Introduction to evidence-based critical care medicine
Evidence based medicine (EBM) is an approach to caring for patients that involves the explicit and judicious use of the clinical research literature combined with an understanding of pathophysiology, our clinical experience and patient preferences to aid in clinical decision making. While many may argue that this definition is what physicians have always done, we believe that the explicit, consciousness reasoning with this approach may improve clinical decision making by 1) incorporating the best available scientific literature, 2) reducing bias that occurs when we base our medical decision making on our most recent patients, 3) explicitly weighing the risks and benefits of a clinical decision, and 4) incorporating patient preferences into our risk/benefit assessment.
While EBM has been championed in internal medicine, there exists a great need to extend the evidenced based approach to other disciplines including critical care. For example, during rounds we will frequently hear house staff and fellows say that the patient is volume overloaded based on either physical exam or pulmonary artery catheter data without knowledge of the sensitivity, specificity, or likelihood ratio of that particular test. That is, we often assume a test is 100% sensitive and specific and thus, if the result is positive or negative, the diagnosis is ruled in or out. The link between the pretest probability, the likelihood ratio of a test, the resultant posttest probability and our treatment threshold often resides in our subconscious rather than in an explicit statement. By encouraging physicians to explain consciously and explicitly their medical decision-making, including the literature on which their decisions were based, we may improve our clinical decision making. Often times this means stating that there is no evidence for a particular decision but acknowledging this and using pathophysiologic reasoning helps clinical decision making. Although the need for EBM occurs throughout medicine, the need for evidenced based critical care is increasing.
Critical care is a relatively new field but is growing rapidly and now consumes 1% of the gross domestic product in the United States1. With the aging of the population, the resources devoted to critical care will likely increase. Despite this investment in resources, most therapies have not been rigorously evaluated.2 Indeed, many decisions in the ICU are based on pathophysiologic reasoning often obtained from animals or healthy volunteers. It is this type of reasoning that demands explicit consideration of patient preferences, and risk and benefits including the risk of being a false positive (ie. extubating somebody who will fail extubation) or a false negative (i.e. leaving somebody intubated who could be extubated).
Dispersed literature sources, poorly defined diseases, and clinical research challenges all increase the need for evidenced based critical care. In addition, the need to make decisions rapidly may complicate the application of EBM in the ICU setting.
Transdisciplinary nature of critical care leads to dispersed literature
Critical care is truly transdisciplinary in that it is practiced by internists, pediatricians, anesthesiologists, and surgeons, each with a particular focus based both on prior training and on particular aspects of the patients they treat3. Indeed, the body of literature pertaining to critical illness can be found in general medical journals, medical subspecialty journals, critical care journals, anesthesiology journals, pediatric journals, and surgery journals. The diverse sources of clinical research information in critical care increases the need for efficient methods to access and search relevant literature. Many clinicians rely on alternative strategies for accessing relevant clinical literature such as evidence summaries (Intensive Care Monitor) that screen journals relevant to critical care and summarize the information to facilitate critical appraisal. Both the breadth of knowledge required for critical care and the diverse sources of clinical literature, demand efficient access and evaluation of relevant literature.
Critical illness poorly defined
Many of the diseases we treat as critical care physicians are poorly defined. For example, the diagnosis of sepsis, a common syndrome in critical care, is controversial and still debated4, 5, 6. The variability in the diagnosis of this disease and the ensuing misclassification of patients enrolled in clinical trials could bias these studies. Additionally, there are often multiple etiologies for a particular critical illness and there may be an interaction between the etiology and the treatment. For example, sepsis due to a primary pulmonary infection may be different than sepsis from infected pancreatic necrosis. Since there may be an interaction between disease etiology and intervention, stratified analyses are often performed but these increase the sample size needed to detect a treatment effect in each subgroup as well as the expense of the trial. Due to these issues, the need to evaluate the quality of clinical literature in critical care becomes paramount.
Clinical research challenges
Critical care is a relatively new specialty7. This, combined with the transdisciplinary nature of critical care and the imprecise definition of critical illness has lead to a poorly defined research agenda with the resultant lack of detailed knowledge about several clinical problems. Critical care presents several unique challenges for clinical research. To formulate a research question, an investigator must be able to define a patient population, treatment or intervention, and outcome. Operationalizing these variables especially a diseased patient population, for critical care clinical research is complex. Compare the investigator studying outcomes after myocardial infarction to one studying the same research question in sepsis. Myocardial infarction is a common disease that is readily diagnosed with several laboratory tests and thus it is easy to define a population of patients with the disease. Patients with myocardial infarction receive treatments that are relatively specific for the diagnosis and are generally cared for by cardiologists, internists or family physicians. Sepsis, on the other hand, is defined by clinical criteria which continue to evolve and that does not have an accepted diagnostic test or unique treatment, and thus it would be difficult to define a diseased population or to assemble a cohort at the same stage in the disease process.4, 5, 6 Critically ill septic patients may receive their care from a variety of physicians including surgeons, anesthesiologists, family physicians, general internists, pulmonologists, and pediatricians. Critical care is a challenge to the clinical researcher precisely because the key variables of disease, patient population, disease stage, therapy, and provider are difficult to define and these challenges increase the risk for bias. We need to increase the number of valid clinical trials in critical care medicine. However, randomized clinical trials are sometimes difficult in critically ill patients, and the results of randomized clinical trials may not be generalizable. Therefore, we may need observational studies to supplement clinical trials to inform clinical practice.
Potential limitations of applying the results of clinical trials to populations
Many clinicians view the "randomized double blind placebo controlled clinical trial" as a seal for a high quality trial and accept ex officio that the inferences from the study are "true". A valid randomized clinical trail is our most powerful tool to evaluate a therapy (is a better than b). Nevertheless, the quality of the trial can significantly impact the validity of the studies inferences. Additionally, we sometimes need observational studies because 1) the results of clinical trials may not be generalazible to the entire population, 2) some clinical questions are more appropriately addressed by observational studies, and 3) clinical trials in critically ill patients are sometimes difficult to perform for ethical or logistic reasons. This book contains guidelines on how to evaluate whether the results of a clinical trial apply to your individual patient or whether your patient is so different that the study results would not apply. In most cases, the answer is usually a quantitative difference (one of degree) rather than a qualitative difference (a different conclusion). This book also provides methods to estimate this quantitative difference. Nevertheless, we must be cautious about assuming we can replicate the results of a clinical trial on a population of patients when creating a practice guideline, or performing a decision analysis or a cost-effectiveness analysis. Most clinical trials, are efficacy studies in that they evaluate how something works in an artificial world where we control multiple variables. While this control is necessary to maintain internal validity of the study, it may limit our ability to replicate the results of the clinical trial on a population of patients outside the trial. On the other hand, effectiveness studies evaluate how a treatment works "in the real world" and have less internal validity but have increased generalizability or external validity. This difference between efficacy studies and effectiveness studies and the potential problem of assuming we can replicate results of a randomized clinical trial for a population of patients is illustrated in the carotid endarterectomy story. The Asymptomatic Carotid Atherosclerosis Study(ACAS) stated that the 30-day mortality for asymptomatic patients having carotid endarterectomy was 1/10008. However, an effectiveness study of Medicare patients revealed that the mortality was 1/100, 10X higher than in the clinical trial9. This difference in mortality is likely due to the strict entry criteria for the clinical trial. Patients with comorbid diseases and patients over 79 years old were excluded from the clinical trial. The carotid endarterectomy story demonstrates the need for observational studies in addition to randomized trials when creating practice guidelines or performing decision analysis or cost effectiveness analysis. If we assume we can replicate the results of a RCT on a population of patients, we must evaluate whether in deed we are achieving the same results.10 This difference between efficacy and effectiveness studies may be particularly large in critical care where there is wide variability in the organization of ICUs and this variability is associated with in-hospital mortality.11 Additionally, as we change the way health care is organized, financed, and delivered, it is important to evaluate the impact of these changes on patient care. Yet, it is difficult to evaluate these types of changes with a randomized clinical trial. For example, we may want to evaluate the impact of adding intensive care physicians to an ICU on in-hospital mortality and complication rates. While a randomized clinical trail of this organizational change may be difficult, we can perform an observational study. Moreover, we need to evaluate the impact of these changes on long-term patient outcomes and this evaluation would be difficult to do with a RCT. For example, critical care researchers have largely failed to explore the long-term outcomes associated with ICU care. Without explicit knowledge about what is occurring in "reality", it may be difficult to make optimal clinical decisions that best meet the needs of the patient, the family, and society.
Need for rapid decisions
Furthermore, the need to make rapid clinical decisions may complicate the applicaton of EBM in the ICU setting. Recently, Sackett et al explored the ability of an evidence-cart to enhance the use of the literature in patient clinical decision-making. This work was partly driven by the concern that EBM, though attractive in theory, may be limited in application simply because of the time and effort required to systematically appraise the literature before a decision is made. Though their study showed such a cart may help, it was conducted on a general medical ward. In the ICU, decisions may need to be made even faster than in the normal care of hospitalized patients. In such a situation, there is little data regarding whether EBM can be applied practically. There are some data suggesting that pre-scripted information (e.g., EBM-based treatment and diagnostic protocols) may be used effectively even in the treatment of acute processes such as ARDS12. But, it is likely that, without such support systems, practical applications of EBM in the ICU will be limited to non-time critical decisions.
Future Directions of Evidenced Based Critical Care
Patients, payers and insurers are demanding that we practice evidenced based medicine and this trend will likely increase. Additionally, several countries now have projects to create limited data sets for ICU patients that can we used to help practice evidenced based medicine13. For example, data from these data sets can be used to create empiric probability distributions for pretest probabilities. We must remember that EBM is an approach to patient care that explicitly and judiciously incorporates the best evidence from clinical research with our clinical judgement and patient preferences to help make patient care decisions. Critical care is ripe for the practice of EBM because we frequently diagnosis, prognosis, treat, and weigh risks and benefits of clinical decisions and these decisions often have significant impact of patients, their families, and the cost of care. By making the clinical decision making processes explicit, incorporating the best available evidence and patient preferences, EBM can help improve the care we provide to patients.
- Jacobs P, Noseworthy TW: National estimates of intensive care utilization and costs: Canada and the United States. Crit Care Med 1990:18;1282-1286.
- Smith R. Where's the widsom: The poverty of medical evidence. BMJ 1991;303:798-799.
- Cook DJ, Levy MM. Evidence-Based Medicine: A Tool for Enhancing Critical Care Practice. Critical Care Clinics 1998:14; 353-57.
- Bone RC. Sepsis, the sepsis syndrome, multi-organ failure: a plea for comparable definitions. Ann Intern Med 1991: 114:332-333.
- Bone RC. Toward an epidemiology and natural history of SIRS (systemic inflammatory response syndrome) JAMA 1992:268;3452-3455.
- Vincent JL. Dear SIRS, I'm sorry to day that I don't like you. Crit Care Med 1997:24:372-374.
- Kelly MA. Critical Care Medicine- a new specialty. N Engl J Med. 1988;318:1613-1617.
- Executive Committee for the Asymptomatic Carotid Atherosclerosis Study. Endarterectomy for asymptomatic carotid artery stenosis. JAMA 1995;273:1421-1428.
- Wennberg DE, Lucas FL, Birkmeyer JD, Bredenberg CE, Fisher ES. Variation in Carotid Endarterectomy Mortality in the Medicare Population. JAMA 1998; 279:1278-1281.
- Groeger JS, Strosberg MA, Halpern NA, et al: Descriptive analysis of critical care units in the United States. Crit Care Med 1992;20:846-863.
- Pronovost PJ, Jenckes, MW, Dorman T, Garrett E, Breslow MJ, Rosenfeld BA, Lipsett PA, Bass EB. Organizational characteristics of Intensive care units related to outcomes of abdominal aortic surgery. JAMA 1999;281:1310-1317.
- East TD. Bohm SH. Wallace CJ. Clemmer TP. Weaver LK. Orme JF Jr. Morris AH. A successful Computerized protocol for clinical management of pressure control inverse ratio ventilation in ARDS patients. Chest. 1992; 101:697-710.
- Le Gall, JR. Results of the pilot study. European Consortium of intensive care data. 1996.