Longer incubation period of coronavirus disease 2019 (COVID‐19) in older adults

Abstract Objective The aim of this study was to explore any age‐related change in the incubation period of COVID‐19, specifically any difference between older (aged ≥65 years) and younger adults. Methods Based on online data released officially by 21 Chinese cities from January 22 to February 15, 2020, the incubation period of COVID‐19 patients who had travelled to Hubei was studied according to age. Previous studies were reviewed and compared. Results The study recruited 136 COVID‐19 patients who had travelled to Hubei during January 5‐31, 2020, stayed for 1‐2 days, and returned with symptom onset during January 10‐February 6, 2020. The median age was 50.5 years (range 1‐86 years), and 22 patients (16.2%) were aged ≥65 years. The age‐stratified incubation period was U‐shaped with higher values at extremes of age. The median COVID‐19 incubation period was 8.3 (90% confidence interval [CI], 7.4‐9.2) days for all patients, 7.6 (90% CI, 6.7‐8.6) days for younger adults, and 11.2 (90% CI, 9.0‐13.5) days for older adults. The 5th/25th/75th/90th percentiles were 2.3/5.3/11.3/14.2 days for all, 2.0/5.0/10.5/13.2 days for younger adults, and 3.1/7.8/14.4/17.0 days for older adults. There were 11 published studies on COVID‐19 incubation periods up to March 30, 2020, reporting means of 1.8‐7.2 days, and medians of 4‐7.5 days, but there was no specific study on the effect of age on incubation period. One study showed that severe COVID‐19 cases, which included more elderly patients, had longer incubation periods. Conclusion Based on 136 patients with a travel history to Hubei, the epicenter of COVID‐19, the COVID‐19 incubation period was found to be longer in older adults. This finding has important implications for diagnosis, prevention, and control of COVID‐19.

1 133 758 globally in over 100 countries and territories, associated with 62 784 deaths. 3 Older adults are at higher risk of contracting COVID-19, developing a severe disease, and dying from the disease, [4][5][6] similar to what was experienced during the severe acute respiratory syndrome (SARS) epidemic in 2003. 7,8 In China, 90% of the COVID-19 deaths occurred in those older than 60 years, and 20% of all deaths were people aged older than 80 years. 4 In Europe, where the proportion of elderly population (19% aged over 65 years) is higher, the impact is greater: over 95% of the COVID-19 deaths has occurred in those aged over 60 years, and more than 50% of all deaths have been people aged over 80 years. 6 Because of immune senescence and common occurrence of secondary immunodeficiency in old age, the presentations and disease course of older adults in response to infectious disease may be altered. 9,10 A review of SARS in old age 7 notes that, according to clinical experience, its incubation period, defined as the time between infection exposure and symptom onset, is apparently longer in older adults, but there is a lack of studies on the incubation periods of SARS-associated coronavirus, specifically in older adults. Knowing the incubation period of this novel virus is relevant for diagnosis, surveillance, prevention, and control of COVID-19, and it is important to know whether its incubation period in older adults, who represent a vulnerable group to this disease, differs from that in younger adults.
The objective of this study was therefore to explore whether there is any age-related change in the incubation period of COVID-19, specifically any difference between older (aged ≥65 years) and younger adults.

| Data source and collection
For nearly every city in China, daily information on the list of COVID-19 cases is released officially to the Chinese social media WeChat accounts of respective cities. However, only a minority of cities include in their official release clear information on the day of symptom onset, which is required in estimating incubation period. The information released between January 22 and February 15, 2020 was captured and compiled into a list of patients with COVID-19 from cities that reported the day of symptom onset and those who had travelled to Hubei, the epicenter of the COVID-19 epidemic. The following data were entered into an Excel (Microsoft) spreadsheet for COVID-19 cases reported between January 22 and February 15, 2020 from 21 Chinese cities outside Hubei: patient case number, age, sex, first day to Hubei, last day in Hubei, and first day with symptoms. The data were collected by another person acknowledged at the end of this paper.

| Data analysis
For this study, only those COVID-19 patients who stayed in Hubei for at most two calendar days were included. The day of exposure was taken as the first day to Hubei if the patient had stayed in Hubei for one calendar day; or as the middle of the first day and second day in Hubei if the patient had stayed for two calendar days. By excluding COVID-19 patients who had stayed in Hubei for more than 2 days, a narrowly defined exposure window was ensured, thus reducing the uncertainty in estimating the day of exposure. The incubation period for each COVID-19 patient was derived from the number of days between exposure and symptom onset. Parametric and non-parametric methods were used in the statistical analysis. The COVID-19 incubation periods for different age groups of 15-year interval size were analyzed in terms of their respective mean, median, 25th, and 75th percentiles. Next, the frequency distributions of COVID-19 incubation periods of the three age groups-children (age 0-14 years), younger adults (age 15-64 years), and older adults (aged ≥65 years)-were plotted and compared. The cumulative frequency distributions for the incubation periods for both younger and older adult groups were analyzed using the cumulative frequency analysis software CumFreq. 11 This computer program fits the observed cumulative frequency data into a best-fit theoretical distribution by the regression method. Observed data are plotted in ranked order of increasing probability, and the most appropriate theoretical distribution is selected based on the lowest average of absolute values of the differences between observed and calculated cumulative frequency values. Further details are available at the CumFreq website. 11 With the use of CumFreq, probability distribution curve fitting was performed for the incubation periods for both younger and older adult groups, and their medians, quartiles, 5th, and 90th percentiles estimated.

| Literature review
The author searched the studies reporting on incubation periods of COVID-19 archived in PubMed (published) and medRxiv (unpublished) until March 30, 2019. The following search terms were used: ("incubation") and ("COVID-19" or "SARS-CoV-2"). Studies reporting on the incubation period of COVID-19 were included for further evaluation and comparison with this study. Information on study place, sample size, mean/median age, percentage of subjects studied who were elderly, and incubation period was extracted from these previous studies.

| RE SULTS
A total of 136 COVID-19 patients were recruited into this study.
They had travelled to Hubei from 21 Chinese cities between January 5 and January 31, 2020, had stayed there for at most two calendar days, and had returned to their respective cities with symptom onset between January 10 and February 6, 2020. The median age was  Figure 2). There was a skew towards the left for younger adults, but for older adults, a skew towards the right with wider variation was noted. Cumulative frequency distributions of the entire study population, the younger (15-64 years), and older (≥65 years) adult age groups were further analyzed by the CumFreq 11 software ( Figure 3). Kumaraswamy distribution returned as the best fit for the entire study population and the younger adult group, while the mirrored generalized Gumbel (also known as the log-Weibull distribution) returned as the best fit for the older adult group. This gave estimates of the non-parametric values of the COVID-19 incubation period for the total study population, and the younger and older adult age groups, as shown in Table 1. The medians were 8.3 (90% CI, 7.4-9.2) days for all patients, 7.6 (90% CI, 6.7-8.6) days for younger adults, and 11.2 (90% CI, 9.0-13.5) days for older adults.
The literature review retrieved 11 published 5,12-21 and two unpublished 22,23 studies that reported on COVID-19 incubation periods. Their details are listed in Table 2.

| D ISCUSS I ON
The first estimate of the COVID-19 incubation period was reported by Li et al 12 (Table 2). This is slightly longer than the incubation period estimated for SARS, with a median of 4.0 days (95% CI, 3.6-4.4) and a 95th percentile of 10.6 days (95% CI, 8.9-12.2). 24 Subsequent published studies 5,12-21 on the COVID-19 incubation period (Table 2) have reported means varying from 1.8 days to 7.2 days, medians of 4-7.5 days, and 95th percentiles of 3.2-14.6 days, which may be due to differences in methodologies and patient samples. In this study, the incubation period of COVID-19, when age-stratified, displayed a U-shaped curve with higher values at the extremes of age for the pediatric and geriatric age groups (Figure 1)    Although previous studies on COVID-19 incubation period did not look specifically at the effect of older age on incubation period, I reviewed these studies for any such clues. Of the ten published studies 5,12-20 on COVID-19 incubation period ( Table 2) that included adults, eight had information on mean or median age. Except for the first study by Li et al, 12 which included much older subjects in the parent population but without age information for the 10 patients  Figure 4), it was noted that the incubation period for the severe group (whose subjects were older with a median age of 61.4 years and 43.5% aged over 65 years) had a longer median incubation period of 7.5 ± 7.2 days when compared with 6.5 ± 4.6 days for the milder group (whose subjects were younger with a median age of 44.5 years and 13% aged over 65 years). The unpublished study by

TA B L E 2 Previous studies on COVID-19 incubation period
Jing et al 22 reports on a long COVID-19 incubation period (mean, F I G U R E 4 COVID-19 incubation period (IP) according to age in previous and present studies: (A) mean or median age, (B) percentage aged ≥65 years. Substudies are indicated by arrows from main study. "Age" refers to age of study sample of incubation period when available, or to age of parent sample when age is not available for study sample 8.62 days; median, 8.13 days) that is close to the present figures, but their sample was younger (median age 40 years; 13.2% aged over 60 years) and a different methodology was used as discussed above.

The incubation period of an infection is influenced by infectious
dose and immune response. Thus, the shorter COVID-19 incubation period among travelers to Hubei compared with non-travelers in the study by Leung 19 was attributed to exposure to a higher viral load among travelers to Hubei, the epicenter. The longer COVID-19 incubation period observed in this study for older adults may be accounted for by the blunted immune response due either to age-associated immune senescence or secondary immunodeficiency in old age. 7,9,10 The lack of a fever response in elderly subjects, the non-specific geriatric presentations in an infectious illness (such as falls and delirium), and multi-comorbidities might result in a delayed awareness of disease onset and its detection by a clinician. 6  Third, the study sample was drawn from a listing of hospitalized individuals. Thus, milder community cases of COVID-19 were not represented in this study.

ACK N OWLED G M ENTS
I would like to acknowledge Dr Ally Bi-zhu Jiang, PhD, for providing collected data for my study and analysis.

CO N FLI C T S O F I NTE R E S T
Nothing to disclose.

AUTH O R CO NTR I B UTI O N S
The author is responsible for design, conceptualization, literature review, data analysis, and writing of paper.