Overall, we found a higher rate of attrition in UC HAT compared to BAC. Intention to treat analyses did not demonstrate improved disease activity, disease-specific quality of life, or adherence in UC HAT compared to BAC. However, at baseline immune suppressant use was twice as high and IBDQ scores were lower in UC HAT compared to BAC. After adjustment for baseline IBDQ scores, trial completers in the UC HAT arm had decreased disease activity scores at 12 months compared to baseline. After adjustment for baseline disease knowledge, UC HAT trial completers experienced significant gains in disease-specific quality of life from baseline compared to BAC trial completers.
There are several possible explanations for our findings. First, our study was powered to detect moderate differences in disease activity, quality of life, and adherence rates. Unfortunately, we were not able to recruit the number of participants needed to detect a moderate effect size in adherence rates between groups. However, we did have sufficient power to detect moderate differences in disease activity and disease-specific quality of life between groups. In addition, we experienced higher than expected attrition rates which further hampered our ability to detect moderate differences for all outcome measures. Furthermore, despite the randomization process, significant differences were present in the groups at baseline. Notably, UC HAT participants had higher use of immune suppressants and had lower baseline IBDQ scores. These results suggest that UC HAT participants had more severe disease at baseline than BAC participants. This may explain why analyses of trial completers adjusted for baseline differences between the groups demonstrated improvement in Seo Index and IBDQ scores from baseline in the UC HAT arm.
Our trial also had higher rates of attrition in the UC HAT arm than expected. There were several reasons for participants withdrawing in the UC HAT arm. Four participants were withdrawn by the research team because they were nonadherent with weekly self-testing. Two participants withdrew because they changed their mind about participating, one patient moved out of the country, and one patient developed severe depression unrelated to the testing and was withdrawn from the study. Participants who continued to use UC HAT experienced significant improvements in quality of life over time. It is not clear if patients will use a system like UC HAT long term. It is plausible that a variation of UC HAT that does not require home installation and that can be accessed via the web may be more successful for increasing recruitment and in retaining participants in the study. A recently published study from Europe reported an attrition rate of only 24% at one year in UC participants that used a web-based telemedicine system; interestingly attrition rates were higher in the telemedicine arm in this study also19
. In addition, decreasing the frequency of self-testing may also improve participant retention and adherence with self-testing. Another reason for lack of improvement in clinical outcomes between groups could be that participants seemed to experience little variability in their disease course during the trial. Seo index and IBDQ scores during the trial were consistent with patients in clinical remission. This is likely a result of our inclusion criteria, as we did not require patients to have active disease at baseline. Therefore, it is possible that if we recruited patients with active disease and/or those with shorter periods of remission at baseline, we might have seen more variability in Seo and IBDQ scores and differences in clinical outcomes. Conversely, the low Seo index and high IBDQ scores throughout the study could also be a reflection of care provided by referral centers for IBD. Thus, it is possible that UC HAT is less effective when combined with specialty IBD care as opposed to when it is applied in community practice. It is also possible that UC HAT would be more effective in a homogenous UC population with less severe disease. For example, a web-based telemedicine system in European patients with UC demonstrated improved outcomes. These patients all had mild to moderate disease and were only on aminosalicylates (5-ASA) at baseline. Including UC patients with mild to moderate disease on 5-ASA only allowed providers in this study to utilize a single, rapid treatment approach (high dose 5-ASA) for most flares19
. Despite these explanations, it is possible that UC HAT does not improve outcomes in patients with UC.
To our knowledge, our trial is the second randomized, controlled trial to assess the impact of a telemedicine system on clinical outcomes in patients with UC. Our study has several strengths. First, the study was a randomized, controlled trial. Second, we concealed allocation to groups, in that researchers were masked to group assignment at the time of study visits. Third, to make our results generalizable, we included a subset of patients from the VA as they tend to have less severe disease and are more representative of a community practice of UC patients. Fourth, we gave action plans and an education curriculum to controls so that we could evaluate the impact of the interaction between participants and UC HAT and the importance of frequent monitoring and prompting. By creating a BAC arm, we could potentially avoid criticism that the success of the system was solely due to use of action plans or the educational curriculum. There were several weaknesses to our study. First, we were underpowered to detect small to moderate differences in outcomes between groups. Second, we included patients in remission for varying intervals in the study. Including patients with long-term remission may have biased our results to the null as these patients experience little variability over time. Third, participants were not masked to their group assignments. We discussed giving all patients UC HAT but only monitoring the “intervention” group. However, it is likely that participants over time would have recognized that they were not being monitored. It is possible that participants in the UC HAT arm responded to disease activity and quality of life indices differently at study visits because of a bias towards a perceived benefit of UC HAT. If this is true, are results would be biased away from the null hypothesis. Our participants had high rates of immune suppressant and infliximab use consistent with a tertiary referral population. Therefore, our findings are probably not generalizable to the entire community with UC.
In summary, we did not demonstrate significant improvements in disease activity, quality of life, or adherence in participants using UC HAT. However, UC HAT trial completers experienced improvements in Seo index and IBDQ score from baseline compared to BAC trial completers after adjustment for baseline quality of life scores and knowledge scores respectively. Our results are suggestive that telemedicine may improve outcomes in patients with UC. Newer telemedicine systems that do not require home installation and that can be accessed from anywhere via the web need to be developed to enhance ease of use, as well as participant recruitment and retention. Larger studies are needed to demonstrate small to moderate differences in disease outcomes in IBD. Further, consideration of length of time in remission and the mix of community and referral patients will need to be addressed for future trials.