Influenza‐associated mortality in South Africa, 2009‐2013: The importance of choices related to influenza infection proxies

Abstract

BackgroundRegression modeling methods are commonly used to estimate influenza‐associated mortality using covariates such as laboratory‐confirmed influenza activity in the population as a proxy of influenza incidence.ObjectiveWe examined the choices of influenza proxies that can be used from influenza laboratory surveillance data and their impact on influenza‐associated mortality estimates.MethodSemiparametric generalized additive models with a smoothing spline were applied on national mortality data from South Africa and influenza surveillance data as covariates to obtain influenza‐associated mortality estimates from respiratory causes from 2009 to 2013. Proxies examined included alternative ways of expressing influenza laboratory surveillance data such as weekly or yearly proportion or rate of positive samples, using influenza subtypes, or total influenza data and expressing the data as influenza season‐specific or across all seasons.ResultBased on model fit, weekly proportion and influenza subtype‐specific proxy formulation provided the best fit. The choice of proxies used gave large differences to mortality estimates, but the 95% confidence interval of these estimates overlaps.ConclusionRegardless of proxy chosen, mortality estimates produced may be broadly consistent and not statistically significant for public health practice.

Estimates of influenza‐associated hospitalisations in tropical Singapore, 2010‐2017: Higher burden estimated in more recent years

“…Generalised additive models (GAMs) have been used in multiple studies for estimating influenza burden. 4,8,16,17 An advantage of the GAM is that it allows the data to suggest an appropriate functional form for the relationship between an explanatory variable and the response using a penalised likelihood function. 26 We also included a spline function to account for the non-linear long-term time trend.…”

“…We used a generalised additive negative binomial model to estimate the weekly number of P&I hospitalisations associated with influenza during the 8‐year study period. Generalised additive models (GAMs) have been used in multiple studies for estimating influenza burden . An advantage of the GAM is that it allows the data to suggest an appropriate functional form for the relationship between an explanatory variable and the response using a penalised likelihood function .…”

“…There is a widespread consensus that influenza imposes a considerable burden on public health, including substantial numbers of severely ill patients resulting in hospitalisations and deaths. [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19] The impact of influenza on public health varies across seasons, mainly due to the varying influenza virus types/subtypes, vaccine uptake in the community and the match between the recommended vaccine with the circulating viruses. [4][5][6][7][8] Quantifying the burden of influenza disease is important so as to assess the severity of the virus, evaluate how well public health services are coping with the additional burden or determine whether additional interventions should be considered to reduce the impact.…”

Advances in measuring influenza burden of disease

“…Eighteen population‐specific articles have been included in this Special Edition—including five from Asia, four from Africa, four from Europe, three from North America, and one each from South America and the Middle East . (Figure and Table ) This reflects a broad geographical breadth of new research, and these data will add to our understanding of the burden of disease worldwide, especially in LMICs.…”

“…The study from India found that while the Sample Registration System provided the most appropriate national mortality data set, other mortality data sources could be used for subregional estimates . The South African study showed that weekly proportion and influenza subtype‐specific proxies provided the best model fit with non‐significant differences in the estimates …”