Hydroxychloroquine in COVID-19 Treatment, Actual Usage in the USA

By Leo Goldstein – Re-Blogged From WUWT

Preprint. August 23, 2020.

Key Words: hydroxychloroquine, COVID-19, SARS-CoV-2, Wuhan



Three population surveys were performed, seeking information about the drugs prescribed for COVID-19 patients. The August 16 national survey (USA-0816, 868 valid responses) and the August 3 national survey (USA-0803, 1,059 valid responses) covered the entire US. Another smaller survey (TX-0711, 116 valid responses) covered the state of Texas. All responses to all three surveys are attached in anonymized form for further analysis by the scientific community as one of the deliverables.

The analysis was focused on Hydroxychloroquine (HCQ). This study has found that Hydroxychloroquine (HCQ) was used for the treatment of COVID-19 in the US since January 2020. From January to August 16, 13.5% of COVID-19 patients ages 40+ were prescribed Hydroxychloroquine in the US.

The New England and Middle Atlantic census divisions suffered from the largest COVID-19 mortality and accounted for most COVID-19 deaths from mid-March through mid-June. This study has found that they had the lowest utilization of HCQ (average 6.1% for patients ages 40+) in the matching period early March — late May.

Everywhere in the US, prescribing HCQ nearly ceased in the last third of May but resumed in June and have been fluctuating around 16%, for patients ages 40+.

The author declares no competing interest.

No funding was provided for this work.

All relevant ethical guidelines have been followed.


Hydroxychloroquine + Azithromycin (with or without Zinc), given upon early symptoms of COVID-19, have been reported to provide significant benefits in clinical trials 1 2, improving patients’ odds  up to 5 times 3 3b 4 5. Since Hydroxychloroquine (HCQ) had no sponsor who would determine the best treatment regimen and conduct clinical trials accordingly, many treatment regimens were tried. Unfortunately, some meta-reviews commingled results from various regimens and included borderline fraudulent papers. That created confusion about the effectiveness of early HCQ-based treatment for COVID-19.

In some countries, HCQ-based treatment for COVID-19 became a de-facto standard6. At the same time, only a small number of relatively small randomized controlled trials were performed. Conducting further RCTs with endpoints in the patient’s health, length or strength of symptoms, hospitalization, or mortality is unethical and impractical.

Luckily, modern science is based on using real-world evidence, rather than on regurgitating prior literature. The gap between the clinical practice and academic world can be bridged by surveying physicians who treated COVID-19, and patients who received treatment from COVID-19. In one survey7, published on April 8, 60%-70% of physicians reported that they would take HCQ and give it to family members on symptoms of COVID-19. Sermo released regular doctor surveys regarding the drugs used for COVID-198. These surveys 9 10 11, now discontinued, have shown that doctors of the world used HCQ very broadly and rated it as being very effective or extremely effective against COVID-19. A systematic review of surveys of physicians12 was conducted by the author and confirmed these conclusions.

In academic meta-analyses of studies repurposing existing drugs for COVID-19, Hydroxychloroquine was not studied a lot13. A registry of self-reported use of medications by physicians14 is not very helpful. Surprisingly, statistics on COVID-19 patients treated with HCQ in the US are hard to find. The pharmacy’s data is not publicly available, shared insufficiently and selectively, and the shared information is more15 or less16 useful. Under the US FDA Emergency Use Authorization, the Strategic National Stockpile dispensed about 2.4 million HCQ 7-day treatment courses to state and local authorities from March 28 to May 22. Then the FDA claimed no knowledge of how these doses were used17.

This study collected data directly from the people who knew COVID-19 patients personally or otherwise, including information about the drugs prescribed, period, region, and the patient’s age. Limited analysis was performed, attempting to quantify the actual usage of HCQ.  The word “patient” is used throughout the paper to stress that the surveys asked only about people who saw a doctor for COVID-19 symptoms.


Data Collection

All three surveys were conducted using SurveyMonkey. Each survey was sent to the general US population. Each survey asked the respondent the following:

  1. Whether he or she knew anybody diagnosed with or treated for COVID-19. Depending on the answer, the response was assigned weight from 1.0 (personal knowledge, friends, or family) to 0.2 (second-hand knowledge) or disqualified at all. Qualified respondents were asked to provide information about a single case best known to them.
  2. The age bracket of the patient was selected from the options <40, 40-49, 50-59, 60-69, 70+.
  3. When the treatment took place.
  4. Which of the following drugs were prescribed or recommended:

Hydroxychloroquine (Plaquenil)

Ibuprofen (Advil)

Acetaminophen (Paracetamol, Tylenol)



The order of the drugs was random, except for Other. The respondent was asked to select everything that applied. TX-0711 also included Azithromycin as an option. In the analysis, it was considered as any drug other than Hydroxychloroquine.

The geographical location of the respondent was provided by SurveyMonkey, at the resolution of census division (i.e., multiple states). It was assumed that the patient was in the same division. Also, TX-0711 was limited to Texas. USA-0816 has explicitly asked about the state of the patient.

Responses that were fulfilled in less than 30-40 seconds (depending on the survey length) or gave the date of treatment in the future or before 2020 were discarded as invalid.

The author has not conducted and did not have access to any other similar surveys.

Specific Details per Survey

USA-0816: A small number of responses were received the next day, on 08/17.

USA-0803 had an additional question Q7. How severe was the disease? (select one: Mild – like a common cold; Moderate; Severe; but without hospitalization; Severe, with hospitalization)

TX-0711 was sent to the ages 40-90 years in Texas.

See the Attachment for the exact wording of questions, audiences, and other details of the surveys.


The data were analyzed with the time granularity of one-third of a month: 1-10, 11-20, 21-end. The January and February treatments were included in the data for the first third of March. The results from the first third of March to the second third of August have been calculated.

A data entering anomaly was detected in USA-0803 and US-0816. A disproportionately large number of responses were on the 3rd and 16th of each of the previous months for these surveys, respectively. On the assumption that some of the respondents wanted to specify a month, but the not exact day, the excess data on these days were spread among all thirds of the same month, proportionately to the weighted number of responses in these thirds, with additional weighing for USA-0803.

The patients younger than 40 were included in the analysis, but weighted down with the coefficient 0.5, except when stated otherwise. There are different ways to think about the treatment of such patients. On the one hand, they have a low risk of death or hospitalization. On the other hand, they might want such treatment anyway. Also, they are capable of the coronavirus transmission, so early anti-viral therapy of such patients might benefit the public at large. Finally, some respondents might have entered <40 in error.

The responses do not provide information on how early a patient was able to obtain HCQ if prescribed. Also, the study did not use any data on what share of symptomatic COVID-19 infected persons consulted a doctor. That limits its interpretation. To partially compensate for that, the data on the total number of the cases in a certain location/time was used and expressed as a percentage of the total number of treatments in that location. See even lines (small font) in the attached Summary.xlsx.

Super Regions

To achieve better statistical significance, and because of strong interdependence between New England and Middle Atlantic, some census divisions (called “regions” by SurveyMonkey) were combined, yielding six super-regions.

Table 1. Regions and Super Regions

Super-Region Region States
NE + MA New England Maine, New Hampshire, Vermont, Massachusetts, Rhode Island, Connecticut
NE + MA Middle Atlantic New York, New Jersey, Pennsylvania
East North Central East North Central Ohio, Indiana, Illinois, Michigan, Wisconsin
South Atlantic South Atlantic Delaware, Maryland, District of Columbia, Virginia, West Virginia, North Carolina, South Carolina, Georgia, Florida
WNC + ESC + Mountain West North Central Minnesota, Iowa, Missouri, North Dakota, South Dakota, Nebraska, Kansas
WNC + ESC + Mountain East South Central Kentucky, Tennessee, Alabama, Mississippi
WNC + ESC + Mountain Mountain Montana, Idaho, Wyoming, Colorado, New Mexico, Arizona, Utah, Nevada
Pacific Pacific Washington, Oregon, California, Alaska, Hawaii
West South Central West South Central Arkansas, Louisiana, Oklahoma, Texas


The data from USA-0803 and USA-0816 were combined. The results were calculated for the US and each of six super-regions, broken down by month thirds.

Separately, the data from TX-0711 and USA-0816 were combined to calculate results for Texas.


The responses do not explicitly reveal which HCQ-based treatment regimen was used. Outside of clinical trials16, most doctors probably have been prescribing HCQ + AZ, with or without Zinc, starting at the end of March. Also, responses do not reveal how early or late in the disease, the treatment was prescribed.

Suspension of the HCQ usage in the last third of May coincides with the publication by The Lancet of Mehra et al.19, published on May 22. That paper was retracted on June 4, after the damage had been done.

It is unclear how the number of patients (i.e., individuals who were infected, had symptoms of COVID-19, and consulted with a doctor) relates to the number of infection cases. Some individuals with COVID-19 did not get tested or received false-negative results. There were also false positives. Some COVID-19 sufferers never sought testing or treatment. The share of people who were incorrectly diagnosed or did not seek treatment has been changing throughout time.

The data on small size HCQ prescriptions20 allows us to estimate that ~25,000 small prescriptions were filled weekly in April, translating into an average 3,600 prescriptions per day. Additionally, some patients were treated with HCQ in hospitals, allowing the number of daily prescriptions to be rounded up to 4,000. There were, on average, 30,000 new daily cases reported in April21. 4,000 daily HCQ prescriptions are 13.3% of the 30,000 daily positive test results. In this study, 15.9% of the patients were prescribed HCQ in April. That might be interpreted as 84% of individuals with positive test results consulted with a doctor, in the absence of other factors.

From late March to early May, about 150,000 US patients received HCQ for COVID-19. HHS OSE found 97 adverse reports22 of all kinds (misspelled as 347 in the FDA Memorandum17) associated with HCQ and chloroquine during that period.

This study shows that Remdesivir was widely used in March-April before its emergency approval on May 1. Surprisingly, its usage throughout the epidemic was, on average, 70% of the HCQ usage. It sharply declined in the last third of June, possibly on disappointing clinical results23 and evidence of liver toxicity24, but resumed at the nearly previous level in July.

Study Limitations

There are limitations usual for studies based on a population survey. Most values in the Summary spreadsheet are computed from small sample sizes. Other limitations are mentioned in subsection Processing.


Patients’ side statistical information about the use of hydroxychloroquine for COVID-19 patients was collected. Using it, this study has found:

  • HCQ was used for the treatment of COVID-19 in the US since January 2020. From January to August 16, 13.5% of COVID-19 patients ages 40+ were prescribed Hydroxychloroquine in the US.
  • The New England and Middle Atlantic census divisions suffered from the largest COVID-19 mortality and accounted for most COVID-19 deaths from mid-March through mid-June. They also had the lowest utilization of HCQ (average 6.1% for patients ages 40+) in the matching period early March — late May.
  • Everywhere in the US, prescription of the HCQ nearly ceased in the last third of May but resumed in June and has been fluctuating around 16% for patients ages 40+.

The raw responses data is attached. It can be mined further, especially when combined with publicly available statistics on the COVID-19 hospitalizations, deaths, tests, infection cases, and how many days pass from the first COVID-19 symptoms and the start of HCQ based treatment. Eventually, more data would allow testing hypotheses:

  • Early HCQ-based treatment of adults of all ages with COVID-19 symptoms correlates with decreased COVID-19  infection and/or hospitalizations cases 10-20 days later25.
  • Early HCQ-based treatment of ages 40+ with COVID-19 symptoms correlates with decreased COVID-19 deaths and/or ICU admissions 20-35 days later.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s