A Review of Recruitment, Adherence and Drop-Out Rates in Omega-3 Polyunsaturated Fatty Acid Supplementation Trials in Children and Adolescents

Introduction: The influence of n-3 long-chain polyunsaturated fatty acids (n-3 LCPUFA) supplementation on health outcomes has been studied extensively with randomized controlled trials (RCT). In many research fields, difficulties with recruitment, adherence and high drop-out rates have been reported. However, what is unknown is how common these problems are in n-3 LCPUFA supplementation studies in children and adolescents. Therefore, this paper will review n-3 LCPUFA supplementation studies in children and adolescents with regard to recruitment, adherence and drop-out rates. Methods: The Web of Science, PubMed and Ovid databases were searched for papers reporting on RCT supplementing children and adolescents (2–18 years) with a form of n-3 LCPUFA (or placebo) for at least four weeks. As a proxy for abiding to CONSORT guidelines, we noted whether manuscripts provided a flow-chart and provided dates defining the period of recruitment and follow-up. Results: Ninety manuscripts (reporting on 75 studies) met the inclusion criteria. The majority of the studies did not abide by the CONSORT guidelines: 55% did not provide a flow-chart, while 70% did not provide dates. The majority of studies provided minimal details about the recruitment process. Only 25 of the 75 studies reported an adherence rate which was on average 85%. Sixty-five of the 75 studies included drop-out rates which were on average 17%. Conclusion: Less than half of the included studies abided by the CONSORT guidelines (45% included a flow chart, while 30% reported dates). Problems with recruitment and drop-out seem to be common in n-3 LCPUFA supplementation trials in children and adolescents. However, reporting about recruitment, adherence and dropout rates was very heterogeneous and minimal in the included studies. Some techniques to improve recruitment, adherence and dropout rates were identified from the literature, however these techniques may need to be tailored to n-3 LCPUFA supplementation studies in children and adolescents.


Introduction
Fatty acids, and especially the omega-3 long-chain polyunsaturated fatty acids (n-3 LCPUFA), are being researched extensively for a wide array of health outcomes varying from, but not exclusive to, cardiovascular diseases, depression and cognition [1][2][3]. As in every health related field, randomized controlled supplementation trials are the gold standard to demonstrate efficacy of n-3 LCPUFA [4]. For these trials, voluntary participants are needed, however recruitment of participants can be challenging, especially when it involves research in children and adolescents (<18 years) [5]. It has been reported 3. Results

Study Characteristics
The original search led to 2163 hits. Upon first screening, 1656 articles were excluded, additional screening of the whole articles led to a further exclusion of 173 articles. Additional checking of the reference lists of reviews, included articles and forward checking led to 15 more studies being included (see Figure 1, adapted from [57]). Thus, in total, 90 papers, describing 75 studies, were included in this review. The characteristics of these studies can be found in Table 1. Fifteen studies focussed on healthy children. The other 60 studies focused on children with a disorder or disease, with attention deficit hyperactivity disorder (ADHD) being the most studied disorder (n = 21) (see Table 1). The majority of studies focussed on children (defined as aged between 2 and 12 years, n = 38) or both children and adolescents (n = 31). A minority of studies focussed only on adolescents (n = 6) (see Table 1). Duration of study varied from 4 to 52 weeks, with the majority of studies lasting 26 weeks or less (n = 59, 79%, see Table 2. Number of measurement moments (i.e., how often did participants have to come to the research facility/how often did they have to fill out questionnaires) varied from 2 to 16 with a mean of 3.7 (SD 2.7), the number of different measurements per moment varied from 1 to 19 with a mean of 4.9 (SD 3.7).

Recruitment
Most of the studies included in this review did not report the number of children or adolescents that were invited to participate in the study, as only 11 out of 75 studies mentioned the number of participants that were invited. The total number of people invited to participate varied from 46 to 3562 (Mean (M) = 804.5, SD = 1083.28). The percentage of invited participants that eventually started the study ranged from 2.4 to 87% (see Table 3).
Forty out of 75 studies mentioned the number of participants that responded to the invitation or were screened for the study and this varied from 30 to 1556, with 12 to 100% of these people actually starting the study.
Most studies did not specify the exact method(s) of recruitment, mostly just mentioning the recruitment setting. Most studies recruited their participants from a hospital or outpatient clinic setting (n = 33). Other settings from which participants were recruited were schools (n = 23) and the community (n = 15). Nine studies reported multiple settings for recruitment; one study recruited participants from a summer camp for children with ADHD and other disorders; one study recruited from an online registry; one study recruited participants from those who participated in earlier studies; and eight studies did not mention the recruitment setting.

Supplementation
Most studies used capsules as the form of supplementation (n = 57), however there were also some other approaches (see Table 2). The number of capsules that participants were instructed to take also varied widely from 1 to 12 capsules a day, with some studies basing the dose per body weight of the participant (see Table 2). Moreover, a huge range of different placebos was used (see Table 2).

Adherence
The included studies mentioned a wide variety of methods to measure adherence: capsule count (or product weighting) (n = 30), diaries or tick-off forms (n = 13), interviews face to face/via phone/ via e-mail (n = 11), taking the capsules under supervision (n = 8), and blood values (n = 5) (see Table 4). Thirteen studies used more than one method to assess adherence. Furthermore, 23 studies did not specify how or whether they assessed adherence. The way in which adherence was reported in the studies also varied greatly. Some studies mentioned percentages of capsules taken, the average number of capsules taken per day, blood values, or just mentioned that adherence was good or mentioned how many students were excluded due to non-adherence.
Twenty-five studies mentioned a specific percentage of adherence, which varied from 60 to 97%, mean 85% (SD 10.1). In addition, the levels of capsules that needed to be taken to be considered as being adherent differed per study, varying from 65 to 90%. Other studies defined adherence as the number of days of not taking capsules.            Looking at the adherence percentage between studies in healthy and diseased children, there seemed to be a slightly lower average adherence in diseased children (M = 83.7%, SD = 11.9), compared to healthy children (M = 87.6%, SD = 7.1). When we looked at the different age groups recruited, there seemed to be a lower average adherence in the child only group (M = 82.5%, SD = 9. . Seven studies, that reported adherence, reported that participants consumed capsules more than once a day while 12 studies, that reported adherence, mentioned that the capsules were only taken once a day. There was no difference in average adherence between those two methods of supplementation (M = 87.3%, SD = 9.4 vs. M = 87.6%, SD = 7.8). Fifteen studies, reporting adherence, mentioned talking to parents or participants either via telephone or face to face (or via e-mail) during the study about the supplementation to increase adherence [61,67,69,70,72,90,91,107,111,113,131,[134][135][136]143]. The studies that included a phone call did not have a higher average adherence rate (M = 81.5%, SD = 9.5) than those that did not include a phone call (M = 86.2%, SD = 10.3). There were three studies that provided some form of incentive [98,107,146], however only one of these studies reported an adherence percentage. Forty-six studies mentioned that they took either blood or cheek samples, but only five studies mentioned that they used blood as an adherence measure [61,85,86,117,134].

Drop-Out
Sixty-five of the 75 included studies mentioned a drop-out rate or included numbers which made it possible to calculate the drop-out rate. The average drop-out was 17% (SD 13%), but it varied between 0% and 58% (see Table 4). There was no clear difference in average drop-out rate between studies in healthy (mean = 16.5%, SD = 11.5) and diseased populations (M = 17.9%, SD = 13.7). There was a difference in average drop-out with regard to the recruited age group: children M = 15%, SD = 11.1), adolescents (M = 12.3%, SD = 7.3) or both (M = 21.5%, SD = 14.9); with a higher average drop-out rate in the combined age group.
There was also no clear difference in mean drop-out between recruitment setting: hospital (M = 15%, SD = 11. Eight studies who reported drop-out rate mentioned that capsules were taken under supervision, this seemed to lead to somewhat lower average drop-out rate (M = 13%, SD = 15.6), compared to the 57 studies in which participants did not take the capsules under supervision (M = 17.9%, SD = 15.6). Sixteen studies that reported drop-out rate divided the capsules over multiple intake moments (M = 17.2%, SD = 9.3). This did not seem to increase or decrease the average drop-out rate if compared to those studies that specified one intake moment (M = 17.1%, SD = 13.6). Fourteen studies that noted drop-out rate reported that they contacted the participants during the study. Studies that did so seemed to have a slightly higher average drop-out rate (M = 20.4%, SD = 11.4) than studies that did not contact participants during the study (M = 16.5%, SD = 14.4). Of the studies that reported giving participants an incentive, two mentioned a drop-out rate, this was on average 15.3% (SD 20.4). Studies that did not state an incentive had an average drop-out rate of 17.4% (SD 12.7). Of the 65 studies that mentioned a drop-out rate, 50 specified a reason for drop-out (six did not have drop-out, and nine did not specify the drop-out). Fifty-two different reasons for drop-out were mentioned, with the most common reasons mentioned being lost to follow-up, poor or no compliance or inability to take supplement.

Discussion
We conducted a thorough review to examine recruitment, adherence and drop-out rates in n-3 LCPUFA supplementation studies in children and adolescents, in order to identify strategies which can be implemented to improve those rates. Even though the CONSORT guidelines clearly state what data need to be included in the report of a RCT, the majority of the included studies did not provide a flow-chart (55% did not) or the dates defining the period of recruitment and follow-up (70% did not).

Recruitment
The majority of studies provided minimal details about the recruitment process. The low number of studies that reported the number of participants that they invited and screened is, however, not uncommon in research studies as similar numbers were reported by Toerien et al. who studied 129 studies in six major journals [148]. The literature does give some suggestions for methods that could increase recruitment; for example, telephone calls to those who do not reply, an opt-out system (participants contact the researchers if they do not want to participate, please do note that this is not legal in all countries), including incentives, making trials open, and in person recruitment [149]. The use of clinical referral is also suggested to be related to higher recruitment rates, as most patients will have a trusting relationship with their doctor [150]. When we looked at the research setting (hospital, community, school), though, the mean started/invited rate and mean started/responded rate seemed to be slightly higher in the school setting. However, in the studies that looked at diseased populations, the average percentage efficiency of started/invited and started/responded was higher than studies looking at healthy populations (M = 43.5% vs. M = 36.2% and M = 63.5% vs. M = 53.1%, respectively).
It has been shown that in adolescents, giving monetary incentives does improve response rates and has a positive effect on their willingness to participate in studies [151]. However, the provision of monetary incentives might be considered unethical in children/adolescents [152,153]. One might thus consider a form on non-monetary incentive, for example in Food2Learn participants received a cinema voucher [19]. In the current review, there were only three studies that provided an incentive and these studies did not have remarkably higher recruitment rates. Hence, more studies that do provide incentives are needed to elucidate whether or not incentives improve recruitment. Moreover, there are myriad reasons as to why somebody would or would not participate in a study. There are participant characteristics which in adults have been associated with a higher chance of non-participation such as younger age, being male, lower social economic status, and lower education level [154,155]. However, in the limited number of studies on recruitment in children, no association between age or sex has been seen, although the education level of parents was associated with higher enrolment rates [5].
Beliefs about the effectiveness of the treatment may also play a role. Examples of reasons as to why adolescents did not participate informally given in Food2learn included: (1) the belief that n-3 LCPUFA are not effective in improving health; (2) the belief that they already consume sufficient amount of n-3 LCPUFA/already eat healthy; (3) the belief that participation will take too much time/effort; and (4) lack of interest in research in general. These factors should be taken into account during the research process and it seems wise to include explanations that most people do not get enough n-3 LCPUFA in their diet as well as elaborating on the possible health benefits of n-3 LCPUFA specific to the age group being assessed.

Adherence
Just 25 studies mentioned a specific adherence percentage, which varied between 60% and 97% with a mean of 85%. Moreover, most studies included in the current review used indirect adherence assessment methods (i.e., diaries, interviews, and capsules counts) which are all subject to problems with reporting bias and errors or intentional manipulation [156]. More direct methods such as the determination of fatty acid levels in the blood seems to be the most reliable method to assess adherence, which was done in only five studies. However, it should be noted that taking blood samples in younger children might not be acceptable for all parents or ethical committees and could therefore lead to lower recruitment numbers.
In the current review, there was no difference in mean adherence in those studies where participants received a telephone call to try and increase adherence compared to those in which participants received no telephone call (M = 81.5% vs. M = 87.7%). There were only three studies that provided an incentive and only one of these studies provided an adherence percentage, which was 75%. There seemed to be a higher average adherence of capsules (M = 88.2%) compared to other forms of supplementation (M = 74.8%, M = 81.5%, M = 80.3%, for food, drink and other forms, respectively). Lastly, there was no difference in the mean adherence between those who took capsules multiple times a day compared to those who took capsules only once a day (M = 87.3% vs. M = 87.6%). It is however important to remember that all these findings are based on only 25 studies that mentioned an adherence rate.
Other studies suggest factors that are associated with higher adherence in children and adolescents, these include: sociodemographic factors (i.e., older children and older adolescents are less likely to be adherent, and boys are less likely to be adherent), disease associated factors (i.e., if the disease also has positive symptoms the person is less likely to be adherent), the belief and attitude that a person has towards the treatment (i.e., those that belief that the treatment will be effective are more likely to be adherent), their mood (i.e., those with depression are less likely to be adherent) and the social context (i.e., those who are supported by family and friends are more likely to be adherent) [157,158]. Methods to increase adherence rates have also been suggested. Methods that have been employed to increase adherence include: educating participants about adherence, making medicine (or supplementation) more palatable, providing incentives/tokens, and involving parents or schools [159,160]. However, one must take into consideration that the vast majority of studies looking at which methods can help increase adherence have been executed in a medical setting with patients requiring medications and these results do not by definition translate to nutritional interventions in healthy participants or those with diagnosed disorders such as ADHD.
Some suggestions for improving adherence for n-3 LCPUFA supplementation studies may include: providing sufficient information about the importance of adherence (i.e., explaining the importance of adherence to get valid results), getting parents involved, and providing appropriate incentives [159,160].

Drop-Out
In the current review, the average drop-out was 17% (range 0-58%). Three studies mentioned some form of incentive [98,107,146] and they reported a slightly lower average drop-out than those that did not use (or did not report) an incentive (M = 15.3% vs. M = 17.4%). There were differences in average drop-out rates between continents, with drop-out rates being higher in Australia (M = 34.9%), USA/Canada (M = 23.1%), and the Middle East (M = 20.1%) compared to Europe (M = 13.3%), Africa (M = 11.6%) and Asia (M = 6.8%). We can only speculate about explanations for this difference (e.g., individualistic vs. collective societies) and do point out that these differences have to be interpreted with caution as the number of studies per continent did differ greatly. A number of methods to decrease drop-out in studies involving adults has been suggested. They include emphasizing the benefit of participation, flexible scheduling of appointments, regular positive communication from the research team to the participants (e.g., birthday and Christmas cards, newsletters, etc.), a consistent research staff so participants can build a bond with the researchers, and appropriate incentives [150,[161][162][163]. Other strategies that have been suggested include decreasing the complexity of the treatment and limiting the number of follow-up visits to the bare minimum [164]. Furthermore, a combination of multiple strategies is suggested to be most effective in increasing retention [161,164]. All these methods to decrease drop-out have been studied in adults; more research on methods to decrease drop-out in children and adolescents in RCT is warranted.
Suggestions for decreasing drop-out in n-3 LCPUFA supplementation trial include: keeping in regular contact with the participants, providing flexible appointment possibilities, providing incentives for participants and providing reminders. With regard to the supplement, one should keep the regime as simple as possible e.g., one (concentrated) capsule per day [150,[161][162][163][164].

Strengths and Limitations
Limitations of the current review include the fact that many of the included studies did not report all data on recruitment, dropout and (assessment of) adherence. Due to the incomplete reporting of data, results should be viewed with caution. The main advantage of the current review is the fact that we included all studies investigating n-3 LCPUFA supplementation in children/adolescents regardless of whether they were healthy children/adolescents or children/adolescents with a disease or disorder.

Conclusions
The conclusions drawn are based on minimal reporting from the included studies in this review. Less than half of the included studies abided by the CONSORT guidelines. Problems with recruitment and drop-out seem to be common in n-3 LCPUFA supplementation trials in children and adolescents. However, since the reporting about recruitment, adherence and dropout rates was very heterogeneous and minimal in the included studies, we cannot provide specific suggestions to improve LCPUFA supplementation studies in children and adolescents.

Recommendations
It is important for future studies to report on recruitment effort and rate, adherence (including the method of assessing adherence) and drop-out rates according to the CONSORT Guidelines.
Suggestions from other scientific areas to increase recruitment, adherence and minimize drop-out include: the provision of sufficient information about the importance of adherence (i.e., explaining the importance of adherence to get valid results), getting parents involved, provision of appropriate incentives, emphasizing the benefit of participation, being flexible with the scheduling of appointments, the research team engaging in regular positive communication with the participants, having a consistent research staff member so participants can build a bond with the researchers and to keep the supplementation regime as simple as possible.