Pre-Registering Myself: Potato Edition

Back in July, I put up a post about some possible interesting ideas about the obesity crisis, which included a lot of discussion about potatoes. At the time the blog Slime Mold Time Mold was running a trial for people interested in eating nothing but potatoes for a month, and they’ve now published the results. The results were intriguing, and the post about it is long. I appreciate that they posted successes as well as failures.

While I was interested in the discussion, I knew that I would likely not attempt to eat nothing but potatoes for 28 days. I have a few health conditions that seem to contraindicate this, and also it just doesn’t sound that fun. I figured this was just an interesting note until I saw someone Tweet about trying the “minimum viable potato diet” to great effect. Basically this person ate an extra 100g of potatoes (about 3.5 oz) per day with no other restrictions and lost 8lbs in 6 weeks.

Given my success with improving my glucose control via potato starch, I loved this idea. Most clinical trials for drugs try to establish a minimum effective dose of the substance in question, why not do this with potatoes? Every substance can have some bad effect if you take enough of it, so restricting your intake to the minimum effective dose helps maximize the impact while minimizing the side effects. While potatoes aren’t going to kill you, that still seemed like a good strategy. Additionally, there are multiple theories for why the potato diet might work well outlined in the SMTM post. Some of them (like monotony) would require a 100% potato based diet to work well, but others (some intrinsic property of potatoes themselves) wouldn’t need you to eliminate all other foods to work.

So basically, I’m going to try this out. I already use 4 tbsp of potato starch at breakfast, and while it has improved my glucose control, I have not seen notable weight loss. For the next 8 weeks I am going to vary the level of cold boiled potatoes I eat per day, and take daily measurements of a few endpoints from the SMTM study spreadsheet:

  • Daily weight
  • Compliance (actual)
  • Energy
  • Mood
  • Ease of Diet (subjective)
  • Any other events of note

My current plan is to test various amounts in 2 week blocks, and I will admit I actually started on 9/5/22. So here’s the schema (adding the potato starch in, though I have been doing that for 2 months already):

WeeksCold Boiled Potatoes
(ounces per day)
Potato Starch
5 oz4 tbsp
10 oz4 tbsp
15 oz4 tbsp
20 oz4 tbsp

This will take me from about 250 calories/day of potato based things to about 550 calories/day. Given that I am a short not-terrifically active female, this would be anywhere from 15-35% of my daily calories. So by the end it will be a substantial chunk of my intake, but nowhere near the 100% done in the SMTM trial.

I plan on posting results. A few notes that I’m also thinking of:

  1. I plan on spreading the potatoes out throughout the day. I started the 5 oz at lunch only, as I have the potato starch with breakfast already. When I go to 10 oz I will add 5 oz to dinner. I do not plan on doing one meal of 20 oz of potatoes or anything like that.
  2. My primary endpoints will be weight loss, mood and energy. Ease of compliance and actual compliance will be secondary. My rationale is that if I have higher weight loss/energy/mood with imperfect compliance to 20oz than I do at perfect compliance with 10oz, I will stick with 20 oz most days.
  3. I reserve the right to stop a phase if I feel terrible. If that happens I will drop to a lower phase until things improve.
  4. Once I find an effective dose, my hope is to continue for 30 days at just that dose to track those results.
  5. I will make no other intentional dietary or activity changes. However, my energy is low so if I start feeling like being more active, I’m not going to intentionally restrict activity either. If all this does is send my energy up and I am more active, that’s a valid finding.
  6. I’m going to try to post updates. We’ll see how often that happens, but I think at least a 4 week, 8 week and 12 week update would be good. Sustainability is another factor for me, so I would like to see how this plays out over several months.

So far week one is going well, so we’ll see how this goes! Looking forward to it. If anyone else would like to try something similar btw, let me know and happy to post your results as well and/or provide more details.

State Level Excess Mortality – 8/31/22

Alright folks! It’s been a good summer but it went by too quickly and I’m realizing it has yet again been too long since I’ve posted an update. And this is still a pretty good update! When I posted for 6/1, we were at 1.125 million excess deaths. Now 3 months later we are at 1.18 million excess deaths. That’s is not bad at all!

We also confirmed that from around Feb-April country-wide we actually had no excess mortality at all, the first time we’ve been there for that long since the pandemic hit. Pretty good! So how does this look on a state level? Let’s see!

Excess Mortality Above Average

First up, the map. When I posted on 6/19, the range at the bottom was 1047-5823 deaths/million residents. Now it’s 1188-6139 deaths/million.

A quick eyeball suggests we are not seeing substantial changes in relative position. Will this play out when we look at the numbers? Let’s see!
StateTotal Excess Mortality
per million 2/1/20-9/1/22
Change from 6/1/22Change in Rank
West Virginia58562340
New Mexico51152670
South Carolina4708421+4

Not a lot of change in rank going on! I took a look and out of the 52 regions listed (50 states + Puerto Rico and DC), 40 were within +/- 2 spots of where they were 3 months ago. So basically we are no longer seeing the strong swings we saw before, it seems like things have settled in to somewhat of a pattern

This raises an interesting point Henry Willmore raised to me a few weeks ago: how well correlated are vaccination rates and things like obesity by state? And if we can get both of those teased out, how much excess mortality seems to be explained by both? I had looked at obesity vs excess mortality about a year ago, but it seems like a good time to look at it again, huh? Let’s go!

Excess Mortality, Obesity and Vaccination

So first up, when I looked at state level obesity rates vs excess mortality a year ago, the correlation was pretty weak. Some high obesity states (like Alaska) still had low excess mortality then, so it wasn’t clear how much this was impacting things. Now the correlation is much higher. Here are the graphs of obesity rate and vaccination rate next to each other. To save you the math, correlation for obesity vs excess mortality is r=.57 and for vaccination rate vs excess mortality it’s r=-.63. Vaccination rates by state pulled from here, obesity rates pulled from here.

So how well correlated are obesity and vaccination rates with each other? Even more strongly correlated than either are with excess mortality, r = -.71:

So can we predict approximate excess mortality using obesity rates and vaccination rates? Well, setting up a model here is a little tricky because our two independent variables are correlated (multicollinearity), but we should end up with a model that looks something like this:

State level excess mortality/million = 5045 + 5876(% Obese) – 5016(% Fully vaccinated)

So a percent drop in obesity is better than a percent increase in vaccination (58 vs 50), though of course vaccination rates have changed quite a bit since 2/1/20.

Testing the model out for MA, we get:




Which is not too far from our actual total of 2063!

Now this model is only moderately well fit. Some states outperformed this by quite a bit: New Hampshire, Hawaii, Nebraska, Iowa, Utah, and Minnesota all had far fewer deaths than this model would predict. On the other end, New Mexico, Arizona, West Virginia, Mississippi, Vermont and New York all had quite a bit more excess mortality than this model would predict. I don’t have a lot of theories for what these particular states have going on, but it is interesting to note that in general those two factors do a moderately job at predicting all cause excess mortality.

Alright, that’s all I have for now! Stay safe our there.

Obesity, Potatoes, and Fermented Foods (Oh My)

This is yet another post that started as a lengthy email to a smaller group, that I thought I’d put up to see if anyone has any more information/links to send my way. The following is mostly just stuff I’m interested in right now, I do not consider any of the below proven hypotheses. It’s just interesting to me and I thought I’d share.

As some background, any long time reader knows I’m pretty fascinated with blood sugar. Mine tends to run high, regardless of my weight or other adjustments I’ve made. My A1C is fine, but my doctor generally does not have any good advice for me on this. Additionally, as someone who has struggled with weight for years, I am also fascinated by the study of obesity and the fact that we have pretty much no public health measures that work to reverse it. And that’s not me conceding defeat, that’s from the Lancet piece “The global obesity pandemic: shaped by global drivers and local environments“. Key quote: “unlike other major causes of preventable death and disability, such as tobacco use, injuries, and infectious diseases, there are no exemplar populations in which the obesity epidemic has been reversed by public health measures.” Validating and depressing all at once there Lancet!

Given how much I’ve thought about this, I was surprised when around this time last year I came across a blog post series from Slime Mold Time Mold explaining how the obesity crisis was even weirder than is commonly believed. The whole thing is lengthy (though worth it IMHO), but it ends up leaving one with the impression that we have focused way too much on the “moral model” of obesity (aka “if fat people would just stop eating and love exercise, things would be better”) and that we should actually take a good look around and make sure there’s nothing actually driving this that we don’t know about yet. They have a couple ideas (lithium as a potential food contaminant/source of weight gain emerges as their top contender), but the overall idea is intriguing. If you could find some simple source of at least some weight gain, you could possibly help millions of people. Beats the hell out of telling them to all go on diets, even the most intensive of which are pretty ineffective over a long time frame.

So where do we look for something simple? Well, there’s a couple interesting leads. First, there seems to be something going on with resistant starch, particularly the kind found in potatoes. This was an internet craze back in 2013 or so, but it appears ingesting potato starch and/or cold potatoes worked really, bizarrely well for some people when it came to lowering blood glucose. A few months ago now, I got the chance to wear a continuous blood glucose monitor myself, and noted that adding 4 Tablespoons of potato starch to my yogurt in the morning absolutely improved my glucose curves to an extremely noticeable degree. The link I provided also links to some other anecdotes, and some discussion about how this failed for some people. It’s not clear why, but there’s a theory the starch might feed some gut bacteria that is good, but also might feed one that is bad, and maybe some people have one or the other.

Actual research in to this has shown that people do in fact react differently to cold potatoes (lots of resistant starch) vs things like hot potatoes or cold noodles (no resistant starch). Intriguing! On top of that, there’s a lot of anecdotes about people eating some variation of the “all potato diet” and doing great. Even healthy weight obesity researchers have tried it out and found it surprisingly pleasant. While anecdotes are not data, the anecdotes are intriguing enough that Slime Mold decided to do a crowd sourced trial on this to see if they could get some data and perhaps spur someone else to figure out what the hell is going on. They have a REALLY good round up of why the idea is so interesting here, and their Twitter feed is RTing real time updates from people who tried it. If this sort of thing interests you, I’d check it out.

I find this idea intriguing because it’s interesting to think that it may not be the foods we eat in particular that are hurting us, but the way they are cooked/eaten. That sort of information would be hidden by current nutrition research, as no one is asking if you had a hot potato or a cold potato when you ate.

This brings me to my other point of fascination: fermented foods. These are pretty trendy at the moment, and have been for a while, but they raise some interesting thoughts. First, fermented foods were a staple of many diets prior to refrigeration. It was literally how things were kept from spoiling. The idea that bacteria we ingest can have a big impact on other foods has been borne out by research. In one study, upping people’s fiber had much less impact on their health markers than telling them to up their intake of fermented foods. The theory was that the extra fiber was only minimally beneficial if people didn’t have the gut microbes to break it down. It’s also been found that skinny people have more gut microbes to break down starches.

I’ve been intrigued by this, especially because I have a lot of trouble digesting grains. Well guess what? Turns out the traditional way of eating grains (at least where my DNA hails from) was to ferment them first. This actually change the composition of grains, decreasing sugar and increasing proteins. Example for spelt here.

Other intriguing facts: in nearly every culture with a “signature” fermented food, those foods tend to be socially associated with long life. Miso, kefir, kvass, kimchi, etc. Kimchi was investigated as a reason why Korea had such low rates of SARS in 2003. I don’t buy the “it’s a cure!” part, but we do know that sometimes non-harmful microbes can edge out harmful microbes with good results for humans. That’s sort of what penicillin is based on, and there is an absolutely wild story about this happening during the Civil War here (note: no one figured out what was going on for over 150 years).

So where does this leave me? Well sadly, we don’t have solid evidence for how to proceed with any of this. Hopefully Slime Mold Time Mold will put something up about their potato results, in the meantime this book is interesting. Personally, I am continuing with the potato starch, as I do have proof that works for me. I am also attempting to eat at least one fermented food at every meal. So far I seem to have seen some positive impacts, but I’m hoping to get another continuous glucose monitor to see if I can more closely watch the effects. Kefir, sauerkraut and miso seems to be the most effective so far. Kombucha doesn’t seem to have much of a benefit, except it does seem to be a very pleasing swap out for alcoholic drinks, and with about 20% of the calories. It seems likely that’s at least partially due to it having a mild-but-still-present amount of alcohol in it.

I am also going to start trying to ferment my grains, to see if I have an easier time with them. I am practicing making my own kefir, and a few other things too. I will continue to eat some fermented something at every meal, and report back if the effects persist.

So folks, any good links or thoughts to add? I find the whole topic fascinating, so send any points of interest my way!

State Level Excess Mortality – 6/1/22

Well hello again! Apparently I’m just falling behind all over the place with this. An update a month was a nice aspiration, but not one I’m managing. Moving on! Last time I posted we were just under 1.1 million excess deaths since 2/1/20, and as of 6/1, we are at 1.125 million. That actually seems….pretty good comparatively? I only have the numbers from 6/1 because the CDC is doing some sort of work on their database and won’t have updated numbers until next week. We’ll see when I get around to looking at those.

Alright, on with what we have!

Excess Mortality Above Average

First up, the map. When I posted on 3/23, the range at the bottom was 1020-5729 deaths/million residents. Now it’s 1047-5823 deaths/million. For a 10 week gap that is….not a bad change. Certainly better than we’ve seen since I’ve been doing this. So where are the bad states?

Interesting. West Virginia is….not doing well? I don’t remember it popping out like that before. Sure enough, here’s what it looked like in my last map:
So where are they at numbers-wise? Well, here’s our top 10:
StateExcess Deaths
Above Average
Change from
Change in Rank
Mississippi5823+94No change
West Virginia5622+912+2
New Mexico4848+206+1
South Carolina4287+85-1

Wow…so that was a jump. They jumped last time as well, so they are moving quite rapidly. What other big movers were there?

State6/1/22 Excess3/23/22 ExcessChangeJune RankMarch Rank
West Virginia5622471091224
North Carolina336925568132442
Puerto Rico197214075654851
New Mexico4848464220645

Interesting, so West Virginia and North Carolina are our two big jumpers here. I’m not clear why that is, but it’s worth noting that North Carolina had been outperforming it’s neighbors for quite some time, and is still outperforming them now. Peurto Rico was also doing very well and it’s jump has it doing only slightly less well.

It’s also worth noting that 9 states lost excess deaths in the last 10 weeks. We had wondered if we were going to see this effect start to happen, as this is something that could occur if some of the people who died initially were those who were close to death already. These states were: Rhode Island (-205), Ohio (-72), Maryland (-62), New Jersey (-52), Massachusetts (-34), Michigan (-31), Illinois (-20), Idaho (-19) and Pennsylvania (-17).

It will be interesting to see if more states start to slip backward as the summer goes on.

Percent Excess Mortality – 2020 and 2021

Alright, so hopefully most states are done updating their numbers from 2020 and 2021 by this point right? Who’s still at it? Well, really only Alaska (+2%), North Carolina (+7%), North Dakota (+8%) and West Virginia (+4%). All other states have very small changes or no change in the last 10 weeks. Top 10 states for each year are highlighted and bolded below, though 2021 had a 3 way tie for 10th so there are actually 12 states there.

State2020 deaths – expected2020 deaths – actual% change2021 deaths – expected2021 deaths – actual% change
District of Columbia695973786%6495713010%
New Hampshire13127134352%13464137752%
New Jersey766869462123%78694834976%
New Mexico191802284219%196162443325%
New York10170511827416%10317911583812%
New York City548708166049%556226325914%
North Carolina999771089169%10029811889319%
North Dakota7233879322%750880657%
Puerto Rico30574320565%30695330908%
Rhode Island103991205416%10877115987%
South Carolina513805967616%527846426022%
South Dakota84561005219%8447936211%
United States2956302335378913%2958796346855317%
West Virginia230332532310%236052867921%

So there we go! The good news is things actually do look to be finally slowing down quite substantially in most places. A few states still look to be struggling, though at this point it’s unclear what’s driving that.

As always, add any questions in the comments or shoot me a message!

State Level Excess Mortality – 3/23/22

Well hello again! Paid work delayed this post for a little bit, so I’m interested to see after 7 weeks where we’re going to land. Last time I posted we had just gone over 1 million excess deaths since 2/1/20, and as of this week we are just under 1.1 million. Hopefully things are settling down now, and I’ll be interested to see where the 2021 numbers are as well. I’m also going to throw in an extra bit about official COVID deaths vs excess mortality, as my state made some major adjustments to the official numbers this month.

Ready? Let’s go!

Excess Mortality Above Average

First up, the map. When I posted 7 weeks ago, the range at the bottom was 881-5245 deaths/million residents. Now it’s 1020-5729 deaths/million. The top states continue to rise faster than the bottom ones. It’s amazing to think that in the top states one out of every 200 people who was alive at the beginning of the pandemic is now an excess death. As always, that’s in addition to those expected to die anyway.

I was surprised to see West Virginia suddenly sticking out more than previously, and was curious to see how that showed up in the numbers. The difference was pretty clear:

StateExcess Deaths
Above Average
Change from
Change in
Mississippi5729+484No change
Arizona4948+328No change
Alabama4768+463No change
West Virginia4710+1081+6
New Mexico4642+455-1
South Carolina4202+833+10

West Virginia and South Carolina had a lousy winter it appears. They were the two states that gained the most excess deaths in the last 7 weeks. Here are the rest of the top 10:

DifferenceMar RankFeb Rank
West Virginia471036291081410
South Carolina42023369833919
North Carolina255620275294245
New Mexico4642418745554
Rhode Island303325934403134

Looking back at my old post, I had noted that West Virginia reporting had been quite sparse since Thanksgiving 2021, so it’s likely some of that jump is them catching up with deaths they should have filed much earlier. Let’s see if that shows up in the 2021 totals.

Percent Excess Mortality – 2020 and 2021

Looks like some of them did! North Carolina, South Carolina and West Virginia added 10+ percentage points to their 2021 excess mortality total since 7 weeks ago. Seems like a data dump. Again, the asterisked states added 2% or more to their total, and the ones highlighted in green are in the top 10.

State2020 deaths – actual2020 deaths – expectedPercent Increase 20202021 deaths – reported2021 deaths – expectedPercent Increase 2021
District of Columbia737868817%7116636912%
New Hampshire13435131152%13766133393%
New Jersey946217682423%83181782126%
New Mexico228421925419%242381961124%*
New York11827410163516%11580210308012%
New York City816605488949%631135556814%
North Carolina108916998539%11202610038212%*
North Dakota8793721422%74647566-1%
Puerto Rico32056304825%32959305268%
Rhode Island120541043915%11566109266%
South Carolina596765151716%642155271122%*
South Dakota10052839820%9348848910%
United States3353789294306914%3454320295500417%
West Virginia253232302810%277132362917%*

Now one more analysis before we go!

Excess Mortality vs Official COVID deaths and Vaccination Rates

So on this blog, we’ve been talking strictly about excess mortality. I started using this metric because I found the discrepancies between state death definitions a bit annoying. Two weeks ago, I heard that Massachusetts was revising their official COVID death count downward by about 3,700. Massachusetts had long been one of the states that appeared to be overcounting COVID deaths, so I was not concerned about this change. While MA has been hanging out in the bottom 5 for excess mortality for months, in terms of official COVID deaths they had been top 10-15 for 2 years now. With this change our official count has dropped to #32, much more in line with the likely true count. I decided to do a quick correlation between states to see how excess mortality lined up with official COVID death counts. The correlation is a stunning r = .83 between reported COVID deaths and excess mortality. Massachusetts highlighted in yellow, suggesting even with the reduction we are still slightly over counting:

That outlier of undercounting is Vermont btw, not sure what they’re up to.

So then, since this question comes up a lot, I decided to do a correlation between vaccine uptake by state and excess mortality since 2/1/20. Even with no vaccines in 2020, we still see a moderately strong negative correlation r = -.65. In general something is called strong at r = .7 or .75:

Vaccine data source here.

Stay safe out there!

The Spanish Project – March 2022

I haven’t posted about personal projects in a while, but I’ve updated a few people in my life about this recently and thought others might be interested as well. After finishing up my stats degree a few years ago, I realized I was almost certainly never going back to school again. That was fine for a while, and while I was dealing with some health issues some rest seemed to be exactly what I needed. However, a few months ago I started itching for a new intellectual project, and realized that I would love to become trilingual. Worldwide, speaking more than one language is the norm. I think it’s actually a little unclear what the percentages are, but the estimates seem to cluster around 40-50% of people worldwide speak 2 languages, and 15-20% more speak 3+. This leaves around 40% only speaking one language. In the US however, only about 20% of people can speak 2 (or more) languages, and this number represents a big increase from prior years. Now some of this could be the lack of regional language variation in the US (you can drive for 3000 miles without hitting a new language), but still, I thought it would be fun to be able to converse in another language. So I decided to do some research.

Because this was just a self driven goal, I decided a few things:

  1. I was going to start with Spanish (4 years of high school learning, but I was never comfortable with it)
  2. I’d give myself 5 years to get conversationally fluent in each language
  3. I’d give myself 3 months to figure out what method I would use
  4. I was really okay if I just became bilingual, so making as much progress as I could in one language was cool too.

The internet has a plethora of information about language learning, so I had a lot to read. Some of the most helpful stuff I came across said I needed to define my goals/reasons to help figure out my approach, so that led me to a couple other rules:

  1. I wanted to be able to comfortably watch Spanish language movies without subtitles
  2. Speaking immediately wasn’t important to me, as I am not planning travel any time soon
  3. I wanted something I could do on my own schedule, and that would be interesting enough for me to see the project through.

This led me to Comprehensible Input, or the input hypothesis. This is a language learning theory (or group of theories) that essentially states that at first, listening is more important than speaking. It’s based on the work of Stephen Krashen, who noted that listening before speaking is how we all acquire our first language, and maybe we should try to mimic that when we acquire a second language. There’s a lot of ins and outs to the theory from a linguistic perspective, many of which are on his website. While this method is a little hard to use in a traditional classroom, it’s exploded in popularity among independent learners. In the age of the internet, getting your hands on media in your target language is easier than ever and more fun than sitting in a classroom. I decided to go for it using a website called Dreaming Spanish, which makes videos specifically designed for adult learners looking to use this method. They specialize in videos that are easier than “native” media, to help you get up to that level. Pablo (the owner) explains the method here, along with the estimated number of hours of viewing it will take you to get to each level. Being numbers driven, I really liked the idea of being able to track hours to monitor my progress, so I decided to go for it. Most of the beginner videos are free, and once you hit the intermediate level it was $7/month (now up to $8/month) to get access to most of the intermediate videos. Less than $100/year to learn a language was a heck of a lot cheaper than grad school, so I went with the subscription.

Pablo estimates that you will need 1000 hours of input to be conversational in Spanish (based on being an English speaker), and 1500 hours to be essentially fluent. I decided to set a goal of 20 hours/month, which would put me at conversational in a little over 4 years and fully fluent in a little over 6 years. I decided to start September 1st, 2021, though I had already watched 40 hours during my investigation period. Here’s how I’m doing so far:

So far I have met or exceeded my goal every month. I started out 42 hours ahead of schedule, and now I am 64 hours ahead of where I thought I’d be. My new goal is to get to Level 4 by the one year mark, which I will meet if I continue to hit 20 hours/month.

More important than just the numbers however, I would say my progress is tracking with Pablo’s estimates. At level 3 (my current level) he estimates I should be able to understand topics adapted for learners, which I can. By level 4 (300 hours) I should be able to understand patient native speakers. At Level 5 (600 hours) I should be able to understand full speed native speakers, and a lot of media will be easier to use for learning.

Also important, the amount of time I spend on this has actually gone up in the last 6 months, which proves this method is engaging, at least for me! Here are my hours/month since I started, not counting the hours I put in before I decided to go with this method:

Interestingly, the jump in hours approximately correlates with hitting the first intermediate level, where I could watch faster videos with fewer drawings/hints for words. I did notice I was more excited to watch the more I felt myself improving in comprehension, which explains why February (one of the shortest months) was my highest number of hours to date.

I won’t pretend I understand all the linguistic debates over whether or not this method is truly superior, but I do have to think that having students get excited over their learning method is a key marker of success. I will never learn Spanish the way I want if I give up after a year, so any method that gets more exciting over time is a plus.

I’ll be updating periodically on my progress.


State Level Excess Mortality – Feb 2nd, 2022

Well hello again! It’s time for another update to state level excess mortality. The last data update contained a dubious distinction, as for the first time our excess mortality since 2/1/20 surpassed 1 million people: 1,006,393. I recall at the very early in the pandemic that someone sent around an email pointing out that in the Hong Kong Flu epidemic of 1968 killed 100,000 people and nothing was canceled. At the time I responded that the 100k death count came over the course of 3 separate calendar years, and that we would have to check back in after 2 years to be equivalent. I feel pretty good about that email now, though I’m not happy I was correct.

Anyway, with the Omicron surge we’ve had some updates to mortality, so let’s get on with it!

Excess Mortality Above Average

Alright, first, here’s the map. When I last posted this 5 weeks ago, the range was about 872-4962 excess deaths/million residents. Now it’s 881-5245. Last time there were some adjustments in the way the data was counted, but this time data calculation methods have stayed steady:

Not a lot of changes in the hotspots, as we can see in the table with numbers:

StateExcess Deaths Above Average/Million,
Change from 12/29/21Change in Rank
Mississippi5245+284No change
Arizona4620+382No change
Alabama4305+105No change
New Mexico4187+459+3
Arkansas3973+210No change
Montana3879+263No change
Wyoming3821+211No change
West Virginia3629+216No change

Interesting that many of the already bad states did not appear to get off the hook in the current wave. Mississippi for example added the 16th most deaths in the last few weeks.

I continue to be pleased that Massachusetts is 47th (out of 52) and New Hampshire is 51st.

The states that added the most deaths in the last 5 weeks are:

State2/2/22 Total12/29/21 TotalDifferenceFeb rankDec Rank
New York355429026521322
New Mexico4187372845947
Rhode Island259322233713440

New York moving up again, interesting.

Percent Excess Mortality, all states 2020 and 2021

I had mentioned last post that I was going to update the totals for all states excess mortality on a regular basis now as well. Some states are still adding to their totals from 2021, so these will continue to change. West Virginia for example appears to have only done 1 week of reporting since Thanksgiving. North Dakota appears they’re backlogged after Halloween. Not sure what’s going on with that, but I highlighted the states with the top 10 totals in green (2021 had a 4 way tie for 10th, so there are 13 states there), and asterisked any state that changed by more than 2 percentage points since the last update. All revisions were upward:

State2020 deaths – actual2020 deaths – expectedPercent Increase 20202021 deaths – reported2021 deaths – expectedPercent Increase 2021
District of Columbia737867769%7033629112%*
New Hampshire13435130923%13766132014%
New Jersey946217681123%83091775947%
New Mexico228421925319%238581961622%
New York11827410167416%11578610274913%
New York City816605496649%631535535714%
North Carolina1089169942010%1004121003120%*
North Dakota8793718522%74697609-2%
Puerto Rico32056301186%32780303308%
Rhode Island120541046815%10889108940%*
South Carolina596765185015%594315250913%
South Dakota10052834920%931985409%*
United States3353789293941814%3427210294848616%
West Virginia253232303010%24911235676%*

An interesting note that in 2020 it took 18% excess mortality to be top 10, and in 2021 it took 21% or more. The US total went from 14% excess to 16% in those years. As more data comes in it will be interesting to see if that moves even further up.

Alright, that’s it for now! Hopefully things will only calm down from here.

All State Excess Mortality 2020 and 2021 – Reported as of January 19th, 2022

I got a request from Henry to post all 50 states excess mortality numbers, both raw and percent above baseline for 2020 and 2021. This seemed a reasonable request, so I pulled it together. To note, data from the end of 2021 is still being compiled by some states. There were 7 that had notes that their data from the end of the year is likely still quite incomplete: Alaska, DC, North Carolina, Rhode Island, South Carolina, Utah and West Virginia. At least 2 of those states are currently showing a death drop, so that’s likely where that’s coming from.

The percent increase calculation changed slightly from my last post, so some of the numbers are slightly different – between 0-2% for 48 out of 54 districts, 3 or 4% for the other 6. Additionally, I used predicted deaths to try to compare in the last post, but here the 2021 numbers are actual reported. This means these numbers will only go up in subsequent weeks. I will probably redo this data in another month or two to see how the slow reporting states (and everyone else) changed.

With that out of the way, here’s the table. I bolded the data for the whole US, and highlighted any percent increase that was top 10 for the year. For 2021 it ended up being top 11 actually, as there was a 3 way tie for 9th. Data set was downloaded here this morning, and leave me any questions in the comments!

State2020 deaths – actual2020 deaths – expectedPercent Increase 20202021 deaths – reported2021 deaths – expectedPercent Increase 2021
District of Columbia737867529%665161598%
New Hampshire13435130923%13756131794%
New Jersey946217677123%82949775177%
New Mexico228421925419%235591960920%
New York11827410163016%11564010271013%
New York City816605495549%631235533214%
North Carolina1089169930310%96231100320-4%
North Dakota8793717922%74697623-2%
Puerto Rico32056300617%32588303068%
Rhode Island120541047415%1062410888-2%
South Carolina596765180515%586125248412%
South Dakota10052834820%911685447%
United States3353789293687814%3403931294807715%
West Virginia253232302610%24240235663%

State Level Excess Mortality – December 29th, 2021

Well hello and happy new year! I hope everyone has had a delightful holiday season and is doing well. As promised, I am back with a state level excess mortality update. Now, I didn’t get to this for a few weeks due to some aforementioned personal life things, and while I was gone I discovered the CDC had update the way they calculated excess mortality and was releasing slightly different numbers from the ones I was previously looking at. You can read their full explanation here, but here’s the gist:

Excess mortality is calculated by taking the prior 4 years worth of deaths and averaging them together to get a baseline of how many people you’d expect to die in a state in any given week. When the pandemic started, the CDC stopped including new deaths in their baseline, because of course we’re all hoping this current mortality level doesn’t become the baseline. Now that the pandemic has gone on for nearly 2 years however, this meant that they were only using 2 years worth of data to determine the “expected” number of deaths. So they decided to go back 6 years (while still excluding our 2 pandemic years, so basically 4 years of data) to get a better baseline. This changed everyone’s excess counts a bit because the baseline was now a bit different. They note that on average this slightly lowered excess mortality estimates by about 2%. In this post I’m going to take a look at if the new calculations substantially changed anything we were seeing before.

To note: they are now only releasing “deaths above average” so that’s what I’m posting here, rather than both deaths above 2SD and above average like I was before. Additionally, this death count is probably going to go up quite a bit in the next 4 weeks as it includes deaths that were reported during Christmas week, which tend to be artificially low.

Excess Mortality Above Average

Alright, first, here’s the map. When I last posted this 6 weeks ago, the range was about 953-4784 excess deaths/million residents. Now it’s 872-4962. So some states clearly lost and some gained:

The hotspots appears approximately the same, with some states changing a bit.

Here are the top 10, along with their change from the mid-November data:

StateExcess Deaths Above Average/Million, 2/1/20-12/27/21Change from 11/10/21 Change in Rank
Mississippi4962+178No change
New Mexico3728+432+9
West Virginia3412+556+16

I looked at Wyoming and West Virginia in particular to see if the change in rank was due to the recalculation or reported deaths, and both states have been running at 50-75% excess mortality since September. With reporting delays, those are likely real increases.

I also looked at the top 10 states that increased their excess mortality count. The ones that showed big increases but didn’t make the top 10 overall were: Alaska (+864, 35th place), Vermont (+362, 25th place), Maine (+331, 45th place), Wisconsin (+314, 41st place), Michigan (+274, 17th place), and Minnesota (+231, 48th place).

I was quite thrilled to see Massachusetts is now 49th in the nation, though the CDC list includes Puerto Rico and DC, so that’s out of 52. New Hampshire is 51st.

Percent Excess Mortality, 2020 vs 2021

A new metric included in the data is the percent excess for each state by week. I thought this was interesting, because some states had a very different 2020 vs 2021. The average percent excess mortality for all states in from 2/1/20 to 12/31/2020 was 16.4%, the average so far for 2021 is 15.6%. Here are the top states in 2020, and how they fared in 2021:

State/Territory% Excess 2/1/2020 to 12/31/2020% Excess 2021 (reported so far)Difference
New York City53.614.6-38.9
New Jersey27.38.2-19.2
North Dakota24.68.2-16.4
South Dakota24.19.2-14.9
New Mexico21.621.5-0.1

Now here’s the reverse: top % excess in 2021, vs how they did in 2020.

State/TerritoryAverage % Excess 2/1/2020 to 12/31/2020Average % Excess 2021 (reported so far)Difference

Unsurprisingly, having 2 bad years appears to land you on the overall top 10 list pretty quickly. I’ll be updating this again to see what 2021 comes in at when we have more reported. With the holidays and the pre-existing reporting delays, this should be relatively straightforward to get.

As always, let me know if there are any questions! Stay safe out there.

Actuarial Tables

So my regular mortality data posting is going to be delayed, as unfortunately we’ve had two pretty close to home deaths within 10 days of each other. Neither were COVID related, but some deaths just make you feel like a light has gone out in the world. Both of these deaths were of that sort, and it’s going to be a dimmer Christmas without them.

Having lost two family members back to back, I’ve gotten the question a few times “was this expected?”. For one it certainly was, for the other not as much. But given that my stress response is often to look at numbers, I did get curious what the probability we were working with was at baseline.

I had posted the Social Security Administrations Actuarial table in one of my posts, so I turned there first. This pleasant little table is broken down by male/female and for each year of life gives you the chance you’ll die in the next year, the number of people (out of 100k) in your age cohort who are still alive, and the number of years you likely have left. All the data is for 2019, so COVID is not included here.

Graphing the probability of death in the next year, it looks like this:

Men get to a 1% annual mortality rate at age 59, women at 66. That goes to 5% by age 79 and 82, respectively. For 10% it’s 86 and 88.

Those seem like decent odds, especially since it’s not random. It is very likely that some chunk of the people in your age category who will die before their next birthday already know, or at least have some serious hints. Terminal cancer diagnoses, major medical events, etc, tend to give a little warning.

Do you want to get even more morbid while we learn about the power of compounding percentages? Good! Here’s the graph of how the death patterns will likely go for 100k people born the same year as you:

So if you are a man born in a particular year, you won’t lose your first 10k cohort members until you’re about 55. In the next decade by age 65, you’ll lose another 10k. The next 10k only take until 72, then 77, 81, 84, 87, 90, and at 94 there will be less than 10k left. For women those numbers are….why don’t I just put this in a table:

# of birth year cohort remaining (of each 100k)Male – ageFemale – age

It’s interesting that the male/female difference appears to come primarily from young age deaths – things actually even out quite a bit as they get older.

I apologize I don’t have a happier/more interesting post. If it helps, you can read about the practice of meditating on your death to help you focus on what’s important: Memento Mori.

I’ll get back on my regular schedule some time in the new year. Stay safe everyone.