Effectiveness of a behavioural intervention delivered by text messages (safetxt) on sexually transmitted reinfections in people aged 16-24 years: randomised controlled trial

Abstract Objective To quantify the effects of a series of text messages (safetxt) delivered in the community on incidence of chlamydia and gonorrhoea reinfection at one year in people aged 16-24 years. Design Parallel group randomised controlled trial. Setting 92 sexual health clinics in the United Kingdom. Participants People aged 16-24 years with a diagnosis of, or treatment for, chlamydia, gonorrhoea, or non-specific urethritis in the past two weeks who owned a mobile phone. Interventions 3123 participants assigned to the safetxt intervention received a series of text messages to improve sex behaviours: four texts daily for days 1-3, one or two daily for days 4-28, two or three weekly for month 2, and 2-5 monthly for months 3-12. 3125 control participants received a monthly text message for one year asking for any change to postal or email address. It was hypothesised that safetxt would reduce the risk of chlamydia and gonorrhoea reinfection at one year by improving three key safer sex behaviours: partner notification at one month, condom use, and sexually transmitted infection testing before unprotected sex with a new partner. Care providers and outcome assessors were blind to allocation. Main outcome measures The primary outcome was the cumulative incidence of chlamydia or gonorrhoea reinfection at one year, assessed by nucleic acid amplification tests. Safety outcomes were self-reported road traffic incidents and partner violence. All analyses were by intention to treat. Results 6248 of 20 476 people assessed for eligibility between 1 April 2016 and 23 November 2018 were randomised. Primary outcome data were available for 4675/6248 (74.8%). At one year, the cumulative incidence of chlamydia or gonorrhoea reinfection was 22.2% (693/3123) in the safetxt arm versus 20.3% (633/3125) in the control arm (odds ratio 1.13, 95% confidence interval 0.98 to 1.31). The number needed to harm was 64 (95% confidence interval number needed to benefit 334 to ∞ to number needed to harm 24) The risk of road traffic incidents and partner violence was similar between the groups. Conclusions The safetxt intervention did not reduce chlamydia and gonorrhoea reinfections at one year in people aged 16-24 years. More reinfections occurred in the safetxt group. The results highlight the need for rigorous evaluation of health communication interventions. Trial registration ISRCTN registry ISRCTN64390461.

prevent this, hold the tip of the condom between your forefinger and thumb and roll it down, making sure there are no air bubbles.

Day 40
When you just start seeing someone, it can be awkward to bring up condoms. Most people are happy to talk about condoms though.
More than likely they're thinking the same thing and will be relieved that you brought it up first. It can help to think about what you'll say beforehand.

Example control group message:
Day 60-Hi, it's Ona here. Thank you for taking part in the texting study. Remember to let us know if your contact details have changed by replying to this text or emailing safetxt@lshtm.ac.uk

Analysis of the intermediate outcomes
The intermediate outcome measure comprised multiple ordinal scales. Using data from the first 1025 randomised participants, we assessed the construct validity of the intermediate outcomes and refined them using confirmatory factor analysis (CFA). The originally specified CFA model was based on the a priori factor structure of the model (which items loaded on which factors), as shown in Table S1. For this original model, the goodness of fit indices indicated borderline fit (. root mean square error of approximation RMSEA: 0.083; comparative fit CFI 0.936; Tucker-Lewis index TLI: 0.923). After examining the modification indices to identify sources of poor fit, the model was revised. The variable "most people who have an STI will tell their partner" was dropped from the model due to having a low factor loading of 0.287 on attitudes to partner notification. The variable relating to how easy or difficult it would be to "Put a condom on" was dropped due to cross loadings (indicating a lack of discriminant validity) between the 'Correct condom use self-efficacy' factor and the 'Self-efficacy in negotiating condom use' factor. Finally, we allowed the error terms of the variables "How easy or difficult would it be to tell the last person you had sex with that you had an STI" and "How easy or difficult would it be to tell the last person you had sex with to get treatment" to correlate; and did the same for the equivalent variables that referred to a 'new partner'. We considered this appropriate given that the correlations of error terms between these pairs of variables is likely to be a case of an 'item priming effect' 65 . It seems reasonable that the answer to the first question in each of these pairs will directly affect how the respondent answers the 'treatment' item, as informing a partner of one's infection is a prerequisite to informing them that they will need treatment. Once these changes had been applied, the revised model showed good fit to the data (RMSEA: 0.052, CFI: 0.980, TLI: 0.975). Furthermore, multi-group analyses across genders, sexual orientation, and mode of questionnaire (phone versus written) indicated measurement equivalence across these groups.
The impact of the intervention of these refined intermediate outcome measures was examined. To aid interpretability, we present the results of two analyses. One is based on summing the responses to each item contributing to that intermediate measure, and using a linear regression to test for a difference in mean scores between the arms. The second analysis extends the CFA measurement model described above into a structural equation model, using the allocation as the main predictor variable, thereby estimating the impact of the intervention on the intermediate outcomes in the absence of measurement error. These regressions were adjusted for the same covariates as the primary analyses.

Intermediate outcome results
The effects of the intervention on the intermediate outcomes (measured by summing items) are reported in Table 3 in the paper. Table S2 presents Table 3, with the intervention resulting in a small increase in knowledge related to STIs and in correct condom use self-efficacy.

Trial -Additional Sensitivity analyses
We performed additional non-pre-specified sensitivity analyses under different assumptions from the primary analysis MAR assumption. Sensitivity analysis 1: We completed the MI model including the clinic testing variable an additional covariate. On this imputed dataset, we conducted one sensitivity analysis with the new imputations from this model where all negative clinic tests who have missing outcome data were considered positive. The result from this analysis was OR 1.13 (95% CI 0.997 to 1.28, p=0.05). Sensitivity analysis 2: Using the same imputed dataset (with the clinic testing variable as an additional covariate), we conducted a second sensitivity analysis where all negative clinic tests who had missing outcome data were considered negative. The result from this analysis was OR 1.12 (95% CI 0.97 to 1.29, p=0.13). Sensitivity analysis 3. Sensitivity analysis 3. We followed the primary analysis that assumed missing at random but in imputing missing values, controlled the odds of STI diagnosis to be ¼, ½, 1, 2, and then 4 times as large as that predicted by the imputation model; these sensitivity parameters were varied factorially for the two randomised groups (giving 24 sensitivity scenarios besides the primary analysis). The results were identical to the primary outcome result: OR 1.13, 95% CI 0.98 to 1.31, p= 0.085. This was due to 1) perfect prediction in the imputation model and 2) using the same random number seed to start each sensitivity analysis.

Including baseline number of partners in the imputation model.
We conducted a post hoc analysis replicating the analysis but adding baseline number of partners (< or ≥2 partners) to the imputation model as an additional covariate for both the primary outcome and for the outcome number of partners. The OR for cumulative incidence of chlamydia/gonorrhoea for this analysis was 1.13 (95% CI 0.98-1.31, p=0.087). The OR for number of partners was 1.10 (0.98 -1.23; p=0.11).

Per protocol analysis
We conducted a per protocol analysis where participants who had 12 month primary outcome data were classified as having received the treatment they were allocated to according to the following criteria: 1) they did not stop the messages; 2) they were not among the few participants that did not receive any messages and 3) they reported that they read all or most of the messages. Baseline characteristics among these participants were similar between the groups (Table S3). The OR for cumulative incidence of GC/CT for this analysis was 1.17 (95% CI 0.99-1.38, p=0.06).

Pooled analysis with the safetxt pilot trial data.
We conducted a pooled analysis with all the main trial and pilot trial data from participants diagnosed with an STI at baseline (where the intervention group had been allocated to receive content targeting partner notification, condom use and STI testing. The pooled odds ratio was 1.12 (95% CI 0.99-1.26), P=0.08, I 2 =0% (figure S1).