Background
32
Understanding and monitoring the demographics of SARS‐CoV‐2 infection can inform strategies 33
for prevention. Surveillance monitoring has suggested that the age distribution of people 34
infected with SARS‐CoV‐2 has changed since the pandemic began, but no formal analysis has 35
been performed. 36
Methods
37
Retrospe
ctive review of SARS‐CoV‐2 molecular testing results from a national reference 38
laboratory was performed. Result distributions by age and positivity were compared between 39
early period (March‐April 2020) and late periods (June‐July 2020) of the COVID‐19 pandemic. 40
Additionally, a sub‐analysis compared changing age distribu tions between inpatients and 41
outpatients. 42
Results
43
There were 277,601 test results of which 19320 (7.0%) were positive. The median age of 44
infected people declined over time (p < 0.0005). In March‐April, the median age of positive 45
people was 40.8 years (Interquartile range (IQR): 29.0 – 54.1). In June‐July, the median age of 46
positive peop
le was 35.8 years (IQR: 24.0 – 50.2). The positivity rate of patients under 50 47
increased from 6.0 to 10.6 percent and the positivity rate for those over 50 decreased from 6.3 48
to 5.0 perc ent between the early and late periods. The trend was only observed for outpatient 49
populations. 50
Conclusions
51
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
3
We confirm that there is a trend toward decreasing age among persons with laboratory‐52
confirmed SARS‐CoV‐2 infection, but that these trends seem to be specific to the outpatient 53
population. Overall, this suggests that observed age‐related trends are driven by changes in 54
testing patterns rather tha n true changes in the epidemiology of SARS‐CoV‐2 infection. This calls 55
for caution in interpretation of routine surveillance data until testing patterns stabilize. 56
57
Introduction
58
Understanding of the demographics of persons infected with SARS‐CoV‐2 is essential for 59
informing the public health response to the COVID‐19 pandem ic. Age is a major factor in determining 60
the risk of severe illness outcomes[1, 2], so data on the age distribution of infected persons can help 61
guide expectations about demands on hospital resources. Knowledge of which age groups are highly 62
infected is also important for designin g effective interventions.[3] In the United States, surveillance data 63
suggest that mean age of infected patients is decreasing compared to the early stages of the COVID‐19 64
pandemic. In Washington State, for example, 35% of detected cases in March were aged 60 years or 65
older, compared to 12% in July .[4] 66
However, testing practices for SARS‐CoV‐2 have changed dramatically over the course of the 67
COVID‐19 epidemic in the United States. The average daily number of SARS‐CoV‐2 tests has increased 68
from approximately 35,000 in March to 676,000 in July.[5] Thus, it is uncl ear whether changes in the age 69
distribution of detected cases represent changes in the epidemiology of COVID‐19, changes in testing 70
practices, or a combination of the two. We used SARS‐CoV‐2 testing data from a national reference 71
laboratory to characterize the age distribution of dete cted cases between March and July of 2020. 72
73
74
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
4
Methods
75
The study was covered under exemption umbrella deidentified data (Utah IRB 00082990). 76
Population: We retrospectively reviewed of all SARS‐COV‐2 test results performed at ARUP 77
Laboratories from March 10, 2020 to July 8, 2020. ARUP Laboratories (Salt Lake City, Utah) is a national 78
Reference
laboratory that provides clinical testing for over 1000 hospitals across the United States. ARUP 79
has offered SARS‐CoV‐2 testing since March 2020. Patients who received positive test results were 80
presumed to been infe
cted. We refer to this group as the positive population which is an estimate of 81
the infected population. 82
Testing: All testing was performed on combination of three high throughput, automated 83
molecular assays: Hologic (75%), Roche (18%) and ThermoFisher (7%). These assays have similar limits 84
of detection and are among the most sensitive assays developed to date. 85
Statistical Analysis: We divided the results into two periods: 1) March 10th ‐ April 30th (early 86
period), 2) June 1st ‐July 8th (late period). We compared the age distribution of positive cases in the 87
early and late periods by calculating the medians and interquartile ranges (IQR). To explore differential 88
changes in testing for severely ill patients vs. patients with milder disease, we also stratified tests 89
ordered within the University of Utah He althcare system by inpatient vs. outpatient facilities. 90
The age distributions (early period vs late period) were tested for equality using the Kruskal‐91
Wallis test. The Wilk‐Shapiro test was used to test for normality. The median test, which is based on 92
Pearson’s chi square test, (as imple mented in Stata) was used to test for equality of medians. 93
Calculations were performed using Stata 16 (Stata Corporation, College Station, TX). 94
95
96
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
5
Results
97
Population characteristics: Combined, 277,601 test results were reported in the early and late 98
periods. Of these,19,320 (7.0%) were positive. Approximately half (48%, N=134,253) of the results were 99
from Utah. Samples were obtained from 40 additional states. In addition to Utah, 22 states had over 100
1000 results each, which accounted for 49% of the total results (Table 1). The percentage of positive 101
cases in the Utah and non‐Utah cohort
s and was 6.94% and 6.97% (p=0.81). The median age of positive 102
patients in the non‐Utah cohort (42.1 years IQR: 27.1 – 58.4) was greater than the Utah cohort (35.8 103
years; IQR: 24.2 – 48.0) (𝜒ଵ
ଶ=254, p<0.0005). 104
We were able to determine inpatient status for 6973 of 9327 (75%) positive patients in the Utah 105
cohort, of whom, 6865 (98.5%) were outpatients and 108 (1.5%) were inpatients. The median ages of 106
positive outpatients and inpatients were 35.4 years (IQR: 23.8 – 47.8) and 53.6 years (IQR: 43 .1 – 67.4), 107
respectively (p < 0.001). 108
Change in age distribution of positive cases over time: The overall median age for positive cases 109
was 38.4 years (IQR: 25.7 – 53.3). The median age for all positive cases decreased by 5.0 years between 110
the early and late periods from a median age of 40.8 years (IQR: 29.0 – 54.1) to 35.8 years (IQR: 24.0 – 111
50.2) late period (June‐July) (p < 0.001) ( Figure 1, Table 2). The pattern was similar when comparing 112
Utah to all other states combined (Figure 1, Table 2). 113
Within Utah, the overall median age for all positive results decreased from 38.5 to 34.3 years. 114
When the Utah cohort was stratified based on inpatient and outpatient me dical care, the median age of 115
inpatients increased by 5.8 years between the early and late period (Table 2, Figure 2). In contrast, the 116
median age of outpatients decreased by 3.9 years. 117
Change in infection rates by age group over time: The positivity rate increased over time for 118
almost all age groups; however, the increase was greatest among younger people (Table 3). For 119
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
6
example, among those younger than 18 years, the positivity rate increased from 3.3% to 10.0% between 120
the early and late periods. Similarly, the positivity rate increased from 6.1% to 11.5% and from 6.2% to 121
10.1% for people in the 18‐29 and 30‐39 age groups. In contrast, the infe ction rate decreased from 5.5% 122
to 4.4% for people within 60‐69 age group and decreased from 6.1% to 3.6% for those 70 years old and 123
older. Overall, the positivity rate of patients under 50 years old increased from 6.0% to 10.6% and the 124
percent of positive results for those over 50 years old decr eased from 6.3% to 5.0%. (Patients aged 125
50.00 – 50.99 years were included in the over 50 age group). 126
127
Discussion
128
Surveillance data in the United States have shown a trend toward decreasing age among 129
persons with laboratory‐confirmed SA RS‐CoV‐2 infection. This study found a similar pattern among 130
patients tested by a national reference laboratory, with the median age among patients testing positive 131
being five years lower in June and early July compared to March and April. This pattern holds true both 132
for patients in Utah (which has not yet experie nced a major COVID‐19 epidemic) and for others states in 133
aggregate (which include states with early major epidemics, recent major epidemics, and no major 134
epidemics to date).[6, 7] 135
If the median age of SARS‐CoV‐2 infection were truly declining over tim e, we would expect to 136
see decreases in the median age of COVID‐19 patients in both inpatients and outpatients. Instead, we 137
found that the median age of COVID‐19 patients showed opposite patterns over time between 138
inpatients and outpatients in Utah: the median age of inpati ents increased while the median age of 139
outpatients decreased. This finding suggests that the overall decreasing age of persons with detected 140
SARS‐CoV‐2 infection is driven by changes in SARS‐CoV‐2 testing rather than a true change in the 141
epidemiology of COVID‐19. Given the striking incr ease in risk of severe COVD‐19 with increasing age, our 142
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
7
findings suggest that the increasing availability of SARS‐CoV‐2 tests has increased testing of low‐acuity 143
patients or asymptomatic persons in ambulatory care settings, who tend to be younger than the more 144
severely ill patients. 145
A key limitation this study is that, although ARUP is a national re ference laboratory, a plurality of 146
specimens (and the only specimens with data on inpatient vs. outpatient providers) were available from 147
Utah. Although the aggregated non‐Utah specimens show similar age‐related trends to the Utah 148
specimens, results from this study may not generalize to any specific non‐Uta h state. 149
Understanding how SARS‐CoV‐2 infection varies across the age spectrum is key for developing 150
responses to the COVID‐19 epidemic. Our findings suggest that age‐related differences in infection from 151
the early epidemic until now are driven by changes in testing patterns rather than tr ue changes in the 152
epidemiology of SARS‐CoV‐2 infection. This calls for caution in interpretation of routine surveillance data 153
until testing patterns are stabilized with regard to illness acuity. 154
FUNDING 155
This work was supported by internal funds at Kaiser Permanente Washington (MLJ, DG). No 156
specific funding was provi
ded to (RLS, JCD, DRH). 157
158
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
8
References
159
160
1. Lewnard JA, Liu VX, Jackson ML, et al. Incidence, clinical outcomes, and transmission dynamics 161
of severe coronavirus disease 2019 in California and Washington: prospective cohort study. BMJ 162
2020; 369: m1923. 163
2. Docherty AB, Harrison EM, Green CA, et al. Features of 20 133 UK patients in hospital with 164
covid‐19 using th
e ISARIC WHO Clinical Characterisation Protocol: prospective observational 165
cohort study. BMJ 2020; 369: m1985. 166
3. Bhopal R. Covid‐19 worldwide: we need precise data by age group and sex urgently. BMJ 2020; 167
369: m1366. 168
4. Washington State Department of Heal th. "2019 Novel Coronavirus Outbreak (COVID‐19)". 2020. 169
https://www.doh.wa.gov/Emergencies/Coronavirus. Accessed 5 June 2020. 170
5. The COVID Tracking Project. Available at: https://covidtracking.com/. Accessed 16 July 2020. 171
6. Richardson S, Hirsch JS, Narasimhan M, et al. Presenting Characteristics, Comorbidities, and 172
Outcomes Among 5700 Patients Hospitalized With COVID‐19 in the New York City Area. JAMA 173
2020. 174
7. Myers LC, Parodi SM, Escobar GJ, Liu VX. Characteristics of Hospitalized Adults With COVID‐19 in 175
an Integrate
d Health Care System in California. JAMA 2020. 176
177
178
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
9
Table 1: Geographical Distribution of Results 179
State Test
Results
Percent
of Total Test
Results
Percent of
Total
Positive
Results
UT 134,253 48.36 9,327 6.9
FL 25,517 9.19 1,185 4.6
NY 14,316 5.16 370 2.6
WI 13,585 4.89 613 4.5
CA 11,465 4.13 1,235 10.8
TX 8,506 3.06 1,576 18.5
OH 7,447 2.68 345 4.6
IL 7,364 2.65 483 6.6
GA 5,821 2.1 686 11.8
AZ 5,766 2.08 556 9.6
PA 5,501 1.98 615 11.2
KY 3,891 1.4 95 2.4
MD 3,425 1.23 716 20.9
TN 3,284 1.18 49 1.5
WY 3,060 1.1 65 2.1
LA 3,039 1.09 310 10.2
180
181
182
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
10
Table 1 (continued) 183
OR 2,894 1.04 134 4.6
ID 2,876 1.04 172 6.0
CO 2,705 0.97 93 3.4
VA 2,484 0.89 40 1.6
SD 2,327 0.84 25 1.1
NM 1,820 0.66 38 2.1
NC 1,134 0.41 75 6.6
Other 5121 2.25 517 10.1
Total 277601 100.0 19320
100.0
184
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
11
Table 2: Summary Statistics for the Age Distribution for Positive Covid Cases. Data is for results 185
between March 10, 2020 and July 8, 2020. Early Period is March 10 to April 30. Late Period is June 1 to 186
July 8. Each cell contains the median age, the interquartile range, and the num ber of positive cases. All 187
changes in the median between the early and late periods were statistically significant (p<0.001). 188
Cohort Median Age of Positive Patients
(Interquartile range)
Early Period
(March‐April)
Late Period
(June – July)
All patients 40.8
(29.0 – 54.1)
3,263
35.8
(24.0 – 50.2)
12, 026
Utah 38.5
(27.1 – 50.8)
2,305
34.3
(23.1 – 46.3)
5,465
NonUtah 47.4
(34.6 – 60.2)
958
37.4
(24.9 – 53.7)
6,561
Utah Inpatient 50.0
(42.6 – 68.3)
39
55.8
(45.2 – 70.2)
47
Utah Outpatient 37.9
(26.9 – 50.6)
1,377
34.0
(23.0 ‐ 46.5)
4,196
189
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
12
Table 3: Change in Distribution of Infection by Age over Time. Data is for results between March 10, 190
2020 and July 8, 2020. Early Period is March 10 to April 30. Late Period is June 1 to July 8. 191
192
Age Group
(years)
Percent positivity
Absolute
positivity
Percent of To
tal
Positive Cases
Early Late Early Late Early Late
< 18 3.3 10.0 150 1,183 4.6 9.8
18 ‐ 29 6.1 11.5 731 3,438 22.4 28.6
30‐39 6.2 10.1 697 2,318 21.4 19.3
40 ‐ 49 7.0 10.0 661 2,035 20.3 16.9
50 ‐ 59 7.1 6.9 506 1,467 15.5 12.2
60 ‐ 69 5.5 4.4 289 848 8.9 7.1
70 and over 6.1 3.6 229 737 7.0 6.1
Total or Average 6.1 8.3 3,263 12,026 100.0 100.0
193
194
195
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
13
FIGURE LEGENDS 196
Figure 1: Comparison of Age Distribution of Positive Covid Cases. Late Period (June – July 8, 2020). 197
Solid Line = Early Period (March 10 – April 30, 2020). All p‐values (median age of early vs late period) are 198
below 0.001. 199
200
201
0 20 40 60 80 100
Age
Early Late
All Cases
0 20 40 60 80 100
Early Late
Utah Cases
0 20 40 60 80 100
Early Late
nonUtah Cases
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
14
Figure 2: Change in Age Distribution of Positive Utah Inpatients and Outpatients Over Time. Late 202
Period (June – July 8, 2020, N=3263 positive cases). Solid Line = Early Period (March 10 – April 30, 2020). 203
All p‐values (median age for early period vs late period) are belo w 0.001. 204
205
206
0 20 40 60 80 100
Age
Early Late
Inpatients
0 20 40 60 80 100
Early Late
Outpatients
. CC-BY-ND 4.0 International licenseIt is made available under a
is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review)
The copyright holder for this preprint this version posted July 24, 2020. ; https://doi.org/10.1101/2020.07.22.20160119doi: medRxiv preprint
Text is read by the "Ask this paper" AI Q&A widget below.
Extraction quality varies by source — PMC NXML preserves structure
cleanly, OA-HTML may include some navigation residue, and OA-PDF can
have broken hyphenation. The publisher copy
(via DOI)
is the canonical version.