1

In this video Funeral Homes in Mexico Reveal Coronavirus’ Hidden Death Toll | Coronavirus News says that government is underestimating the death toll.

In this part says that health analyst from the university of Washington project the epidemic could kill as many as 45,000 by the end of summer.

This is like 500 daily deaths on June, July and August.

Is this a reasonable prediction?

I likeThatMeow
  • 265
  • 1
  • 17
  • 1
    The population of Mexico is 126 million. If half of those get COVID and it has a 1% fatality rate, that would be 630,000 deaths. How rapidly that may play out is hard to guess. – Daniel R Hicks Jun 05 '20 at 20:24
  • Note: Regarding the [mark](https://youtu.be/a_58Q0H7KoI?t=301) YouTube video, while that's a common practice on _certain_ places, that's certainly not common on other places. So don't generalize, not all mexican people think like that guy. – I likeThatMeow Jun 09 '20 at 16:21

1 Answers1

3

The reference given is a bit vague.

There is an organisation at the University of Washington called the IHME

The Institute for Health Metrics and Evaluation (IHME) is an independent population health research center at UW Medicine, part of the University of Washington, that provides rigorous and comparable measurement of the world's most important health problems and evaluates the strategies used to address them.

Given they are health analysts from the University of Washington and they model the COVID-19 pandemic and produce projections, I am making the jump that this is to whom the New York Times video is referring.

Their current predictions for COVID-19 deaths in Mexico by August 4 are far more optimistic than 45,000.

Chart modelling deaths in Mexico

The project 6859 deaths, with a 95% uncertainty interval ranging from 3,578 - 16,795. It predicts until August 4, not the end of (Northern Hemisphere) Summer.

This projection was made in May 29, 2020. (The video was published June 5, 2020.)

This answer does NOT attempt to show that:

  • There wasn't another health analyst at UW that made the bigger prediction.
  • That the model that the UW prediction is based on is "good", or that it is more accurate than a model that predicts the 45,000 estimate. [Update! See below]

It is intended to suggest that the New York Times reported prediction isn't in line with IHME modelling.

That said, the IHME projection seems to far too optimistic when compared with later data:

The ncov19.live dashboard puts Mexico's COVID-19 deaths at 12,545 (Hat tip to @DanielRHicks.) The John Hopkins Coronavirus resource center agree.

The best graph I could find to correspond with the prediction was from Wikipedia:

enter image description here

It shows that by May 21, the actual deaths were outside the May 29 projections!

I currently have no explanation for this discrepancy. It makes this answer far muddier.

Community
  • 1
Oddthinking
  • 140,378
  • 46
  • 548
  • 638
  • 1
    Do I need to mention that 6,859 deaths is still a tragedy? I hope this isn't seen as dismissive. – Oddthinking Jun 05 '20 at 20:14
  • 2
    The [Coronavirus Dashboard](https://ncov2019.live/) reports 12,545 deaths to date. – Daniel R Hicks Jun 05 '20 at 20:26
  • @DanielRHicks: !!! Interesting. I haven't an explanation for this discrepancy. – Oddthinking Jun 05 '20 at 21:08
  • @Oddthinking Reporting holes or classification differences? –  Jun 05 '20 at 21:26
  • @fredsbend: Yes, could be. Either of those undermine the answer - 45,000 may well be plausible. (Notwithstanding that I don't attempt to dismiss 45K, merely suggest the experts at UW don't project that.) – Oddthinking Jun 06 '20 at 04:00
  • @Oddthinking Thank you for your answer. It's strange that IHME is wrong by much – I likeThatMeow Jun 08 '20 at 15:20
  • 2
    https://covid19-projections.com/about/#historical-performance has been pretty critical of the IHME models; they are updated far too infrequently and their parameters seem quite off, because they very quickly begin to fail after updates. The comparisons are for US-based models but it seems they've been tracking other countries accurately as well. For [Mexico](https://covid19-projections.com/mexico) they're predicting 45,000 by mid/late July and >80,000 through August. The lower bound of their confidence interval is also in the range of 45k by the end of summer. – Bryan Krause Jun 09 '20 at 18:36
  • @BryanKrause Thank you for the information. From you point of view, don't you think they are exaggerating in their predictions? I did the calculations and they are predicting ~70,000 deaths until september 1st. That's like 777 daily deaths on June, July and August. In addition that doesn't match at all with the 45k mentioned on the YouTube video. – I likeThatMeow Jun 09 '20 at 21:22
  • I think they are exaggerating because their prediction to Mexico it's close to the prediction to [US](https://covid19-projections.com/us) but that's strange because US has 331,814,684 and Mexico has 127,575,529 – I likeThatMeow Jun 09 '20 at 21:30
  • @América So far that model has been pretty good at predicting. It's been off at times, but not by too much; for evaluating their models since late April for actual US deaths on 6/6, they were never off by more than 26%. The site itself has more information on how those predictions are made. And yes, for Mexico right now they predict deaths to peak around 1000 per day in early August, with a confidence interval of 400-1300. Deaths per day right now in Mexico are >500 per day over the past few days, and R_t>1 so cases are spreading and these numbers aren't surprising. – Bryan Krause Jun 09 '20 at 21:30
  • @América Huh? They are predicting about 180k US deaths by the end of August. Also US and Mexico are not the same country: different population distribution, different health care, different response to the epidemic. The models are based on the rate of spread of cases and the rate of deaths. And you're asking (in the original question) if 500 deaths per day is reasonable when *it is already happening right now*. – Bryan Krause Jun 09 '20 at 21:32
  • @BryanKrause Ok. AFAIU the number of deaths from now until september 1st is 180k minus the 110k so 70k, that is, basically the same prediction for Mexico. – I likeThatMeow Jun 09 '20 at 21:42
  • @América Things were worse in the US earlier than in Mexico. Please just look at the data, it works so much better than trying to reason about what sounds reasonable without using evidence. – Bryan Krause Jun 09 '20 at 21:45
  • Note that the [Coronavirus Dashboard](https://ncov2019.live/) now has the number of deaths in Mexico up to 14,053. – Daniel R Hicks Jun 09 '20 at 21:58
  • @BryanKrause Ok. I am certainly looking at the graphs, I don't like them at all. The health subsecretary said that this week, we were going to be in the _red_ spot. We have 4 colors so far, green, yellow, orange and red. He said that this colors were going to change each week :) but according to that graph, seems like we're going not going to see green for so long. – I likeThatMeow Jun 09 '20 at 22:08
  • @América Pandemics don't care very much whether you like their effects or not. – Bryan Krause Jun 09 '20 at 22:09
  • @BryanKrause: I would love to see that link in an answer. I think it wipes the floor with my answer. Are you willing to post it? While it doesn't appear to be a UW scientist, in the About section they cite "an analysis of our model by Dr. Carl T. Bergstrom, Professor of Biology at the University of Washington." which could be the vector for the claim. – Oddthinking Jun 09 '20 at 22:20
  • Sure, I thought it could just be an addendum to your answer but I'm happy to make a new post with a different data source as well. That said, it might be worth updating your image from the https://covid19.healthdata.org/mexico link; I don't know whether IMHE updated their model or just accounted for the last 4 days of data (almost certainly the former), but their predictions now also align with the statement by OP (specifically, their average prediction is now 51,912 deaths by Aug 4th). – Bryan Krause Jun 09 '20 at 22:26