Clinical outcomes of intravenous immunoglobulin therapy in COVID-19 related acute respiratory distress syndrome: a retrospective cohort study

We retrospectively analyzed electronic health record data of patients admitted to ICUs at Hazm Mebaireek General Hospital, Qatar, between March 7, 2020 and September 9, 2020. The Medical Research Center (MRC) at Hamad Medical Corporation, Qatar, approved this study and waived the requirement for informed consent (protocol ID MRC-01-20-853). Inclusion criteria were: COVID-19 positivity as determined by reverse-transcription polymerase chain reaction (RT-PCR) of nasopharyngeal swabs, age above 18 years, respiratory failure requiring invasive mechanical ventilation, and moderate-to-severe ARDS (PaO₂/FiO₂ ≤ 200 mm Hg) as defined by the Berlin criteria [11]. Patients who received IVIG for indications other than COVID-19 related ARDS or had cardiac arrest before ICU admission were excluded. The study population was divided into two groups based on receiving routine care or IVIG plus routine care during their ICU stay. All patients received steroid and antiviral therapy as per the hospital's policy for COVID-19 management unless contraindicated. Prophylactic-, intermediate- or therapeutic-dose of anticoagulation was administered based on the patient's risk stratification as low-, intermediate- or high-risk for thromboembolic disease (anticoagulation protocol in Additional file 1: Fig. S1). IVIG was given to patients at the treating physician's discretion if there was a persistent increase in oxygen requirement and worsening laboratory parameters (C-reactive protein and serum ferritin level), suggestive of disease progression. The IVIG treatment group received a minimum one dose of 0.4 g/kg of IVIG. Further doses of IVIG were given on consecutive days, to a maximum of 5 doses, based on the treating physician's decision. Patients were closely monitored for immediate adverse effects like skin rash, arrhythmias, hypotension, and anaphylaxis.

Baseline data were collected at the time of ICU admission. Information collected were demographic characteristics, comorbid conditions, blood test results, PaO₂/FiO₂ ratio, vasopressor use, sequential organ failure assessment (SOFA) score, immune-modulating, and antiviral drugs received.

The primary outcome studied was all-cause ICU mortality. Secondary outcomes included ventilator-free days and ICU-free days at day-28, and incidence of acute kidney injury (AKI) defined by the Kidney Disease Improving Global Outcomes (KDIGO) criteria as an increase in serum creatinine level by ≥ 0.3 mg/dl (≥ 26.5 μmol/L) within 48 h or increase in serum creatinine to ≥ 1.5 times baseline [12]. Sub-group analysis was done for patients based on PaO₂/FiO₂ ratio, serum ferritin level, time from ICU admission to IVIG therapy, and doses of IVIG received.

We used propensity score matching to account for the non-random treatment allocation and to adjust for confounders. Propensity scores were generated based on age, sex, body mass index (BMI), comorbidities, SOFA score, PaO₂/FiO₂ ratio, baseline laboratory results, and other drug therapies. The propensity score overlap was assessed graphically. One-to-one matching with a caliper width of 0.2 of the standard deviation of the logit of the propensity score and no replacement was used to select a matching control group [13]. The post-matching covariate balance was assessed using the standardized differences. The primary outcome was compared using survival analysis. We included the treatment with IVIG as a discrete time-dependent covariate to minimize the risk of immortal time bias. Because the discharge from ICU does not fulfill the assumption of non-informative censoring, Fine-and-Gray's competing risk model was used for the primary outcome with ICU discharge as a competing event. The treatment effect on the primary outcome was described as a sub-distribution hazard ratio (sHR) with a 95% confidence interval. A robust variance estimator was used to account for correlations resulting from matching. Multiple imputation procedure was used to deal with missing data (BMI, seven values; D-dimer, nine values; and ferritin, three values). P values of less than 0.05 were considered statistically significant. All statistical analyses were conducted using Stata/MP 16.0 for Windows.

During the study period (between March 7, 2020 and September 9, 2020), 1417 patients were admitted to ICUs at Hazm Mebaireek General Hospital with the diagnosis of COVID-19. Seven hundred eighty-seven patients received invasive mechanical ventilation, of which 595 (75.6%) had moderate-to-severe ARDS. We excluded three patients who received IVIG for other indications from the treated group and two patients admitted to the ICU post-cardiac arrest from the control group. The remaining cohort of 590 patients comprised 190 treated with IVIG and 400 in the routine care group (Fig. 1).

A comparison between unmatched and propensity score-matched groups is shown in Table 1. Before matching, the patients in the IVIG group were older, had a higher prevalence of hypertension, dyslipidemia, and hemodialysis, had lower PaO₂/FiO₂ ratio, lesser vasopressor use, lower SOFA score, raised alanine transaminase (ALT), more use of dexamethasone and lesser use of methylprednisolone, hydrocortisone, tocilizumab, lopinavir-ritonavir, and oseltamivir. In the IVIG group, the median time from ICU admission to initiation of IVIG therapy was 6.28 days (IQR 2.1–11.9 days). The median cumulative dose of IVIG received was 150 g (IQR 105–235 g), and the median number of doses received was 4 (IQR 3–5).

Propensity score matching generated 118 matched sets. Post matching, the two groups did not have any statistically significant differences (Table 1). In the matched cohort, the median time from ICU admission to initiation of IVIG therapy was 6 days (IQR 2.2–11.1 days). The median cumulative dose of IVIG received was 152.0 g (IQR 108.0–235.0 g), and the median number of doses received was 5.0 (IQR 3.0–5.0).

The all-cause ICU mortality for COVID-19 pneumonia patients admitted with respiratory failure requiring invasive mechanical ventilation for moderate-severe ARDS was 27.1%, and for the matched cohort, it was 25.8%. ICU mortality was significantly higher in the IVIG group (36.4% vs. 15.3% for IVIG and routine care, respectively; sHR 3.5; 95% CI 1.98–6.19; P < 0.001). In the propensity score-matched and propensity score-adjusted survival analysis, there was an increased risk of death in the IVIG group compared to the routine care group (Table 2). Results were similar, with an increased risk of death in the IVIG-treated patients for the subgroups based on PaO₂/FiO₂ ratio and serum ferritin level. Association of IVIG with increased ICU mortality was significant regardless of the time from ICU admission to IVIG therapy and the number of doses received (Table 3).

Compared to routine care group, ventilator-free days at day-28 were lower in the IVIG group [median (IQR); 0 (0–18) vs. 22 (7–24) days; P < 0.001], and ICU-free days at day-28 were also lower in the IVIG group [median (IQR); 0 (0–8.8) vs. 16 (0–20) days; P < 0.001]. Additionally, the incidence of AKI was significantly higher in the IVIG group (85.6% vs. 67.8% for IVIG and routine care, respectively; P = 0.001).