Health Survey of the Lancaster County Plain Communities and Propensity Score Matching for Asthma in Amish Women
Show FileMIME type:application/vnd.openxmlformats-officedocument.wordprocessingml.documentFile Size:67.6 Kb
Show FileMIME type:application/pdfFile Size:124.7 Kb
Show FileMIME type:application/pdfFile Size:132.6 Kb
MetadataShow full item record
We are distributing a health survey to the Lancaster County Old Order Amish and Old Order Mennonites covering physical and mental health, health care access, environmental exposures, and attitudes toward responsibility for one's health. We hope to understand the current health and health needs of the Plain communities and to measure how perceptions of modern medicine and technology are changing the Plain way of life. Surveys will be mailed to a random sample of households in Plain communities with follow-up in-person with non-responders. We will then survey an over-sampling of Clinic for Special Children patients. This group will receive additional questions about genetics, to be compared to genomic data from exome sequencing of patient-parent trios. New Clinic parents will be surveyed and trios sequenced as part of intake procedures. We also performed propensity score matching and subsequent analysis comparing asthma rates of Amish and non-Amish women, using data from the 2008 Central Pennsylvania Women’s Health Study. Matching was performed using non-Amish population as control, Amish population as treatment, predicted based on the covariates of farm dwelling, household size, body mass index, diet, depression, and social support, although social support was excluded from the final predictive model due to insignificant change in the slope of the logistic regression (∆-2LL = 2.71). We found significantly less asthma in the Amish population, using Fisher’s Exact Test, pre-matching (N = 1566, 1-sided p = 0.000, Non-Amish = 9.6% prevalence, Amish = 1.1% prevalence), post-matching (N = 548, 1-sided p = 0.010, Non-Amish = 4.7% prevalence, Amish = 1.1% prevalence), and post-matching with Amish propensity threshold of 0.5 (N = 316, 1-sided p=0.033, Non-Amish = 4.4% prevalence, Amish = 0.6% prevalence). Despite high statistical significance, we stress that the purpose of these results is exploratory and primarily for proof of concept. A follow-up survey, like the one discussed here, will be necessary to reevaluate the prevalence of asthma in the Amish, although we have established clear evidence of the usefulness of the propensity score matching procedure.
- F&M Theses Collection