Comparison of a Daily Smartphone App and Retrospective Questionnaire Measures of Adherence to Nicotine Replacement Therapy within Pregnant Women (Preprint)

BACKGROUND Few studies have investigated how to best measure adherence to smoking cessation medications, but continuous usage measures are recommended. OBJECTIVE In this first study of its kind, we compared methods for measuring adherence to nicotine replacement therapy (NRT), investigating the completeness and validity of data collected from daily assessments using a smartphone app versus data collected from retrospective questionnaires. METHODS Women aged ≥16, who were daily smokers <25 weeks pregnant, were offered cessation counselling and encouraged to use NRT. Women set quit dates (QD) and, for 28 days afterwards, were asked to report NRT use daily to a smartphone app and to questionnaires administered in-person or remotely at 7 and 28 days. For both data collection methods, we provided up to £25 in compensation for time taken providing research data. Data completeness and NRT use reported to app and questionnaires were compared. For each method, we also correlated mean daily nicotine doses reported within 7 days of QD with Day 7 saliva cotinine concentrations. RESULTS Of 438 women assessed for eligibility, 40 participated and 35 accepted NRT. More participants (31/35) submitted NRT usage data to the app by Day 28 (median days submitted=25 [IQR 11]) than completed the Day 28 questionnaire (24/35). Data submitted to the app showed a lower reported duration of NRT use compared to the questionnaire (median days NRT for app=24 [IQR 10.25]; questionnaire=28 [4.75], P=.007), and there appeared to be specific cases of overreporting to the questionnaire. Mean daily nicotine doses (mg) between QD and Day 7 were lower when calculated using app data (median mg for app=40 [52.1]; questionnaire=40 [63.1], P=.001), and some large outliers were evident for the questionnaire. Mean daily nicotine doses, adjusted for cigarettes smoked, were not associated with cotinine concentrations for either method (app r=0.184, P=.55; questionnaire r=0.031, P=.92). CONCLUSIONS Daily assessment of NRT use via a smartphone app facilitated more complete data (a higher response rate) than questionnaires, and reporting rates over 28 days were encouraging among pregnant women. Although data from neither method was significantly associated with cotinine concentrations, app data had better face validity; retrospective questionnaires appeared to overestimate NRT use for some.