The results of this study showed that Medicaid beneficiaries in the direct group experienced improved access to timely and effective primary care in the 24 months post-disaster. This study assessed the long-term impact of an environmental disaster on primary healthcare utilization for a health system serving the medically underserved. Our results are an original contribution to the disaster epidemiology literature. We provide evidence that a health system serving a MUA can prove resilient and display improved adaptive capacity under adverse circumstances (i.e., technological disaster) to ensure access to primary care for vulnerable sub-groups. Controlling for recent surprises in the local health system, we conclude that the permanent reduction in mean ACSC visits post-disaster was likely due to the impact of recovery activity associated with the chlorine spill and is not likely due to an alternative unobserved factor.
We believe reliance on the local FQHC post-event buffered the disaster’s impact on health system performance by providing an alternate site of preventive and primary care for low-income residents directly affected by the chlorine spill. A disaster declaration was denied following the accident and therefore no federal funds were issued to aide response efforts. The local health department and response personnel were focused on the initial characterization of the event for the first two months following the spill. In March 2005, three town-hall meeting were convened with an expert panel to answer the questions of community members. At that time, local officials provided outreach information on general resources within the community, including mention of a FQHC as a medical resource. Note the immediate increase in FQHC attendance for April and May 2005 (). In June a health registry [37
] was initiated and health screenings were made available to registrants within 1 mile [38
] in late August for 10 consecutive weeks. Each screened community member was provided a local community resource flier that included information on the area FQHC and if necessary, screened registrants were referred for follow-up medical or mental health care. Note the subsequent spike in FQHC visits in November 2005 following the conclusion of health screenings and patient referrals (). Our results corroborate previous research that medically underserved populations (MUP) served by a FQHC have significantly lower ACSC rates compared to MUA populations without an FQHC [39
4.1. Strength and Limitations
One strength of our study design is that autoregressive integrated moving average (ARIMA) models are a robust method for assessing ecological changes at the population level. We included a nonequivalent control group to account for secular changes in ACSC visits over time and to rule out threats to internal validity (e.g., history, maturation, testing, instrumentation, and regression) [33
]. In the case of a natural experiment (e.g., a disaster), randomization of cases and controls is often not possible. Therefore we improved upon the non-random control group design by including successive monthly observations before and after the intervention (i.e.
, serial pre-test and post-test measures). One additional advantage to using Medicaid data was the ability to track health care visits within and outside the enrollee’s service area, including visits in surrounding counties and states. Therefore, we were able to capture all ACSC-related discharges and primary care visits independent of service location.
There are a few limitations to our study. Due to small sample size, monthly ACSC rates could not be analyzed separately for ACSC-related hospital or Emergency department (ED) discharges. Combining ACSC visits possibly underestimated the individual impact of the disaster on ACSC volume for hospitals and EDs separately. However, the routine collection of ACSC admissions in hospital administrative data may be used as a public health surveillance measure to estimate population changes in unmet need throughout disaster recovery. Surveillance data can also be compared to ACSC benchmarking in other areas or hospitals, as ACSCs are reported at the local, county, state, and federal levels. Our results also may not be generalizable to other MUA/Ps without a local FQHC. Finally, additional factors affecting monthly ACSC rates should be accounted for in future models, including primary care provider shortages, individual characteristics, disease burden, and the estimation of unmet demand absorbed in receiving areas (neighboring counties, states, and distant areas).
4.2. Directions for Future Research
This study is a preliminary step in understanding a disaster’s impact on the performance capacity of the impacted health system(s). More research is needed to test the application of ACSC as a metric for primary care access in other vulnerable disaster populations, including the uninsured and privately insured. Health system characteristics (e.g., hospital bed availability, primary care provider supply, MUA/HPSA, and availability of a FQHC or Rural Health Center (RHC)) and socio-ecologic characteristics of the population (e.g., household income, poverty, and unemployment) should be included in intervention studies as explanatory variables to help capture the context of health service utilization for future health response planning. Retrospective analysis should be performed in areas experiencing repeat occurrences of specific seasonal disaster events including flooding, tornadoes, and hurricanes to establish the ACSC as a reliable marker of access and explore the impact of persistent health system strain on adaptive capacity across disaster populations and scenarios.
The safety net system, a system designed to ensure health care access for the most vulnerable, is not currently addressed in recovery plans. For the coordination and provision of equitable long-term primary care, we propose that areas designated as medically underserved and/or with health professional shortages (e.g., MUAs, HPSAs) incorporate a safety net response plan to ensure operational continuity throughout recovery efforts. This plan would include the expansion of local safety net services (i.e., community health centers, FQHCs, rural health centers) for up to one year post-disaster. Input from local safety net providers, hospitals, and health systems alongside the residents they serve are needed during the planning process to ensure long-term provision of medical care for vulnerable populations, provider support, response capabilities, and post-event implementation. Commitment to, routine exercise of, and regular updates to the safety net plan would allow communities to prospectively better assess baseline health disparities, risk level, community priorities, and medical resources. Ultimately, without a safety net plan communities can do little more than react in the event of a public health catastrophe, resulting in poor decision making, encumbered post-disaster action, and the compromised health of vulnerable residents.