The role of environmental factors on sporadic Creutzfeldt-Jakob disease mortality: evidence from an age-period-cohort analysis

Sporadic Creutzfeldt-Jakob disease (sCJD) is the most common form of prion diseases. The causes of sCJD are still unknown and exogenous factors may play a role. Worldwide, the number of patients with sCJD has progressively increased over time. This increase can be partly explained by increasing life expectancy and better case ascertainment, but a true increase in the number of sCJD cases cannot be excluded. We estimated mortality rates from sCJD in France (1992–2016) and studied variation in mortality rates by age, period, and time. We included all cases aged 45–89 years old who died with a probable/definite sCJD diagnosis based on the French national surveillance network. We used age-period-cohort (APC) Poisson regression models to study variation in mortality rates by sex, age, period, and time. A total of 2475 sCJD cases aged 45–89 years were included. Mortality rates increased with age, reached a peak between 75 and 79 years, and decreased thereafter. Mortality rates were higher in women than men at younger ages and lower at older ages. The full APC model with a sex×age interaction provided the best fit to the data, thus in favour of sex, age, period, and cohort effects on mortality rates. In particular, mortality rates increased progressively with successive birth cohorts. Based on 25 years of active surveillance in France, we show evidence for sex, age, period, and cohort effects on sCJD mortality. The identification of cohort effects suggests that environmental exposures may play a role in sCJD etiology. Supplementary Information The online version contains supplementary material available at 10.1007/s10654-023-01004-5.


Alternative parametrizations of the age-period-cohort model
In sensitivity analyses using the three time variables as factors, we constrained the regression coefficients of the two last periods of deaths (2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012) to be equal and reached similar conclusions (Table S5; Figure  S4). The full APC model had a better fit compared to the A (P<0.001), AD (P<0.001), AP (P=0.040), and AC (P<0.001) models. In addition, including an interaction between age and sex in the APC model provided a better fit to the data compared to the APC model without the interaction (P=0.027).

Analyses restricted to sCJD cases with the M1 strain
Analyses restricted to sCJD cases with the M1 strain are based on 972 cases (438 men, 534 women; 39% of all sCJD patients in this age group) aged 45-89 years old. M1 status was predicted by the algorithm for 609 (63%) of them. The overall mortality rate was 1.91 per 1,000,000 person-years (95% CI=1.78-2.04; Table S6). The agestandardized mortality rate was 1.86 per 1,000,000 person-years (95% CI=1.69-2.02) in women and 1.99 per 1,000,000 person-years (95% CI=1.79-2.20) in men. Unlike analyses based on all sCJD cases, there was no significant interaction between age and sex (P=0.550), although age-specific rates tended to be higher in men than women after 80 years ( Figure S5). APC analyses restricted to the M1 strain showed that the full APC model without the interaction between age and sex had the smallest AIC value and provided the best fit to the data (Table S7).
%" Table S1. Age-specific mortality rates of sCJD per 1,000,000 person-years, overall and by sex (France, 1992-2016). a Global test for the interaction between sex and age. c The relative risk for male sex compares mortality in men to mortality in women at the age of 45 years.

Men
'" N, number of sCJD cases; CI, confidence interval. a Age-standardized mortality rates using the French population as the reference. b Age-and sex-standardized mortality rates using the French population (1992-2016) as the reference.
(" N, number of sCJD cases; MR, mortality rate; CI, confidence interval computed using Poisson regression. )" Table S5. Age-period-cohort analysis using factor variables: assessment of the goodness-of-fit of the models. DF, Degree of freedom; AIC, Akaike's information criterion; A, Age; D, drift; P, Period; C, Cohort; S, sex; A×S, interaction between age and sex.   Age-and sex-standardized mortality rate overall using the French population as reference (grey line) and age-standardized mortality rate by sex using the French population as reference (men, blue; women, red; Table S3).
!#" !$" Figure S3. Age, period, and cohort effects for sCJD mortality rates: alternative parametrization. The left part of the graph shows sCJD age-specific mortality rates (per 1,000,000 person-years; shaded area, 95% confidence intervals) in men (blue) and women (red). The right part of the graph shows relative risks (orange; shaded area, 95% confidence intervals) for periods (reference, 2007), and average cohort effects constrained to be zero on average (green).
!%" Figure S4. Age, period, and cohort effects for sCJD mortality rates: APC modeling using factor variables. The left part of the graph shows sCJD age-specific mortality rates (per 1,000,000 person-years; shaded area, 95% confidence intervals) in men (blue) and women (red). The right part of the graph shows relative risks (shaded area, 95% confidence intervals) for periods (orange; reference, 2007) and cohorts (green; reference, 1937). Periods of death 2002-2006 and 2007-2011 are constrained to be equal (Table S5).
Age-specific mortality rates of sCJD with the M1 strain per 1,000,000 person-years in men (blue) and in women (red) with their 95% confidence intervals (Table S6).