Methods
We surveyed a nationally representative sample of dentists across the United States. The survey examined dentists' knowledge, attitudes, beliefs, and willingness regarding the possibility of offering routine HIV rapid testing in the dental care setting. We developed the survey using a modified version of the Theory of Planned Behavior as a heuristic framework. This framework posits that human action is guided by 3 types of considerations: (1) beliefs about likely outcomes of the behavior and evaluation of these outcomes, (2) beliefs about the normative expectations of others and motivation to comply with these expectations, and (3) beliefs about the presence of factors that may facilitate or impede performance of the behavior and perceived power of these factors. Our modified framework also encompassed a feedback loop pertaining to structural factors, such as reimbursement mechanisms, that could influence providers' knowledge, attitudes, and willingness to offer testing; in turn, dentists' willingness to offer HIV screening, and patients' demand for screening, may influence how other stakeholders develop public health and financial policies toward HIV screening.
Participants
We obtained the sampling frame from the American Dental Association (ADA) Survey Center, which maintains a sample frame of professionally active dentists, dental students, and retired dentists independent of membership in the ADA. The ADA Survey Center provided a stratified random sample of dentists based on 2 variables: urbanicity and practice type. The type of practice variable mandated 2 separate strata: general dentists practicing in (1) private practice settings and (2) the public health sector. Private practitioners constitute 95% of all dentists surveyed by the ADA, a sample of roughly 130 000 individuals.
To capture dental practices in areas most prominently affected by the HIV epidemic, we drew 75% of the sample from Ryan White CARE Act eligible metropolitan areas (Ryan White CARE Act, Pub L No. 101–381, 104 Stat. 576, enacted August 18, 1990). The remaining 25% of the sample was taken from outside metropolitan areas as defined by Beale's Rural–Urban Continuum Code. An additional stratum of sampled dentists consisted of 80% of the 383 dentists in the United States who identified by self-report as "public health dentists." These individuals represent the professional group potentially most amenable to population screening.
Data Collection
The survey instrument consisted of 38 questions assessing HIV rapid testing, other screening behaviors and intentions, attitudes on HIV testing, subjective norms, perceived behavioral control, knowledge, background, demographics, dental practice, patient populations, and system and structural factors. In addition, we assessed perceived barriers to offering HIV testing in the dental care setting based on items derived from a qualitative elicitation study that was conducted to inform the survey development. The National Opinion Research Center at the University of Chicago conducted the survey from November 2010 through November 2011. The survey contacted a sampling frame of 2876 dentists, using the standard 5 contacts in the order recommended by Dillman: prenotification letter, initial questionnaire mailing, thank-you or reminder letter, signature-confirmed replacement questionnaire, and final follow-up via telephone. Working with staff at the National Opinion Research Center, we supplemented follow-up contacts with additional mass faxes, e-mail correspondences, and postcards. All methods of communication contained a toll-free number and e-mail address for respondents to use if assistance was needed. Additionally, respondents were given the option of completing the survey via a password-protected Internet site.
Each survey included an initial cash payment of $10. Respondents who completed the survey in a timely fashion received an additional $20. To increase the response rate, we offered chronic nonrespondents a bonus of $50 as time progressed and a maximum of $100 to hard-to-reach nonrespondents. Remuneration was provided only upon successful completion of the research study.
Measures
The 2 primary outcomes were (1) whether dentists were currently offering HIV rapid testing to their patients in their dental offices or clinics and (2) whether those dentists not currently offering testing would be willing to do so over the next year. In these analyses, the dependent variables were (1) a 0–1 indicator of whether the respondent currently offered HIV rapid testing and (2) a Likert 5-point indicator of whether the respondent might be willing to offer HIV rapid testing in the future.
In the multivariable analysis, we also examined whether dentists were willing to offer HIV rapid oral testing and whether they believed this testing should be part of dentists' professional role. As guided by the Theory of Planned Behavior, we explored the influence of normative expectations of others (specifically, of other dental professionals and of patients), likely outcomes, and impeding or facilitating factors associated with HIV testing in the dental setting.
Statistical Analysis
In calculating descriptive statistics, we adjusted for the weighted and stratified nature of the sample using the SVY routines in Stata version 10.0 (StataCorp LP, College Station, TX). We performed unweighted multivariable analyses, incorporating variables used for stratification and weighting as covariates. We also included as covariates individual sociodemographic characteristics, characteristics of the dental setting and patient population, and personal experiences and knowledge regarding HIV. As a robustness check, we performed the multivariable regression analyses accounting for weighting and stratification. Point estimates were extremely similar to those reported. Multivariable analysis enabled investigation of and adjustment for potential confounders with examination of all binary outcomes. We used multiple logistic regression analysis to examine the ordinal dependent variables proposed by the theoretical model and ordinary least squares regression to examine continuous outcomes.