Context and setting
The setting of this case study is a quality improvement project (PREDICT ) conducted in 14 rural counties in Oregon, USA. The project is being implemented by a dental care organization, Advantage Dental Plans, LLC, Redmond, Oregon USA, The organization is a privately-owned for-profit company and one of the largest providers of dental care in Orgaon State for low income families insured by the United States’ joint federal-state program, Medicaid. In 2016, the organization had 42 group practices and contracted with approximately 200 affiliated smaller, largely rural primary care practices. Providers and staff members from the affiliated practices are not salaried employees of the organization and these practices receive from the organization either per capita monthly payments for patients assigned to them or discounted fee-for-service. Before this quality improvement project began, most clinical care was provided at dental clinics with limited outreach, and incentives paid only to providers were limited to only one performance measure based on access to care to pregnant women.
PREDICT involves changes at the system, community, provider and staff and patient levels . The delivery system changes are intended to increase, substantially, community-based mobile care. At community settings, such as public schools and prenatal and early childhood nutrition and prevention agencies (e.g. Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) and Head Start), Expanded Practice Dental Hygienists provide dental care needs assessment, risk-based preventive services, caries arresting, interim therapeutic restorative services and referrals to care to the company owned clinics or affiliated private practices. PREDICT follows a risk- and evidence-based prevention protocol, in which low risk children receive an assessment and fluoridated toothpaste once a year; moderate risk children receive the services provided to low risk children, as well as topical applications of 38% silver diamine fluoride (SDF) twice a year or 5% sodium fluoride varnish four times a year to the teeth; high risk children receive, in addition to the above, topical applications of 10% povidone iodine and 5% sodium fluoride varnish twice a year. Children with untreated active tooth decay receive 38% silver diamine fluoride and/or interim therapeutic restorations of glass ionomer cement to arrest the decay in specific teeth. The Expanded Practice Dental Hygienists who provide these services are supported by regional community liaisons responsible for establishing and maintaining the local partnerships needed to provide services in the community settings and by case managers who can follow-up with parents to obtain consent for community-based care and assist them with scheduling office appointments, as needed, for more complex care.
Changes in compensation are also part of PREDICT. Compensation changes are based, in part, on pay for performance and the performance metrics are designed to reduce disparities in access to care faced by the low-income patients. Provider and staff incentives are funded centrally from funds withheld on the payments to clinics. A broad range of employees are eligible to receive the incentive funds, including those working directly with clients at the community level, staff with the central administration (e.g., supervisors, case managers and information technology specialists) and clinic dental providers and staff (within the group and affiliated practices). The delivery system and compensation changes that comprise PREDICT are supported by an information technology system that facilitates ongoing performance feedback and quality improvement.
The study design was a cross-sectional survey conducted prior to implementation of PREDICT. The survey instrument and evaluation plan was submitted to the Institutional Review Board of the University of Washington for consideration. It was determined that this effort did not meet the definition of human subjects research.
Sample selection and recruitment
The sample frame was all providers and staff in the group and affiliated private dental practices in the 14 counties and administrative staff at the company’s headquarters. Potential participants were invited to participate via email and the company’s internal newsletter. The company sent invitations to email addresses of all employees from staff model clinics and headquarters and sent invitations to one email address in the private dental care practices (either the dentist-owner or the practice’s generic email). Invitations to the private practices asked the email recipient to share the email with their employees. The invitations were repeated three times to encourage participation. Potential participants were informed that completion of the survey was voluntary, responses would be confidential and that no one within the company would review individual responses. Upon completion of the survey, respondents could provide their contact information (kept separate and unlinked from their responses) to enter a drawing for a tablet computer. Data was collected from July 15, 2015 through September 15, 2015.
Procedures and instruments
Data collection was by a web-based questionnaire implemented by company information technology staff members using SurveyMonkey® (Palo Alto, CA). Usability and technical functionality of the questionnaire was tested in advance with company volunteers.
We developed a questionnaire comprised of 14 constructs; most item response options were 5- or 4-point Likert scales. Readiness for change was assessed by the Organizational Readiness for Implementing Change (ORIC) scales . This validated 10-item questionnaire includes two scales that capture employees’ opinion that the people of the company are committed to the proposed changes (change commitment: Cronbach’s α=0.95) and the belief that the company can handle the adjustments needed for smooth and effective implementation (change efficacy: α=0.93). Response options for these items are 5-point Likert-scales that range from strongly disagree to strongly agree (1 to 5 points).
We developed specific items to assess opinions about the process and practice of dental care provided in the PREDICT model; for these, item response options used a 5-point Likert scale (from strongly agree to strongly disagree). Based on the results of a factor analysis, we created a 4-item scale representing support for the PREDICT model; the four items were respondent’s opinion about: 1) the company’s responsibility for obtaining parental consent for child’s dental care, 2) delivery of risk-based preventive care, 3) a priority focus on preventive care for children with greatest risk and 4) timely restorative and urgent care (α=0.79). Questions about employee’s agreement with the company’s mission statement and the use of methods for caries arrest or stabilization, such as silver diamine fluoride, were assessed separately.
The majority of constructs examined were adapted from the Minimizing Errors/Maximizing Outcomes (MEMO) questionnaire . Organizational climate was measured with a single item about office or practice chaos and the organizational climate scale (α=0.95) that includes 5 subscales: workplace emphasis on quality (11 items, α=0.87), workplace cohesiveness (3 items, α=0.79), trust in the organization (5 items, α=0.86), workplace emphasis on information and communication (4 items, α=0.56) and leadership and governance alignment (8 items, α=0.87) . Other workforce issues were assessed by a validated job stress scale (4-items, α=0.84) , a job satisfaction scale (5-items, α=0.81) , and single items on burnout  and likelihood to leave practice . Data on demographics (i.e., respondent age, gender and race/ethnicity), job role, tenure, part-time or full-time employment and office or practice location were also collected. We categorized place of work as being “central” or “local.” Respondents’ with job responsibilities directly related to the company’s administration were considered “central”. We considered community liaisons and Extended Practice Dental Hygienists to be part of the “central” work group also because they were directly supervised by central administration staff. Providers and staff whose primary place of work was a dental practice were considered “local”. The survey questionnaire is available online as an Additional file 1.
We estimated that a sample size of 51 participants would yield 80% power to detect a 1-point difference in ORIC mean scores between the two study groups (central vs. local), assuming an intra-county correlation (ICC) of 0.05 and standard deviation of 1.25. Frequencies and percentages for categorical responses and medians and interquartile ranges for continuous responses were calculated. If a respondent failed to complete more than 50% of the survey or if more than 30% of respondents failed to complete a specific question, that participant or question was excluded from the analysis. We calculated summary scores for the scales as the mean of the items. If more than 50% of items from a scale for a respondent was missing, the score was set to missing. If less than 50% of items were missing, we imputed the simple mean of the non-missing items for each respondent. Some item responses were reversed so that a higher summary score for the scale reflected a more positive attribute. We used linear regression models to investigate the association between organizational readiness for change and support for PREDICT with workforce characteristics and work environment. We tested for differences in readiness for implementing changes between respondents working at the central administration level and at the local dental practice level using linear regression models with an interaction term between central/local work group and workforce characteristics and work environment.