Introduction

A decade ago, two landmark studies demonstrated the survival and neurologic benefits of therapeutic hypothermia (TH) in comatose survivors of out-of-hospital cardiac arrest (OHCA), and since then TH has become standard of care in this patient population [1, 2]. Although both the American Heart Association (AHA) and the International Liaison Committee on Resuscitation (ILCOR) [3, 4] offer guidelines for early goal-directed hemodynamic optimization, there is scant literature examining the relationship between hemodynamics and outcomes in post-cardiac arrest syndrome (PCAS) patients treated with TH. Instead, these guidelines are based on studies that have found similarities between PCAS and severe sepsis [5, 6], adopting the basic tenets of goal-directed hemodynamic optimization used to treat severe sepsis patients [7]. The 2010 AHA recommendations for hemodynamic optimization include fluid administration and vasopressor, inotropic, and vasodilator agents titrated as needed to optimize blood pressure, cardiac output, and systemic perfusion (Class I, LOE B). “Reasonable values” include a mean arterial pressure (MAP) ≥65 mmHg or systolic blood pressure ≥90 mmHg and central venous oxygen saturation (ScvO2) ≥70 %. The 2010 ILCOR guidelines are more detailed, recommending MAP of 65–90 mmHg, central venous pressure (CVP) of 8–12 mmHg, ScvO2 >70 %, hematocrit >30 %, lactate ≤2 mmol/L, and urine output ≥0.5 mL/kg/h.

However, PCAS is a more complex clinical entity than a sepsis-like syndrome alone. Brain injury and cardiovascular instability are the leading causes of mortality in cardiac arrest survivors admitted to an intensive care unit (ICU) [8]. As Nolan and Soar discussed in an editorial advocating a post-arrest care bundle, an optimal MAP has to “maintain an adequate cerebral perfusion without exposing the myocardium to excessive afterload” [9], since PCAS is complicated by myocardial stunning, hemodynamic changes and instability within the first 24 h [10]. No studies have addressed the efficacy or goals of early hemodynamic optimization in PCAS patients [11], and only a minority of TH implementation studies state specific hemodynamic goals and outcomes [12]. It is possible that hemodynamic optimization and current goal values may help, harm, or have no impact on outcomes. We hypothesized that patients with higher MAPs would have better survival and neurological outcomes, and increasing requirement for vasoactive agents would be associated with poor outcomes.

Methods

All consecutive comatose PCAS patients undergoing TH in our health system from May 2005 to October 2011 were identified using data collected from the Penn Alliance for Therapeutic Hypothermia (PATH) registry. PATH was established in 2010 as a multi-center U.S.-based registry hosted by the University of Pennsylvania Health System, serving as a clinical data repository for cardiac arrest and post-arrest care. The PCAS protocol followed by the three tertiary care urban teaching hospitals includes early goal-directed hemodynamic optimization (CVP 8–20 mmHg; MAP 80–100 mmHg; ScvO2 ≥65 %) addressed in an algorithmic fashion. Approval was obtained from the Institutional Review Board at our university prior to data collection. Patients were included in the study if they qualified for our health system’s hypothermia protocol (previously described) [13] after out-of-hospital or in-hospital cardiac arrest or were transferred to our health system within 6 h of return of spontaneous circulation (ROSC). Exclusion criteria were initiation of emergency cardiopulmonary bypass (ECPB), sepsis as etiology of arrest, neurological emergency leading to cardiac arrest (i.e., stroke, bleed, trauma), late transfer to our institution (>6 h), or preexisting Do Not Resuscitate order (Fig. 1).

Fig. 1
figure 1

Study flow chart. PCAS Post-cardiac arrest syndrome, PATH Penn alliance for therapeutic hypothermia, TH Therapeutic hypothermia, SAH Sub-arachnoid hemorrhage, CVA Cerebral vascular accident, CSI Cerebral spine injury, OSH Outside hospital, ROSC Return of spontaneous circulation, ECPB Emergency cardiopulmonary bypass, CABG Coronary artery bypass graft, GI Gastrointestinal

The PCAS protocol includes the following care bundle: (1) ≥2 L cold normal saline peripheral infusion accompanied by application of external cooling body wraps (Stryker Industries, Kalamazoo, MI, USA) to achieve a target temperature of 33 °C (range 32–34 °C) as soon as possible after ROSC; the target temperature is maintained for 24 h followed by active rewarming at 0.33 °C to 0.5 °C/h, with a goal of rewarming over a minimum of 8 h; (2) early percutaneous coronary intervention (PCI) if indicated; (3) continuous hemodynamic monitoring with arterial and oximetric triple lumen central venous catheters for CVP, MAP, and ScvO2 monitoring; (4) low tidal volume (6–8 cc/kg) ventilator management; (5) evaluation for relative adrenal insufficiency; (6) glucose control with a goal of <150 mg/dL; and (7) seizure management and neuromonitoring with continuous electroencephalography and bispectral index monitoring.

Vasoactive agents included dobutamine, norepinephrine, dopamine, vasopressin, epinephrine, phenylephrine, milrinone, esmolol, lopressor, sodium nitroprusside, and nitroglycerin. For the purposes of analysis, patients were considered as requiring a vasoactive agent if they were on any agent at 1, 6, 12, or 24 h after ROSC. In addition, we performed a sub-group analysis limiting vasoactive agents to traditional vasopressors and inotropes (dobutamine, norepinephrine, dopamine, vasopressin, epinephrine, phenylephrine, and milrinone).

Patients were admitted to a variety of ICU settings, and hemodynamic data were recorded on standardized flow sheets by registered nurses as part of their patient care duties. MAP and vasoactive agent data were gathered by retrospective chart review. Time variables are reported at 1, 6, 12, and 24 h after ROSC. The closest value available to a time point was recorded (±2 h range). The primary outcome was survival to hospital discharge; secondary outcome was cerebral performance category (CPC) scores, which were directly extracted or inferred from daily primary team and neurology progress notes, discharge summaries, and physical/occupational therapy assessments. A CPC 1 or 2 was considered a good neurological outcome. Data were analyzed using logistic regression analysis, controlling for arrest-level factors (initial rhythm, no/low flow time, bystander CPR, location of arrest), as well as patient-level factors (age, gender, race), independent sample Student’s t test and two-way ANOVA for repeated measures. Univariate analysis was completed to assess for association. Logistic regression was then completed using variables from univariate analysis that were significant (p < 0.1). No variables were forced into the model. Regression results are reported as odds ratios with a 95 % confidence interval; significance was achieved if the confidence interval did not cross 1.00. For ANOVA, cutoff for main effect and higher order term significance was p = 0.05 and 0.2, respectively. Statistical analysis was completed using STATA 12 software (College Station, TX, USA).

Results

Demographics

A total of 201 consecutive comatose PCAS patients underwent TH in the study period and were screened for inclusion. Of those, 33 were excluded because of delayed transfer from an outside hospital, neurologic etiology of arrest, or early implementation of ECPB, resulting in 168 patients meeting study inclusion criteria. By 24 h, 151 patients remained in the PCAS protocol (Fig. 1). Mean age of patients was 58 years; 57 % were male and 49 % African American (Table 1). Overall, 45 % (75/168) of patients survived, and 35 % (58/168) were CPC 1–2; and among survivors, 77 % had a good neurological outcome at hospital discharge.

Table 1 Demographic characteristics, co-morbidities and arrest variables

Cardiac arrest characteristics

The majority (61 %) of arrests were of cardiac etiology, 77 % (129/168) occurred out-of-hospital or in the ED, 81 % were witnessed, and 20 % of OHCA arrests received bystander CPR (an additional 30 % of OHCA received CPR immediately by a health care professional). Initial rhythm was pulseless ventricular tachycardia/ventricular fibrillation (pVT/VF) in 38 %, pulseless electrical activity (PEA) in 43 %, and asystole in 19 %. Mean times to ROSC for those with witnessed and unwitnessed CA were 22 and 38 min, respectively (Table 1).

Hemodynamic variables, logistic regression, student’s t-test, and ANOVA results

MAP at 1, 6, 12, and 24 h post-ROSC were normally distributed with mean values of 89 ± 28, 92 ± 22, 85 ± 18, and 81 ± 17 mmHg, respectively. Survivors had a higher MAP than non-survivors at all time points as determined by independent sample Student’s t-tests: 1 h: 96 ± 3 vs. 84 ± 3 mmHg (p = 0.008); 6 h: 97 ± 2 vs. 88 ± 2 mmHg (p = 0.005); 12 h: 88 ± 2 vs. 82 ± 2 mmHg (p = 0.051); and 24 h: 85 ± 2 vs. 78 ± 2 mmHg; p = 0.012). At all the analyzed time points, over half of patients required a vasoactive agent (ranging from 50.6 to 58.7 %), and more then 20 % of this cohort required ≥2 agents (ranging from 20.6 to 26.1 %). Twenty-four patients (14 %) required agents to lower elevated MAPs into the target range. A majority (13/24) of these patients also required a vasopressor during one of the other time points to achieve the goal MAP for our PCAS bundle of 80–100 mmHg. Of those receiving an antihypertensive, 67 % (16/24) survived, and 75 % (12/16) had a CPC 1–2. Among the 11 patients purely requiring antihypertensives (never receiving a vasopressor or inotrope), 73 % (8/11) survived, and 75 % (6/8) had a CPC 1–2.

Univariate analysis was completed to determine the association between outcome and arrest-level factors (initial rhythm, no/low flow time, bystander CPR, location of arrest) and patient-level factors (age, gender, race). When accounting for shockable initial rhythm, no/low flow time and witnessed cardiac arrest, increasing vasopressor use was negatively associated with survival at all time points (OR 0.42, 95 % CI 0.27–0.67 at 1 h; OR 0.44, 95 % CI 0.28–0.69 at 6 h; OR 0.66, 95 % CI 0.45–0.96 at 12 h; and OR 0.55, 95 % CI 0.38–0.80 at 24 h). Increasing vasopressor support was negatively associated with good neurologic outcome when adjusting for potential confounders (no/low flow time, bystander CPR, race and initial shockable rhythm) (OR 0.23, 95 % CI 0.11–0.46 at 1 h; OR 0.43, 95 % CI 0.25–0.75 at 6 h; OR 0.61, 95 % CI 0.39–0.97 at 12 h; and OR 0.54, 95 % CI 0.35–0.83 at 24 h) (Fig. 2).

Fig. 2
figure 2

Increasing use of vasoactive agents is associated with mortality at all time points and poor neurological outcome [cerebral performance category (CPC)] at early time points [reported as odds ratios (OR) per vasoactive agent]

Because the effect of vasoactive agents was highly significant in prior logistic regression, a two-way ANOVA for repeated measures was completed in order to account for vasopressor effects on MAP. We analyzed the MAPs of 154 patients who survived the first 24 h of therapy, and determined that survivors had a higher MAP than non-survivors at all time points except 12 h post-ROSC: 1 h (96 vs. 84 mmHg, p < 0.0001), 6 h 96 vs. 90 mmHg, p = 0.014), and 24 h (86 vs. 78 mmHg, p = 0.15) (Fig. 3). Next, we evaluated the MAP for patients requiring and not requiring vasoactive agents in relation to survival as a two-way factorial analysis of variance. Among survivors, there was no statistically significant difference in MAP at any time point whether or not vasoactive agents were required. Among those requiring vasoactive agents, survivors had higher MAPs than non-survivors at 1 h (97 vs. 82 mmHg) and 6 h (94 vs 87 mmHg) post-ROSC (p = <0.0001, p = 0.05) (Fig. 4). Identical results were seen when analyzed with good neurologic outcomes as the primary end-point. For example, patients requiring vasopressors during admission who were discharged with CPC 1–2 had higher MAPs than those with CPC 3–5 at 1 h (97 vs. 84 mmHg, p = 0.001) and 6 h (95 vs 88 mmHg, p = 0.04) post-ROSC. This effect was not seen at 12 and 24 h post-ROSC for either outcome. Among those not requiring vasoactive agents, there was no difference in MAP between those with good and poor outcomes.

Fig. 3
figure 3

Survivors have higher mean arterial pressure (MAP) than non-survivors at all time points except 12 h post-return of spontaneous circulation (ROSC)

Fig. 4
figure 4

Interaction between vasoactive agents, mean arterial pressure (MAP), and survival

Discussion

In this retrospective, single-center study, we investigated the relationship between MAP at various time points during the first 24 h post-ROSC, vasoactive agents, and survival as well as neurological outcomes for PCAS patients and found that survivors had a significantly higher MAP than non-survivors and increasing use of vasoactive agents was strongly associated with mortality at all time points and lower CPC scores. At all time points, there was no difference in mean MAPs of survivors whether or not they required vasoactive agents. Further, those not requiring vasoactive agents had the same mean MAPs whether or not they survived. Thus, the difference in mean MAP values between survivors and non-survivors was driven by the group of non-survivors who required vasopressors yet had lower mean MAP values than the rest of the patients.

These findings can be interpreted that a higher MAP is a surrogate for hemodynamic stability and therefore improved survival. However, this does not explain the finding that there is a group of hemodynamically stable patients that do not require vasoactive agents but expire and a group of hemodynamically unstable patients that require vasoactive agents and survive. Instead, our results suggest that achieving goal MAP is necessary but not sufficient for survival.

In the TH era, there have been limited studies of early goal-directed hemodynamic optimization. Gaieski et al. [13] demonstrated the feasibility of implementing TH and hemodynamic optimization simultaneously, while Sunde et al. [14] demonstrated superior results of a bundled care package including hemodynamic goals when compared to historic controls. Cardiac arrest patients with cardiogenic shock treated with TH had higher MAPs and decreased inotropic requirements compared with historic controls [15]; similarly, others have demonstrated that TH has a stabilizing effect on hemodynamic parameters of cardiac arrest patients both with [16] and without cardiogenic shock [17].

Influenced by the work of Rittenberger et al. [18], we interpret our findings to support the hypothesis that there are different types of PCAS patients who can be classified by their combined degree of neurologic and cardiovascular injury. Although these patients fall along continua of neurologic and cardiovascular injury, they can be divided into four distinct categories of patients at presentation and in relation to outcomes (Fig. 4): (1) moderate neurological and mild cardiovascular injury (survivors without vasoactive agent requirement); (2) severe neurological and mild cardiovascular injury (expired without vasoactive agent requirement); (3) moderate neurological and moderate to severe cardiovascular injury (survived with vasoactive agent requirement); and (4) severe neurological and moderate to severe cardiovascular injury (expired with vasoactive requirement).

These heterogeneous PCAS patients may benefit from different interventions. For example, it is possible that those with severe cardiovascular injury would benefit from early catheterization to determine the need for PCI or coronary artery bypass graft surgery, in addition to hemodynamic optimization with fluids and vasoactive agents, increased use of intra-aortic balloon pumps to augment coronary and cerebral perfusion, and selective implementation of ECPB for refractory shock. Those with severe neurological injury may benefit from therapeutically-induced mild hypertension to increase cerebral perfusion and mitigate cerebral injury, longer periods of cooling to optimize treatment of delayed reperfusion injury, and prophylactic as well as therapeutic anticonvulsant therapy [1719].

Another finding of clinical interest is the timing of hemodynamic interventions. The time period encompassing “early” goal-directed hemodynamic optimization is not defined for PCAS patients. In this study, those responsive to vasoactive agents at 1 and 6 h post-ROSC had better survival and CPC scores; this did not hold true for the 12 or 24 h marks. Linear regression data for use of vasoactive agents also indicate that the early time points of 1 and 6 h were more important for survival and CPC 1–2 outcomes. These data support the AHA recommendations of a minimum MAP of ≥65 mmHg, beginning in the pre-hospital setting. In the era before TH, Laurent et al. [10] found PCAS patients had significant reversible (over 24 h) hypotension beginning 4–7 h after ROSC, occurring in conjunction with the development of reperfusion-related myocardial stunning. Their study found an association between low cardiac index and early death by multiorgan failure, but no association between hemodynamic instability and poor neurological outcomes. In contrast, one large retrospective study by Trzeciak et al. [19] demonstrated that ≥1 event(s) of hypotension within 1 h of ICU arrival was associated with increased mortality and worse neurological outcomes in post-arrest patients prior to the TH era. Our study helps to identify the first 6 h as most critical and advocates for early goal-directed hemodynamic optimization in the first hours post-ROSC for PCAS patients.

The findings of this study suggest benefits from a goal MAP that is higher than the current AHA and ILCOR guidelines. Among patients requiring vasoactive agents, survivors had higher MAPs than non-survivors. Furthermore, the MAPs were higher, at 97 versus 82 mmHg at 1 h, and 94 versus 87 mmHg at 6 h post-ROSC, than the minimum of 65 mmHg recommended in the AHA and ILCOR guidelines. This suggests that high—perhaps higher than originally thought—MAPs are associated with improved survival for those requiring vasoactive agents and undergoing TH. There is a precedent for and prior evidence of neurological benefits of hypertension in PCAS. Although both landmark TH trials excluded patients presenting in cardiogenic shock [1, 2], the Bernard et al. study maintained patients at a goal MAP of 90–100 mmHg with either epinephrine or nitroglycerin. The HACA study does not mention hemodynamic goals. In a retrospective review of PCAS patients, Mullner et al. [20] found that patients with a higher MAP averaged over the first 2 h post-ROSC, but not 5 min post-ROSC, had better neurological outcomes following a protocol of epinephrine administration to a goal MAP of 75–100 mmHg. There is also animal model evidence that hypertension may be of neurological benefit in PCAS. In a dog model for comatose survivors of cardiac arrest, Safar et al. [21] found that an experimental group treated with mild hypothermia, hemodilution, and hypertension had improved neurological outcomes compared with controls treated with normothermia and normotension. Our study suggests a higher minimum MAP may be indicated than the >65 mmHg suggested by AHA guidelines and the 65–90 mmHg range suggested by the ILCOR guidelines, especially in the early hours of PCAS.

It is possible that a MAP sufficient for survival is not adequate for good neurological outcome. However, we did not find this to be the case in this study. The mean MAP associated with survival and CPC 1–2 was the same, 97 mmHg, at 1 h post-ROSC, and 94 (survival) and 95 (good neurological outcome) at 6 h post-ROSC. Further, a balance needs to be struck between the optimal MAP for patients needing early PCI and those with severe anoxic encephalopathy needing optimized cerebral perfusion [9, 22].

Limitations

The retrospective single-center nature of the study limits analysis; this was necessary as a starting point because no other retrospective, prospective, or randomized control trial has been published to date focusing on the interaction of TH and early goal-directed hemodynamic optimization in PCAS patients. Clinicians are left to determine resuscitation end points that are not based on data from the disease population they are treating. In addition, although we know that all patients had arterial and oximetric triple lumen central venous catheters for CVP, MAP, and ScvO2 monitoring placed within the first few hours after ROSC, strict adherence with the recommended post-arrest protocol goals is at the attending physician’s discretion, undocumented protocol violations are possible, and changes in vasoactive medications may have occurred based on prognosis. Another limitation is the non-continuous hemodynamic variables recorded at 1, 6, 12, and 24 h post-ROSC. Utilizing continuous monitoring that could be downloaded and analyzed may have expanded our findings. Mean arterial pressure may be too crude a measurement to target in post-arrest patients, and it may be more clinically beneficial to examine the relationship between increasing or decreasing MAP and other tissue perfusion variables including CI, regional blood flow, and microcirculatory flow [23]. Further, although the majority of patients had a CPC score assigned during their hospitalization by the managing clinicians, researchers retrospectively assigned values to those for whom none was available. These values may not be accurate but do correspond accurately with the discharge location (home, rehabilitation, nursing home) of the patients. Finally, this study does not address a fundamental question facing clinicians taking care of individual PCAS patients: what should my target MAP be for optimal outcome in this specific patient? Is a MAP of 95 mmHg better than 85 or 75 or 65 mmHg, for any, all, or some patients? Should we be using agents that decrease blood pressure to achieve target MAPs when the patient’s MAP is elevated, or is this harmful? Should we a priori classify patients into the four categories described above, taking into account the individual patient’s degree of neurologic and cardiac dysfunctions, and develop unique resuscitation strategies for each of these groups? We recommend that future prospective trials incorporate this study’s findings and make use of continuous time data and specific hemodynamic goals while investigating survival and neurologic outcomes.

Conclusions

In conclusion, this retrospective single center study suggests that (1) vasoactive agent use in PCAS patients is, overall, associated with mortality and poor neurological outcomes; (2) those requiring vasoactive agents are more likely to survive and have good neurological outcomes if they respond to vasoactive agents at early hours post-ROSC as evidenced by higher MAP values; (3) there may be several types of PCAS patients that could be categorized by the degree of neurological and cardiovascular injury and these patient groups may benefit from tailored therapy; and (4) those with good outcomes have MAPs in the 90–100 mmHg during the first 6 h post, ROSC.