The ideal panel size is the holy grail of primary care redesign. If we could cap our panels at just the right number, we would provide better care for our patients and have more joy in our work, or so the conventional wisdom goes. But in leading the empanelment effort in our health care system of more than 20 sites and 130 primary care clinicians, we came to realize that panel size alone should not determine if a primary care clinician can accept new patients.
Having done the work of formalizing our empanelment process,1 attributing patients to each clinician, and finding that some panels were larger than others, we were left with an important question: Which panels should be closed to new patients? To determine this, we developed a rubric based on an adjusted panel size, plus each clinician's upcoming openings and productivity. Here's how it works.
Deciding whether a physician's panel should be open or closed depends on more than just determining the number of patients the physician is already seeing.
This rubric accounts for variables like how often the physician sees patients in clinic, how backlogged the physician's schedule is, and how productive the physician is.
The rubric categorizes physicians' panels as “low,” “medium,” or “high,” depending on statistical ranges that can be modified to meet the needs of individual practices.
THREE OBJECTIVE VARIABLES
To begin, we needed to understand the consequences of a panel that is the wrong size. If clinicians take care of too many patients, their patients will struggle to get timely follow-up appointments and quality of care will suffer.2 If clinicians cap their panel sizes too low, the primary care workforce shortage will be exacerbated and a larger percentage of the population will not have a primary care clinician.3 Furthermore, clinicians with panel sizes that are too small will not fill their schedules.
But just knowing the raw number of patients in a panel without context is not enough to determine whether that panel is the right size. Every group of patients has different needs and every clinician works at a different pace. 4 A clinician who cares for thousands of patients but ends each day with unused appointment slots can take new patients. Conversely, a clinician who has a panel of only a few hundred patients but no available appointments for three months should not accept new patients.
We were tempted to adjust panel size for the age, gender, or morbidity of patients, but we concluded these calculations would not help determine if a specific panel size is the right number for a certain clinician. The purpose of adjusting panel sizes by those measures is to promote fairness – e.g., my panel size might be smaller than yours, but my patients are sicker. However, one patient panel might skew older, while another has more comorbidities, a third generates more patient portal messages and telephone calls, a fourth has higher rates of addiction, and so on. Any adjustment for one factor can undervalue the others.
Instead, we focused on three objective variables:
Panel size adjusted for number of clinic sessions per week. The panel size of a clinician who is in clinic one day per week should be interpreted differently from that of a clinician who is in clinic five days per week. So we took the raw number of patients in each clinician's panel and adjusted it based on the number of half-day sessions per week that the clinician is in clinic seeing the clinician's own patients.
Days until third next available appointment. The number of days until a clinician's third next available appointment is a measurement of each clinician's backlog.5 We use the third next appointment because the first or second may have become available because of last-minute cancellations and would overstate a clinician's availability.
Specialty-specific relative value unit (RVU) productivity percentile. To come up with relevant specialty-specific RVU productivity percentiles, our system uses weighted averages of national surveys from two sources: the Medical Group Management Association and the Association of American Medical Colleges.6,7
In the rubric we developed, we slotted each clinician's numbers into “low,” “medium,” or “high” categories for each of the three variables (see “Opening and closing panels”).
For adjusted panel size, 1,500 patients or fewer is low, 1,501 to 1,899 is medium, and 1,900 or greater is high.
For days until third next available appointment, 20 days or less is low, 21 to 29 is medium, and 30 or more is high.
For productivity percentile, less than 50th percentile is low, 50th to 60th percentile is medium, and greater than 60th percentile is high.
If two or three variables land in one category (low, medium, or high), we designate the clinician's panel as that category overall. If one variable falls into each category, we designate the clinician's panel as medium overall.
If a panel is designated low overall, the clinician can accept two new patients per half-day clinic session. If it's medium overall, the clinician can accept one new patient per session. If it's high overall, the panel is closed to new patients. The three variables (adjusted panel size, appointment availability, and productivity) change over time, so we recalculate average scores and reevaluate whether to open or close panels every six months. But practices could choose to reevaluate more or less frequently, depending on their resources and preferences.
OPENING AND CLOSING PANELS
Practice leaders who use this rubric to determine panel status may adjust variables and ranges to their specific environment.
|Adjusted panel size*||≤1500||1501–1899||≥1900|
|Third next available appointment||≤20 Days||21–29 Days||≥30 Days|
|Productivity percentile||<50th percentile||50th–60th percentile||>60th percentile|
|Two or more in the “Low” category||Panel is OPEN to two new patients per half-day clinic session|
|Two or more in the “Medium” category||Panel is OPEN to one new patient per half-day clinic session|
|One in each category||Panel is OPEN to one new patient per half-day clinic session|
|Two or more in the “High” category||Panel is CLOSED: No new patients per half-day clinic session|
Other practices may also choose to use different ranges for the three variables to decide what qualifies as low, medium, or high. The panel size cutoffs we chose were specific to the distribution of patients in our system. Using those ranges, approximately one-third of our clinicians were slotted into the low panel size category, one-third into the medium category, and one-third into the high category. The access and RVU cutoffs we chose were similarly specific to our system goals: Clinicians with more than a 30-day wait for an appointment will be more likely to have a closed panel and should not accept new patients if we want their access to improve. Measuring productivity and RVUs understandably elicits negative reactions from most clinicians. But we hope that by including these factors in this equation, we can better align revenue generation, complexity documentation, and empanelment.
Our model has several limitations. First, compiling the data and communicating the changes to more than 130 clinicians takes several days of work. That's one reason we reevaluate panel status every six months, instead of more frequently. Each adjustment requires us to communicate to a wide array of stakeholders (marketing, call center, specialists, clinical staff, etc.) about who is accepting new patients. For us, changing the panel status more than twice a year risks creating confusion.
Third, practices in rural or underserved areas, or that are not part of a larger system, might not be able to close their panels to new patients — especially if those patients have no other option in the area. In those cases, as well, optimizing performance and staffing are likely the best available solutions.
How will you know if our rubric or a modified version is working for your system? Existing patients should have better access to appointments with their own primary care clinicians, while new patients should have the opportunity to schedule with a primary care clinician who has time for additional patients. Clinicians should feel the process is fair, transparent, and accommodates different practice styles. Primary care teams should be able to deliver high-quality care to their entire panel of patients.
The quest for the ideal panel size is motivated by a desire to deliver better care. Burnout rates are high, and too few physicians choose careers in primary care. Finding the ideal panel size might improve the care of our patients and bring us more joy in practice. But panel size is just a number, and we believe that the ideal number is unique to each clinician, primary care team, and group of patients. No amount of risk adjustment will bestow greater meaning to the one number, but system leaders can use these three variables or other relevant measures to help deliver better access to primary care.