Study setting and design
HealthyHearts New York City (HHNYC) is a partnership between the New York University School of Medicine; the Primary Care Information Project (PCIP), a bureau of the New York City Department of Health and Mental Hygiene. The study was a stepped-wedge, cluster randomized controlled design in which sites were randomized into one of four 12-month intervention waves to evaluate the effect of practice facilitation (implementation strategy) to help small independent practices adopt CVD guidelines (intervention). A new wave of practices was enrolled every 3 months. The protocol for the PF implementation strategy consisted of a minimum of 13 in-person, on-site PF visits to an assigned panel of practices within a one-year intervention period to assist practices with their implementation of system changes consistent with the patient-centered medical home and the Chronic Care Model (CCM)[18,19]. For example, in alignment with the CCM’s “delivery system redesign” component, facilitators promoted the use of patient registries to identify high-risk patients. PFs scheduled two visits to their assigned practices in the first month and monthly visits for the remaining 11 months. For the cost analysis, we included the 261 small independent practices from the PCIP network that were enrolled during the first year of the study intervention phase (12/1/15-11/30/16). Sixteen practice facilitators were assigned to work across the practices.
Nearly two-thirds of the 261 practices were solo clinicians (n=149), 37% were in accountable care organizations[20], almost half (47%) were in Medically Underserved Areas [21] and/or designated as patient-centered medical homes, and 45% of patients had Medicaid.
Data sources
There were several data sources for the cost analysis, comprising categories for start-up costs, implementation and intervention costs, and practice staff costs.
Start-up costs. The program manager tracked all start-up costs in a database that included materials, supplies, toolkits for the facilitators and for each practice to implement the CCM and evidence-based guidelines for the ABCS measures, computers for use by the facilitators, webinar hosting, and program staff time and expenses for kickoff events, training, and meetings prior to program launch, based on invoices from and communication with PCIP. Total start-up costs included all of the above-mentioned costs, but excluded time costs for practices, because while we documented the number of staff and/or clinicians present, we did not know which specific staff members attended.
Implementation and intervention costs. Facilitators used an electronic customer relationship management (CRM) system (i.e., Salesforce™) to document the amount of time spent at each visit and other interactions including email and phone calls. The facilitator costs also included other meetings related to the project that did not involve site contact, such as supervision and project management meetings with other facilitators and project or program staff. Facilitator travel time was incorporated as a regular part of their workday and not calculated separately. Further, research costs were not included, because they would not be part of a future implementation.
In addition to facilitators, we accounted for the time of facilitator managers and several other program staff (e.g., outreach coordinators). Data for salaries were obtained either from published salary ranges from the New York City Department of Health and Mental Hygiene or estimated from the 2016 data of the U.S. Bureau of Labor Statistics, where we assigned a dollar value for each hour of staff time based on the national wage and fringe rate of those who could perform the activity in the same setting. We estimated the staff time cost with a lower and upper bound based on the given New York City Department of Health and Mental Hygiene salary range (e.g., $74,700-85,700/year for a practice facilitator including fringe) or from possible occupation types from the U.S. Bureau of Labor Statistics. Several occupation types could fulfill the duties of different program staff, such as nurse practitioner ($49.54/hour including fringe), registered nurse ($31.68/hour including fringe), or medical assistant ($15.76/hour including fringe), and all have different wage levels [22].
Practice staff costs. The HHNYC practice member survey was used to estimate the amount of time physicians and their practice staff spent on activities that were implemented as part of the intervention. (Questions were essentially, “Does your practice site conduct [daily huddles or outreach to high-risk patients]?” with a skip pattern allowing for recall of number of minutes per most recent full day or week spent on the activity.) The surveys were completed by all providers and staff at 3 times (baseline, post-intervention, and 6-month follow-up). We used 2016 data from the U.S. Bureau of Labor Statistics to estimate hourly wages including fringe for practice staff time involved in the program [22]. Examples of personnel included medical biller, financial controller, front desk staff, general and operations manager, medical assistant, medical secretary, nurse practitioner, office manager, physician assistant, physician, registered nurse, and secretary/administrative assistant. We used the national average wage as the upper bound on time costs and 80% of that as a lower bound.
Analytic approach
Our analyses distinguished between start-up costs (e.g., kick-off events, program staff time), program staff costs (i.e., facilitator and associated staff time), and practice costs (e.g., provider and practice staff)[23–25]. These categories allowed for estimating the perspectives of both the entity that must decide to pay for the initial program and of the practice. We estimated the total one-year costs to operate the program, and calculated the mean and range of the cost-per-practice by quarter of the intervention, knowing that facilitators increased their volume of practices as the year progressed (i.e., started at less than full capacity). That is, Wave 1, quarter 1, had the lowest number of practices per facilitator, and this increased through the year. By estimating the cost-per-practice, we indicate how program efficiency over time might change the cost-per-practice over one year duration, as facilitators operated at closer to full capacity. We estimated the lower and upper bounds for all salary expenses, rounding to the nearest $100.