Exploring excess mortality during the covid-19 pandemic with seasonal arima models

This chapter analyses the daily and the weekly deaths in Germany, Sweden and Spain between 2016 and 2019. It gives an estimate of the future number of deaths in 2020 in those countries, with a special focus on uncertainty, and thereby presents alternative models and methods for estimating the excess mortality in 2020, the year of the Covid-19 pandemic. Suitable seasonal auto-regressive integrated moving average (ARIMA) models are sought that allow the best possible fit to the available time series, in the sense that the properties of the resulting residual processes are compatible with those of white noise. It can be seen that only deaths in the age class 0-30 can be satisfactorily presented by a binomial mortality model. ARIMA is one of the most widely used forecasting methods for predicting univariate time series data. © ISTE Ltd 2022.

[Factors Influencing Results of Mortality Measurement in the Corona Pandemic: Analyses of Mortality in Germany in 2020]. / Einflussfaktoren auf Ergebnisse der Sterblichkeitsmessung in der Corona-Pandemie: Auswertungen zur Sterblichkeit in Deutschland im Jahr 2020.

INTRODUCTION: (Excess) mortality and years of life lost are important measures of health risks from the Corona pandemic. The aim of this paper was to identify methodological factors that affect the calculation of mortality and further to point out possible misinterpretations of years of life lost. METHODOLOGY: Standardized mortality ratios (SMRs) can be used to compare mortalities (e. g., an SMR of 1.015 means excess mortality of 1.5%, an SMR of 0.990 means that mortality is reduced by 1.0%). In this study, SMRs as a measure of association for mortality in Germany were calculated for 2020 using different methods. In particular, the influence of different data sources and reference periods was examined. Furthermore, its influence on the calculated mortality was also examined to take into account increasing life expectancy. In addition, published results on years of life lost were critically analyzed. RESULTS: Using January 2022 data from the Federal Statistical Office on mortality for 5-year age groups resulted in higher SMR values than using preliminary data from February 2021 with 20-year age groups (SMR=0.997, 95% confidence interval (CI): 0.995-0.999 versus SMR=0.976 (95% CI: 0.974-0.978)). The choice of the reference period had a large impact on calculated mortality (for men, SMR=1.024 (95% CI: 1.022-1.027) with 2019 as the reference year versus SMR=0.998 (95% CI: 0.996-1.001) with 2016 to 2019 as the reference period). Analyses in which declining mortality in 2016 to 2019 was carried forward into 2020 when calculating expected deaths resulted in significantly higher SMR values (for men SMR=1.024 (95% CI: 1.021-1.026) with, and SMR=0.998 (95% CI: 0.996-1.001) without carrying forward declining mortality). Figures for pandemic-related years of life lost per person who died from COVID-19 should be interpreted with caution: Calculation from remaining life reported in mortality tables can lead to misleading results. CONCLUSION: When calculating mortality and years of life lost during the pandemic, a number of methodological assumptions must be made that have a significant impact on the results and must be considered when interpreting the results.

Willingness to participate in a COVID-19 follow-up study and symptoms 1.5 years after infection : Sara Schramm

Introduction Data on willingness to participate in population-based long-COVID studies are sparse. We invited all citizens of Essen aged 18-74 years with a positive SARS-CoV-2 PCR test between Mar-Aug 2020 and assessed COVID-related symptoms in responders ∼1.5 years after infection. Methods The invited population included 1282 infected citizens (48% women). At the time of testing 64% reported symptoms. We asked responders about past and current symptoms, hospitalization, smoking, sport, pre-existing conditions (heart attack, stroke, diabetes), subjective health status as compared to before infection, assessed BMI, and performed descriptive statistics. Results We investigated 255 participants (50% women, 19-73 years, response rate 20%) ∼20 month (median) after the PCR test. 95% reported symptoms at the time of testing: 67% fatigue, 58% taste disorders, 56% limb pain, 55% odor disorders, 54% headache, 50% cough, 43% fever;10% needed hospitalization, 3% intensive care, 1.6% artificial ventilation. Compared to the non-hospitalized the formerly inpatients were more often male (62% vs 49%), older (56±13 vs 49±14 years), less often never smokers (42% vs 53%), had a higher BMI (31±7 vs 28±5 kg/m2), and more pre-existing conditions (23% vs 10%). Compared to before infection, 53% rated their current health worse, with a higher rate among inpatients (81%). After ∼1.5 years, 55% still reported symptoms: 25% fatigue, 20% concentration disorder, 18% breathing problems, 13% odor and 11% taste disorders. Persistent symptoms were more common in inpatients than in non-hospitalized (69% vs 53%). Conclusions Symptomatic individuals are more likely to participate in a COVID19 follow-up study than asymptomatic ones. This may overestimate the number of individuals with long-term symptoms in population-based long-COVID study populations. However, persistent symptoms seem to be more likely in formerly inpatients compared to non-hospitalized individuals with former SARS-CoV-2 infection. Key messages • Symptomatic individuals are more likely to participate in a COVID19 follow-up study than asymptomatic ones. • Persistent symptoms seem to be more likely in formerly inpatients compared to non-hospitalized individuals with former SARS-CoV-2 infection.