Pilot study of psychotherapeutic text messaging for depression

Abstract

Background

Text messaging services could increase access to psychotherapeutic content for individuals with depression by avoiding barriers to in-person psychotherapy such as cost, transportation, and therapist availability. Determining whether text messages reflecting different psychotherapeutic techniques exhibit differences in acceptability or effectiveness may help guide service development.

Objectives

We aimed to determine: (1) the feasibility of delivering a psychotherapy-based text messaging service to people with depression identified via the internet, (2) whether there is variation in satisfaction with messages according to the type of psychotherapeutic technique they represent, and (3) whether symptoms of depression vary according to receipt of each message type and participants’ satisfaction with the messages they received.

Methods

For this study 190 US adults who screened positive for a major depressive episode (Patient Health Questionnaire (PHQ-9) score ≥10) were recruited from online advertisements. Participants received a daily psychotherapy-based text message 6 days per week for 12 weeks. Text messages were developed by a team of psychiatrists, psychologists, and social workers to reflect three psychotherapeutic approaches: acceptance and commitment therapy (ACT), behavioural activation, and cognitive restructuring. Each week the message type for the week was randomly assigned from one of the three types, allowing for repeats. Participants were asked daily to rate each message. On the 7th day of each week, participants completed a two-item depression screener (PHQ-2). Web-based surveys at baseline, 6, and 12 weeks were used as the primary measure of depressive symptoms (PHQ-9).

Results

Of the 190 participants enrolled, 85 (45%) completed the 6-week web survey and 67 (35%) completed the 12-week survey. The mean baseline PHQ-9 score was 19.4 (SD 4.2) and there was a statistically significant mean improvement in PHQ-9 scores of −2.9 (SD 6.0; p < 0.001) at 6 weeks and −4.8 (SD 6.6; p < 0.001) at 12 weeks. Mean satisfaction ratings did not vary across message types. There was generally no association between the number of weeks in which participants received a given message type and PHQ-9 scores at follow-up. However, among individuals with more severe depression at baseline (PHQ-9 ≥ 20; n = 30), the number of weeks of behavioural activation messages received was associated with less symptom improvement (ρ = 0.37; p = 0.04) at 12 weeks. Participants who gave higher satisfaction ratings to messages regardless of type had greater reductions in PHQ-9 scores at 6 weeks.

Conclusions

Study participants exhibited borderline clinically significant improvement in depressive symptoms at 12 weeks suggesting the effects of the intervention, if any, were small. Although there was no overall variation in changes in PHQ-9 according to users’ exposure to the three message types, effectiveness among severely depressed participants could potentially be improved by tailoring towards fewer behavioural activation messages. Controlled studies to determine effectiveness of texting interventions such as this one are indicated considering that even small effects may be cost effective given the low cost of delivering text messages.

Keywords

Mobile technology psychotherapy SMS depressive internet

Introduction

Psychotherapy is effective for treating depression, but less than a third of individuals with a depressive disorder in the US receive any psychotherapy, and only half of those who receive any psychotherapy receive an adequate course.^1,2 Antidepressant medication use has increased substantially since the 1990s, while the percentage of antidepressant medication users receiving psychotherapy has declined to less than 20%.³ Although individuals with depression often state a preference for psychotherapy over medication treatment, depressed patients encounter a number of barriers to psychotherapy including cost, availability, stigma, and physical or health restrictions to attending visits.⁴ New methods for delivering psychotherapy content that circumvent these barriers – such as through mobile health technologies – are therefore indicated to meet patients’ preferences for alternatives to medication and to augment antidepressant pharmacotherapy.

Automated mobile health technologies could provide the benefits of psychotherapy without the access barriers associated with meeting with a therapist at specific locations and appointment times. An estimated 91% of US adults own a cellular phone and therefore could access psychotherapy content via a mobile device.⁵ Automated methods for delivering psychotherapy may afford recipients greater privacy and control compared with meeting with a therapist.⁶ Mobile health technologies might also allow greater opportunities for psychotherapy skills or concepts to be practised in vivo as these technologies are integrated into people’s daily lives. Despite the potential advantages, the evidence to date for effective automated mobile health-delivered psychotherapy for depression is limited.

Computerized cognitive behavioural therapy (c-CBT), typically delivered via desktop computers but capable of being delivered through mobile devices, has been shown to be effective in clinical trials and was adopted by the United Kingdom’s national health system.^7,8 However, a recent independent study of two c-CBT programmes demonstrated a lack of effectiveness with poor completion rates of the c-CBT programmes despite weekly telephone support.⁹ Qualitative analyses revealed several factors limiting programme completion, including too much flexibility with scheduling time to work through the programme and a lack of accountability in completing modules, and perceptions that engaging with a computer was impersonal or felt too much like “work”.¹⁰ Considering that people increasingly use text messages for a wide range of personal communication needs, texting may be perceived as a more satisfying and engaging method for receiving therapeutic content. Because text messaging is a “push” technology that can deliver psychotherapeutic content proactively, it may circumvent some of the barriers to c-CBT by not depending on individuals to schedule their own time to complete modules and by delivering content in smaller but more frequent increments.

Prior research supports the notion that text messaging may be an effective means for delivering psychotherapeutic content. In one study of individuals with depression and alcohol use disorders, the delivery of twice-daily supportive text messages was associated with improvements in depression after 3 months, but the effects were not sustained 3 months after the intervention was discontinued.^11,12 The current study sought to extend this prior work with the long-term goal of increasing the effectiveness of text messaging for depression by personalizing the types of text messages individuals receive according to the psychotherapeutic technique most likely to be helpful to them. As an initial step towards this goal, a team of psychiatrists, psychologists, and social workers constructed text messages of three different types based on three psychotherapeutic techniques: acceptance and commitment therapy (ACT), behavioural activation, and cognitive restructuring.^13,14 ACT emphasizes mindfulness and acceptance of distressing thoughts while taking actions that are consistent with one’s values. Cognitive restructuring focuses on identifying and correcting maladaptive thought patterns, and behavioural activation encourages individuals to engage in more meaningful or pleasurable activities. This study tested whether individuals with depression displayed differences in satisfaction, psychotherapy-related self-management skills, or depression symptoms based on their experiences with each of the three message types. A secondary goal of the study was to test the feasibility of delivering the intervention and all related study activities (e.g. enrolment, measurement collection) via the internet prior to larger-scale trials or service implementation.

Methods

Study design

This was a single-arm trial of a 12-week therapeutic text messaging intervention for individuals seeking information about their depressive symptoms via the internet.

Participants

Inclusion criteria were: (1) age ≥18 years, (2) Patient Health Questionnaire (PHQ-9) score ≥10, (3) a valid email address, and (4) access to a cellular phone capable of receiving text messages. Based on the methods described by Morgan et al., participants were recruited via online advertisements that were shown to individuals residing in the United States who conducted a Google internet search using terms related to depression.¹⁵ The advertisement offered a “depression test” or “quiz”, thus individuals were not made aware of the research component or the monetary incentive until after clicking on the advertisement. The advertisement was shown to 91,310 individuals between April and June of 2015, of whom 9922 clicked on the advertisement. Upon clicking on the advertisement, individuals were directed to a study website that obtained a research consent for screening and then administered the PHQ-9. Of these individuals, 2027 had a PHQ-9 ≥10 and were provided the opportunity to participate in the trial, and 190 people completed all enrolment and consent procedures and initiated the intervention. All individuals who scored PHQ-9 ≥10 were provided with a link to the Substance Abuse and Mental Health Services Administration’s treatment services locator, and those who indicated any suicidal ideation were provided the national crisis line website and phone number. The study was conducted in accordance with local institutional review board approval.

Procedures

Participants received a therapeutic text message 6 days out of the week at 8am Pacific Standard Time. Participants received a second “rating” text message at 3pm Pacific Standard Time which asked them to rate that day’s therapeutic message. On every 7th day, participants were asked to complete the PHQ-2, a 2-item screening measure for depression,¹⁶ by responding to two sequential text messages.

The therapeutic text messages were developed by mental health clinicians and researchers including psychiatrists, psychologists, and social workers. These clinicians were directed individually and in small groups to develop messages that were aligned with cognitive restructuring, behavioural activation, or ACT. After an initial pool of messages was created, the messages were rated from 1 to 5 (5 being the best) by the mental health clinicians on the extent to which they reflected the expected therapeutic approach, and messages with a mean rating below 3.0 were discarded. The therapeutic messages were not interactive nor tailored to the recipient’s characteristics. Examples of text messages for each technique are in the Online Appendix.

For each week of the 12-week text messaging service, the message type for the week was determined randomly for each participant, allowing for repeats. So, for example, a participant could have been randomly assigned to receive behavioural activation messages for week 1, then randomly assigned to receive behavioural activation messages again for week 2, and then randomly assigned to receive cognitive restructuring messages for week 3 and so on. This randomization scheme allowed us to analyse the effects of varying amounts of each message type without pre-specifying the sequences. For each day a participant was scheduled to receive a therapeutic text message, one message was randomly selected from the pool messages of the weekly assigned message type and delivered to the participant. Individual messages could also be repeated.

Requests for satisfaction ratings were delivered daily, separately from the therapeutic message. These messages addressed participants by their preferred name and included a brief uplifting comment, such as “hope your day went well”, and then asked, “How would you rate today’s morning message from 0 (worst) to 5 (best)?”.

Participants could text “stop” at any time to prevent any additional text messages from being sent. Participants who texted “help” were provided the suicide crisis line phone number and were contacted by the study staff by phone within the next business day. Instructions for texting “stop” and “help” were provided at study enrolment, and reminders regarding these instructions were provided when participants texted unrequested messages to the study team. There were no incidents of participants who requested help due to a suicidal crisis.

Measurement

The amount of exposure to randomly assigned message types was measured in weeks. Individual message ratings and PHQ-2 scores were collected via text messages. Additional measures described below were collected using internet-based surveys by emailing the survey link to participants. Participants were provided a $15 online gift card for an internet retailer for completing those surveys at baseline, 6 weeks, and 12 weeks.

Demographic characteristics assessed at baseline included age, gender, race/ethnicity, marital status, education, and employment status.

Depression history and treatment history assessed at baseline included the number of lifetime depressive episodes; duration of the current episode; prior treatment with psychotherapy; counselling from someone other than a licensed therapist; medication, herbal or dietary supplement use; attendance at in-person support groups and online support groups; receipt of self-help materials (printed or online); exercise; inpatient treatment for depression; or other forms of depression treatment. Barriers to psychotherapy were assessed at baseline among participants not currently receiving psychotherapy using 12 items developed by Mohr et al.⁴

Depression symptoms were assessed at baseline, 6 weeks, and 12 weeks using the PHQ-9. The PHQ-9 is a nine-item instrument that assesses the symptoms of depression corresponding to the Diagnostic and Statistical Manual Version IV (DSM-IV) diagnostic criteria for a major depressive episode.¹⁷ Each symptom is rated on a four-point scale ranging from 0 to 3, and item scores are summed resulting in a score range from 0 to 27. A score of 10 or greater is 88% sensitive and 88% specific for a diagnosis of major depression compared with a gold-standard diagnostic interview by a mental health professional.^17,18 The PHQ-2 which includes two of the PHQ-9 items measuring depressed mood and anhedonia is a validated screening measure of depression and was used to assess weekly depression status via text messages. A PHQ-2 score of 3 or greater has a sensitivity of 83% and specificity of 92% for major depression.¹⁶

Depression self-management skills typically taught in psychotherapy were assessed at baseline, 6 weeks, and 12 weeks using the Cognitive Behavioural Therapy Skills Questionnaire (CBTSQ) and a shortened version of the Five Facet Mindfulness (FFM-S) scale. The CBTSQ assesses behavioural activation (seven items) and cognitive restructuring (nine items) skills on a 5-point scale and has been shown to be sensitive to change during a course of cognitive behavioural therapy.¹⁹ The FFM-S was constructed in order to shorten the 39-item Five Facet Mindfulness scale to 15 items, with each item scored on a 5-point scale. Items were selected by the investigators based on high factor loadings according to work by Baer et al. and with the removal of items perceived to be overly similar in wording to other included items, while maintaining three items per each of the five facets.²⁰ The 15-item measure in this study had a Cronbach’s alpha of 0.66.

Satisfaction with the overall text messaging service was assessed at 12 weeks using four items scored on a 5-point scale. Open-ended semi-structured items were also administered at 12 weeks and used to assess what participants liked best and least about the service.

Analyses

Means and frequencies were used to describe the baseline characteristics of the study sample, completion rates for the 6- and 12-week assessments, and participant satisfaction. Completion bias was assessed by comparing the baseline characteristics of participants who completed versus did not complete the 6- and 12-week measures using chi-square and Wilcoxon rank sum tests. The main quantitative analyses were separated into overall outcomes and outcomes related to message type. Overall outcomes included the change in depression symptoms (PHQ-9) and psychotherapy-related skills between baseline, 6 weeks, and 12 weeks of the intervention, assessed using ANOVA. The relationship between mean message ratings (across message types) and change in PHQ-9 score was assessed using Spearman’s correlation.

Differences in outcomes by message type were assessed using Spearman’s correlation to determine whether change in depression symptoms was associated with the number of weeks received of each of the three message types (random for each participant). In secondary analyses, this same relationship was examined for the subset of participants with severe (PHQ-9 ≥20) or less than severe depression (PHQ-9 <20) at baseline. To examine differences in ratings by message type, we used a linear regression model with message rating as the outcome and message type as the predictor, accounting for clustering by participant and day of message since start of study. Changes across individual weeks were analysed using linear regression with week-to-week change in PHQ-2 score as the outcome and message type for the week and mean message rating for the week as predictors, accounting for clustering by participant and day of message since start of study.

At 12 weeks, participants responded to open-ended questions regarding their experience with the intervention. These responses were analysed using a cutting and sorting method in which each response was placed on a single strip of paper.²¹ The strips were then sorted into piles representing similar responses. Themes were generated to characterize these piles, and the themes with the most responses are reported here.

Results

Sample characteristics and participation

In total, 190 people with PHQ-9 scores of 10+ enrolled in the study and completed baseline surveys. The study population was 84% female, 68% White, 13% Black, 11% Hispanic, 9% other race or ethnicity, and 88% were between the ages of 18 and 54 years old. One in four (25%) participants were married, 47% were employed, and 66% had some education beyond high school (Table 1). In terms of participants’ depression histories, 72% reported three or more prior depressive episodes, and 52% reported their current episode lasting more than 2 years. The most common treatment for depression prior to study enrolment was medications (44%), followed by psychotherapy (40%) and written or online self-help materials (34%). Only 24% of participants reported no prior depression treatment. The most common barriers to in-person psychotherapy endorsed by participants who were not currently receiving psychotherapy were lack of energy/motivation (52%), cost (43%), daily responsibilities and activities (33%), and the expectation that psychotherapy would not be helpful (30%) (Table 2). The mean baseline PHQ-9 score was 19.4 (SD 4.2), which represents moderately severe to severe depression.

Table 1.

Participant demographic characteristics (N = 190).

Variable	N (%)
Age
18–24	28 (15)
25–34	46 (24)
35–44	50 (26)
45–54	44 (23)
55–64	15 (8)
65+	7 (4)
Female Gender	160 (84)
Race/ethnicity
White	129 (68)
Black	24 (13)
Hispanic	20 (11)
Other	17 (9)
Marital Status
Married	47 (25)
Divorced	47 (25)
Widowed	3 (2)
Never Married	66 (35)
Living together, not married	27 (14)
Education Attainment
Less than high school graduate	17 (9)
High school graduate	48 (25)
Some college	69 (36)
College graduate	42 (22)
Advanced degree	14 (7)
Currently Employed	89 (47)

Table 2.

Participant depression histories and barriers to in-person psychotherapy.

Variable	N (%)
Number of prior depressive episodes
0	11 (6)
1	10 (5)
2	32 (17)
3 or more	137 (72)
Duration of current episode
Less than 1 month	6 (3)
1 to 6 months	37 (19)
6 months to 1 year	23 (12)
1 to 2 years	26 (14)
More than 2 years	98 (52)
Prior treatment
Psychotherapy	76 (40)
Medication	84 (44)
Self-help using book or computer	65 (34)
Online support from others	13 (7)
Barriers to in-person psychotherapy*
Transportation	30 (18)
Daily responsibilities and activities	56 (33)
Cost	73 (43)
Availability in area	27 (16)
Don’t expect to be helpful	51 (30)
Lack of energy/motivation	87 (52)
Heard about bad experiences	8 (5)
Physical limitations	7 (4)
Problems not severe enough	25 (15)
Therapy means I can’t solve own problems	23 (14)
Other people would know	22 (13)
Talking about problems makes it worse	30 (18)

Among those not currently receiving psychotherapy.

The completion rate of the 6-week outcome survey of depressive symptoms was 45% and completion at 12 weeks was 35%. There were no statistically significant differences in baseline depression symptoms, demographic characteristics, or depression histories between completers and non-completers except that completers of the 6-week assessment were more likely to have previously received online support (11% vs. 4%; p = 0.04), and completers of the 12-week assessment were more likely to be age 45–64 (45% vs. 21%; p = 0.002). At least one message satisfaction rating was completed by 81% of participants, and among these participants the mean number of message ratings was 33 (SD 23). At least one valid PHQ-2 was completed via text message by 64% of participants, and among these participants the mean number of valid PHQ-2s completed was 5 (SD 3).

Overall outcomes

The mean change in PHQ-9 score from baseline to 6 weeks was −2.9 (SD 6.0) and from baseline to 12 weeks was −4.8 (SD 6.6). The improvement in PHQ-9 scores across time points was statistically significant (p < 0.001) (Table 3); however, only the change at 12 weeks constituted a clinically significant improvement of 5 points or more (with rounding).²² The mean improvement in behavioural activation skills was 1.6 points (SD 4.9) at 6 weeks and 2.5 (SD 4.7) at 12 weeks. The mean improvement in cognitive restructuring skills was 1.0 (SD 6.1) at 6 weeks and 2.0 (SD 5.8) at 12 weeks, and the mean improvement in mindfulness skills was 0.3 (SD 6.6) at 6 weeks and 2.9 (SD 7.2) at 12 weeks. Differences across time were statistically significant for each skill (p < 0.01).

Table 3.

Participant depression symptom and psychotherapy skills outcomes.

Outcome	Baseline	6 weeks	12 weeks	p-value*
Outcome	N = 190	N = 85	N = 67	p-value*
PHQ-9, M (SD)	19.4 (4.2)	15.8 (6.7)	13.7(6.6)	<0.001
Mindfulness skills¹, M (SD)	41.0 (7.7)	41.3 (8.0)	44.7 (8.9)	<0.001
Behavioural skills², M (SD)	15.6 (5.5)	17.7 (4.5)	19.1 (5.6)	<0.001
Cognitive skills², M (SD)	22.5 (7.2)	23.9 (6.3)	25.4 (7.4)	<0.01

ANOVA comparing outcomes across time points

PHQ-9: Patient Health Questionnaire

From the shortened Five Factor Mindfulness scale

From the Cognitive-Behavioural Therapy Skills questionnaire

Participants who tended to rate messages higher had greater reductions in PHQ-9 scores at 6 weeks with a −0.31 correlation (p = 0.007), whereas there was not a statistically significant correlation at 12 weeks. In analyses of weekly PHQ-2 scores, higher average message ratings for the week were associated with greater reductions in PHQ-2 scores (β = −0.13, SE = 0.06, p = 0.04). In post-hoc analyses, we examined whether participants with greater improvement in depression rated the same messages higher compared with those with less improvement or whether they received more of the more highly rated messages. We divided the sample at the median for PHQ-9 depression symptom improvement at 6 weeks and identified the top 25% of messages based on overall mean ratings. We found participants in the top half of depression symptom improvement gave a mean rating of 4.0 vs. 3.9 (p = 0.13) to the top messages and received a mean 6.2 vs. 5.7 (p = 0.64) of the top messages compared with participants in the bottom half of symptom improvement.

Outcomes related to message type

The mean number of weeks in which participants received ACT-based messages was 1.7 at 6 weeks (SD 1.2) and 3.3 at 12 weeks (SD 1.9). For behavioural activation messages, participants received an average of 1.8 weeks (SD 1.2) after 6 weeks of follow-up and 3.5 weeks (SD 2.0) after 12 weeks of follow-up. Participants received 1.9 weeks (SD 1.3) and 3.5 (SD 2.1) weeks of cognitive restructuring messages at 6 and 12 week follow-ups, respectively. Each message type had nearly identical mean ratings of 3.5 (out of 5) with a standard deviation of 1.3, and message type did not predict message rating in the regression model. There was no statistically significant correlation between the number of weeks received of any individual message type and change in PHQ-9 within the total sample (Table 4). In subgroup analyses, weeks of behavioural activation messages received was associated with less improvement in depression at 12 weeks (ρ = 0.37; p = 0.04) among those with severe depression but not among those with less than severe depression (ρ = −0.03; p = 0.87). Weeks of ACT-based messages had a correlation suggesting greater improvement in depression symptoms at 12 weeks among those with more severe depression (ρ = −0.34); however, this finding was just above the threshold for statistical significance (p = 0.06). Weeks of cognitive restructuring messages received was not associated with changes in depression among either subgroup. The only statistically significant correlation between the number of weeks of a given message type and skills change was a −0.25 correlation between the weeks of ACT-based messages and change in behavioural activation at 6 weeks. Weekly message type did not predict week-to-week changes in PHQ-2.

Table 4.

Weeks received of each message type and correlations with change in PHQ-9 score and change in psychotherapy skills.

		Δ PHQ-9 Score Correlation, ρ			Δ Psychotherapy Skills Correlation, ρ
Message type	Mean (SD) Weeks Received	Total Sample	Baseline PHQ-9 <20	Baseline PHQ-9 ≥20	Mindfulness skills	Behavioural skills	Cognitive skills
6-week outcomes
ACT-based	1.7 (1.2)	0.09	0.19	0.01	−0.05	−0.25*	−0.17
Behavioural activation	1.8 (1.2)	0.01	−0.09	0.12	0.10	0.13	−0.04
Cognitive restructuring	1.9 (1.3)	−0.18	−0.16	−0.21	−0.05	0.13	0.20
12-week outcomes
ACT-based	3.3 (1.9)	−0.08	0.21	−0.34	0.10	0.05	−0.11
Behavioural activation	3.5 (2.0)	0.16	-0.03	0.37*	-0.03	0.09	0.10
Cognitive restructuring	3.5 (2.1)	−0.05	−0.08	−0.03	−0.03	−0.14	0.09

p < 0.05

Overall programme satisfaction

Some 45% of participants strongly agreed and 30% somewhat agreed that they were satisfied with the programme; 22% of participants strongly agreed and 48% somewhat agreed that the service was helpful for improving their overall mood, 43% strongly agreed and 31% somewhat agreed that they would recommend the service to others, and 3% strongly agreed and 6% somewhat agreed that the programme was a hassle. What participants liked most about the programme based on responses to open-ended items at follow-up included the fact that the programme provided new ideas or positive thoughts for the day, a daily reminder to take care of themselves, the sense that they are not alone and that someone cared, and that they had something to look forward to each day. The most common complaints about the service were that it only lasted 12 weeks, that the messages came at inconvenient times, and that there was no interactivity except for rating the messages. Complaints about the content of the messages were that messages were repeated and some suggestions in the messages were not helpful, relevant, or realistic for people with depression. The most common suggestions for changing the content of the messages were that messages be more positive or uplifting, more motivating to take action, and that they better represent how a truly depressed person thinks by not “sugar coating” the illness or trying to “fix” a chronic condition.

Discussion

This pilot study demonstrated the feasibility and acceptability of delivering psychotherapy-based text messages to individuals with depression identified via the internet. Study participants were able to access and receive evaluative surveys and psychotherapeutic messages completely via internet and mobile health technologies, indicating that such a service could be provided to those who have difficulty accessing in-person mental health services. We note the study sample was ethnically diverse and similar to the general US population.

Although the decrease in depression symptoms after receiving the therapeutic text messages for 12 weeks was of marginal clinical significance and could have been due to the natural progression of the illness or other factors unrelated to receipt of the text messages, we note that the study population had moderately severe to severe depression that was frequently recurrent or chronic. In this population, a mean decrease of 5 points on the PHQ-9 could indicate a meaningful benefit, particularly in light of the low-cost, high-reach potential of a text messaging service. Controlled trials are therefore indicated to determine the effectiveness of the service.

The decrease in symptoms in this study was proportionately less than reported in the Agyapong study of individuals with depression and alcohol use disorders.^11,12 The difference in effect size across the two studies could reflect differences in the study population (e.g. it may be that participants in the current study had more chronic depression), different measures of depression (PHQ-9 vs. the Beck Depression Inventory–II), or differences between the messages provided throughout the intervention. One key difference between the studies was that in the Agyapong study text messages reinforced lessons participants had learned during a dual diagnosis treatment program, whereas in the current study there was no common prior treatment. Other investigators have studied the use of text messaging as an augmentation to in-person treatment;²³ however, the goal of the current research was to develop and pilot test a service that could be provided to individuals with difficulty accessing in-person treatments. Such services may have more limited effect sizes. Future versions of “stand-alone” text messaging services might consider addressing barriers and facilitating linkages to treatment, particularly for individuals with severe or chronic symptoms.

This study found that participants did not distinguish in their daily satisfaction ratings between messages based on ACT, behavioural activation, or cognitive restructuring. In fact, the means and variability of the ratings were nearly identical. Weeks received of each message type also had no association with depression improvement in the total sample. Weeks of ACT-based messages received was modestly associated with less change in behavioural activation at 6 weeks; and at 12 weeks, more weeks of behavioural activation messages was associated with less improvement in PHQ-9 scores among individuals with severe depression. The finding of no difference between message types in the total sample is consistent with a study by Ly et al. that found no statistically significant difference in depression improvement between smartphone applications based on behavioural activation vs. mindfulness.²⁴ However, our subgroup analyses were in contrast to the Ly study, which found that individuals with more severe depression (PHQ-9 ≥10) responded better to the behavioural activation smartphone application whereas those with less severe depression (PHQ-9 <10) responded better to the mindfulness application. The difference in findings could be related to differences in symptom severity between the two study populations such that the less severely depressed participants in the current study (PHQ-9 <20) were perhaps similar to the more severely depressed participants in the smartphone application study (PHQ-9 ≥10). There were also key differences in the interventions. In addition to differences in the means of content delivery (text messages vs. smartphone application), the text messages related to mindfulness also included other elements of ACT psychotherapy beyond mindfulness, such as identifying values, and perhaps this additional content was more helpful to individuals with severe depression than behavioural activation messages.

Given the overall modest differences in effectiveness between message types based on different psychotherapeutic techniques, other avenues for improving message effectiveness, such as the message tone (e.g. uplifting) or tailoring to an individual’s characteristics (e.g. age, gender, physical disability, degree of social support) may be more important. Tailoring of text messages for behavioural change has generally been shown to improve engagement, though the most important tailoring elements for depressed individuals have yet to be established.²⁵ Repetition of messages was clearly disliked and should be avoided.

This study was limited in that only 35% of participants completed the final outcome assessment. However, we note that there were very few demographic or depression history differences between completers and non-completers. Although conducting the study without face-to-face or live telephone communication provided certain advantages in terms of efficiency, the lack of a personal connection with the study staff likely limited the completion of follow-up measures. The financial incentive of $15 may also have been insufficient. The study is also limited in the use of the PHQ-9 to determine depression status to the extent that the findings may not be generalizable to the specific disorder of major depression because participants may have had elevated depression symptoms for other reasons that were not assessed, such as bipolar depression.

In conclusion, this study found completers of a 12-week daily text messaging intervention for depression had clinically significant improvement in depression symptoms and were generally satisfied with the service. Overall, the psychotherapeutic technique on which messages were based was not associated with participants’ satisfaction with the messages or change in depression symptoms. However, participants with severe depression reported less benefit when they received a greater proportion of behavioural activation messages. Text messaging has the potential to reach patients with severe and chronic depression who might not otherwise receive mental health services. Further study is necessary to establish the effectiveness of text messaging delivered in the absence of other services.

Footnotes

Acknowledgements

The authors would like to acknowledge Heather Walters for her assistance in preparing the manuscript, and consultants Susan Murphy, Daniel Almirall, and Hyungjin Myra Kim.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by the Department of Veterans Affairs, Health Services Research and Development Service (CDA 10-036-1), the University of Michigan Department of Psychiatry, and the National Institute of Mental Health (R01 MH096699). John Piette is a Senior VA Research Career Scientist.

References

Young

Klap

Sherbourne

. The quality of care for depressive and anxiety disorders in the United States. Arch Gen Psychiatry 2001; 58: 55–61.

Cuijpers

Smit

Bohlmeijer

. Efficacy of cognitive-behavioural therapy and other psychological treatments for adult depression: Meta-analytic study of publication bias. Br J Psychiatry 2010; 196: 173–178.

Olfson

Marcus

. National Patterns in Antidepressant Medication Treatment. Arch Gen Psychiatry 2009; 66: 848–856.

Mohr

Duffecy

. Perceived barriers to psychological treatments and their relationship to depression. J Clin Psychol 2010; 66: 394–409.

Duggan M. Cell Phone Activities 2013. Pew Research Center, 2013, http://pewinternet.org/Reports/2013/Cell-Activities.aspx.

Knowles

Toms

Sanders

. Qualitative meta-synthesis of user experience of computerised therapy for depression and anxiety. PLoS One 2014; 9: e84323–e84323.

Andrews

Cuijpers

Craske

. Computer therapy for the anxiety and depressive disorders is effective, acceptable and practical health care: A meta-analysis. PLoS One 2010; 5: e13196–e13196.

National Collaborating Centre for Mental Health. Depression: the treatment and management of depression in adults. National Clinical Practice Guideline 90. National Institute for Health and Clinical Excellence (NICE). 2009.

Gilbody

Littlewood

Hewitt

. Computerised cognitive behaviour therapy (cCBT) as treatment for depression in primary care (REEACT trial): Large scale pragmatic randomised controlled trial. BMJ 2015; 351: h5627–h5627.

10.

Knowles

Lovell

Bower

. Patient experience of computerised therapy for depression in primary care. BMJ Open 2015; 5: e008581–e008581.

11.

Agyapong

VIO

Ahern

McLoughlin

. Supportive text messaging for depression and comorbid alcohol use disorder: Single-blind randomised trial. J Affect Disord 2012; 141: 168–176.

12.

Agyapong

VIO

McLoughlin

Farren

. Six-months outcomes of a randomised trial of supportive text messaging for depression and comorbid alcohol use disorder. J Affect Disord 2013; 151: 100–104.

13.

Hayes

Strosahl

Wilson

. Acceptance and commitment therapy: The process and practice of mindful change, (2nd edition). New York, NY: The Guildford Press, 2012.

14.

Beck

. Cognitive behavior therapy: Basics and beyond, (2nd edition). New York NY: The Guilford Press, 2011.

15.

Morgan

Jorm

Mackinnon

. Internet-based recruitment to a depression prevention intervention: Lessons from the Mood Memos study. J Med Internet Res 2013; 15: e31–e31.

16.

Kroenke

Spitzer

Williams

JBW

. The Patient Health Questionnaire-2: Validity of a two-item depression screener. Med Care 2003; 41: 1284–1292.

17.

Kroenke

Spitzer

Williams

JBW

. The PHQ-9 - Validity of a brief depression severity measure. J Gen Intern Med 2001; 16: 606–613.

18.

Manea

Gilbody

McMillan

. Optimal cut-off score for diagnosing depression with the Patient Health Questionnaire (PHQ-9): A meta-analysis. Can Med Assoc J 2012; 184: E191–E196.

19.

Jacob

Christopher

Neuhaus

. Development and validation of the cognitive-behavioral therapy skills questionnaire. Behav Modif 2011; 35: 595–618.

20.

Baer

Smith

Hopkins

. Using self-report assessment methods to explore facets of mindfulness. Assessment 2006; 13: 27–45.

21.

Ryan

Bernard

. Techniques to identify themes. Field Methods 2003; 15: 85–109.

22.

Kroenke

Spitzer

. The PHQ-9: A new depression diagnostic and severity measure. Psychiatr Ann 2002; 32: 509–515.

23.

Aguilera

Munoz

. Text messaging as an adjunct to CBT in low-income populations: A usability and feasibility pilot study. Prof Psychol Res Pr 2011; 42: 472–478.

24.

Truschel

Jarl

. Behavioural activation versus mindfulness-based guided self-help treatment administered through a smartphone application: A randomised controlled trial. BMJ Open 2014; 4: e003440–e003440.

25.

Fjeldsoe

Marshall

Miller

. Behavior change interventions delivered by mobile telephone short-message service. Am J Prevent Med 2009; 36: 165–173.