Staff collected a urine sample from the youngest potty-trained child in the home. If no children in the home were potty-trained, a urine collection kit was provided to the parent for sample collection. The kit included two all-natural cotton pads, gloves, specimen bags, and an instruction sheet. Participants were instructed to wear gloves to prevent contaminating the sample with nicotine on their hands and to place a pad in the child’s diaper until it was urine soaked. Participants were encouraged to collect a sample overnight, the night before the scheduled visit for heavy soaking, and collect an additional sample the day of the visit. The pad was collected during the visit and transported to our offices where the urine-soaked pad was cut into 0.5-inch wide strips using scissors and placed in a syringe. The syringe was used to extract the urine from the soaked pad. The urine was expelled into a bio specimen cup and stored in a –80°C freezer until it was sent to the laboratory for assaying.

If the identified child was potty-trained, parent assistance was requested to place a specimen commode (aka “hat”) in the toilet for collection and to assist the child in sample collection. Each urine sample was analyzed for the presence of nicotine, a nicotine metabolite (cotinine), and NNAL.

At each assessment visit, staff hung a passive air nicotine dosimeter [31], which is an objective, validated method to quantify the level of ambient nicotine present in the air. The dosimeters consist of a filter treated with sodium bisulfate and contained in a 4-cm polystyrene cassette. Nicotine passively diffuses to the dosimeter and binds to the filter. The nicotine dosimeter was hung in the main activity room of each home, as well as the family car, if available. The dosimeter remained in the home and car for a minimum of 7 days for sample collection. After 7 days, staff collected and shipped the dosimeter to the University of California, Berkeley, for analysis using gas chromatography to quantify the presence of nicotine in the air.

Surface dust sampling was collected from two locations—the home and the dashboard of the car—to assess for the presence of THS or residual nicotine and NNK remaining on surfaces after the cigarette was extinguished. A presoaked all-natural cotton wipe was used for surface dust collection. The cotton wipe was soaked with a solution consisting of 1.5 mL of 1% ascorbic acid mixed in reagent-grade water for sample preservation [44]. A 10 × 10 cm wire frame was created for precise marking of the sampling area. The moistened cotton wipe was swabbed over the 100-cm² area on a porous location with visible dust. This sample was taken from the same room that the nicotine dosimeter was placed in as well as another central location in the home.

In addition to the brochures detailed above, participants randomized to the biomarker feedback (active) condition were provided with baseline results from the child’s nicotine, cotinine, and NNAL urine analysis and nicotine levels found in the home and car from the nicotine dosimeter. The information was detailed in a Participant Home Report, and information was verbally reviewed by the counselor. The target behavior in counseling, determined by the participant, was either smoking cessation or the institution of complete home smoking restrictions. Counseling was administered using MAPS counseling techniques [30]. MAPS is a unique approach, derived from a motivational interviewing framework, which utilizes motivationally based techniques to enhance commitment and intrinsic motivation for change in combination with cognitive-behavioral techniques to target self-efficacy, coping, stress, and negative effects. Participants, in collaboration with their counselor, developed a wellness plan that addresses not only smoking cessation or eliminating home smoking, but also other concerns and barriers to changing their smoking behaviors, such as parenting or relationship stress.

All counseling was delivered by a female study counselor who acted as the sole counselor for all participants randomized to the active condition. The study counselor received approximately 40 hours of face-to-face MAPS counseling training, followed by additional training until performance criteria for competence and adherence to the assigned protocol were reached. The counselor participated in monthly, individual supervision to maintain performance standards. In order to monitor protocol fidelity, all counseling sessions were audiotaped, a random sample of two of the counselor’s monthly sessions were coded, and detailed feedback was provided using a modified version of the Motivational Interviewing Treatment Integrity (3.1.1) [45] coding and feedback system to ensure adequate competence and adherence to the protocol.

The counselor followed a strict protocol for completion of the counseling visits and to maximize session adherence/retention. All five sessions were conducted within a 12-week window. The first, second, and fifth sessions were conducted in the participants home, and the third and fourth sessions were conducted over the telephone. The window allowed for sessions to be scheduled every other week until the end of the counseling window. Visits were scheduled at times negotiated between the participant and the counselor. The counselor made daily attempts to reach participants to schedule or reschedule a visit. In addition, contact letters were mailed to the participant’s current address and attempts were made to reach alternate contacts of persons listed by the participant, who would always know how to contact the participant.

Participants in the comparison condition received health education, which included a review of the aforementioned four health education brochures at the baseline assessment. Participants randomized to the control condition also received the baseline biomarker feedback information but at the completion of their last follow up session, to avoid influencing the study outcomes, but as a gesture of good will and gratitude for participation.

At each of the three assessment visits, participants were given a choice of receiving a gift card to either a local grocery chain or a discount department store. At baseline, participants received a US $50 gift card and an additional US $5 gift card as compensation for any cell phone usage incurred from study-based calls. Participants were also provided a small gift bag including a manicure kit, a small hair towel, a bath pillow, and a body sponge. Children were given a stuffed bear named Crystal, which was designed and marked by the American Lung Association. The bear included a short poem inscribed on the tag, which read, “Second hand smoke is not a joke! Please keep clear of smoking my dear.” After completing the 16-week home visit, participants received a US $25 gift card, and a US $50 gift card was offered for completion of the 26-week visit.

The primary outcome for this study is change in home SHS exposure measured by the passive air nicotine dosimeter at 26 weeks poststudy enrollment. The specificity of the nicotine dosimeter as a marker of SHS is an appropriate marker to measure a reduction in home air nicotine. Reduction in child’s cotinine and NNAL levels and parental report of smoking behavior change (ie, cessation, quit attempts, reduction, and restrictions) served as secondary outcomes. Specifically, carbon monoxide–verified, 7-day, self-reported point-prevalence abstinence assessed 6 months after study enrollment and defined as no smoking, not even a puff, for 7 consecutive days prior to the final assessment point at month 6 served as the smoking cessation measure [47]. Using a conservative approach, participants who dropped out or were lost to follow-up were considered to be smoking. Adoption of home smoking restrictions was assessed by a question modified from the Current Population Survey Tobacco Use Supplement [48]: “Which statement best describes the rules about smoking in your home?” Participants reported whether there were no bans (“Smoking is permitted anywhere”), some restriction (”Smoking is allowed in some places or at some times”), or complete restriction or ban (“No one is allowed to smoke anywhere”). Smokers were also asked to report the number of cigarettes smoked per day and the number of quit attempts in at least 24 hours since enrollment and their use of programs (eg, Helpline) or pharmacological therapy (eg, nicotine replacement therapy and other cessation medications) at weeks 8 and 26.

In order to examine psychological mediators and moderators of the proposed intervention (exploratory aims), key factors related to important determinants of smoking behavior change were selected for each set of analyses. These included proposed moderators of effect including sociodemographic variables for parent and child, baseline smoking, and social environmental variables. Participants were asked to detail the number of days they smoked in the last 30 days, the number of cigarettes they smoked on days they smoked, use of smokeless tobacco products and other tobacco used in the last 30 days, whether they used mentholated cigarettes, the number of 24-hour quit attempts they made in the past year, the longest time period they had without smoking a cigarette, and their lifetime nicotine replacement therapy use. Other tobacco-related variables included nicotine dependence, measured using “time to first cigarette” [49] and readiness to quit smoking [50]. Alcohol use [51] (ie, number of days one drank at least one drink, number of days one drank five or more drinks) was also assessed. Social environmental variables included the number of smokers living in the home, number of adult nonsmokers living in the home, outdoor smoking options, location where most smoking occurs, and adoption of home smoking restrictions in the home and car. Child exposure to tobacco smoke was also measured (ie, exposure to tobacco smoke in the home and outside the home in the past week). To assess any potential illnesses or health conditions that might impact the intervention or its effectiveness, parents were asked the following: “Has any health care provider ever told you that your child has any of the following illnesses: (eg, asthma, allergies, cancer, heart problems)?” Psychosocial variables included depressive symptoms (assessed using the 10-item Center for Epidemiological Studies Depression Scale [52]) and stress (measured using the 4-item Perceived Stress Scale [53]). Social influence was measured by asking how many of the participant’s five closest friends smoked cigarettes [54], and social support was measured via the Partner Interaction Questionnaire (Short-Form) [55]. Participants’ view of their social standing in the community and the nation was assessed using the MacArthur Scale of Subjective Social Status [56].

Key mediators are important behaviors, attitudes, and beliefs related to both the proposed intervention and the outcomes. Optimism bias was assessed using a modified version of the Smoking Hazards Scale [57], a 12-item questionnaire with four scales designed to assess perceptions of obvious and subtle health risks due to smoking and other health behaviors. We modified the existing measure to assess perceptions regarding the child’s exposure to SHS. The Passive Smoking Outcomes Expectations Scale [17] was used to assess parents’ expectations of the outcomes that may result from their children’s exposure to SHS. Intrinsic motivation was assessed with the Treatment Self-Regulation Questionnaire [58-60]. We used this tool to measure intrinsic and extrinsic motivation in order to adopt home smoking bans and quit smoking. Smoking abstinence self-efficacy was assessed using the 12-item Smoking Self-Efficacy Questionnaire, and self-efficacy to engage in behaviors to decrease child SHS exposure (eg, adopt home smoking bans) was measured using the Passive Smoking Outcomes Expectations scale [17]. To assess whether receipt of biomarker feedback was associated with changes in motivation and confidence to quit smoking/adopt home smoking bans, participants were asked “How motivated are you to quit smoking/adopt bans?” and “How confident are you in your ability to quit smoking/adopt bans?” In order to assess message salience and comprehension, at the conclusion of each of the three counseling sessions, participants were asked a series of five questions to assess their comprehension and ability to relate to the material presented during that session. Finally, social support for both cessation and adopting home smoking bans was measured using an abbreviated version of the Partner Interaction Questionnaire [61], consisting of six items designed to assess both positive and negative support behaviors [62].

The sample size was chosen to ensure adequate power in order to examine the efficacy of tobacco-specific biomarker feedback in terms of the difference of SHS home exposure (primary outcome) between the two treatment groups at week 26. The power calculation was based on two-sided, two-sample t test assuming a type I error of 0.05. A total sample size of 180 (90 per treatment group) was determined to achieve 85% power in order to detect an expected difference (between 1.15 [SD 1.58] μg/m³ for the biomarker feedback group and 1.89 [SD 1.68] μg/m³ for the health education group) in air nicotine levels measured by dosimeters, which was derived from the assumed adoption rates of home smoking bans of the two treatment groups (55% vs 25%).

For the continuous end points such as reduction (ie, change score) in SHS home exposure (primary outcome), child urinary cotinine levels (secondary outcome), and parental self-report amount of smoking, the primary analysis will be the two-sided t tests for the measurements at week 26. In case the distribution of nicotine or cotinine level shows nonnormality, the logarithm transformation on the biomarkers or nonparametric Wilcoxon rank sum test will be applied. Categorical variables such as the adoption of home smoking restriction policies and parental quitting behaviors will be analyzed using a Chi-square test (or the Fisher exact test if there are cells with a frequency ≤ 5).

Supportive analyses will include multivariable regressions accounting for the effect of relevant baseline sociodemographic, tobacco-related (eg, cigarettes smoked per day), and psychosocial variables (eg, depression), especially if significant effects on the outcomes or imbalance between the study groups are discovered. To utilize the repeated measures data more efficiently, we will also use linear mixed models for the continuous endpoints and the generalized linear mixed models for the binary endpoints. All regression models will include treatment, time, and possibly interactions between treatment and time, after adjusting for other aforementioned covariates. The model-based and empirical estimates of the endpoints for both treatment arms at each of the three time points will be generated and displayed with appropriate tabular and graphical methods.

We will use the intent-to-treat method to include all subjects who were randomized in the final data analysis. For subjects with missing data at week 26, a zero change score will be imputed. As a supplementary analysis, we will also use the completers-only method to analyze variables with missing values.

Potential treatment effect moderators will be tested by including their main effect and interaction with the treatment group in multivariable regressions. For the analysis of potential mediators, we will employ the four-step logistical process of testing mediational effects developed by Baron and Kenny [63] and others and summarized by Frazier et al [64]. We will also use structural equation modeling techniques to examine the full pattern of predicted mediational pathways. We propose to develop a series of three models to examine how the proposed interventions influence the process of smoking behavior change. The first step in the process is to examine if there is any relationship between the proposed intervention and behavioral determinants as defined by our theoretical model. This step will involve the development of models, with each key behavioral determinant being the dependent variable (Model A). The next model will examine how changes in behavioral determinants are related to interim outcomes, that is, week 16 outcomes (Model B). The final model will examine how changes in behavioral determinants at the interim (16 weeks) evaluation are related to the outcomes at the final evaluation (26 weeks; Model C).