In this retrospective cohort study of children hospitalized with moderate to severe MIS-C, hospital stay was longer in patients who received anakinra compared to patients who did not receive anakinra. Also, there was no difference in vasoactive support duration, VIS, fever duration, CRP reduction, or ICU length of stay in patients who received anakinra compared to patients who did not receive anakinra. This finding does not support improved outcomes with anakinra in the propensity score matched cohort.

The combination therapy of IVIG and glucocorticoids, when compared to IVIG alone, has been associated with a range of improved outcomes in patients with MIS-C. These include a more favorable fever course [23], lower frequency of treatment escalation [24], reduced need for hemodynamic support [24], shorter stays in the intensive care unit [9], a shorter time to recovery of cardiac function, and a lower risk of serious short-term outcomes [25]. A recent study showed evidence that administering glucocorticoids earlier in the disease process of MIS-C may limit the inflammatory cascade that leads to tissue injury, organ dysfunction, and shock, thereby leading to a faster recovery [26]. However, McArdle et al. did not find any evidence of differences in outcomes between treatment with glucocorticoids or IVIG as single agents, or between the single-agent and dual-agent primary treatments [27]. While there are discrepant results, the general recommendations are to include IVIG and steroids in the initial treatment of MIS-C.

Cole et al. found that patients with MIS-C who were treated with infliximab in addition to IVIG were less likely to receive supplementary treatment, had a shorter duration of ICU length of stay, had decreased development of LV dysfunction, and demonstrated a more rapid decline in CRP levels [19] than those who were only given IVIG. Notably in this study, steroids were not used as part of the treatment. Comparatively, in our study almost every child received steroids whether they also received anakinra or not. Our findings showed no difference in CRP reduction, worsening LV function, or ICU length of stay between the two groups. These findings highlight the potential benefits of combining different therapies in certain cases, and further emphasize the importance of individualized treatment plans based on patient needs and characteristics.

Limited data exist regarding the use of anakinra in MIS-C, however a few case reports demonstrated clinical improvement, and a reduction in inflammatory markers [28, 29] as well as rapid cardiac function improvement [25] after anakinra initiation. A recent study showed that early anakinra treatment improves cardiac outcome regardless of disease severity [30]. Our findings showed no difference in CRP reduction or improving LV function between the two groups. Interestingly, our cohort had a higher proportion of patients (50%) with vasoactive need, whereas 19% in the Taddio study had hypotension or cardiogenic shock. Furthermore, their anakinra cohort (9%) was smaller than ours (41%). Given the different definitions of MIS-C by the Royal College of Paediatrics and Child Health [31] and the CDC [21], and the unstandardized definitions of severity, it is possible that the heterogeneous populations account for our different findings.

In our matched cohort, the anakinra group had a longer median length of hospital stay. Additionally, anakinra was not associated with improvement in vasoactive medication duration, time to fever resolution, organ support and ICU length of stay. The absence of statistically or clinically meaningful improved outcomes in the matched cohort suggests that anakinra, when given in addition to IVIG and steroids, does not improve patient outcomes in MIS-C. Since 93% of all patients received IVIG and 99% received steroids, the possibility of a confounding effect from those immunomodulators is minimized. However, with its relatively low risk profile, anakinra may still be beneficial in managing severe disease states.

To address the variability in factors that can impact time of admission, we performed a sensitivity analysis using length of stay with the starting from time of initial steroid administration to time of discharge and confirmed a longer stay in the anakinra group. The longer hospital stay in the anakinra group may be attributed to the duration of the taper and need for observation to monitor rebound signs and symptoms. Anakinra was administered with a delay of approximately 12–14 h following the start of IVIG or steroids and about 24 h after admission. This timing may be due to several factors: (1) anakinra may have been started in patients who exhibited an inadequate response to initial first-line therapies, (2) anakinra required a rheumatology specialist to order it, which may not have happened overnight with as much ease as ordering IVIG and steroids, and (3) clinician behavior changing over time, with later administration in disease course early in the pandemic but earlier adoption of anakinra later in the pandemic. This timing gap of 12–14 h may provide an additional explanation for the prolonged hospitalization seen in the anakinra group, but it would not account for the difference of 2–3 days. Interestingly, we found that there was about 48 h between the last dose of anakinra and time of discharge. This result in combination with the fact that ICU LOS was the same in the matched groups, suggests that the difference in hospital length of stay stems from the end of the hospital stay. This longer length of stay may be for clinical observation purposes once anakinra was stopped.

The main strength of this study is being one of the largest single institutional cohorts evaluating the effectiveness of anakinra use in MIS-C. In addition, the application of propensity score-matched analysis was important in controlling for confounding factors.

This study has several limitations. First, as a single center study in the USA from 2020 to 2021, the results may not be generalizable to other regions or countries, or to MIS-C in the future. Furthermore, the knowledge and familiarity with this new illness changed over time and it is impossible to ascertain how this may have affected practice patterns or patient outcomes. Second, it is unknown how the unmatched patients in the cohort may have impacted the results. It is possible that in severe but unmatched cases, anakinra was beneficial. Third, the dosage and duration of steroid and timing of anakinra treatment used was variable, and the small size of the study population precludes comparison of anakinra as first line or second-line treatment or by dose of anakinra. It is unclear if early use of anakinra as opposed to second-line use would lead to different outcomes. While our propensity score method can mitigate disparities between the groups at the study’s outset, they may not fully address discrepancies that may arise after admission. This scenario could introduce an unmeasured ‘confounding by indication’ bias that persists even after propensity score matching. Our primary outcome, length of stay was measured from time to arrival to our health system to time of discharge. While there may be unmeasured factors that affect time of discharge, such as social or logistical factors, they would likely similarly impact both anakinra and non-anakinra cohorts. Finally, severity classification has no quantitative thresholds and is based on expert consensus, which makes comparison of various studies difficult.