4.1 Issues related to recruitment, retention, and adherence
Although many people with MS express an interest in dietary modification as a potential for disease modification, participating in a dietary intervention study may present considerable burden for participants, thus presenting challenges to recruitment, retention, and adherence. It is imperative that prior to enrollment, potential participants fully understand the design of the study and the potential dietary interventions they may be asked to follow, what will be expected of them with respect to time, finances, and travel (to controlled feeding sites, training sessions, other study visits, etc.).
There are many threats to retention in dietary intervention trials, some of which are generalized and others of which are specific to dietary and other behavioral studies. In randomized dietary intervention trials, in which blinding may be difficult, members of a placebo arm may be more likely to lose interest in completing study-related assessments if they don't perceive some benefit of participation. A common strategy to address this includes a “wait-list control” option, in which the participants initially randomized to usual care/placebo ultimately gain access to the (in this case, dietary) intervention (
Yadav et al., 2016
Low-fat, plant-based diet in multiple sclerosis: a randomized controlled trial.
). However, the study's outcome should be considered if the wait-list control option is employed, especially if longer-term outcomes like disability are desired as these participants may have a prolonged wait-list period in which drop-out or other temporal changes may occur. Another option involves an active control arm, in which participants learn, for example, about some other aspect of MS other than diet or other lifestyle interventions. The possibility of this leading to greater engagement in healthy eating or other behavioral changes that may impact the outcomes of interest (e.g. a control intervention improving adherence to medications and, therefore, reducing subsequent MS activity) must be strongly considered. It may be best to account for this possibility when calculating sample size estimates, as well as to deliver similar education to the active arm participants in order to minimize differences in the arms of the study other than the dietary intervention. Retention rates in dietary interventions are often variable, which also should be considered when calculating sample size estimates that adequately account for potential dropout.
Several strategies are often employed in dietary intervention studies to promote adherence and retention. These strategies include maintaining regular contact with study participants (assessing and encouraging adherence), ensuring study staff are readily available to answer questions quickly (by email and/or phone), and continuously encouraging and expressing appreciation for participants. Whether such strategies are truly beneficial in the MS population is not known. For example, in one of our pilot studies, participants found regular phone calls assessing adherence to a self-directed dietary intervention to be burdensome (
Fitzgerald et al., 2018
Effect of intermittent vs. daily calorie restriction on changes in weight and patient-reported outcomes in people with multiple sclerosis.
) and in another small pilot study, regular text messages as a strategy to improve adherence did not, in fact, appear to be associated with an improvement therein (
Roman Fox et al., 2017
Roman Fox, S. et al. A pilot study evaluating changes in clinical outcomes with weight loss in people with multiple sclerosis. (2017).
Additional activities that may improve adherence and retention include providing ongoing education regarding the scientific rationale for conducting the study, both to participants and, potentially, to other members of the household who may be affected by the changes required of the study participant. Group-based study activities may provide additional motivation, but the composition of the groups must be considered carefully. For example, participants or study staff may be blinded to the specifics of the intervention, or active arm or control participants may become dissuaded from continuing in their assigned group based on interacting with members of the opposite arm. In all such endeavors, care must be taken to minimize any differential treatment of the study arms while not dissuading one group (typically, the control group) from becoming less interested in continued participation.
Even if participants do not adhere to the diet as dictated by the study, rigorous studies will attempt to continue to engage participants to complete study-related assessments so as to minimize bias due to loss to follow-up. Randomized trials typically assess the outcome from such participants using an “intention to treat” methodology, though sensitivity analyses may consider including the degree of adherence to the prescribed intervention in order to assess the impact of the intervention in those individuals who followed the plan more consistently.
Measuring adherence to prescribed diet plans, whether in a controlled feeding setting or in a self-directed study, can be challenging. Even subjects participating in a traditional controlled feeding study in which consumption of most study-provided foods is monitored may still deviate from the prescribed intervention in the off-site setting. While no true gold standard for assessing dietary intake exists, a widely accepted dietary intake assessment method involves the collection of multiple 24-hour recalls by a trained research dieticians (
Buzzard and Sievert, 1994
- Buzzard I.M.
- Sievert Y.A.
Research priorities and recommendations for dietary assessment methodology. First International Conference on dietary assessment methods.
). This methodology introduces expense and requires coordination of schedules between the study dietician and the participant, as well as time for the actual interview (e.g. 30 to 40 min), which in our experience was burdensome for many participants. A web-based, automated 24-hour recall is an alternative to the in-person interview (
Subar et al., 2001
Comparative validation of the block, willett, and national cancer institute food frequency questionnaires : the eating at America's Table Study.
), however, some research suggests they may have reduced accuracy when compared to other validated dietary assessment methods (
Yuan et al., 2018
Relative validity of nutrient intakes assessed by questionnaire, 24-hour recalls, and diet records as compared with urinary recovery and plasma concentration biomarkers: findings for women.
). Asking participants to complete detailed diet records for specified period of time is another widely-accepted dietary assessment method (
). However, a study dietician with access to nutrient databases is still required to analyze recorded intakes, which introduces additional expense. With diet records, it is possible that participants will incompletely record dietary intake (especially if the record-keeping period extends for a long period of time). Repeated food frequency questionnaires (FFQs) (
Subar et al., 2001
Comparative validation of the block, willett, and national cancer institute food frequency questionnaires : the eating at America's Table Study.
Willett et al., 1985
Reproducibility and validity of a semiquantitative food frequency questionnaire.
Rimm et al., 1992
Reproducibility and validity of an expanded self-administered semiquantitative food frequency questionnaire among male health professionals.
) may reduce the need for substantial study staff involvement and may improve timing flexibility for study participants, but they still do require 30–60 min to complete. FFQs can be administered online or as a paper hardcopy. Keep in mind that participants may lose paper forms if they take them home, and study staff may need to remind participants repeatedly to complete the surveys. Hardcopy FFQs are often completed on scantron forms (e.g. machine-readable answer sheets for multiple-choice surveys), which can be challenging for people with reduced manual dexterity or limited upper limb mobility, and study staff may need to assist some participants with data collection. Brief FFQs or screeners are designed to capture the foods that contribute to > 90% of a particular nutrient, such as vitamin D, sodium, or dietary fat. These can reduce participant burden but may also limit functionality of the measure.
Whether smartphone apps or other technology-enabled measures are valid remains to be determined but may, in theory, offer improved affordability and patient satisfaction. Additionally, for each type of self-reported dietary assessment method, participants may be less likely to report intake of “undesirable” foods (e.g. foods which are not consistent with the study diet). While biological markers of nutritional intake may be reasonable to consider for some types of dietary interventions (e.g. 24-h urine sodium measures in a low-salt dietary intervention or detection of urinary ketones in a ketogenic dietary intervention (
)), studies of broad dietary change often do not have a “gold standard” biological measure that is useful as a marker of adherence.
Of note, how an investigator would like to monitor adherence can also help guide investigators in choosing among the different dietary assessment methods. Theoretically, 24-hour recalls and diet records measure actual consumption, while FFQs measure typical intake over a period of time; the reference intake timeframe for biological markers can vary depending on the marker (
). Thus, the investigator's choice of how to monitor adherence (e.g. measuring average consumption of a macronutrient over time vs. following a prescribed diet plan) will determine choice of measure.
Because FFQs ask people to reflect on intakes over long periods of time, they are less susceptible than the 24-hour recall or the food record to errors that arise from day-to-day variability in diet (
). However, attempting to average a respondent's intake over longer time intervals introduces new errors of estimation; it can be difficult for respondents to conceptualize average intakes over long periods of time.
In addition to measuring adherence, it is important to measure potential confounders, both at baseline and, when relevant, throughout the course of the study. For example, timing of meal intake may influence some outcomes and thus, depending on the specific study, measuring typical intake timing may be useful. Other lifestyle factors, such as exercise habits, body weight at baseline, cigarette smoking, alcohol intake, vitamin D supplement use, and sleep, all of which may vary with diet, should be assessed so that differences in such behaviors can be assessed and, if needed, accounted for in statistical analyses.
In summary, well-designed research studies that employ optimal methods to improve retention and measure adherence are needed to advance our understanding of the impact of diet on MS. People with MS are more likely to experience depression, anxiety, fatigue, and mobility restrictions, challenges that must be considered in the design of intervention administration and follow-up. The insights gained from such studies may also help to promote incorporation of an identified beneficial diet, or of a more broadly healthy diet, in the real-world setting for people with MS.
4.2 Issues related to the collection of biospecimens
Many dietary intervention studies, particularly those of short duration, may seek to assess the impact of the intervention on biological outcomes, whether as a primary aim of the study or after acquiring funding for future, secondary investigations. Careful planning and conduct of specimen collection will enhance the quality of subsequent studies using previously collected biological samples. In addition, investigators should thoughtfully balance the potential “ideal” of repeated collections with the impact of this decision on study participants. For example, collecting samples too frequently may deter recruitment, reduce retention, increase missing data, and/or diminish the representativeness of the study sample (i.e. enrolling those willing to incur more invasive data collection can introduce selection bias).
Collecting blood is a nearly ubiquitous feature of studies that investigate biological outcomes. In MS, processing blood to obtain serum or plasma is common, and these subcomponents are often used to conduct metabolomic and lipidomic analyses. Many MS diet studies also report serum cytokine levels, however it may be more informative to assess peripheral blood mononuclear cells (PBMCs) in order to characterize an immunophenotype and document change related to a dietary intervention (
). Collecting PBMCs, however, requires a substantial investment of time and expertise in order to process the specimens and conduct subsequent studies. Some studies may wish to collect genetic material, such as DNA or RNA, particularly if the sample size is large enough to support analyzing genetic contributions to the effects of the diet or change in gene expression as a result of dietary intervention.
Standardization of blood collection and processing is critical to minimizing heterogeneity that may make results less interpretable. Study personnel should record subject-specific information and be aware that this information may influence future studies using the samples, including the sample collection date and time and the time of the last meal/fasting status. Ideally, single-center studies will use the same personnel to collect and process blood, although this is not always feasible. Whether collecting blood at one or multiple sites, it is good practice to use the same blood collection and processing materials and to minimize heterogeneity in sample handling (e.g. create standards for collection, processing, transfer, and storage of specimens). Multisite studies should also standardize the timing and methods for specimen shipping, considering the weather (changes in temperature may cause shipping delays and/or sample degradation) and the timing of weekend and holiday schedules for shippers and receivers.
Urine samples can provide a quantitative measure of adherence for some dietary intervention studies (e.g. low-sodium diet). Collecting 24-h samples is often ideal (e.g. quantifying sodium intake), but whether to combine urine samples into one collection kit or to collect each sample in a separate container so that circadian changes can be assessed may depend on the anticipated use of the samples. Of additional consideration for people with MS is that urinary frequency, retention, and nocturia may have implications for studies seeking to take into account the time of day in which samples are collected. Urinary symptoms or their treatment should be considered in the study design phase and documented at the time of sample collection (
Mauruc et al., 2017
Nocturnal urinary disorders and multiple sclerosis: clinical and urodynamic study of 309 patients.
Stool collection for gut microbiota analysis may help elucidate mechanisms of dietary interventions and aid in adherence assessment. The first step in planning stool collection involves making decisions regarding anticipated sample use. For example, if the anticipated use is solely 16S sequencing for bacterial identification, a simple dry collection swab will likely be sufficient and the least costly option. If, however, experiments are planned that require the microbes to function upon thawing (for example, experiments in germ-free mice), a larger sample using a “wet” kit is needed. Thorough review with collaborators should direct kit selection. The same holds true for planning sample storage conditions. Samples should be stored and processed together to avoid batch effects, when feasible.
Once a kit has been selected, it is important to continue with the same exact kit through study completion. Uniformity is crucial for participant collection and shipping. Providing participants a detailed yet simple and clear instruction sheet with illustrations will increase the likelihood that the sample is collected correctly. Instructions should include acceptable days of the week for sample shipment to avoid arrivals to the lab on weekends and holidays. Depending on anticipated use of the samples, instructions should include preferred time of day for sample collection (often first bowel movement of the day is recommended) and considerations for constipation, which is often an issue for people with MS (
Comparison of prevalence, related medical history, symptoms, and interventions regarding bowel dysfunction in persons with multiple sclerosis.
). A member of the research team should review the instructions with each participant and remain available for questions. This is particularly important for participants reporting cognitive issues. Providing participants with a pre-paid label affixed to the return shipping envelope or box can increase the return rate. When selecting a shipment method, cost and the preservation of sample quality must be considered. Prior to settling on a shipment mechanism, investigators should explore different options with a “dummy” package to discern the length of transit time and ensure the sample arrives at the correct temperature (for most kits this means the included ice packs are still cold and the sample is still frozen though, there are some room temperature kits available). Many institutions have access to discounted rates with particular shipping providers, which should be investigated prior to the initiation of the study.