Appendix A, https://links.lww.com/AA/E819.
Data Extraction
Data were extracted by 3 independent reviewers (B.G., I.J., and M.L.). In addition to extracting quantitative values necessary to perform the meta-analyses, the type of surgery performed, number of patients with and without diabetes per study arm, restrictions on the type of diabetic patients allowed to participate, and dexamethasone dosing schedule were obtained. Secondary outcomes were extracted from all studies, regardless of whether they contributed to the qualitative synthesis. Given that many studies reported the inclusion of, but not the explicit effect on diabetic patients, qualitative extraction of overall results (not specific to diabetics) was performed, with findings organized into those that were superior, noninferior, and inferior. Qualitative extraction on the diabetic patient subpopulation was also extracted when available.
In an attempt to obtain the greatest number of data points for meta-analysis, authors were contacted to request additional data for all included studies. Several different outcomes were considered for potential analysis including opioid consumption and length of stay. However, only outcomes for which 3 or more studies could be reasonably compared were pursued, which left only the primary outcome of surgical site infections and the secondary outcome of composite adverse events for final quantitative analysis. Elements of the composite were selected based on consideration of the available data and with the aim of approximating treatment adverse events, as defined as harmful and negative outcomes that are at least possibly related to the dexamethasone intervention. All included and excluded composite elements were reported to ensure outcome reporting transparency.
Data Synthesis
Quantitative assessments were performed exclusively on the diabetic subpopulation. Meta-analyses on study outcomes were performed using inverse variance with random effects. Heterogeneity was assessed via standard χ and I tests, where I >50% indicated high heterogeneity. A funnel plot was not utilized to assess for publication bias as there were fewer than 10 studies included in quantitative analysis. We defined significance as P < .05 and performed all statistical analyses using STATA (version 17.0; StataCorp).
RESULTS
A total of 16 randomized controlled studies qualified for inclusion (Table). All authors were contacted for additional study details related to the diabetic subpopulation, 5 of which responded and 2 of which provided data that could be included in the meta-analysis.
Eight thousand seven hundred fifty-four individuals were randomly assigned to dexamethasone treatment, while 8358 received a placebo. Among the diabetic patient subset (n = 2592), 1315 received dexamethasone and 1277 received placebo. Of the 2592 diabetic patients, 2232 came from 3 large studies included in both the qualitative and quantitative analysis. Only 1 study excluded insulin-dependent diabetics, 5 studies excluded uncontrolled diabetics, and 1 study excluded both insulin-dependent and uncontrolled diabetics. Uncontrolled diabetes was defined differently among authors and often without explicit numerical thresholds, although 2 studies set the exclusion limit as glycated hemoglobin test (hemoglobin A1c) > 9.0%.
Considerable heterogeneity was observed in dosing, timing, and number of dexamethasone administrations. Dosages ranged from 4 mg to a maximum of 100 mg. Dexamethasone was typically administered after induction and before incision, although 4 studies administered dexamethasone twice and 1 study administered dexamethasone 3 times over 48 hours. A wide range of surgery types were included, comprising both cardiac and noncardiac surgeries. Specifically, Dieleman et al, Hanafy et al, and Murphy et al studied dexamethasone in cardiac surgery, with Dieleman et al and Hanafy et al using high-dose regimens. Notably, among these 3 cardiac studies, only Dieleman's data were available for quantitative analysis.
Overall, methodological quality was rated as "good" for the majority of included studies (Supplemental Digital Content 1, Appendix A, https://links.lww.com/AA/E819). No studies were rated as "poor" and, of the 4 studies rated as "fair," none contributed to the quantitative synthesis.
Quantitative Analysis on Risk of Surgical Site Infections and Composite Complications
Heterogeneity in the quantitative data for surgical site infections was minimal. Both the Perioperative Administration of Dexamethasone and Infection (PADDI) trial first published by Corcoran et al in 2021 and the trial published by Dieleman et al provided data on 30-day events. Gasbjerg et al provided data up to 90 days and the Perioperative Administration of Dexamethasone And Blood Glucose (PADDAG) trial which was first published by Corcoran in 2020 provided data up to 72 hours or at final hospital discharge, whichever was longer. No differences in log odds ratio (LOR) of surgical site infections was observed in the dexamethasone-treated group compared to the control group on meta-analysis of the 4 studies meeting criteria for inclusion (LOR, -0.10, 95% confidence interval [CI], -0.64 to 0.44, P = .72) with acceptable heterogeneity (I = 25.92%) (Figure 2).
There was extensive heterogeneity in the quantitative data for the composite outcome, which was minimized in part by manually selecting outcomes for inclusion. To ensure reporting transparency, Supplemental Digital Content 2, Supplemental Table 1, https://links.lww.com/AA/E820 lists all available outcomes, as well as those that were excluded from the quantitative analysis. Lower odds of composite adverse events were observed in the dexamethasone-treated group on meta-analysis of the 4 studies meeting criteria for inclusion (LOR, -0.33, 95% CI, -0.62 to -0.05, P = .02) with acceptable heterogeneity (I = 0.00%) (Figure 3).
Qualitative Overview of Clinical Outcomes
In the qualitative review of overall study outcomes including both diabetic and nondiabetic patients, the dexamethasone intervention consistently demonstrated positive-to-neutral benefits while maintaining a robust safety profile. Dexamethasone was found to reduce the risk of PONV, postoperative pain, and length of stay in the Intensive care unit (ICU) and hospital. Notably, 1 study reported a significant reduction in major adverse cardiac events, while another demonstrated decreased postoperative infection risk and respiratory complications.
Among the 16 studies, only 4 explicitly commented on clinical outcomes specific to diabetic patients. These results were consistent with those reported for the overall population, with the outcomes comparing dexamethasone to placebo for diabetic individuals being statistically insignificant across all assessed outcomes.
DISCUSSION
This study of 2592 diabetic patients randomized to receive either dexamethasone or placebo across a variety of different surgical specialties found no increased risk of surgical site infections and lower risk of total complications among patients who received dexamethasone. This finding is notable, as previous attempts to assess infection risk in diabetics failed due to limited available data. Moreover, a qualitative assessment of overall trial data found that dexamethasone was superior or similar across all clinical outcomes when compared to placebo, despite increasing blood glucose levels. While strong conclusions are limited by available data and study heterogeneity, this systematic review and meta-analysis emphasizes the lack of data to support withholding dexamethasone in diabetic patients due to concerns related to infection risk.
A statistically significant reduction in composite major adverse events is perhaps the most unexpected and noteworthy finding of this study that sets this study apart from those that have been published previously. However, there are several methodological and data considerations that should temper strong conclusions. First and foremost, there was a disproportionate contribution of data from a single, well-designed trial published by Dieleman et al, which was conducted on patients undergoing cardiopulmonary bypass. There are more than 30 years of high-quality data linking inflammatory changes to cardiopulmonary bypass. Indeed, unlike other surgery types, cardiopulmonary bypass results in pathophysiologic changes that are very similar to those seen in systemic inflammatory response syndrome. Nevertheless, it is also true that increased surgery-associated inflammation is linked to surgical trauma and various comorbidities, particularly diabetes. Moreover, all of the other studies contributing to the composite outcome supported dexamethasone having a neutral-to-positive overall effect with respect to major adverse events. Thus, the observed reduction in composite major adverse events should be interpreted similarly to the finding that there was no increased risk of surgical site infections. Specifically, this study reports findings that are contrary to conventional wisdom and highlights the need for further investigation.
A composite outcome was used to provide a more comprehensive assessment of dexamethasone and avoid overreliance on any single factor that may not adequately describe the individual patient experience. The principal limitation of any composite outcome is that all events are given the same importance, despite having differential importance to patients, their families, and clinicians. While composite outcomes can be particularly useful when both benefits and harms are considered, this study aimed to make a concerted effort to ensure that the harmful effects of dexamethasone were not overshadowed in the composite by well-established treatment benefits such as PONV. Specifically, the composite outcome minimizes the possibility of a type II error at the cost of a more frequent type I error. In other words, the composite outcome in this study maximized the chances of finding dexamethasone-associated adverse events, yet still failed to find any significant increased risks.
While most of the studies did not explicitly exclude patients with uncontrolled diabetes, several studies, including some of the largest and best designed, did have exclusion criteria in place for patients with poorly controlled diabetes. Importantly, 3 of the 5 studies that contributed data to the quantitative analysis excluded uncontrolled diabetics. Thus, while dexamethasone likely has positive-to-neutral effects in diabetic patients overall, the risk in poorly controlled diabetics remains largely unknown, particularly for patients with A1c >9.0%. However, it does appear that insulin-dependent diabetes alone is not a reason to withhold perioperative dexamethasone. Only 2 of the 16 studies explicitly excluded insulin-dependent diabetics, representing <1% of the total diabetic patients identified in this study.
The qualitative outcomes, while unable to provide direct information about the effect of dexamethasone in diabetic patients, bolster the findings from the quantitative analysis. Specifically, dexamethasone did not lead to detrimental effects across a wide array of clinical outcomes. Furthermore, dexamethasone likely provides benefits beyond PONV, as it has been shown to improve postoperative pain, length of stay, and patient perception of their own recovery in certain surgical contexts. Hyperglycemia, while explicitly reported as an outcome of interest in 2 of the included studies, did not translate into worse clinical outcomes.
This study has several limitations. The most prominent limitation was the lack of available data for quantitative analysis. Published reports to date have generally failed to report quantitative data for the diabetic subpopulation, and, despite attempts to contact all 16 study authors, only 2 were able to provide additional data on the diabetic subpopulation that could be included in the quantitative analysis. This poor response rate raises concerns regarding the generalizability and robustness of the findings. Nevertheless, the 2 fulfilled data requests yielded unpublished data on 900 diabetic patients, which is more than 4 times the total number of diabetic patients in the 11 studies not included in the quantitative assessment. This not only highlights the disproportionate influence that a small group of studies has in understanding the safety and efficacy of perioperative dexamethasone in diabetics, but also partially addresses concerns related to the lack of available data for quantitative analysis and any associated concern that the data may not be representative of the literature overall. Indeed, this study was largely successful in performing quantitative assessment on the vast majority of diabetic patients participating in randomized, placebo-controlled trials, and the inclusion of the unanalyzed patients likely would not have meaningfully changed the results. The focus explicitly on dexamethasone, while not necessarily a limitation, should also be noted. Large, well-designed trials using alternative steroids have been published. Finally, heterogeneity is a major limitation. Indeed, the only major exclusion criteria imposed to safeguard against heterogeneity was the focus exclusively on short-course (<72 hours) perioperative dexamethasone. This was done to ensure that the well-established, deleterious long-term effects of corticosteroids were not conflated with their short-term effects when used as perioperative adjuncts aimed at blunting the acute inflammatory surgical stress response. While limitations due to study heterogeneity are largely expected with any attempt to aggregate the total available literature, the heterogeneity should not be overlooked. Indeed, there are important questions that remain unanswered, particularly with respect to dosing and the specific surgical context.
Based on the available data, perioperative dexamethasone may be given safely to diabetic patients without an increased risk of infectious complications. As such, the narrative surrounding the use of dexamethasone in diabetics should shift away from assumptions that reinforce unfound concerns about its appropriateness, and instead focus on exploring the validity of such concerns. Prospective investigations aimed at optimizing dose, frequency, and timing are needed, as well as studies aimed explicitly at exploring the use of dexamethasone in patients with poorly controlled diabetes.
ACKNOWLEDGMENTS
The authors would like to acknowledge with gratitude Drs Jan M. Dieleman and Kasper S. Gasbjerg as well as their respective teams for their collaborative efforts, which were integral to the completion of this study.
Contribution: This author helped in project idea, project design, literature review, data extraction, and article authorship.
Name: Michael A. LoBasso, MD.
Contribution: This author helped in project idea, project design, literature review, data extraction, and article authorship.
Name: Julian Wier, MD.
Contribution: This author helped in statistical analysis, figure generation, and article revisions.
Name: Brandon S. Gettleman, BS.
Contribution: This author helped in the literature review, data extraction, and article revisions.
Name: Mary K. Richardson, BS.
Contribution: This author helped in the literature review, data interpretation, and article revisions.
Name: Christina E. Ratto, MD.
Contribution: This author helped in data interpretation and article revisions.
Name: Jay R. Lieberman, MD.
Contribution: This author helped in data interpretation and article revisions.
Name: Nathanael D. Heckmann, MD.
Contribution: This author helped in project idea, project design, literature review, data extraction, and article authorship.