Designing and conducting clinical trials is challenging for some institutions and researchers due to associated time and personnel requirements. We conducted recruitment, screening, informed consent, study product distribution, and data collection remotely. Our objective is to describe how to conduct a randomized clinical trial using remote and automated methods.

A randomized clinical trial in healthcare workers is used as a model. A random group of workers were invited to participate in the study through email. Following an automated process, interested individuals scheduled consent/screening interviews. Enrollees received study product by mail and surveys via email. Adherence to study product and safety were monitored with survey data review and via real-time safety alerts to study staff.

A staff of 10 remotely screened 406 subjects and enrolled 299 over a 3-month period. Adherence to study product was 87%, and survey data completeness was 98.5% over 9 months. Participants and study staff scored the System Usability Scale 93.8% and 90%, respectively. The automated and remote methods allowed the study maintenance period to be managed by a small study team of two members, while safety monitoring was conducted by three to four team members. Conception of the trial to study completion was 21 months.

The remote and automated methods produced efficient subject recruitment with excellent study product adherence and data completeness. These methods can improve efficiency without sacrificing safety or quality. We share our XML file for researchers to use as a template for learning purposes or designing their own clinical trials.

To outline the development of a smartphone-based tool to collect thrice-repeated 24 h dietary recall data in rural Nepal, and to describe energy intakes, common errors and researchers’ experiences using the tool.

We designed a novel tool to collect multi-pass 24 h dietary recalls in rural Nepal by combining the use of a CommCare questionnaire on smartphones, a paper form, a QR (quick response)-coded list of foods and a photographic atlas of portion sizes. Twenty interviewers collected dietary data on three non-consecutive days per respondent, with three respondents per household. Intakes were converted into nutrients using databases on nutritional composition of foods, recipes and portion sizes.

Dhanusha and Mahottari districts, Nepal.

Pregnant women, their mothers-in-law and male household heads. Energy intakes assessed in 150 households; data corrections and our experiences reported from 805 households and 6765 individual recalls.

Dietary intake estimates gave plausible values, with male household heads appearing to have higher energy intakes (median (25th–75th centile): 12 079 (9293–14 108) kJ/d) than female members (8979 (7234–11 042) kJ/d for pregnant women). Manual editing of data was required when interviewers mistook portions for food codes and for coding items not on the food list. Smartphones enabled quick monitoring of data and interviewer performance, but we initially faced technical challenges with CommCare forms crashing.

With sufficient time dedicated to development and pre-testing, this novel smartphone-based tool provides a useful method to collect data. Future work is needed to further validate this tool and adapt it for other contexts.