Journal of Correctional Health Care - April 2023 - 147

PULSE OXIMETRY IMPROVES COVID SCREEN SENSITIVITY
147
Table 2. Results of Retrospectively Applied Screening Algorithms for COVID-19
Screening algorithm
TPa
Temp 100.4F
Temp 100.4F and symptoms
Temp 99.6F
Temp 99.6F and symptoms
Temp 99.6F, symptoms, and SpO2£94%
Temp 99.6F, symptoms, and SpO2£95%
Temp 99.1F
Temp 99.1F and symptoms
Temp 99.1F, symptoms, and SpO2£94%
Temp 99.1F, symptoms, and SpO2£ 95
14
22
20
26
37
48
36
38
45
50
TN
81
81
81
81
79
77
78
78
76
74
FP
2
4
3
3
5
7
FN
43
35
37
31
20
9
21
19
12
7
Sensitivity [95% CI]
24.6 [14.1-37.4]
38.6 [26-52.4]
35.1 [22.9-48.9]
45.6 [32.4-59.3]
64.9 [51.1-77.1]
84.2 [72.1-92.5]
63.2 [49.3-75.6]
66.7 [52.9-78.6]
79 [66.1-88.6]
87.7 [76.3-94.9]
aThe seventh day of illness was used as a cut-off for consideration of a positive screening for all algorithms.
CI, confidence interval; TP, true positive; TN, true negative; FP, false positive; FN, false negative.
exclusively on symptoms questionnaire and temperature
cut-offs of 99.6F or 100.4F yielded poor sensitivities,
although there were no false positives. An algorithm
that used symptoms and a more aggressive temp cut-off
of 99.1F had a much higher sensitivity with a very low
false positive rate. Using a temperature cut-off of
99.1F combined with SpO2 had the highest sensitivity
but with the trade-off of higher false positive rates.
There were seven (12%) individuals with COVID-19 infection
who would not have been screened as positive by
any algorithm within 7 days of their initial positive lab
test for COVID-19.
Discussion
Our study supports the finding that a large portion of people
with COVID-19 infections are asymptomatic and afebrile
throughout their disease and would be missed by
traditional screening algorithms that rely only on the presence
of symptoms and a temperature check with a cut-off
of either 99.6 or 100.4. Although both of these traditional
algorithms yielded a 0% false positive rate in this study,
the algorithms also missed 31 (54%) and 35 (61%) of infections
using cut-off measures of 99.6F and 100.4F, respectively.
The real-world application of these traditional
screening algorithms has little value given that these cutoffs
for temperature are unlikely to identify asymptomatic
infections as evidenced by our study where only four
(11%) of asymptomatic patients had a T-max ‡99.6F.
Furthermore, there was a high prevalence of fever
among symptomatic people and therefore nearly no difference
between the screening algorithms of fever alone
versus fever and symptoms. This is one of the reasons
why traditional screening algorithms miss so many people
with infections; if a person is symptomatic, then
they are likely not traveling or entering public places
where they are likely to be screened and instead are quarantining
at home.
This study also supports the finding of asymptomatic
hypoxemia in COVID-19 while going one step further
in showing that a large percentage of people with asymptomatic
infections have hypoxemia to at least a mild degree.
Our data also clearly show a subtle yet statistically
significant decrease in SpO2 and increase in temperature
in AC. Enhanced screening algorithms that use a combination
of symptoms, pulse oximetry, and temperature
with more sensitive thresholds significantly increase
screening sensitivity while also maintaining a low false
positive rate. By providing multiple screening algorithms,
one can select the screen that best suits one's
needs for balancing sensitivity and false positive rate.
In settings with a high-risk population when the prevalence
of COVID-19 in the community is high, or in a
confined population with a high risk for transmission, a
more sensitive screen is likely well worth the small
false positive rate.
Screening algorithms that use objective measures are
valuable in settings where the screened population is unable
or unwilling to report a symptom or a known exposure.
In one meta-analysis, the reported rate of asymptomatic infections
in four state prisons was 96% (So & Smith, 2020).
However, one critique of this study was the lack of followup
to determine how many individuals may have been
presymptomatic.
Regardless of this point, the fact remains that due to
complex psychosocial factors, it is reasonable to assume
that individuals are much less likely to report a symptom.
A primary benefit of not reporting a symptom is avoiding
placement in isolation due to an infection. In our study,
because infected individuals were followed daily throughout
their isolation, the secondary benefit of hiding symptoms
is lost because they are already placed in isolation.
The use of pulse oximetry with a cut-off of 94%-95%
will increase the rate of false positive screening in individuals
who are older, obese, or have chronic cardiopulmonary
comorbidities. However, this is the group known to
have a greater severity of illness and mortality from
COVID-19. Careful screening, thought, and consideration
of this group is warranted for precisely these reasons.
False positive rate (%)
2.5
4.9
3.7
3.7
6.2
8.6

Journal of Correctional Health Care - April 2023

Table of Contents for the Digital Edition of Journal of Correctional Health Care - April 2023