INTRODUCTION

Patients benefit from guideline-based preventive, chronic disease, and acute care,1, 2 but many do not receive it.3 A limitation to providing high-quality care is insufficient time.4, 5 In 2003, Yarnall et al. estimated it would take 7.4 h/day for a primary care provider (PCP) to provide preventive care for a 2500 patient panel.6 A study by Privett et al. found it would take 8.6 h/day.7 A complementary study in 2005 by Østbye et al. calculated 10.6 h/day were needed for a PCP to manage the top ten chronic diseases.8 These studies suggested at least 18 h/per day are needed for a PCP to provide preventive and chronic care management, not accounting for other tasks.

Since 2005, there have been changes in the structure of primary care. Traditionally, PCPs cared for patients primarily by themselves. However, new team-based care models, such as the Comprehensive Primary Care Initiative Plus (CPC+), increasingly involve other care providers in providing patient care.9–11 Many healthcare organizations have encouraged this shift; for example, a National Academy of Medicine report urged new payment models to “pay for primary care teams to care for people, not doctors to deliver services.”12 These models may save PCP time by shifting tasks traditionally performed by the PCP to other members of the care team. Theoretically, the time savings may allow PCPs to focus on more advanced care, see more patients, or increase the delivery of guideline-based care. Because of the potential for team-based care to change primary care practice, we re-investigated the amount of time needed to provide preventive, chronic disease, and acute care without and with team-based care.

METHODS

Patient Panels and Study Design

Two recent surveys of family practice physicians calculated their average patient panel sizes as approximately 230013 and 290014 patients, respectively. A recent meta-analysis determining the optimal panel size was inconclusive, but panels ranged from 1200 to 3600.15 Given the uncertainty in the literature, we modeled patient panels of 2500 patients, and varied this assumption in sensitivity analysis. We created 1000 hypothetical panels of 2500 patients from the 5856 adult participants in the National Health and Nutrition Examination Survey (NHANES) 2017–2018 cohort.16 The chances of inclusion in the panels were proportional to each participant’s sampling weight,17 and each individual selected from the survey to include in a panel was replaced into the survey prior to selecting the next patient. For each panel, we quantified the time needed to address preventive, chronic disease, and acute care assuming the PCP worked alone and with a multidisciplinary team (Fig. 1).

Fig. 1
figure 1

Schematic displaying the creation of patient panels from an NHANES cohort, and the calculation of annual PCP preventive and chronic disease care time

Full size image

Primary Care Services

Preventive Care

We included the Grade A and B preventive care services and immunizations recommended by the United States Preventive Services Task Force (USPSTF) and Advisory Committee on Immunization Practices (ACIP), respectively, in 2020 (Appendix Table 1).14, 15 For each preventive care service and immunization, we determined each NHANES participant’s eligibility based on their age, gender, and presence of chronic conditions/comorbidities. Information to determine eligibility for all services were publicly available in NHANES 2017–2018, except for sexual history, for which we substituted the distribution of responses from the publicly available NHANES 2015–2016 data. For each NHANES participant, we calculated the annual time needed to provide preventive care based on the annual time required for each service they were eligible for (Fig. 1). If the PCP could provide the service herself, that time was calculated. If the service required a referral, the time required for the PCP to coordinate the referral was calculated. These times were summed for each participant to calculate the annual time needed to provide preventive care for each hypothetical panel of 2500 patients.

The total annual time for each service was calculated based on USPSTF guidelines, their citations, and a literature search (Appendix Table 2) describing their annual visit frequency and time per visit. For the literature search, we prioritized in order meta-analyses, clinical trials, review articles, and cohort or case studies. If we found no evidence, we assumed the service would take the same amount of time as another similar service (Appendix Table 2). If a service required screening and follow-up care, the time for both components was calculated independently. If screening resulted in the diagnosis of a chronic disease, we assumed that time would be part of chronic disease care (see below).

Because of heterogeneity in data availability for time estimates, we classified the quality of evidence as “strong,” “moderate,” or “weak,” depending on the evidence source (Table 1). Data from a meta-analysis or randomized control trial was considered “strong”; evidence from cohort or case studies was considered “moderate.” Weak evidence was based on expert opinion or author estimates. If estimates from a similar service were used, we downgraded the strength of evidence by one level.

Table 1 Estimated Time Needed to Provide Guideline-Based Preventive and Chronic Disease Care for an Average US 2500 Adult Patient Panel
Full size table

Chronic Disease Care

We selected the ten most prevalent chronic diseases based on published data from the 2014 Medical Expenditure Panel Survey (Table 1).18 We defined each chronic disease using NHANES data (Appendix Table 3), distinguishing between “controlled” and “uncontrolled” disease based on lab or physical exam data (such as a HbA1c >7 for diabetes). Then, for each participant in NHANES, we assessed whether they had each chronic disease, and whether it was “controlled” or “uncontrolled.” The annual time needed to address each chronic disease for each patient was based on the annual visit frequency, which varied based on whether the disease was controlled or uncontrolled, and the time per visit. These times were summed for each participant to calculate the annual time needed to provide chronic disease care for each hypothetical panel.

To estimate the annual visit frequency, we reviewed published guidelines (Appendix Table 4), and collected data on the number of visits required to address controlled vs uncontrolled disease.8, 19–27 If no evidence existed, clinical authors (J.P, C.B, N.L.) independently estimated the annual visits needed, and discussed until consensus. Little data existed quantifying the time needed per visit for each chronic disease. Similar to previous analyses,8 we assumed 10 min per visit to address each chronic disease and tested this assumption in sensitivity analyses.

Acute Care

To account for acute care, we examined nationally representative studies of primary care office visits. These studies found PCPs had approximately 637 visits per 1000 adult patients per year,28 and a mean visit duration of approximately 21 min.29 The time needed to address acute care for each panel was calculated by taking the product of the average number of annual acute care visits per patient, the number of patients per panel, and the mean visit duration.

Documentation and Inbox Management

We estimated this time based on evidence that PCPs spend 3.2 h per day on average on documentation, paperwork, and after visit care.30, 31

Team-Based Care

Preventive Care

We re-estimated the time needed to provide care when dividing time between a PCP and other care team members. We modeled estimates based on the team members in CPC+ because this advanced model allowed us to estimate times for a wide scope of disciplines.9, 32 We determined which preventive care tasks could be done by non-PCP team members by reviewing scope of practice guidelines for each discipline, and calculated the time spent for each team member. We used national guidelines published by governing bodies for each discipline, or Illinois guidelines if national guidelines were not available.33–36 We assumed non-PCP team members performed all services within their scope of practice instead of the PCP.33–36

Chronic Disease and Acute Care

Estimates for the time spent on chronic disease and acute care that could be offset by team-based care models were determined by literature review. For seven of ten chronic diseases, we found evidence non-PCP team members could perform some chronic disease management independently from the PCP,37–43 but little data existed quantifying the time saved. We assumed 50% of PCP time could be saved for chronic disease and acute care, and examined this assumption in sensitivity analyses.

Documentation and Inbox Management

We found evidence medical assistants could screen inboxes for PCPs,44–46 but little data existed quantifying the time saved. We assumed 20% of PCP time could be saved and examined this assumption in sensitivity analysis.

Analysis

For each of the 1000 hypothetical panels of 2500 NHANES participants, the time needed to address preventive care, chronic disease care, acute care, and documentation/inbox management was calculated as outlined above. The means of these data were aggregated, and are displayed in our results.

Sensitivity/Scenario Analysis

We conducted additional analyses to evaluate the robustness of the results. We varied patient panel size between 1500, 2000, and 3000 patients. We also re-estimated the chronic disease visit time, assuming treating multiple concordant chronic conditions would take less time than treating each condition separately.47, 48 We created four clusters of related conditions: “metabolic disorders” (hypertension, hyperlipidemia, diabetes, and coronary artery disease), “psychiatric disorders” (mood disorders and anxiety disorders), “respiratory disorders” (asthma and other upper respiratory disorders), and “musculoskeletal disorders” (osteoarthritis and inflammatory joint disorders). The time needed for each cluster was based on the sum of each patient’s annual visits needed to address each chronic disease within the cluster, assuming that the first three visits would take 10 min each and each additional visit would take 5 min. We created a separate sensitivity analysis assuming chronic disease visits would take 5 min, instead of 10 min. Lastly, we varied the percentage of PCP time for chronic disease and acute care saved by team-based care by ±25% (75% and 25%), and for documentation/inbox management by ±10% (30% and 10%).

RESULTS

For an average 2500-patient panel, a PCP would require 26.7 h/day to provide preventive, chronic, and acute care with documentation/inbox management (Fig. 2). Preventive care alone would require about 14.1 h. Each preventive care task time varied from less than 1 s/day (medication to reduce breast cancer) to 4.1 h/day (obesity counseling) (Table 1). Over 10 h would be spent on counseling tasks, especially dietary or obesity counseling. Chronic disease care alone required about 7.2 h/day. Acute care would require 2.2 h/day, while documentation and inbox management would require 3.2 h/day.30, 31

Fig. 2
figure 2

Primary care provider time needed to provide care for average US adult panel of 2500 patients.

Full size image

With team-based care, the total PCP time decreased to about 9.3 h/day. Non-PCP team members could partially or completely perform 29 preventive care tasks, leaving about 2.0 h of preventive care tasks per day for the PCP. The majority of time reduction was due to 10.8 h of counseling tasks being transferred to dieticians or counselors. Chronic disease care would decrease to 3.6 h/day, acute care to 2.2 h/day, and documentation/inbox management to 2.6 h/day. Overall, 17.4 h could be shifted to other members of the health care team (Figs. 2 and 3).

Fig. 3
figure 3

Additional preventive care time shifted to non-PCP team members in a team-based care model. Abbreviations: RN, registered nurse; LPN, licensed practical nurse; MA, medical assistant.

Full size image

Sensitivity/Scenario Analysis

Using best-case assumptions (a 1500-patient panel, team-based care, 5 min per chronic disease visit, 75% of chronic disease/acute care time, and 30% of documentation time performed by other team members), a PCP would require 4.5 h/day. However, using the worst-case assumptions (a 3000-patient panel, no team-based care), total time increased to 32.0 h/day.

Changing the panel size to 1500, 2000, and 3000 patients resulted in −10.7, −5.3, and +5.3 h change in PCP time without team-based care, and −3.7, −1.9, and +1.9 h change with team-based care (Figs. 4 and 5). Accounting for multimorbidity by “clustering” chronic diseases decreased the PCP time by 0.6 and 1.2 h with and without team-based care, respectively. Assuming a 5-min chronic disease visit time decreased the PCP time by 1.8 h with and 3.6 h without team-based care. Assuming 25% or 75% of chronic disease care could be done by non-PCP team members resulted in a ±1.8 h change in PCP time.

Fig. 4
figure 4

Sensitivity/scenario analyses: change in PCP time with PCP-only care. Abbreviations: PCP, primary care provider.

Full size image
Fig. 5
figure 5

Sensitivity/scenario analyses: percent change in PCP time with team-based care. Abbreviations: PCP, primary care provider.

Full size image

DISCUSSION

Our study found that a PCP would need an infeasible 26.7 h/day to provide preventive, chronic disease, and acute care for an average US adult patient panel. If a PCP worked in a team-based care model, up to 65% of his services could be performed by other healthcare team members. Only under the best-case assumptions could total PCP time fit into a standard 8-h workday.

Our estimates for preventive care time (14.1 h) were higher than estimates by Yarnall et al. (7.4 h) or Privett et al. (8.6 h).6, 7 This difference is largely explained by differences in calculating the time needed for counseling tasks. We based our estimates on USPSTF meta-analyses, which demonstrated that effective counseling interventions are extremely time intensive.49–51 For example, obesity counseling requires “12 or more sessions” per year to provide evidence of benefit,52 with most studies taking more time than our conservative estimate of 72 total minutes per patient per year.50, 52 Privett estimated counseling times more conservatively. Our estimates for chronic disease care (7.2 h) were lower than Østbye et al.’s estimates, as current chronic disease guidelines recommend fewer visits per year.20, 21

Given the large gap between the time required to provide guideline-based care and the limits of a clinic day, many clinicians are likely not completing specific services, not completing them according to the guidelines, or working overtime. If time pressures are driving a gap between guideline-based and clinical medicine, it might explain why national health outcomes are worse than expected.53 It may also drive physician burnout, which is often associated with large panel sizes54 and the subjective feeling of being overworked.4 Low income and racial/ethnic minority status are strongly associated with insufficient access to primary care.55 To the extent that the excessive time required to provide guideline-based care is exacerbating the physician shortage, it may be contributing to unequal access to care. If clinical guidelines do not consider the time opportunity cost of an intervention, the gap between guideline-based and clinical medicine will persist.

Advanced team-based care may reduce the time needed for PCPs to deliver guideline-based care. Counseling tasks are extremely time intensive,49–51, 56 but the majority of PCP time saved occurred by shifting those tasks to other team members. Our findings support a 2012 study which concluded that non-PCPs could complete 77% of preventive care guidelines and 47% of chronic disease care.57

Team-based care models may save PCPs time at the expense of fragmented patient care. However, the efficiency gains from providers working at the top of their license and well-functioning team-based care models may limit this downside. Overall, team-based care models may be associated with increased adherence to quality measures at marginally lower total costs.58

We chose the CPC+ model as it was a functioning care model that incentivized a broad range of team-based care services. Other models in the community (and even individual CPC+ clinics) may not have access to all the team members envisioned by our idealized model. As a result, their ability to save PCP time may be more limited. Conversely, if a team-based care model included a broader range of team-members, such as dentists or physical therapists, more PCP time could be saved. Our estimates of PCP time with and without team-based care represent two polar scenarios for a PCP to provide guideline-based care, and individual results may vary.

Some innovative care models incentivize adoption of team-based care;32, 58 however, fee-for-service healthcare reimbursement does not allow many non-PCP team members to be paid for services.59, 60 Dietician performed dietary counseling is only reimbursed by Medicare if the patient has diabetes, renal disease, or a renal transplant,61 though USPSTF guidelines also recommend dietary counseling for other populations. Even with a transition to value-based care, the debate of whether each potential team member adds sufficient value to be hired will remain.

The challenge of implementing team-based care models is compounded by low PCP compensation. PCPs are reimbursed at lower rates than specialists, are more likely to maintain high-volume clinics,62 and spend large amounts of time doing uncompensated between-visit work.63 In this setting, it would be a challenge for a PCP to handle the costs of hiring new employees without additional payment. Unless new compensation models are implemented, these team-based care models might not be feasible.

Significantly lowering the panel size to less than 1500 patients may allow time to provide guideline-based care, and the emergence of “direct primary care” models may reflect this opportunity.54 However, the membership fees necessary to support these models may preclude their dissemination, and smaller panels would exacerbate the current PCP shortage.

Our study had several limitations. Little evidence existed on the time needed to address each chronic disease at each visit, but even assuming a 5-min visit time or accounting for time saved from clustering concordant chronic diseases did not change our conclusions. Assuming different panel sizes resulted in proportional changes to our results, but even a 1500-patient panel required 16.0 h/day of PCP time. Under every scenario our conclusion was the same: providing ideal guideline-based preventive, chronic disease, and acute care services places an unreasonable time burden on a PCP that is only partially mitigated by team-based care models and smaller panel sizes.

Our calculations represent a lower bound estimate for PCP time. We consistently chose the most conservative time estimate for which there were data, included only ten chronic diseases, limited our preventive care services to USPSTF recommendations, and did not account for breaks in time between patients. We also did not account for inefficiencies associated with combining visits for all the separate preventive care/chronic disease/acute care services into discrete clinic time slots. Additionally, the 2020 COVID-19 pandemic has at least temporarily increased the workload for PCPs, leading to large increases in electronic work which is not accounted for in this analysis.64, 65 Finally, many common patient-specific factors and conditions requiring PCP time were not addressed. Issues such as aging/frailty, language concordance, or homelessness were indirectly addressed insofar as they are associated with a higher risk for chronic diseases, but were not accounted for independently.

Overall, it would take an infeasible 26.7 h per day for a PCP to provide guideline-based care for a 2500 patient panel in 2020. This time could be decreased to 9.3 h per day with team-based care. These findings explain why improvements in the quality of primary care have eluded the USA for the last two decades. Models of primary care that leverage and reimburse appropriately for interdisciplinary teams can only partially rectify the US healthcare system.