How Large Was the Mortality Increase Directly and Indirectly Caused by the COVID-19 Epidemic? An Analysis on All-Causes Mortality Data in Italy

Abstract

Objective: Overall mortality is a relevant indicator of the population burden during an epidemic. It informs on both undiagnosed cases and on the effects of health system disruption. Methods: We aimed at evaluating the extent of the total death excess during the COVID-19 epidemic in Italy. Data from 4433 municipalities providing mortality reports until April 15th, 2020 were included for a total of 34.5 million residents from all Italian regions. Data were analyzed by region, sex and age, and compared to expected from 2015-2019. Results: In both genders, overall mortality was stable until February 2020 and abruptly increased from March 1st onwards. Within the municipalities studied, 77,339 deaths were observed in the period between March 1st to April 15th, 2020, in contrast to the 50,822.6 expected. The rate ratio was 1.11 before age 60 and 1.55 afterwards. Both sexes were affected. The excess was greater in the regions most affected by COVID-19 but always exceeded the deaths attributed to COVID-19. The extrapolation to the total Italian population suggests an excess of 45,033 deaths in the study period, while the number of COVID-19 deaths was 21,046. Conclusion: Our paper shows a large death excess during the COVID-19 epidemic in Italy; greater than the number attributed to it. Possible causes included both the undetected cases and the disruption of the Health Service organization. Timely monitoring of overall mortality based on unbiased nationwide data is an essential tool for epidemic control.

Keywords: COVID-19 epidemic; Italy; SARS-CoV-2; mortality; national statistics.

Figure 1

Daily mortality rates (5-day moving average) in the total sample population (red lines: mean and 95% confidence interval) and in the average of 2015–2019 periods (blue line), and COVID-19 mortality rates (green line).

Figure 2

Differences between the 2015–2019 and 2020 Standardized Mortality Rates versus COVID mortality rates per 100,000 inhabitants calculated via weighed ratio estimator. The weights correspond to the probability of swab testing among the population for Italian regions (as of April 30^th). The size of the circle is proportional to the number of swabs over 1000 inhabitants. The label of the circle includes the name of the region and the number of swabs over 1000 inhabitants. The estimated linear regression line is reported (intercept t = 7.8, slope = 0.47, p value < 0.001).