Beyond these studies of individual patients, we have seen what happens in large populations when these drugs are used. These have been "natural experiments." In the northern Brazil state of Pará, COVID-19 deaths were increasing exponentially. On April 6, the public hospital network purchased 75,000 doses of azithromycin and 90,000 doses of hydroxychloroquine. Over the next few weeks, authorities began distributing these medications to infected individuals. Even though new cases continued to occur, on May 22 the death rate started to plummet and is now about one-eighth what it was at the peak.
A reverse natural experiment happened in Switzerland. On May 27, the Swiss national government banned outpatient use of hydroxychloroquine for COVID-19. Around June 10, COVID-19 deaths increased four-fold and remained elevated. On June 11, the Swiss government revoked the ban, and on June 23 the death rate reverted to what it had been beforehand. People who die from COVID-19 live about three to five weeks from the start of symptoms, which makes the evidence of a causal relation in these experiments strong. Both episodes suggest that a combination of hydroxychloroquine and its companion medications reduces mortality and should be immediately adopted as the new standard of care in high-risk patients.
Right now, there is more observational evidence that suggests no effect than effect, whether it's used for prevention or for treatment.
To counter those claims from the Newsweek article, US saw decreases in deaths from April to May. It does not mention control for confounding between the two points in Brazil (e.g., if more old people had cases in April then most young people had cases in May).
Swiss one doesn't account for what outpatient or inpatient use of the drug was like before or after the ban. Additionally, the ban was only on outpatient use, so the critically ill could still receive it inpatient. Also, the deaths did quadruple, but it was from about 2 to 8. Switzerland has been relatively flat from about May 12 onward.