Abstract
Research into the role of diet in health faces a number of methodologic challenges in the choice of study design, measurement methods, and analytic options. Heavier reliance on randomized controlled trial (RCT) designs is suggested as a way to solve these challenges. We present and discuss 7 inherent and practical considerations with special relevance to RCTs designed to study diet: 1) the need for narrow focus; 2) the choice of subjects and exposures; 3) blinding of the intervention; 4) perceived asymmetry of treatment in relation to need; 5) temporal relations between dietary exposures and putative outcomes; 6) strict adherence to the intervention protocol, despite potential clinical counter-indications; and 7) the need to maintain methodologic rigor, including measuring diet carefully and frequently. Alternatives, including observational studies and adaptive intervention designs, are presented and discussed. Given high noise-to-signal ratios interjected by using inaccurate assessment methods in studies with weak or inappropriate study designs (including RCTs), it is conceivable and indeed likely that effects of diet are underestimated. No matter which designs are used, studies will require continued improvement in the assessment of dietary intake. As technology continues to improve, there is potential for enhanced accuracy and reduced user burden of dietary assessments that are applicable to a wide variety of study designs, including RCTs.
Keywords: study design, epidemiologic studies, observational studies, randomized controlled trials, dietary assessment methods, informed consent, blinding, behavioral interventions
Introduction
Literature has accumulated over the past 3 decades to highlight confusing results from epidemiologic studies of diet and health (1–17) and errors in measuring dietary intake (18–28). Some of the investigators of these studies and others also have questioned the value of observational studies of diet and health, with some advising to limit diet-related research to randomized controlled trials (RCTs)10 (28–31). The RCT often is considered to be the strongest study design in biomedicine (32–34), one that might provide a broad-based solution for addressing methodologic problems encountered in nutrition research.
RCTs provide exact and prescriptive protocols to ensure scientific rigor in the most transparent of ways, by randomly allocating treatment. When factors that may bias the estimate of the effect of the intervention on the primary outcome are randomly distributed across intervention and comparison arms of an RCT, there is assurance that results derived are not subject to confounding bias. The allocation of the intervention by the investigators also reduces the probability of selection bias, by assigning people to specified study conditions rather than allowing them to choose.
Despite the apparent advantages of RCTs, only a small fraction of all human studies use randomized designs. However, much is known about relations between risk factors and disease. Recommendations on the role of human behaviors as determinants of health are made across a wide variety of risk factors and disease outcomes with little or no RCT-derived evidence. One of the best examples is contained in the US Surgeon General’s 1964 report on Smoking and Health. On the basis of Hill’s Criteria for Judging Causality (35, 36), the expert panel concluded that RCTs were not necessary to assert that tobacco “causes” an array of health outcomes, including lung cancer (37). This showed that strong, persuasive evidence can come from sources other than RCTs, which may be difficult or impossible to conduct for a variety of ethical or logistical reasons (28, 38–42).
Although the method of allocation (i.e., randomization compared with self-selection) and the nature of the trial (i.e., explanatory compared with pragmatic) are conceptually orthogonal, the reality is that rarely in normal clinical or community practice would a treatment be allocated at random. In those rare instances when this happens, trials tend to be cluster randomized (43), focused on supplementation (see The need to focus section) (44), or lack distinguishing features (thus facilitating blinding; see Blinding section) (45). In the work on pragmatic clinical trials by Peikes et al. (46), the implicit assumption is that such trials cannot use random allocation of treatments. Therefore, for most practical purposes, randomization is the exclusive province of explanatory trials, which tend to be favored by regulatory bodies and methodologic purists (47). Because it is much easier to randomize in the context of an explanatory trial, most RCTs tend to be explanatory. This is consistent with the edict that trials of health care interventions with well-understood mechanisms of action should lie toward the explanatory end of the trial continuum (32). This often is not the case for diet-related interventions.
In their seminal work nearly half a century ago, Schwartz and Lellouch (48) were concerned mainly with the distinction between internal validity and external validity (generalizability) and the tendency for many explanatory trials to produce results irrelevant to real-world needs. They also described pragmatism as an attitude rather than a characteristic of the trial. The reality is that trials lie on a continuum from purely explanatory to purely pragmatic (32, 49, 50). Despite the demand to use intention to treat as the first-line analyses as in any RCT (51), usually data from dietary trials are subject to post hoc analyses that do not require strict adherence (52–54). In some ways, results that take into account incomplete adherence may resemble those of pragmatic trials. Of course, caution must be exercised when interpreting results, because statistical power may be greatly diminished.
RCTs of behavioral interventions, in general, and ones that focus on diet, in particular, face a number of challenges because of the high level of participant commitment and involvement required. Previously, we delineated a number of problems commonly encountered in research into the role of diet in health and therein described a variety of solutions (55). We note that the study by Satija et al. (56) also touches on points related to RCTs. Our focus here is confined to describing the limitations, inherent and practical (57), in the use of RCTs to determine the role of diet in health and alternative designs for allocating dietary exposures or treatments. Advantages and disadvantages of various study designs are given in Table 1.
TABLE 1.
Design Type (references) | Advantages | Disadvantages |
Conventional RCT (32–34) | Random allocation of exposure; theoretically provides a “clean” comparison between intervention and control arms | Selective dropout/retention (58); limited ability to control for multiple exposures by design; incomplete adherence in the intervention arm; control group reactivity; inability to blind complex, behavioral exposures; limited generalizability for the specific exposure and study sample; questions about real-world effectiveness (where people choose therapies) (32, 59) |
Randomize-before-consent RCT (38, 60) | Random allocation of exposure; subject does not know alternative, thus reducing problems associated with motivation and expectation; may reduce selective dropout as it obviates problems associated with being given something seen as inferior | Inability to account for factors that might lead to selective dropout associated with knowing the alternative; study assignment cannot be blinded (although the subject is not initially aware of the alternative) |
SST (61) | Participants choose their preferred study arm; similar to real-world settings; popular among highly informed/engaged individuals who refuse randomization (e.g., HIV/AIDS as activists) | Inability to control for personal factors related to expectation and motivation that may be expressed as the “placebo” effect |
Hybrid RCT/SST (62) | Incorporates advantages of the RCT and SST; allows for control of individual factors related to motivation and expectation | Disadvantages of both RCT and SST in each respective arm (see above); expense (doubles required study size) |
Adaptive intervention (63–65) | Realistic, efficient, and practical; opportunistic (e.g., taking advantage of clinical or public health system changes); allows for changes in protocol to fit participant need | Poor control for extraneous factors that may not be captured well in clinical systems; likely limited opportunity for measuring dietary exposures of interest and important potential confounders; need to document and analyze for changes in protocol |
“N-of-1” design (66) | Sensitive to the needs of individual participants; comparison within individuals provides statistical power and control for unmeasured confounders | Economy of scale for measuring devices (including for diet); potential carryover effects; not blinded to participant; need to determine sequence of treatments and washout periods |
Observational study, case-control | Inexpensive and may be the only practical method for rare conditions (e.g., pancreatic cancer) | Selective recruitment that may be related to condition under study; retrospective assessment of exposures, including diet, may lead to information biases that are differentially recalled according to disease status |
Observational study, cohort | Allows for measuring exposures before disease onset (although “toggling” back in time to etiologically relevant period may be an issue) | Selective recruitment may exist, although if well designed and conducted it cannot be related to condition(s) under study; recall of exposures, including diet, may lead to information biases (although not directly to disease status) |
Cross-sectional studies | Potentially useful for hypothesis generation | Inability to control for temporal relations/causal sequence; because data typically are collected for other purposes, usually there is poor quality control for information on diet or important potential confounders |
Ecologic studies | Potentially useful for hypothesis generation | No direct use of dietary/nutritional information; instead, these are based on economic data (e.g., FAO Food Balance Sheets) |
1RCT, randomized clinical trial, SST, self-selection trial.
Special Considerations for RCT Designs in the Study of Diet and Health Outcomes
The need to focus
As a practical matter, RCT designs allow for only a limited number of factors (usually 1 or 2) to be allocated at a time. Although RCTs can be used for “whole-diet” approaches [e.g., the Women’s Health Initiative (WHI) Dietary Modification arm (67, 68) and the PREDIMED (PREvencion con DIeta MEDiterranea) trial (69, 70)], more typically they focus on a single food or one or a few nutrients. This may reflect the perception that obtaining adherence to a request or demand to change one’s entire diet is difficult and therefore neither feasible except under exceptional conditions nor readily translatable to public health practice. Despite the appeal to focus narrowly , doing so does not represent a realistic way to make meaningful change to prevent chronic disease, especially when the preponderance of evidence indicates that eating patterns associated with whole foods are much more strongly predictive of health outcomes than are individual foods or nutrients (70–73). Indeed, the National Cancer Institute Chemoprevention Program, which focused on key nutrients that could be isolated and tested in trials, had limited success (74). Whole diet or whole lifestyle approach represents an alternative perspective to single-agent strategies. However, they pose an additional set of challenges related to making extensive changes in diet that may be particularly relevant to influencing disease course.
Choice of subjects and exposures
Dietary intervention trials also may be of limited value because they inadvertently study the wrong population or the wrong type of exposure at the wrong point in the disease process. For reasons of cost, efficiency, and interpretability, trials generally are designed to study relatively homogeneous populations at relatively high risk of the outcomes of interest, testing narrowly defined exposures for a limited period of time (75). The answers they provide are more definitive for those conditions, but there may be severe limitations in how well the findings can be generalized. Sometimes they get it wrong.
For example, the ATBC (Alpha-Tocopherol, Beta-Carotene Cancer Prevention) study and the β-Carotene and Retinol Efficacy trial (76, 77) unexpectedly found evidence for a detrimental effect of β-carotene supplementation on subsequent risk of lung cancer in older smoking men. The results obtained were inconsistent with those of hundreds of observational studies that showed protective effects of whole-food diets rich in antioxidant and anti-inflammatory micronutrients on cancers of various sites (78–81). Although only partially understood, the reasons for these paradoxical results almost certainly include design decisions made for efficiency and cost. These reasons include studying only high-risk populations that may have differing responses to dietary exposures. Subsequent studies have shown that high-dose β-carotene in heavy smokers may induce alterations of retinoid metabolism and signaling pathways that favor cancer promotion, whereas more moderate doses (dietary amounts) in nonsmokers have beneficial effects (82).
Commonly, trials test only higher doses of isolated nutrients that may not have the same effect as more modest intake of nutrients in foods that naturally combine other bioactive constituents. This is true of the Supplémentation en Vitamines et Minéraux AntioXydants (SU.VI.MAX) study, in which we also conducted a post hoc analysis to test the effect of the dietary inflammatory index on metabolic syndrome (83). Most trials of chronic disease prevention study exposures relatively late in terms of disease latency because it is generally impractical to conduct trials for more than a few years. However, observational studies compare food and nutrient intakes at the time of measurement that serve as indicators of relative long-term exposures that extend back many years. Even in the ATBC study that found an adverse effect of supplemental moderate-dose β-carotene on cancer risk, retrospective measures of dietary intake of the nutrient at recruitment showed an inverse association with subsequent cancer risks. Thus, trials on the effects of supplemental nutrients later in life provide limited evidence on the benefits of sustained dietary practices during periods of etiologic relevance for most primary prevention. These problems were not foreseen at the time these trials were initiated, and the mistakes were costly. Protocols of large explanatory trials cannot be modified in response to new information without compromising study power and time required to complete.
Blinding
Unlike single-agent trials that, at least theoretically, can be double blinded, participation in behavioral trials requires obvious commitment that precludes blinding from the perspective of the participant. The inability to blind the participant to the active ingredient(s) of the intervention forestalls one of the major advantages of the RCT, which is that there is no discernable difference in exposure by treatment allocation. This is an important point of distinction in that studies on the pragmatic trial end of the continuum would allow for unblinding in a manner uncharacteristic of explanatory trials.
Perceived asymmetry of treatment in relation to need
Especially in studies of individuals with conditions perceived to be life-threatening, there will be major concerns about asymmetry for the control condition, notwithstanding common attempts to devise an attentionally equivalent control for comparison (84–86). Individuals that might be important to study, such as persons with anemia or pregnant women, may be excluded for ethical reasons. Even for studies in which there may be no major concern about perception of such vulnerability, the intention to engage in behaviors that are alleged to affect long-term chronic disease risk requires substantial commitment and associated motivation and expectation. In essence, potential participants will only seek RCT dietary trials because they are looking for solutions to what they perceive to be a behavior that needs improvement. Thus, in any RCT of dietary factors there will be controls who seek dietary constituents that mimic the intervention (because they are now suitably informed and have already expressed an interest in participating). Likewise, there will be participants randomly assigned to the intervention who adhere only incompletely or not at all because they are unwilling, even after providing informed consent, to commit fully to participating in the intervention.
Even if procedures are put in place to not exclude individuals at the outset, thus avoiding selection before recruitment or selective dropout, there could be other problems with adherence. These factors would tend to “wash out” effects that otherwise might be observed in a self-selection trial (or even through careful observation). Ideally, this would happen in a nondifferential manner, but it is likely that this occurs in a way that leads to bias toward no effect of the intervention. For example, we have found that a greater percentage (59.5%) of control than intervention (49.1%) participants in a community-based RCT (33) lost weight. This phenomenon among controls is in stark contrast to findings from population-based surveys or observational studies in which adults typically gain weight at a rate of ∼1 pound (∼0.5 kg)/y (87–91). What may be driving this finding is that 39% of controls did not return for follow-up measures (12 wk after baseline), whereas only 21% of intervention participants did not return.
Another example of this phenomenon is the early discontinuation of the Look AHEAD (Action for Health in Diabetes) trial, the largest and longest trial to examine an intensive lifestyle intervention for weight loss and cardiovascular disease prevention compared with usual care (92). The trial was discontinued early (median of 9.6 y) because of a lack of difference in cardiovascular endpoints between the 2 groups (93). The researchers cited possible reasons for the lack of differences as the impact of the minimal education sessions (≤2/y) offered and increased use of statins in the control group (93). The most obvious problem with Look AHEAD is that the event rate was much lower than “expected” (because it was overestimated). In essence, the study pitted diet change against a large pharmacologic effect, which dramatically lowered the expected event rate: 50% were on statins, 75% were on antihypertensive agents, and the mean circulating LDL concentration at baseline was 112 mg/dL. This is seen in every modern study of cardiovascular disease (94).
The effect of this apparent asymmetry between control and treatment arms also could explain why we observed a much stronger effect in men with rising prostate-specific antigens after prostatectomy who self-selected a diet-physical activity-stress reduction intervention (95) than in similar men randomly assigned to a comparable intervention (96). It also explains diminution of effects in long-term trials, ranging from the Multiple Risk Factor Intervention Trial to the WHI. These common field experiences result in a loss of the original scientific rigor that the RCT was supposed to impart to the study. This may lead to a sense of “failure” on the part of the participant and the study team when, in reality, the dietary or other lifestyle intervention does indeed have a positive health benefit that simply cannot be observed with the use of an RCT design (97). This point is especially relevant among under-represented and vulnerable populations who have been witness to gross mistreatment by the research community. These community partners often come reluctantly to research and want to see that their hard work and efforts (both individually as participants and collectively as recruiters and advocates for the study) has realized a positive benefit for their community.
Temporal relations between dietary exposures and putative outcomes
Most chronic disease outcomes present logistical problems in terms of temporal control because they usually occur only after suitably long latency or incubation periods (which tend to be longest for cancer) (97). The tighter the control (e.g., with metabolic ward studies being the most extreme), the greater the logistical complexity required, including the need to follow participants over long time periods. Practical problems that plague these studies include fatigue related to long-duration involvement (often interacting with the condition under study) (64, 98–102) and attempts on the part of participants to compensate for one behavior change by making another change that may countervail or amplify the effect of the first (103, 104). In addition, efforts to exert more control by design interject a set of selection factors, related both to subjects’ participation and exposures, that greatly limit real-world relevance (including those that influence reporting accuracy) that negatively affect translatability. For chronic diseases with long latencies, such as atherosclerosis or cancer, it is reasonable to question how making changes late in the natural history of the disease process will translate into meaningful reductions in risk of these chronic diseases (69, 105–109). This also applies to analyses of data from observational studies in which long-term follow-up data on relevant exposures may not exist; however, dietary exposures tend to track over time, with intakes of individuals being highly correlated over long periods from childhood through adulthood (110–112).
Immutability of the treatment
Typically, individuals who participate in RCTs will receive treatment(s) that remains unchanged throughout the study, regardless of response. Although needed to ensure rigor for statistical power, this is the opposite of what is recommended for evidence-based care, whereby individuals are regularly reassessed to determine whether the treatment is effective, the dose should be changed, or another treatment should be substituted (113). In addition, subjects may be excluded from trials because of comorbidities or other conditions that can confound the evaluation of treatment effects and, as a result, compromise the external validity and limit the usefulness of the research findings for clinical practice (114).
The need to maintain methodologic rigor
No matter a study’s design, dietary exposures may modify or confound the effect of the exposures targeted by the intervention (whether or not they focus on diet). Therefore, it is important to identify potential confounders at the design stage and provide means for measuring and controlling them analytically. This could include medications and other factors that might affect nutrient uptake and utilization. It also could include nutrients such as α-tocopherol and β-carotene that were used in the ATBC study (115) and selenium and vitamin E in the Selenium and Vitamin E Cancer Prevention Trial (116), which can come from dietary sources and from supplements. Individuals can (and often do) change their behaviors to modify risk. Participants in the control group of a dietary trial may be motivated to change their diet or supplement use to decrease risk. Participants in the intervention group may compensate intake in more subtle ways to account for changes in taste, satiety, or other attributes of diet. In the ATBC study the total daily dose of 25 mg β-carotene was equivalent to only 3 large carrots, underscoring the need to measure diet carefully to conduct meaningful post hoc analyses. Another example of how prescribing a supplement or pill-based intervention can have an unintended impact on diet is revealed by examining the trends in dietary intake among statin users. Between 1999 and 2010, individuals who began using statins had significant increases in energy and dietary fat consumption compared with nonstatin users (117).
As with other studies of diet and health, RCTs must face the need to measure diet. Besides the normal issues concerning measurement bias related to subject-specific factors (21, 118–121), RCTs are uniquely susceptible to errors in self-report related to implementing a focused intervention and monitoring adherence (68). The increased susceptibility is due to participants being sensitized to the dietary hypothesis being tested. For example, in the WHI, we found that individuals who were eligible for the diet modification arm overestimated their self-reported dietary intake by ∼169 kcal/d in comparison with estimated metabolic requirement relative to women who were ineligible (67). Possible measurement bias could help explain why the dietary modification arm of the WHI provided only ambiguous, uncertain results for the benefits of diet, despite the enormous expense and time the trial required. In addition, the primary question tested (total dietary fat reduction) was considered outdated (supplanted by alterations in type of fat and growing concern about the effects of simple carbohydrates) by the time the results went to press (17, 122). This problem is certainly not unique to the WHI and will likely apply to other large-scale, long-term trials of diet on chronic disease risk. So, pointing to the inability to account and control for measurement error as an argument in favor of conducting RCTs so as to avoid measuring diet is misguided in light of the available evidence.
Summary of preceding points
RCTs designed to study diet among free-living people face a host of problems in attempting to create large contrasts in dietary exposures (123, 124). Changing behaviors is challenging, and these trials may require intense commitment to make and sustain large changes. Furthermore, some individuals who are willing to accept randomization likely would either lack the motivation to persevere if randomly assigned to an intensive intervention or to seek other means for achieving change if randomly assigned to a “no-treatment” control.
Placing additional emphasis on RCTs to answer questions that relate diet to health outcomes, as has been suggested (29), would delay the scientific process, leaving us with little additional evidence on diet and health for many years and serious questions about the future relevance of questions asked now on the subject of diet and health. Despite the success of PREDIMED in showing that the Mediterranean diet can prevent cardiovascular disease (69, 70), there are many other examples of expensive and lengthy trials and large-scale observational studies that have failed to provide definitive answers to the questions they set out to answer. For many dietary issues, trials are neither feasible nor ethical, and they may be limited in the generalizability of their findings even if they can be implemented (125–127).
Alternatives to the RCT
Rarely would a behavioral intervention of any kind be able to strictly enforce and monitor adherence in a rigorous way. In addition, the matter of selective recruitment and dropout would tend to undermine the explanatory imperative. Largely, this is because in such trials the design is not matched to the decision-making needs of those people using the protocol under study (32, 46, 49). Trials at the explanatory end of the continuum that attempt to guarantee internal validity are prone to be undermined by external influences, including a lack of participant adherence, which would obscure a true effect of diet on study outcome.
By contrast, a pragmatic attitude would be much more highly tuned to the needs of patients. The reality of clinical and community practice, however, would not be easily amenable to randomization under such circumstances (46). Although some may argue that randomization would factor out the placebo effect, the reality is that motivation and expectation are important factors in any behavioral intervention, including one focused on diet. If we force randomization as part of an explanatory attitude (128), it is likely that we would have neither internal nor external validity.
So, what are we to do? Pragmatic trials may make allowance for individual tailoring of the intervention. In instances in which such alterations may take place, however, there is an additional requirement for measurement and monitoring (32, 47).
Given that results from observational studies produce results that are consistent with those from trials if the exposure level and time of exposure are the same (129), in many instances it is reasonable to continue to use and improve on observational study designs. There also is a growing interest among intervention researchers to use adaptive, nonstatic research designs (130). Adaptive interventions were described as “operationalized and individually tailored strategies for prevention and treatment of chronic, relapsing disorders” (64). Examples of adaptive intervention designs include both the Multiphase Optimization Strategy and the sequential multiple assignment randomized trial designs (131). These designs could be applied to different nutrition- or diet-based intervention components, such as randomly assigning individuals to consume certain foods or diets, and be used to examine behavioral strategies, such as participant motivation or adherence. These strategies are beginning to be used frequently in studies that involve mobile health technology because of the need for research to keep pace with technology (132). Another design being used by mobile health studies is an “n-of-1” design (133). This design addresses the need for patient-centered outcomes research and the need to rapidly iterate interventions (66). As opposed to standard RCTs, n-of-1 trials use crossover between treatments to address the problem of patient-by-treatment interaction (66). Multiple n-of-1 studies could be conducted and jointly analyzed, a strategy that might help identify carryover effects and provide reasons for why certain individuals respond to treatments (134).
Another alternative to the RCT that is applicable to nutrition research is the use of a randomized encouragement design (135). Although nutrition studies can use methods to carefully control what participants consume, such as housing participants in metabolic wards (136) or conducting feeding studies that provide prepared, pre-proportioned meals to participants (137), these methods are not real-world tests of diet. Randomized encouragement design studies focus less on adherence and more on randomly assigning participants to differing advice or recommendations (135).
Along with alternatives to RCT designs, we also need alternatives to the way we assess nutrition data during research studies. The RCT mindset has led to a rigid assessment of nutrition, collecting data typically before and after intervention with most of what happens in between being unknown. But human behavior, including nutrition-related behavior changes induced by an intervention, can be fluid and dynamic. Finding ways to capture real-time, continuous data are important because this will allow researchers to deliver more adaptive, just-in-time interventions (138).
Future Directions
It is important to carefully consider the overall goals of diet-related research and the design of dietary intervention studies, recognizing the inherent challenges and limitations of conducting meaningful RCTs on diet and health. Observational studies and pragmatic trials are not intrinsically flawed. Both reflect exposures as they are allocated in the real world. Efforts should be aimed at better understanding the challenges involved in enhancing their performance and improving methods of dietary assessment and study design. Humans are complex, and their behaviors are subject to multiple influences at many different levels. We advocate for greater creativity among investigators and increased transdisciplinary dialogue. In this way, advances in nutritional sciences can be realized.
Given high noise-to-signal ratios interjected by using inaccurate assessment methods in studies with weak or inappropriate study designs (including RCTs), it is likely that the effects of diet are underestimated. So, no matter which design is chosen, we must continue to work diligently to address acknowledged problems with the measurement of diet (55, 139). For the foreseeable future there will be no avoiding the use of these methods in studies of the effects of diet on health, no matter their design. Therefore, effort should be devoted to understanding and controlling these errors. This should include continued investigation of reporting biases with the use of a variety of criteria, but probably mainly construct validation measures, extending past work in the area (21, 118–121, 140, 141).
Research that focuses on technologic improvement also seems well advised. Digital food photography via smartphone camera has the potential to allow for just-in-time food recording (142) that could assist with assessing adherence. Research currently is under way to examine the use of photography to estimate the nutrient content of foods and beverages consumed (142, 143). Studies that use digital food photography with smartphones or wearable cameras have either solely relied on food photos by the user as a digital food record or included user photos as a way to enhance 24-h recalls or food records (144, 145). To analyze the nutrient content of the foods and beverages present in the photos, studies have either used trained raters to view photos and to enter the foods into a nutrient database or have relied on image processing by computers to determine what foods and beverages are present and the portion sizes (143, 145). There are other technologies currently being explored to capture dietary data, including interactive websites (146), wearable devices (147), digital audio recorders (146), scanning or sensor-based technologies (146), and expanded use of social media (148). As technology continues to improve, there is potential for enhanced accuracy and reduced user burden of dietary assessment. Many of these newer methods/approaches will have their limitations, however, and biases known to be associated with structured assessment methods may be evident. For example, people could change their eating behaviors toward socially desirable norms when being followed so closely with the use of more invasive data collections methods (e.g., pictures and cameras). Additional methodologic research will be needed as the field incorporates these technologic innovations into study designs.
In conclusion, the RCT is a powerful tool for health research, but it may be particularly limiting for diet-related studies. We have described many alternatives to this design that need further exploration and consideration. No matter which design is used, studies will require continued improvement in the assessment of dietary intake. Future knowledge on the health effects of diet is likely to come from a varied and dynamic range of methods, including observational and experimental strategies.
Acknowledgments
All authors read and approved the final manuscript.
Footnotes
Abbreviations used: AHEAD, Action for Health in Diabetes; ATBC, Alpha-Tocopherol, Beta-Carotene Cancer Prevention; PREDIMED, PREvencion con DIeta MEDiterranea; RCT, randomized controlled trial; WHI, Women’s Health Initiative.
References
- 1.Feinstein AR. Scientific standards and epidemiologic methods. Am J Clin Nutr 1987; 45(5 Suppl)1080–8. [DOI] [PubMed] [Google Scholar]
- 2.Feinstein AR. Scientific standards in epidemiologic studies of the menace of daily life. Science 1988;242:1257–63. [DOI] [PubMed] [Google Scholar]
- 3.Savitz DA, Greenland S, Stolley PD, Kelsey JL. Scientific standards of criticism: A reaction to “scientific standards in epidemiologic studies of the menace of daily life”, by A.R. Feinstein. Epidemiology 1990;1:78–83. [DOI] [PubMed] [Google Scholar]
- 4.Taubes G. Epidemiology faces its limits. Science 1995;269:164–9. [DOI] [PubMed] [Google Scholar]
- 5.Taubes G. The (political) science of salt. Science 1998;281:898–901. [DOI] [PubMed] [Google Scholar]
- 6.Taubes G. The soft science of dietary fat. Science 2001;291:2536–45. [DOI] [PubMed] [Google Scholar]
- 7.Taubes G. Nutrition. The epidemic that wasn’t? Science 2001;291:2540. [DOI] [PubMed] [Google Scholar]
- 8.Taubes G. Nutrition. What if Americans ate less saturated fat? Science 2001;291:2538. [DOI] [PubMed] [Google Scholar]
- 9.Trichopoulos D. Epidemiology faces its limits [letter]. Science 1995;269:1326.7660106 [Google Scholar]
- 10.Willett W, Greenland S, MacMahom B, Trichopoulos D, Rothman K, Thomas D, Thun M, Weiss N. The discipline of epidemiology [letter]. Science 1995;269:1325–6. [DOI] [PubMed] [Google Scholar]
- 11.Gori GB. The discipline of epidemiology [letter]. Science 1995;269:1327– 8. [DOI] [PubMed] [Google Scholar]
- 12.Miller RW. The discipline of epidemiology [letter]. Science 1995;269:1327. [DOI] [PubMed] [Google Scholar]
- 13.Rapp J. The discipline of epidemiology [letter]. Science 1995;269:1327. [DOI] [PubMed] [Google Scholar]
- 14.Saah AJ. The discipline of epidemiology [letter]. Science 1995;269:1327. [DOI] [PubMed] [Google Scholar]
- 15.Fraser GE. A search for truth in dietary epidemiology. Am J Clin Nutr 2003; 78(3 Suppl)521S–5S. [DOI] [PubMed] [Google Scholar]
- 16.Stevens J, Taber DR, Murray DM, Ward DS. Advances and controversies in the design of obesity prevention trials. Obesity (Silver Spring) 2007;15:2163–70. [DOI] [PubMed] [Google Scholar]
- 17.Yngve A, Hambraeus L, Lissner L, Serra Majem L, Vaz de Almeida MD, Berg C, Hughes R, Cannon G, Thorsdottir I, Kearney J, et al. . The Women’s Health Initiative. What is on trial: nutrition and chronic disease? Or misinterpreted science, media havoc and the sound of silence from peers? Public Health Nutr 2006;9:269–72. [DOI] [PubMed] [Google Scholar]
- 18.Freudenheim JL, Marshall JR. The problem of profound mismeasurement and the power of epidemiological studies of diet and cancer. Nutr Cancer 1988;11:243–50. [DOI] [PubMed] [Google Scholar]
- 19.Errors in reporting habitual energy intake. Nutr Rev 1991;49:215–7. [DOI] [PubMed] [Google Scholar]
- 20.Beaton GH, Burema J, Rittenbaugh C. Errors in the interpretation of dietary assessments. Am J Clin Nutr 1997;65:1100S–7S. [DOI] [PubMed] [Google Scholar]
- 21.Hebert JR, Ebbeling CB, Matthews CE, Ma Y, Clemow L, Hurley TG, Druker S. Systematic errors in middle-aged women’s estimates of energy intake: Comparing three self-report measures to total energy expenditure from doubly labeled water. Ann Epidemiol 2002;12:577–86. [DOI] [PubMed] [Google Scholar]
- 22.Prentice RL. Measurement error and results from analytic epidemiology: dietary fat and breast cancer. J Natl Cancer Inst 1996;88:1738–47. [DOI] [PubMed] [Google Scholar]
- 23.Wu ML, Whittemore AS, Jung DL. Errors in reported dietary intakes: I. short-term recall. Am J Epidemiol 1986;124:826–35. [DOI] [PubMed] [Google Scholar]
- 24.Archer E, Hand GA, Blair SN. Validity of U.S. Nutritional surveillance: National Health and Nutrition Examination Survey caloric energy intake data, 1971–2010. PLoS One 2013;8:e76632. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Boeing H. Nutritional epidemiology: new perspectives for understanding the diet-disease relationship? Eur J Clin Nutr 2013;67:424–9. [DOI] [PubMed] [Google Scholar]
- 26.Hite AH. Food frequency questionnaires: small associations and large errors. Nutrition 2013;29:925–6. [DOI] [PubMed] [Google Scholar]
- 27.Mitka M. Do flawed data on caloric intake from NHANES present problems for researchers and policy makers? JAMA 2013;310:2137–8. [DOI] [PubMed] [Google Scholar]
- 28.Casazza K, Fontaine KR, Astrup A, Birch LL, Brown AW, Bohan Brown MM, Durant N, Dutton G, Foster EM, Heymsfield SB, et al. . Myths, presumptions, and facts about obesity. N Engl J Med 2013;368:446–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Ioannidis JP. Implausible results in human nutrition research. BMJ 2013;347:f6698. [DOI] [PubMed] [Google Scholar]
- 30.Mattes RD, Shikany JM, Kaiser KA, Allison DB. Nutritively sweetened beverage consumption and body weight: a systematic review and meta-analysis of randomized experiments. Obes Rev 2011;12:346–65. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Cope MB, Allison DB. White hat bias: a threat to the integrity of scientific reporting. Acta Paediatr 2010;99:1615–7. [DOI] [PubMed] [Google Scholar]
- 32.Treweek S, Zwarenstein M. Making trials matter: pragmatic and explanatory trials and the problem of applicability. Trials 2009;10:37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Adams SA, Heiney SP, Brandt HM, Wirth MD, Khan S, Johnson H, Davis L, Wineglass CM, Warren-Jones TY, Felder TM, et al. . A comparison of a centralized versus de-centralized recruitment schema in two community-based participatory research studies for cancer prevention. J Community Health 2015;40:251–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Nagraj SK, Naresh S, Srinivas K, Renjith George P, Shrestha A, Levenson D, Ferraiolo DM. Interventions for the management of taste disturbances. Cochrane Database Syst Rev 2014;11:CD010470. [DOI] [PubMed] [Google Scholar]
- 35.Hill AB. Observation and Experiment. N Engl J Med 1953;248:995–1001. [DOI] [PubMed] [Google Scholar]
- 36.Hill AB. The environment and disease: association or causation? Proc R Soc Med 1965;58:295–300. [PMC free article] [PubMed] [Google Scholar]
- 37.US Department of Health Education and Welfare. Smoking and health: report of the Advisory Committee to the Surgeon General of the Public Health Services. Washington (DC): US Department of Health and Human Services; 1964. [US DHHS publication (PHS) 1103.] [Google Scholar]
- 38.Armstrong PW, Watts DG. Clinical trials: randomization before consent. Biomedicine. 1981;34:65–6. [PubMed] [Google Scholar]
- 39.Ellis PM. Attitudes towards and participation in randomised clinical trials in oncology: a review of the literature. Ann Oncol 2000;11:939–45. [DOI] [PubMed] [Google Scholar]
- 40.Fisher LD. Advances in clinical trials in the twentieth century. Annu Rev Public Health 1999;20:109–24. [DOI] [PubMed] [Google Scholar]
- 41.Moyer A, Knapp-Oliver SK, Sohl SJ, Schnieder S, Floyd AH. Lessons to be learned from 25 years of research investigating psychosocial interventions for cancer patients. Cancer J 2009;15:345–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Smith SA, Blumenthal DS. Community health workers support community-based participatory research ethics: lessons learned along the research-to-practice-to-community continuum. J Health Care Poor Underserved 2012; 23(4 Suppl)77–87. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43.Robroek SJ, Polinder S, Bredt FJ, Burdorf A. Cost-effectiveness of a long-term Internet-delivered worksite health promotion programme on physical activity and nutrition: a cluster randomized controlled trial. Health Educ Res 2012;27:399–410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Avenell A, Campbell MK, Cook JA, Hannaford PC, Kilonzo MM, McNeill G, Milne AC, Ramsay CR, Seymour DG, Stephen AI, et al. . Effect of multivitamin and multimineral supplements on morbidity from infections in older people (MAVIS trial): pragmatic, randomised, double blind, placebo controlled trial. BMJ 2005;331:324–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Riecke BF, Christensen R, Christensen P, Leeds AR, Boesen M, Lohmander LS, Astrup A, Bliddal H. Comparing two low-energy diets for the treatment of knee osteoarthritis symptoms in obese patients: a pragmatic randomized clinical trial. Osteoarthritis Cartilage 2010;18:746–54. [DOI] [PubMed] [Google Scholar]
- 46. Peikes D, Geonnotti K, Wang W. Using pragmatic clinical trials to test the effectiveness of patient-centered medical home models in real-world settings. Rockville (MD): Agency for Healthcare Research and Quality; 2013. Contract 13–0030-EF.
- 47.Zwarenstein M, Treweek S.. ACP Journal Club. What kind of randomized trials do patients and clinicians need ? Ann Intern Med 2009;150:JC5–2, JC5–3. [DOI] [PubMed] [Google Scholar]
- 48.Schwartz D, Lellouch J. Explanatory and pragmatic attitudes in therapeutical trials. J Chronic Dis 1967;20:637–48. [DOI] [PubMed] [Google Scholar]
- 49.Patsopoulos NA. A pragmatic view on pragmatic trials. Dialogues Clin Neurosci 2011;13:217–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 50.Thorpe KE, Zwarenstein M, Oxman AD, Treweek S, Furberg CD, Altman DG, Tunis S, Bergel E, Harvey I, Magid DJ, et al. . A pragmatic-explanatory continuum indicator summary (PRECIS): a tool to help trial designers. J Clin Epidemiol 2009;62:464–75. [DOI] [PubMed] [Google Scholar]
- 51.Hollis S, Campbell F. What is meant by intention to treat analysis? Survey of published randomised controlled trials. BMJ 1999;319:670–4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 52.Antwi SO, Steck SE, Zhang H, Stumm L, Zhang J, Hurley TG, Hebert JR. Plasma carotenoids and tocopherols in relation to prostate-specific antigen (PSA) levels among men with biochemical recurrence of prostate cancer. Cancer Epidemiol 2015;39:752–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Ockene IS, Hebert JR, Ockene JK, Saperia GM, Stanek E, Nicolosi R, Merriam PA, Hurley TG. Effect of physician-delivered nutrition counseling training and an office support system on saturated fat intake, weight, and serum lipid measurements in a hyperlipidemic population: the Worcester-Area Trial for Counseling in Hyperlipidemia (WATCH). Arch Intern Med 1999;159:725–31. [DOI] [PubMed] [Google Scholar]
- 54.Tabung FK, Steck SE, Ma Y, Liese AD, Zhang J, Caan B, Hou L, Johnson KC, Mossavar-Rahmani Y, Shivappa N, et al. . The association between dietary inflammatory index and risk of colorectal cancer among postmenopausal women: results from the Women’s Health Initiative. Cancer Causes Control 2015;26:399–408. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Hébert JR, Hurley TG, Steck SE, Miller DR, Tabung FK, Peterson KE, Kushi LH, Frongillo EA. Considering the value of dietary assessment data in informing nutrition-related health policy. Adv Nutr 2014;5:447–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Satija A, Yu E, Willett WC, Hu FB. Understanding nutritional epidemiology and its role in policy. Adv Nutr 2015;6:5–18. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 57.Sanson-Fisher RW, Bonevski B, Green LW, D’Este C. Limitations of the randomized controlled trial in evaluating population-based health interventions. Am J Prev Med 2007;33:155–61. [DOI] [PubMed] [Google Scholar]
- 58.Goode AD, Winkler EAH, Reeves MM, Eakin EG. Relationship between intervention dose and outcomes in living well with diabetes-a randomized trial of a telephone-delivered lifestyle-based weight loss intervention. Am J Health Promot 2015;30:120–9. [DOI] [PubMed] [Google Scholar]
- 59.Attanasio OP. Evidence on public policy: Methodological issues, political issues and examples. Scand J Public Health 2014; 42(13 Suppl)28–40. [DOI] [PubMed] [Google Scholar]
- 60.Lubowitz JH. Randomize, then consent: a strategy for improving patient acceptance of participation in randomized controlled trials. Arthroscopy 2006;22:1007–8. [DOI] [PubMed] [Google Scholar]
- 61.Doyle JS, Degenhardt L, Pedrana AE, McBryde ES, Guy RJ, Stoove MA, Weaver ER, Grulich AE, Lo YR, Hellard ME. Effects of HIV antiretroviral therapy on sexual and injecting risk-taking behavior: a systematic review and meta-analysis. Clin Infect Dis 2014;59:1483–94. [DOI] [PubMed] [Google Scholar]
- 62.Brewin CR, Bradley C. Patient preferences and randomized clinical trials. BMJ 1989;299:313–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Stallard N, Hamborg T, Parsons N, Friede T. Adaptive designs for confirmatory clinical trials with subgroup selection. J Biopharm Stat 2014;24:168–87. [DOI] [PubMed] [Google Scholar]
- 64.Deshpande S, Rivera DE, Younger JW, Nandola NN. A control systems engineering approach for adaptive behavioral interventions: illustration with a fibromyalgia intervention. Transl Behav Med 2014;4:275–89. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Lai TL, Lavori PW, Shih MC. Adaptive trial designs. Annu Rev Pharmacol Toxicol 2012;52:101–10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 66.Kravitz RL, Duan N, Duan N, Eslick I, Gabler NB, Kaplan HC, Kravitz RL, Larson EB, Pace WD, Schmid CH, et al. . Design and implementation of N-of-1 trials: a user’s guide. Rockville (MD): Agency for Healthcare Research and Quality; 2014. [AHRQ publication13(14)-EHC122-EF.] www.effectivehealthcare.ahrq.gov/N-1-Trials.cfm.
- 67.Hebert JR, Patterson RE, Gorfine M, Ebbeling CB, St. Jeor ST, Chlebowski RT. Differences between estimated caloric requirements and self-reported caloric intake in the Women’s Health Initiative. Ann Epidemiol 2003;13:629–37. [DOI] [PubMed] [Google Scholar]
- 68.Prentice RL, Anderson GL. The Women’s Health Initiative: Lessons learned. Annu Rev Public Health 2008;29:131–50. [DOI] [PubMed] [Google Scholar]
- 69.Estruch R, Ros E, Salas-Salvado J, Covas MI, Corella D, Aros F, Gomez-Gracia E, Ruiz-Gutierrez V, Fiol M, Lapetra J, et al. . Primary prevention of cardiovascular disease with a Mediterranean diet. N Engl J Med 2013;368:1279–90. [DOI] [PubMed] [Google Scholar]
- 70.Garcia-Arellano A, Ramallal R, Ruiz-Canela M, Salas-Salvado J, Corella D, Shivappa N, Schroder H, Hebert JR, Ros E, Gomez-Garcia E, et al. . Dietary Inflammatory Index and incidence of cardiovascular disease in the PREDIMED study. Nutrients 2015;7:4124–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 71.Singh PN, Arthur KN, Orlich MJ, James W, Purty A, Job JS, Rajaram S, Sabate J. Global epidemiology of obesity, vegetarian dietary patterns, and noncommunicable disease in Asian Indians. Am J Clin Nutr 2014;100 Suppl 1:359S–64S. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 72.Pimenta AM, Toledo E, Rodriguez-Diez MC, Gea A, Lopez-Iracheta R, Shivappa N, Hebert JR, Martinez-Gonzalez MA. Dietary indexes, food patterns and incidence of metabolic syndrome in a Mediterranean cohort: The SUN project. Clin Nutr 2015;34:508–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 73.Barbaresko J, Koch M, Schulze MB, Nothlings U. Dietary pattern analysis and biomarkers of low-grade inflammation: a systematic literature review. Nutr Rev 2013;71:511–27. [DOI] [PubMed] [Google Scholar]
- 74.Potter JD. The failure of cancer chemoprevention. Carcinogenesis 2014;35:974–82. [DOI] [PubMed] [Google Scholar]
- 75.Rothman KJ, Lash TL, Greenland S. Modern epidemiology. 3rd ed. Philadelphia: Lippincott Williams & Wilkens; 2013. [Google Scholar]
- 76.Omenn GS, Goodman G, Thornquist M, Grizzle J, Rosenstock L, Barnhart S, Balmes J, Cherniack MG, Cullen MR, Glass A, et al. . The beta-carotene and retinol efficacy trial (CARET) for chemoprevention of lung cancer in high risk populations: smokers and asbestos-exposed workers. Cancer Res 1994;54:2038s–43s. [PubMed] [Google Scholar]
- 77.The effect of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. The Alpha-Tocopherol, Beta Carotene Cancer Preventio Study Group. N Engl J Med 1994;330:1029–35. [DOI] [PubMed] [Google Scholar]
- 78.Aune D, Chan DS, Vieira AR, Navarro Rosenblatt DA, Vieira R, Greenwood DC, Norat T. Dietary compared with blood concentrations of carotenoids and breast cancer risk: a systematic review and meta-analysis of prospective studies. Am J Clin Nutr 2012;96:356–73. [DOI] [PubMed] [Google Scholar]
- 79.Birt DF. Update on the effects of vitamins A, C, and E and selenium on carcinogenesis. Proc Soc Exp Biol Med 1986;183:311–20. [DOI] [PubMed] [Google Scholar]
- 80.Rock CL, Flatt SW, Natarajan L, Thomson CA, Bardwell WA, Newman VA, Hollenbach KA, Jones L, Caan BJ, Pierce JP. Plasma carotenoids and recurrence-free survival in women with a history of breast cancer. J Clin Oncol 2005;23:6631–8. [DOI] [PubMed] [Google Scholar]
- 81.Ziegler RG. Epidemiologic studies of vitamins and cancer of the lung, esophagus, and cervix. Adv Exp Med Biol 1986;206:11–26. [DOI] [PubMed] [Google Scholar]
- 82.Goralczyk R. Beta-carotene and lung cancer in smokers: review of hypotheses and status of research. Nutr Cancer 2009;61:767–74. [DOI] [PubMed] [Google Scholar]
- 83.Neufcourt L, Assmann KE, Fezeu LK, Touvier M, Graffouillere L, Shivappa N, Hebert JR, Wirth MD, Hercberg S, Galan P, et al. . Prospective association between the dietary inflammatory index and metabolic syndrome: Findings from the SU.VI.MAX study. Nutr Metab Cardiovasc Dis 2015;25:988–96. [DOI] [PubMed] [Google Scholar]
- 84.Henderson VP, Massion AO, Clemow L, Hurley TG, Druker S, Hebert JR. A randomized controlled trial of mindfulness-based stress reduction for women with early-stage breast cancer receiving radiotherapy. Integr Cancer Ther 2013;12:404–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 85.West DS, DiLillo V, Bursac Z, Gore SA, Greene PG. Motivational interviewing improves weight loss in women with type 2 diabetes. Diabetes Care 2007;30:1081–7. [DOI] [PubMed] [Google Scholar]
- 86.Pagoto SL, McDermott MM, Reed G, Greenland P, Mazor KM, Ockene JK, Whited M, Schneider K, Appelhans B, Leung K, et al. . Can attention control conditions have detrimental effects on behavioral medicine randomized trials? Psychosom Med 2013;75:137–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 87.Du H, van der A DL, Ginder V, Jebb SA, Forouhi NG, Wareham NJ, Halkjaer J, Tjønneland A, Overvad K, Jakobsen MU, et al. . Dietary energy density in relation to subsequent changes of weight and waist circumference in European men and women. PLoS One 2009;4:e5339. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 88.Hutfless S, Gudzune KA, Maruthur N, Wilson RF, Bleich SN, Lau BD, Fawole OA, Anderson CA, Segal J. Strategies to prevent weight gain in adults: a systematic review. Am J Prev Med 2013;45:e41–51. [DOI] [PubMed] [Google Scholar]
- 89.Lindvall K, Jenkins P, Emmelin M, Scribani M, Norberg M, Larsson C, Weinehall L. Primary weight maintenance: an observational study exploring candidate variables for intervention. Nutr J 2013;12:97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 90.Vergnaud AC, Norat T, Romaguera D, Mouw T, May AM, Travier N, Luan J, Wareham N, Slimani N, Rinaldi S, et al. . Meat consumption and prospective weight change in participants of the EPIC-PANACEA study. Am J Clin Nutr 2010;92:398–407. [DOI] [PubMed] [Google Scholar]
- 91.Vistisen D, Witte DR, Tabák AG, Herder C, Brunner EJ, Kivimäaki M, Færch K. Patterns of obesity development before the diagnosis of type 2 diabetes: the Whitehall II cohort study. PLoS Med 2014;11:e1001602. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 92.Look AHEAD Research Group. Eight-year weight losses with an intensive lifestyle intervention: the look AHEAD study. Obesity (Silver Spring) 2014;22:5–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 93.Look AHEAD Research Group, Wing RR, Bolin P, Brancati FL, Bray GA, Clark JM, Coday M, Crow RS, Curtis JM, Egan CM, et al. . Cardiovascular effects of intensive lifestyle intervention in type 2 diabetes. N Engl J Med 2013;369:145–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 94.Brancati FL, Evans M, Furberg CD, Geller N, Haffner S, Kahn SE, Kaufmann PG, Lewis CE, Nathan DM, Pitt B, et al. . Midcourse correction to a clinical trial when the event rate is underestimated: the Look AHEAD (Action for Health in Diabetes) study. Clin Trials 2012;9:113–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 95.Saxe GA, Hebert JR, Carmody JF, Kabat-Zinn J, Rosenzweig PH, Jarzobski D, Reed GW, Blute RD. Can diet, in conjunction with stress reduction, affect the rate of increase in prostate specific antigen after biochemical recurrence of prostate cancer? J Urol 2001;166:2202–7. [PubMed] [Google Scholar]
- 96.Hébert JR, Hurley TG, Harmon BE, Heiney S, Hebert CJ, Steck SE. A diet, physical activity, and stress reduction intervention in men with rising prostate-specific antigen after treatment for prostate cancer. Cancer Epidemiol 2012;36:e128–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 97.Kones R, Rumana U, Merino J. Exclusion of ‘nonRCT evidence’ in guidelines for chronic diseases - is it always appropriate? The Look AHEAD study. Curr Med Res Opin 2014;30:2009–19. [DOI] [PubMed] [Google Scholar]
- 98.Hawkes AL, Chambers SK, Pakenham KI, Patrao TA, Baade PD, Lynch BM, Aitken JF, Meng X, Courneya KS. Effects of a telephone-delivered multiple health behavior change intervention (CanChange) on health and behavioral outcomes in survivors of colorectal cancer: a randomized controlled trial. J Clin Oncol 2013;31:2313–21. [DOI] [PubMed] [Google Scholar]
- 99.Wolin KY, Schwartz AL, Matthews CE, Courneya KS, Schmitz KH. Implementing the exercise guidelines for cancer survivors. J Support Oncol 2012;10:171–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 100.Lengacher CA, Johnson-Mallard V, Post-White J, Moscoso MS, Jacobsen PB, Klein TW, Widen RH, Fitzgerald SG, Shelton MM, Barta M, et al. . Randomized controlled trial of mindfulness-based stress reduction (MBSR) for survivors of breast cancer. Psychooncology 2009;18:1261–72. [DOI] [PubMed] [Google Scholar]
- 101.Haseen F, Murray LJ, O’Neill RF, O’Sullivan JM, Cantwell MM. A randomised controlled trial to evaluate the efficacy of a 6 month dietary and physical activity intervention for prostate cancer patients receiving androgen deprivation therapy. Trials 2010;11:86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 102.Nail LM. Fatigue in patients with cancer. Oncol Nurs Forum 2002;29:537. [DOI] [PubMed] [Google Scholar]
- 103.Pan A, Hu FB. Effects of carbohydrates on satiety: differences between liquid and solid food. Curr Opin Clin Nutr Metab Care 2011;14:385–90. [DOI] [PubMed] [Google Scholar]
- 104.Urban N, Self S, Kessler L, Prentice R, Henderson M, Iverson D, Thompson D, Byar D, Insull W, Gorbach SL, et al. . Analysis of the costs of a large prevention trial. Control Clin Trials 1990;11:129–46. [DOI] [PubMed] [Google Scholar]
- 105.Sacks FM, Svetkey LP, Vollmer WM, Appel LJ, Bray GA, Harsha D, Obarzanek E, Conlin PR, Miller ER 3rd, Simons-Morton DG, et al. . Effects on blood pressure of reduced dietary sodium and the Dietary Approaches to Stop Hypertension (DASH) diet. DASH-Sodium Collaborative Research Group. N Engl J Med 2001;344:3–10. [DOI] [PubMed] [Google Scholar]
- 106.Appel LJ, Van Horn L. Did the PREDIMED trial test a Mediterranean diet? N Engl J Med 2013;368:1353–4. [DOI] [PubMed] [Google Scholar]
- 107.Barnard ND, Cohen J, Jenkins DJ, Turner-McGrievy G, Gloede L, Green A, Ferdowsian H. A low-fat vegan diet and a conventional diabetes diet in the treatment of type 2 diabetes: a randomized, controlled, 74-wk clinical trial. Am J Clin Nutr 2009;89:1588S–96S. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 108.Jenkins DJ, Jones PJ, Lamarche B, Kendall CW, Faulkner D, Cermakova L, Gigleux I, Ramprasath V, de Souza R, Ireland C, et al. . Effect of a dietary portfolio of cholesterol-lowering foods given at 2 levels of intensity of dietary advice on serum lipids in hyperlipidemia: a randomized controlled trial. JAMA 2011;306:831–9. [DOI] [PubMed] [Google Scholar]
- 109.Sacks FM, Bray GA, Carey VJ, Smith SR, Ryan DH, Anton SD, McManus K, Champagne CM, Bishop LM, Laranjo N, et al. . Comparison of weight-loss diets with different compositions of fat, protein, and carbohydrates. N Engl J Med 2009;360:859–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 110.Jain M, Howe GR, Harrison L, Miller AB. A study of repeatability of dietary data over a seven-year period. Am J Epidemiol 1989;129:422–9. [DOI] [PubMed] [Google Scholar]
- 111.Bertheke Post G, de Vente W, Kemper HC, Twisk JW. Longitudinal trends in and tracking of energy and nutrient intake over 20 years in a Dutch cohort of men and women between 13 and 33 years of age: the Amsterdam growth and health longitudinal study. Br J Nutr 2001;85:375–85. [DOI] [PubMed] [Google Scholar]
- 112.Mikkilä V, Räsnän L, Raitakari OT, Marniemi J, Pietinen P, Rönnemaa T, Viikari J. Major dietary patterns and cardiovascular risk factors from childhood to adulthood. The Cardiovascular Risk in Young Finns Study. Br J Nutr 2007;98:218–25. [DOI] [PubMed] [Google Scholar]
- 113.Weisz JR, Chu BC, Polo AJ. Treatment dissemination and evidence-based practice: strengthening intervention through clinician-researcher collaboration. Clin Psychol Sci Pract 2004;11:300–7. [Google Scholar]
- 114.Green LW, Glasgow RE. Evaluating the relevance, generalization, and applicability of research: issues in external validation and translation methodology. Eval Health Prof 2006;29:126–53. [DOI] [PubMed] [Google Scholar]
- 115.The alpha-tocopherol, beta-carotene lung cancer prevention study: design, methods, participant characteristics, and compliance. The ATBC Cancer Prevention Study Group. Ann Epidemiol 1994;4:1–10. [DOI] [PubMed] [Google Scholar]
- 116.Lippman SM, Goodman PJ, Klein EA, Parnes HL, Thompson IM Jr, Kristal AR, Santella RM, Probstfield JL, Moinpour CM, Albanes D, et al. . Designing the Selenium and Vitamin E Cancer Prevention Trial (SELECT). J Natl Cancer Inst 2005;97:94–102. [DOI] [PubMed] [Google Scholar]
- 117.Sugiyama T, Tsugawa Y, Tseng CH, Kobayashi Y, Shapiro MF. Different time trends of caloric and fat intake between statin users and nonusers among US adults: gluttony in the time of statins? JAMA Intern Med 2014;174:1038–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 118.Hebert JR, Clemow L, Pbert L, Ockene IS, Ockene JK. Social desirability bias in dietary self-report may compromise the validity of dietary intake measures. Int J Epidemiol 1995;24:389–98. [DOI] [PubMed] [Google Scholar]
- 119.Hebert JR, Ma Y, Clemow L, Ockene IS, Saperia G, Stanek EJ, Merriam PA, Ockene JK. Gender differences in social desirability and social approval bias in dietary self report. Am J Epidemiol 1997;146:1046–55. [DOI] [PubMed] [Google Scholar]
- 120.Hébert JR, Peterson KE, Hurley TG, Stoddard AM, Cohen N, Field AE, Sorensen G. The effect of social desirability trait on self-reported dietary measures among multi-ethnic female health center employees. Ann Epidemiol 2001;11:417–27. [DOI] [PubMed] [Google Scholar]
- 121.Hebert JR, Hurley TG, Peterson KE, Resnicow K, Thompson FE, Yaroch AL, Ehlers M, Midthune D, Williams GC, Greene GW, et al. . Social desirability trait influences on self-reported dietary measures among diverse participants in a multicenter multiple risk factor trial. J Nutr 2008;138:226S–34S. [DOI] [PubMed] [Google Scholar]
- 122.Howard BV, Van Horn L, Hsia J, Manson JE, Stefanick ML, Wassertheil-Smoller S, Kuller LH, LaCroix AZ, Langer RD, Lasser NL, et al. . Low-fat dietary pattern and risk of cardiovascular disease: the Women’s Health Initiative Randomized Controlled Dietary Modification Trial. JAMA 2006;295:655–66. [DOI] [PubMed] [Google Scholar]
- 123.Lanza E, Yu B, Murphy G, Albert PS, Caan B, Marshall JR, Lance P, Paskett ED, Weissfeld J, Slattery M, et al. . The polyp prevention trial continued follow-up study: no effect of a low-fat, high-fiber, high-fruit, and -vegetable diet on adenoma recurrence eight years after randomization. Cancer Epidemiol Biomarkers Prev 2007;16:1745–52. [DOI] [PubMed] [Google Scholar]
- 124.Schatzkin A, Lanza E, Corle D, Lance P, Iber F, Caan B, Shike M, Weissfeld J, Burt R, Cooper MR, et al. . Lack of effect of a low-fat, high-fiber diet on the recurrence of colorectal adenomas. Polyp Prevention Trial Study Group. N Engl J Med 2000;342:1149–55. [DOI] [PubMed] [Google Scholar]
- 125.Gibson TM, Ferrucci LM, Tangrea JA, Schatzkin A. Epidemiological and clinical studies of nutrition. Semin Oncol 2010;37:282–96. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 126.Blumberg J, Heaney RP, Huncharek M, Scholl T, Stampfer M, Vieth R, Weaver CM, Zeisel SH. Evidence-based criteria in the nutritional context. Nutr Rev 2010;68:478–84. [DOI] [PubMed] [Google Scholar]
- 127.Byers T. What can randomized controlled trials tell us about nutrition and cancer prevention? CA Cancer J Clin 1999;49:353–61. [DOI] [PubMed] [Google Scholar]
- 128.Zwarenstein M, Treweek S, Gagnier JJ, Altman DG, Tunis S, Haynes B, Oxman AD, Moher D; CONSORT group; Pragmatic Trials in Healthcare (Practihc) group. Improving the reporting of pragmatic trials: an extension of the CONSORT statement. BMJ 2008;337:a2390. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 129.Hernán MA, Alonso A, Logan R, Grodstein F, Michels KB, Willett WC, Manson JE, Robins JM. Observational studies analyzed like randomized experiments: an application to postmenopausal hormone therapy and coronary heart disease. Epidemiology 2008;19:766–79. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 130.Lei H, Nahum-Shani I, Lynch K, Oslin D, Murphy SAA. “SMART” design for building individualized treatment sequences. Annu Rev Clin Psychol 2012;8:21–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 131.Collins LM, Murphy SA, Strecher V. The multiphase optimization strategy (MOST) and the sequential multiple assignment randomized trial (SMART): new methods for more potent eHealth interventions. Am J Prev Med 2007; 32(5 Suppl)S112–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 132.Riley WT, Rivera DE, Atienza AA, Nilsen W, Allison SM, Mermelstein R. Health behavior models in the age of mobile interventions: are our theories up to the task? Transl Behav Med 2011;1:53–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 133.Kumar S, Nilsen WJ, Abernethy A, Atienza A, Patrick K, Pavel M, Riley WT, Shar A, Spring B, Spruijt-Metz D, et al. . Mobile health technology evaluation: the mHealth evidence workshop. Am J Prev Med 2013;45:228–36. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 134.Lillie EO, Patay B, Diamant J, Issell B, Topol EJ, Schork NJ. The n-of-1 clinical trial: the ultimate strategy for individualizing medicine? Per Med 2011;8:161–73. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 135.West SG, Duan N, Pequegnat W, Gaist P, Des Jarlais DC, Holtgrave D, Szapocznik J, Fishbein M, Rapkin B, Clatts M, et al. . Alternatives to the randomized controlled trial. Am J Public Health 2008;98:1359–66. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 136.Hodges RE. The role of a metabolic ward in nutritional studies. Am J Clin Nutr 1971;24:930–3. [DOI] [PubMed] [Google Scholar]
- 137.Mozaffarian D, Micha R, Wallace S. Effects on coronary heart disease of increasing polyunsaturated fat in place of saturated fat: a systematic review and meta-analysis of randomized controlled trials. PLoS Med 2010;7:e1000252. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 138.Spruijt-Metz D, Hekler E, Saranummi N, Intille S, Korhonen I, Nilsen W, Rivera DE, Spring B, Michie S, Asch DA, et al. . Building new computational models to support health behavior change and maintenance: new opportunities in behavioral research. Transl Behav Med 2015;5:335–46. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 139.Hébert JR, Hurley TG, Steck SE, Miller DR, Tabung FK, Kushi LH, Frongillo EA. Reply to E Archer and SN Blair: implausible data, false memories, and the status quo in dietary assessment. Adv Nutr 2015;6:230–3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 140.Heitmann BL. Social desirability bias in dietary self-report may compromise the validity of dietary intake measures. Implications for diet disease relationships [letter; author's response] Int J Epidemiol 1996;25:222–5. [DOI] [PubMed] [Google Scholar]
- 141.Hebert JR, Ma Y, Ebbeling CB, Matthews CE, Ockene IS. Self-report data. In: Ockene IS, Burke LE, editors. Compliance in healthcare and research. Armonk (NY): Futura; 2001. p. 163–79. [Google Scholar]
- 142.Stumbo PJ. New technology in dietary assessment: a review of digital methods in improving food record accuracy. Proc Nutr Soc 2013;72:70–6. [DOI] [PubMed] [Google Scholar]
- 143.Thompson FE, Subar AF, Loria CM, Reedy JL, Baranowski T. Need for technological innovation in dietary assessment. J Am Diet Assoc 2010;110:48–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 144.Gemming L, Doherty A, Kelly P, Utter J, Ni Mhurchu C. Feasibility of a SenseCam-assisted 24-h recall to reduce under-reporting of energy intake. Eur J Clin Nutr 2013;67:1095–9. [DOI] [PubMed] [Google Scholar]
- 145.Gemming L, Utter J, Ni Mhurchu C. Image-assisted dietary assessment: a systematic review of the evidence. J Acad Nutr Diet 2015;115:64–77. [DOI] [PubMed] [Google Scholar]
- 146.Illner A-K, Freisling H, Boeing H, Huybrechts I, Crispim S, Slimani N. Review and evaluation of innovative technologies for measuring diet in nutritional epidemiology. Int J Epidemiol 2012;41:1187–203. [DOI] [PubMed] [Google Scholar]
- 147.Dong Y, Hoover A, Scisco J, Muth E. A new method for measuring meal intake in humans via automated wrist motion tracking. Appl Psychophysiol Biofeedback 2012;37:205–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 148.Hingle M, Yoon D, Fowler J, Kobourov S, Schneider ML, Falk D, Burd R. Collection and visualization of dietary behavior and reasons for eating using Twitter. J Med Internet Res 2013;15:e125. [DOI] [PMC free article] [PubMed] [Google Scholar]