Are Direct Primary Care Practices Located in Health Professional Shortage Areas?

DISCUSSION

In this analysis, we found that DPC practices were less likely to be in HPSAs overall and less likely to be in a high-priority–need HPSA but more likely to be in a rural or partially rural HPSA compared to NPI-identified PCPs in the United States. We offer several observations on these findings. Optimistically, there is evidence that DPC practices are located in nonurban areas, where there is a need for primary care physicians, yet overall, DPC practices are not as geographically dispersed as PCPs across the United States with respect to HPSAs. However, geographic availability might not equate to accessibility, which is driven by multiple factors such as cost, coverage, distance, patient volume, patient comfort, etc. We also recognize that DPCs in a high-priority–need HPSA might in fact not serve the target population if the DPC draws from (the relatively small number of) people who already have access to primary health care in that area or the cost of the DPC practice precludes those without primary health care from participating.

We did not include concierge medicine practices in this analysis because they tend to skew toward higher-income households—22% of adults in the top 1% income bracket participate in a concierge practice⁹—with average fees more than double those of DPC.¹⁰ Their geographic distribution across the United States might mirror that of PCPs in general. A survey found that approximately 33% of concierge medicine offices operated in an urban area, 42% in a suburban area, and 25% in a rural area.¹¹ We are unaware of equivalent data regarding the income distribution of DPC vs non-DPC patients; this is an important area for follow-up research because it might relate to access to primary care.

There are several potential limitations to our analysis. First, our data identified DPC practices and not individual DPC physicians; thus, the actual number of physicians is likely greater unless all DPC practices comprise solo practitioners. Second, the volunteer nature of DPC Frontier Mapper might result in an underestimation of practices; this might be especially true for DPC practices that marketed to businesses as opposed to individuals. The Direct Primary Care Coalition, an advocacy organization for DPC, estimated there to be 1,600 practices in the United States,¹² suggesting that DPC Frontier Mapper might be a more complete data source. It might also be possible that DPC Frontier Mapper practice addresses do not align with patient catchment areas, for example, if there are multiple practice locations. Third, the self-reported nature of the NPI taxonomy could result in a misclassification of primary care physicians. If misclassification of either data source is differential by HPSA characteristic, our results might be biased by the degree of misclassification. However, others have extolled the use of publicly available data on NPIs for identification of PCPs,¹³ and a validation study comparing the taxonomy to Medicare charges reported a sensitivity of 0.9 (95% CI, 0.8-1.0) and a specificity of 0.8 (95% CI, 0.7-0.8), suggesting high agreement.¹⁴ Finally, patient panels tend to be smaller at DPC (and concierge) practices;¹⁵ thus, to balance out demand should practices transition from a traditional fee-for-service model would require additional DPC physicians.

In conclusion, there remains ample opportunity to grow DPC presence in many HPSAs, especially high-priority HPSAs in urban areas. As others have stated, there is a need to show that expansion of DPC practices into HPSAs will not further marginalize groups that have historically lacked access to primary care.¹⁶