To check on the newest robustness of your model and you will possible inconsistencies inside research, i recalibrated the fresh model and you will calculated the fresh involved change in the newest chance speed off syphilis, the fresh new suitable variables, and also the god regarding match (amount of squared adjusted residuals) (Desk D during the S1 Text).
Instead way of get the brand new difference from the investigation, i suitable an unit providing there is absolutely no difference in sign speed away from syphilis certainly MSM
When assuming reported non-steady condomless anal intercourse (nsCAI) as the transmission risk instead of nsP, we estimated that MSM with nsCAI had 1.71 [IQR (1.65, 1.81)] times the incidence of MSM without nsCAI. However, the goodness of fit for this model was considerably worse than for the main model (378.8 vs. 148.3). Secondly, we assessed the impact of the assumption of different transmission rates of syphilis among MSMwHD and MSMw/oHD. In the main model we had to assume different transmission rates, because of an apparent discrepancy in the data: despite similar frequency of reported nsP among MSMwHD and MSMw/oHD, we observed much higher incidence of syphilis among MSMwHD than MSMw/oHD even when taking serosorting into account (Table B in S1 Text). To resolve this contradiction, in the main model, we assumed different transmission rates of syphilis (beta) in MSMwHD and MSMw/oHD in our model (Table A in S1 Text). In the absence of this assumption, the model was unable to reproduce the syphilis epidemic in Switzerland and led to either a large overestimation of syphilis incidence in MSMw/oHD or a large underestimation in MSMwHD (Fig N in S1 Text). wHD and MSMw/oHD but instead including a parameter to account for possible underreporting of syphilis cases to FOPH among MSMw/oHD (Fig O in S1 Text). However, the goodness of fit for this model was worse than for the main model (sum of squared weighted residuals, 206.6 vs. 148.3) and required the assumption of an unrealistically high underreporting factor (40.8). w/oHD. We fitted models assuming the proportion of MSMw/oHD with nsP to be 25%, 50%, and 75% lower than the proportion of datingmentor.org/bisexual-dating MSMw/oHD with nsP estimated from data published in the Gaysurvey (Fig P in S1 Text). This corresponds to proportion of MSMw/oHD with nsP in 2017 to be 0.54, 0.36, and 0.18 vs. 0.73, respectively. The estimated goodness of fit values for these models were slightly better than the goodness of fit for the main model (136.4, 124.4, and 123.6 vs. 148.3 respectively) suggesting that model is robust to the hypothesized overestimation.
Thirdly, i examined the newest effect from a possible overestimation off alert risk among MSM
Fourthly, we assessed the impact of the duration of syphilis infectiousness on the dependency between the screening frequency and the syphilis incidence. Instead of assuming zero infectiousness in the latent stage of syphilis (>3 months since infection), we varied the infectiousness of syphilis in the latent stage to be 1% and 10% of that in the primary and/or secondary stage of syphilis (? 3 months since infection) (Fig Q in S1 Text). The goodness of fit values for these models were worse than for the main model (154.3 and 397.6 vs 148.3 respectively). As depicted above in the counterfactual scenario, increase in the frequency of screening for syphilis among MSMwHD with nsP led to reduced simulated syphilis incidence in 2017. Increased infectiousness during the latent stage led to a smaller estimated reduction in syphilis incidence in 2017 in the counterfactual scenarios (% and % vs. % by assuming the infectiousness in latent stage of syphilis to be 1% and 10% vs. 0% of that in primary and secondary stage of syphilis respectively) (Fig 5)