Bacon's Rebellion

Democracy Thrives in Sunlight

Lies, Damn Lies, and Race-Obsessed Statistics

—

by

James A. Bacon
in ,

by Carol J. Bova

A March 3 Virginia Department of Health (VDH) blog post discusses racial/ethnic “health and disease” disparities in light of the COVID-19 epidemic. It states that COVID case rates and hospitalization rates for blacks and Hispanics in the United States and Virginia are substantially higher than for whites.

“Social determinants of health are maldistributed,” concludes the blog post. “These disparities will continue health problem by health problem until there is more equity in the distribution of social determinants of health across racial/ethnic groups.”

Not so fast. There are two problems with this framing of the issue. First, the Northam administration’s obsessive focus on the color of peoples’ skin distracts from targeting the real factors influencing COVID mortality such as rates of obesity and diabetes. The second is that, ironically, VDH isn’t even doing a good job of measuring race. The assertions about differential case and hospitalization rates are based on deficient data.

As an example of the data issues Virginia is dealing with, the Virginia Hospital and Healthcare Association report of COVID hospitalizations shows 49,126 persons were discharged and 1,013 were still hospitalized as of March 14. The VDH hospitalization number of 25,323 is half that. because they count only those hospitalized when they first learn of a case If VDH is ignoring half of all hospitalizations, can any conclusions be drawn from the data?

Similarly, a Feb. 5 Centers for Disease Control (CDC) Morbidity and Mortality Weekly Report (MMWR), “Demographic Characteristics of Persons Vaccinated During the First Month of the COVID-19 Vaccination Program,” highlights problems with the statistics underpinning assumptions about racial disparities.

While data on sex was reported for 97% of recipients and age for 99.9%, the CDC article stated that data on race/ethnicity was available for only 51.9%. The range across jurisdictions administering vaccine doses was 0.2% to 100%. The large percentage of unknown race/ethnicity, says the CDC, means that “the findings presented in this study might not be generalizable to all persons initiating COVID-19 vaccination in the United States.”

VDH does concede some limitations to its data. In the fine print of a chart showing the Age-Adjusted COVID-19 Death Rates by Race/Ethnicity, VDH indicates that it uses the 2019 National Center for Health Statistics for “bridged” race and ethnicity, and the deaths per 100,000 are age adjusted to 2000 U.S. population standards.

Then there’s the problem of unreported data. The VDH Death Rates chart goes through February 22, 2021, and on that day, VDH was still saying, “VDH is now processing 2021 death certificates related to the post-holiday surge of COVID-19 cases. As a result, a larger number of deaths is expected to be added by report date.” How many of the 2,402 deaths added to the total since then were prior to February 22?

VDH also acknowledges “missing race and ethnicity data continues to be an issue. While death records are mostly complete, the case, hospitalization, and vaccination data reported above still reflect significant amounts of missing data. VDH has adopted an imputation method to fill these gaps. More information using this method will be available in the coming weeks.” (My emphasis.)

An article in the American Journal of Clinical Nutrition said: “The highest rates of obesity occur among population groups with the highest poverty rates and the least education.” Again, obesity and conditions associated with it such as diabetes and hypertension increase the risk of COVID-19 complications and death. The CDC notes that the risk factors can affect all adults, not only people of color.

Instead of working with flawed racial/ethnic demographics, VDH should target vaccine efforts to neighborhoods with high rates of poverty where COVID-19 risk conditions are most likely to be found — whatever their racial/ethnic composition. But don’t hold your breath. That approach might get vaccinations to people who need them most, but it won’t advance the Governor’s racial disparity agenda.

Share this article

ADVERTISEMENT

(comments below)

ADVERTISEMENT

(comments below)

Comments

24 responses to “Lies, Damn Lies, and Race-Obsessed Statistics”

  1. James C. Sherlock Avatar
    James C. Sherlock

    Great reporting.

  2. LarrytheG Avatar
    LarrytheG

    ” Few states are accurately tracking coronavirus vaccinations by race. Some aren’t at all.

    Most states are not publicly reporting racial data on people receiving coronavirus vaccines, despite disproportionate covid-19 death rates for Black and Hispanic people and rising concerns about who has access to — and is willing to take — the vaccine.

    Seven weeks after the first shots were administered, just 20 states include race and ethnicity data on their vaccine dashboards, even though it is required by the federal government. And even those states have major gaps in their data, with “unknown” being the first- or second-most frequent category in almost every state.

    https://www.washingtonpost.com/local/social-issues/race-data-vaccine-states/2021/01/27/fc9df6de-6001-11eb-afbe-9a11a127d146_story.html

  3. Dick Hall-Sizemore Avatar
    Dick Hall-Sizemore

    Carol, I appreciate the time you are spending on this reporting and the detail you provide.

    Like you and probably everyone else, I find the confusion in the data frustrating. But, I am not sure that it is all VDH’s fault. Take the discrepancy between VDH and the Hospital Association regarding hospitalizations, for instance. Are hospitals required to report COVID hospitalizations, or any hospitalizations, for that matter, to the state? If not, that is probably the reason for the discrepancy.

    I also sometimes get tired of the frequent emphasis on race. I agree with you that communities of poverty best defines those most vulnerable to COVID. But poverty and race are highly correlated; therefore, describing the situation from the perspective of race is virtually the same as using poverty. Furthermore, because the needs of Black communities have been historically overlooked, it is understandable that there is emphasis on race now.

    1. CJBova Avatar
      CJBova

      Using percentages of estimates of total Virginia population by race/ethnicity, and rounding off comparable percentages of those in poverty by race/ethnicity from the Kaiser Family Foundation reports,

      16.1% or about 258,000 Black Virginians and

      12.5% or 102,000 Hispanic Virginians are in poverty.

      This overlooks 7.7% or 402,000 White Virginians who are impoverished, and about

      7% or 39,00 Asians, and about

      43,000 Native Americans, Other Races, or Two+ races who are also in poverty.

      Altogether, about 9.9% of 8,52MM or 843,500 Virginians are impoverished.

      1. Dick Hall-Sizemore Avatar
        Dick Hall-Sizemore

        Looked at another way, about 43% of the impoverished are Black or Hispanic and 48% are white. It looks like my perception of the category of “impoverished” as being a surrogate for minority populations is faulty.

        However, as I think about this more, there is another justification for the emphasis on race. Although “Blacks and browns” and whites comprise roughly the same proportions of the impoverished, members of the former group have a higher incidence of COVID infection and death, other than residents of nursing homes and assisted living centers. I am assuming that is because they are the ones who dis proportionally had to go out to work and could not work from home. They are the grocery store clerks, bus drivers, maintenance crews, etc.

        1. CJBova Avatar
          CJBova

          We don’t know the actual incidence of cases by race because 22% are unidentified. Look at the post again. The death statistics which are 98.6% identified by race/ethnicity show 2.6 times as many White persons died as Black persons. Or twice as many White deaths as Black and Hispanic combined. If you want to compare deaths to percentage of population:
          Asian 6.5% of population 3.7% CV deaths.

          Black 18.8% population. 23.9% deaths
          Latino 9.5% population 6.6% deaths

          Whites 61.3% population, 63.2% deaths

          The numbers have shifted since last July, but I wrote then about the number of Hispanic deaths being so out of proportion for the 35 to 64 age group. And I raised the issues of comorbidities, the Hispanic work ethic, and type of work in long term care and meat or poultry processing plants as possible factors. I couldn’t do anything about the work places, but I did arrange with a pastor in the Northern Neck with a mostly Hispanic congregation to have him share where his congregants could go for free health services to deal with diabetes and other health issues that put them at risk.

          I said then, and I say now, “Everyone needs to stop blaming systemic racism as a cause and relying on
          news media and politicians posturing about inequities. We need to
          demand the Office of Health Equity investigate and document the facts
          and then formulate real life solutions.”

      2. WayneS Avatar
        WayneS

        What would be nice to know is how closely the infection rates and death rates among impoverished blacks, impoverished hispanics and impoverished whites follow their respective percentages of the population.

        1. CJBova Avatar
          CJBova

          Income isn’t on the reporting form. Don’t know if there’s a way to identify zip codes with income levels and then you could compare COVID numbers by zip code.

          1. WayneS Avatar
            WayneS

            That’s why I said it would be nice to know. I don’t expect to ever know it. 🙂

    2. CJBova Avatar
      CJBova

      To answer your questions, Dick, there are reportable diseases which must be reported within three days by any healthcare provider of suspicion or confirmation, and rapidly reportable diseases which “require immediate communication to the local health department by the most rapid means available upon suspicion or confirmation.” Severe coronavirus (SARS-CoV) infection is in the rapidly reportable group, preferably by telephone. The details are in “Regulations for Disease Reporting and Control.”

      The form for reporting is the Confidential Morbidity Report, and it asks for full details, whether deceased or hospitalized, what tests, what symptoms, what treatments, with full patient, provider and lab details and contacts.

      So yes, VDH should have been notified of every COVID hospitalization. It has been the VDH policy since the beginning of the pandemic to only track hospitalization if that is their first report of the illness.

      1. Dick Hall-Sizemore Avatar
        Dick Hall-Sizemore

        I assume that the phrase “first report of the illness” is significant. So, if someone tests positive, which is reported, and is later hospitalized, she does not show up in the VDH’s hospitalization list? Similarly, if someone is hospitalized, which is reported, then discharged, but has to be readmitted, she does not show up on the VDH hospitalization list on the second go-around?

        1. LarrytheG Avatar
          LarrytheG

          There is a term used in IT and database technology called “data normalization”. It’s essentially a precise definition of what each data is (and is not), how it is collected and incorporated and provided as a product on the other end.

          It’s what IT needs in order to deploy a database that provides correct and consistent data.

          But IT cannot “force” those who are providing the data to necessarily follow the “rules”. When data comes from all over the place from a lot of different people and venues… it’s like herding cats.

          Think of it this way. You’re filling in a blank on a form and it asks for a date of some action and that date needs to be consistent with other tasks and dates and you put in a date that is clearly not the date intended.

          OR the form asks for something like race AND if you don’t provide, it refuses to accept the record.

          What is worse? Not getting the record at all or a incomplete record?

          Spread this out across the state a thousand times and you get the sense of the problem – not only in Virginia – in many states collecting data from all over the state.

          It’s just not a perfect system – and really, not likely to become one.

          1. CJBova Avatar
            CJBova

            Larry, this is nonsense. The reports are not identified by date alone. VDH accepts incomplete reports and that’s what goes into the datasets. The VDH form is the same for all reporting. That’s how VDH is able (according to them) to only count one positive test per person, no matter how many times they’re tested. The earlier backlogs of entries started with faxed reports that had to be manually entered. Now, all the forms are supposed to go directly into the electronic system from the health care provider.

            That doesn’t excuse using partial data and declaring unsupported conclusions that dictate policy.

            It’s up to the Commissioner of Health to see that health regulations are enforced.

          2. LarrytheG Avatar
            LarrytheG

            Carol – I was giving an EXAMPLE of how inconsistent or incorrect data can get into the database from disparate field sites.

            The actual data definitions and rules are administratively done, then implemented by IT.

            You have a choice on folks filling in forms. You can reject them if they are incomplete or you can accept them as incomplete. Either way, there are downsides and the data is not perfect.

            In terms of what the results are used for – politically or not – that’s beyond the realm of IT. They slice and dice the data per the administrative policy and rules then others grab that data – imperfect as it might be and use it for political purposes.

            AGAIN – Virginia is the norm – not the outlier on these issues. They are COMMON in many other states and vary only by degree or issue.

        2. CJBova Avatar
          CJBova

          As far as VDH numbers go, that person only shows up as a case, unless they die. Then they’d show as a VDH case and death. Never as hospitalized…unless the hospital reports the readmission as a new case.

  4. LarrytheG Avatar
    LarrytheG

    In terms of what is going on in Virginia with Northam and the rest of the country – does not look like Virginia/Northam are really that far out from the others.

    I found this article interesting given all the stuff aimed at Va and Northam from critics:

    ” Faulty Software Snarls Vaccine Sign-Ups
    Health departments continue to grapple with delays caused by technical problems with numerous websites used for making appointments.

    When coronavirus vaccines first became available, state health officials in Virginia turned to software recommended by the Centers for Disease Control and Prevention to schedule appointments. But people complained that the software, called VAMS, was too confusing for older adults to use.

    So the state switched to another system, PrepMod — but that had problems, too. Links sent to seniors for their appointments were reusable and found their way to Facebook, leading to one vaccination event in Richmond with dozens of overbookings. Some of those people threatened health care workers when they were turned away.

    “It was a nightmare scenario,” said Ruth Morrison, the policy director for the Richmond and Henrico County health district. “People showing up confused, irate, thinking they had an appointment.”

    https://nyti.ms/38A40RY

    1. LarrytheG Avatar
      LarrytheG

      ” State and local health departments around the country continue to face delays dispensing shots, in part because flaws remain in the appointment software tools like those used in Richmond. The problems threaten to slow the vaccine rollout even as supplies and distribution are picking up quickly across the country.”

      1. CJBova Avatar
        CJBova

        True, Larry. But vaccine scheduling software is not the topic of this post.

        1. LarrytheG Avatar
          LarrytheG

          It’s in the same church different pew though in terms of criticism of VDH, Northam and Virginia.

          Few states are doing this “right” – it’s not that Virginia is terrible among the states and that’s what I object to in commentary here.

          There are many parts to “IT” – but think of it this way. If an organization has complete control of the data from it’s source then they could be criticized for their flubs. But when data is being “collected” from all over the place from different kinds of entities and the “theory” is that they are “supposed” to provide it on a timeframe – then IT does not have real control of the data.

          So for example, if the source of the data does not provide some attributes, like race, how would IT deal with that?

          How would ANY state health department deal with it ?

          I just think the continuing drumbeat against VDH and Virginia on this as if Virginia alone had these problems is overly critical for what it is.

          It’s a problem and it needs to be fixed but it’s not exactly a “failure of govt” or on that level worth all the condemnation they are receiving.

          IMHO.

  5. StarboardLift Avatar
    StarboardLift

    The COVID data in Virginia and elsewhere may be fubar and even if not, it is too soon to make conclusions about race. And one underreported feature of the most recent glitch while merging 3 dB back to VAMS is that people deep in 1c got VAMS invite to make appointments for vaccine, and at least some received vaccinations. Frustrating for the 80 somethings still wrestling with the system.

    1. CJBova Avatar
      CJBova

      You’re right. The data for deaths will eventually be available, but historically, it takes two years to get the CDC analysis from death certificates. Some stats, like case numbers by race, will never be sorted out.

      The amount of money the Commonwealth pays out for IT consultants and services that don’t do the job is a criminal waste.

  6. Eric the half a troll Avatar
    Eric the half a troll

    “If VDH is ignoring half of all hospitalizations, can any conclusions be drawn from the data?”

    Unless there is something about how they are ignoring them that skews the data, yes, a 50% sampling of the data would be more than adequate to draw reliable conclusions.

  7. Stephen Haner Avatar
    Stephen Haner

    Thank you, Carol, for raising issues that few are willing to touch. I had a half-written piece following one of the more ridiculous RTD racist rants but never finished or posted, dreading the response. There is too much political capital to be mined by making this all about race. Not so much if you make it about poverty and other health conditions. Even the medical establishment is afraid to discuss the role of obesity. CDC had some data up on that, the additional risk it caused, but a few days later I couldn’t find it.

    But the conventional wisdom that this disease is twice, or even more than twice as deadly for black Americans is deeply ingrained now. You hear it all the time. I look at the case/death data and see little variation in “death rate.” One of the better questions is why the rate is so much lower for Hispanics. High cases, but lower deaths. Perhaps just a reflection of lower median age.

    1. CJBova Avatar
      CJBova

      In the next to the last paragraph, the link at COVID-19 will take you to the CDC page on obesity and CV.

Leave a Reply

Read More