Coronavirus disparities in Palestine and in Michigan

I wanted to share two articles that were sent to me recently, one focusing on data collection and one focusing on data analysis.

On the International Statistical Institute blog, Ola Awad writes:

The Palestinian economy is micro — with the majority of establishments employing less than 10 workers, and the informal sector making up about a third of the economy. It is primarily a service-based economy, and also a consumer-based economy with a consumption rate of about 116% of the gross domestic product. . . . Those most affected are the most vulnerable sectors of society such us Palestinians living in refugee camps[3] and in Area C. This area is home to an estimated 180,000-300,000 Palestinians who are suffering from demolitions and forced evictions that deprive people of their homes and disrupt livelihoods, leading to entrenched poverty and increased dependence on aid. . . .

About 29% of those employed in the private sector receive less than the minimum wage (1,450 NIS = USD 426 per month), and to make matters worse — 57% of the employed are considered informal employees, meaning they do not receive formal basic work rights such as employment contracts, paid leave, sick leave or social retirement.

The fragile economy faced even more tragic conditions after the COVID-19 pandemic hit Palestine and the rest of the world. . . . During the lock-down, face-to-face data collection was no longer an option and we had to come up with creative ways like hand held devices, phones, and the use of registers when available. We also performed a rapid assessment to offer real time data capturing the effect of the pandemic. . . . It is vital for governments to have data on the most marginalized groups which are expected to fall deeper into vulnerability due to the pandemic. In Palestine, this includes women heading households, workers of the informal sector, and workers at Israeli settlements, refugees, and the population in Area C. . . .

The lockdown has widened the poverty gap. Families who were on the edge are falling into poverty, leading to the emergence of new groups of poor people, especially in refugee camps and Area C. Around 109,000 women working in the private sector have lost their jobs due to closure measures. . . .

Meanwhile, at the University of Michigan, Jon Zelner, Rob Trangucci, Ramya Naraharisetti, Alex Cao, Ryan Malosh, Kelly Broen, Nina Masters, Paul Delamater write:

Racial disparities in COVID-19 mortality are driven by unequal infection risks.

Geographic, racial-ethnic, age and socioeconomic disparities in exposure and mortality are key features of the first and second wave of the U.S. COVID-19 epidemic. We used individual-level COVID-19 incidence and mortality data from the U.S. state of Michigan to estimate age-specific incidence and mortality rates by race/ethnic group. Data were analyzed using hierarchical Bayesian regression models [using rstanarm], and model results were validated using posterior predictive checks. In crude and age-standardized analyses we found rates of incidence and mortality more than twice as high than Whites for all groups other than Native Americans. Of these, Blacks experienced the greatest burden . . . We also found that the bulk of the disparity in mortality between Blacks and Whites is driven by dramatically higher rates of COVID-19 infection across all age groups, particularly among older adults, rather than age-specific variation in case-fatality rates. Interpretation. This work suggests that well-documented racial disparities in COVID-19 mortality in hard-hit settings, such as the U.S. state of Michigan, are driven primarily by variation in household, community and workplace exposure rather than case-fatality rates.

P.S. In response to some comments (see below), Zelner writes:

I [Zelner] agree 100% about the challenges of knowing what is driven by testing vs infection. But to the naysayers I’d say good luck getting the negative test data you’d need to figure it out! This has been a huge challenge and basically every chance I have I harass the people who could give it to us, but it is actually a legal and logistical morass to get it… For me it’s sort of the perfect vs the good. Plus the effect sizes are so enormous it’s hard for me to believe it’s all testing. Re: pop density I’d reject that out of hand; Dense population in SE Michigan is like a suburb anywhere else.

41 thoughts on “Coronavirus disparities in Palestine and in Michigan

  1. It seems that michigan is 80% white but the most population dense area (ie, detroit) is 80% black and that is where most of the cases were:

    https://www.michigan.gov/coronavirus/0,9753,7-406-98163_98173—,00.html
    https://datamapper.geo.census.gov/map.html
    http://2.bp.blogspot.com/-SJtKrucwlp4/T4cve2PNarI/AAAAAAAAAEQ/B-26rRtSAuc/s1600/Choroplethic_map.jpg

    So I don’t see why there is a mystery about more cases in blacks in Michigan.

  2. There are huge differences in the mortality rates vs reported case rates between the first say 100 or 150 days of the pandemic and the current period (say from 150 days onward). In the case of the state of Michigan, according the the Johns-Hopkins data, the change has been a factor of ten or maybe a little more (more commonly in the US, the change has been between a factor of 2 and 5). Similar changes can be seen in data from Western Europe.

    One can imagine many plausible reasons for these changes, but I would say that any study that tries to make claims such as those reported in this post first needs to provide a reasonably well supported explanation for these changes.

    • Could you cite a source so I can see what “huge changes” you are talking about?
      There have been changes in some European countries in the crude Case Fatality Rate because of changes in how infections and deaths are tracked; e.g. France used to only count hospitalized cases.

      For more general aspects, there’s a “rasonably well supported explanation” here: https://www.psychiatrictimes.com/view/covid-19-has-mortality-rate-declined

      There are definitely variations in the underlying mortality, but I wouldn’t call them “huge”.

      • In the US mortality rate is about 25-50% what it was back in April (see eg page 2-3 xayadata.com/covidstates.pdf). This is mainly due to:

        1) Do not legally require nursing homes to take covid patients
        2) Do not put covid patients on ventilators right away

        If they also gave people pulse oximeter and corrected the oxygen deficiency before it became severe, or severe oxygen deficiency with HBOT, and severe vitamin C deficiency with IV vitamin C we would probably have near zero deaths from covid…

        But there really does seem to be no interest in solving this problem unless it helps someone grab political power or makes someone money. I thought when the measurements of vitamin c levels finally got poublished (which took until August, when it was obvious they were low back in Feb) the healthcare industy would be forced to address this but even 95% of covid patients with undetectable levels seems to have had zero effect on what they do.

        100% of severe COVID patients have severe vitamin c deficiency: https://pubmed.ncbi.nlm.nih.gov/32847620/

        50% reduction in mortality using either 25 g IV vitamin C per day or HBOT:
        https://www.researchsquare.com/article/rs-52778/latest.pdf
        https://www.uhms.org/images/UHM-Journal/PRE-PROOF_-_HBO2_for_COVID_-_47-3_THIRD_QUARTER_2020_print_version_version_47-3.pdf

        Various smaller supporting studies:

        Vitamin C
        https://pubmed.ncbi.nlm.nih.gov/32828741/
        https://pubmed.ncbi.nlm.nih.gov/32709838/

        HBOT
        https://pubmed.ncbi.nlm.nih.gov/32574433/
        https://pubmed.ncbi.nlm.nih.gov/32412891/
        http://www.medgasres.com/preprintarticle.asp?id=282177
        https://www.ihausa.org/covid19-hyperbaric-therapy/

        No one else needs to die of COVID if tomorrow they start correcting these two deficiencies as early as possible.

        • I haven’t evaluated the vitamin C angle closely, but the vitamin D angle looks pretty compelling. The usual caveats of small samples, need to be replicated, etc. all apply, but the observational combined with small RCT here now look like enough to justify vitamin D prophylaxis and treatment (guided by bloodwork ideally to prevent overshoot):
          https://www.sciencedirect.com/science/article/pii/S0960076020302764?via%3Dihub

          The thing about ascorbate is that it is involved in a TON of reactions, and it is not at all surprising that it would be severely depleted under prolonged immune stress. Like vitamin D, it is a low cost intervention. It may not be a silver bullet, but it’s so cheap and safe that should be part of standard of care unless there’s some really good evidence to the contrary…

        • Vitamin C is the terminal extracellular antioxidant and a cofactor required for proper collagen formation (1/3 the protein in your body). When a tissue does not have enough it literally starts disintegrating: https://pubmed.ncbi.nlm.nih.gov/2762330/

          Eventually you will drop dead (as seen in many covid patients):

          One of the striking observations in the older literature is of death being quite sudden in apparently stable individuals. One of the best descriptions of this was in a passage from Ansons’ A Voyage Around the World (1740–1744): “For many of our people, though confined to their hammocks, appeared to have no unconsiderable share of health; for they ate and drank heartily, were cheerful, and talked with much seeming vigor, and with a loud strong tone of voice; and yet on being the least moved, though it was only from one part of the ship to the other, and that in their hammocks, they have immediately expired. And others, who have confided in their seeming strength, and have resolved to get out of their hammocks, have died before they could well reach the deck. And it was no uncommon thing for these who would do some kind of duty, and walk the deck, to drop down dead in an instant on any endeavors to act with their utmost vigour.” 42 Lind also commented that “they are apt, upon being moved, or exposed to fresh air, suddenly to expire.” 4 It is recorded that only 4 men died from enemy action during Anson’s voyage, with more than 1300 dying of disease, which was primarily scurvy. 4

          https://pubmed.ncbi.nlm.nih.gov/30422823/

          In fact many of the symptoms of covid are also symptoms of scurvy:

          Examination showed gingival inflammation (Appendix 1, available at http://www.cmaj.ca/lookup/suppl/doi:10.1503/cmaj.190934/-/DC1) and diffuse lower-extremity nonpalpable purpura in a perifollicular distribution (Figure 1). Laboratory investigations showed microcytic anemia, with normal platelet count and coagulation studies. Serologic testing for vasculitides, hepatitis B and C viruses, HIV and cryocrit were negative. A skin biopsy showed erythrocyte extravasation, hemosiderin deposition, and fibrin thrombi within small blood vessels; there was no evidence of dermatitis or vasculitis.

          https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6989015/

          Covid:

          Other maculopapules (47%). Some of them showed perifollicular distribution and varying degrees of scaling (Fig 2a). Some had been described as similar to pityriasis rosea. Purpura may also be present, either punctiform or on larger areas. A few cases showed infiltrated papules in the extremities, mostly dorsum of the hands, that look pseudovesicular (Fig 2b) or resemble erythema elevatum diutinum or erythema multiforme (Fig 2c).

          https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7267236/

          Additional frequent findings included erythrocyte extravasation (82%) (Fig 3); papillary dermis edema (76%), which was massive in 4 cases, resulting in the formation of subepidermal pseudobullae (Fig 1); endothelial cell swelling (65%) (Fig 3); and moderately increased interstitial mucin deposition (41%). Less frequent, although remarkable, findings included the presence of vascular microthrombi within superficial dermal capillaries and more rarely in dermal venules

          https://www.sciencedirect.com/science/article/pii/S0190962220310227

          Hemosiderin-laden macrophages (4/8), hemorrhage (4/8), mucus aspiration (3/8), emphysema (2/8), and microthrombi (1/8) were seen (Figure 2).

          https://wwwnc.cdc.gov/eid/article/26/9/20-2095_article

          While it is widely suspected that an abnormal host response such as a ‘cytokine storm’ is the driving force in those precariously ill,5 there are peculiarities in radiological findings and ventilator mechanics that are atypical of the usual viral pneumonia and acute respiratory distress syndrome (ARDS). Increasingly, published data and anecdotal observations indicate that the pathogenesis may lie primarily in the pulmonary vasculature with the newly observed tendency for thrombi formation.

          https://journals.sagepub.com/doi/10.1177/2045894020931702

          Vitamin D is more like a hormone that signals a tissue to do something (as far as I know, maybe some other role has been missed). While every vitamin and mineral deficiency should be corrected, vitamin C deficiency is up there with water, oxygen, glucose, etc.

        • Indeed huge differences in action there. I remember reading Linus Pauling back in the day – he argued that ascorbate is basically a missing metabolite given that we only lack a couple enzymes to synthesize it from glucose like most other mammals do. The evidence for benefits of chronic high doses that he advocated is not awesome, but it is interesting to me that benefits for modest amounts have mostly stood the test of time, and that we are now starting to see therapeutic IV-range dosing come back in some circles. Personally I feel better when I supplement ~0.5-1g doses when I start to feel ‘run down’ or craving sour.

        • I have found that when my allergies are bad (which are atypical allergies, not itchy eyes and sneezing, more like atopic dermatitis, eczyma and similar issues) it really helps to take both vit C (500-1000 mg in the range of 1 to 3 times a day) as well as N-actyl-cysteine (NAC) which I use in raw crystal form so I’m not sure the dose exactly, but say 1/4 teaspoon of crystals 1 to 2 times a day in water… it’s probably 200-500mg, tastes a bit nasty but it’s temporary.

          NAC is a precursor to glutathione which is the main cell-internal antioxidant. It also directly reduces mucus production, so it’s helpful that way too. If I remember correctly vit C is used in a recycling reaction to convert spent glutathione back to active form as well.

          In college I was once diagnosed with Vit C deficiency (had micro-bruising all around my arms after a day of minor exercise moving boxes between dorm rooms, and complaints of bleeding gums). Eating vit C tablets twice a day solved all that within 3 days. My wife now gives a talk at med school about how her husband is a scurvy sea dog (I am a sailor, at least, as often as I can).

          I’m convinced those with high levels of immune activity like atopic dermatitis, rheumatoid arthritis, irritable bowel syndrome, etc etc they should all be taking both vit C and NAC in the hundreds of mg per day range regularly (not necessarily every day, but at least a few times a week)

          My impression is that the mechanism is Immune activity -> production of oxidative chemicals -> depletion of antioxidants.

          Vit D seems to work by modulating the immune activity, so you don’t produce so much toxic chemistry… Vit C and NAD seem to work by replenishing antioxidant power.

          For people without high levels of immune activity, the effect of vit C and NAC is probably minimal. So if you just run stupid NHST type experiments on the general population, you’d likely conclude they do nothing much.

        • Hi Daniel, yes that makes sense to me. Ascorbate indeed involved in recycling glutathione. My guess is that immune activation or inflammatory stress of any sort increases need for/benefit from it. I eased back on supplementation (from 1-2g/day down to 0-500 mg) a few years ago when a few studies came out suggesting that could blunt molecular signaling needed to benefit from exercise (at least in mouse models lol). The whole hormesis concept, etc. Recently my solution to that is just not supplement in the hours immediately before or after exercising. Blood levels normalize pretty quickly so hard to imagine it having impact outside that windown.

        • The evidence for benefits of chronic high doses that he advocated is not awesome, but it is interesting to me that benefits for modest amounts have mostly stood the test of time, and that we are now starting to see therapeutic IV-range dosing come back in some circles. Personally I feel better when I supplement ~0.5-1g doses when I start to feel ‘run down’ or craving sour.

          Yes, the mistake that was made in the vitamin C literature IMO is looking at pharmacokinetics in healthy people and then extrapolating to sick people.

          It is true that if you take more than your body needs you either don’t absorb it so it ends up in diarrhea or what does get absorbed ends up in the urine. For healthy people about 100 mg/day oral vitamin C seems to be sufficient: https://pubmed.ncbi.nlm.nih.gov/8623000/

          However, things are very different when you get sick or injured and oxidative stress levels rise. Its now well established that at least 2-3 grams per day IV (20-30x the oral dose required for a healthy person) is needed for critically ill people to avoid deficiency:

          Overall, the critically ill patients exhibited hypovitaminosis C (i.e., < 23 μmol/L), with a mean plasma vitamin C concentration of 17.8 ± 8.7 μmol/L; of these, one-third had vitamin C deficiency (i.e., < 11 μmol/L). Patients with hypovitaminosis C had elevated inflammation (C-reactive protein levels; P < 0.05). The patients with septic shock had lower vitamin C concentrations and higher C-reactive protein concentrations than the non-septic patients (P < 0.05). Nearly 40% of the septic shock patients were deficient in vitamin C, compared with 25% of the non-septic patients. These low vitamin C levels were apparent despite receiving recommended intakes via enteral and/or parenteral nutritional therapy (mean 125 mg/d).

          https://pubmed.ncbi.nlm.nih.gov/29228951/

          Our study confirms previous findings that intravenous doses of 2 to 3g/d are required to normalize vitamin-C plasma concentrations in critically ill patients . However, optimal plasma concentrations during overwhelming oxidative stress are not known, nor whether peaks are more effective than lower but stable plasma concentrations. Two recent studies using 100mg/kg/day and 6g/day as intermittent boluses in sepsis reported a reduction in mortality, but these findings as well as the dose-effect relationship need confirmation in larger randomized controlled trials15,16.

          https://pubmed.ncbi.nlm.nih.gov/29522710/

          Our study implies vitamin C hypovitaminosis or deficiency are common in septic shock patients and that 6-hourly dosing at 1.5 g intravenous of vitamin C achieves and maintains normal or supranormal vitamin C levels in patients with septic shock. The observation of an increase in volume of distribution in the multiple dose cohort (39.9 L) compared with the first-dose cohort (23.3 L) implies either changes in volume of distribution secondary to fluid resuscitation or other septic shock-associated mechanisms such as changes in vascular permeability.42,43 The observation that clearance was reduced in the multidose cohort implies either loss of renal clearance due to worsening function or decreased vitamin C consumption, as treatment with antibiotics decreased oxidative stress.

          https://pubmed.ncbi.nlm.nih.gov/31778629/

          Plasma vitamin C levels in all patients were subnormal at enrollment (<28 μmol/L), with no significant difference between groups (median for vitamin C–infused patients vs placebo, 22 vs 22 μmol/L [interquartile range {IQR}, 8-39 vs 11-37]; P = .49). (To convert vitamin C to mg/dL, divide values by 56.78.) Plasma vitamin C levels sampled at trough periods before infusion had increased significantly in vitamin C–infused patients by hours 48 (median, 166 μmol/L; IQR, 88-376) and 96 (median, 169 μmol/L; IQR, 87-412) compared with placebo patients by hours 48 (median, 23 μmol/L; IQR, 9-37) and 96 (median, 26 μmol/L; IQR, 9-41) (Figure 3). At hour 168, plasma vitamin C level in placebo patients remained low (median, 29 μmol/L; IQR, 12-39). After cessation of vitamin C infusion at 96 hours, vitamin C levels declined but remained significantly elevated at hour 168 (median, 46 μmol/L; IQR, 19-66) compared with placebo

          https://jamanetwork.com/journals/jama/fullarticle/2752063

          Correctly timed baseline vitamin C levels were available for 22 patients in the treatment group; the mean level was 14.1 +/- 11.8 uM (normal, 40-60 uM), with no patient having a normal level.

          https://journal.chestnet.org/article/S0012-3692(16)62564-3/fulltext

          Our study confirms previous findings that IV doses of 2 to 3 g/d are required to normalize vitamin C plasma concentrations in patients who are critically ill. 6 However, optimal plasma concentrations during overwhelming oxidative stress are not known, nor is whether peaks are more effective than lower but stable plasma concentrations. Two recent studies using 100 mg/kg/d and 6 g/d as intermittent boluses in patients with sepsis reported a reduction in mortality,but these findings and the dose-effect relationship need confirmation in larger randomized controlled trials. 15,16

          https://pubmed.ncbi.nlm.nih.gov/29522710/

          Our study implies vitamin C hypovitaminosis or deficiency are common in septic shock patients and that 6-hourly dosing at 1.5 g intravenous of vitamin C achieves and maintains normal or supranormal vitamin C levels in patients with septic shock.

          https://pubmed.ncbi.nlm.nih.gov/31778629/

          Patients with pneumonia had depleted vitamin C status compared with healthy controls (23 ± 14 μmol/L vs 56 ± 24 μmol/L, P <0.001). The more severe patients in the ICU had significantly lower vitamin C status than those recruited through AMAU (11 ± 3 μ mol/L vs 24 ± 14 μmol/L, P = 0.02). The total pneumonia cohort comprised 62% with hypovitaminosis C and 22% with deficiency, compared with only 8% hypovitaminosis C and no cases of deficiency in the healthy controls. The pneumonia cohort also exhibited significantly elevated protein carbonyl concentrations compared with the healthy controls (P < 0.001), indicating enhanced oxidative stress in the patients. We were able to collect subsequent samples form 28% of the cohort (mean 2.7 ± 1.7 days; range 1-7 days). These showed no significant differences in vitamin C status or protein carbonyl concentrations compared with baseline values (P = 0.6). Overall, the depleted vitamin C status and elevated oxidative stress observed in the patients with pneumonia indicates an enhanced requirement for the vitamin during their illness.

          […]

          Although the samples in our study were collected prior to the SARS-CoV-2 outbreak, it is likely that people with COVID-19-associated pneumonia and sepsis would have similarly low vitamin C status and high oxidative stress. Early case reports from the 1940s indicated that IV administration of gram doses of vitamin C to cases of viral pneumonia rapidly improved common symptoms [35]. There are currently a number of intervention trials up and running around the world which will specifically test IV vitamin C for COVID-19-related pneumonia and sepsis. Furthermore, it is likely that patients with other severe infectious conditions may also have low vitamin C status. This has been previously demonstrated in patients with tuberculosis, bacterial meningitis, tetanus and typhoid fever [11, 12]. These patients would also likely benefit from additional vitamin C supplementation.

          https://pubmed.ncbi.nlm.nih.gov/32384616/

          Also, most of the pharamacokinetics is done on IV sodium ascorbate. I’ve still yet to see anyone check the original claims that people with various illnesses could take tens to hundreds of grams of ascorbic acid (NB: not sodium ascorbate*) orally per day to suppress their symptoms:

          In 1970, I discovered that the sicker a patient was, the more ascorbic acid he would tolerate by mouth before diarrhea was produced. At least 80% of adult patients will tolerate 10 to 15 grams of ascorbic acid fine crystals in l/2 cup water divided into 4 doses per 24 hours without having diarrhea. The astonishing finding was that all patients, tolerant of ascorbic acid, can take greater amounts of the substance orally without having diarrhea when ill or under stress. This increased tolerance is somewhat proportional to the toxicity of the disease being treated. Tolerance is increased some by stress (e.g., anxiety, exercise, heat, cold, etc.)(see FIGURE I). Admittedly, increasing the frequency of doses increases tolerance perhaps to half again as much, but the tolerances of sometimes over 200 grams per 24 hours were totally unexpected.

          https://pubmed.ncbi.nlm.nih.gov/7321921/

          So the failures to check the pharmacokinetics in sick/injured people and not use that to determine the dosing schedule, and only studying sodium ascorbate instead of ascorbic acid* have really set back the study of vitamin C. There is a very large literature of crappy studies that amount to putting random amounts of water on burning houses then saying water doesnt put out house fires.

          * Since it is uncharged at gastric pH, the latter could get absorbed passively and quickly as a bolus in the stomach instead of slowly and actively in the intestine. So the pharmacokinetics and biological effect of large doses could be very different.

        • Thank you for that source!

          It counts people dying “in hospitals in England”. It does not count people dying outside of hospitals, and it does not show age demographics. In Germany, we know that the age demographics of the infections has shifted: as the number of cases went down, the proportion of elderly that were infected also declined, on a delay. (The RKI-published data does look like a high rate of cases in the general population does lead to a high rate of infections in the elderly.) Any fatality rates that don’t consider age demographics are incomplete, as Covid-19 mortality is highly dependent on patient age.

          I also hope we’re seeing benefits from all of these drug trials; some drugs have been reported to improve survival chances quite a bit, so I’d be worried if we didn’t see that (and other improvements in treatment methods) in the data.

        • Oh sure. I am not making any particularly strong claims myself, but there are numbers out there from reputable sources reporting what I would call “huge” changes in mortality. Yes, this could be driven more by age than by improvement in treatment (but, as you say, dexamethasone has been shown in clinical trial to reduce mortality, and remdesivir looks helpful if not quite p. <05 for mortality specifically).

          Here's a paper claiming dramatic mortality drop among hospitalized patients in Italy: https://www.tandfonline.com/doi/pdf/10.1080/20477724.2020.1785782?needAccess=true&amp;

        • The Italy data requires a grain of salt; it’s data from a single hospital in Milan. In the fourth quartile, they admitted only 4 people to the ICU, and only one patient died. In the previous quartiles, more patients died than were admitted to the ICU, and the first quartile was at the height of the outbreak in the region, where severe cases from smaller hospitals in Lombardy may have been transferred to Milan? (Note that there is no decline from the second to the third quartile.) I would not be surprised if it turned out that the mortality decline was in part a consequence of a less severly ill patient population in that hospital as case numbers in Italy had declined.

        • Yeah much of this needs at least a grain of salt, there are tons of confounding factors.

          But I do think that combined with the summer peak in the US being about half the deaths/day of the spring peak, despite those states hit hard in the summer having a larger population than those hit hard in the spring, overall there is decent (though not great) evidence for COVID IFR in “developed Western nations” having fallen.

  3. For the Michigan study, my concern is the same as always. We do not know relative rates of infection in Black and White. We know relative rates of positive tests. They are talking about the different household, community and workplace exposures of Black and White persons which very well may lead to different infection rates. But they can’t know infection rates from their data, therefore any reasoning about mortality conditional on infection is suspect.

    • In addition, all positive tests are not created equal. There have been a number of stories recently about the PR test in particular. Some (many?) labs amplify the amount of virus in the sample so much that the lab reports positive when the amount a virus in the patient is so small that it cannot be infectious. And many of the newer quick tests supposedly miss a lot of true positives.

      Just to add to the mass of confusion.

      • Are you talking about the PCR test? The basic principle of this test (more properly RT-PCR) is to amplify the virus RNA present in the sample to make it detectable. All PCR tests amplify the RNA a huge amount, and always have.
        If you manage to swab an amplifyable amount of RNA from a patient’s throat, you can be fairly sure that it got there because it multiplied in the person itself: the chance of otherwise getting it on the swab and getting it amplified is just too low. Who exactly is claiming that a true positive PCR test can sense an amount that is “not infectious” for the carrier?

        Or are you talking about people who have actually undergone an infection, but where the virus concentration in their samples is so low that they probably won’t infect others, and may be made up of “dead virus” still present in some nooks and crannies of the lung? Because the PCR test doesn’t distinguish between complete, infectious viruses and RNA fragments from inactive viruses. This has been known all along, and no virologist is “confused” about this. Any “confusion” on these issues comes from not learning the available information.
        And obviously these issues do not affect the case fatality rate in any way.

        • Who exactly is claiming that a true positive PCR test can sense an amount that is “not infectious” for the carrier?

          Or are you talking about people who have actually undergone an infection, but where the virus concentration in their samples is so low that they probably won’t infect others

          First you ask who is claiming pcr can amplify noninfectious numbers of viruses, then you say that has been “known all along”. Very strange post.

          I’ve been saying since the beginning these pcr tests were never properly studied (they just appeared out of nowhere without discussion in Jan) but if they ever are we will find it’s something like 70% false positives and 70% false negatives. Error rates in a lab are not the same as in the field.

          Some are saying now that 90% of the results are effectively false positives:

          https://www.nytimes.com/2020/08/29/health/coronavirus-testing.html

        • > it’s something like 70% false positives and 70% false negatives.

          In that case inversing the output will be enough to get a much better test.

        • Not really, it is more like GIGO.

          The false positives are due to contamination, faulty negative controls, and presence of the RNA without infectious virus (or at least enough of it to worry about).

          The false negatives are due to sampling a tissue without virus when it is present elsewhere, or improper handling of the sample leading to degradation.

          So the same sample can be both a false negative and false positive.

          The much better test is a pulse oximeter. If you feel sick and your spO2 is below 95% consistently it’s probably covid and that is the time to figure out a way to avoid chronic oxygen deficiency.

        • Another way to look at it:

          Say on avg there are three bad storms each month that require you to take some action. The weatherman also predicts 3 bad storms each month but 2/3 times nothing happens and 2/3 of the actual bad storms are missed so if you rely on the prediction your trash cans float away and car gets flooded (or whatever).

          Are you just going to reverse the test and act like a storm is coming every day not predicted to be a bad storm and vice versa? No, you eventually just ignore the predictions altogether.

        • > Say on avg there are three bad storms each month that require you to take some action. […] Are you just going to reverse the test and act like a storm is coming every day not predicted to be a bad storm and vice versa?

          In that example you have 70% false positives and 7% false negatives. Not quite the same thing.

        • In that example you have 70% false positives and 7% false negatives. Not quite the same thing.

          Lets see. They missed 2/3 real storms and only hit on 1/3 predicted storms. You added in the assumption of one storm/action taken per day from which you got 2/27 stormy days that were predicted non-stormy.

          What I meant is that if there is a storm you will not be warned 66% of the time, and if you are warned of a storm none will show up 66% of the time.

          For covid tests:

          If you have covid you will test negative 70% of the time, and if you test positive you dont have covid 70% of the time.

          So that is false negative rate and false discovery rate, I think.

        • Ok. I thought that “70% of false negatives” meant “70% of negative results are wrong” in the same way that “70% of false positives” meant “70% of positive results are wrong”.

        • Ok. I thought that “70% of false negatives” meant “70% of negative results are wrong” in the same way that “70% of false positives” meant “70% of positive results are wrong”.

          Are you responding to this?

          If you have covid you will test negative 70% of the time, and if you test positive you dont have covid 70% of the time.

          So that is false negative rate and false discovery rate, I think.

        • You said “it’s something like 70% false positives and 70% false negatives.” I understood something, you meant something else. Case closed.

        • The question is what are you testing for?

          The PCR test tests for the presence of viral RNA in the PCR well. It can get there a number of different ways:

          1) you were sick with the virus, and had sufficient virus to be infectious to others and it was detected (this is the best case).

          2) You were sick with the virus a while back, by now you are not infectious to others, but the RNA is still there due to viral fragments etc.

          3) You were not sick at all, didn’t have the virus, and the sample was contaminated at the lab, so that some RNA got into the PCR well.

          Only case (1) helps us control the pandemic, and only if the result is returned in 24-48 hours not 7 to 10 days or more.

          The CDC is saying that around 10 days after infection you’re pretty much non-contagious, but typically you’d detect the viral RNA for much longer, say 20 days after infection or more for some people.

          If you understand the test as “testing for the presence of viral RNA” then false positive and false negative rates are probably a few percent.

          If you understand the test as “testing to see if you are infectious to other people” then yes, you might get 70% false positives. People will be getting the test, they’ve had the virus for 10 days already, they test positive, but they don’t have sufficient quantity of active infectious virus to be dangerous to people. That can happen for sure. Also, you could be very sick, but you get your positive test results 15 days after being swabbed. Even if you had plenty of infectious virus at time of sample collection, by the time you get the result you’re not a danger to others…

        • I do have to wonder what the “true” false positive rate (genuinely no viral RNA) is – given human error, etc., not the *theoretical* specificity of the method itself.

          Most places it’s probably not enough to matter… but in places like New York where they’re at something like 1% positive tests, how many of those are real?

        • I believe the false positive contamination rate probably varies with the testing positivity rate. A major way you contaminate is you take a positive sample and drip a little into the wrong well, or something similar (forget to change a pipette tip, or something). That’ll be less common when positive samples are less common.

          People I respect who have lab experience say on the order of 2% of PCR reactions tend to be contaminated… but this is when working with things where you have say one positive and a matched control, so there is plenty of material to contaminate with. I can’t imagine zero percent of NYC tests are true positive. If 1% are reported positive, then probably no more than 0.5% are false, I’d guess 0.1% or something. This is percent *of tests*. But even 0.1% of tests falsely positive is 10% of positives are false.

        • We are writing to share the analysis of positive results in our Health Board from 20/6/2020 to 21/7/2020 during a period of very low prevalence in this area. Over this period 31 throat swabs tested positive for SARS-CoV-2 by Real-time polymerase chain reaction (RT-PCR) on a number of platforms (Seegene (n=15), Cepheid (n=15), Luminex (n=1)). These platforms use a combination of targets including the N and E gene (Cepheid and Luminex) and the N, E and RdRp gene (Seegene).

          Our results show that 26/31 were positive at low level (CT value >35) in a single gene and five were positive in more than one gene: three were positive in two genes (Cepheid n=2, Seegene n=1), and two positive in three genes (Seegene).

          Of the 26 single gene low level positive results, 19 were repeated and all 19 were negative on repeat testing. Of these 26, 17 were asymptomatic and underwent screening for the following indications: pre-operative (5), nursing home residents (5), transplant waiting list (3), pre-discharge (2), pre-delivery (1), and nursing home key worker (1).

          Nine of the 26 were symptomatic. The threshold values (Ct) in these patients ranged from 36 to 43, with a mean score of 39.5. One had a previous positive result and the result likely reflected residual genetic material. Five symptomatic individuals were re-tested and all re-tests were negative. Two were re-tested on the same day, one the following day, one after five days and one after eight days.

          Five were positive in more than one gene. Two were asymptomatic pre-discharge patients. One was most likely a true positive case. He was asymptomatic at the time of swabbing but was symptomatic on admission with CXR changes and lymphopenia. His initial negative test at the time of admission was likely a false negative. The other asymptomatic individual was asymptomatic at the time of swabbing and was negative on repeat swabs on the same day and two days later. However, he was positive three months earlier and was antibody positive and so we think the positive result likely reflects the presence of residual RNA. Three were symptomatic. One was positive in two genes and two were positive in three genes. Overall, there were 12 symptomatic patients (nine were positive in a single gene, three in more than one gene).

          https://www.bmj.com/content/369/bmj.m1808/rr-22

        • I wasn’t really suggesting zero true positives.

          Just that the proportion of positives which are actually false positives might be high enough to make the reported numbers pretty meaningless.

        • > First you ask who is claiming pcr can amplify noninfectious numbers of viruses, then you say that has been “known all along”. Very strange post.

          Anoneuoid, you are misquoting me.
          I was making a distinction between the notion “the level is low now, so the tested person was never really infected and doesn’t need to be contact traced” (which is ludicrous/reckless) and “the level is low now, so the person can be discharged from quarantine because there is little risk they’ll infect anyone”.

          Tom’s “it cannot be infectious” is ok if it refers to the latter, and that was always known. But the shortened statement seems to suggest that low levels indicate that the person whose test comes back positive never really had an infection, and that’s a dumb assumption.

          Throwing the 70% number into the discussion that has no basis in reality is very poor discussion style.

          rt-PCR tests for other viruses are in common use, which is why they were able to be rolled out so quickly in many countries: it was just a matter of choosing the right gene sequence to detect to adapt this method to SARS, which was done by January 12th. The labs already had machines and personnel who could operate them. rt-PCR doesn’t “appear out of nowhere”.

        • False negative rate 70%:
          https://www.acpjournals.org/doi/10.7326/M20-1495

          False positive (or is it false discovery?) rate 70%: https://www.bmj.com/content/369/bmj.m1808/rr-22

          False positive rate 100%: https://apnews.com/5d15429433051c78f2ca42fc08fbf835

          And if you have a bit of virus RNA in you but never got a sick and were never at risk of transmitting to anyone else, that is also effectively a false positive. For no disease before this one in history would that be counted as a case.

          However, my original 70% “false positive” number was based on the 80-90% mild/asymptomatic cases along with the very skewed distribution of transmission. Most “cases” do not appear to infect anyone else:

          https://www.researchsquare.com/article/rs-29548/v1
          https://www.medrxiv.org/content/10.1101/2020.05.21.20104521v1

        • Figure 2 in your first source supports a false negative rate of close to 25% using nose swabs in a phase where the virus is expected to replicate in the nose; as the virus moves into the lungs later, obviously this number is expected to get worse over time. I don’t see how you get 70% from that.

          The second source is a letter; it is unclear how thoroughly the rate of false positives was determined. The first source confirmed the diagnoses via antibody tests, this one did not. I don’t trust it.

          The NFL anecdote does not support 100% false positive rate, and it is misleading of you to suggest it does. For one, it mentions that only one of 5 labs testing the NFL had the contamination event; and secondly, this is a self-selected and not a random sample.

        • On your two sources at the bottom of your comment:

          “Full genome viral sequences inform patterns of SARS-CoV-2 spread into and within Israel” appears to have no bearing on false positives, as I can’t imagine a full viral sequence being a false positive?

          The Hongkong cluster study had the purpose of identifying superspreading events (and IMO failed at that, as “events” included a large number of secondary infections); they excluded a large amount of data that was not conducive to this purpose, and the problem with that is that “no recorded transmission pair” does not mean that transmission did not occur. Also, under lockdown conditions, with R at or below 1, and with many people infecting more than 1 other person, we expect the majority of infectious persons to not infect anyone else; but that’s not because they weren’t infectious, it’s because social distancing worked, and randomness. “Did not infect anyone” does not imply “was not infectious”, and does not imply “needn’t have isolated”.

        • Thirdly, the “80-90% mild/asymptotic cases” feels also unsupported/unsupportable and at the very least begs the question which cases are counted as “mild” and why these would not be considered infectious when asymptomatic transmission has been documented; contradicted on the face by Diamond Princess data, I believe; my personal guess is that at most 50% of all SARS-CoV-2 infections are asymptomatic, and maybe as few as 25%.

        • Figure 2 in your first source supports a false negative rate of close to 25% using nose swabs in a phase where the virus is expected to replicate in the nose; as the virus moves into the lungs later, obviously this number is expected to get worse over time. I don’t see how you get 70% from that.

          Not sure why you chose to look at only one figure about nasal swabs which almost no one is doing so is largely irrelevant. Why would I be referring to that?

        • I don’t see where figure 2 in the body of that article is referring to “nasal” swabs; I thought it reflected all of the data, and it provides the best information on WHEN the PCR test will detect viruses in the upper respiratory tract.

Leave a Reply to Tom Passin Cancel reply

Your email address will not be published. Required fields are marked *