A comparison of smartphone and paper data-collection tools in the Burden of Obstructive Lung Disease (BOLD) study in Gezira state, Sudan

• Published in: PloS one 2018-03, Vol.13 (3), p.e0193917-e0193917
• Main Authors: Ahmed, Rana, Robinson, Ryan, Elsony, Asma, Thomson, Rachael, Squire, S Bertel, Malmborg, Rasmus, Burney, Peter, Mortimer, Kevin
• Format: Article
• Language: English
• Subjects: Accuracy; Biology and Life Sciences; Cellular telephones; Chronic obstructive lung disease; Comparative analysis; Cough; Data collection; Data collection devices; Data integrity; Engineering and Technology; Global positioning systems; GPS; Handbooks; Health aspects; Human error; Laboratories; Lung diseases; Medicine; Medicine and Health Sciences; Methods; Obstructive lung disease; Open data; People and Places; Polls & surveys; Population studies; Questionnaires; Research and Analysis Methods; Researchers; Rural areas; Rural environments; Smart phones; Smartphones; Smoking; Social Sciences; Statistical tests; Wheezing
• ISSN: 1932-6203; 1932-6203
• DOI: 10.1371/journal.pone.0193917

Data collection using paper-based questionnaires can be time consuming and return errors affect data accuracy, completeness, and information quality in health surveys. We compared smartphone and paper-based data collection systems in the Burden of Obstructive Lung Disease (BOLD) study in rural Sudan. This exploratory pilot study was designed to run in parallel with the cross-sectional household survey. The Open Data Kit was used to programme questionnaires in Arabic into smartphones. We included 100 study participants (83% women; median age = 41.5 ± 16.4 years) from the BOLD study from 3 rural villages in East-Gezira and Kamleen localities of Gezira state, Sudan. Questionnaire data were collected using smartphone and paper-based technologies simultaneously. We used Kappa statistics and inter-rater class coefficient to test agreement between the two methods. Symptoms reported included cough (24%), phlegm (15%), wheezing (17%), and shortness of breath (18%). One in five were or had been cigarette smokers. The two data collection methods varied between perfect to slight agreement across the 204 variables evaluated (Kappa varied between 1.00 and 0.02 and inter-rater coefficient between 1.00 and -0.12). Errors were most commonly seen with paper questionnaires (83% of errors seen) vs smartphones (17% of errors seen) administered questionnaires with questions with complex skip-patterns being a major source of errors in paper questionnaires. Automated checks and validations in smartphone-administered questionnaires avoided skip-pattern related errors. Incomplete and inconsistent records were more likely seen on paper questionnaires. Compared to paper-based data collection, smartphone technology worked well for data collection in the study, which was conducted in a challenging rural environment in Sudan. This approach provided timely, quality data with fewer errors and inconsistencies compared to paper-based data collection. We recommend this method for future BOLD studies and other population-based studies in similar settings.