Study Recruitment and Enrolment
Six large rent- subsidized senior apartment buildings in central Connecticut of 125 to 250 apartments were grouped into pairs matched by size and randomized by the study’s biostatistician into two groups of three buildings each. A sample size of 123 per group provided >90% power to detect mean differences of 0.25 for GI and 0.66 for PS based on data and SDs from a pilot study, using a 2-group t-test with a two-sided alpha of 0.05. An attrition rate of 10% was also assumed. Residents included older adults, aged 62 and above, and people with disabilities under the age of 62, who were restricted after the first cycle to approximately 30% of the study population to ensure adequate enrolment of those 62 and over. The study recruited and enrolled 331 participants from these buildings from 2015 – 2017, 175 in the AMI counseling intervention and 157 into the campaign intervention. Of these, 165 completed the AMI and 76 unduplicated participants attended at least one of the three campaign events based on sign-in registration at the event. Three hundred and six participants completed the T1 assessment. Inclusion criteria included: =>18 years of age, two teeth or more, no conservator; able to comprehend consent form. Exclusion criteria included: temporary building resident; conservatorship; inability to respond correctly to five questions about the study during the consenting process; edentulous; history of infective endocarditis, past six months prosthetic cardiac valve replacement, past six-week myocardial infarction or arterial stent insertion; on dialysis. Eligible candidates signed an informed consent form and completed a baseline (T0) and post intervention (T1) survey and clinical assessment, as well as the intervention assigned to their group. The T1 assessments were administered approximately one month after completion of each intervention.
Ethical Approval and Consent to Participate. The study was reviewed annually by the University of Connecticut Health Center IRB and by NIDCR. All procedures were performed in accordance with the US HHS Belmont Report, 1991, and the Revised Common Rule, 2018 and with requirements of the revised the annually approved study protocol.
Intervention approaches
Intervention activities and processes in both interventions addressed each of the cognitive/emotional and behavioral mediators as mechanisms of intervention in the adapted IM model.
- The face to face Adapted Motivational Interviewing Intervention (AMI: This 45-60 minute counseling approach was administered by trained bilingual English/Spanish speaking oral health educators no more than one month after the baseline survey. It was guided by the IM model. To prepare for the counseling intervention administration, study PIs established cutoff points for scale means in the pre-intervention survey below which it was determined that participants needed intervention(21). Interventionists followed an intervention protocol that began with cognitive/emotional mechanisms, followed by a behavioral intervention instructional component. First, participants were asked to describe their oral health concerns which the interventionist matched with the mechanisms that scored under the cutoff. Next interventionists discussed with participants the mechanisms scoring under the cutoffs and both engaged in a conversation to determine how to best address them, helped by an explanatory script for each mechanism. Next interventionists proceeded to the behavioral instruction component of the intervention. First they reviewed with the participants the pre-intervention plaque scoring record which illustrated in red those teeth had plaque accumulations. This record helped the participant to visualize where to target more effective brushing. The second step involved showing two brief videos in English or Spanish to demonstrate correct brushing and flossing techniques. Then participants practiced brushing and flossing on a typodont (model) and were scored and provided with feedback until they had mastered these activities to the best of their ability. This process was referred to as “practice to mastery” (PM). Finally, with the interventionist, participants created their own plan for oral health improvement with strategies for improving brushing, flossing and selected mechanisms, and kept a copy for themselves. All but one enrollee completed the AMI.
- The Oral Health Campaign intervention consisted of three oral health fairs held three to four weeks apart, facilitated by a committee of trained peer volunteers in each building with intervention team support. Bilingual campaign committees consisted of 10 – 12 volunteers who did not meet study eligibility criteria and represented a diverse cross-section of residents. The volunteer training program began immediately after the baseline assessment was completed for the entire study sample in each building. It consisted of 12 sessions completed over six to eight weeks (21). Session topics included defining a campaign, team building, oral health and hygiene, and knowledge about all intervention domains (mechanisms) in the IM model and their relationship to the desired clinical outcomes. Once completing these basic instructional sessions, volunteers, with staff support, developed messages for residents based on each of the intervention mechanisms, along with interactive games and other activities. Finally they prepared recruitment strategies and a plan for implementing each of the fairs. Most committee members remained with the campaign for all three oral health fairs. The entire training protocol is posted on the study website (www,projectgoh.com).
Fairs were conducted in English and Spanish simultaneously. The protocol for each fair
included a standard presentation on oral hygiene by dental hygienists delivered in English and Spanish, followed by a question/answer period. Campaign Committee members assisted by project personnel staffed twelve tables, each with a different message associated with a specific mechanism along with related games and informational handouts. Attendees rotated from table to table querying Committee members. At one of the tables, they were instructed on brushing and flossing using a typodont (model). Both enrolled study participants and non-enrolled residents were welcome at the fairs. Attendance was recorded on a sign-in sheet. Each attendee recorded their visit to each table, evaluating their experience with a “passport”. Their assignment was to complete the passport before leaving the fair. The passport was turned into project staff on departure from each fair. It served as a record of attendance and provided one measure of dosage. All fairs included a raffle and refreshments. Attendance at each fair averaged 45 people including enrolled participants and visitors and a number of attendees attended 2 or all three fairs. Of enrolled participants (n=143), 95 attended at least one fair. Among the anecdotally derived reasons for non-attendance were disability, depression, other obligations (doctor appointments, work schedule), preference for avoiding public events in their building and the incorrect perception that the fair would be conducted in a language they did not understand.
Measures
Most cognitive/emotional domains in the IM model were adapted from pre-existing literature on factors shown to be associated with oral hygiene behavior. Several indices were based on formative research or pilot testing with the study population including fear of oral diseases and worries about oral health self-management. The latter has been validated and published.
Baseline covariates included demographics (age <61 vs 62 and older, gender, income <$900.00 or =>$900, ethnicity (Black non-Hispanic, Hispanic, White non-Hispanic plus other) perceived oral health status rated on a four-point Likert scale as poor (1), fair (2), good (3) or excellent (4), and treated as both a continuous and categorical variable, dichotomized as poor/fair vs good/excellent (22), number of diagnoses that interfered with daily activities (0 and 1or more) and depression measured with the CES-D short form (=>4 high versus <4 low) (23).
Cognitive/emotional Mechanisms included oral health self-efficacy: 5 items, (a .603) and locus of control: seven items (a .72) with responses as 4-point Likert scales 1 (low) – 4 (high) for both scales (24); perceived oral health risks (chances of getting specific health problems associated with oral health)(22): five questions with responses rated 1 – 4 with 4 as least chances (a .76); fears of oral diseases: 4 items rated 1 – 4, with 4 as no fear (a .82); intentionality (25): a 6 question scale with responses rated as 0 (no intention) to 2 (high intention) (a 72); importance of oral health behavior (26). 9 items rated from 1 not important to 5 important (a .672.); oral hygiene self-management worries scale (OHWSMS): 19 questions with responses rated 1-4 with 4 as least worried (a .93) (27).
Behavioral mechanisms included sugar intake: five questions asking about frequency of consumption of sugar and starch (0 is never to 4 as > five times a day); brushing often (1 = <2/day, 2=2+/day); and flossing often (0=<1/day v. 1=1+ per day).
Outcome measures consisted of the Gingival Index (GI) (28) and Plaque Score (PS) (29), both assessed by two trained dental hygienists calibrated each year against an experienced dental examiner. The GI assessed the status of gingiva associated with 6 surfaces of each tooth, three buccal and three lingual, by scoring for gingival inflammation from 0= no visible inflammation to 3=overt inflammation and spontaneous bleeding. The index mean and individual scores were calculated by summing all surface GI scores and dividing by the total number of surfaces. The PS was obtained by the examining hygienist who identified and recorded dichotomous presence or absence scores for bacterial plaque on each of 6 tooth surfaces. PS is expressed as a percentage of surfaces stained red with plaque over total number of surfaces, or a ratio. Reliability of the clinical assessments was assessed prior to T0 and T1. Two hygienists conducted the clinical assessments with the dental director of the study as the gold standard. After training and prior to T1, Kappa improved from a difference of 0.45 to 0.54 to a difference of 0.72 to 1.00 for the Gingival Index and from a difference of 0.46 to 0,78 to a difference of 0.77 to 0.94 for plaque scores. Measures and calibration procedures are further described elsewhere (21)
Fidelity measures for the AMI counseling intervention included a record of domains covered in each administration, whether the prepared script was utilized, duration of intervention, record of brushing and flossing skills, and a documented plan,. These were reviewed for completeness and accuracy by PIs during each cycle along with reviewing 10% of audio recordings of AMI administrations in English and Spanish. All participant files included recorded plans. Fidelity measures for the oral health fairs included a record of slide presentations at campaign events, attendance via registration plus passport record of attendance, exposure to each message table, and observations of each campaign.
Statistical Analysis
To investigate within-group change separately by intervention group, paired t-tests are reported for clinical outcomes GI and PS (Table 1) and for intervention mechanisms (Table 2). These tests analyze changes from baseline, but are not meant for comparing the two interventions. To make inferential, adjusted assessments of the intervention, we used repeated measures generalized linear mixed models (GLMMs) with main effects of time (0 vs 1), intervention (AMI vs. Campaign), the (time × intervention) interaction plus covariates of interest (e.g., demographics and health status variables). The interaction terms assess the extent to which the outcomes differ between groups and across time, and interpretations for significant interactions are provided for GI and PS (See Table 4, and Figures 2 and 3). A final reduced model was fit for including significant mediators, moderators and main effects plus interaction for both GI and PS (Table 5). These GLMM models were estimated in the MIXED procedure in SAS version 9.5. For binary outcomes (brushing and flossing) the general estimating equations (GEE) approach was used in the GENMOD procedure in SAS (30). A two-sided level of significance of 0.05 was used to determine statistical significance.