Research Article

A Convenience Sample Survey of Risk Factors Associated with a Positive Test for SARS-CoV-2 in Asymptomatic Students of a Large Mid-Western University

David Claborn*, David EA Johnson, Karen McKinnis, David A Hall

Department of Public Health and Sports Medicine, Missouri State University, Missouri, USA

*Corresponding author: David Claborn, Department of Public Health and Sports Medicine, Missouri State University, 901 South National Avenue, Springfield, MO 65897, Missouri, USA. Tel: +1417-836-8945; E-mail: davidclaborn@missouristate.edu

Citation: Claborn D, Johnson DEA, McKinnis K, Hall DA (2021) Convenience Sample Survey of Risk Factors Associated with a Positive Test for SARSCoV-2 in Asymptomatic Students of a Large Mid-Western University. J Glob Epidemiol Environ Health 2021: 01-06. doi: https://doi.org/10.29199/GEEH.101024

Received: 02 August, 2021; Accepted: 20 September, 2021; Published: 27 September, 2021

Abstract

A convenience sample was used to survey the student population for SARS-CoV-2 at a large state university in the mid-western United States. The survey was done on the Missouri State University campus in Springfield, Missouri, during the fall of 2020 and it sampled students who were not seeking treatment for symptoms of SARS-CoV-2. For this study, these students were considered “asymptomatic” because they were not seeking treatment, though some of them did actually report symptoms in an accompanying questionnaire. Infection status was determined using a PCR-based test of a saliva collection. The only symptoms that were associated with a positive infection status in this population were the loss of taste or smell and shortness of breath. Having experienced diarrhoea within the two weeks prior to the donation of the saliva sample was associated with reduced odds of a positive test. Odds of infection seemed to be reduced in housing arrangements that limited the number of people the subject was exposed to within the residence.

Keywords: SARS-CoV-2, symptoms, housing.

Introduction

The pandemic of severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) has disrupted the operation of colleges and universities across much of the world. In the United State of America (USA) alone, closure of campuses has affected about 26,000,000 students [1]. As these institutions of higher education planned for re-opening of the campuses, screening techniques evolved to help administrators and public health workers design safe re-population or return-to-campus programs. These techniques included a variety of screening tests and symptom-based evaluations. The need for and utility of these tests is evident for overt symptoms of SARS-CoV-2, but there is confusion regarding these techniques and their application to decision making with regard to the asymptomatic population as well as that of the pre-symptomatic and clinic-averse population. Specifically, the use of symptoms to screen the peripatetic, seemingly healthy population outside the clinical setting requires additional research to elucidate the reliability of symptoms as a predictor of infection status. Information on demographic and behavioral issues that may be associated with increased risk of disease is also incomplete, including the role of student housing. We describe here a surveillance study for SARS-CoV-2 in the student population of a large, mid-Western university (Missouri State University (MSU)). The study’s purpose was to identify symptoms associated with SARS-CoV-2 infection in a student population which has not sought clinical care. It was also designed to identify risk factors involved with different housing arrangements.

Background

Screening for infected persons is an important component of any SARS-CoV-2 prevention program. Infected persons can be isolated and treated to prevent further transmission. Just as important, contacts of the infected person can be identified and quarantined through a contact tracing effort. However, it is more difficult to identify asymptomatic or pre-symptomatic persons, as well as the role these people play in the spread of the virus. An asymptomatic person has been described as “someone who is carrying the virus but not showing symptoms such as fever or gastrointestinal or respiratory disorders and has no significant abnormalities on chest radiography [2].” This definition is obviously impractical in screening programs for students returning to large campuses because most students will not have a recent radiography examination. Nevertheless, the Centers for Disease Control and Prevention (CDC) has stated that testing can slow or even stop the spread of SARS-CoV-2 on campus if the institution of higher education conducts both diagnostic and screening tests of students, faculty, and staff [3]. This perspective acknowledges that symptom screening alone will fail to identify some people who have the virus. Using viral tests as a screening tool allows public health workers to identify infected people who are asymptomatic and who can be placed in isolation to impede the spread of the virus. Symptomatology of symptomatic patients in the clinic is much better defined than is that for the seemingly healthy population on campus. In fact, in many cases, the “asymptomatic” population is not actually without symptoms. In one study of healthcare workers in a hospital in England, about 3% of the workers tested positive for SARS-CoV-2 and nearly 40% had displayed at least one symptom consistent with clinical disease within the seven days prior to testing [4]. The authors of that hospital study and others stated that mass screening and isolation of asymptomatic individuals could be an effective way of preventing spread, perhaps by as much as 23%, if the results of testing are available within 24 hours of the initial specimen collection.

Student populations have some unique demographics that may also affect viral transmission. Student populations in the USA are typically young, disproportionately female, and relatively affluent. They often have communal living arrangements such as residence halls or dormitories. This latter risk factor is of particular interest given the potential for crowded living conditions to affect virus transmission.

The current study was performed to help define risk factors of a positive SARS-CoV-2 test in the student population of a large mid-Western public university. Because some of the students actually had symptoms for which they were not seeking treatment, they are referred to here as “asymptomatic” (in quotes).

Materials & Methods

Site of the study

Missouri State University (MSU) is a public university with an enrolment of 23,504 students in the Fall of 2020. Females constitute 60% of the student populations; 83.5% of the students are undergraduates. The student body is made up of 82.8% Missouri residents; about 80% are white. The largest minority population is reported as Hispanic (4.2%) followed by black/African American (3.8%). Approximately 2,000 students live on campus during most Fall and Spring semesters. The university offers approximately 100 undergraduate majors. It also has about 60 master’s level programs and seven doctoral degree programs.

MSU is located in Springfield, Missouri. The city had a population of about 168,000 in 2019 but the metropolitan area had about 462,369. It has three university campuses as well as a large community college. The city lies on a plateau within the Ozarks region of southwest Missouri and is in the northern limit of a humid subtropical climate. The economy is based on health care, manufacturing, retail, education, and tourism.

Collection of data

An initial attempt to randomly test the student population was unsuccessful due to lack of participation, so a convenience sample was used instead. The Office of Safety and Emergency Services set up collection tables outside major buildings on campus two to three times a week during the Fall 2020 and Spring 2021 semesters. The placement of the tables rotated between several buildings during the two semesters. Initially, the test that was used in targeted testing of high-risk groups was a PCR-based test that required a nasopharyngeal swab. This test was somewhat invasive and expensive, so by the time the convenience sample survey started in the first week of the Fall semester, 2020, another test was selected; that test was used for the entirety of the Fall and Spring semesters. Data presented here are all from the same test. The selected test was from Clinical Reference Laboratory (CRL) with laboratory facilities in Lenexa, KS. The PCR-based test required the collection of a standardized aliquot of saliva using an existing Canadianmade collection kit (Omnigene Oral). Saliva tests have shown a pooled sensitivity of 87% and a specificity of 98% [5]. The collection kit allowed the volunteer to register the specific sample through a personal cellular telephone so that results of the test were returned directly to the volunteer as well as to MSU. Collections were non-invasive and simple; with the only real limitation being that volunteers could not have eaten nor had anything to drink within 30 minutes prior to supplying the saliva sample. Volunteers with positive results were reported to the Springfield-Greene County Health Department as required by law. Results were also provided to the MSU COVID-19 Response Team for recording and analysis. The results were linked to a questionnaire on demographics, symptoms, and housing arrangements that the volunteer filled out at the time of the specimen collection. Written permission to use the results was also obtained on that questionnaire. A copy of the questionnaire is appended to this publication. Volunteers were allowed to submit samples as often as they wished during the survey, but only the most recent test for a specific individual was used for data analysis.

In addition to the collection of data from students who were not seeking treatment, that is the “asymptomatic” population, the number of confirmed cases in the university population was collected by the local health department and the university. Data on those cases discovered on campus were shared with the local health department. The data used on confirmed cases were taken from the MSU Office of Safety and Emergency Services data base which is available online [6]. For the analysis, MSU cases were subtracted from the health department data to avoid double counting those patients. The data for Greene County, which surrounds MSU, were collected from the Springfield-Greene County COVID dashboard’s daily reports which were also available online [7]. These were combined into weeks to compare with the data from MSU.

Statistical analysis

Data were examined to ensure that only one entry per individual was used for analysis. Also, data lines that were incomplete or did not have explicit permission to use the data were removed before analysis. Data on “asymptomatic” testing were analyzed using the LOGISTIC and GLM procedures of SAS [8]. For comparison of trends between the campus and the surrounding county population, a fourteenweek window from early September 2020 to mid-December 2020 was used for the fall semester. For the spring semester, the asymptomatic sample was too small to allow the same analytical procedure as the fall. Data on symptomatic cases from MSU and the surrounding counties were collected from online sources for each organization.

Results

In total, 1,178 “asymptomatic” samples were submitted for analysis. After removal of multiple tests for individuals, lines without complete information, and information for which no permission to use was granted, data for a total of 953 individuals were available. A total of 27 tested positive using the PCR-based test on donated saliva. Thus, about 2.8% of the “asymptomatic” population became positive at some time during the testing period.

 

Study Participants

Student Population  (Spring 2020)

Age

< 25

881 (92%)

17,897 (80.5%)

25-39

70 (7.3%)

3,289 (14.8)

40-59

2 (<1 %)

879 (4%)

59+

0 (0%)

156 (<1%)

Sex

Female

640 (67%)

13,229 (60%)

Male

314 (33%)

8,989 (40%)

Race/ethnicity1

Asian2

49 (5.1%)

469 (2.1%)

Black

74 (5.2%)

732 (3.2%)

Hispanic

35 (3.7%)

324 (3.7%)

Mixed

25 (2.6%)

780 (3.5%)

White

770 (80.7%)

17,721 (80.0%)

1The university demographics have a separate category for international students that were not used in the convenience sample. 

2Asian and Pacific Islander

Table 1: Demographics of participants in a convenience sample of university students given a PCR- based COVID-19 test in a non-clinical setting (Fall 2020 to Spring 2021).

Table 1 presents the demographics of the convenience sample population for both semesters of the study and the demographics of the student population for the entire university. The total student population for Spring 2020 is reported here because that reflects the population before the pandemic affected the size of the on-campus population. This comparison is provided as an estimate of how well the convenience sample represented the university population. The volunteers who participated in this study of the “asymptomatic” population did not submit their specimens in a clinical setting, but rather in an open setting, usually outside of a classroom building.

For the population of students, “sex” was not a predictor of disease in this study (p = 0.88) nor was “race”, which was dichotomized as “white” and “all others” (p = 0. 97). Also, age dichotomized at “26 and older” was not a predictor of positive infection status (p = 0.98).

Type of housing interval 

Odds Ratio compared to all other housing

95% Confidence

Residence Hall

0.80

0.35 – 8.81

Single family home

0.22

0.15 – 0.32

Apartment alone

0.65

0.43 – 0.98

Fraternity/sorority house

0.79

0.51 – 1.21

Apartment with one roommate

0.47

0.23 – 0.93

Table 2: Odds of testing positive for COVID-19 as associated with housing arrangement for students at a large state university (n = 954) (Fall 2020 – Spring, 2021).

Table 2 reports the odds ratio and 95% confidence interval for odds of infection associated with different categories of student housing, with each variable compared to all other types of housing. Housing types that limited the number of human interactions generally had lower odds ratios and seemed to have a protective effect. The types of housing with the lowest odds ratios included singlefamily homes, apartments in which the volunteer was the only occupant, and an apartment with only one roommate. Communal living with more individuals seemed to reflect higher odds of a positive test.

Symptom

Odds ratio

95% Confidence Interval

Loss of taste or smell

2.44

1.08 - 5.50

Protracted cough

1.07

0.36 - 3.19

Diarrhea

0.18

0.53 - 0.61

Fatigue

1.11

0.24 - 5.03

Headache

1.00

0.48 - 2.10

Nasal congestion

1.12

0.63 - 2.00

Sneezing

0.95

0.62 - 1.47

Shortness of breath

1.68

1.07 - 2.63

Table 3: Odds of testing positive for COVID-19 as associated with self-reported symptoms by volunteers in a convenience sample of university students on a large mid-western state university. (n = 954) (Fall 2020-Spring, 2021).

Table 3 reports the odds ratio for a positive test by those reporting a specific symptom compared to all others. Over 20% of the volunteers reported at least one symptom even though they were walking around campus and probably attending classes. Most of the symptoms addressed in the questionnaire were not associated with increased or decreased odds of a positive test; however, “loss of taste and smell” and “shortness of breath” were associated with increased odds of a positive test as indicated by a confidence interval that did not include the value of 1. Odds of a positive test as estimated by the odds ratio in those with the loss of one or both senses were about 2.44 times higher than in those who did not display that symptom. Surprisingly, the odds of a positive test for the virus were lower in those who reported the symptom of diarrhea within the seven days prior to donation of the sample. There was about an 80% reduction in odds of a positive test in those reporting this symptom.

The data from Greene County cases, MSU symptomatic cases diagnosed clinically, and MSU’s “asymptomatic” survey cases were examined by semester for correlations to examine any potential linkages between the three populations. Figure 1 shows a graph of the Greene County cases and the MSU “asymptomatic” cases for the fall 2020 semester. Analysis revealed a small, statistically insignificant Pearson correlation of 0.257 (p=0.376) between these two populations.

      

Figure 1: Number of on-campus “asymptomatic cases” compared to the number of clinically diagnosed Greene County Cases (GC cases divided by 50). Week one started on September 9, 2020.

When the Fall 2020 Greene County cases were compared to the MSU symptomatic cases, figure 2, a slightly negative Pearson correlation of -0.461 (p=0.097) was found. This may be an indication that the mitigative measures taken on campus were effective for that time period. When comparing the MSU symptomatic cases and Greene County cases the following semester, Spring 2021 (January-May), there was a strong Pearson correlation of 0.890 (p<.001) between the number of campus and community cases as both decreased. This trend is shown in figure 3. This close correlation may demonstrate the impact of vaccination on both populations. The vaccination rate in Greene county increased during this period from 1.82% in January to 34.56% on May 9, 2021. An increase in the number of cases for MSU students in week 11 was not reflected in the county population. Several students chose to be tested prior to going home for the Thanksgiving holidays and this may explain the difference.

              

Figure 2: MSU Symptomatic Cases and Confirmed Greene County Cases (GC cases divided by 3). Week 1 started on September 9, 2020.

      

Figure 3: Spring 2021 MSU Symptomatic Cases and Cleaned Confirmed Greene County Cases/10.

Discussion and Conclusion

In general, the trends of infection status over time were similar between the campus and the non-campus populations. The overall positivity rate of 2.9% in “asymptomatic” university students was consistent with at least one other study done at the same time [9]. In that study at two universities in Wisconsin, which was more limited in time over two weeks, 2.4% of participants tested positive with an antigen test and 2.0% tested positive with a RT-PCR test. The positivity rate reported for the MSU study covered a much longer period of time, from September 2020 to May 2021. The 2.8% positivity rate does not include the number of symptomatic persons who sought treatment in the campus clinic or elsewhere.

This study failed to identify increased odds of disease in demographic groups that have been identified at higher risk in other studies. For instance, one other study reported that African Americans and males were at higher risk of infection [10]. This finding was not confirmed in the study reported here, though it is important to note that the population of college students is very different from the broader population of adults in the country or world, being younger and probably more affluent in general. Also, the odds ratios reported here are for the population of students attending classes and not seeking treatment at the time of the specimen collection. In the MSU study, race was analyzed as a dichotomous variable because there were so few positive cases in the “asymptomatic” population and a relatively small population of non-white participants. Being white was neither protective nor did it increase the risk of infection. Similarly, sex was analyzed as a dichotomous variable and did not show any association with increased or decreased risk. It should be noted that the female population was probably oversampled slightly in the convenience survey.

The finding that loss of taste or smell is predictive of positive disease status is consistent with other studies [11]. The same is true of shortness of breath. However, the negative association between diarrhoea and a positive test for the virus is not consistent with other findings. Diarrhoea has been reported as a less common symptom positively associated with SARS-CoV-2 [12]. At present we do not have any hypothesis to explain this finding and it may be an anomaly. Also, students were requested to report their symptoms for a period of time of several days prior to collection of the saliva specimen, and this length of time may not have been appropriate for all symptoms.

A trend that is obvious in the analysis of housing arrangements with risk of disease is the reduced risk associated with a lower number of people living in the same room or house. Early in the epidemic, the administration at MSU emphasized the use of single occupancy rooms in the residence halls whenever possible. This study validates the efficacy of such efforts in the reduction of the spread of disease, suggesting that reduced exposure to roommates or housemates is associated with reduced risk of disease.

There are some weaknesses to this study, foremost of which is the use of a convenience sample. For the most part, the convenience sample resulted in a sample population that reflected the population on the campus, though females and some minorities were slightly oversampled. One group that was under-sampled was the population of non-traditional students over the age of 40. These students often attend classes at night or exclusively online and would be unlikely to be on campus during the day when saliva samples were collected from participants. Another weakness to the study is the low number of positives. Odds ratios, as used in this study, are appropriate for analysis of relative rare events, but low numbers may make some findings tentative.

Another weakness to this study is that it was done relatively early in the pandemic and the findings reported here may not reflect what would be found with later variants of the virus.

References

  1. Van Pelt A, Glick HA, Yang W, Rubin D, Feldman M, et al. (2021) Evaluation of COVID-19 testing strategies for repopulating college and university campuses: A decision tree analysis. J Adolesc Health 68: 28-34.
  2. Han D, Li R, Han Y, Zhang R, Li J (2020) COVID-19: Insight into the asymptomatic SARS-CoV-2 infection and transmission. Int J Bio Sci 16: 2803-2811.
  3. Rivett L, Sridhar S, Sparkes D, Routledge M, Jones NK, et al. (2020) Screening of healthcare workers for SARS-CoV-2 highlights the role of asymptomatic carriage in COVID-19 transmission. eLife 9: e58728.
  4. Imperial College COVID-19 Response Team. Report 16: Role of testing in COVID-19 control.
  5. Warsi I, Khurshid Z, Shazam H, Under MF, Imran E, et al. (2021) Saliva exhibits high sensitivity and specificity for the detection of SAR-CoV-2. Diseases 9: 38.
  6. Missouri State University COVID Dashboard.
  7. Springfield-Greene County COVID-19 Dashboard.
  8. SAS Institute Inc. (2004) SAS/ETS® 9.1 User’s Guide. Cary, NC: S Istituten Inc.
  9. Pray IW, Ford L, Cole C, Lee C, Bigouett JP (2021) Performance of an antigen-based test for asymptomatic and symptomatic SARS-CoV-2 testing at two university campuses-Wisconsin, September-October 2020. MMWR 69: 1542-1647.
  10. Jehi L, Xinge J, Milinovich A, Erzurum S, Rubin B, et al. (2020) Individualizing risk prediction for positive coronavirus disease 2019 testing: results from 11,672 patients. Chest 158: 1364-1375.
  11. Gautier J, Ravussin Y (2020) A new symptom of COVID-19: Loss of taste and smell. Obesity 28: 848.
  12. Chen N, Zhou M, Dong X, Qu J, Gong F, et al. (2020) Epidemiological and clinical characteristics of 99 cases of 2019 novel coronavirus pneumonia in Wuhan, China: a descriptive study. Lancet 295: 507-513.