JFOA Journal

Introduction

The Center for Medicare & Medicaid Innovation (CMMI) awarded Connecticut a State Innovation Model (SIM) grant in 2014 to support healthcare payment and delivery model reforms aimed at bolstering the role of primary care to improve health system performance, increase quality of care, and decrease costs. Many of the CMMI initiatives emphasized furthering team-based, patientcentered care models.

In recent years, there has been increasing emphasis on the importance of patient-centered care, or care that is responsive to individual patient preferences and needs [1]. In addition to being an important indicator of patient-centered care, positive patient experiences are positively associated with patient safety and clinical effectiveness across a wide range of disease areas, settings, outcome measures, and study designs [2,3]. Furthermore, the Centers for Medicare & Medicaid Services (CMS) and other payers have incorporated patient experience measures into performance-based incentive systems [1].

Team-based hospital care has been shown to be positively associated with better patient experiences, but less is known about how care team composition impacts patient experiences in outpatient settings, particularly in primary care [4]. Despite a policy on implementing primary care models that incorporate nonphysician staff and the emergence of patient experience surveys as an important quality measure, there have been inconsistent results on the association between these primary care models and patient experiences of care [5-7]. In primary care settings, patients may have longstanding, trusted relationships with their physicians [8] and some have suggested that using clinical staff for certain functions may disrupt the relationship between a patient and his or her primary care provider (PCP) by splintering care across multiple providers [9,10]. Furthermore, research on patient experience has focused on patient predictors of patient experience (e.g., patient age, health, or race) rather than organizational factors (e.g., practice size, specialty, or hospital affiliations) [11-14]. This has led to a lack of information about whether organizational commitment to and investment in the mix of providers meaningfully affects patient experiences in those settings.

In this study, we use data from a SIM project in Connecticut to evaluate the association between an indicator of team composition, the number of non-physician staff and the number of PCPs, and patient care experiences in primary care practices. We also evaluate whether the relationship between care team composition and patient experiences differs by patients’ reported health status. Our hypothesis was that there would be a greater benefit of additional staff for care among sicker patients.

Methods

Data Sources

The data for these analyses were collected as part of a Connecticut SIM project. Data were collected using two surveys: a Primary

Care Organization Survey and a Clinician and Group Consumer Assessment of Healthcare Providers and Systems (CG-CAHPS) survey.

The 40-question Primary Care Organization Survey evaluated four aspects of primary care organizations: 1) organizational characteristics and governance; 2) health information technology; 3) quality improvement; and 4) clinical care delivery. Primary care practices in Connecticut that were engaged in the SIM transformation project and participating in one or more value-based contracts with commercial, Medicare and/or Medicaid payers were involved in the SIM project. Twenty primary care organizations with commercial contracts and 17 Federally Qualified Health Centers (FQHCs), which predominantly serve Medicaid patients, were invited to participate. Individuals who completed the survey on behalf of their organizations were Chief Executive Officers, Chief Operating Officers, or Chief Quality Officers. Data were collected via the online survey platform, Qualtrics, from October 2019 to January 2020. Practices were reminded by phone calls to complete the survey in December 2019. Eleven primary care organizations with commercial contracts and 12 FQHCs responded to the survey, totalling 23 respondent organizations (62% response rate). A total of N=19 organizations had data on all variables of interest and were included in the analyses.

The CG-CAHPS survey asked patients about care experiences, augmented with questions about aspects of care most salient to PCMH-designated organizations. The surveys were administered to a probability sample of primary care patients in CT. The sample was a stratified cluster sample with clustering of patients by primary care organization. A sample of Medicaid patients and patients covered by three large commercial insurance plans was drawn from each participating primary care organization in the state. Approximately equal numbers of patients (unless constrained by total number of patients) were selected from each primary care organization. Those with Medicaid were surveyed by telephone from July 2019 to October 2019. Those with commercial insurance were surveyed by mail from November 2019 to January 2020. The numbers of patient respondents from organizations that participated in the organizational survey were: Medicaid: 3,087 (out of 87,160 called); Commercial: 3,675 (out of 28,852 mailed). Only respondents with complete survey responses in the N=19 organizations of interest were included in this analysis, leading to a final sample of 2,982 Medicaid and 3,125 Commercial patients (total patients N=6107).

Measures

We calculated the number of clinical support staff and PCPs (employed and affiliated) within an organization using responses to two questions from the Primary Care Organization Survey. The questions that were used to construct this measure were:

• … indicate the number of providers employed and affiliated within your network who are predominantly dedicated to providing primary care. (MD / DO)
• … estimate the unduplicated number of staff members by professional category dedicated to primary care. …

(Categories of professionals included: Employed & affiliated PAs, employed & affiliated APRNs, Care Management/Coordination Registered Nurses (RNs) & Licensed Practical Nurses (LPNs), Care Management/Coordination Social Workers or similar, Licensed Practical Nurses (Clinical), Registered Nurses (Clinical), Patient Navigators, Chronic Disease Educator (e.g., Asthma, Diabetes))

For the main analyses we defined primary care practitioners (PCPS) as MDs, DOs, PAs, and APRNs. For the measure of clinical staff, we included: Care Management/Coordination RNs & LPNs, Care Management/Coordination Social Workers or similar, LPNs (Clinical), RNs (Clinical), Patient Navigators, and Chronic Disease Educators (e.g., Asthma, Diabetes). In sensitivity analyses we used the number of only MDs and DOs for the measure of PCPs and included PAs and APRNs in the measure of staff (Total Staff).

The dependent variable was the quality of patient experiences, assessed at the patient level using responses from the 2019 wave of the SIM CG-CAHPS survey. We used four summary measures of patient experiences: timeliness of care, communication, coordination, and staff courtesy [15]. Each summary measure was the average of valid responses to questions in that dimension. Each measure was calculated on a 0-100 scale, with 100 representing the most positive experience.

Covariates also included organizational-level variables. Organizational type was included, as respondent organizations included both organizations which predominantly served commercially insured patients, and Federally Qualified Health Centers (FQHCs), which predominantly served Medicaid patients. Given inherent differences in patient mix, organizational size, and access to specialty providers that might differ by hospital affiliation status, commercial organizations were divided into two groups: no hospital affiliation and hospital affiliation. None of the FQHC organizations had a hospital affiliation (Table 1).

Analyses

Descriptive statistics were used to describe patient and organizational characteristics, as well as overall CG-CAHPS results. Multi-level linear regression models were used to assess the association between care team composition (number of clinical support staff and number of PCPs) and patient experiences, controlling for potentially confounding variables. Patient-level covariates included patient sex, age, patient physical health status, patient mental health status, patient race (white vs non-white), and education, all of which have been found in prior research to be associated with patient experience [15-19].

Multi-level models treated patients as clustered within primary care organizations. Models were first estimated with just main effects for care team composition and then an interaction between number of staff and health status was included to investigate whether the care of patients in poor health was affected more by number of staff.

Results

The mean number of MD/DOs per organization was 98.3 (median 40.0) and the mean number of total staff was 72.9 (median 40) (Table 1). The average ratio of total staff to MD/DOs was 1.8 (median 1.1). There were no practices with zero total staff.

Table 1: Organizational Characteristics; N=19.

The majority of patients (Table 2) were female (62.8%) and under 65 years of age (88.7%). The majority of patients were white (63.8%) and had less than a college education (64.6%). Most patients described their physical health as “Very Good” (34.1%) or “Good” (35.7%). Patients generally reported positive healthcare experiences. The mean CG-CAHPs score was above 85 (on a 0-100 scale) for all four of the outcome scales (Table 3). “Timeliness of care” had the lowest mean (87.1), and “Communication” had the highest (93.9).

Table 2: Characteristics of Patients; N=6107.

Table 3: CG-CAHPS Scores.

Multi-level linear regression models of the relationship between care team composition and patient experience (Table 4) indicated that having more clinical staff was significantly associated with higher patient experience scores for: communication (coefficient = 0.02, p=0.02), coordination (coefficient = 0.03, p=0.005), and staff courtesy (coefficient = 0.02, p=0.01). That is, an increase in the number of staff by 100 (about 2 SD) is associated with an increase in communication and staff courtesy scores by 2 points and an increase in coordination by 3 points. The association of staff with timeliness of care was not significant (p=0.12).

Table 4: Predictors of CAHPS Scores.

Compared to FQHCs, organizations with a hospital affiliation that predominantly served commercially insured patients were associated with poorer experience scores for coordination. Experience scores also differed by hospital affiliation status for staff courtesy, with patients of commercial primary care groups that were hospital affiliated reporting better experiences than those with no hospital affiliation. Worse patient health was associated with worse experience across all four models, consistent with earlier findings [15]. “Fair” patient mental health (rather than “Excellent”) was also negatively associated with worse experiences for all four of the outcomes modeled.

Models with an interactive term between patient health and number of clinical staff showed that the overall interaction between health and clinical staff was not significant for timeliness of care (p=0.12), although the number of clinical staff was more important for timeliness of care for patients in poor health than for those in excellent health (p=0.01) (Table 5). The overall interaction term was also not statistically significant for the other three outcomes (communication (p=0.64), coordination (p=0.61), staff courtesy (p=0.46)).

Table 5: Multi-Level Regression Results for Models with Interactions.

When we re-estimated the models using a PCP variable composed of just MDs and DOs and a separate variable representing all staff (total staff), the results were comparable (See Supplemental Tables S1 and S2).

Supplemental Table S1: Predictors of CAHPS Scores (using number of total staff and number MD/DO).

Supplemental Table S2: Multi-Level Regression Results for Models with Interactions (using number of total staff and number MD/ DO).