Hypothesis

Counties containing or adjacent to DOD installations with confirmed PFAS contamination in groundwater (PFOA/PFOS > 70 ppt) exhibit age-adjusted autoimmune disease mortality rates 15-30% higher than demographically matched counties without military PFAS sites, after controlling for income, insurance coverage, urbanicity, and industrial co-exposures.

What We Know

PFAS (per- and polyfluoroalkyl substances) are immunotoxicants at environmentally relevant doses. Grandjean et al. (2012, JAMA) demonstrated that PFOS doubles at serum concentrations above 20 ng/mL suppress vaccine antibody responses in children. Epidemiological studies near Parkersburg, WV (C8 Science Panel) linked PFOA exposure to ulcerative colitis and thyroid disease. The DOD has identified over 700 installations with known or suspected PFAS releases from aqueous film-forming foam (AFFF), primarily at fire training areas. These sites create persistent groundwater plumes that contaminate surrounding municipal and private wells, often for decades before detection.

Autoimmune diseases -- including rheumatoid arthritis, lupus (SLE), type 1 diabetes, and inflammatory bowel disease -- share immune dysregulation pathways (Th1/Th17 skewing, regulatory T-cell suppression) that PFAS compounds are known to perturb in vitro and in animal models (Dong et al., 2013, Environ Health Perspect).

The Gap

No population-level study has systematically tested whether the geographic footprint of DOD PFAS contamination predicts spatial clustering of autoimmune mortality. Existing studies focus on single sites (e.g., Pease AFB) or single diseases. The 700+ DOD sites create a natural experiment with built-in controls: military bases without PFAS releases, and civilian counties with similar socioeconomic profiles.

Testable Predictions

1. Counties within 10 miles of a DOD PFAS site (n ~ 350) will show autoimmune disease mortality (ICD-10 codes M05-M06, M32, K50-K51, E10) 15-30% above expected rates vs. propensity-score-matched non-PFAS counties, using Poisson regression with overdispersion.
2. A dose-response gradient will emerge: counties with PFAS groundwater concentrations > 400 ppt (top quartile of DOD sites) will show stronger associations (IRR > 1.30) than those at 70-400 ppt (IRR 1.10-1.20).
3. Lag analysis will show the association strengthening at 10-20 year latency windows, consistent with chronic immune disruption rather than acute toxicity.
4. The effect will persist after excluding counties with known industrial PFAS sources (3M, DuPont facilities) identified via EPA TRI.

Falsification

The hypothesis is falsified if: (a) autoimmune mortality IRR for PFAS-exposed counties falls below 1.05 with 95% CI crossing 1.0 in the fully adjusted model; (b) no dose-response gradient is observed across PFAS concentration quartiles (trend p > 0.10); or (c) the association disappears after adding industrial PFAS co-exposure as a covariate, indicating confounding rather than military-specific exposure.

Data Sources

• DOD PFAS sites: DOD PFAS Task Force installation list + EPA PFAS Analytic Tools (public, updated quarterly)
• Groundwater concentrations: USGS PFAS groundwater monitoring data + state-level UCMR3/UCMR5 detections via EPA SDWIS
• Autoimmune mortality: CDC WONDER Compressed Mortality (county-level, ICD-10, 1999-2022)
• Demographics/controls: Census ACS 5-year (income, race, insurance, housing), EPA TRI (industrial co-exposures), AHRF (physician density)
• Military base locations: DOD BRAC and installation geospatial data (public)

Why This Matters

DOD has allocated $34B for PFAS remediation but prioritizes cleanup based on drinking water contamination, not disease burden. If autoimmune clustering is confirmed, it reframes remediation urgency from environmental compliance to public health triage -- and opens VA disability claims for autoimmune conditions in veterans and dependents who lived on contaminated bases. The 700-site natural experiment is unrepeatable; the data exist now and are free.