Long COVID prevalence and risk factors in adults residing in middle- and high-income countries…, 2025, Hermans+

[SNT Gatchaman]

Long COVID prevalence and risk factors in adults residing in middle- and high-income countries: secondary analysis of the multinational Anti-Coronavirus Therapies ACT trials
Lucas Etienne Hermans, Sean Wasserman, Lizhen Xu, John Eikelboom

BACKGROUND
During the recent COVID-19 pandemic, reports of long-term persistence or recurrence of symptoms after SARS-CoV-2 infection emerged, which are now collectively referred to as ‘long COVID’. Most descriptions of long COVID originate from patients residing in high-income countries. We set out to characterise long COVID in a large-scale clinical trial that was conducted in low-middle, high-middle and high-income countries.

METHODS
The Anti-Coronavirus Therapies trials enrolled 6528 adult patients with symptomatic COVID-19 in Argentina, Brazil, Canada, Colombia, Ecuador, Egypt, India, Nepal, Pakistan, Philippines, Russia, Saudi Arabia, South Africa and the United Arab Emirates. Long COVID was defined as the presence of patient-reported symptoms at 180 days after enrolment. Multivariable logistic regression was used to evaluate associations of baseline characteristics with long COVID.

RESULTS
Of 4697 included participants, 1181 (25.1%) reported long COVID symptoms. The most frequently reported symptoms were sleeping disorders (n=601; 12.8%), joint pain (n=461; 9.8%), fatigue (n=410; 8.7%) and headaches (n=382; 8.1%). Long COVID prevalence was higher in participants from lower middle-income compared with high-income countries (29.8% (850/2854) vs 14.4% (102/706); adjusted OR (aOR) 1.53 (1.10 to 2.14); p=0.012). Prevalence also varied between participants of different ethnic backgrounds and was highest (36.1% (775/2145)) for patients of Arab/North African ethnicity. Patients requiring inpatient admission were at increased risk of long COVID (aOR: 2.04 (1.63 to 2.54); p<0.001). Other independent predictors of long COVID were male sex, older age and hypertension. Vaccination, prior lung disease, smoking and diabetes mellitus conferred protective effects.

CONCLUSION
Symptoms of long COVID are reported in a quarter of cases of symptomatic COVID-19 in this study and were significantly more prevalent in participants from countries with lower income status and in patients of Arab/North African ethnicity. Research to further assess the health burden posed by long COVID in low-and middle-income countries is urgently needed.

Link | PDF (BMJ Global Health) [Open Access]

 

[Utsikt]

Long COVID was defined as the presence of patient-reported symptoms at 180 days after enrolment.

Just symptoms, not new onset?

 

[Hutan]

Patients requiring inpatient admission were at increased risk of long COVID (aOR: 2.04 (1.63 to 2.54); p<0.001). Other independent predictors of long COVID were male sex, older age and hypertension. Vaccination, prior lung disease, smoking and diabetes mellitus conferred protective effects.

It looks as though the Long Covid here has a lot to do with severe acute disease - with the predictors of persisting symptoms being male sex, older age and hypertension. That would be consistent for poorer quality medical care during the acute infection, as might be expected in a comparison of less wealthy and more wealthy countries.

(Edit - see my post below, the abstract is misleading, persisting symptoms were more commonly reported by females, although the effect was not very large. )

Prior lung disease, smoking and diabetes protective against Long Covid? That's a bit weird. Perhaps they did have a requirement for new onset of symptoms? That might mean that people who had some breathlessness before the infectionwould not qualify for Long Covid, even if they became more breathless.

________

It's a shame this study doesn't look to be very relevant to ME/CFS-type Long Covid, as it would be really interesting to get some idea about differences in prevalence by country.

 

[Hutan]

This study emphasises that the morbidity burden conferred by long COVID may fall disproportionately on countries with the least capacity to carry it, and frequently affects individuals that are likely to be under-represented in interventional studies that are being conducted in the global North.

This study is a secondary analysis of data collected in the Anti-Coronavirus Therapies (ACT) trials. The ACT trials were two international multicentre parallel group 2×2 factorial trials, respectively enrolling 3917 outpatients and 2749 inpatients with symptomatic COVID-19.

For the outpatient trial, patients were eligible for inclusion if they were symptomatic with laboratory-confirmed COVID-19 and within 7 days (ideally 72 hours) of diagnosis or worsening clinically (but not requiring hospitalisation). Patients above 30 years of age were eligible, and those below 70 years had to have at least one additional risk factor for disease progression, including male sex, body mass index of at least 30 kg/m², chronic cardiovascular, respiratory, or renal disease, active cancer or diabetes. For the inpatient trial, patients were eligible for inclusion if they were symptomatic with laboratory-confirmed COVID-19 disease and aged 18 or older. Patients were eligible for recruitment within 72 hours of their admission to the hospital or, in patients already hospitalised, within 72 hours of clinical worsening (eg, requiring ventilation).

Based on published data, we included the following as symptoms of long COVID6: muscle pain, shortness of breath with activities of daily living, anosmia/ageusia, diarrhoea, fatigue, headaches, sleeping disorder, palpitations, altered concentration, memory impairment and joint pain. We defined long COVID as the presence of at least one long COVID-associated symptom at 180 days after enrolment.

 

[Hutan]

Other independent predictors of long COVID were male sex, older age and hypertension. Vaccination, prior lung disease, smoking and diabetes mellitus conferred protective effects.

That makes it sound as if being male made it more likely that people met the Long Covid definition. But, it didn't.

In multivariable analysis, male sex was associated with a reduced risk of long COVID (aOR: 0.81 (0.68 to 0.95); p=0.009), and more advanced age was associated with an increased risk of long COVID (aOR: 1.02 (1.02 to 1.03); p<0.001). Participants from lower middle-income countries more frequently reported symptoms than patients in upper middle-income countries and high-income countries (29.8% (850/2854) vs 20.1% (229/1137) vs 14.4% (102/706)).

51.6% of the people with Long covid were male; 63.1% of the people without Long Covid were male.

 

[Hutan]

Age
The effect of age was statistically significant very small. The mean age of the people with Long Covid was 48.0, the mean age of the people without Long covid was 45.2.

In multivariable analysis, male sex was associated with a reduced risk of long COVID (aOR: 0.81 (0.68 to 0.95); p=0.009), and more advanced age was associated with an increased risk of long COVID (aOR: 1.02 (1.02 to 1.03); p<0.001).

Ethnicity
There's a really big difference.
For Arab/North African: 65.6% of the people with Long Covid were Arab/North African; only 39% of the people without Long Covid were Arab/North African
For South Asian: 9.6% of the people with Long Covid were South Asian; 24.1% of the people with Long Covid were South Asian
A South Asian person had an adjusted odds ratio of getting Long covid of just 0.18 compared to an Arab/North African.

The increased risk of long COVID for participants residing in lower middle-income countries, the lowest stratum of country income level represented in this cohort—Egypt, India, Nepal, Pakistan and the Philippines—is a concerning signal. Given the significant variability in symptom reporting between participants of different ethnicities, with very high rates of long COVID reported by participants of Arab/North African ethnicity, this variable was included in the logistic regression model.

While ethnicity did attenuate the effect of country income level on the outcome, a statistically significant risk remained. The increased risk for long COVID in lower middle-income countries may reflect underlying health disparities between these and high-income countries. The encountered variability in the prevalence of long COVID among different ethnical groups encountered in this study may be the result of underlying environmental or genetic variation. Alternatively, they may be explained by cultural and linguistic differences, resulting in differential subjective symptom reporting rates.

So, there was a relationship with the country income level, although ethnicity also had an impact. The authors note the differences could be environmental or genetic, but there may also be cultural and linguistic effects on symptom reporting rates.

I'd also possibly add regional variation in sex differences in study participation and symptom reporting to that list. I didn't see the percentages of males from each country or ethnicity reported.

 

[Hutan]

Vaccination, prior lung disease, smoking and diabetes mellitus conferred protective effects.

Vaccination
There was a big difference, vaccination does appear to be protective, and it makes sense that it would be. Vaccinated people would be less likely to have a severe illness leaving tissue damage. But, vaccination is surely correlated with a good health systems, and personal wealth, so I don't think they should be saying that vaccination definitely is protective.

Vaccination conferred a protective effect when compared with unvaccinated participants (aOR: 0.34 (0.27 to 0.43); p<0.001 for fully vaccinated individuals

Prior lung disease
It's true that prior lung disease was associated with not having Long Covid, but the numbers were fairly small. There could be some gender effect, or what I mentioned before - if you were breathless before getting Covid, being breathless afterwards probably was not counted as a Long Covid symptom.

Smoking
The so-called protective effect of smoking was pretty marginal I think, and might just be the result of not enough adjustment for things like sex differences in smoking rates.

Diabetes mellitus
The so-called protective effect of diabetes is also very debatable. An association of diabetes with an increased risk of Long Covid (OR=1.07, with the 95% Confidence limits crossing 1) changed to an association of diabetes with a decreased Risk of Long Covid after adjustment (aOR of 0.74).

There is also the complication that to be included in the outpatient trial you had to have a risk factor for severe disease - be male, have diabetes etc. So, there was probably a higher percentage of people with diabetes in the outpatient trial than in the inpatient trial. And people in the inpatient trial had a much greater risk of "Long covid".

[QUOTE="for inclusion in outpatient trial" ]Patients above 30 years of age were eligible, and those below 70 years had to have at least one additional risk factor for disease progression, including male sex, body mass index of at least 30 kg/m², chronic cardiovascular, respiratory, or renal disease, active cancer or diabetes.[/QUOTE]

 

[Hutan]

Last comment:

This study is a secondary analysis of data collected in the Anti-Coronavirus Therapies (ACT) trials. The ACT trials were two international multicentre parallel group 2×2 factorial trials, respectively enrolling 3917 outpatients and 2749 inpatients with symptomatic COVID-19. Both trials compared colchicine and usual care versus usual care alone.

The outpatient trial further compared aspirin 100 mg daily and usual care versus usual care, and the inpatient trial compared the combination of aspirin 100 mg daily and rivaroxaban 2.5 mg two times per day and usual care versus usual care.12 In both trials, none of the tested interventions were shown to be of benefit in reducing the primary outcome.13 14

This is an interesting study because it was undertaken simultaneously in a diverse set of countries. I would like to understand the differences by ethnicity - was there a difference in how males and females reponded? Was there a difference in how the symptom descriptions were understood? Is there a real difference in Long covid prevalence by ethnicity? I don't think we have seen much epidemiology coming out of South Asia yet.

The treatments were not found to be useful for acute Covid-19, and neither did they help protect against Long Covid.
Aspirin can thin the blood and reduces pain and swelling
Colchicine is said to reduce inflammation and the build-up of uric acid
Rivaroxaban is an anti-clotting medicine