Association between hydroxocobalamin administration and acute kidney injury after smoke inhalation: a multicenter retrospective study

We conducted a retrospective, multicenter study in 21 ICUs in France. All consecutive adult (≥ 18 years old) patients admitted between January 2011 and December 2017 with the final diagnosis of smoke inhalation were included in this study. Patients discharged alive at home within 24 h from admission (reflecting absence of severity) were excluded. Patient files were retrieved using our national coding for smoke inhalation injury (code CIM T599 for all patients and associated: X00·0, X 09·0, X09·9, X47·0, X47·08, X47·9, X67·0, X67·9, T58) (Additional file 1: Table S1). This academic investigator-driven study was registered at ClinicalTrials.​gov, NCT03558646. The study protocol was approved by our local research ethical committee (Comité de Protection des Personnes 2013/17NICB). The use of patient’s data was allowed in cases of death and/or when proxies could not be contacted.

The primary outcome was AKI. AKI was defined according to Kidney Disease Improving Global Outcome (KDIGO) within 7 days following admission using the serum creatinine criteria [6].

Secondary outcomes were severe AKI (severe AKI defined patients by AKI stage 2 or 3), major associated kidney events (MAKE) in the ICU, which includes death and/or renal replacement therapy (RRT), and/or persistent AKI at ICU discharge. Persistent AKI was defined as an elevated SCreat level from baseline by > 1.5-fold or > 0.3 mg/dL (26.3 μmol/L) at ICU discharge or RRT at ICU discharge and survival in the ICU.

Data were collected through a standardized case report form. Each form was manually encoded in each participating center using the initials and year of birth of the patient. Admission serum creatinine (SCreat) was used for baseline SCreat. Severe burn was defined as total body surface area (TBSA) burn ≥ 20% and/or deep TBSA burn ≥ 10% with organ support on admission (i.e., mechanical ventilation or need for vasopressors) [7]. When a fiber-optic bronchoscopy was performed, inhalation injury was classified according to fiber-optic bronchoscopy inhalation injury classification [8].

Due to the exploratory design of this study, no sample size could be calculated. Categorical variables were presented using percentages and counts, and continuous variables were presented using means and standard deviations or medians with the 25th and 75th percentiles. Categorical variables were compared using the chi-square test or Fisher’s exact test as appropriate. Continuous variables were compared using the Student t test or the Mann–Whitney U test as appropriate.

Actuarial mortality was plotted using the Kaplan–Meier estimator. Administrative censoring was applied at day 90. The impact of hydroxocobalamin administration on AKI was estimated using a multivariate logistic regression model adjusting age, comorbidities, aminoglycoside, vancomycin and iodine contrast agent utilization during hospitalization, peak value of creatinine phosphokinase, prehospital cardiac arrest, prehospital minimal GCS, severe burn, initial sequential organ failure assessment (SOFA) score without the renal item, lactate at admission, need for catecholamine infusion at admission, the SAPS2 score calculated over the first 24 h, the interaction between hydroxocobalamin administration and prehospital cardiac arrest, and the center. The impact of hydroxocobalamin administration on secondary outcomes (severe AKI, MAKE, survival) was estimated using a multivariate logistic regression model adjusting for the same confounders. An interaction term between prehospital cardiac arrest and hydroxocobalamin was also introduced in the model. Odds ratios were provided together with their 95% confidence intervals.

Missing data were handled using multivariate imputation by chained equations (mice package for R, 50 imputations) [9].

Because plasma lactate level was shown to correlate with cyanide, we performed subgroup analysis in patients with plasma lactate level at admission above median of the entire cohort and above 8 mmol/L [10, 11]. A subgroup analysis was also performed in patients with severe burns.

Two hundred eighty-eight (39%) patients developed AKI in the first week, including 186 (25.2%) patients with severe AKI. In univariate analysis, the use of hydroxocobalamin was associated with AKI, severe AKI, and RRT (Table 1). After adjusting for potential confounders, hydroxocobalamin remained associated with AKI and severe AKI and RRT (Table 2, Fig. 1 and Additional file 3: Table S3). Severe burns, aminoglycoside administration, admission plasma lactate level, chronic hypertension, and SAPS2 were also associated with AKI (Table 2).

Three hundred and thirteen (42.4%) patients met the criteria for MAKE during ICU stay. Hydroxocobalamin was not associated with the risk of MAKE in multivariate analysis (OR = 0.784, LCI = 0.456, UCI = 1.349, p = 0.379). Factors associated with MAKE in ICU are summarized in Additional file 4: Table S4.

Mortality in ICU was 32.9% (N = 243). Patients who received hydroxocobalamin had higher severity scores and higher mortality rate. Factors associated with mortality in univariate analysis are summarized in Table 3 and Fig. 2. In multivariate analysis, hydroxocobalamin was not associated with survival (OR = 1.114, LCI = 0.691, UCI = 1.797, p = 0.657). Age, severe burns, SAPS2, plasma lactate level at admission, and the center were associated with mortality in ICU (Table 4). Three hundred and ninety-two patients (53%) had admission plasma lactate level above the median (median = 3.0 [1.8–5.2] mmol/L), and 74 patients had admission plasma lactate level ≥ 8 mmol/L (median = 10.5 [8.9–12.8] mmol/L) (among 681 (92.2%) patients with plasma lactate available at admission, Additional file 5: Table S5). No association between hydroxocobalamin use and survival was observed after adjustment for confounding factors in the subgroups of patients with admission plasma lactate level above the median (OR = 0.76; LCI = 0.398, UCI = 1.451, p = 0.403) or in the group of patients with admission plasma lactate level ≥ 8 mmol/L (OR = 2.604, LCI = 0.361, UCI = 18.79, p = 0.336). In multivariate analysis, hydroxocobalamin was neither associated with survival (OR = 0.85, LCI = 0.5, UCI = 1.447, p = 0.549) among 405 (54.8%) patients with severe burns (Additional file 6: Table S6).