This study examined if hurricane damage is associated with increased CVD mortality rates in high and low damage counties compared to counties that were undamaged, and if social capital reduces CVD mortality. We found that, net of controls, low damage counties experience higher levels of CVD mortality that low damage counties, but high damage counties affected by Hurricane Matthew experienced lower overall CVD mortality. Institutional confidence was associated with reduced CVD mortality immediately following Hurricane Matthew, but no measure of social capital was associated with reduced CVD mortality trajectories after Hurricane Matthew. These findings suggest that the effects of hurricane damage on CVD mortality may be heavily influenced by location and out-migration before and after hurricanes.
With regard to low damage counties, our findings are consistent with previous literature on hurricanes and CVD mortality, which suggest that CVD events occur at an elevated rate after major hurricanes [3, 8, 9, 35]. It is likely that the emotional and physical stressors, combined with interruptions in normal routines and coping resources exacerbate conditions associated with CVD mortality [81–85].
Our results indicating that high damage counties experienced lower rates of CVD mortality after Hurricane Matthew are likely a result of out migration from the storm area. One study has documented that upwards of 67% of people in high damage counties may have evacuated before Hurricane Matthew, some of which would not return [86]. Although this reflects as a decrease in CVD mortality rates in high damage counties, it is likely that there is simply a smaller pool of potential CVD events that can occur.
Only the institutional confidence sub-index had a significantly reduced CVD mortality rates, and only in low damage counties. This finding is consistent with other social capital and health in disaster research [87]. Areas with high levels of institutional confidence ought to be more likely to follow the advice of public health messaging, may be better equipped to know what to do after hurricanes, and may have an easier time advocating for themselves in the post-disaster process [33, 88]. It follows that CVD mortality rates would decrease as rates of institutional confidence increase. Although intuitional confidence is not associated with reduced trends in CVD mortality after hurricanes, the finding that institutional confidence is related to lower CVD mortality immediately after hurricanes warrants further investigation.
It is also worth noting that among the RA models, the informal civil society sub-index showed consistent positive associations with CVD mortality rates. This may indicate that the informal civil society may isolate individuals or transmit negative health behaviors after hurricanes, in line with the “dark side,” or “Janus-faced nature,” of social capital [22, 26].
Our findings point towards social factors, specifically age and income distribution in the community, as important factors for CVD mortality after hurricanes. Higher levels of income, wealth, and education have been consistently associated with a decreased risk of all-cause and CVD mortality, as well as improved health outcomes in hurricane survivors [89–93]. Communities with a high number of elderly residents may be at particularly high risk for elevated CVD mortality in the wake of hurricanes. This finding may be driven by the relationship between age and CVD mortality, but may also be related to social isolation and a lack of mobility associated with aging [69, 94, 95]. Communities with a high proportion of residents aged 65+ may do well to make sure that elderly citizens are able to evacuate if necessary and are able to access food, water, and medical resources after hurricanes.
Previous research also suggests that pre-hurricane institutional confidence, in the form of civic participation or engagement with government programs, is strongly associated with pre-hurricane preparedness and post-hurricane resilience [96–99]. Cultivating institutional confidence increases community trust in local officials, as well as doctors, and disaster management agencies, potentially limiting the initial shock of high levels of hurricane damage. We did not find evidence for a relationship between social capital and post-disaster CVD mortality trajectories in this study. However, our research provides a foundation for future research that intends to examine the complex relationship between social capital and disaster recovery in small geographic areas using concrete, repeatable, and theoretically grounded measures.
5.1 Limitations
There are several limitations to the study. First, there are data capture issues. The CDC WONDER system redacts data from counties when raw CVD mortality is below ten people under specific sets of parameters. Since complete case analysis is used, data are more likely to be from counties with larger populations. In a similar vein, age-adjusted mortality rates are not available. Data for social capital, even though they are not redacted, are also limited. Most social capital indicators are available on an annual basis, but data for some indicators is available for only a single time point throughout the study. For this project, monthly data collected before and after hurricanes would be ideal. It is possible that a different social capital measurement strategy could pick up associations in the current study. An institutional confidence indicator that was collected more consistently, for example, might show an effect on CVD mortality, which is not seen here. It is also possible that a social capital measure that reflects a different conception of social capital, such as a resource generator, may be helpful in highlighting the specific resources that improve CVD outcomes after hurricanes. Data limitations notwithstanding, our research provides evidence for the relationship between high levels of hurricane damage and CVD mortality after Hurricane Matthew.
The data also do not account for migration away from, and return to, disaster areas. Many people probably left heavily damaged areas and did not come back during the 18 month observation period [100]. Since all population estimates are computed from the midyear interval, the number of people in a county before each hurricane is greater than the number of people in counties after hurricanes. Thus, post-hurricane mortality estimates are likely conservative. This may have played a role in the lack of relationship between CVD mortality and most social capital measures in all models.
The JEC social capital measure also has shortcomings. One of the contributions of our study is highlighting limited utility of the family unity index, which uses a limited definition of family and may be politically biased. In the 2019 US Census, 6.6% of households had a single parent, 39% of women aged 35-44 were unmarried, and 33.7% of women aged 15-50 years that gave birth were unmarried [60]. The JEC-based family unity measure is restricted to nuclear-family ties and is biased in favor of two-parent childrearing households, reflecting conservative political leaning. The measurement gap is reflected in the JEC validation studies, where the family unity sub-index scores the lowest compared to other sub-indices ([49]; Table 5). While we do not doubt the usefulness of close others and family ties in disaster recovery, we do not think that current JEC measure fully captures the diverse nature of family relationships in the United States. Ideally, family unity would reflect diverse interpersonal relationships that generate social capital. A better measure would represent the full spectrum or diversity of family ties that help to form social capital and be supported with empirical data. This would require improved in family measures in the Census/American Community Survey, allowing for an expanded definition of family, per other research [101].
5.2 Implications
Further research is needed to disentangle the specific mechanisms through which social capital operates to protect and, sometimes, harm health after disasters, both in general and in different populations. Specifically, future research could look at how social capital moderates health outcomes in populations that may be more vulnerable to natural hazards due to socioeconomic status, race and ethnicity, or urbanicity. It is unlikely that social capital, and its constituent indices, have a uniform effect for all populations across all outcomes. Additional research can identify the circumstances and populations that are helped, and potentially harmed, by social capital. Doing so will allow researchers to provide valuable insights and perspectives for public health agencies, community leaders, and researchers as severe hurricanes become more common and increase the burden of CVD and other non-communicable diseases.