Evidence-based and Good Sense-based Critical Care Medicine
Critical care medicine has improved greatly, thanks to the implementation of guidelines and good clinical practice protocols. For this reason, the present clinical practice requires evidence-based medicine to answer unresolved debates and to further improve applied protocols. Randomized controlled trials are (even if not always) essential in order to achieve a high level of scientific evidence good enough to support the application of new life-saving strategies or to change significantly current practice. Such trials are now strongly needed because mortality has improved to such a level that large investments will be required to achieve relatively small improvements in clinical outcomes. Nevertheless, this approach has justified and encouraged changes in therapeutic strategies sometimes contradicted after only a few years. This not only demonstrates how difficult it is to reach a definitive consensus in the world of critical care medicine but it may also induce perplexity and diffidence among the operators of the field.
A good example of such a problem is represented by the issue of tight glycaemic control in critically ill patients. Hyperglycaemia associated with insulin resistance is common in critically ill patients, even in those who have not been previously diagnosed with diabetes. It has been reported that pronounced hyperglycaemia may lead to significant complications. In diabetic patients with acute myocardial infarction, therapy to maintain the blood glucose at a level 7< 215 mg/dL improves the long-term outcome. In nondiabetic patients with protracted critical illnesses, high serum levels of insulin-like growth factor-binding protein 1, which reflect an impaired response of hepatocytes to insulin, increase the risk of death. Critical illness-related polyneuropathy and skeletal muscle wasting are associated with hyperglycaemia and the prolonged need for mechanical ventilation. Moreover, in hyperglycaemic critically ill patients, increased susceptibility to severe infections and failure of vital organs, particularly the kidneys, occur and amplify the risk of an adverse outcome. One landmark trial in 2001 showed significantly decreased mortality in a surgical intensive care unit (ICU) targeting blood glucose to 80-110 mg/dL with intensive intravenous insulin therapy. A reduction in organ dysfunction and ICU length of stay (LOS) (from a median of 15 to 12 days) was also observed in the subset of patients with an ICU LOS > 5 days. A second randomized trial of intensive insulin therapy, following the same protocol, enrolled medical ICU patients with an anticipated ICU LOS of > 3 days. Overall, mortality was not reduced but ICU and hospital LOS were reduced and associated with earlier weaning from mechanical ventilation and less acute kidney injury (AKI). In patients with a medical ICU LOS > 3 days, hospital mortality was reduced with intensive insulin therapy. However, the investigators were unsuccessful in predicting ICU LOS and 433 patients (36%) had an ICU LOS of < 3 days. Furthermore, use of the intensive insulin strategy in the medical ICU resulted in a nearly 3-fold higher rate of hypoglycaemia than in the original experience (18% versus 6.2% of patients). Interestingly, a recent survey conducted in 29 Australian and New Zealand ICUs with 911 admissions showed that few of those ICUs had adopted intensive insulin therapy. That study could not separate insulin administration and highest daily blood glucose concentration in their association with hospital mortality. Other clinical trials performed over the last 7 years reported conflicting results about the impact on patient outcome of tight glycaemic control, and dangerous hypoglycaemia was frequently observed in tight glycaemic control regimes. The Surviving Sepsis Campaign now recommends glucose control in critically ill patients below a glucose level of 150 mg/dL. The data underlying this recommendation were critically evaluated from Wiener et al. who searched for studies in which adult intensive care patients were randomly assigned to a tight versus usual glucose control. Of 1358 identified studies from all languages, 34 randomized trials (23 full publications, 9 abstracts, 2 unpublished studies) met the inclusion criteria. Twenty-nine randomized controlled trials (RCT) totalling 8432 patients contributed data for this meta-analysis. Hospital mortality did not differ between tight glucose control and overall usual care. There was also no significant difference in mortality when stratified by glucose goal (very tight: < 110 mg/dL or moderately tight: < 150 mg/ dL) or intensive care unit setting (surgical, medical, medical-surgical). Tight glucose control was not associated with a significantly decreased risk for dialysis requirement, but was associated with a significantly decreased risk of sepsis and significantly increased risk of hypoglycaemia (glucose < 40 mg/dL). The authors conclude that in critically ill adult patients, tight glucose control is not associated with significantly reduced hospital mortality but with an increased risk of hypoglycaemia.
In a recent article, Schetz et al. reported on the combined analysis of the two Belgian studies that investigated the renal effects of intensive insulin therapy in critically ill patients. The analysis of the cumulative data showed that patients randomized to treatment with intensive insulin therapy were less likely to develop higher classes of a modified RIFLE (Risk, Injury, Failure, Loss and End-stage renal disease) criteria for AKI and were less likely to have oliguria or to develop the need for renal replacement therapy (RRT).
A large RCT is finally planned to compare glucose levels of 80-110 mg/dL versus 140-180 mg/dL and will recruit more than 6000 patients (Normoglycaemia in Intensive Care Evaluation and Survival Using Glucose Algorithm Regulation, or NICE-SUGAR). This RCT is scheduled to randomize its last patient before the end of the year 2008. The trial's primary outcome is 90-day all-cause mortality, and information on renal function is among the secondary aims of the study. The maintenance of blood glucose levels within the conventional range (110-190 mg/dL) seems justified until the results on glucose control become available.
Interestingly, not only therapeutic approaches have been re-evaluated in the recent literature, but lack of evidence also exists for some monitoring systems and their utility in ICU patients. A recent analysis was aimed at determining which monitoring techniques might improve outcomes in ICU patients. The authors retrieved about 4000 articles, of which 67 evaluated the impact of monitoring in acutely ill adult patients. There were 40 studies related to haemodynamic monitoring, 17 to respiratory monitoring and 10 to neurological monitoring. Positive non-mortality outcomes were observed in 17 of 40 haemodynamic studies, 11 of 17 respiratory and in all 10 neurological studies. Mortality was evaluated in 31 haemodynamic studies, but a beneficial impact was demonstrated in only 10. For respiratory monitoring, seven studies evaluated mortality, but only three of them showed an improved outcome. No neurological monitoring study assessed mortality. The authors conclude that there is no broad evidence that any form of monitoring improves outcomes in the ICU and that most commonly used devices have not been evaluated by RCTs. In particular, pulmonary artery catheters (PACS) have been considered for almost 40 years as a useful tool in the acutely ill patient to achieve information on intrathoracic intravascular pressures, cardiac output and mixed venous oxygen saturation (SvO2). Over the last 10 years, different trials were able to show that PACS do not improve outcome and can even be dangerous: in 1996, a multicentre observational study by Connors et al. suggested an increased mortality with PACS. As a consequence, multiple randomized trials were performed and a Cochrane Collaboration meta-analysis has recently shown that this technology has no impact on mortality in different critically ill patient populations. Recently, a time trend analysis on national estimates of PAC utilization from 1993 to 2004 was performed. A primary analysis focused on admissions with a medical diagnosis and a secondary analysis on surgical admissions. Between 1993 and 2004, PAC use decreased by 65% from 5.66 to 1.99 per 1000 medical admissions. Among the subgroup of patients who died during hospitalization, but whose disease severity was considered constant across time, the relative decline was similar, decreasing from 54.7 to 18.1 per 1000 deaths. A significant change in trend occurred following the 1996 trial. The decline in utilization was similar in surgical patients. The decline of PAC utilization was most prominent for myocardial infarction, which decreased by 81%, and least prominent for septicaemia, which decreased by only 54%. Use of the PACS was previously a hallmark of critical care practice, and this significant trend towards decreased use in the United States in the last decade is possibly due to the evidence that this invasive monitoring does not reduce mortality. A group of experts in the field, however, met and wrote an interesting point of view article, strongly criticizing these conclusions. They correctly stated that, 'If we apply the argument that there is a lack of evidence showing a mortality reduction with PAC use, then practically every monitoring technique used in critical care should be abandoned'. Furthermore, if it is true that PACS are applied less often (especially in the United States), this might mean that they have been previously overused. If European trends were analysed, they would probably be less spectacular than those described above. Finally, evidence-based medicine can be interpreted in different ways: if it is true that several RCTs have indicated that using a PAC does not influence outcomes, they did not show that PACS are either inherently dangerous or beneficial (except for Connors' study) and, so far, a PAC is still considered the gold standard of haemodynamic monitoring compared to alternative and less invasive systems. It is difficult, based on RCTs, to develop PAC-guided treatment protocols that are applicable across a heterogeneous population of acutely ill patients with complex comorbidities. Indeed, the way in which PAC findings are interpreted and used requires deep knowledge of the physiology and integration of the three PAC elements (pressures, cardiac output and SvO2): such interpretation may vary and often be inadequate among different operators. If a properly trained physician believes that invasive haemodynamic data are necessary for the management of a specific patient, then the use of the PAC is justified in that patient. The use of PACS, according to these authors, should still be considered a good medicine, and it should be utilized every time a pathophysiological rationale exists. Doctors should still be trained in PAC positioning and correct data interpretation.