Author: Tucker Matthews
Lit Review: OHCA Resuscitation Termination
I’ve started thinking more about calling codes. It’s always been odd to me how lacking we are in concrete rules to help us decide when another round of CPR is useless, and when to just end our efforts. In particular, as ED folks, most of what we see involves patients who had an out of hospital cardiac arrest (OHCA), and so I started looking into the protocols that exist for terminating resuscitation efforts for those patients. Not surprisingly, the current answer is just to keep on transporting for the most part, but here are four recent studies that I found that tried to investigate the utility of these protocols on different populations of patients.
Paper 1: Goto Y, Funada A, Maeda T, Goto Y. Termination-of-resuscitation rule in the emergency department for patients with refractory out-of-hospital cardiac arrest: a nationwide, population-based observational study. Crit Care. 2022 May 16;26(1):137. doi: 10.1186/s13054-022-03999-x. PMID: 35578295; PMCID: PMC9109290.
This is a national observational study out of Japan, where their EMS providers aren’t allowed to terminate OHCA in the field, and so a termination protocol was made previously by the authors, which they updated in this paper.
The initial rule was that CPR could be terminated if there was no prehospital ROSC, an initial non-shockable rhythm, and the arrest was unwitnessed. They wanted to increase the specificity by finding other variables that could be included. They used a group of patients to test their proposed variables (age, initial rhythm, witnessed status, bystander CPR, prehospital shock delivery, duration of EMS CPR, prehospital ROSC), and then performed a validation against a separate group of patients.
It’s a massive study, with 465,657 total patients (split pretty evenly in the two groups).
Their analysis developed a new criteria for predicting 1 month mortality, consisting of initial asystole, unwitnessed arrest, EMS CPR duration >20 minutes, and the absence of prehospital ROSC. Using these criteria, the specificity was greater than 99% in both their development and validation groups. That’s significantly more specific than the previous rule, and other rules out of Korea. Of note, these criteria aren’t rare; 27% of patients meet all 4 variables.
The original criteria had actually been externally validated in North America, so it’s likely that these new criteria may also be helpful. Obviously we have different EMS setups here, but I really like this kind of cognitive offloading and improved resource utilization. They also mention that they weren’t able to look at end-tidal CO2 data, but an initial ETCO2<10 would be an interesting additional data point.
Paper 2: Lin YY, Lai YY, Chang HC, Lu CH, Chiu PW, Kuo YS, Huang SP, Chang YH, Lin CH. Predictive performances of ALS and BLS termination of resuscitation rules in out-of-hospital cardiac arrest for different resuscitation protocols. BMC Emerg Med. 2022 Mar 27;22(1):53. doi: 10.1186/s12873-022-00606-8. PMID: 35346055; PMCID: PMC8958476.
This is a retrospective study out of Taiwan that looked at the performance of the BLS and ALS termination of resuscitation (TOR) protocols at two different timepoints (2020 and 2015). Part of their rationale for these timepoints specifically was the development of the AHA guidelines in 2015, which provided new emphases and protocols to EMS providers, which may impact OHCA outcomes. This city does not have existing rules for prehospital TOR.
The BLS TOR rule recommends termination if three criteria are met: arrest unwitnessed by EMS, absence of ROSC prior to transport, absence of shock delivery prior to transport. The ALS TOR rule recommends termination if four criteria are met: arrest unwitnessed, absence of bystander CPR, absence of pre-transport ROSC, absence of defib pre-transport.
1260 patients in the 2015 group, and 979 in 2020. Unfortunately, there are loads of differences between the characteristics in each group that don’t necessarily pertain to updates to the guidelines. For example, 48% of patients in 2020 got bystander CPR, as opposed to 28% in 2015. That is a massive difference in something that we know significantly impacts outcomes. The good news is that these sort of differences shouldn’t really matter overall, since the more important investigation is just that the scoring tools still work, no matter who the patients are.
The only thing that really matters is specificity here. What’s terrifying is how low the specificities are for the BLS rules. For survival to discharge, they are just 40% specific in 2015, and up to 69% in 2020. The ALS rule was actually way better, at 100%, up from 91% in 2015. It’s odd to me that they would have these improved numbers to a score characteristic, and I can’t really think of a great explanation. The authors propose that it may be explained by the 2020 cohort having more time in the field, and thus more resuscitation before the rules could not still be applied. That’s definitely possible, but it should only impact whether ROSC was obtained.
Overall, what this really tells me is that these rules for TOR are extremely dependent on variables that might not be immediately obvious, and so external validation studies are extremely important. And on an individual case basis, even the cases that meet the rule should still be thought over an only terminated if that logically seems the correct decision.
Paper 3: Nazeha N, Ong MEH, Limkakeng AT Jr, Ye JJ, Joiner AP, Blewer A, Shahidah N, Nadarajan GD, Mao DR, Graves N. A hypothetical implementation of ‘Termination of Resuscitation’ protocol for out-of-hospital cardiac arrest. Resusc Plus. 2021 Mar 3;6:100092. doi: 10.1016/j.resplu.2021.100092. PMID: 34223357; PMCID: PMC8244430.
This is a paper out of Singapore, also looking at the AHA TOR criteria (terminate resuscitation if arrest not witnessed by EMS/bystander, no bystander CPR, no pre-transport shocks, and no pre-transport ROSC). They are specifically trying to investigate the possible cost savings to not transporting patients who meet the criteria.
In 2019, the TOR protocol was adopted in Singapore, so these authors were able to use data from 2014-2017, when OHCA patients were required to be transported to the hospital with ongoing resuscitation. The used a total of 5396 cases in their analysis. The proportion of patients transported to the hospital decreased by 11.2%. Not surprisingly, there wasn’t a big difference between in-hospital costs between the groups, since those not transported are generally not surviving to admission.
I really had some frustrations reading this paper. They focus a lot on cost, but don’t actually study that at all. And they make multiple comments on the expense of the cases where a neurologically devastated patient is kept inpatient, but that’s not really the function of TOR protocols. TOR protocols are specifically trying to stop patients who will be dead in the ED from being transported. The patients who achieve ROSC and are admitted even though their prognosis is terrible, are a totally different category, that a separate scoring tool would be needed for. And if anything, this tells me that neuroprognostication needs to improve so that ICU doctors who have a lot more time are able to consider terminating care with the aid of time-consuming goals of care discussions. Their data showing the 11% reduction in transports is great, but it’s consistent with previously published data from other areas, so it’s definitely nothing particularly ground-breaking.
One other note is that they note 2 patients who would have been terminated on scene with the TOR protocol, but actually survived with decent functional status. This speaks to the importance of extremely high specificity numbers for a score like this. It also kind of annoys me that they didn’t say anything about those cases. Why do they think they were false positives?
One thing that I did like though is that they brought up on limitation to these kinds of on-scene termination protocols, which is that it’s not like the paramedics/EMTs just get to pronounce and then leave (pronounce then bounce?). They have to wait for police, and also have that discussion with the family about why there won’t be any hospital transport. I think it’s important to consider those sorts of outcomes when we consider just denying care in some way.
Paper 4: Harris MI, Crowe RP, Anders J, D’Acunto S, Adelgais KM, Fishe J. Applying a set of termination of resuscitation criteria to paediatric out-of-hospital cardiac arrest. Resuscitation. 2021 Dec;169:175-181. doi: 10.1016/j.resuscitation.2021.09.015. Epub 2021 Sep 20. PMID: 34555488.
Not surprisingly, there weren’t any prehospital resuscitation protocols for pediatric patients for a long time. In 2019, a pediatric version of the TOR criteria were published by the Maryland Institute of EMS Systems. This study tried to retrospectively apply these criteria to OHCA pediatric patients over the 2019 calendar year through one of the EMS EHR suppliers (ESO).
The specific criteria are having two 15-minute CPR cycles, 3 or more doses of epinephrine, asystole, and a terminal EtCO2 < 15 mmHg. There are subtle differences for traumatic patients, but to me that’s a totally separate patient population (especially in kids), and should really have it’s own analyses.
Because this is a prehospital EMR, they were severely limited in their choice of primary outcomes, and settled on the occurrence of prehospital ROSC. This is incredibly misleading as an outcome. If we are trying to determine if a patient needs transport to a hospital for continued vital therapy, then our outcome variable can’t preclude that therapy. So without data on neurologic outcomes and survival to hospital discharge, there just isn’t any way to actually calculate useful test characteristics.
The total number of cases was 1595, with a median age of 1 year. 23% had prehospital ROSC. Not shockingly, the protocol correctly predicted 99.7% of the patients who had prehospital ROSC as patients who shouldn’t have protocoled TOR in the field. Unfortunately though, we don’t know how specific it was for patients who maybe got ROSC in the hospital. We also don’t know anything about the future outcomes for these patients.
All in all, this study does nothing for me. They also made some stretches using surrogate data in place of definitive measures of the actual pTOR criteria. For example, they used scene time > 30 minutes in place of having two 15 minute CPR cycles. That’s just not so straightforward. To actually get some meaningful data on this, they would need to partner with ESO and make sure the correct data points were routinely collected/input by the EMS providers, and then would also need some sort of eventual hospital-based clinical information for these patients too. This is a topic where we just can’t have false positives (where a kid is just left dead when they could have been resuscitated), and until a really well-done study can show that the specificity is basically 100%, I want every kid transported.