JMIR Publications

ABSTRACT

Background:

Systematic and paper-based perinatal mental health screening may overburden health care systems lacking workforce and financial resources. Short Message Service (SMS) text message screening may provide a novel alternative, increasing the efficacy and accessibility of screening for patients and healthcare providers. Before bringing texting-based screening protocols to scale, it is critical to understand the factors that may limit their feasibility and appropriateness within diverse populations.

Objective:

Our study aimed to 1) examine sources of selection bias in patient engagement and compliance in a texting-based perinatal depression and anxiety screening protocol, and 2) determine whether participant SMS text message response rate was a better predictor of postpartum mental health symptoms than scores from the texting-based brief screening measures themselves.

Methods:

Perinatal participants from the Montreal Antenatal Well-Being Study (MAWS, n=1130) completed brief screening questionnaires assessing depression (Whooley Questions) and anxiety symptoms (Generalized Anxiety Disorder 2-Item questionnaire: GAD-2) at baseline and then at 14-day intervals via SMS text message. T-tests and Fisher’s tests were used to determine the sociodemographic and mental health profile of participants who responded to the SMS text message-based questions. Hurdle regression analyses were used to determine if baseline depression and anxiety symptoms were associated with number of SMS text message responses. Measures of model fit were used to determine the added predictive value of participants’ texting response rate on their depression and anxiety symptoms in the postpartum period (assessed by the Edinburgh Postnatal Depression Scale [EPDS] and the State-Trait Anxiety Inventory [STAI-S]).

Results:

Participants who responded to the SMS text messages (n=933) were more likely than non-respondents (n=114) to be Caucasian (n=587, 64.72% vs. n=39, 40.63%; P<.001), have higher educational attainment (post-graduate: n=268, 29.48% vs. n=15, 15.96%; P=.005), and higher income levels ($150,000 or more: n=176, 21.15% vs. n=10, 11.91%; P<.001). There were no significant differences in subclinical symptoms of depression and anxiety between the two groups at baseline and postpartum. However, worse depression and anxiety symptoms at baseline (a one-point increase in Whooley, EPDS and STAI-S scores) were associated with fewer SMS responses overall (respective decrease of 3.1%, 1%, and 0.3% in the average number of SMS answered by participants). Scores on the GAD-2 questionnaire sent via SMS text message in the first eight weeks postpartum were the best predictor of depression and anxiety symptoms assessed between eight weeks and six months postpartum; participant SMS response rate did not impact this association.

Conclusions:

Findings from this study cautiously support the use and feasibility of brief screening questionnaires sent via SMS text message to screen for perinatal depression and anxiety symptoms. However, findings also highlight how screening and service delivery via digital technology could exacerbate disparities in mental health between certain patient groups.