Mild therapeutic hypothermia shortens intensive care unit stay of survivors after out-of-hospital cardiac arrest compared to historical controls

Persistent coma is a common finding after cardiac arrest and has profound ethical and economic implications. In a significant proportion of patients, neurological status rather than specific treatment of the underlying disease affects the outcome after cardiac arrest [1]. Recent randomized controlled trials have demonstrated that therapeutic hypothermia is highly effective in improving the neurological outcome in patients after cardiac arrest [2, 3]. In 2003, the International Liaison Committee on Resuscitation (ILCOR) recommended this treatment for all comatose survivors of out-of-hospital cardiac arrest due to ventricular fibrillation [4]. Although only a minority of such patients are currently treated with therapeutic hypothermia [5, 6], recent efforts aim to implement therapeutic hypothermia as a routine procedure in patients after cardiac arrest [7].

The optimal method for controlled and safe application of therapeutic hypothermia is still under debate [8, 9]. For intravascular cooling devices as well as for device-controlled surface cooling methods, the efficacy has been demonstrated in different trials [10‐13]. Of course, other cooling methods like crushed-ice, towels pre-soaked in ice water, or simple cooling blankets may be very effective as well but the temperature range is more difficult to control [14, 15]. For out-of-hospital cooling, new devices and methods for fast induction of hypothermia are of increasing importance [16, 17]. However, using advanced cooling methods, equipment, and manpower required for the application of therapeutic hypothermia generates higher treatment costs. Length of intensive care unit stay (ICU LOS) has been identified as a major determinant of total treatment costs after survived cardiac arrest [18]. Therefore, it was investigated whether therapeutic hypothermia influences ICU LOS and time of mechanical ventilation in patients after out-of-hospital cardiac arrest.

The study protocol was approved by the local ethics committee on human research. The need for informed patient consent was waived by the committee. We conducted our study in an urban area with a two-tiered medical emergency system: basic life support, including automated defibrillation, is offered by ambulances, and advanced life support procedures were performed by qualified emergency physicians at the mobile ICU. All patients with cardiac arrest and return of spontaneous circulation (ROSC) in the field were directly admitted to our medical ICU (MICU). Pre-hospital cooling procedures were not applied during the study period.

Between January 2006 and January 2007, a total of 52 patients were admitted to our MICU after out-of-hospital cardiac arrest. All patients received therapeutic hypothermia according to the current ILCOR recommendations. A historical control group in the era prior to hypothermia treatment was identified in a cohort of 74 patients admitted to our MICU between 2003 and 2005 after out-of-hospital cardiac arrest. Detailed characteristics for all study patients are presented in Table 1.

Therapeutic hypothermia was initiated after admission with an intravenous infusion of cold saline (4°C, 1,000 to 1,500 mL bolus) followed by surface cooling with commercially available non-invasive devices (CritiCool^®; MTRE, Yavne, Rehovot, Israel, or ArcticSun2000^®; Medivance, Inc., Louisville, CO, USA). After the start of cooling, the target temperature of 33°C was reached in a range of 180 to 300 minutes and maintained for 24 hours. Re-warming was performed at a rate of 0.25°C per hour. Intravenous sedation was induced in all patients by a combination of midazolam (0.125 mg/kg per hour) and fentanyl (0.002 mg/kg per hour) with dose adjustment as needed. Patients undergoing hypothermia received muscle relaxation with repetitive pancuronium (0.1 mg/kg) administration in order to prevent shivering. All patients completed therapeutic hypothermia without complications or overcooling. With the exception of therapeutic hypothermia, there was no further difference in the standard of intensive care treatment between the two groups. To discover early-onset infection and pneumonia, laboratory markers of systemic inflammation (c-reactive protein/leukocyte count) and radiological findings of pneumonia in chest x-ray were analyzed on day 3. Neurological outcome was assessed at the time of discharge from the ICU according to the Pittsburgh cerebral performance category (CPC) score [19]. CPC 1 and 2 were classified as a favorable neurological outcome, whereas CPC 3 and 4 were regarded as an unfavorable outcome.

Mild therapeutic hypothermia is recommended by current ILCOR guidelines and has become an essential part in post-resuscitation care. Although the benefit concerning neurological outcome could not be demonstrated in a few single centers, most other trials could show an improved neurological outcome under hypothermia even in unselected effectiveness trials [20, 21]. Nevertheless, technical, logistical, and financial barriers may limit the transfer into daily practice [22]. Therefore, for the first time, the present study investigates the impact of therapeutic hypothermia on ICU LOS in patients with out-of-hospital cardiac arrest. The major determinants of short ICU LOS and ventilator time in patients after cardiac arrest were found to be either early death during ICU stay or rapid neurological recovery. While comparison of all patients (survivors and non-survivors) did not reveal a statistical difference regarding ICU LOS, subgroup analysis showed that short ICU LOS is related to an improved neurological outcome in the hypothermia group. In contrast, short ICU LOS in the control group was associated with a high rate of non-survivors. This striking result was confirmed with a multivariate regression model emphasizing the impact of therapeutic hypothermia on ICU LOS.

A high APACHE II score at admission was associated with a prolonged ICU stay in our study. It seems plausible that patients with higher severity of illness depend more on critical care facilities; however, it is known that the predictive value of the APACHE II score has various limitations after cardiac arrest [23, 24]. Our present results suggest that prognostic application of current scoring systems may be even more difficult in patients treated with therapeutic hypothermia. Both ICU LOS and time of mechanical ventilation tended to be longer in non-survivors of the hypothermia group compared with non-survivors of the control group, which may reflect the difficulty of outcome prediction in patients routinely requiring sedation and muscle relaxation in the first days of intensive care. Beside clinical evaluation, biochemical markers (neuron-specific enolase, protein S100B) and electrophysiological studies (somatosensory-evoked potentials) are established tools for outcome prediction in patients after cardiac arrest [25, 26]. However, most of these parameters were studied in patients not undergoing hypothermia treatment. At present, little data on whether therapeutic hypothermia treatment may influence these biomarkers are available [27, 28]. Therefore, physicians should be careful in the prognostication of patients treated with therapeutic hypothermia [29].

It should be mentioned that our analysis has various limitations. First of all, results obtained with an observational study design using a historical control group generally require further validation by a randomized controlled trial. However, ethical concerns will most likely prevent a further randomized trial that withholds hypothermia treatment from a control group. Second, our local treatment standards may have affected ICU LOS and patient outcome. There is considerable variation regarding end-of-life decisions and care practices. Thus, the approach used by a specific ICU on deciding whether to withhold or withdraw critical care in patients after cardiac arrest may have an important impact on ICU LOS. This may help to explain data from a Canadian survey reporting that patients with a higher Glasgow Coma Scale score had a longer ICU stay [30].

Moreover, differences in ICU LOS may also be related to a different use and allocation of ICU capacities in general. Decision making potentially may be influenced by a variety of factors that are independent of the individual patient characteristics, such as the availability of ICU beds and/or of facilities of 'step-down' care. It is of interest to note, however, that the necessity of mechanical ventilation, which is a rather objective criterion for the necessity of ICU treatment, is reduced by hypothermia treatment. Since impaired neurological status has been identified as a predictor of extubation failure or weaning failure [31, 32], this finding may be associated with the more favorable neurological outcome in the hypothermia group.

Another issue with impact on ICU LOS and time on ventilator might be systemic inflammation after cardiac arrest. Both infectious and non-infectious systemic inflammations are a frequent problem in these patients [33, 34]. As demonstrated in the HACA (Hypothermia after Cardiac Arrest) trial, there may even be a trend toward a higher rate of infections in patients treated with therapeutic hypothermia. However, we found no difference between the two study groups with respect to laboratory markers of inflammation (C-reactive protein/leukocyte count) and radiological findings of early-onset pneumonia. Therefore, early infection or pneumonia had no impact on different mechanical ventilator times between the groups.

In our analysis, we may have suffered from a selection bias because some baseline characteristics (bystander CPR and rate of ventricular fibrillation) tended to be more favorable in the hypothermia group although APACHE II score was lower in the control group. Finally, we cannot exclude the fact that the introduction of therapeutic hypothermia itself has focused the attention of the physicians in charge to a more sophisticated post-resuscitation care in general, which has been described [35]. However, similar promising results were also reported from a large hypothermia registry [36].

In summary, patients treated with therapeutic hypothermia showed both an impressive improvement of neurological outcome as well as an increased 1-year survival rate. Furthermore, therapeutic hypothermia did not prolong ICU stay or time of mechanical ventilation; rather, these parameters were reduced in survivors when therapeutic hypothermia was applied. Although we did not directly calculate the ICU treatment costs, we believe that this could be an additional argument for the application of therapeutic hypothermia in patients after cardiac arrest.