Table 2.

Sensitivity analyses of the association of heavy precipitation and shigellosis during a Shigella sonnei outbreak – Oregon, July 2015–June 2016.

Type of analysis	All cases RR [95% CI]	Homeless persons RR [95% CI]	Housed persons RR [95% CI]
Primary analysisa	1.18 [1.06–1.33]	1.36 [1.17–1.59]	1.04 [0.86–1.25]
Exposure window from days −7 to −1	1.13 [0.99–1.28]	1.24 [1.04–1.50]	1.08 [0.90–1.29]
Exposure window from days −3 to −1	1.19 [0.93–1.51]	1.48 [1.05–2.10]	1.06 [0.74–1.50]
Exposure window from days −10 to −4	1.16 [1.02–1.31]	1.29 [1.08–1.54]	1.07 [0.89–1.28]
Exposure window from days −10 to −8	1.28 [1.02–1.61]	1.58 [1.15–2.16]	1.04 [0.73–1.49]
Limited to the rainy seasonb	1.19 [1.03–1.37]	1.35 [1.10–1.66]	0.99 [0.78–1.26]
Maximum (instead of minimum) temperature	1.14 [1.01–1.29]	1.27 [1.09–1.48]	0.98 [0.80–1.21]
Limited to Multnomah County cases	1.27 [1.13–1.43]	1.35 [1.16–1.57]	1.15 [0.95–1.40]
Including a holidays indicator variable	1.14 [1.01–1.29]	1.27 [1.09–1.49]	1.00 [0.82–1.23]
Using NOAA data instead of PRISM data	1.14 [1.03–1.27]	1.27 [1.11–1.47]	1.02 [0.87–1.21]

^aThe primary analysis used an exposure period from days −14 to −8, including cases during the entire outbreak (July 2015–June 2016) and from all Oregon locations, controlled for minimum temperature (average from days −14 to −8) by using PRISM data, and did not include a holidays indicator variable. To account for autocorrelation, all analyses controlled for daily counts at lags in which counts were correlated with counts on a given day (lag 0); these included lags 1 and 7 (days −1 and −7) for the analyses with all persons and homeless persons and lag 7 (day −7) for the analysis with housed persons.

^bRainy season in Oregon is November–March.

RR, risk ratio; CI, confidence interval; NOAA, National Oceanic and Atmospheric Administration; PRISM, PRISM Climate Group, Oregon State University.