Family Access to Dentist Study

Purpose:

The purpose of this study is to evaluate the effectiveness of new referral approaches to increase receipt of dental care among inner-city urban and rural elementary school children who were screened at school and have restorative treatment needs. The primary outcome is the assessment of dental care receipt through dental exams. The secondary outcomes are change in caregiver illness perception, measured through the IPQ-RD, and behavioral intention measured through items in the caregiver questionnaire.

Study Design:

The study will utilize a multi-site, double-blind, parallel arm, community-based randomized trial design. The experimental intervention will be based upon the Common Sense Model of Self-Regulation (CSM) to address parent/caregiver illness perception and navigation of resources to increase receipt of dental care during a 7-month follow-up period for children with restorative dental needs. In this five arm study the new CSM theory-driven referral letter alone, the new CSM referral letter and Dental Information Guide (DIG), the new CSM theory-driven reduced referral letter alone, and the new CSM theory-driven reduced referral letter and reduced Dental Information Guide (DIG) will each be compared to a standard referral letter to increase receipt of dental care among inner-city urban African American, and rural Hispanic and White elementary school children.

Participants:

Subjects will be parents/caregivers and their children in grades K-4 in one of four school districts; East Cleveland City School District (OH), Cowlitz County School Districts (WA), Lewis County School Districts (WA), and Bedford (OH). The study will be offered to all caregivers and their children including those children with special health care needs. At the Ohio site, only English-speaking caregivers and their children will be enrolled while both English- and Spanish-speaking caregivers and their children will be included at the Washington site. At the beginning of the school year dental screening, primary caregivers will be randomized if their child has a tooth with an International Caries Detection and Assessment System (ICDAS) active lesion score of 2 or greater. Caregivers with multiple children in K-4 will receive the same intervention assignment. The second year of recruitment included Bedford School District in Ohio and East Cleveland School District KG and other grade families if they did not enroll in the first year.

Procedures:

Caregivers will be recruited at the beginning of the school year and complete baseline forms (consent form, Illness Perception Questionnaire-Revised for Dental (IPQ-RD) questionnaire, and caregiver questionnaire) prior to dental screening and randomization. Children in K-4 whose parents have consented will receive a screening exam in early October 2015 (October 2016 for the second year of recruitment). After screening exams, eligible caregivers of children with restorative referrals will be randomized into one of five study arms. Different referral letter with/without Dental Information Guide according to study arms will be sent home with the child on the day of screening and mailed to the caregiver address within 24 hours.

Caregivers will be followed at 2 time points: 2 weeks after the receipt of the intervention letter and 2 weeks prior to the final (exit) study visit examinations in May 2016 (May 2017 for the second year of recruitment)-approximately 7 months after the initial dental exam. At both time points caregivers will complete the IPQ-RD and the caregiver questionnaire. A small sub-sample of 60 caregivers will be randomly sampled from three arms (standard letter, intervention letter, and intervention letter+DIG) to complete the IPQ-RD 1 month after enrollment for fidelity purposes. All questionnaires will be mailed to the home address and asked to be returned to the class-room teacher or the outreach staff. Further, outreach staff will ensure that the caregivers received and returned the materials through follow-up calls. Children will have a follow-up dental exam in May 2016 (May 2017 for the second year of recruitment).

Analysis Plan:

Primary Statistical Analysis: Descriptive statistics will include frequencies (categorical variables) and means (continuous variables) of baseline covariates among the five intervention groups. Chi-square tests (for categorical variables) and t-tests (for continuous variables) will be performed to test for group differences (by site and overall). In our primary analysis, we will fit multivariate generalized linear (specifically, logistic regression) models with dental care receipt as the (binary) outcome and the child as the unit of analysis (as receipt of dental care will be determined for each child). A generalized estimating equations (GEE) approach (with an exchangeable working correlation structure) will be used to allow for correlations among multiple children for a given caregiver. Secondarily, we will fit a model with a site by intervention group interaction, and the latter will be tested to formally assess possible intervention effect heterogeneity over sites.

We will also consider working correlation structures (e.g., alternating logistic regressions) that allow for second-level clusters, namely schools (within which the first-level cluster, caregiver, is nested). We will determine final models based on the quasi-likelihood information criterion (QIC). Intervention (and other covariate) effects will be tested using generalized score statistics.

Analysis of Secondary Outcomes (Questionnaire data): Summary statistics (including means and standard errors) for questionnaire responses for both the IPQ-RD and behavioral intention items within the caregiver questionnaire will be calculated by site and intervention group. This will be done for the overall questionnaire (IPQ-RD and behavioral intention), for each of the five IPQ-RD cognitive constructs, and the two behavioral intention questions. Analyses will be done by site and overall (controlling for site). A summary measure approach, based on the mean questionnaire response, will be used to compare questionnaire outcomes across intervention groups, for the overall questionnaire and by construct. Specifically, for each construct (or overall) a linear regression model will be used with the mean response as the dependent variable, and intervention group indicators and selected baseline (control) covariates as independent variables. Inferences (that is, p-values and confidence intervals) will be performed assuming summary measures are normally distributed. This assumption will be tested using the Shapiro-Wilk statistic and if violated a suitable transformation of the summary measure used or a nonparametric approach (e.g., using ranks of the summary measures) conducted.

Missing Data:

In the event of missing questionnaire responses, we will first assess whether the data are missing completely at random (MCAR), that is, whether missingness for a given item is dependent only on participant baseline characteristics and not further on any other item responses. This will be done using a generalized estimating equations (GEE) approach, that is, by fitting a multivariate generalized linear (e.g., logistic regression) model to missing data indicators for each item, including as explanatory variables selected baseline covariates and other item responses, and testing for an effect of the latter.

If the MCAR assumption is deemed reasonable (no significant effect found), we will conduct the above regression analyses with mean outcomes based on available data. A weighted analysis (that is, regression with the use of the weighted least squares criterion) will be conducted to adjust for differential missingness (thus, non-homogeneous summary measure variances) among participants. If the MCAR assumption is found to be violated we will use a multiple imputation approach (in conjunction with the summary measure approach) using a suitable imputation model (i.e., using other item responses). For each construct (or overall) we will perform an F-test for a difference in the mean summary measure among the three intervention groups, and obtain nominal 95% confidence intervals for pairwise mean differences. If necessary, we will include a random effect for school.

Data Management:

Data for this study are captured using examinations, forms, questionnaires, and audio recordings. Study data will be collected and stored using the REDCap platform. REDCap is a secure, web-based application designed to support remote data capture for research studies, providing: 1) an intuitive interface for validated data entry, 2) audit trails for tracking data manipulation and export procedures, 3) automated export procedures for seamless data downloads to common statistical packages, and 4) procedures for importing data from external sources.

Study forms will be completed by study personnel on paper, and subsequently entered into REDCap by study staff. Because randomization is contingent upon the computer, no paper randomization forms will be completed. Paper forms will be securely stored in a locked file cabinet at both sites when they are not in use. Completed paper forms will be transferred between sites via courier or other secure document service. Similarly, questionnaires used in the study (IPQ-RD, Caregiver Questionnaire) will be completed by subjects on paper sheets and returned to Outreach Workers in each school. Study staff will enter the information into the database. All data forms, questionnaires will be entered into database stored in a secure password protected computer. Recorded audios will be stored in secured/locked cabinets in one of the Project Coordinators' offices.

Dental examination and study questionnaire data will be reviewed for accuracy and completeness after each study visit. The Outreach Workers will review the questionnaires after the participants have completed them, and the Project Coordinators and the Data and Statistics Team will do a final review after the documents have been transmitted. An annual report (or as requested by the DSMB or NIDCR Program Official) will also be generated for sharing progress of the study.

The Data \& Statistics Team will generate regular reports showing enrollment and potential data anomalies, which will be sent to PIs, Project Coordinators, and other relevant study staff. The Data Manager will work with the PIs at each site to resolve any data anomalies that arise during recruitment. The Data \& Statistics Team will also respond to data queries generated by the PIs, Project Coordinators, or other study staff.

Statistical analyses and summary reports will be generated by the study Biostatistician during the course of the study. At the end of the study, the study Biostatistician will conduct analyses of the data and assist in preparation of study publications and presentations. The Data Manager will provide technical and data support for the Biostatistician throughout the study.