| Sign In to gain access to subscriptions and/or personal tools. |
Evidence-Based Medicine for Nutrition Support: An Overview of the ProcessDepartment of Human Nutrition, University of Illinois at Chicago, Chicago, Illinois Correspondence: Correspondence: Carol Braunschweig, PhD, RD, Department of Human Nutrition, University of Illinois at Chicago, 1919 W Taylor (m/c 517), Chicago, IL 60612. Electronic mail may be sent to braunsch{at}uic.edu. On a daily basis, clinicians make decisions regarding therapies to result in the best outcome for their patients. These decisions should be based on the evidence in the literature, indicating a therapy will cause the best outcome. To facilitate this, many professional societies and scientific journals have published technical and scientific reviews, as well as evidence-based standards of care focused on many issues of nutrition support practice. This paper provides an overview of how these reviews and standards of care are derived to promote both the understanding of what they can and cannot do to enhance clinical practice. Evidence-based medicine (EBM) has been defined as "the conscientious, explicit, and judicious use of current best evidence in making decisions about the care of individual patients."1 To practice EBM, clinicians integrate their clinical expertise with the best available evidence from the medical literature to provide the most optimal, cost-effective patient care. Due to the thousands of research articles published each year, many with conflicting findings, knowing which treatment is the most effective can be a complex process. To promote EBM, many professional societies publish technical reviews and recommendations for standards of care according to a comprehensive review of the literature. The purpose of this article is to provide an overview of how technical scientific review and evidence-based guidelines are developed using examples from recent reviews and guidelines that have been published in the area of clinical nutrition.
The initial step in the development of a technical review is to formulate the questions to be addressed. This is done by establishing a workgroup with expertise in the area of concern that defines each question and the types of studies that will be reviewed for evidence. The workgroup is made up of individuals with research and clinical expertise in the area of interest and may also include stakeholders such as representatives from industry or government. The question formulation process involves defining the 4 major parts listed below, which has been termed PICO (problem, intervention, control, outcome):
The question that is ultimately formulated by the workgroup will determine the types of studies reviewed for establishing the best answer. Questions regarding treatments generally are resolved by reviewing randomized controlled trials (RCTs) that have been conducted for the treatment of interest. For example, to determine if enteral nutrition affected clinical outcomes, a recent scientific review limited the studies reviewed to RCTs.2 Questions focusing on diagnosis, screening, or prevalence of a disease/condition base their review on cross-sectional studies. Investigators attempting to determine the prevalence of malnutrition in hospitalized patients reviewed cross-sectional studies that had assessed nutrition status using various assessment tools in hospitalized patients.3 Questions involving the overall prognosis for a patient population are resolved by reviewing prospective cohort studies. These types of studies were reviewed to establish an association between malnutrition and poor outcomes.4 When questions are focused on disease etiology or harm, cohort studies (both prospective and retrospective) and case-control studies are reviewed. It is not uncommon for many questions in clinical nutrition to have few studies available for review. In this situation, the team of experts may decide to include all investigations conducted in a specific area, regardless of the study design, to derive their conclusions. The Canadian guidelines for nutrition support in intensive care units based their recommendations on a review of all types of studies conducted in this area.5 When this approach is used, the findings are graded according to a hierarchy of evidence for causation developed by Elwood and Elwood.6
Once the types of studies to be included in the review are established, the quality or degree of rigor used in each investigation is assessed. The findings from studies with higher quality are given greater weight in the review than those with lower quality. Because the purpose of this review is to describe how EBM is used to develop standards of care, which is typically done by reviewing RCTs in the area of interest, description of how quality evaluation is determined will be limited to this type of design. Quality scores for RCTs include the components listed in Table 1. Well-defined eligibility criteria for participation allow the outcomes of the intervention to be generalized to other similar populations. Concealment of randomization prevents selection bias from occurring in the allocation of participants to study groups; thus, the outcomes observed will not be due to the selection of particular patients to receive a given therapy. Blinding of investigators or participants from their treatment allocation prevents observation bias, which may arise from differences in investigator probing (interviewer bias) or differences in participants' recall of events (recall bias) if their study group is known or the exposure of interest is perceived to be either detrimental or beneficial to health status. Unfortunately, it generally is not practical to blind the treatment allocation from the investigators or the participants for nutrition interventions; thus, to the extent possible, it is important that the delivery of the intervention and outcomes assessed be objective and obtained or confirmed from an individual that was blinded to the treatment allocation. The intensity of monitoring between groups should be the same to avoid imbalances in detection of the exposures and outcomes of interest. A complete description of each outcome and the intervention provides assurance that all data are similarly obtained and that the intervention is delivered similarly to all participants. A description for the adequacy of sample size recruited in each group is needed to ensure the power of the study was adequate to detect a difference if a difference exists. The number of individuals recruited compared with the number who complete or adhere to the protocol can influence results. Intent to treat implies "as randomized, so analyzed." Analysis includes all participants initially randomized to each group, regardless of their compliance to the protocol, their exposure to the treatment, or their loss to follow-up. Use of intent-to-treat analysis prevents overstatement of the clinical effectiveness of the treatment.
The overall goal of scientific and technical reviews is to discern if those individuals exposed to the treatment in question have a better outcome than those not exposed (ie, did the treatment "cause" an improvement in outcomes). Determination of causality involves evaluating the potential for bias, confounding factors (ie, the effects of additional variables that might be responsible for the association observed), and chance to have caused the results observed via the studies' quality, as described in the previous paragraph. It also involves assessing for the presence of the positive features of causation. These include the presence of (a) biologic plausibility, (b) temporality (ie, the outcome occurred after the exposure of interest), (c) degree of association, (d) a dose-response, (e) consistency of findings in various subgroups of participants, and (f) consistency of results with other similar studies. The biologic basis for the treatment is typically provided in the studies' introduction. Discerning the temporal response (the treatment precedes the outcome of interest) is a component of evaluating the quality of an RCT. Well-designed trials provide baseline profiles of participants in the trial to demonstrate they are all either free of the outcomes of interest or similar in their distribution between groups before the intervention. The stronger the association observed between the treatment and the outcome of interest, the more likely it is to reflect a causal relationship. The dose-response is similar to the strength of the association and refers to the assumption that if there is a causal relationship, there will be an increase in outcome with an increase in exposure. The reverse direction may also support a causal relationship. Consistency of the association across subgroups (eg, men and women or across races/ethnicities) of the sample population supports causation. It is important that this type of subgroup analysis be done; however, this is often a problem in clinical trials because a very large data set may be required to have sufficient numbers within each strata for meaningful subgroup analysis. Evaluation of the studies' external validity involves determining if the results can be applied to all the eligible population and, more broadly, to other relevant populations. The primary issue is not to determine whether the participants included in the trial were "typical," but whether the association between outcome and exposure is likely to apply to other groups. Results that were observed for issues of basic physiology are generally quite easy to extrapolate to other groups, whereas those involving cultural and psychosocial results are more problematic.
To evaluate the evidence for a nutrition intervention, a data abstraction sheet is developed by the workgroup for data collected from each of the studies. This will include those items listed in Table 2; the items are used to evaluate the quality of the study and all of the data that the workgroup selected to assess the outcomes of interest and the intervention. Data extraction is tedious and can be fraught with error because the outcomes of interest are reported in different places within each article. To minimize errors in the data extracted, each article should be reviewed by at least 2 individuals, and differences in reporting should be resolved by consensus. Once the data have been extracted, they are compiled and summary recommendations are generated. Typically, a definition of the language of the recommendation is provided and the number and quality of the studies included for each intervention question are described. All materials are then presented to the workgroup. They review the individual evaluations and develop conclusion statements and evidence summaries. Once the working group is satisfied with the evidence summary and conclusion statement for a given topic, they assign a grade or recommendation to indicate the overall strength of the evidence. In the Canadian guidelines for enteral nutrition in the ICU patient, the intervention was "strongly recommended" if there were no reservations about endorsing it; "recommended" when evidence was supportive but there were uncertainties about the safety, feasibility, or costs; "should be considered" when supportive evidence was weak or there were major uncertainties about the safety, feasibility, or costs; or "no recommendation" when there was inadequate or conflicting evidence about the intervention.5
When a decision is required on whether a therapy will improve a patient's condition, EBM should be used whenever possible. This paper provides an overview of how evidence-based standards of care and technical and scientific reviews are conducted. Incorporating the evidence-based process into routine clinical practice requires clinicians to become efficient in searching the literature, appraising the evidence for its validity and applicability to their particular situation, and ultimately integrating the evidence into practice. These summaries by themselves will not make the decision for clinicians on how to ultimately treat their patients; however, they can help support this process. When fully integrated into clinical decisions, they enhance the opportunity for optimal clinical outcomes.
1 Sackett DL, Rosenberg WM, Gray JA, Haynes RB, Richardson WS.
Evidence based medicine: what it is and what it isn't.
BMJ. 1996;312:71
–72.
Nutrition in Clinical Practice, Vol. 22, No. 6,
599-601 (2007)
|
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
Top
Defining the Question
Study Inclusion Criteria