• No se han encontrado resultados

3.3. Clasificación del Daño

3.5.2.4. Naturaleza Extrapatrimonial de Daño Moral y

To substantiate the claim that previous research recovers biased estimates of PCP supply on ACSC admis- sions, I use the sample data to compare the results from univariate OLS (physicians only), multivariate OLS (includes county-level controls), and the RD method in Table B.12. Similar to the existing literature, the results are mixed across different methods. The coefficients for the univariate OLS are positive and significant in lev- els. Although adding additional independent variables and controlling for omitted variable bias decreases the positive bias of the coefficient, the coefficients are insignificant in multivariate OLS. The only results that are significant and negative (the theoretically expected result) are from the RD method. Table B.13 shows results from the RD regression with covariates from all stages. Table B.14 further break down the categorization of ACSC into chronic, acute, preventable and multiple conditions using RD. Additional analyses by age group,

race and gender are also performed to examine the heterogeneous effect of PCP on ACSC admissions.

Looking at Table B.13, the results from the RD model display a significant negative relationship between PCP supply and ACSC admissions. A one unit increase in PCP decreases ACSC admissions by almost 148 incidences. This means an additional PCP, or an average of 3.5 percent increase in rural PCP supply, decreases ACSC admissions by 4.2 percent on average in rural counties. In the first stage, the coefficient of the true shortage indicator derived from PCP supply and population is significantly positively correlated with the HPSA designation by the government but is less than one. This further substantiates the usage of “fuzzy” discontinuity design because if a county is below the cutoff, the county is more likely to be a HPSA. In the second stage, there is a significant positive relationship between the HPSA designation and the number of PCPs, more specifically, a county with a HPSA designation at the threshold would have four more PCPs than its non-HPSA counterpart. There is also a significant positive relationship between numbers of short-term care beds and ACSC ad- missions. Short-term care beds or hospital beds are an important control as the number of beds providing the opportunity for patients to be treated, which leads to the positive relationship. However, the indicator variable for whether or not the county is urban or rural is not significant in our analysis. Similar to what Rickett found in his paper, whether or not a county is rural or urban does not significantly affect the number of ACSC. Holding economic characteristics such as per capita income and number of people below the poverty level constant, a rural county does not differ greatly from an urban county in terms of preventable hospitalization.

Examining subcategories of ACSC, Table B.14 shows that PCP supply has the largest effect on chronic ACSC while the effect is the smallest and insignificant with preventable ACSC admissions. Because patients with chronic conditions require more frequent outpatient care in order to maintain their health, easy access to a PCP prevents more ACSC admissions for those at-risk patients than for the rest of the population. Since preventative care is more dependent on the patients willingness to care for themselves, the effect of additional PCPs is small and insignificant on preventable ACSC. This result means that treatment for preventable ACSC and the corresponding patient welfare is not dependent on the access to PCPs. Comparing the models with lagged and contemporaneous policy, it is evident that lagged policy has a stronger effect on ACSC admissions because it allows the county more time to increase its PCP supply. Thus, a county with a longer HPSA duration is able to attract more physicians and decrease the number of ACSC admissions by a larger magnitude across all types of ACSC.

Since prior research has shown that ACSC affects patients differently across their lifecycle, I divided the sample into five age groups: pre-adolescent (age 0-18), adult (age 19-39), prime age (age 40-64), post-retirement (age 65-84), elderly (age 84+). As seen in Table B.15, the effect of access to PCPs on ACSC is more prominent

in the prime age and post-retirement age group. This finding is very different from other papers. The result is logical, however, because aging adults are more vulnerable to limited access than their younger counterparts since their failing health makes them more prone to be admitted to the hospital for ACSC.

When comparing the effect of physician supply across genders (shown in Table B.15), I find females benefit more from an increase in PCPs than their male counterparts in ACSC admissions and its subcategories. Women may make more use of their PCPs than males. I performed some analyses by race using the same dependent vari- ables; however, no significant differences were found. This result means that the effect of access to physicians is equally important to all races after controlling for income and other economic variables.

For robustness, I consider a window of data symmetric about the discontinuity to test whether the result is sensitive to the bandwidth for the first and second stages. At a small bandwidth, the estimation produces similar results and discontinuity estimates only fluctuated slightly due to the inclusion of a limited additional number of counties at the boundary. However, because the small numbers of counties below the cutoff in the data, a larger bandwidth size would render the estimation insignificant.

3.8 Conclusion

Since the creation of ACSC, many costly federal and state policies have been implemented to increase PCPs in the U.S. with the aim of improving patient welfare and decreasing preventable hospitalizations. However, the relationship between PCP supply and preventable hospitalizations (ACSC admissions) is a debated topic in the health policy literature due to the mixed empirical results. Because the number of PCPs may affect hospitaliza- tion for ACSC and physicians tend to locate in areas where there is a high incidence of these or any other illness, there is a reverse causality problem in existing studies. This chapter addresses the simultaneity problem by using timing discrepancies related to HPSA designations through RD design. Using the population of physician and patient data from North Carolina and controlling for endogeneity, I find a significant negative relationship be- tween PCPs and number of ACSC admissions, such that an increased access to PCPs significantly decreases the number of preventable hospitalizations in NC. Estimates using different population groups indicates that female patients benefit more from access to PCP than their counterparts but the effect of PCP are the same across races. Out of the three types of ACSC patients, chronic ACSC patients benefit more from PCP presence than acute ACSC and preventable ACSC patients.

APPENDIX A

APPENDIX FOR CHAPTER 2: DETERMINANTS OF PHYSICIAN LABOR SUPPLY DYNAMICS

Documento similar