Research Signal – Source date 2022-12-13
Age is one of the clearest risk factors for severe COVID-19, but birthdays do not fully describe a person’s health. People of the same chronological age can have different histories, conditions and physical resilience. A December 2022 study asked whether measurements associated with biological ageing could help explain some of that variation.
The analysis used participants in the UK Biobank who had been assessed at centres in England. It compared three ageing-related measures with COVID-19 outcomes: a biomarker-based estimate of biological age, relative telomere length and a questionnaire measure of perceived facial age. The outcome was defined as COVID-19-related hospitalisation or death.
This was an observational analysis, not a test of an anti-ageing treatment. Baseline biological information came from assessments conducted in 2006–2010. COVID-19 records covered the period from 16 March 2020 through the UK Biobank’s December 2021 update. By then, 27,909 participants had either a positive test or a clinical diagnosis of COVID-19, and 4,016 had experienced the defined severe outcome.
Birthdays capture only part of ageing
Chronological age is easy to record, but it is an imperfect summary of biological condition. Two people with the same number of years since birth may differ in blood markers, immune function, accumulated disease or other features that are not visible in their age alone.
The researchers therefore examined whether people who appeared biologically older than their chronological age were more likely to experience hospitalisation or death after diagnosed COVID-19. They adjusted their statistical models for factors including sex, ethnicity, smoking, alcohol status, body mass index, deprivation and pre-existing conditions. Chronological age remained strongly associated with the outcome: after adjustment for the study’s covariates, its odds ratio was 1.080, with a 95% confidence interval of 1.074–1.086.
An odds ratio describes the relationship estimated by a model; it is not the same as an individual’s probability of hospitalisation or death. The main question was whether any ageing measure added information after the model already accounted for chronological age and other characteristics.
Three different views of biological age
The first measure was phenotypic age acceleration. PhenoAge combines nine biomarkers with chronological age. The researchers compared that estimate with age at assessment and expressed the difference in years. A positive value indicated that a participant’s measured phenotype appeared older than expected for their chronological age; a negative value indicated the opposite. It was a calculated score, not a direct reading of an underlying age inside the body.
The second measure was relative leukocyte telomere length. Telomeres are structures at the ends of chromosomes that help protect chromosome integrity. In this study, telomere length in white blood cells was measured using a relative ratio adjusted for technical influences. Telomere length has been proposed as an ageing-related measure because it is connected with cell division and cellular senescence, but it represents a different biological feature from the multi-biomarker PhenoAge estimate.
The third measure was much less laboratory-based. Participants answered whether people said they looked younger than their age, about their age or older than their age. The study treated these answers as categories of perceived facial age. The measure captured a participant’s report about how other people described their appearance; it was not a facial photograph analysis.
These instruments could all be described as ageing-related, but they do not measure the same thing. One combines biomarkers and age, one measures a chromosome-end structure in blood cells, and one records a subjective perception of appearance. Comparing them offered a way to see whether the same association appeared across very different measures.
The adjusted models did not tell the same story
Phenotypic age acceleration showed the clearest and most persistent association. In a model adjusted for chronological age and the study’s other covariates, each additional year of phenotypic age acceleration was associated with an odds ratio of 1.029 for severe COVID-19 outcomes. The 95% confidence interval was 1.020–1.039.
In plain terms, the model associated a one-year difference in the score with about 2.9% higher odds of the defined severe outcome. That estimate remained statistically significant after chronological age was included. The association became smaller as more variables were added, but it did not disappear.
The telomere result changed more substantially with adjustment. Longer relative telomere length was associated with lower odds of severe outcomes in simpler models. After chronological age was included, however, the association was no longer statistically significant. This does not show that telomere biology has no role in COVID-19. It shows that this measure did not retain a clear independent association in the study’s final age-adjusted model.
Perceived facial age produced another pattern. Participants who reported that others considered them younger than their age had an adjusted odds ratio of 0.847, with a 95% confidence interval of 0.772–0.929. The category describing people as older-looking was not statistically significant in the final model. The younger-looking result is therefore an association with a questionnaire category, not evidence that looking younger protects someone from severe infection.
Chronological age remained relevant throughout these comparisons. A biological-age estimate can be associated with an outcome after age adjustment without replacing age as a useful risk factor. Nor does an adjusted association by itself show that the additional measure improves a clinically validated prediction model.
Ethnic-group patterns need careful interpretation
The researchers also repeated parts of the analysis within White, Black, Asian and Other ethnic groups. The adjusted association between phenotypic age acceleration and severe COVID-19 outcomes was positive in White participants, with an odds ratio of 1.031 and a 95% confidence interval of 1.021–1.042. It was also positive in Black participants, with an odds ratio of 1.049 and a 95% confidence interval of 1.001–1.100. Results for Asian and Other participants were not statistically significant in the main subgroup models.
Those findings do not establish that the measure works differently across ethnic groups. A statistically significant result in one group and a nonsignificant result in another are not, on their own, proof that the underlying effects differ. The size and precision of each estimate matter, as do the differences in the people represented in each analysis.
The UK Biobank sample was highly selective, with a reported response rate of 5.5%, and it contained about 94% White participants. That composition does not represent the ethnic distribution of the wider UK population. Smaller numbers in minority groups make estimates less precise and can make it harder to detect an association even when one may exist. The models that included comorbidities also reduced the number of participants by more than 10% in most groups, adding another complication when results are compared.
The subgroup analysis therefore provides a useful question for future research rather than a settled explanation of ethnic differences in COVID-19 outcomes. It points to an association that appeared in two groups in this dataset, while leaving the broader pattern inconclusive.
Possible biology is not the same as demonstrated causation
The authors discussed two connected processes that might help explain why accelerated ageing could be associated with severe disease. One is inflammaging, the tendency for the body’s pro-inflammatory state to increase with age. A more inflammatory baseline could contribute to excessive inflammation during infection. The other is immunosenescence, age-related change in immune function. The paper linked this idea to reduced T-cell proliferation and a weaker adaptive response, particularly in the context of shorter telomeres and the loss of T cells that can occur during COVID-19.
These mechanisms are plausible explanations for an observed relationship, but the statistical design cannot decide whether they caused the outcomes. The analysis measured ageing-related features before the pandemic and then linked them with later records. It did not test whether changing a person’s ageing score, lengthening telomeres or altering facial appearance would change the course of infection. The pre-pandemic measurements also cannot establish that COVID-19 accelerated participants’ ageing.
Several details limit how confidently the findings can be applied. The phenotypic-age calculations used baseline measurements collected roughly 10–15 years before the pandemic, so they may not have represented participants’ biological state in 2020 or 2021. The cohort was healthier and more selective than the general UK population. The analysis also did not adjust for every potentially relevant factor, including cancer and vaccination status, according to the authors’ discussion. Facial age was self-reported, and the telomere measure has limitations as a general biological-age indicator.
For those reasons, this study did not establish a personal risk test, prove that biological age outperforms chronological age, or demonstrate benefit from senolytics, anti-ageing drugs or telomerase-directed treatment. Its more limited contribution was to show that different ageing-related measures behaved differently when placed in the same observational analysis.
What biological age could add to COVID-19 risk assessment
A useful biological-age measure could eventually help clinicians identify people whose vulnerability is greater than their birthdays alone suggest. In a group of people with the same chronological age, a validated measure that adds reliable information could help focus preventive attention on those most likely to need it.
That possibility depends on demonstrating a real improvement beyond established risk factors, not simply finding a statistically significant association. The measure would also need to perform reliably across relevant populations, including groups that were underrepresented in this UK Biobank analysis. As of the study’s December 2022 publication, the evidence supported further investigation of that capability, not its use as an established clinical tool.
Key Sources
- Primary source: Accelerated ageing is associated with increased COVID-19 severity and differences across ethnic groups may exist.
DOI or Journal Page · Peer-Reviewed Paper · 2022-12-13
Keep exploring