New Estimates of Excess Mortality on Puerto Rico

It’s easy to forget that thousands of American died in the wake of Hurricane Maria, and more are set to die as the administration opposes further assistance to the Commonwealth. Yet, new estimates confirm earlier estimates, and give detail on the causes.

Figure 1 summarizes some key estimates, including that from a recent study by Cruz-Cano and Mead.

Figure 1: Cumulative excess deaths from September 2017, for simple time dummies OLS model (blue), OLS model adjusting for population (green), and Quantile Regression model adjusting for population (red), Milken Institute point estimate (black square) and 95% confidence interval (gray +), Santos-Lozada, Howard letter (chartreuse triangle), Cruz-Cano and Mead (pink squares), Kopits (teal triangle). Not pictured: Kopits estimate of 300-400 for October 2018. Source: author’s calculations, Milken Institute (2018)Santos-Lozada and Howard (2018), Cruz-Cano and Mead (2019), and Kopits (2018).

The Cruz-Cano and Mead study breaks down excess fatalities.

Source: Cruz-Cano and Mead (2019)

I find it striking how many excess fatalities are attributable to diabetes — presumably some nonsignificant share is due to the lack of power for refrigeration and access to dialysis treatment. Heart disease is the single largest category. Lack of treatment was likely the cause, according to the article.

And yet, I suspect Mr. Trump would assert his administration deserves an A+ for disaster response. As for the current situation, from AJMC:

Meanwhile, recovery efforts to repair the $90 billion worth of damage are lagging; most recently, Congress and President Donald Trump failed to resolve an impasse over disaster aid. Rep. Nita Lowey, D-New York, has introduced a bill for $17.2 billion in emergency supplemental appropriations, as well as $3 billion for the Midwest, which recently flooded.

The administration has dismissed additional funding for food assistance as “excessive and unnecessary“. So expect excess mortality levels to remain elevated.


40 thoughts on “New Estimates of Excess Mortality on Puerto Rico

  1. Moses Herzog

    Is it possible to have several seconds of chuckles followed immediately by a clearly audible groan?? That was my reaction upon seeing the teal green triangle.

    What precipitated Kopits’ change in the number from October 2018 to now?? Did it finally dawn on him he was making a complete jack-a$$ out of himself?? Because I know it wasn’t a sudden discovery of morals, because I’m pretty sure at this stage of his life that one is beyond him.

    Has Kopits come up with any solutions for the most undesirable and unskilled of the illegal immigrants?? The ones who knowingly break federal laws related to their status??

    I suspect IP theft could be a large problem with one weak loser immigrant in particular:

    Maybe it excites her that the ugly fat orange mammal she’s married to molests women?? Guess you’d have to ask her.

      1. Moses Herzog

        “Ostensibly” is a very interesting word choice for a man who spends a large degree of his time exaggerating his personal profile. Is it “circulating at the White House”?? Probably not, for if you had even brought it with you to be turned away at the White House guard shack entrance you’d have been telling us the President had been obsessed with your work for several months.

        YOU are a laugh-riot dude.

      2. pgl

        You’re kidding? #1 – I doubt this has made it to the White House but don’t let that stop your self promotion parade.

        #2 – the White House is full of morons and racists. So if your garbage made it there – maybe they would like it. Which of course would add to the insanity of an already awful Administration.

  2. Moses Herzog

    “I find it striking how many excess fatalities are attributable to diabetes — presumably some nonsignificant share is due to the lack of power for refrigeration and access to dialysis treatment ”

    This isn’t a criticism, more a highlight of what my comment is in relation to. I’m certainly no expert on diabetes (although arguably I should be, as many of my relatives have had it). But let’s even say by some miracle you have a nearby refrigerator that works—however unrealistic that is, let’s just take that as an assumption. How do you get supplies of insulin when you can’t even get your hands on food?? Nevermind how much your sugar readings would be tracking off the regular readings because you’re not even eating.

    I’ve made it clear here I’m no wiz at math. But I think if you were to take the median or average number of poor blacks or poor hispanics % the overall population with diabetes (choose a comparable country or region to Puerto Rico—maybe a small Central American country–whatever—you choose. Or 5 of those countries and then take the average)—take that number however you were to calculate that. Then take that % to multiply it by the total population to get an estimate of the total number of Puerto Ricans with diabetes. Then use that number as your denominator and the best estimate of deaths by complications of diabetes related to post-Maria effects as your numerator to that denominator—-and my guess is the opposite of Menzie’s—that the number of deaths (considering the post-Maria environment/context) are actually shockingly on the low end.

    I rarely disagree with Menzie, and it’s not that I even think the numbers are wrong—-I am just saying I am not shocked at how high those diabetes death numbers are—but would be more apt to be shocked how low those numbers are. People (and I am not talking Menzie here, I’m talking the “average person”) don’t really realize how dependent some people are on insulin. I mean, not having insulin handy for some people is near to putting an air-tight baggy over their head—it’s not if they are gonna die but when.

  3. pgl

    “excessive and unnecessary”? But of course Stephen Miller would explain – these are BROWN people after all so they do not count. Besides – Team Trump can always rely on the “policy” analysis of Princeton Steve to give them cover. Back to shouting NO COLLUSION!

  4. Moses Herzog

    Coffee and chocolate wafers, still highlighting the redacted Mueller Report. Generally enjoying it in a demented kind of way. Reading glacially slow. My Dad was a slow reader. I often feel some embarrassment over my slow reading—but I will say this—I bet with my slow reading I am catching some parts of this that others who have already finished the report are missing. Some of this stuff (as with many things in life) you really have to slow-walk through your mind to grasp it all. Couple things was super tempted to comment on, but I’m gonna do my best to reserve thoughts until I finish this.

      1. Bob Flood

        I’d like to see someone – maybe a student – do the following:
        1. This should be a state by state panel VAR
        2. equation 1 is mortality
        3 equation 2 is exogenous disaster measure, eg. max wind or storm category
        4. equation 3 is federal aid.
        Bunch of ‘controls, e.g., per capita income….

        Then conditional on the hurricane we can look at the shock to federal aid and trace out implications.

        With the panel we can compare PR vs states. When estimated state by state including PR, we cn get an idea of ways PR is different from states…..

  5. Erik Poole

    One could argue that Puerto Ricans died in good numbers because of poor lifestyle choices. Does that mean that fellow Americans should abandon these people?

    Similarly one could argue that obese North Americans with COPD are ill and more likely to die early because of lifestyle choices. Does that mean the rest of us should abandon these people? Does that mean that just because small particulate matter (e.g., PM<2.5) is not visible that North America should maintain the lowest excise taxes on automobiles among the rich OECD members?

    The Camp Fire that devastated Paradise California last year and left 85 dead is classical case of supreme ecological arrogance: low-density suburbs pushed into semi-arid forests. Does that mean that the state of California should not bother fighting these fires? Or does it mean that we should finger-point global warming and then try to figure out to save the planet by imposing zero burdens on most American citizens? (Apparently Paul Krugman lives in this fantasy land.)

    1. pgl

      “One could argue that Puerto Ricans died in good numbers because of poor lifestyle choices.”

      One could argue a lot of BS. Does that justify FEMA abandoning them simply because its boss hates brown people? Yea – I know, Team Trump has tons of lame excuses including your pathetic comment here.

      1. ilsm

        pgl, do have anything more than a middle school rank out response?

        I outgrew t those in the early 1960’s.

  6. Steven Kopits

    My general comments:

    Cruz-Cano and Mead is largely in line with our earlier assertions.

    By their count, 86% of excess deaths occurred in the 60+ age group. And as their footnote ‘b’ states: ‘No excess deaths were observed in other age groups.” They estimate 54% of those who died were in vulnerable health groups, with death attributed to heart conditions, diabetes, kidney failure (septicemia) and Alzheimer’s. ‘Other’ takes an additional 17%, but it’s not entirely clear if these are aged-related infirmities or something else.

    It’s not entirely clear whether they have fully accounted for departures from the island in the immediate aftermath of the storm.

    For Sept-Oct, they come out with excess mortality of 1205. Ours through December is 1439, and through February, 1525.

    It is reasonably clear that the Milken numbers look high. There is virtually no way to generate excess deaths on-island in Jan-Feb, and thus the Milken numbers for Feb look very high.

    Finally, it appears we have jettisoned the Harvard MPR study numbers entirely, as they would not even fit on the scale of the graph. I believe this qualifies under the definition of ‘garbage’.

    1. 2slugbaits

      Steven Kopits it appears we have jettisoned the Harvard MPR study numbers entirely,

      I’m not following this. The confidence interval for the Harvard MPR study for Sep – Dec 2017 covers all of the other point estimates from other studies. Indeed, my recollection is that you criticized the Harvard study because you felt the confidence interval was too large, but here you seem to have forgotten about the confidence interval.

      In your study you estimated expected deaths sans hurricane using various linear trends; however, in the Cruz-Cano and Mead paper they emphatically reject a linear trend in favor of the old Box-Jenkins Airline Model; i.e., ARIMA(1,1,1 1,1,1). Those are two very different worlds.

      There is virtually no way to generate excess deaths on-island in Jan-Feb

      This is not obvious, so you should probably explain yourself. Why would it be possible to have excess deaths on 31 Dec 2017 but not 1 Jan 2018???

      1. Steven Kopits

        “There is virtually no way to generate excess deaths on-island in Jan-Feb. This is not obvious, so you should probably explain yourself.”

        At the link on the comment to which you replied, please see the second graph, “Deaths in Puerto Rico: Jan-Feb of Various Years”. The data and analysis underpinning these numbers can be found on the spreadsheet linked at the bottom of the post. See tab ‘Deaths’ from line 64.

      2. Steven Kopits

        As for confidence intervals, you are correct that the MPR bands are so wide as to capture just about anything–which means it’s essentially useless for decision-making. Here’s the text from the study:

        “From the survey data, we estimated a mortality rate of 14.3 deaths (95% confidence interval [CI], 9.8 to 18.9) per 1000 persons from September 20 through December 31, 2017. This rate yielded a total of 4645 excess deaths during this period (95% CI, 793 to 8498), equivalent to a 62% increase in the mortality rate as compared with the same period in 2016. However, this number is likely to be an underestimate because of survivor bias. The mortality rate remained high through the end of December 2017, and one third of the deaths were attributed to delayed or interrupted health care.”

        The mean estimate is well off Menzie’s graph.

        This is the devastating sentence: “However, this number [of excess deaths] is likely to be an underestimate because of survivor bias.” This statement was wildly inaccurate, and knowable to be inaccurate from data found within the MPR study itself.

        The confidence intervals are also useless. Recorded excess deaths were already well above the 793 lower bound by the time the study was published; and notion of nearly 8500 excess deaths at the high end is utterly absurd. So you have a study with a mean estimate firmly established in la la land, and upper and lower bounds to the confidence interval which are provably impossible on either side.

        Again, I have no objection to trying a new method to gather statistical data in the absence of other more reliable approaches. The Harvard study, however, was grossly oversold, both in its own text, and even more importantly, in the media.

      3. Steven Kopits

        Finally, let me comment on confidence intervals.

        On Menzie’s graph above, note the confidence intervals for the Milken (GWU) study.

        The lower bound confidence interval, even in December, is well above mine, or that of Santos-Lozado Howard, or even two of three of Menzie’s OLS lines.

        The Milken GWU February numbers look well into implausible land. The lower bound of the confidence interval is well above Menzie’s highest estimate.

        Thus, one is left with the impression that the whole Milken construct is wrong and the confidence intervals are flat-out incorrect. And yet, a 95% CI would have you believe the real value for the population must be somewhere in that interval, with overwhelming probability. Thus, a CI can give the impression of accuracy and precision where neither is true in the least. Statistical analysis is valid if the data is reliable and the model appropriately constructed. If it’s not, then it’s all GIGO, and the CI is misleading, rather than informative.

  7. joseph

    Menzie Chinn is being overly generous to Kopits by not plotting his original absurd estimate of 300-400.

    And to this day, even as the count moved up into the thousands, Kopits insists that these are not “excess deaths”, they were merely “premature deaths.” He’s a monster.

    1. Steven Kopits

      The 300-400 (did I revise that?) were as of Oct. 2018. I personally would be quite interested to see how these numbers developed. I would add these would have to be adjusted for the whole PROMESA (probably better named AMENAZA) thing. But, yes, the issue remains as to whether these were principally premature deaths or additional deaths. I think the one year horizon would make a reasonable time frame for an estimate. Again, adjustments for interim population changes and effects of austerity would be material, I would think.

      1. Menzie Chinn Post author

        Steven Kopits: On 5/31 you gave a definitive estimate of no more than 300-400 by October 2018. On 6/1 you admitted you misinterpreted the data, and raised your estimate to 1400 in light of additional data. A discussion of PROMESA and exchange between you and me is here. I let readers decide what estimates make more sense, including those regarding pre-Maria excess mortality.

          1. Moses Herzog

            @ “Princeton”Kopits
            A blog link directly to bullcr*p is not a “model”. Unless you’re willing to give Betsy Devos some Sallie Mae money she can run off to the bank with, and then she’ll let you call it “The FauxEmpirical Buttmunch” or whatever else strikes your fancy.

    2. pgl

      Ah yes – Princeton Stephen are now arguing it was their fault because they were OLD.

      “He’s a monster.”

      Monsters worldwide are insulted at this comparison.

      1. Moses Herzog

        I’ll never be able to do a 3.a.m. chug-a-lug of Monster drink again. (joking I can, I can drink 2 even)

  8. joseph

    Oh, and the major reason for Trump’s refusal to fund recovery efforts? In his delusional brain he is convinced that the money will be used to pay down Puerto Rico’s debts. Trump, of course, has stated publicly that he favors defaulting on the debt, because stiffing his creditors has been his business operating practice for decades.

  9. Moses Herzog

    Does anyone have an accounting of the average age of those who have been “officially” tabulated by the Puerto Rican government?? No, “Princeton”Kopits, I’m not asking you your opinion or for a link to your trash blog. But it’s kind of you for looking for any excuse.

    If the age of the average death fell below the average Puerto Rican citizen lifespan say for the last 15 years, wouldn’t that strongly imply other causes??

    1. Moses Herzog

      —–In the last sentence of my above comment I mean to say “other causes than old age or ‘death by natural causes’ ” obviously.

    2. pgl

      ‘“Princeton”Kopits, I’m not asking you your opinion or for a link to your trash blog. ‘

      Look above – he has a link to his trash blog begging us to read his latest intellectual garbage. Yea – I’m skipping it too as I always do!

  10. Moses Herzog

    OK, this is admittedly a trivial item taken in context with the other things such as obstruction of justice by donald trump and later his quasi-personal defense attorney Bill Barr’s obstruction of justice in his attempts to filter the Mueller Report. But an item on page 28 I thought was humorous. Even “shocking”. The great and patriotic Sean Hannity retweeted political comments made from Russian controlled accounts multiple times. Gee, imagine that. Also “the great patriot” Roger Stone was retweeting Russian accounts, in addition to Stone’s interactions with Guccifer 2.0. Roger Stone would sell his own mother into prostitution for his DC taxi-fare or a paypal donation for a Richard Nixon bobblehead. Sean Hannity wouldn’t sell his own Mom though. Hannity would just encourage one of his white trash listeners to sell their Mom for a $50 pillow plus shipping.

  11. joseph

    Kopits: “Cruz-Cano and Mead is largely in line with our earlier assertions.”

    Our? Who talks like that? The queen? Or pretentious hacks?

    1. Moses Herzog

      @ joseph
      Be gentle. This is a sensitive time for “Princeton”Kopits. The unverified rumor is his “Uncle” Larry was arrested recently:

      I hope “Princeton”Kopits turns a negative into a positive, fights his way through this emotionally grueling time, and realizes even some of the weakest characters and rock-bottom losers in life can make money with Paypal and Crowdfunding. This could be an impetus for great things ahead for “Princeton”Kopits. A guiding light.

      “Uncle” Larry—“Princeton” before “Princeton” was “Princeton”—this one’s for you buddy:

Comments are closed.