COVID: hydroxychloroquine trial in Detroit

A quasi-randomized trial of hydroxychloroquine in patients with COVID-associated pneumonia in Detroit, Michigan found worse outcomes with the med (see covid hydroxychloroquine RTC nejm2020 in dropbox).  This study was just submitted to the New England Journal of Medicine, as a pre-publication and pre-peer reviewed submission

Details:

-- 63 consecutive adults with viral pneumonia, secondary to SARS-CoV-2 infection (documented by PCR) during a two-week period in March 2020

-- the trial was done in 2 different hospitals in Detroit, 7 miles apart, and serving the same general population

-- 41% female (though 53% were on hydroxychloroquine/29% on standard care), baseline absolute lymphocyte count 0.92K/mL, baseline absolute neutrophil count 5.7K/mL, baseline respiratory support requirement 1.7 (statistically significant difference of 1.94 on hydroxychloroquine vs 1.52 on standard care,  p=0.012):  respiratory support was quantified as room air (level I), supplemental oxygen from 1 to 15 L without intubation (level II), intubation and mechanical ventilation (level III), failure to support respiratory needs resulting in death (level IV)

-- number of high-risk conditions was 1.2  (includes asthma, COPD, heart failure, diabetes, hematologic malignancy, or immune compromised in which patients were taking prednisone equivalent to 20 mg/day)

-- 32 patients were started on hydroxychloroquine (400 mg b.i.d. for 1 to 2 days, followed by 200 mg to 400 mg once daily for 3 to 4 subsequent days plus supportive care) vs 31 patients on supportive care alone 

-- in one hospital, patients received hydroxychloroquine shortly after admission, and in the other it was initiated days later when PCR results were available

-- primary endpoint: need to escalate respiratory support, change in lymphocyte count, and change in neutrophil-to-lymphocyte ratio, over the 5-day hospital course

-- subgroup analysis was done on patients who had no high-risk comorbid conditions and were not intubated on admission

Results:

-- at 5 days, those on hydroxychloroquine vs just supportive care:

    -- needed more respiratory support, in particular an impressive trend to higher risk for intubation, p=0.051

        -- change in respiratory support level +0.93 on hydroxychloroquine, +0.16 on supportive care, p=0.013

-- there was no difference in absolute lymphocyte change between the groups, though a significant trend to being worse in hydroxychloroquine (+9.6 vs +1.6), p=0.051

-- hydroxychloroquine was associated with worsening neutrophil-to-lymphocyte ratio

-- no mortality benefit (mortality rate: 4 of 31 on hydroxychloroquine, 1 of 32 on supportive care)

-- by assessing matched subgroups (given some of the large differences in baseline demographics), which include 38 of the 63 total patients:

    -- mortality rate: no difference, 2 of 17 on hydroxychloroquine vs 1 of 21 on supportive care

    -- rate of intubation: 7 of 17 (41%) vs 2 of 21 (10%), p=0.051

    -- change in respiratory support level: +0.76, vs +0.24, p-=0.041

    -- change in neutrophil-to-lymphocyte ratio +15 vs +6 ( p=0.053); no difference in the change in absolute lymphocyte counts

-- baseline respiratory support requirements were inversely related to the likelihood of requiring increased level of supportive care throughout admission, odds ratio 0.27 (0.08-0.97), p=0.044

-- post hoc analysis: those with neutrophil-to-lymphocyte ratio greater than 3.13 (which correlates with a strong cytokine response), and age greater than 50, were more likely to need more supportive care throughout admission (46% vs 18%) p=0.031.

Commentary:

-- hydroxychloroquine and chloroquine have been advanced as a potential therapy for Covid-19, largley based on reasonable animal and in vitro models. And, there have been pretty dramatic results from observational studies. 

    -- a Chinese study (also quasi-randomized, with some details missing) seemed to increase hydroxychloroquine from 5 days to 10 days, finding improved clinical outcomes: http://gmodestmedblogs.blogspot.com/2020/04/covid-hydroxychloroquine-helped-in.html

    -- chloroquine has multiple potential beneficial antiviral effects: http://gmodestmedblogs.blogspot.com/2020/04/covid-chloroquines-antiviral-effects.html

--there are several limitations to this Detroit study:

   -- this is a small study, with some pretty big differences in the patients receiving hydroxychloroquine vs just supportive care, especially for baseline respiratory support level required

        --and, they did find that the baseline respiratory support requirement was inversely related to the likelihood of requiring increased level of supportive care throughout the admission (ie, the hydroxychloroquine group was pretty disadvantaged in this, and with a really large odds ratio of 0.27!!)

    -- and, being such a small study, they were not able to disaggregate several factors: the gross total number of comorbidities may obscure differences (eg, perhaps heart failure or COPD is more of a risk factor than diabetes?), or other criteria (eg, needing 15L of oxygen vs 1L may be an outcome predictor, though they are in the same respiratory risk category)

    -- the differences in the characteristics of the experimental groups are likely to be greater in a small study (less regression to the mean)

 their mathematical modeling to "match" the subgroups did not include 25 of the patients (40% of the total), which might distort these results (ie, cherry-picking). 

            -- a prior blog reveals the distortion that can occur with such propensity score matching: see http://gmodestmedblogs.blogspot.com/2020/03/tramadol-fo-oa-inc-mortalityprobs-with.html 

    -- this was a  a retrospective study, and lacking real randomization (and did not even having standardization of dosing), raising the possibility of residual confounders or unevenness in implementation of the interventions

        -- and, individual physicians were not mandated to obtain confirmatory PCR results prior to initiating hydroxychloroquine (a portion of the patients in respiratory distress may have been empirically started on hydroxychloroquine), potentially skewing the results to the disadvantage of hydroxychloroquine (and no data provided about what the actual treatment courses were actually given; and, as noted, the dosing of the hydroxychloroquine was not consistent, as would happen in a prospective RCT)

    -- they themselves referred to this study as a “quasi-RCT”, since in one hospital they waited 7 days to get the results of the PCR documenting Covid-19, and they combined these results with the hospital having more immediate dosing (no separate analysis, though pretty small number of patients)

        --and what about having to wait 7 days to get the Covid-19 results in one of the hospitals?????  this is egregious. it does seem that Trump got his results later the same day. Without going out too much on a limb, my guess is that this is not just happenstance... (will do a blog soon on disparities in covid-19 death rates)

-- the authors do note that the incidence of Covid-19 is increasing rapidly in Detroit, but they thought that this early publication might be useful. and they do reinforce that a prospective randomized clinical study would be necessary to have a clear conclusion.

so, lots of concerns about the actual design and conduct of this study (which really was a retrospective one, and without sufficient data on what actually transpired). it does, however, raise the flag of caution in assuming that hydroxychloroquine is beneficial, and may in fact make things worse (though there was a significant bias against hydroxychloroquine, as above). to me, we really do need true prospective RCTs with reasonably large numbers of patients and consistent medication dosing to look at the utility of hydroxychloroquine in COVID-19, and in particular:

--does it work better if given much earlier in the course of covid?

    -- and perhaps it works prophylactically, since there are some data suggesting that it impairs viral entry into target cells (interferes with binding to ACE2)?

--what is the correct dose and time course? the Chinese study suggested that a 10-day course was better than a 5-day one (that is my assumption from the study, which was not written up so clearly). the initial suggestion i heard was 200 mg bid for 10 days. which i assume is based on as much of a guess as any other regimen. But, as noted, in this quasi-randomized Chinese study there was apparent major benefit to adding 5 more days to the initial 5 day therapy

--i believe that these and other studies are in the works, and likely will have more robust results on which to base therapy

hopefully we will get more convincing data soon, though these preliminary data on hydroxychloroquine are certainly troubling....

geoff​

If you would like to be on the regular email list for upcoming blogs, please contact me at gmodest@uphams.org

to get access to all of the blogs:

1. go to http://gmodestmedblogs.blogspot.com/ to see them in reverse chronological order

2. click on 3 parallel lines top left, if you want to see blogs by category, then click on "labels" and choose a category​

3. or you can just click on the magnifying glass on top right, then  type in a name in the search box and get all the blogs with that name in them

please feel free to circulate this to others. also, if you send me their emails, i can add them to the list