Article Text

Randomised pilot trial of cash incentives for reducing paediatric asthmatic tobacco smoke exposures from maternal caregivers and members of their social network
1. Mandeep S Jassal1,
2. Cassia Lewis-Land1,
3. Richard E Thompson2,
4. Arlene Butz1
1. 1 Department of Pediatrics, Johns Hopkins Medicine, Baltimore, Maryland, USA
2. 2 Department of Biostatistics, Johns Hopkins University Bloomberg School of Public Health, Baltimore, Maryland, USA
1. Correspondence to Dr Mandeep S Jassal, Johns Hopkins Medicine, Baltimore, MD 21287-0010, USA; mjassal1{at}jhmi.edu

## Abstract

Background The primary aim was to evaluate the efficacy of financial incentives for reducing paediatric tobacco smoke exposures (TSEs) through motivating cigarette usage reduction among low-income maternal caregivers and members of their social network.

Design Randomised control pilot trial over a 6-month study follow-up time period. The study was undertaken from May 2017 to -May 2018. Once monthly follow-up visits occurred over the 6-month study period.

Setting Baltimore City, Maryland, USA.

Participants We grouped 135 participants into 45 triads (asthmatic child (2–12 years of age), maternal caregiver and social network member). Triads were assigned in a 1:1 allocation ratio. The maternal caregiver and social network members were active smokers and contributed to paediatric TSE.

### Outcomes

The primary outcome was the monthly mean change in children’s monthly salivary cotinine levels over the 6-month study period. Triads were excluded from analyses if we were unable to collect at least 3 monthly salivary cotinine from the group’s child member over the 6-month study. Secondary outcomes included smoking characteristics of adult participants, social network–caregiver interactions, paediatric asthma control and caregiver mental health outcomes (anxiety and depression).

### Statistical analysis

We compared the baseline characteristics of cohorts in the RCT trial using summary statistics (eg, mean or medians). Regression models were fit using generalised estimating equations (GEEs)51 to model the longitudinal relationship between the trial’s monthly follow-up visits and cotinine levels among children, caregivers and social network participants. Standard GEE models allowed for missing data to be modelled by making use of all available data without the need to use imputations to replace missing values.52 Data were analysed on an intention-to-treat basis.

Non-transformed, raw paediatric cotinine values were positively skewed, and log transformation of the monthly data did not increase the symmetric distribution. Non-transformed and log-transformed data did not pass the Shapiro-Wilk normality test (see online supplemental figure 1). Transformation of the data does not alter the results for the primary outcome and the GEE modelling displayed henceforth using non-transformed data.

### Supplemental material

Descriptive statistics using measures of central tendency described baseline paediatric asthma control and caregiver mental health (anxiety and depression). Inferential statistics of these two clinical outcomes were not reported since valid statistical comparisons are not possible, given the lack of significant decrease in paediatric cotinine levels in intervention and control cohorts. Given that our study’s primary outcome was a decrease in paediatric cotinine for secondary effects on improvement in both asthma control and caregiver mental health, we could not satisfy this premise, given the lack of fulfilment of the first objective, and therefore no further inferential analyses were performed. Due to budget restraints, we also did not collect the multitude of socioeconomic, environmental and clinical data from adult participants that would have allowed for more meaningful understanding of caregiver mental health outcomes—all of which are needed elements since we could not ascribe the results to an intervention.

Power calculations were based on the hypothesis of a 50% reduction in salivary cotinine levels at the 6-month study interval between study groups—derived from previous nicotine measurements in CM and smoking ban studies by our group.4 To detect a mean monthly cotinine difference of 75% (salivary cotinine SD 2.2 ng/mL) difference between groups, with 80% power using a cut-off for statistical significance of 0.05, we needed to have a sample size of at least 45 caregiver–child triads. We attempted to increase the enrolled triad goal by 10% to account for loss to follow-up but we were unable to consistently maintain the additional required five triads in the study. Missing data were limited to approximately 10% over the total 6-month study. Missing data were not imputed in the analyses due to the relatively low missing data, as well as the possibility that there is no random loss to the missing data. There were no observed differences in patterns of missing data among randomised children, caregivers and social network members between the two cohorts. Given our lower-income study participants who may have multiple clinical and non-clinical factors for undertaking smoking behaviours and missing appointments, we believe that potential bias could be introduced if imputation processes were undertaken. If individual missing study visits were more likely to be engaging in more smoking behaviours or allowing children to be in high TSE environments, imputation methods could have have resulted in downward biases Individuals not available for follow-up could have been a different subset of participants (eg, higher psychosocial stressors, poorer medical supervision of paediatric asthmatics or lesser interest in personal well-being), whose inclusion could have yielded lesser monthly change in individual and paediatric cotinine levels over the 6-month study follow-up period. Statistical significance was defined as two-sided, with significance level set at p<0 .05. Statistical analyses were conducted using STATA V.15.1.

## Results

### Participant characteristics

Among the 418 individuals assessed for eligibility, 64% were excluded for predominantly the lack of fulfilment of inclusion criteria (online supplemental figure 2). Approximately 23% of maternal caregivers screened had expressed interest in the study but were excluded due to the inability to locate a member of their social network who is known to smoke in proximity to their child and willing to participate in the study. Less than 5% of participants were lost to follow-up but no reasons could be ascertained due to the inability to contact the participants. Only participants randomised to the intervention were lost to follow-up. We could not ascertain the reasons for participant attrition after multiple unsuccessful attempts at contacting them through home visits and phone messaging. No difference in baseline characteristics were noted in those lost to follow-up, compared with those who remained in the study. No attrition was noted among the control group despite being paid less than the intervention cohort, a monetary feature that the cohort was made aware in the institutional review board-approved informed consent forms.

### Supplemental material

The intervention cohort (total n=63) consisted of 21 participants equally distributed among linked triads of children, caregivers and social network members (table 1). The control cohort (total n=72) consisted of 24 participants within each triad population. The gender distribution of enrolled children was approximately equal, and >70% resided in households with an annual income that was below the federal poverty level.53 Greater than 90% of the maternal caregivers were the biological mothers. There was an equal gender distribution among social network participants and represented a broad spread of assigned social roles. The social network member was in the majority of occasions a first-degree female relative of the maternal caregiver (eg, grandmother and aunt), while approximately 20% were the biological father. The median PHQ-4 scores among caregivers was suggestive of a diagnosis of mild anxiety; no indications of depression were present on baseline screening.

Table 1

Baseline characteristics of study participants (N=135) participants were recruited in triads and randomised by incentive administration

Smoking characteristics were equally distributed across adult participants with most reporting continued cigarette usage at each follow-up time point to address stress or a combination of stress and addiction. Median FTND scores indicated a moderate to high level of nicotine dependence in both social network and caregiver participants. Greater than 70% of participants reported living with at least one smoker and >65% did not enforce a home smoking ban. Younger children (<4 years) were classified as having mild suboptimal asthma control based on the TRACK questionnaire; however, children aged ≥4 years assessed using the ACT showed on average adequate asthma control. High levels of TSE was noted in both cohorts (median levels of >5 ng/mL), though median cotinine levels were 2 ng/mL higher in children randomised to the intervention. Median cotinine and exhaled CO levels confirmed the presence of active, daily smoking status in all caregivers and social network participants.

### Cotinine outcome measures

No significant difference was demonstrated in the intervention cohort’s mean monthly child cotinine levels over the 6-month follow-up study period, compared with the control cohorts (difference in slope (control–intervention)=−0.86 ng/mL/month, p=0.098, CI −0.160 to 1.887) (figure 2). The upward trend in cotinine levels within both paediatric cohorts could not be attributed to smoking patterns in maternal caregivers or social network members, both of whom had declining cotinine values over time. Maternal caregivers in the intervention group had no significant differences in their decline of cotinine values when compared with the caregivers in the control group (difference in slope (control–intervention)=3.30 ng/mL/month, p=0.144, CI −7.717 to 1.127). Social network participants allocated to the intervention also had no significant differences in cotinine values compared with their counterparts in the control group (difference in slope (control–intervention)=−1.59 ng/mL/month, p=0.546, CI −3.569 to 6.745). No adult participant reported use of the state tobacco quit line and subsequent prescription of NRTs to potentially confound cotinine data.

Figure 2

Trend of cotinine levels among triads (child, maternal caregiver and social network member) randomised to the incentive-based intervention, compared with the control cohort. Children, maternal caregivers and social network members in the intervention cohort did not have a significant difference in the mean monthly cotinine levels, compared with the control population. The effect size, p value and CI for each of the populations were the following: children (difference in slope (control–intervention)=−0.86 ng/mL/month; p=0.098, CI −0.160 to 1.887), maternal caregiver (difference in slope (control–intervention)=3.30 ng/mL/month; p=0.144, CI −7.717 to 1.127) and social network (difference in slope (control–intervention)=−1.59 ng/mL/month; p=0.546, CI: −3.569 to 6.745). Colourful lines represent individual trajectories of cotinine over time. Solid black line represents the regression line based on generalised estimating equation models, with CIs designated with dashed lines Arrows represent the cumulative trend in cotinine levels.

## Acknowledgments

We thank Mary Gates, Bonnie Addison and Caryn Gates for data collection and their assistance in data extraction. We are also grateful to Kimberly Lawler of Johns Hopkins University’s Department of Art as Applied to Medicine, who assisted us in developing the manuscript figures.

## Footnotes

• Contributors MSJ, CL-L and AB conceptualised and designed the study. MSJ, CL-L, RET and AB were involved in the analysis and interpretation of the data. MSJ drafted the initial manuscript. CL-L, RET and AB revised the manuscript critically for important intellectual content. All authors agreed to be accountable for all aspects of the work and provided approval of the final manuscript version.

• Funding This work was funded by the National Institute of Environmental Health Sciences (K23ES023814). All phases of this study were supported by the National Institute of Environmental Health Sciences (K23ES023814).

• Competing interests None declared.

• Patient consent for publication Not required.

• Ethics approval The protocol was approved by the institutional review board at The Johns Hopkins University School of Medicine (IRB00064875).

• Provenance and peer review Not commissioned; externally peer reviewed.

• Data availability statement All data relevant to the study are included in the article or uploaded as supplementary information.

## Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.