key: cord-0867047-1mdoy6w5
authors: Tleyjeh, Imad M.; Tlayjeh, Haytham
title: The perceived efficacy of hydroxychloroquine in observational studies: The results of the confounding effect of "goals of care"
date: 2021-02-17
journal: Int J Antimicrob Agents
DOI: 10.1016/j.ijantimicag.2021.106308
sha: 9766448e852cebbbeb51496ead8877ab5010bff6
doc_id: 867047
cord_uid: 1mdoy6w5

nan

We read with interest the study by Catteau et al. [1] about the association of hydroxychloroquine (HCQ) with in-hospital mortality among 8075 COVID-19 patients from the Belgian national COVID-19 hospital surveillance data. Using a competing risks proportional hazards, the mortality was lower in the HCQ group compared to the non-HCQ group (adjusted cause-specific hazard ratio (cs-HR) 0.684, 95% confidence interval [CI] 0.617-0.758). The authors concluded that treatment with HCQ was associated with reduced in-hospital mortality.

Nevertheless, we believe that the authors have made spurious inferences because their findings are susceptible to treatment selection bias or "confounding by unmeasured goals of care " [2] as illustrated by the study data. The authors did not account for "goal of care" information, that is usually not readily available in national surveillance data and is of utmost importance during pandemics peaks and scarcity of resources. Decisions to limit life-sustaining therapies, commonly made during this pandemic, are strongly associated with high risks for death in ways that are unaccounted for by routine measures of illness severity. In fact, in this study by Catteau et al., 90% of non-survivors were older than 65 years and 80.4% of nonsurvivors did not receive ventilatory support prior to death. The data summarized in the (Table) support our contention that physicians' allocation of patients to the non-HCQ group was biased to the elderly and frail patients and most (80%) of these patients did not receive mechanical ventilation and likely died outside the ICU setting. Death could not be explained by the clinical features of severe ARDS due to COVID-19 as displayed in table 1 or table 2 of the study [1] . This practice of triaging process in allocating critical care resources was seen in many countries during the peak of the COVID-19 pandemic. The COVID-19 pandemic has put tremendous strain on healthcare systems in different countries, including Belgium, and has led to shortage of ventilators and ICU resources. During this period, an additional triaging process that involves medical and moral choices was added to patients' preferences for life sustaining treatments decisions [3]. Many countries have put guidelines for triaging those patients depending on short term survival chance versus long term prognosis criteria that can limit life expectancy [4] . For example, medical urgency, frailty, comorbidities and cognitive impairment (Table non-HCQ group) were used as criteria to limitation of care in Belgium [4] . Although age was not one of the criteria, older patients and patients with preexisting serious illness who are more likely to die with an acute illness are more likely to have treatment limitations [2] .

On the other hand, we command the authors for accounting for two important biases that are frequently ignored by investigators: survivor bias and competing risk bias. We have illustrated that survivor bias, which occurs because patients who live longer are more likely to receive treatment than those who die early, could change associations from benefit to harm [5] . In a sensitivity analysis that considered survivor bias, Catteau et al. reported a cs-HR 0.816, 95% CI 0.751-0.887; (a 20% increase in the HR). We believe that this should be the primary analysis of the study.

A common approach used by investigators, like the study by Catteau et al., is to examine patients' in-hospital mortality data without any follow-up beyond hospital discharge. In this scenario, discharged patients are treated as censored observations when using survival analysis. The fundamental assumption that the death hazard remains the same after censoring is violated here; discharged patients have usually recovered and thus have lower death hazards than patients who remain hospitalized. If no follow-up beyond hospital discharge is available, discharge from hospital and in-hospital death are therefore considered "competing events". A competing event is an event whose occurrence precludes the occurrence of the event of interest. Investigators should calculate and report cs-HR for each event to help interpret the associations: HR for death (the event of interest) and HR for discharge (the competing event).

Although Catteau et al. have calculated cs-HR for hospital mortality, they have not reported csHR for discharge for any risk factor including the use of HCQ.

Finally, we argue that the "low-dose" regimen of HCQ (400mg twice on day 1, followed by 200mg twice a day from day 2 to 5) is unlikely to achieve SARS-CoV-2 antiviral activity. We have recently reviewed [6] HCQ pharmacological data and concluded that, even using the lowest in vitro HCQ inhibitory concentration, it seems not possible to achieve minimum clinical 4 therapeutic concentration of HCQ with dosing of HCQ of >400 mg twice a day. This is because the free lung extracellular trough concentrations from a 400 mg twice daily dose, are well below the in vitro half-maximal effective antiviral concentrations EC50 values (Reviewed in [6] ).

We call for investigators to use more vigilance in analyzing data and interpreting findings from observational studies especially when they show results contradicting those of randomized trials.

Ethical Approval: Not required 

Low-dose Hydroxychloroquine Therapy and Mortality in Hospitalized Patients with COVID-19: A Nationwide Observational Study of 8075 Participants

Accounting for Patient Preferences Regarding Life-Sustaining Treatment in Evaluations of Medical Effectiveness and Quality

Ethics guidelines on COVID-19 triage-an emerging international consensus

Conclusion about the association between valve surgery and mortality in an infective endocarditis cohort changed after adjusting for survivor bias

It is time to drop hydroxychloroquine from our COVID-19 armamentarium