Published drought-whiplash effects gave the direction of statewide California coccidioidomycosis changes in 2020-2024 but no county forecast gain over persistence; cases stayed high after wet years
Question: Do published climate relations for coccidioidomycosis (Valley fever) in California forecast later years? We tested the pooled fall incidence rate ratios of Head et al. (2022; winter precipitation 1.45, summer temperature 2.02, winter temperature 0.74 per IQR, with a 1.36 times larger precipitation effect after two dry winters) and the statewide forecast of Camponuri et al. (2025). Data: CDPH final county and monthly cases 2001-2024, provisional January-September 2023-2025, and NOAA nClimGrid county climate below 500 m. Results: For April 2024 to March 2025, the published forecast was 3% low and inside its 90% interval (12,244 against 12,685); persistence was 17% low. Its 2023-24 estimate was 14% high and outside its interval. The published IRRs gave the direction of statewide fall changes (r = 0.60 in 2002-2019, 0.68 in 2020-2024), but their size was too large after 2019 (skill against no change +0.31, then -0.49). They predicted a 37% decrease in 2024, but fall cases increased by 20% to a record. In 10 endemic counties, frozen 2003-2019 models did not forecast 2020-2024 calendar-year counts better than persistence (skill -1.58 with published IRRs; -0.12 to +0.06 for refits; placebo 95th percentile 0.21). After the dry 2024-25 southern winter, the published IRRs predicted 45% fewer January-September cases in 21 counties; provisional cases increased by 2%. The excess after a whiplash winter did not decrease in the next year (carry-over ratio 1.36, 90% CI 0.82-1.76, 21 counties, 2003-2024). Thus the relation is a valid signal of statewide direction, but not an operational county or year-ahead forecast in its published form. Limitations: public data are county calendar years, not census tracts by month, so the IRR transfer is approximate. There are only five forward years and a partial, provisional 2025. Regions of the published forecast were not scored.
References
8- reviews, challengesdoi:10.1016/S2542-5196(22)00202-9
- reviews, supportsdoi:10.3201/eid3105.241338
- doi:10.1016/j.lana.2024.100864
- doi:10.15585/mmwr.mm6948a4
- A 2001-2019 dengue temperature curve gave half the possible 2020-2024 forecast gain in Brazilian municipalities, but its 20-26 C slope rose from near 0 to 0.8-1.0 per C after 2012
- Published drought and immunity models of US West Nile neuroinvasive disease fitted to 1999-2013 did not forecast 24 state counts in 2014-2025 better than simple baselines
- useshttps://www.cdph.ca.gov/Programs/CID/DCDC/Pages/ValleyFeverDashboard.aspx
- useshttps://www.ncei.noaa.gov/products/land-based-station/nclimgrid-monthly
Cited by
0 publicationsNo incoming citations on Sciscape.