In a 25-week crossover tea trial a very high attrition rate, especially among smokers, was observed. The dropout rate was highest between the first and second tea treatments. Black tea treatment was associated with the highest rate of attrition among smokers. Demographic and behavioral factors other than smoking, although suggestive, were not statistically significantly associated with attrition or compliance in our study.
Few previous crossover studies have explicitly reported on participant attrition and compliance in tea trials and none of them have reported a dropout/non-adherence rate as high as seen in this study [7
]. None of those studies had four crossover treatments, and all were significantly shorter in duration compared to our study. In addition, none of the studies had a smoking population as large as this study.
Research on the association of smoking and attrition in clinical trials is primarily based on smoking cessation studies, which have shown that pre-inclusion (during screening or intake evaluations) attrition is generally higher (in the 30-50% range) as compared to post-inclusion (intervention or post-intervention follow-up) attrition which varies from less than 10% to 50% [13
]. These rates are comparable to the rates seen in the present study. Rates of attrition for both smokers and non-smokers in this study were higher for the first treatment as compared to the subsequent crossover treatments.
Black tea, and to a lesser extent, green tea were associated with the highest rates of attrition among smokers but not among non-smokers. Five smokers and three non-smokers in the current trial complained of an unpleasant/bitter taste while on either the green tea or black tea treatment. However, only one of the smokers reported not complying as a result of taste, whereas the others complied with the treatments. Although smoking is common in tea houses in some cultures in Asia and Africa, and there is no research to suggest that smoking alters the taste of tea, it is possible that the polyphenols, theaflavins or other substances in black tea interact with smoking to cause a bitter taste, and might reduce the tolerance of smokers to tea treatments. Efforts to minimize attrition in future CAM studies of tea should focus primarily on smokers.
In addition to smoking, other possible reasons for this relatively high rate of non-adherence could be related to the frequent doses of treatment (5 doses per day) that the participants were required to take as part of the trial, the crossover design, the long duration of the study (16
weeks of treatment), the strong taste of the black and green tea, and reports of teeth staining as mentioned by some participants.
Given the 51% attrition seen in this study, estimates of sample size for similar crossover tea studies would essentially double, and make the crossover design less efficient than the parallel design in our opinion. In addition, lack of a carryover effect in a parallel group design and more practical treatment administration, especially with multiple treatments, are other advantages of the parallel compared to the crossover trial design. Although the washout periods included in the crossover trials of tea are a safeguard against carryover effects, lack of knowledge on the intensity and duration of carryover effects limits the utility of the washout periods. Additionally, in most crossover trials, including the present study, washout periods between treatments are of the same duration thus assuming similar duration of carryover effects for all treatments in the trial which might not be true. For these reasons multi-arm chemoprevention trials of tea might be more practical and achieve better compliance if conducted using the parallel group design rather than the crossover design.
Efforts to make the treatments more palatable, either by providing the active ingredients in a pill or capsule form or by masking the taste of tea, might improve retention rates in tea trials. In addition, more research is needed to determine the dosage required for physiologic changes such that the strength of the tea and frequency of intake might be modified to ensure compliance in such trials. In this study participants were asked to drink five packets of tea every day. Since most participants were not habitual drinkers of tea, drinking five cups is a burden that might have resulted in the high percentage of dropouts seen in this study. In addition, participants were prescribed to drink tea using a “standard method” that asked them to hold the tea in the mouth for 30–60 seconds prior to swallowing. Although this method ensures standardization of dose delivery to oral cells for all participants it increases participant burden. Use of a placebo run-in period prior to randomization is another way to exclude participants who are less likely to be compliant over the period of the study. A short-term placebo run-in might be an effective way to decrease attrition and increase compliance in a parallel study, and also in a crossover study because the results from this study indicate that most of the dropouts occur after between the first and second treatments. However, participant exclusion based on a placebo run-in might limit the generalizability of the study results if participation and compliance are systematically associated with participant characteristics, such as, education, socioeconomic status, smoking, diet, and other behaviors. Crossover trials afford the advantage of having a participant as his/her own control in a different time period but multi-arm trials with assessment of more than two treatments might make the trial lengthy and increase participant burden.
Strengths of this study include conduct of a four-arm crossover trial of tea that has not been previously reported. In addition, data on participant characteristics, compliance, and dropout was accurately measured throughout the trial. However, this study has limitations. Although the four treatments were designed to be as similar to each other as possible in taste and color, it might not have led to 100% success with participant blinding because participants previously exposed to green or black tea could have easily identified the tea treatment. In addition, although some participants reported the cause of non-compliance and dropout from the study, a formal attempt was not made to determine the cause of dropout of all participants who left the study early.