Hydrocortisone with Fludrocortisone for Adult Patients with Septic Shock: A Systematic Review and Meta-Analysis

Background : The efficacy of corticosteroids for patients with septic shock remains controversial. Therefore, evaluating a specific corticosteroid treatment could solve the conflicting results of past studies. We performed this systematic review and meta-analysis to elucidate the beneficial effects of a dual corticosteroid treatment regime involving administration of both hydrocortisone and fludrocortisone for adult patients with septic shock on mortality. Methods : We searched the Medline, Cochrane CENTRAL, and ICHUSHI databases from inception to April 2019. We included peer-reviewed randomized controlled trials that compared the use of both hydrocortisone and fludrocortisone with either corticosteroid-free or hydrocortisone-only treatments on adult patients with septic shock. Three researchers independently reviewed the studies for design, eligibility criteria of patients, dose of each corticosteroid, and duration of corticosteroid therapy following PRISMA guidelines. The random-effects models and Grading of Recommendations Assessment, Development and Evaluation were applied to rate the quality of the evidence. Primary outcome was 28-day mortality, and secondary outcomes were in-hospital mortality, long-term mortality, shock reversal, and adverse events. Results : Among the four studies eligible for data synthesis, we included 2050 patients from three studies for quantitative synthesis. All studies used similar regimens (50 mg intravenous bolus of hydrocortisone every 6 h and 50 μg tablets of fludrocortisone once daily for 7 days without tapering). The 28-day mortality rate was reduced after dual corticosteroid treatment [risk ratio, 0.88; 95% confidence intervals (CI), 0.78–0.99]. The heterogeneity between the studies was low (I² = 0%). Furthermore, the certainty in the effect estimates was high. Patients that underwent dual corticosteroid treatment had lower long-term mortality rates (RR, 0.90, 95% CI, 0.83–0.98) and higher rate of shock

Conclusions : The available evidence suggests that a combination of fludrocortisone and hydrocortisone is more effective than adjunctive therapy and may be recommended for patients with septic shock.
Some reasons for these contradictory findings include differences in patient populations and the variation in corticosteroid treatments. Low-risk-of-bias (RoB) RCTs recruited only patients with septic shock and investigated mortalities as their primary outcomes; however, their definition of refractory shock differed in the doses of vasopressors required [15,19,26,27,33]. The durations, amounts, and type of corticosteroids also differed, due to which optimal corticosteroid treatments remain unclear [26,27,33,34]. Moreover, most systematic reviews involved heterogeneous studies, and they examined specific populations or particular corticosteroid therapies only through subgroup analyses [5,6,8].
Among the various corticosteroid treatments, the dual treatment with hydrocortisone and fludrocortisone for septic shock has shown promising results; one large RCT in 2018 found 4 that administration of both hydrocortisone and fludrocortisone reduced the 90-day mortality of patients with septic shock [27]. Hydrocortisone, the name for cortisol hormone as a medication, has both glucocorticoid and mineralocorticoid activities; whereas fludrocortisone, a synthetic corticosteroid, possess very potent mineralocorticoid activity [35][36][37][38]. While hydrocortisone has been extensively examined in sepsis, fludrocortisone has been used for patients with aldosterone deficiency, and the dual administration of these two medications is recommended for some patients with primary adrenal insufficiency [39,40]. Considering that patients with septic shock have been found to have unexpectedly low aldosterone levels due to HPA axis abnormalities [41], dual treatment with hydrocortisone and fludrocortisone should be further validated as a type of corticosteroid treatment for septic shock.
Accordingly, we conducted a systematic review and meta-analysis to identify beneficial effects of the dual treatment with hydrocortisone and fludrocortisone for patients with septic shock, when compared to treatment with placebo or hydrocortisone alone. We particularly examined clinically significant outcomes such as mortality, vasopressor withdrawal, and adverse events (AEs).

Methods
We report our findings in this review in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analyses Guidelines. The review protocol has been registered with PROSPERO (reference CRD42019139069). Substantive deviations from the published protocol are highlighted with accompanying explanations.

Search Strategy
Three databases were searched in April 2019: Medline, Cochrane CENTRAL, and ICHUSHI.
Our search strategy included Medical Subject Headings search terms and keywords.

Study Selection
We included RCTs that fulfilled the following criteria: 1) full-text publication in peerreviewed journals in English; 2) inclusion of adult patients diagnosed with septic shock, according to accepted criteria; and 3) studies comparing the use of both hydrocortisone and fludrocortisone with a corticosteroid-free or hydrocortisone-only comparator group.
We excluded studies that did not meet those three criteria.
After implementation of the search strategy, reviewers performed screenings in duplicate in two stages. First, two independent reviewers (IN and MT) assessed titles and abstracts of the retrieved literature to identify potentially relevant articles. Then, the reviewers obtained full texts of articles for those possibly eligible for further review and independently assessed them. We recorded reasons for exclusion at the full article review stage. Disagreements between the two reviewers were resolved through discussion among them and a third reviewer (RY) until consensus was achieved.

Data Extraction and Quality Assessment
The three reviewers extracted the data independently and in duplicate using predefined data abstraction forms. Any discrepancies were resolved by group discussion. The extracted data for each study included the first author, year of publication, study design, number of study sites, number of patients, eligibility criteria for patients, demographic 6 data, type of corticosteroids given to patients, dose of each corticosteroid, and duration of corticosteroid therapy. In addition, researchers also extracted the following clinical outcomes: 28-day mortality as short-term mortality, in-hospital mortality, long-term mortality (longer than 90 days), shock reversal, and any AEs related to the corticosteroid therapy.
RoB was then evaluated, independently and in duplicate, for each outcome of individual studies using the Cochrane risk of bias assessment tool, which assesses randomization, allocation concealment, blinding of the study participants and personnel, blinding of the outcome assessments, incomplete outcome data, selective outcome reporting, and other potential sources of bias [42]. Disagreements for RoB were resolved by discussion.

Data Synthesis and Analysis
The primary outcome was 28-day mortality. The secondary outcomes included in-hospital mortality, long-term mortality (longer than 90 days), shock reversal at day 28 defined as vasopressor withdrawal at day 28, vasopressor-free days up to day 28, and the prevalence of AEs such as superinfection, gastrointestinal (GI) bleeding, hyperglycemia, hypernatremia, and any other events related to corticosteroid treatment. Subgroup analyses were prespecified according to duration of treatment and dose of hydrocortisone and/or fludrocortisone. Sensitivity analyses were performed by repeating meta-analyses, in which we defined control groups as either patients not treated with corticosteroid (placebo) or patients treated only with hydrocortisone (hydrocortisone-only).
We used the Review Manager software (RevMan, Version 5.3: The Cochrane Collaboration, Copenhagen, Denmark) to conduct the meta-analyses. We used a random-effects model to calculate pooled effect sizes and corresponding 95% confidence intervals (CIs) for outcomes in which nonidentical effects of intervention were estimated between studies, whereas we used a fixed-effects model for shock reversal at day 28, in which the effect of intervention was estimated to be in the same direction based on previous studies [1][2][3][4][5][6][7][8].
We presented results as risk ratios (RRs) for dichotomous outcomes and as mean differences (MDs) for continuous outcomes. Extracted data provided as medians with interquartile ranges were changed into means with standard deviation [43]. Heterogeneity between studies was assessed using χ² test for homogeneity, I² statistic, and visual inspection of the forest plots. We considered the heterogeneity of I² values < 25% as low, those between 25 and 50% as moderate, and those > 50% as high. Publication bias was also evaluated by a funnel plot.
The overall certainty of evidence for each outcome was assessed and rated as high, moderate, low, or very low using the Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach based on study limitations, inconsistency, indirectness, and publication bias (44). Disagreements for GRADE assessment were resolved by group discussion.

Study Selection
We identified 94 articles through the Medline search, 35 through the Cochrane CENTRAL search, and two articles through the ICHUSHI search. Eleven studies were considered potentially eligible; then, we excluded seven after the full-text screening (two non-fulltext articles, three non-RCTs, one non-English study, and one study with a different intervention). Among four studies eligible for data synthesis, we included a total of 2050 patients from three studies for our quantitative synthesis (one study did not report any primary or secondary outcome data; Fig. 1).

Description of Eligible Studies
Supplemental Table 3 presents a detailed description of the eligible studies [15,27,45,46]. All studies were conducted at multiple centers, and their eligibility criteria for patients included the requirement of vasopressors to define septic shock. All studies also used the same intervention protocol in terms of type, dose, and duration of the corticosteroid therapy; hydrocortisone was administered as a 50 mg intravenous bolus every 6 h, and fludrocortisone as a 50 µg tablet through a nasogastric tube once daily for 7 days (without tapering). Three of the eligible studies used placebo for the control group [15,27,46], and the other used the hydrocortisone alone therapy (a 50 mg intravenous bolus every 6 h) for the control group [45]. One of the eligible studies reported only hematological and biochemical outcomes [46] obtained from the same population of another included study [15]; therefore, we did not include it in the quantitative synthesis.

Primary Outcome
Two studies reported 28-day mortalities [15,27], and our analyses showed the 28-day mortality rate was lower in the dual corticosteroid treatment patients than in the controls, and the RR of 28-day mortality was 0.88 (95% CI = 0.78-0.99) with low heterogenicity (I² = 0%, p = 0.79; Fig. 2). Publication bias was not estimated using the funnel plot because only two studies were included in the meta-analysis. Prespecified subgroup analysis was not performed regarding the primary outcome because the duration and dose of the corticosteroid treatments were identical between the two studies. Furthermore, sensitivity analysis was not performed on primary outcome because the control groups of both studies were cortisol-free (placebo) populations.
Shock reversal at day 28 was reported in two included studies, whereas vasopressor-free days of up to day 28 were reported in only one included study. Patients in the dual corticosteroid treatment group had a higher rate of shock reversal (RR = 1.06, 95% CI, 1.01-1.12) [15,27] and more vasopressor-free days (MD = 2.0 days, 95% CI, 0.8-3.2 days) [27] than patients in the control group (Fig. 3).
Meta-analyses on AEs by corticosteroid treatment revealed that risks of superinfection [15,27,45], GI bleeding [15,27], and psychiatric disorder [15] were similar between the dual corticosteroid treatment and the control groups (Supplemental Fig. 2). The incidence of hyperglycemia was higher in patients treated with both hydrocortisone and fludrocortisone, although only one study reported hyperglycemia as an AE [27] (Supplemental Fig. 2).
We did not perform prespecified subgroup analyses on secondary outcomes because the duration and dose of corticosteroid treatments were identical among all included studies.
Sensitivity analyses were performed on in-hospital mortality, long-term mortality, and superinfection because the control groups comprised both corticosteroid-free population and hydrocortisone-only population. Meta-analyses comparing the dual corticosteroid treatment with placebo showed that in-hospital and long-term mortalities were lower in the dual corticosteroid group than in the placebo group (RR = 0.88, 95% CI, 0.80-0.98 and RR = 0.89, 95% CI, 0.81-0.97, respectively), whereas AEs were comparable between the groups [15,27] (Supplemental Fig. 3). A sensitivity analysis comparing the dual corticosteroid treatment with hydrocortisone-only therapy revealed a higher superinfection rate in the dual corticosteroid treatment group (RR = 1.54, 95% CI = 1.05-2.26), whereas in-hospital and long-term mortalities were comparable between the groups [45] (Supplemental Fig. 4). Table 1 summarizes results according to control groups.

Risk of Bias and Summary of Findings
A summary of the RoB analysis results is shown in Supplemental Fig. 5. The RoB for mortality was evaluated as "low" for all components of the Cochrane risk of bias assessment, with the exception of the "unclear risk" at "selective outcome reporting" in the study by Annane et al [15] in 2002.
The quality of evidence for each outcome is summarized in Table 2, in which we evaluated the certainty of effect estimates using the GRADE approach. The 28-day and long-term mortalities were significantly reduced by the dual administration of both hydrocortisone and fludrocortisone with high certainty. Among the AEs associated with corticosteroid treatment, the incidence of hyperglycemia was increased by the dual corticosteroid treatment with high certainty.

Discussion
We conducted the current systematic review and meta-analysis to evaluate the effects of dual corticosteroid treatment in terms of clinical outcomes in patients with septic shock. While previous systematic reviews had shown conflicting results, we found a reduced mortality with high certainty: 28-day, in-hospital, and long-term (later than 90 days) mortalities were reduced by the treatment with both hydrocortisone and fludrocortisone.
The main difference between the current study and other meta-analyses is the fact that we examined only the effects of the dual corticosteroid treatment for septic shock. The idea behind the addition of fludrocortisone to hydrocortisone, used as glucocorticoid replacement therapy in patients with adrenal insufficiency, is to enhance the mineralocorticoid activity [35][36][37]. The biological activity of mineralocorticoids is mediated by the mineralocorticoid receptor (MR) [35,47], which exists in various organs, such as the kidneys, cardiovascular, immune, and central nervous systems [48,49]. Animal studies found an association between sepsis and the downregulation of the MR in endothelial cells [50], and mineralocorticoid supplementation lowered IL-6 levels, hastened shock reversal, and improved survival [51,52]. Some clinical studies also revealed inappropriately low aldosterone levels in patients with septic shock, suggesting an impaired adrenal synthesis of aldosterone, that might be associated with increased mortality [41,53]. Although the direct effect of mineralocorticoids in septic patients has not been fully elucidated, our results support a recommendation for the dual administration of hydrocortisone and fludrocortisone to patients with septic shock.
The meta-analyses on the secondary outcomes found that the incidence of AEs was not increased by the dual corticosteroid treatment, except that for hyperglycemia, which is consistent with the results of a systematic review examining all types of corticosteroid therapies for sepsis [5]. In that study, the risks for hyperglycemia, hypernatremia, and neuromuscular weakness were similarly increased by the corticosteroid administration, while the incidence of superinfections, GI bleeding, and psychiatric disorders remained similar to those in control patients. Although there may be additional adverse event risks by adding fludrocortisone on hydrocortisone, our sensitivity analysis comparing the dual corticosteroid treatment with hydrocortisone-only therapy identified only higher superinfection rates in the dual corticosteroid treatment group. Considering that MRs are expressed in monocytes and macrophages that undergo a pro-inflammatory polarization in response to mineralocorticoids [54,55], pathophysiological immunomodulatory changes by the additional mineralocorticoid administration should be further examined.
The recommendation of fludrocortisone use in a previous version of the Surviving Sepsis Campaign guidelines was an optional addition to low-dose hydrocortisone [56], and it got removed from the most recent guidelines in 2016 [28]. Two recent systematic reviews analyzed the dual administration of hydrocortisone and fludrocortisone in subgroup analyses that evaluated heterogenicity in types of corticosteroid treatments, and did not find a credible effect of the specific type of corticosteroid treatment [5,8]. However, these analyses did not examine the direct association between the dual corticosteroid treatment and clinical outcomes, and based on our results the dual corticosteroid treatment (adding fludrocortisone to hydrocortisone) should be considered more than just an adjunctive therapy.
The results in this study must be interpreted within the context of the study design. We found only four eligible studies and included only two in the meta-analysis for the primary outcome [15,27], in part because the additional fludrocortisone has not been extensively examined and because we considered only RCTs. However, our search strategy used a wide variety of search terms and the eligibility criteria were wide enough to capture an article by Laviolle et al [46] that was not included in the recent systematic reviews [5][6][7][8].
Also, the primary outcome had no heterogenicity and we evaluated the RoBs of the studies included as "low" using the Cochrane risk of bias assessment tool.
Another limitation of this study is the fact that the control group in the meta-analyses consisted of both corticosteroid-free and hydrocortisone-only populations, which may hamper the interpretation of our results. Although some secondary outcomes such as inhospital and long-term mortalities and superinfection differed in the sensitivity analyses according to the definitions of the comparator group, the reduced 28-day mortality by the dual corticosteroid therapy resulted only from the comparison with placebo administration. Considering that the current study aimed to find beneficial effects of the dual administration of hydrocortisone and fludrocortisone (rather than effects directly related to only fludrocortisone), we believe that this specific corticosteroid therapy can be recommended for patients with septic shock with high quality of evidence.
Moreover, all eligible studies used the same intervention protocol in terms of type, dose, 13 and duration of the corticosteroid therapy. Although different doses of hydrocortisone and/or fludrocortisone might affect the results, the doses used were those standardly used in the replacement therapy for primary adrenal insufficiency [40,57]. Given that a study on different hydrocortisone-only treatment durations for septic shock revealed no differences in outcomes between 3-day and 7-day regimens [58], a shorter regimen of the dual corticosteroid treatment may be effective and should be investigated.

Conclusions
This systematic review and meta-analysis clarified that hydrocortisone and fludrocortisone treatment reduce the 28-day mortality of patients with septic shock with minimum risk of AEs. Therefore, we recommend this specific corticosteroid treatment for septic shock with high certainty in effect estimates. The pathophysiological mechanisms of the additional fludrocortisone and the duration of treatment should be further studied.

Consent for publication
Not applicable

Availability of data and materials
All data generated or analyzed during this study are included in this published article.

Competing interests
Dr. Fujishima has received personal fees from Asahi Kasei Japan and Takeda