Presence of Primary Care Physicians and Patients’ Ability to Register: A Simulated-Patient Survey in the Paris Region

Article Figures & Data

Table 1.

Comparison of PCP Presence and Willingness to Register Patients for Visits (N = 5,188 Census Blocks)

Indicator of Local PCP Supply	Present	Registering Patients for Office Visits	Registering Patients for Home Visits	P Value^{^a}
Block has ≥1 PCP of this type, No. (%)	2,905 (55.4)	2,004 (38.6)	1,008 (19.4)	<.001^{^b}
Number of PCPs of this type per block per 10,000 inhabitants, mean (SD)	7.59 (18.4)	3.32 (6.10)	1.31 (9.20)	<.001^{^c}

PCP = primary care physician.
Notes: The binary indicators (block has vs does not have at least 1 PCP of given type) reflect the minimum ability of patients to find a physician; the continuous indicators (numbers of PCPs of given type per number of inhabitants) reflect patients’ possible ability to choose among several physicians. See Methods for details.
↵a Indicators of local PCP supply were compared by using empty mixed models (ie, those without any independent variables) with a random intercept and 2 levels (block level and municipality level, to take into account possible municipal policies related to primary care supply).
↵b The 3 binary indicators were compared with a log-binomial model.
↵c The 3 continuous indicators were compared with a Poisson model containing as an offset the logarithm of the number of inhabitants in the block, to take physician density into account.

Table 2.

Inequalities in PCP Supply by Population Density and by Deprivation Level, and Variation of Inequalities Across Indicators (N = 5,188 Census Blocks)

Indicator of Local PCP Supply	Tertile of Municipal Population Density	% (No.) of Blocks	RR (95% CI)^{^a}	P Value	Tertile of Social Deprivation Index	% (No.) of Blocks	RR (95% CI)^{^a}	P Value
Blocks has ≥1 PCP …
… present	Low Medium High	41.6 (715) 56.8 (987) 69.6 (1,202)	ref 1.36 (1.27-1.46) 1.67 (1.56-1.80)	<.001	High Medium Low	49.7 (850) 50.0 (854) 68.2 (1,200)	ref 1.01 (0.94-1.08) 1.33 (1.25-1.41)	<.001
… registering patients for office visits	Low Medium High	26.0 (450) 38.0 (663) 51.6 (892)	ref 1.46 (1.31-1.62) 1.98 (1.79-2.18)	<.001	High Medium Low	32.2 (550) 33.1 (565) 50.5 (889)	ref 1.03 (0.94-1.14) 1.57 (1.42-1.74)	<.001
… registering patients for home visits	Low Medium High	14.1 (244) 17.9 (311) 26.2 (453)	ref 1.26 (1.06-1.50) 1.84 (1.58-2.14)	<.001	High Medium Low	13.5 (230) 16.2 (277) 28.5 (501)	ref 1.20 (1.02-1.42) 2.12 (1.81-2.48)	<.001

Number of PCPs … per block per 10,000 inhabitants	Tertile of Municipal Population Density	Mean (SD)	IRR (95% CI)^{^b}	P Value	Tertile of Social Deprivation Index	Mean (SD)	IRR (95% CI)^{^b}	P Value
… present	Low Medium High	5.45 (20.6) 7.05 (10.7) 10.3 (21.6)	ref 1.06 (0.95-1.18) 1.29 (1.13-1.48)	.001	High Medium Low	6.71 (26.2) 5.90 (11.0) 10.1 (14.4)	ref 1.13 (0.95-1.33) 1.47 (1.28-1.68)	<.001
… registering patients for office visits	Low Medium High	1.97 (5.11) 3.06 (5.93) 4.45 (6.89)	ref 1.19 (1.03-1.38) 1.67 (1.44-1.93)	<.001	High Medium Low	2.42 (5.34) 2.24 (5.70) 4.50 (6.92)	ref 1.02 (0.89-1.18) 1.60 (1.40-1.82)	<.001
… registering patients for home visits	Low Medium High	0.93 (3.33) 1.12 (3.24) 1.91 (15.2)	ref 0.88 (0.70-1.11) 1.28 (1.02-1.60)	.02	High Medium Low	1.18 (15.2) 1.00 (3.37) 1.78 (3.76)	ref 1.13 (0.83-1.54) 1.82 (1.31-2.54)	<.001

IRR = incidence rate ratio; PCP = primary care physician; ref = reference group; RR = relative risk.
Notes: The least dense and the most deprived tertiles are the reference groups. For example, patients in the least deprived vs most deprived tertile were 33% more likely to have access to at least 1 PCP (RR = 1.33), 57% more likely to have access to at least 1 PCP willing to register a new patient for office visits (RR = 1.57), and 112% more likely to have access to at least 1 PCP willing to register a new patient for ongoing home visits (RR = 2.12). All interactions of indicators with tertiles were statistically significant (P <.001).
↵a RRs were calculated by a univariate log binomial mixed model with a random intercept and 2 levels (block level and municipality level, to take into account possible municipal policies related to the primary care supply).
↵b IRRs were calculated by a univariate Poisson mixed model with a random intercept and 2 levels (block level and a municipality level, to take into account possible municipal policies related to the primary care supply). The offset is the number of inhabitants per census block.