Disruption of normal life during the pandemic: How many lives were lost in Arizona?

Eyal Shahar
7 min readAug 31, 2022

Excess deaths — the difference between all deaths and expected (“normal”) deaths — don’t lie, assuming “normal” is estimated reasonably well. Not so for Covid deaths, which may be under-counted (missed Covid deaths) or over-counted (misattributed as a cause of death). It is difficult to get direct evidence for the former, but it is simple for the latter. If the count of Covid deaths is higher than the estimate of excess deaths, then some so-called Covid deaths were not caused by Covid. A true Covid death should count towards excess mortality. It would not have happened “normally” (in the absence of a pandemic.)

For example, suppose we estimate 900 excess deaths in some month and the official number of Covid deaths is 1,000. Then, 10% of the official count was misattributed to Covid (100/1,000). There was over-counting.

We have all heard about excess mortality during the pandemic, which was often higher than a tally of Covid deaths. Three sequential questions may be asked:

1. What was the difference between excess deaths and Covid deaths over time?

2. Is there evidence of over-counting of Covid deaths?

3. How many excess deaths are not attributed to Covid?

The last question is the topic here. The answer will tell us about the death toll of unprecedented interventions which disrupted normal life “to save lives”.

Sources of data

1. The Centers for Disease Control (CDC) excess death file provides weekly data for the US and for each state, using death certificates. The file contains the following relevant variables (among others): all-cause deaths; all-cause deaths, excluding Covid; average expected number of deaths.

2. Arizona Department of Health Services (ADHS) presents a bar graph of weekly Covid deaths, using the date of death. The numbers can be read off the graph.

All data were truncated in September 2021 because flu returned that fall and its share in excess mortality is unknown.

Two key terms

When excess deaths exceed Covid deaths, the difference may be called “unaccounted excess deaths”, because these deaths are not accounted for by the tally of Covid deaths.

When Covid deaths exceed excess deaths, the difference may be called “over-counted Covid deaths”, as explained earlier.

First, what’s in the US data?

The graph below is not the kind of graph that CDC would publish, but it was derived from their data.

It is probably the first time that you are looking at solid evidence of over-counting of Covid deaths in the US. Two periods provide the evidence, but we will ignore the first because the magnitude was small. The second period was about five-month long (roughly January through May 2021), showing substantial, continuous over-counting over many weeks.

Do you think that misattribution of deaths to Covid began suddenly in January 2021? Do you think it was absent before?

Well, that’s implausible. That’s not the way trends work. They don’t usually change overnight. The magnitude of over-counting might have been different, perhaps smaller perhaps larger, but it was probably present as early as two to three months into the pandemic (a subtle indication in May and June 2020.) Over-counting was detected between waves, when the background rate of Covid was low. It was masked otherwise.

Are unaccounted excess deaths missed Covid deaths?

If over-counting of Covid deaths persisted throughout the pandemic, to a varying extent, then unaccounted excess deaths are not missed Covid deaths, contrary to what many have claimed. They are truly unaccounted for, or rather, accounted for by all those interventions that disrupted normal life. Numerous causal pathways have led to these deaths.

Moreover, the true number of unaccounted excess deaths (excess minus Covid) should be larger because the number of true Covid deaths was lower than the official tally. How much larger depends on our estimate of the percentage of over-counted Covid deaths during the pandemic. About 5–10% would be a conservative range. We will use these numbers later, for Arizona.

What happened in June-September 2021?

The last period on the graph (above) is peculiar. Not only do we see, again, unaccounted excess deaths, but their average number per month, and their share of all excess deaths, are much higher than in 2020. What has accounted for so many unaccounted deaths so late in the pandemic? Have they suddenly started missing Covid deaths after a long period of over-counting? As I wrote before, trends don’t usually change overnight.

There are no data-driven answers, but I will offer one speculation, which might explain a drastic change.

That period includes the rising part of the Delta wave. It was the first time that a highly vaccinated country faced a Covid wave, and everyone believed, including physicians, that vaccination offered almost perfect protection (which was false).

Physicians might have changed their behavior when Delta hit. Instead of willingly attributing deaths to Covid as before — correctly and incorrectly — they were now reluctant to attribute Covid to a (true) Covid death of a vaccinated patient, thereby increasing the share of unaccounted excess deaths. In other words, some unaccounted excess deaths were Covid deaths of vaccinated people. Again, that’s no more than one speculation. Perhaps (masked) over-counting continued and other causes were operating.

What happened in Arizona?

The top figure replicates the US graph, using the CDC data for Arizona. The bottom figure used Covid data from the Arizona Department of Health Services (ADHS), but still relying on the CDC estimates of expected (“normal”) deaths.

These graphs, and the data behind them, show the following:

First, the tally of Covid deaths by the CDC was lower than by ADHS (making the bars of unaccounted excess taller). It was lower in every single week throughout the pandemic. We have evidence of persistent over-counting of Covid deaths by an official state agency (ADHS) — evidence provided by no other than the CDC itself! Who will check if the CDC numbers are still too high, at least some of the time? Their own data: a period of over-counting is clearly seen in the top graph.

Second, despite differences in the counts, the general pattern is consistent between the two sources of Covid deaths.

Third, the period of detected over-counting in Arizona was shorter than in the US as a whole: about three months instead of five months.

Data tables for Arizona

The 18-month period (April 2020 — September 2021) was divided into two periods of unaccounted excess deaths and an interim period of over-counting Covid deaths (Tables).

Notice, again, that the CDC Covid death counts are lower than ADHS counts in each period. About 10% of Covid deaths, as counted by the state of Arizona, were deemed to have not been caused by Covid according to the CDC criteria.

Similar to the US data, the share of unaccounted excess death in the last period was much higher than in the first period, and so was their average number per month. In between, we have evidence of substantial over-counting of Covid deaths. (The round numbers in the middle table are a rare feature of data.)

How many lives were lost — beyond Covid — during the 18-month period?

We can only compute a range, given uncertainty about the magnitude of over-counting of Covid deaths in the first period and uncertainty about what happened in the last period.

The ADHS counts were already rejected by the CDC counts. So we may start with the most conservative assumption: Over-counting and any under-counting eventually balanced, and the CDC count of 18,540 Covid deaths in Arizona is correct (Table). On that assumption, over 5,000 deaths (22% of the excess mortality) are attributed to disruption of normal life. That’s about one-fifth.

Under plausible assumptions of 5% and 10% net over-counting, the numbers increase to over 6,000 and over 7,000, respectively, or 26% and 30% of the excess mortality. That’s one-quarter to almost one-third.

Saved lives?

But lives were saved by those “non-pharmacological interventions” that disrupted normal life. Were they not?

First, the lives of those 5,000 to 7,000 Arizonans were not saved. Statistically, only a few of them would have died from Covid. Is it ethical to sacrifice so many people in order to save other people? Is this an accepted conduct of public health? That’s not the case of inevitable, rare fatality of a flu vaccine (one per millions).

Second, suppose we accept the dubious ethic, how do we know that more lives were saved than lost? Do we even have evidence that lives were saved? Or how many?


Pre-pandemic plans rejected everything that was forced upon us by public health officials and elected politicians — fearmongering, lockdowns, masks, and any form of disruption of normal life. It was widely accepted that these measures were useless, and would have grave consequences.

There was never evidence that human beings can control the spread of an airborne virus, and none emerged during the pandemic. The slogan “control the spread” was no more than naïve, wishful thinking. There was no science behind it. None.

It is certain that many lives were lost by the disruption of normal life. It is doubtful that a significant number was saved.


I have written about 20 articles about the pandemic (dating back to March 2020), in Hebrew and in English. My article on Sweden in Medium was read by 50,000. This one will be read by hundreds, perhaps. People have moved on. The pandemic is history. The only visible reminder is the occasional masked customer at the grocery store.

Many aspects of the pandemic cannot be unbiasedly judged by our brainwashed generation. They will be judged by future scientists and future historians. For them, we should leave as much record as possible of what has truly happened. This article is a small part of the record.



Eyal Shahar

Professor Emeritus of Public Health (University of Arizona); MD (Tel-Aviv University, Israel); MPH, Epidemiology (University of Minnesota)