Skip to main content

Lymphopenia as a Biological Predictor of Outcomes in COVID-19 Patients: A Nationwide Cohort Study

Our take —

This cohort study of 4,052 individuals with laboratory-confirmed SARS-CoV-2 infection in South Korea at the start of the COVID-19 pandemic assessed the relationship between absolute lymphocyte count after COVID-19 diagnosis and 28-day mortality. While they found increased 28-day mortality among those with low baseline absolute lymphocyte counts, suggesting the utility of absolute lymphocyte counts as a potentially appealing biomarker of COVID-19 outcomes, their analytic design did not adequately account for potential residual confounding, which limits the ability to make a causal inference about the relationship.

Study design

Retrospective Cohort

Study population and setting

This cohort study included data collected from people with laboratory-confirmed SARS-CoV-2 infection from a national communicable diseases registry in South Korea between January and April 2020, all of whom, regardless of their symptoms, were admitted to designated hospitals for isolation. Among 5628 individuals with registered data, 4,052 had baseline absolute lymphocyte count (ALC) available. The researchers assessed the relationship between baseline lymphopenia (lymphocyte count < 1,000/mm^3) and 28-day COVID-19 mortality. They used multivariable logistic regression to create propensity scores to predict likelihood of baseline lymphopenia. Variable selection was based on univariate associations with lymphopenia and included participants’ age, sex, blood pressure, medical comorbidities, and baseline vital symptoms and blood counts. They used nearest neighbor matching to create two 3:1 propensity-score matched cohorts (group II: ALC 500 to < 1,000/mm^3; group III: ALC greater than or equal to 1,000/mm^3) to the 110 patients with baseline ALCs < 500/mm^3 (group I). They then assessed time-to-event across all participants and propensity-score matched groups with Kaplan-Meier curves and log rank tests. Finally, they used multivariable logistic regression to assess the likelihood of mortality given predictors (including baseline ALC group) selected due to their univariate association with COVID-19 mortality among the propensity score matched cohort (n = 770).

Summary of Main Findings

Among the 4,052 individuals with baseline ALC data, 82.6% (95% CI: 79.8, 85.6) of the 786 participants with baseline lymphopenia (ALC < 1,000/mm^3) survived 28 days from when they were admitted to COVID-19 isolation hospitals compared to 98% (95% CI: 97.4, 98.5) of the 3,266 participants with normal baseline ALC (p-value < 0.001). In the propensity score-matched groups, 62.7% (95% CI: 54.0, 72.9) of individuals in group I survived to 28 days as compared to 79.9% (95% CI: 75.4, 84.7) of individuals in group II and 89.0% (95% CI: 85.6, 92.5) of individuals in group III (p < 0.001). There were also lower rates of mechanical ventilation, intensive oxygen supplementation, and any oxygen supplementation in propensity-matched individuals in group III as compared to groups II and I. Patients who presented with a mental disturbance (OR 11.09, 95% CI: 3.28, 47.25), comorbid dementia (OR 6.55, 95% CI: 3.84, 11.4), or an ALC < 500/mm^3 (group I OR 5.63, 95% CI: 3, 10.72) experienced the highest of odds of mortality during the study period.

Study Strengths

This study included a large number of participants and attempted to use propensity score methods to account for variables that may confound the relationship between baseline ALC and COVID-19 related mortality. Also, because of South Korea’s COVID-19 policy, all participants who tested positive to SARS-CoV-2, regardless of symptom presence or severity, were admitted to isolation hospitals, which allows their results to be more generalizable to those experiencing a variety of COVID-19-related symptoms.


The authors dichotomized a continuous outcome (absolute lymphocyte count), which makes the strong, likely inaccurate, assumption that a participant with an ALC of 999/mm^3 is more similar to a participant with an ALC of 100/mm^3 than to a participant with an ALC of 1001/mm^3. They also dichotomized several of the covariates they used to calculate the propensity scores (age, blood pressure, and vital signs) all of which likely lead to increased residual confounding in propensity score matched individuals. These limitations were compounded by their univariate variable selection technique for the propensity score model and the multivariable logistic regression, which likely underestimated the uncertainty in their already uncertain estimates. Finally, it is unclear whether the participants with low baseline ALC levels had low ALC prior to their infection with SARS-CoV-2 or whether their ALC levels dropped below the normal range in response to SARS-CoV-2 infection, which has important implications for our mechanistic understanding of how ALC may be relevant to COVID-19-related mortality.

Value added

This study, which included a large number of participants with a variety of COVID-19 symptoms, suggests that an easily measured biomarker, absolute lymphocyte count, may have prognostic value in COVID-19-related mortality.

This review was posted on: 22 March 2021