THE AUTHOR REPLIES

I thank Dr. Korman (1) and Drs. Peiffer-Smadja and Costagliola (2) for their observations. Dr. Korman’s thesis is that no available treatments are effective in preventing hospitalization for the overwhelming majority of coronavirus disease 2019 (COVID-19) patients and that potential hazards are associated with use of hydroxychloroquine (HCQ) plus azithromycin (AZ) (1). The studies I reviewed (3) contradict this. Dr. Korman superficially describes the same studies that I discussed at length, except with negative adjectives and numerous terms in quotation marks to imply, without evidence, their lack of validity. He calls all of these studies “anecdotal,” to distinguish them from the “magic” of randomized controlled trials (4), wherein government medical and scientific regulatory agencies of Western countries around the world routinely use epidemiologic evidence to establish facts of causation, benefit, and harm (5). This disingenuous argument has been discussed at length elsewhere (6). Dr. Korman’s only novel point is that macrolide antibiotics such as AZ can lead to development of antibiotic resistance. Such instances can occur but are uncommon, and this issue has seemingly not been of substantial concern in the hundreds of millions of uses of AZ worldwide during the past 30 years.

Drs. Peiffer-Smadja and Costagliola (2) discuss the data in some of the studies that I reviewed. They first question the results of the small nonrandomized trial by Gautret et al. (7). I also have concerns about subject baseline differences between the treated and untreated subjects in that study and thus limit my conclusions to the 26 treated patients. Gautret et al. provided individual-subject data on all 26, which enabled me to carry out my own Cox regression analyses. Fourteen patients received HCQ only, 6 received HCQ + AZ, and under intention-to-treat principles, 6 were lost to follow-up and had received 3 or fewer days of HCQ but were unspecified as to receipt of AZ. I conducted my analyses starting with the 20 subjects with completed medication usage, and then included bracketing of the unknown-exposure subjects, first assuming that all had taken HCQ + AZ and then only HCQ, and in each case, whether the 6 subjects presented with upper respiratory-system versus lower respiratory-system infections. In the 20 main subjects, for HCQ + AZ versus HCQ alone, the hazard ratio for viral clearance was 7.4 (95% confidence interval: 1.12, 48.4). Across the 4 bracketed-combination analyses, the hazard ratios ranged from 3.5 to 8.0 and the P values from 0.078 to 0.021. Drs. Peiffer-Smadja and Costagliola need to use the outcome event times in Cox regression in order to obtain proper P values. They also say that the trial was not conducted in outpatients and therefore cannot be applied to outpatients (2). However, the Marseille, France, hospital was used as both an outpatient clinic and a small inpatient facility for a citywide COVID-19 population screening and treatment program, and patients were seen as daily “inpatients” as well as stayed overnight in it. Many of the patients in the Gautret et al. study (7) were asymptomatic or very mildly symptomatic and would be treated as outpatients in most circumstances.

Second, Drs. Peiffer-Smadja and Costagliola refer to the larger Marseille screening program (8). I have discussed those data at length in my response (9) to Dr. Fleury (10). Third, they label a carefully performed, sequential-patient nonrandomized controlled clinical trial as one of numerous “unpublished, poorly designed studies whose quality was even lower than that of the studies discussed above” (2, p. 000). I disagree with this characterization, as it is unsupported by the evidence. Fourth, Drs. Peiffer-Smadja and Costagliola assert that the Boulware et al. prevention trial (11) demonstrates lack of treatment efficacy. That prevention trial is not relevant to treatment of high-risk outpatients, because virtually all of its subjects were low-risk; it would be difficult for any active treatment to do much better than the 1 hospitalization observed among the 58 test-positive placebo patients. In fact, using a preventive medication that allows subjects to develop antibodies while protecting them from severe disease and hospitalization is a better goal than blocking infection altogether. Fifth, Drs. Peiffer-Smadja and Costagliola take issue with the use of case series of treated patients (6). In the mass-mortality circumstances we face, a cohort of 400 treated high-risk outpatients with 1 or 2 deaths can only be considered informative about the fact of treatment efficacy.

Finally, in pandemic times when months and years of delay cannot be tolerated before large randomized controlled trials are completed, it is possible to quibble with apparent imperfections in almost any study. That misses the forest for the trees. Since my paper discussing 5 studies (3) was published, data from 7 other studies of high-risk outpatients have become available (see Table 1), all showing the same substantial and significant benefit of use of HCQ along with AZ or other companion medications. Two additional large studies of hospital patients given HCQ within 48 hours of admission showed significant benefit after adjustment for age and comorbidity (12, 13), and a meta-analysis of studies carried out to date completely demonstrates this benefit (14). Perhaps even more importantly, the exponential COVID-19 mortality explosion in the northern state of Pará, Brazil (15), reversed direction, trending downward dramatically about 5 weeks after a shipment of 75,000 doses of AZ and 90,000 doses of HCQ began to be distributed to infected individuals (Figure 1). No such decline has been observed in the rest of Brazil. This is a compelling, large-scale experiment demonstrating efficacy of HCQ + AZ in saving the lives of high-risk people infected with severe acute respiratory syndrome coronavirus 2.

Figure 1

Daily numbers of deaths from coronavirus disease 2019 (COVID-19) in the Brazilian state of Pará (solid line) and in the rest of Brazil (dashed line), April 1, 2020–July 9, 2020. Points plotted are 7-day symmetrical moving averages of the raw data, which are available from Coronavírus Brasil as posted daily (15). The daily numbers of newly identified COVID-19 cases in Pará were increasing through May 28, 2020, and have since stayed roughly flat at about 2,100 per day, whereas the daily numbers of new cases in the rest of Brazil have risen throughout the period (data not shown). On April 6, 2020 (gray arrow), the public hospital network of Pará State purchased 75,000 doses of azithromycin and 90,000 doses of hydroxychloroquine and started distributing them to infected individuals over the next few weeks; the Hapvida health maintenance organization hospitals in the state also acquired the medications and started using them during the same period (Alexandre Wolkoff, Hapvida Saúde (Fortaleza, Brazil), personal communication, 2020). Approximately 5 weeks after distribution of the medications began, the numbers of deaths in Pará turned down dramatically. In Pará, the July 2 mortality (n = 40) as a fraction of June 2 incidence (n = 2,068) was 1.9%, whereas the same proportion for the rest of Brazil was 4.3% (1,026/23,733). Brazil outside of Pará was not systematically using hydroxychloroquine and azithromycin during the time period shown in the figure.

Table 1

Studies and Investigators Examining High-Risk Outpatient Coronavirus Disease 2019 Treated Early with Hydroxychloroquine

First Author, Year (Reference No.)	Location	Subject Status	No. of Patients Treated With HCQ	No. of Comparison Subjects	Other Medications Administered With HCQ	Comparison-Subject Medications Administered	Outcome	Direction of Benefit	Results
									RR	95% CI	P Value	No. of Deaths	Total No. of Subjects
Gautret, 2020 (6)	Marseille, France	Mixed older adults	6	14	AZ	HCQ alone	Nasopharyngeal viral clearance	Reduced risk	7.4	1.12, 48.4
Zelenko, 2020 (16)	Kiryas Joel, New York	High-risk	405 (series 1)		AZ, zinc sulfate		Mortality	Reduced risk				2	405
Zelenko, 2020 (PCa)	Kiryas Joel, New York	High-risk	400 (series 2)		AZ, zinc sulfate		Mortality	Reduced risk				0	400
Barbosa Esper, 2020 (17)	São Paulo, Brazil	Mixed older adults	412	224	AZ	SOC	Hospitalization	Reduced risk	0.35	0.14, 0.87
Ahmad, 2020 (18, PCb)	Long Island, New York, nursing home	High-risk	200		DOX		Mortality	Reduced risk				9	200
Lagier, 2020 (19)	Marseille, France	High-risk	199	199	AZ	Propensity-score-matched; HCQ or AZ alone or neither	Mortality	Reduced risk	0.41	0.17, 0.99
Leriger, 2020 (PCc)	Indiana nursing home	High-risk	105	113	None, AZ, DOX	SOC + AZ or DOX	Mortality	Reduced risk			<0.05
Kacmar, 2020 (PCd)	Aurora, Illinois	High-risk	68		AZ		Mortality	Reduced risk				0	68
Fonseca, 2020 (PCe)	Brazilian health maintenance organization	High-risk	159	558	Prednisone	Neither HCQ nor prednisone	Hospitalization	Reduced risk			<0.05
Procter, 2020 (PCf)	McKinney, Texas	High-risk	50		AZ, zinc sulfate, losartan, aspirin		Mortality	Reduced risk				0	50
Crawford, 2020 (20)	Festus, Missouri, nursing home	High-risk	52		Rehydration		Mortality	Reduced risk				0	52
Tyson, 2020 (PCg)	El Centro, California	High-risk	219		AZ (2 with DOX)		Mortality	Reduced risk				0	219

Abbreviations: AZ, azithromycin; CI, confidence interval; DOX, doxycycline; HCQ, hydroxychloroquine; PC, personal communication; RR, relative risk; SOC, standard of care.

^a Vladimir Zelenko, family practitioner (Monroe, New York), personal communication, 2020.

^b Imtiaz Ahmad, 21st Century Oncology, Inc. (Fort Myers, Florida), personal communication, 2020.

^c Monica Leriger, American Senior Communities (Indianapolis, Indiana), personal communication, 2020.

^d Lawrence Kacmar, Center for Primary Care and Sports Medicine (Aurora, Illinois), personal communication, 2020.

^e Silvia Fonseca, Hospital São Francisco (Ribeirão Preto, Brazil), personal communication, 2020.

^f Brian Procter, McKinney Family Medicine (McKinney, Texas), personal communication, 2020.

^g Brian Tyson, All Valley Urgent Care (El Centro, California), personal communication, 2020.

 

ACKNOWLEDGMENTS

Author affiliation: Department of Chronic Disease Epidemiology, Yale School of Public Health, New Haven, Connecticut (Harvey A. Risch).

H.A.R. acknowledges past advisory consulting work with 2 of the more than 50 manufacturers of hydroxychloroquine, azithromycin, and doxycycline. This past work was not related to any of these 3 medications and was completed more than 2 years ago. He has no ongoing, planned, or projected relationships with any of these companies or any other potential conflicts of interest to disclose.