| Sign In to gain access to subscriptions and/or personal tools. |
Outcomes Research in Specialized Nutrition SupportDepartment of Surgery, Division of Surgical Oncology, UMDNJ/Robert Wood Johnson Medical School; and The Cancer Institute of New Jersey, New Brunswick, New Jersey Correspondence: David A. August, CINJ-195 Little Albany Street, New Brunswick, NJ 08903. Electronic mail may be sent to augustda{at}umdnj.edu. Outcomes research can play an important role in the development and evolution of the specialty practice of nutrition support. This manuscript discusses the current state and future prospect of outcomes research in specialized nutrition support. It is concluded that there is an important role in the field for research that focuses on "what works, and what does not" in nutrition support practice. A little over a decade ago, an article was published advocating the creation of a multi-institutional nutrition support outcomes research consortium (NSORC).1 The argument was offered that collection of rigorous outcomes data could be used to improve patient outcomes and the cost-effectiveness of specialized nutrition support (SNS), establish firm clinical evidence for optimal nutrition support practices, and assist professionals in the field to cope with increasing financial and regulatory pressures. Given the compelling nature of the discussion and its potential impact, the author asked, "If not now, when?" With the passage of over 10 years, it may be useful to ask:
Outcomes research has been defined as "the rigorous determination of what works in medical care and what does not."2 The purpose of outcomes research is to collect and analyze data to help patients, providers, payers, policy makers, and administrators make informed choices regarding medical treatment options and healthcare policy. Outcomes research uses analytic methods to produce information about the benefits, risks, and costs of healthcare services.3 It seeks to expand upon studies of clinical efficacy, focusing on effectiveness of treatment options in real-world clinical settings rather than the biologic effects of an intervention under ideal circumstances.4 Whereas traditional clinical trials methods are designed to assure valid and reproducible conclusions, outcomes research gathers data to understand the effect of interventions on patient care and patient outcomes.5 A research shortfall has been identified in the field of SNS, especially relating to clinical outcomes and the data needed to practice evidence-based SNS.6 Nutrition support outcomes research may help bridge this gap. There are many examples of the effect of this research shortfall. Malnutrition remains strikingly prevalent in hospitalized patients (25%–45%)7 and has not changed significantly from the seminal study of Bistrian et al8 in 1976. Despite the recognition more than 20 years ago of the devastating effect of malnutrition on cancer patient mortality, there are no proven interventions to mitigate this harmful circumstance.9 The contribution of nutrition support to the incidence of infectious complications in well-nourished and malnourished patients is poorly understood.10 Deadly errors persist in the formulation and administration of SNS.6 Prospective, randomized controlled trials (PRCTs) have historically ruled the medical research hierarchy (Table 1). An influential study published in 1982 comparing the results of PRCTs with clinical studies that used observational designs involving historical controls demonstrated that patient selection bias and changes in care over time strongly weight historical control studies in favor of new treatments.11 A more recent study, however, failed to demonstrate that similar biases are inherent in observational studies (case-control and cohort studies) with contemporaneous controls. In fact, the authors concluded that "The results of well-designed observational studies... do not systematically overestimate the magnitude of the effects of treatment as compared with those in randomized, controlled trials on the same topic."12
Although the methodologic strengths of PRCTs should not be undervalued, randomized trials are not ideally suited to efficiently improve the outcomes of patients receiving SNS. The clinical benefits of SNS are neither dramatic nor realized in the short term. Except in relatively unusual cases of short gut syndrome, malabsorption, gastrointestinal fistula, hyperemesis gravidarum, and specific nutrient deficiency, when SNS is used as a primary therapy to treat or prevent malnutrition resulting from primary gastrointestinal dysfunction, SNS is an adjuvant therapy. Its attributable effect on patient outcomes may be hidden by the effects of demographic factors, the presence of comorbidities, diagnosis, severity of illness, and disease-specific therapies.1,13 Some studies have shown reductions in morbidity and mortality resulting from the use of "adjuvant" SNS, especially in perioperative settings and in trauma patients.14–16 Many studies of the effectiveness of SNS (most notably the Veterans Administration Cooperative Trial10), however, have been confounded by the inclusion of well-nourished subjects not at nutrition risk, a subset of patients who are unlikely to benefit from SNS.17 Furthermore, most studies have focused on short-term clinical outcomes (primarily morbidity and mortality). More significant benefits may also be seen if longer-term outcomes (eg, 30-day readmission rate, functional status, sense of well-being) are assessed. Although incremental progress in SNS has been achieved through the application of PRCTs, this gold-standard methodology has not been as helpful as might have been hoped. Five major issues have restricted the utility of PRCTs in nutrition support. First, there is ethical reluctance to perform PRCTs out of concern of depriving patients of purportedly effective nutrition care. Because of this reluctance, some have been unwilling to include no-treatment arms in PRCTs of SNS. An observational study of nutritionally-at-risk patients can overcome this ethical dilemma by including patients who go unfed as part of standard, accepted (but possibly suboptimal) medical care who can serve as an untreated comparison group. Second, a treatment protocol used in a large, well-designed PRCT may be outdated by the time the results are reported. For example, the Veterans Affairs Total Parenteral Nutrition Cooperative Group Study, the largest PRCT reported to date of the use of SNS, was first conceived in the early 1980s, enrolled patients between 1984 and 1987, and reached publication in 1991.10 By 1991, the nutrition support protocol that was used was outdated because it included feeding of only marginally malnourished patients, it involved overfeeding of both carbohydrate and lipid calories, and required parenteral (as opposed to enteral) SNS.17 In contrast, an observational study is likely to include at least some subjects who are receiving the best available therapy throughout the duration of the study. Third, the technical complexity of SNS makes selection of an optimal intervention arm for a PRCT problematic. Serious disagreements exist concerning the best timing, composition (protein, carbohydrate, lipid, and micronutrient content), quantity, and route of administration of SNS. A well-designed observational study can help plan more effective PRCTs by identifying the most (and least) relevant treatment factors and the specific patients in whom interventions are most likely to be effective.18 Fourth, given the number of confounders that must be accounted for to assess the effectiveness of SNS in clinically relevant situations, large numbers of subjects are needed to answer the relevant questions in the heterogeneous population nutritionally at risk.19 Twenty thousand or more subjects may be required to answer even the most basic questions concerning the effectiveness and optimal use of SNS.1 Large randomized, multi-institutional trials are prohibitively expensive. Large observational studies using outcomes methods are often more cost-effective and can guide the implementation of subsequent PRCTs.20 Finally, the limited number of consenting, eligible subjects is a major impediment to the conduct of PRCTs, all the more so for the study of a technically complex adjuvant therapy (such as SNS) in patients who are not under the control of the interested clinician-scientist investigators. An observational study is more likely to successfully enroll a greater proportion of eligible subjects. Understanding the impact of quality of nutrition care upon patient outcomes is difficult because it is only one of a complex medley of therapies affecting patient outcomes. The confounders of severity of illness, comorbidity, and randomness must be considered and controlled.21,22 To control for these factors, large numbers of patients are required. The data elements required to accomplish risk adjustment and control for selection bias must be readily and reproducibly available if an outcomes project is to be practical and economically feasible (Figure 1).23–25
Well-conceived and executed outcomes studies have been successfully conducted in other medical specialties that face obstacles similar to those encountered in the field of nutrition support. A notable example is the work of Connors et al,26 in which the researchers studied the effectiveness of right heart (Swan-Ganz) catheterization in critically ill patients. This study successfully implemented an outcomes database using a prospective, multi-institutional, observational design. Multivariate techniques were successfully used to control for confounding factors and selection bias and allowed identification of important relationships between therapeutic interventions and patient outcomes. The Connors study looked at a question involving a complex, unstandardized intervention widely used in heterogeneous patient populations and was able to reach important conclusions. This study led to a clinical reappraisal of the indications for use the of Swan-Ganz catheters and paved the way for focused PRCTs of the use of right heart catheters. This was not felt feasible or ethical before this study.18 Subsequent PRCTs confirmed the counterintuitive hypothesis generated by this landmark clinical investigation.
By their very nature, PRCTs ask specific questions in narrowly defined patient populations receiving tightly controlled therapeutic interventions. They are well suited to incremental improvements in care but do not often result in major changes in outcomes. Carefully designed, prospective cohort studies, such as the proposed NSORC, can more efficiently study large numbers of patients in heterogeneous clinical settings. They are also, by their observational design, well suited to study the effects of actual practices in real-world clinical settings, in contrast to the very proscribed conditions often dictated by a PRCT. Observational studies can generate hypotheses that may subsequently be tested and refined in PRCTs.
It is beyond the scope of this article to comprehensively review the past 10 years' outcomes research in SNS. However, there has been significant progress in evidence-based approaches to the practice of SNS in the critical care setting largely because of PRCTs and meta-analyses. It is illustrative to briefly look at that experience, with particular attention to glycemic control, immune-enhancing formulas (IEF), route of administration, and guidelines development (Table 2).
In 2001, Van den Berghe et al27 published the results of a PRCT, involving 1548 mechanically ventilated subjects, that demonstrated reduction in morbidity and mortality in critically ill patients admitted to a surgical intensive care unit who received intensive insulin therapy with a goal of maintaining the blood glucose level under 100 mg/dL. Mortality was reduced by almost 50% in the intensive care unit and by more than one-third overall. Bacteremic sepsis was also reduced by almost 50%. Although reductions in morbidity were observed in a similar trial involving medical intensive care unit patients, the magnitude of the effect was less and there were no observed differences in mortality.28 A subsequent ad hoc analysis of the same group concluded that metabolic control, as reflected by normoglycemia, rather than the infused insulin dose per se, was the cause of improved outcomes.29 Although substantial reductions in morbidity, and perhaps mortality, are apparently achievable through tight glycemic control in critically ill patients, further studies are needed to determine whether these benefits are achievable outside of a focused investigational setting and whether glycemic control might also benefit the broader population of acutely but not critically ill hospitalized patients.
IEFs have been developed and tested in numerous PRCTs over the past decade.
These enteral formulas typically are supplemented with arginine, A recurring controversy in SNS has been the optimal route of administration of nutrients, enteral nutrition (EN) vs parenteral nutrition (PN). The Canadian critical care guidelines, while acknowledging no differences in mortality, strongly recommend the use of EN over PN when possible because of an apparent reduction in infectious complications and because of "safety, cost, and feasibility considerations."31 Other experts, however, point to the ease of administration of PN as an advantage of that route of administration.32 Simpson and Doig33 criticized the findings of previous meta-analyses addressing this issue because of methodologic choices that resulted in the misidentification of "high-quality trials." Having meta-analyzed 11 trials that they identified as high-quality methodologically, they found a mortality benefit in favor of PN, despite a reduced incidence of infectious complications in EN-treated patients. Again, these results were obtained only in critically ill patients and are not necessarily applicable to acutely ill, hospitalized patients not in an intensive care unit. The Canadian clinical practice guidelines for nutrition support in mechanically ventilated, critically ill adult patients referenced above codified an intensive effort to generate evidence-based nutrition management guidelines for critically ill patients.31 Rigorous methods were developed to identify relevant studies, analyze them systematically, and generate evidence-based clinical practice recommendations. Thirty-one specific questions were addressed, ranging from the timing of initiation of SNS to the use of glutamine supplementation. Because of the rigor of the development process and the broad sample of stakeholders used to review the final recommendations, the authors noted that, "we would expect that this guideline will be accepted by clinicians and that implementation of this guideline will improve outcomes for critically ill patients who need SNS."31 Much to their credit, these investigators subsequently performed a number of studies to determine whether promulgation of these guidelines would actually alter clinical practice and effect patient outcomes. In a prospective, observational study, they compared institutions whose ongoing practices were in greater compliance with the guideline recommendations with those that were less compliant.34 The more compliant units were more likely to feed patients enterally. In a follow-up study, 58 centers were randomized in 50 clusters to be in either an active dissemination arm (a lead dietitian at these sites was actively supported to achieve implementation of the guidelines) vs a passive arm (a copy of the guidelines was mailed to the lead dietitian at these sites).35 Clusters in the active dissemination arm achieved better glycemic control, but there were no other identifiable differences in practice patterns between the 2 groups. Overall, dissemination of the guidelines appeared to improve nutrition practices in both groups, but did not affect clinical outcomes. These latter implementation studies highlight an important issue regarding clinical guideline implementation. If one accepts the premise that "the ends justify the means," that the ultimate goal is to improve patient outcomes, studies must be performed that directly assess the effect of clinical research and guidelines development on patient well-being in broad populations. Principles and practices elucidated in clinical trials (be they PRCTs or otherwise) and rigorous guideline development will not achieve the desired endpoint of better patient outcomes unless they can be generalized and applied to larger numbers of patients. Only more broadly based outcomes studies that investigate ongoing practices and interventions in a general practice setting can determine whether specific interventions or more general changes in practice patterns actually help real-world patients.
We have given a broad overview of outcomes research activities in SNS over the past decade. Although relevant and important trials have been conducted, they have often not provided answers to important questions concerning optimal administration of SNS. Furthermore, there is little evidence that significant findings such as the importance of tight blood glucose control and the role of IEFs have been translated into clinical practice. In fact, the studies of Heyland and colleagues34,35 suggest that there is a dearth of information concerning successful dissemination of clinical trial discoveries and guidelines in a manner that actually changes practice and improves patient outcomes. To affect practice and outcomes, these issues must be considered much earlier in the research process and should influence how research questions are framed and what methods are used.3 These are the very problems that outcomes-oriented research is designed to address. The goal of outcomes research is, in fact, to change practice to achieve better outcomes. It may be useful to conclude by attempting to answer the 4 questions stated in the introductory paragraph of this article: What came of the idea to develop an NSORC? A dedicated group of investigators, with the support of the American Society for Parenteral and Enteral Nutrition, submitted a grant application to the National Institutes of Health for seed funding to assess the feasibility of an NSORC. The grant was not funded, in part because the reviewers felt that a feasibility study was not necessary. A larger grant was never submitted because the investigator group felt it could not generate a credible proposal without significant preliminary data assessing the feasibility of a multi-institutional, prospective, observational trial. Are the factors that motivated an SNS outcomes research initiative 10 years ago still relevant? Yes. As August stated in 1996, "... improvement in quality of nutrition care for our patients and for our professional survival in a rapidly changing healthcare environment demands that the field of nutrition support focus on asking what works and what does not."1 Thoughtful answers to these questions founded in rigorous data can improve patient outcomes, improve the cost-effectiveness of nutrition support, establish a firm scientific basis for the discipline of nutrition support, and protect nutrition support clinicians from the onslaught of economically motivated cutbacks in reimbursement and salary support that characterize the current healthcare climate.1 These issues seem as relevant today as they were a decade ago. What has happened in SNS-related outcomes research over the past decade? The significant advances achieved over the past decade have been due to the success of PRCTs, rigorous meta-analyses, and evidence-based guidelines development in incrementally improving our knowledge of best practices, primarily in critically ill patients. These successes have not been generalized to non-critically ill acute care or ambulatory patients. Furthermore, although these studies have improved our knowledge base, there is little evidence to suggest that this additional wisdom has been translated into improved patient care outcomes. The knowledge gained from these studies must be applied to patient care. What is the future of SNS-related outcomes research? We must build upon the advances of the past decade and focus on the endpoint that truly matters: patient well-being. Whether through development of an NSORC or some alternative approach, the field of SNS must once again assess the effect of our current practices on patient outcomes and validate methods to improve those outcomes. Although the details may change, it is still fair to ask of patient-driven research in SNS, If not now, when?
Nutrition in Clinical Practice, Vol. 22, No. 6,
602-608 (2007) This article has been cited by other articles:
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
|
|
 |
|
Sign In to gain access to subscriptions and/or personal tools.
Top
Why Outcomes Research in...
The Impact of Outcomes...
-3
fatty acids, glutamine, and nucleic acids, nutrients thought to improve immune
function, modulate inflammation, and reduce the incidence of nosocomial
infections. Heyland et
al
