In this retrospective observational study, results have indicated that the serum lactate level measured during CPR correlated with the outcome of survival to hospital discharge for IHCA. A serum lactate level <9 mmol/L within 10 minutes of CPR initiation was positively associated with survival. Moreover, together with other variables, the serum lactate level could be useful in estimating optimal CPR durations before transition to next strategy for different patient groups, which could help clinicians create further resuscitation strategies in addition to the initial conventional CPR.
Excess production of lactate has been reported to be largely due to tissue hypoxia or hypoperfusion with associated anaerobic metabolism . For the past few decades, it has been noted that the blood lactate level might correlate with the duration of no-flow and low-flow status during CPR [23, 24], both of which are known to impact outcomes after cardiac arrest . This biomarker is therefore a promising candidate for a prognostic predictor of CPR. Weil et al.  suggested that the lactate level measured within 10 minutes of CPR could be associated with survival outcomes despite the fact that other important confounders were not considered during their analysis. In subsequent investigations, lower serum lactate measurements or early and effective serum lactate clearance was also shown to correlate with decreased mortality in cardiac arrest patients [12–18]. These studies [12–18] all used post-ROSC lactate level for analysis.
In line with previous studies [11–18], we revealed that the serum lactate level measured during CPR was significantly associated with survival outcomes after adjusting for the confounding effects of multiple factors. There was a near-linear inverse association between the lactate level and the probability of survival. Furthermore, a lactate level <9 mmol/L was significantly associated with higher survival probabilities. This cutoff point was higher than those reported in previous studies [15, 16]. Seeger et al.  reported that a lactate level >6.94 mmol/L was associated with poor neurological outcomes; Kaji et al.  reported that a lactate level <5 mmol/L correlated with a favourable neurological outcome. The difference in threshold values could be explained by the different time points of lactate measurement and outcome selection. Compared to measurements after ROSC [15, 16], in which blood flow had already recovered to ameliorate systemic hypoxia and hypoperfusion, it seems reasonable that lactate levels would be higher when measured during ongoing CPR.
However, the causes of elevated serum lactate do merit further explanation. Firstly, the serum lactate level could indeed be a surrogate marker for duration of no-flow and low-flow status after cardiac arrest [23, 24]. Nevertheless, Müllner et al. also suggested that the lactate level may be just a weak measure of the duration of cardiac arrest . For IHCA patients, various conditions, such as severe septic shock, could also produce similarly high lactate levels even without cardiac arrest. The use of adrenaline had also been shown to increase the serum lactate level . Secondly, the serum lactate level reflected the difference between lactate production and elimination; thus, an increased lactate level may also indicate mechanisms other than cellular hypoxia or hypoperfusion. For example, impaired liver function could lead to reduced lactate elimination and an elevated lactate level, even when the hemodynamic status was not compromised [28, 29]. Despite these considerations, by adjusting the effects of multiple confounders, including hepatic insufficiency, we believe that the serum lactate level measured during CPR, with its complex physiologic and pathologic implications, could still serve as a useful prognostic factor for outcomes.
There are currently no recommendations for optimal CPR duration before transition to the next strategy [3, 4]. In clinical practice, patients usually received a predetermined duration of CPR, often 30 minutes, even if the patients had not exhibited any response to the repeated resuscitation efforts. However, for adult IHCA patients, Goldberger et al. showed that patients would have a higher probability of survival if they were resuscitated at hospitals that tended to implement longer CPR durations . Furthermore, Matos et al.  demonstrated that outcomes after paediatric IHCA could differ significantly between different patient groups, indicating that surgical cardiac patients had the highest probability of a good outcome and trauma patients had the lowest probability, and that CPR >20 minutes may not be futile in certain patient groups. Conversely, Kim et al.  recommended that ECPR should be considered for adult patients with out-of-hospital cardiac arrest (OHCA) when CPR was ≥21 minutes. In addition, in an adult OHCA study, Reynolds et al.  revealed that the probability of favourable outcomes declined rapidly with each minute of CPR and most surviving patients (89.7 %) with favourable neurological outcomes would achieve ROSC within 16.1 minutes of CPR. Therefore, Reynolds et al.  suggested that novel therapeutics, such as ECPR, should be considered early after CPR rather than after the complete failure of resuscitation. This could also hold true for IHCA patients especially for whom applications of ECPR were more available and mature .
Using CPR duration alone to define futile resuscitation may risk delaying timely implementation of ECPR and missing opportunities to improve patient outcomes. However, the classification method of patient groups used by Matos et al.  was somewhat difficult to be generalized and applied in clinical practice. In our current study, in addition to serum lactate level and CPR duration, we also noted that shockable rhythm and hepatic insufficiency were also significantly associated with survival outcomes, both of which have also been reported in previous studies [33, 34]. Based on these variables, we were able to divide patients into groups with different survival probabilities (Fig. 1). In this way, clinicians may be more prepared for determining the optimal CPR duration for each patient group before transition to the next strategy, either a more advanced strategy, such as ECPR, or termination of CPR.
For example, if we choose the average IHCA survival rate 18 % as a reference [1, 2], we would note that the maximally allowed CPR duration was 32.3 minutes for patient group 1 before the survival probability dropped below 18 %; however, for patient groups 7 and 8, the survival probability would be far below 18 % even from the start of CPR (Table 5). From another perspective, if we choose the commonly used CPR duration of 30 minutes, we could note that even if the CPR intervals were the same, there would be a 10-fold difference in the probability of survival between patient groups 1 (23 %) and 8 (2 %) (Table 5). Therefore, it is clear that if clinicians depended solely on CPR duration to define futile resuscitation, it may be premature to give up for some patients and delayed to activate alternative resuscitation measures for other patients. However, because of the limited number of patients in our analysis, the estimations for CPR duration and survival probability can only serve for the purpose of explanation rather than precise estimations.
A lactate level <9 mmol/L was also significantly associated with a 10-minute ROSC. This CPR interval has been used as one of prerequisites prior to activation of ECPR  and termination of IHCA CPR [35, 36]. We suggest that for IHCA patients with a lactate level higher than 9 mmol/L and without exclusion criteria for ECPR, the ECPR team should probably be alerted earlier in order to be able to initiate the ECPR sooner to improve outcomes, as these patients were shown to have less chance of achieving rapid ROSC. Conversely, the clinical decision to terminate IHCA was based solely on witness status, initial arrest rhythm and CPR duration [35, 36]. Using serum lactate level as an additional variable could probably increase the accuracy of this decision and avoid lengthening the dying process of critical patients.
Post-ROSC serum lactate level or the clearance rate of serum lactate has been associated with neurological outcomes [12, 14, 16, 17]. An elevated lactate level may represent a more severe status of tissue hypoperfusion and ischaemia-reperfusion injury, which might aggravate the cerebral dysfunction in post-cardiac arrest syndrome . As there were only 21 patients (6.2 %) in our cohort who regained a favourable neurological outcome, the statistical power may be insufficient to detect the difference between patients with and without a favourable neurological outcome. Nonetheless, the trend of the point estimate of lactate level (odds ratio 1.86) supports our hypothesis. Future studies with more patients would be expected to clarify this issue.
In summary, we found that the serum lactate level measured during CPR was associated with survival outcomes. Along with other variables, the lactate level could help separate patients into several groups with different levels of survival probability. With a predetermined survival probability, clinicians may be able to determine the optimal CPR duration for each group. However, in our current study, we were not aiming to obtain and recommend a fixed CPR duration for each patient group as this was dependent on individual patients. We also did not intend to analyse which strategies should be undertaken following implementation of this optimal CPR duration. What we did emphasize, however, was that the CPR duration alone should not be used for decision-making in CPR. The optimal resuscitation strategy should be personalised and be adjusted to the specific context, taking into account the autonomy of patients, wishes of family members, or the availability of ECPR and other resources of critical care.
First, this was an observational study, which could only establish an association rather than a causal relationship between independent and dependent variables. Second, the quality of CPR could not be retrospectively determined. Currently, there were no studies indicating that the serum lactate level measured during CPR could reflect the quality of CPR. The association between serum lactate level and quality of CPR should be further examined in a prospective method. Third, we could not be sure whether the serum lactate was from arterial or venous samples. Ralston et al.  reported that there was no significant difference in lactate levels between arterial and venous samples during animal CPR. Therefore, this issue may not bias our results to a significant extent. Fourth, the proportion of patients receiving therapeutic hypothermia was low in our cohort. The influence of therapeutic hypothermia on the association between lactate levels and CPR outcomes might be left for future researchers to investigate. Fifth, we used survival to hospital discharge as the primary outcome because the number in the studied cohort was small. Nonetheless, neurologically intact survival might be a more important and meaningful long-term outcome and should be explored as the primary outcome in future research with larger patient populations. Finally, we used IHCA patients for analysis, and so the results might not be applicaple to OHCA patients. In addition, the current analysis was a retrospective study in a single centre with a highly selective cohort, so the probability for selection bias might be high. The lactate levels measured during the initial 10 minutes of CPR were only available in 30 % (340/1123) of the total IHCA patients. Most of the lactate values were reported simultaneously with measurements of blood gases or potassium by point-of-care machines. As resuscitation guidelines [3, 4] do not suggest the measurement of lactate level during CPR, only a prospective study could avoid this kind of selection bias.