Do Physicians Respond to Additional Capitation Payments in Mixed Remuneration Schemes?

Abstract

Mixed remuneration schemes with capitation and fee-for-service (FFS) payments hold financial incentives to add patients to the list and provide services to listed patients. However, as patients with complex needs tend to require longer consultations there is a risk of inequality in access if fees are not adjusted to patient characteristics. In this paper, we assess a natural experiment introducing additional capitation for GPs with a high share of complex patients (moderate scheme) and for GPs in certain geographical areas (intensive scheme). GPs are eligible if the complexity of their listed patients exceeds a threshold, but as the scheme is subject to a national budget constraint, some eligible general practitioners (GPs) are left without additional payment. For the most favored GPs, the reform distributed additional capitation at 8% of the total baseline income. We study the effects on the number of patients per GP and the number of services per patient, applying difference-in-difference (DiD) models. For both schemes (moderate and intensive), we find tendencies of reductions in the number of patients served and the level of service provision per patient. This also holds for complex patients indicating that the reform did not improve equity in access. The effect on income showed a 2.5% increase in the first follow-up year but the effect became insignificant in the second year after the reform. We interpret this result as a sign that GPs trade income increases with leisure as suggested by the target income hypothesis.

Keywords: income target hypothesis; physician incentives; physician remuneration.

FIGURE 1

Illustration of the distribution rule for additional capitation. The x‐axis has one observation per GP clinic sorted by $R_j$, which is the difference between the weighted and the unweighted capitation per full‐time physician—from highest to lowest. The y‐axis shows the value of $R_j$. At some point, the value falls below 0. GPs to the right of this point are not eligible for additional capitation. GPs with $R_j$ above $R^{\max }$ receive only $€13K$. The grayed area represents the total differentiated capitation paid to GPs (€4.8 million). GPs with $R_j$ below $R^{\ast }$ are allocated to the treatment group. GPs with $R_j$ between $R^{\ast }$ and 0 are allocated to the threshold control group. GPs with $R_j$ between $R^{\ast }$ and $R^{\min }$, of which some are non‐eligible, are allocated to a control group that can be used for matching as an alternative to the threshold control group.

FIGURE 2

Flowchart of sample size.

FIGURE 3

Common trend—full pop. Each panel shows the raw outcome values over time for each group (control, moderate treated, and intensive treated).

FIGURE 4

Event study results—threshold. Each panel shows the treatment effects for GPs receiving either moderate or intensive treatment. The vertical lines illustrate 95% confidence intervals.

FIGURE 5

Event study results—matching. Each panel shows the treatment effects for GPs receiving either moderate or intensive treatment. The vertical lines illustrate 95% confidence intervals.

FIGURE A1

Discontinuity plot—threshold. Each panel shows the variable of interest plotted against the running variable $(R-Threshold)$ using the matched population. The title of the panel tells the name of the variable plotted and the beta coefficient and associated p‐value for the discontinuity.

FIGURE A2

Discontinuity plot—matching. Each panel shows the variable of interest plotted against the running variable $(R-Threshold)$ using the threshold population. The title of the panel tells the name of the variable plotted and the beta coefficient and associated p‐value for the discontinuity.

FIGURE A3

Event study results GLM—threshold. Each panel shows the treatment effects for GPs receiving risk‐adjusted capitation or both risk‐adjusted and geographical capitation. The name of the panel shows the outcome, distribution, and link function used in the GLM model. The vertical lines illustrate 95% confidence intervals.

FIGURE A4

Event study results GLM—matching. Each panel shows the treatment effects for GPs receiving risk‐adjusted capitation or both risk‐adjusted and geographical capitation. The name of the panel shows the outcome, distribution, and link function used in the GLM model. The vertical lines illustrate 95% confidence intervals.

FIGURE A5

Event study results—threshold—dropping retiring GPs. Each panel shows the treatment effects for GPs receiving either moderate or intensive treatment. The vertical lines illustrate 95% confidence intervals. The population is restricted to GPs who haven't retired by the end of 2020.

FIGURE A6

Event study results—matching—dropping retiring GPs. Each panel shows the treatment effects for GPs receiving either moderate or intensive treatment. The vertical lines illustrate 95% confidence intervals. The population is restricted to GPs who haven't retired by the end of 2020.