Johnson NJ, Danielson KR, Counts CR, et al. Targeted temperature management at 33 versus 36 degrees: A retrospective cohort study. Crit Care Med. 2020 Mar;48(3):362-369.
Reviewed by Wazim Mohamed, MD
Read the article.*
*You will need to log in to US National Library of Medicine National Institutes of Health to read this article.
Targeted temperature management (TTM) has become standard of care in post-cardiac arrest survivors after two landmark studies in 2002 demonstrated improved neurological outcomes with induced hypothermia to 32-34°C in out-of-hospital cardiac arrest (OHCA) survivors with a shockable rhythm. The 2013 TTM trial explored the differences in outcome comparing 33°C to 36°C and found no difference. However, clinical studies since then suggest higher incidence of fevers, lower protocol adherence and trends toward worse outcome after adoption of 36°C.
The current study sought to explore an association between TTM goal temperature and neurological outcomes. This was a retrospective single center study of adult patients who suffered OHCA in Seattle between 2010 and 2017. Qualifying patients were identified from an existing registry. Prior to April 2014, the center had a TTM protocol with a goal temperature of 33°C (TTM33) and subsequently the goal temperature was changed to 36°C (TTM36). Patients being treated were assumed to be in accordance with the TTM33 protocol prior to this date and with the TTM36 protocol after. Other inclusion criteria included shockable and non-shockable rhythms, and not following simple commands after resuscitation. Life-threatening hemorrhages were an exclusion criteria.
Primary outcome: favorable neurological outcome at discharge (cerebral performance category (CPC) of 1 or 2). Secondary outcomes: in-hospital mortality, emergency department (ED) initiation of TTM, time to TTM initiation, time to lowest temperature and DNR status. Univariate analyses included evaluation of characteristics through Student t test or Wilcoxon rank-sum test. Categorical variables were compared using chi-square or Fisher exact tests. Logistic regression models were run examining the association between TTM goal temperature, favorable neurological outcome and mortality while adjusting for covariates. The covariates included age, sex, arrest location, witnessed arrest, bystander CPR, arrest etiology, initial rhythm and EMS response time.
The registry consisted of 782 patients of which a cohort 453 were included in the analysis. The median age was 54.6±15.6 years, predominantly male (70%) with 56% witnessed arrests and 47% receiving bystander CPR. The median time to initial contact with EMS was 3.9±1.5 minutes and an initial shockable rhythm was present in 38%. The vast majority of patients (91%) had a pulse on arrival to ED and the mean temperature on arrival was 35.5°C. Of the patients that received TTM, 41% survived to hospital discharge. The TTM33 period comprised 258 patients and the TTM36 cohort had 195 patients. The TTM33 group was older (56 vs. 51 yr; p<0.05) and had a higher incidence of cardiac etiology (45% vs. 35%; p<0.05). Other baseline characteristics, including initial cardiac rhythm, were similar between groups.
Primary outcome: In the univariate analysis, a significantly higher proportion of patients in TTM33 group had favorable neurological outcome (39.9% vs. 29.7%; p<0.05). In the multivariate logistic regression model as well, the TTM33 group was associated with higher favorable neurological outcome (OR 1.79; 95% CI 1.09 – 2.93), after adjusting for covariates. Secondary outcomes: There were no significant mortality differences between the groups: TTM33 45% vs. TTM36 36%; p=0.08. In the regression model, after adjusting for covariates, TTM33 did not have a significantly different mortality (OR 1.46; 95% CI 0.92-2.31). A significantly higher proportion of patients in the TTM33 group had TTM initiated in the ED (87% vs. 55%; p<0.001) and the median time to TTM initiation in the TTM33 group was also faster (1.5 hr vs. 3.5 hr; P<0.001). As expected, the lowest temperature was lower (32.5°C vs. 35°C; p<0.001) and the time to lowest temperature was longer in the TTM33 group (6.9 hr vs. 4.1 hr; p<0.001). There were no other significant differences in other aspects of care, DNR status and causes of mortality. The majority of patients that died were from complications of cerebral anoxia in both groups (TTM33 60.1% and TTM36 62.9%).
The authors conclude that a TTM goal temperature of 33°C had significantly higher odds of favorable neurological outcome at discharge compared to 36°C. The hypotheses put forward for this association include a delay in active temperature management for the TTM36 group (as demonstrated in the results), secular changes in patient characteristics over the years and differences in cardiac arrest care. There was 79% higher odds of neurologically intact survival for the TTM33 group in this study, which is in contrast to the 2013 TTM trial. A major criticism that the TTM trial faced was the long durations to induce TTM and achieve that target temperature (~10 hours) compared to the original HACA and Bernard trials from 2002 (2 to 8 hours), which may have negated the positive effect of a lower target temperature. The current study demonstrated a significant delay in starting active temperature management and 32% decrease in active temperature management in the ED for the TTM36 group. These factors could influence the higher the proportion of patients with poor neurological outcomes seen in the TTM36 group. One would expect that instituting TTM would be logistically easier in the TTM36 group; however, the opposite was seen in the current study. This trend has also been well reported in the literature among patients with a target temperature of 36°C subject to decreased active cooling measures, less time at target temperatures and increase incidences of fever.
The current study did not find an association between goal temperature and survival. Nearly 62% of patients in the current study had an initial non-shockable rhythm. Neither the TTM trial nor the HYPERION trial showed a mortality difference between 33°C vs 36°C, but HYPERION did show a higher proportion of patient with good neurologic outcome in the 33°C group. The current study does not include a subgroup analysis of mortality associations in the shockable and non-shockable groups separately. This may have been an interesting finding.
Several studies have documented worsening secular trends in post cardiac arrest survival rate and neurological outcomes since the publication of the TTM trial. One particular publication (Salter, et al) demonstrated an improvement in mortality up until the TTM trial publication with an abrupt reversal and increasing mortality henceforth. This could definitely be attributed to changing patient demographics or other aspects of patients care, but it may also be possible that widespread use of 36°C as target temperature or abandoning TTM altogether may have contributed to this.
The limitations of the current study include its retrospective nature, which restricts analysis to associations and not causation. Unmeasured confounders in either group could have contributed to the outcome; specifically, the incidence of fever rates in either groups have not been mentioned. This was a study conducted at a single high volume, high performing center, so the results may not be widely applicable. The data abstractor was not blinded to patient groups or other aspects of care, and there may have been a bias in CPC assessment. In addition, long-term outcomes (beyond hospital discharge) for these patients were not assessed.
Studies such as this emphasize the need to re-evaluate 36°C as a target temperature for TTM patients and begs the question, “Do the patients in the TTM trial truly represent the patient population that I treat?”