LMS: The excess mortality due to the corona is starting to stand out in the data

The total mortality from the end of March to the end of October is 10.4% higher than the expected mortality in 2020. That is, this is the excess mortality during the Corona period until the end of October

The Central Bureau of Statistics issued a report on the number of deaths and causes of death in Israel during the Corona period. It turns out that in October the percentage of deaths from Corona is 21% of all deaths.

Summary

For all graphs and tables on the CBS website

• From the beginning of 2020 until the end of October, 40,315 residents died in Israel. In the months of March-October 2020, 2,586 more residents died than in these months in 2019 - 9% more will die in 2020. The difference compared to 2019 is mainly in the months of August-October - 2,207 died, 960 of them in October.
• In the first months of 2020 the mortality was low compared to the previous years, in the months of April-May it was slightly higher and in the month of August it was significantly higher.
• By the end of October, 2,537 residents had died from the corona virus. The percentage of deaths from corona virus among the total number of deaths increased over the months and reached 21% in October.
• The total mortality from the end of March to the end of October is 10.4% higher than the mortality Expected in 2020. That is, this is the excess mortality during the Corona period until the end of October.
• The excess mortality is different in the different age groups. The highest excess mortality is found among those aged 79-70 - 17.3%. Among those aged 80 and over, the excess is about 11%. On the other hand, among young people up to the age of 19 the mortality in 2020 is relatively lower than expected, and up to the age of 59 the excess is very small. 

From the beginning of 2020 until the end of October, 40,315 residents died in Israel.[1] In 2019, 38,056 residents died in those months, a difference of 2,259. In the first three months of 2020, the total mortality was low compared to the previous three years (2019-2017), probably due to an early flu season that affected mortality mainly at the end of 2019 and only slightly at the beginning of 2020. In the months of April-May, mortality was slightly higher than in previous years. But from the month of July, and especially from August, the mortality in 2020 is significantly higher. In the month of September, the death rate was 19% higher than the death rate in this month in 2019 (and also from the month of September in the average of the last three years), and in the month of October the rate was 24% higher compared to 2019 (and 21% compared to the 2019-2017 average).

The increase in mortality is very close to the known mortality from the corona virus. The first cases of corona virus were discovered in Israel at the end of February 2020. The first case of death was on March 20. According to the Ministry of Health, 2,537 people died in Israel from March to the end of October residents From corona virus, 6.3% of the total deaths in these months. In the first months since the corona virus entered Israel, the number of deaths from the virus was relatively low and so was the percentage of the total number of deaths. However, from mid-July this percentage is on the rise, and it reached 16% of the deceased in September and 21% in October. However, in the last weeks of October, there was a decrease in the percentage of deaths from Corona among the total number of deaths - from 24% in the first week of October to 15% in the last week of the month.

Deaths among those aged 70 and over

• Most of the deceased in 2020 and most of the deceased from Corona were 70 years old or older. 76% of all deaths in 2020 are of those aged 70 and over.
• In the months of January-October 2020, 30,555 people aged 70 and over died compared to 28,425 in 2019.
• In the first months of 2020, the mortality of those aged 70 and over was relatively low compared to previous years, but from July it is higher.
• In the months of March-October, 24,063 people aged 70 and over died, compared to 21,645 in these months in 2019, a difference of 2,418 people. In these months, 2,054 residents aged 70 and over died from the corona virus - 8.5% of all those who died at this age.
• 81% of all those who died from the corona virus were aged 70 or older. Over the months, there was an increase in the percentage of deaths from the corona virus among all those who died at the age of 70 and over, and it reached 21% in the month of October.

Mortality by sex

• Of all the deceased in the months of March-October - 51% were men and 49% - women.
• 57% of those who died from corona virus were men compared to 43% women. The rate of deaths from the corona virus among men is 25% higher than the corresponding rate among women (0.25 per 1,000 men versus 0.20 per 1,000 women).
• 9% of men who died in the months of March-October died from the virus, compared to 6.9% of women.

Deaths by population group

• By the end of October, 2,076 Jews and others and 456 Arabs had died from the corona virus.
• The death rate of Jews and others from the virus is 1.6 times higher than that of Arabs, but the difference in rates is due to the fact that the Jewish population is older.
• When examining the difference in rates by age - up to age 69 and over age 70, we find that the rate of deaths among Arabs is 1.9-1.8 times higher than the rate among Jews and others in both age groups. Among those who die from the virus up to the age of 69, 64% are Jews (their share in the population in this age group is 78%) compared to 88% Jews among those aged 70 and over (their share in the population is 92%).

excess mortality

In periods of health crisis, it is customary to examine whether there is excess mortality. That is, is the mortality from all causes higher than the mortality that would be expected under normal conditions. In the period of the corona epidemic, excess mortality is a comprehensive and better measure than the direct mortality from COVID-19 to examine the impact of the epidemic on mortality. Measuring the excess mortality makes it possible to estimate the scope of mortality beyond the known mortality from COVID-19, and to include mortality that could result, for example, from deaths that were not mistakenly diagnosed as being due to the virus or from other causes that were indirectly affected by the crisis, for example due to late seeking medical treatment. This index is also better for comparison between countries, because of differences between countries in testing and diagnosis policies.

Excess mortality can be calculated in different ways, two methods will be presented below:

• P-score - a simple index for calculating excess mortality. It is based only on the number of deaths, and includes a calculation of the percentage difference between the number of weekly deaths in 2020 and the average number of deaths in the corresponding weeks in the previous five years (2019-2015).[2]  This is an accepted index for comparison between countries, but its shortcoming is that it does not take into account the changes in the size and composition of the population and the mortality trend. This disadvantage is particularly problematic because differences in trends in mortality and population (size and age composition) between the countries, can be significant even in a period of 5 years.
• A model developed by a group of researchers in the "Climate and Corona" team as part of the Academia IL Collective Impact: Covid19. Using the model it is possible to calculate the difference between the actual mortality in Israel in 2020 and the mortality Expected Considering the downward trend in mortality characteristic of recent years and the increase in the size of the population and its aging. 

Excess mortality according to the P-score method

When examining the difference in the number of deaths in 2020 compared to previous years using the P-score method, you can see the excess in weeks 16-13 (19/4/20-13/3/20) - an average excess of 11.5% and the increase in the excess starting from week 31 (end of July). Starting with week 37 (which started on 7/9/20) except for the last week of October, the surplus in all weeks is higher than 20%. The average surplus in all weeks 44-13 is 13.9%.

However, age differences were found. At the age of 19-0, the mortality in each year of 2020 is significantly lower than the mortality in previous years. At the age of 59-20 there is fluctuation, but the total mortality is similar to the previous years. At age 70 and over, similar to the trend in the general population, there is an average excess of 13% in weeks 17-13. Starting from week 31 there was an increase in the excess, and from week 33 the excess is higher than 30%. In the last two weeks of October the surplus is lower.

In a comparison between Israel and many countries, some of which are shown in Figure 5, it is noticeable that the excess mortality in Israel in the first months of the corona epidemic was low, but from week 31 (end of July) and especially from week 37 (beginning of September) to week 42 (mid-October) the excess mortality in Israel is higher than most the states. At the end of October there is a noticeable decrease in the excess mortality in Israel, compared to some countries where there is an increase.

Excess mortality according to the model of the "Climate and Corona" team

As mentioned, the P-score index based only on the number of deaths and comparing them to the number of deaths in previous years, does not take into account changes in the size of the population and its composition, and does not take into account the existing trend in mortality. Israel is characterized by a relatively high population growth compared to other countries, an aging population (that is, the share of the older age groups is increasing) and a general trend of decreasing mortality over the years. In order to more accurately examine the excess mortality in Israel, a calculation model was developed Number of deaths Was expected In 2020 no plague Considering these factors, as well as climatic factors and the presence of influenza. The observed number of deaths was compared to it. The model was developed by a group of researchers in the "Climate and Corona" team, one of the groups established as part of - Academia IL Collective Impact: Covid19.[3]

Model[4] Based on this data for the years 2019-2000:

1. Daily deaths according to six age groups.
2.  Average population estimates each year.
3. Daily climate data from Beit Dagan station - temperature - maximum and minimum, radiation - daily average and maximum, calculated values ​​of absolute humidity and dew point.[5]
4. The rate of gross referrals per week to "Maccabi Health Services" clinics due to flu-like illness as an indication of flu-like illness.

The results of the model indicate that from the beginning of 2020 until the end of October, a mortality excess of almost 2,300 deaths was observed (6% compared to the expected). However, if you look at the excess only after the first case of death from the corona virus (from week 13) until the end of October, the resulting excess is slightly more than 2,700 deaths - 10.4% more than expected. It should be noted that all the excess mortality in 2020 is after week 13 since in the first 12 weeks of the year there was a lack of mortality of about 440 deaths (3.6%) compared to the expected. From week 13 until the end of July, the excess mortality was about 370 deaths, so most of the excess started in the months of August - September.

Table A - Excess of deaths during the entire period, 44-1 weeks, by age

(Week 1 starts on 30.12.19, week 44 starts on 26.10.20)

age group	Expected number of deaths (95% prediction margin)	Several deaths were observed	excess deaths
the entire population	38,444 (38,060-38,829)	40,741	2,297*
19-0	807 (751-864)	638	169-*
59-20	4,003 (3,878-4,127)	4,137	134*
69-60	4,912 (4,774-5,050)	5,093	181*
79-70	7,717 (7,544-7,890)	8,485	768*
89-80	12,781 (12,559-13,003)	13,511	730*
90+	8,224 (8,046-8,402)	8,877	653*

* Statistically significant at the 0.05>p level of significance

Panel B - Excess deaths before the onset of mortality from coronavirus, Weeks 12-1, by age

(Week 1 starts on 30.12.19, week 12 starts on 16.3.20)

age group	Expected number of deaths (95% prediction margin)	Several deaths were observed	excess deaths
the entire population	12,219 (12,002-12,436)	11,779	440-*
19-0	241 (210 - 272)	204	-37 *
59-20	1,158 1091 – 1225))	1,153	5-
69-60	1,533 (1,456-1,610)	1,459	74
79-70	2,464 (2366-2,561)	2,323	141-*
89-80	4,162 4,035 – 4,289))	3,968	194-*
90+	2,661 2,559 – 2,762))	2,672	11

* Statistically significant at the 0.05>p level of significance

Table C - Excess deaths during the Corona period, weeks 44-13 by age

(Week 13 starts on 23.3.20, week 44 starts on 26.10.20)

Group	An expected number of deaths	Number of deaths	Number of deaths	excess deaths
age	(Predictive profit 95%)	Watched	from corona virus	מספר	Percent
the entire population	26,225 (25,908-26,543)	28,962	2,537	2,737*	10.4
19-0	566 (519-614)	434	3	132-*	-23.3
59-20	2,845 (2,740-2,950)	2,984	169	139*	4.9
69-60	3,379 (3,265-3,494)	3,634	309	255*	7.5
79-70	5,253 (5,111-5,396)	6,162	620	909*	17.3
89-80	8,619 (8,436-8,801)	9,543	882	924*	10.7
90+	5,564 (5,417-5,710)	6,205	552	641*	11.5

* Statistically significant at the 0.05>p level of significance

The excess mortality is different in the different age groups and it exists mainly among those aged 70 and over. During the Corona period (44-13 weeks), the age group where the excess is the largest is 79-70 - an excess of 17.3%. Among those aged 89-80 and 90 and over, the excess is higher than 10%. Among young people up to the age of 20 there was a lack of mortality even during the Corona period (23%).

The deaths from the corona virus explain almost all the excess mortality, at all ages - 94% of the excess is explained by the corona virus. But even here there is a difference between the age groups:

• Until the age of 69, the number of deaths from the corona virus is greater than the excess of deaths. This figure suggests that mortality from other causes was avoided.
• Over the age of 70, the excess of deaths is greater than the number of known deaths from the corona virus. A possible explanation for this figure is deaths from the virus that are unknown and/or from an indirect effect of the epidemic on the worsening of the morbidity for other reasons.

---

[1] The source of the death data - the Population and Immigration Authority, the Population Registry; October figures are not final and may rise further.

[2] Excess mortality during the Coronavirus pandemic (COVID-19)

[3] A unique academic infrastructure common to all research universities in Israel that was established after the outbreak of the Corona epidemic and aims to provide policy makers with a relevant and reliable research infrastructure for making informed decisions

Team members: Prof. David Steinberg (Tel Aviv University), Prof. Chava Peretz (Tel Aviv University), Prof. Manfred Green (Haifa University), Prof. Little Keenan-Booker (National Center for Disease Control),
Dr. Michal Bitan (Tel Aviv University), Avner Porshafan (Meteorological Service) and Naama Rotem (CBS).

[4] To investigate the daily mortality patterns (basic daily mortality curve) and the relationship with the explanatory factors, a Poisson regression model (GAM Generalized Additive Model) was built. A regression model was fitted separately for each of the six age groups, where the dependent variable was the mortality rate. To examine trends over time in mortality rates, the chronological year was introduced as an explanatory variable. Mortality is characterized by an annual cycle similar to that of the weather, with high mortality in winter and low in summer. This cycle is related to the weather changes and the morbidity phenomena that accompany them, but it does not coincide with them exactly and therefore it is modeled separately. Beyond cycles, there may also be targeted associations with unusual weather, for example an extreme heat wave, especially during the transitional seasons, associated with increased mortality. Therefore, the model also includes the possibility of such effects. The model also reflects weekly fluctuations in mortality rates, with a clear decrease on Saturday and a more moderate decrease on Friday. To predict the expected number of deaths in 2020, the basic models built were used. The prediction is calculated for each day and for each age group. The daily predictions were summed up to get weekly predictions and prediction gains were calculated at a confidence level of 95%. Every week the difference between the expected value and the observed value is calculated and a periodic summary is calculated. Weekly predictions for the entire population were calculated from the sum of the predictions for all age groups.

[5] The temperature at which the air cools at a constant pressure and with a constant water vapor content until it reaches saturation, i.e. the state where the water vapor must undergo condensation and turn from a gas to a liquid.