Mixed Methods Analysis of Unprompted Participant Text Messages in an Automated Text Message Program for Smoking
Belinda Borrelli1, Romano Endrighi1, Gavin N Rackoff1
1Boston University, Center for Behavioral Science Research, Henry M. Goldman School of Dental Medicine.
Introduction:
Text messaging programs are effective for smoking cessation. Unprompted, participant-initiated text messages that are unrecognized by rules-based programs (UNRs) provide real-time reactions and feedback regarding program experiences. This mixed methods study characterizes UNRs and examines their prospective relationship with treatment outcome.
Methods:
This is a secondary analysis of a randomized controlled cessation induction trial with adult smokers in a 4-week text message program. UNRs were defined as participant-initiated text messages not matching pre-defined expected responses or not in answer to pre-programmed queries. UNRs were qualitatively coded (if not related to a study payment issue or question) and analyzed for associations with participant and program characteristics as well as outcomes (use of evidence-based treatments [EBTs] for cessation and 7-day point prevalence abstinence from smoking).
Results:
Participants (n=440; mean age = 47.5, 47.4% female, 16.9% motivated to quit) sent 2,264 UNRs (1,785 met criteria for coding) and 59.8% sent > 1 UNR. Sending any UNR or any codable UNR predicted greater odds of EBT use during the 7-month study period (self-reported and objectively verified, all ORs > 1.50). Sending any codable UNR (OR = 1.91 [1.09-3.33]) or a UNR acknowledging a program-initiated text (OR = 2.00 [1.18-3.37]) predicted greater odds of self-reported smoking abstinence during the follow-up. Sending a UNR conveying updates on target behaviors predicted greater odds of biochemically verified abstinence during follow-up (OR = 3.05 [1.24-7.46]).
Conclusions:
UNRs may serve as markers of meaningful engagement and early indicators of subsequent behavioral outcomes in text messaging programs.
