Critical Care: Research Letters
Use of High-Dose Hydroxocobalamin for Septic Shock
Jason R. Carr, Brian Locke, Jayshil J. Patel, et al
Chest 2025; 168: 131-135
To the Editor:
Septic shock, a global health burden with a case fatality rate exceeding 40%, is initiated and maintained by circulating nitric oxide and hydrogen sulfide that are emergent targets of therapeutic interest.1 High-dose (5-gm) hydroxocobalamin (vitamin B12) (HD-HOC) scavenges circulating nitric oxide and hydrogen sulfide and may attenuate the inflammatory response in early septic shock.2 Since 2014, several case reports and case series have reported that HD-HOC is associated with reduced vasopressor requirements in adults with refractory vasodilatory shock (including septic shock).3,4 In 2023, a phase IIa randomized, controlled pilot trial reported that HD-HOC reduced vasopressor dose at 1 and 3 hours in adults with early septic shock (within 24 hours of ICU admission and onset of septic shock).5 Given growing interest in the use of HD-HOC in septic shock, we hypothesized that use of HD-HOC in septic shock is increasing, yet highly variable between institutions.
Methods
We queried the PINC AI Healthcare Database, an enhanced claims-based data set that captures approximately 25% of all US inpatient hospitalizations that contain claims-indexed and hospital day-indexed billing information that was designed for measuring quality and health care use.6 We included adult patients (≥ 18 years) with septic shock who were admitted to an ICU from any location (as in prior work, defined as patients with suspected infection and acute organ failure who received vasopressors) on the first day of hospitalization from 2016 to 2022.7,8 The outcome of interest was receipt of at least 1 dose of HD-HOC during hospital admission. To improve specificity for HD-HOC administered solely for sepsis, we excluded patients with cobalamin deficiency, cyanide poisoning, or cardiac surgery on hospital day 1.
We calculated medians and proportions to summarize patient-level and hospital-level covariables, the proportion of patients who received HD-HOC on each hospital admission day, and annual rates of HD-HOC use. We characterized between-hospital variation in HD-HOC using patient-level hierarchic multivariable logistic regression models with hospital of admission as a random effect, characteristics of the hospital of admission as fixed effects (Table 1), and each patient’s predicted probability of death as a fixed effect (as predicted by patient-level covariables) (Table 1). From the hierarchic model, we identified each hospital’s risk-adjusted rate of HD-HOC use and the intraclass correlation coefficient (a measure of the unexplained between-hospital variation in HD-HOC use). A priori, an intraclass correlation coefficient of > 15% was considered “high variation,” as described in prior work.9 Statistical testing was 2-tailed with alpha set at .05; R (version 4.1.2) was used for analysis. This study was deemed not human participants research by the Boston University institutional review board (#H-41795).
Table 1. Characteristics of Patients Admitted With Septic Shock, 2016-2022
| Variable | Category | ||
|---|---|---|---|
| Patient Demographics and Clinical Characteristics | Overall Population (N = 598,858) | No HD-HOC (n = 598,616) | Received HD-HOC (n = 242) |
| Age, median [IQR], y | 67 [57, 77] | 67 [57, 77] | 59 [44, 66] |
| In-hospital deaths | 178,144 (29.7) | 177,980 (29.7) | 164 (67.8) |
| Sex | |||
| Female | 288,922 (48.2) | 288,841 (48.3) | 81 (33.5) |
| Male | 309,775 (51.7) | 309,614 (51.7) | 161 (66.5) |
| Not specified | 161 (0.0) | 161 (0.0) | 0 (0.0) |
| Race | |||
| Asian | 12,998 (2.2) | 12,996 (2.2) | 2 (0.8) |
| Black | 78,449 (13.1) | 78,407 (13.1) | 42 (17.4) |
| Other | 45,168 (7.5) | 45,134 (7.5) | 34 (14.0) |
| Not specified | 13,766 (2.3) | 13,753 (2.3) | 13 (5.4) |
| White | 448,477 (74.9) | 448,326 (74.9) | 151 (62.4) |
| Gagne Chronic Comorbidity Score, median [IQR] | 5.00 [3.00, 7.00] | 5.00 [3.00, 7.00] | 5.00 [3.00, 7.00] |
| Acute Organ Dysfunction | |||
| Cardiac | 489,034 (81.7) | 488,821 (81.7) | 213 (88.0) |
| Respiratory | 366,188 (61.1) | 365,981 (61.1) | 207 (85.5) |
| Neurologic | 194,274 (32.4) | 194,204 (32.4) | 70 (28.9) |
| Hematologic | 106,999 (17.9) | 106,920 (17.9) | 79 (32.6) |
| Hepatic | 47,844 (8.0) | 47,783 (8.0) | 61 (25.2) |
| Invasive ventilation on day 1, mean (SD) | 293,326 (49.0) | 293,150 (49.0) | 176 (72.7) |
| Dialysis on day 1 | 26,127 (4.4) | 26,088 (4.4) | 39 (16.1) |
| Major surgery on day 1 | 64,353 (10.7) | 64,309 (10.7) | 44 (18.2) |
| Vasoactive support | |||
| Norepinephrine | 509,547 (85.1) | 509,345 (85.1) | 202 (83.5) |
| Epinephrine | 24,557 (4.1) | 24,528 (4.1) | 29 (12.0) |
| Vasopressin | 114,438 (19.1) | 114,307 (19.1) | 131 (54.1) |
| Phenylephrine | 123,316 (20.6) | 123,236 (20.6) | 80 (33.1) |
| Dopamine | 41,254 (6.9) | 41,243 (6.9) | 11 (4.5) |
| Angiotensin II | 888 (0.1) | 882 (0.1) | 6 (2.5) |
| Midodrine | 18,249 (3.0) | 18,237 (3.0) | 12 (5.0) |
| Dobutamine | 12,263 (2.0) | 12,242 (2.0) | 21 (8.7) |
| Milrinone | 3,453 (0.6) | 3,443 (0.6) | 10 (4.1) |
| Hydrocortisone | 88,392 (14.8) | 88,305 (14.8) | 87 (36.0) |
| Methylene blue | 967 (0.2) | 950 (0.2) | 17 (7.0) |
| Characteristics of hospital of admission | |||
| Hospital region | |||
| Midwest | 131,335 (21.9) | 131,252 (21.9) | 83 (34.3) |
| Northeast | 75,783 (12.7) | 75,741 (12.7) | 42 (17.4) |
| South | 284,187 (47.5) | 284,092 (47.5) | 95 (39.3) |
| West | 107,553 (18.0) | 107,531 (18.0) | 22 (9.1) |
| Safety net hospital | 163,338 (27.3) | 163,234 (27.3) | 104 (43.0) |
| Teaching hospital | 297,540 (49.7) | 297,353 (49.7) | 187 (77.3) |
| No. of beds | |||
| < 100 | 23,512 (3.9) | 23,512 (3.9) | 0 (0.0) |
| 100-199 | 76,915 (12.8) | 76,909 (12.8) | 6 (2.5) |
| 200-299 | 102,330 (17.1) | 102,323 (17.1) | 7 (2.9) |
| 300-399 | 110,097 (18.4) | 110,078 (18.4) | 19 (7.9) |
| 400-499 | 75,008 (12.5) | 74,973 (12.5) | 35 (14.5) |
| ≥ 500 | 210,996 (35.2) | 210,821 (35.2) | 175 (72.3) |
| Urban hospital | 524,988 (87.7) | 524,749 (87.7) | 239 (98.8) |
| ICU type | |||
| Cardiac | 54,101 (9.0) | 54,070 (9.0) | 31 (12.8) |
| Medical | 55,486 (9.3) | 55,440 (9.3) | 46 (19.0) |
| Other/general | 468,190 (78.2) | 468,036 (78.2) | 154 (63.6) |
| Surgical | 21,081 (3.5) | 21,070 (3.5) | 11 (4.5) |
Data are presented as No. (%) unless otherwise indicated. Characteristics are shown for the total population and are stratified by HD-HOC receipt. All patient characteristics shown were incorporated into a patient-level index for the probability of death (included as a fixed effect in the hierarchic regression model for HD-HOC receipt). The patient-level hierarchic regression model for HD-HOC receipt also included all hospital characteristics shown (as individual fixed effects) and hospital of admission (as a random effect). HD-HOC = high-dose hydroxocobalamin; IQR = interquartile range.
Results
We identified 598,858 patients (mean age, 67.0 years; interquartile range, 57.0 to 77.0 years; female, 288,922 [48.2%]; Table 1) from 987 hospitals who had been admitted to an ICU with septic shock on the first day of hospital admission. Of these, 242 patients (0.04%) received HD-HOC at 59 hospitals (6%); 47 patients (19.4%) also received methylene blue. When stratified by year, HD-HOC use increased from 0.007% in 2016 to 0.093% in 2022 (relative change, 1,229%) (Fig 1A).
Mortality rate in the overall group was 29.7% (178,144 of 598,858 patients); mortality rate in the group that received HD-HOC was 67.8% (164 of 242 patients). In the hierarchic model, a higher predicted probability of death was associated with increased odds of HD-HOC receipt (adjusted OR, 1.03; 95% CI, 1.02 to 1.04, per 1% increase in predicted probability of death). The intraclass correlation coefficient (that represented unexplained between-hospital variation in HD-HOC use) was 67.7% (95% CI, 67.0 to 68.4) (Fig 1B). No other hospital-level characteristics (Table 1) were statistically significant predictors of HD-HOC use.
Discussion
We examined uptake of HD-HOC for the treatment of patients with septic shock in the United States from 2016 to 2022. Although HD-HOC use remains rare, adoption has increased over time. Greater severity of illness was associated with an increased likelihood of receiving HD-HOC. Most of the remaining variability in whether HD-HOC was used (67%) correlated with the hospital where a patient received care, rather than individual patient characteristics. A minority of hospitals (6.0%) accounted for all HD-HOC prescribing. In unadjusted analysis, patients who received HD-HOC were much more likely to die than those with septic shock who did not receive HD-HOC (67.8% vs 29.7% in-hospital mortality rate), though this likely represents its use as primarily a therapy in refractory shock rather than a causal effect of its use.
Outside of the single phase IIa trial that reported feasibility of recruitment and use in early septic shock that was published after the study period,5 evidence is limited to case reports and small case series. The use of HD-HOC before definitive data that supported efficacy mirrors the observed early adoption of hydrocortisone/ascorbic acid/thiamine for treatment of septic shock in several ways.10 Although overall rates of HD-HOC are (as of 2022) lower than the use hydrocortisone/ascorbic acid/thiamine was in 2018, both therapies were used predominantly in only a few hospitals and used preferentially among patients with greater severity of illness, which suggests that clinical urgency is a major driver of willingness to adopt as-yet unproven therapies. Because the hydrocortisone/ascorbic acid/thiamine combination therapy subsequently was not shown to be efficacious, our findings highlight the importance of conducting phase III efficacy trials of HD-HOC.
Our study has limitations. As in other research that uses administrative data, misclassification of septic shock is possible. However, our definition of septic shock is used commonly in critical care health services research.7,8 We are also unable to assess whether HD-HOC adoption has continued to rise since the publication of the phase IIa trial in 2023. Finally, because of the rare use of HD-HOC in our data, we are unable to explore how other patient characteristics are associated with HD-HOC receipt.
Given rising use, substantial practice variation between institutions, and previously reported changes in vasopressor doses, a pivotal trial that will evaluate the efficacy of HD-HOC in early septic shock is both important and time-sensitive to inform clinical practice.
