A large study published yesterday in BMJ concludes that long-COVID symptoms in patients who had mild infections resolved within a year, but some physicians say the research design was flawed, and the findings don't match their clinical experiences, could provide false assurance, and may have unintended consequences for those with persistent symptoms.
Shortness of breath, weakness
Researchers at the KI Research Institute and Maccabi Healthcare Services in Israel analyzed the electronic health records of 1,913,234 Maccabi members of all ages tested for COVID-19 from Mar 1, 2020, and Oct 1, 2021. The team used a test-negative design to compare the risk of 70 reported long-COVID outcomes in unvaccinated, infected patients with that of uninfected patients, as well as with the risk of breakthrough infection in unvaccinated, infected controls.
The team concluded that COVID-19 was significantly linked to an elevated risk of anosmia (loss of smell) and dysgeusia (loss of taste) in both the early period (first 30 to 180 days; hazard ratio [HR], 4.59; risk difference, 19.6) and late period (181 to 360 days; HR, 2.96; risk difference, 11.0).
The risks were also elevated for cognitive impairment in the early (HR, 1.85; risk difference, 12.8) and late periods (HR, 1.69; risk difference, 13.3) and for shortness of breath (HR, 1.79 and 1.30; risk difference, 85.7 and 35.4, respectively), weakness (HR, 1.78 and 1.30; risk difference, 108.5 and 50.2), and heart palpitations (HR, 1.49 and 1.16; risk difference, 22.1 and 8.3).
The risk of streptococcal tonsillitis and dizziness were lower but still significantly elevated in both periods, while hair loss, chest pain, cough, myalgia, and respiratory disorders were significantly higher only in the early period. Risks among males and females were slightly different, and children had fewer persistent symptoms than adults in the early period, which mostly resolved by 1 year.
The results were similar, regardless of whether the wild-type, Alpha, or Delta SARS-CoV-2 variant was dominant. Vaccinated participants who later were infected had a lower risk of shortness of breath than their unvaccinated, infected counterparts but had a comparable risk of other lingering symptoms.
"This nationwide study suggests that patients with mild COVID-19 are at risk for a small number of health outcomes, most of which are resolved within a year from diagnosis," the authors concluded.
The problem with test-negative designs
Ziyad Al-Aly, MD, clinical epidemiologist at the Washington University School of Medicine in St. Louis and chief of research and development at the Veterans Affairs St. Louis Healthcare System, told CIDRAP News that the conclusions are puzzling. "They don't jibe with our understanding of long COVID, and I think I know why" he said.
The study's fundamental flaw, Al-Aly said, is its test-negative design. "People don't usually test for SARS-CoV-2 or for any other reason without an indication," he said. "People who seek care often do so for other reasons, and they are tested. This is not the normal population; these are people who had the need to test and tested positive."
In other words, "When you test people, you're testing against people who are sick," Al-Aly said. "It makes it look like COVID-positive people have a resolution of symptoms when in reality it's a suboptimal design that may give us a false assurance."
When the pandemic began, he said, many researchers thought a test-negative design was useful, but Al-Aly's group soon ran into confounding when conducting a COVID-19 study. "What we found is that people with a COVID-negative test had a higher risk of cancer, and it turns out that people who had come to the hospital for testing did so for chemotherapy or other reasons," he said. "The patients were not the average person in Israel or the United States or Italy."