Can Using an Intensive Management Program Improve Primary Care Staff Experiences With Caring for High-Risk Patients?
Background: Complex, high-risk patients present challenges for primary care staff. Intensive outpatient management teams aim to serve as a resource for usual primary care to improve care for high-risk patients without adding burden to the primary care staff. Whether such assistance can influence the primary care staff experiences is unknown. The objective of this study was to examine improvement in job satisfaction and intent to stay for primary care staff at the US Department of Veterans Affairs (VA) who sought assistance from an intensive management program.
Methods: Longitudinal analysis of a staff cohort that completed 2 cross-sectional surveys 18 months apart, controlling for outcomes at time 1. Participants included 144 primary care providers at 5 geographically diverse VA health care systems who completed both surveys. Measured outcomes included job satisfaction and intent to stay within primary care at the VA (measured at time 2). Predictors included likelihood of using intensive management teams (measured at time 1). Covariates included outcomes and professional/practice characteristics (measured at time 1).
Results: The response rate for primary care staff that completed both surveys was 21%. Staff who indicated at time 1 that they were more likely to use intensive management teams for high-risk patients reported significantly higher satisfaction and intention to stay at VA primary care at time 2 (both P < .05).
Conclusions: A VA primary care workforce might benefit from assistance from intensive management teams for high-risk patients. Additional work is needed to understand the mechanisms by which primary care staff benefit and how to optimize them.
Data are from PCPs and nurses who completed 2 cross-sectional surveys (online or hard copy). We invited 1,000 PCPs and nurses to complete the first survey (fielded December 2014 to May 2015) and 863 to complete the second survey (fielded October 2016 to January 2017). A total of 436 completed the first survey for a response rate of 44%, and 313 completed the second survey for a response rate of 36%. We constructed a longitudinal cohort of 144 PCPs and nurses who completed both surveys and had data at 2 timepoints. This longitudinal cohort represents 33% of the 442 unique respondents who completed either the first or second survey. Overlap across surveys was low because of high staff turnover between survey waves.
Measures
Outcomes. We examined 2 single-item outcome measures to assess job satisfaction and retention (ie, intent to stay in primary care at the VA) measured in both surveys. These items were worded “Overall, I am satisfied with my job.” and “I intend to continue working in primary care at the VA for the next two years.” Both items were rated on a 5-point Likert scale.
Independent Variable. We assessed proclivity to seek assistance in caring for high-risk patients based on PCPs or nurses indicating that they are likely to either “manage these patients with ongoing care coordination assistance from an intensive management team” and/or “transfer these patients from primary care to another intensive management team or program specializing in high-risk patients.” These 2 items were rated on a 5-point Likert scale; we dichotomized the scale with likely or very likely indicating high proclivity (likelihood) for ease of interpretation of the combined items.
Covariates. We also controlled for indicators of staff demographic and practice characteristics in multivariate analyses. These included gender, staff type (PCP vs nurse), years practicing at a VA clinic, team staffing level (full vs partial), proportion of the panel consisting of high-risk patients (using binary indicators: 11 to 20% or > 20% compared with 0 to 10% as the reference group), and whether or not the site participated in the pilot program offering an intensive management team to support primary care for high-risk patients to distinguish the 8 pilot sites from nonpilot sites.
Statistical Analysis
We used ordinary least squares regression analysis to examine associations between the independent variable measured at time 1 and outcomes measured at time 2, controlling for time 1 outcomes among staff who completed both surveys (eg, the longitudinal cohort). We adjusted for time 1 covariates and clustering of staff within clinics, assuming a random effect with robust standard errors, and conducted multiple imputations for item-level missing data. Poststratification weights were used to adjust for survey nonresponse by staff type, gender, facilities participating in the innovations, and type of specialty PACT. We calculated weights based on the sampling frame of all PCPs and nurses for each survey.
Results
Table 1 shows the proportion of primary care staff responding to the surveys. For the longitudinal cohort, the response by staff type was similar to the sample of staff that responded only to a single survey, but the sample that did not respond to either survey included more physicians. There was also some variation by medical center. For example, a smaller proportion of the cohort was from site D and more was from site E compared with the other samples. The proportion of primary care staff in facilities that participated in the intensive management pilot was higher than the proportion in other facilities. More women (81.9%) were in the longitudinal cohort compared with 77.4% in the single-survey sample and 69.2% in the sample that responded to neither survey.
Both surveys were completed by 144 respondents while 442 completed 1 survey and 645 did not respond to either survey. The cohort was predominantly nurses (64.6%); Of the PCPs, 25% were physicians. Most staff were women (81.9%) and aged > 45 years (72.2%). Staff had practiced at their current VA clinics for a mean of 7.4 years, and most reported being on a fully-staffed primary care team (70%).
