Skip to main content
. 2021 Apr 16;118(16):e2103254118. doi: 10.1073/pnas.2103254118

Table 1.

Results of metaregression of association between famine cohort (F1) TB incidence rate ratio (IRR) and famine intensity (CSSI) across alternative famine intensity estimators

CSSI estimator Type Trend Data used to fit model Adjustments Median CSSI (interquartile range) Estimated log change TB F1 IRR per 100% increase in CSSI (95% CI)
CSSIpre_fixed Average None 1953–1957 cohort None 0.39 (0.34–0.45) 0.41 (−0.02–0.84)
CSSIpre_post_mean Average 15 y linear 1953–1957, 1963–1967 cohorts None 0.49 (0.44–0.52) 0.76 (0.14–1.38)
CSSIpre_post_mean_adjusted Average 15 y linear 1953–1957, 1963–1967 cohorts, 1952–1970 fertility data Potential delayed births subtracted from postfamine cohort 0.47 (0.42–0.51) 0.75 (0.14–1.37)
CSSIpre_5_post_5_lm Linear model 25 y linear 1948–1952, 1968–1972 cohorts Cohorts adjusted for survival from 1964 to 2000 0.50 (0.46–0.54) 0.95 (0.16–1.73)
CSSIpre_53_post_5_gam Generalized additive model Smoothly varying 1900–1952, 1968–1972 cohorts Cohorts adjusted for survival from 1964 to 2000 0.45 (0.42–0.51) 0.93 (0.17–1.68)
CSSIpre_10_post_10_lm Linear model 35 y linear 1943–1952, 1968–1977 cohorts Cohorts adjusted for survival from 1964 to 2000 0.45 (0.40–0.49) 0.82 (−0.03–1.67)
CSSIpre_53_post_10_gam Generalized additive model Smoothly varying 1900–1952, 1968–1977 cohorts Cohorts adjusted for survival from 1964 to 2000 0.43 (0.39–0.46) 0.78 (−0.10–1.67)