Abstract
This diagnostic study describes a dog screening program used to identify COVID-19 infections among schoolchildren.
The California Department of Public Health sponsors a statewide, school-based COVID-19 antigen testing program. Although effective, this program requires personnel, testing resources, and sample collection and generates medical waste. Scent-trained dogs are a strategy for rapid, noninvasive, low-cost, and environmentally responsible COVID-19 screening. We conducted a dog screening program to complement a school antigen testing program.
Methods
We partnered with Early Alert Canines to train 2 medical alert dogs (eFigure 1 in Supplement 1) to identify volatile organic compounds (VOCs) emitted by people with COVID-19. Dog training was similar to that in other studies.1,2,3 This diagnostic study was approved as public health surveillance by California’s State Committee for the Protection of Human Subjects. Electronic informed consent for testing was obtained from participants or guardians. Early Alert Canine is accredited by Assistance Dogs International to ensure ethical oversight of the dogs. We followed the STARD reporting guideline.
In-person dog screening was piloted in a subset of volunteer schools on days when antigen testing was scheduled. Participants were 6 ft apart, and the dogs, led by handlers, sniffed participants’ ankles and feet (eFigure 2 in Supplement 1). Dogs alerted handlers to potential COVID-19 infection by sitting. To protect confidentiality, participants faced away from the dogs. Participants then underwent BinaxNOW (Abbott) antigen testing. Dog and antigen results were recorded in a digital platform (Primary.Health; Primary Diagnostics).
We assessed dogs’ sensitivity and specificity for COVID-19 detection using antigen test results as the comparator. Antigen tests were already deployed in participating schools as their results correlate best with active infection.4 If a dog signaled positive and antigen testing results were negative, the signal was considered falsely positive; if a dog did not signal and antigen testing results were positive, the signal was considered falsely negative. Analyses were completed in SAS, version 9.4 (SAS Institute, Inc).
Results
After 2 months of training on COVID-19 scent samples in the laboratory, the dogs achieved greater than 95% sensitivity and specificity for detection of the virus. Dog screening was then piloted in the field; 50 visits were conducted at 27 schools from April 1 to May 25, 2022. Of 1558 participants (median [IQR] age, 13 [9-17] years; 870 females [55.8%], 670 males [43.0%], and 18 nonbinary, transgender, or undisclosed gender [1.2%]), most (89%) were students and many (68%) were screened at least twice (Table 1). Overall, 3897 paired antigen-dog screenings were completed. The dogs accurately signaled 85 infections and ruled out 3411 infections (overall accuracy, 90%). However, they inaccurately signaled infection in 383 instances and missed 18 infections, resulting in sensitivity of 83% (95% CI, 75%-90%) and specificity of 90% (95% CI, 89%-91%) (Table 2).
Table 1. Demographic and Other Characteristics of Participants in the California Department of Public Health K-12 School Dog COVID-19 Detection Program.
| Characteristic | No. (%) | ||
|---|---|---|---|
| Students (n = 1383) | Staff (n = 145) | Community members (n = 30) | |
| Age, y | |||
| 5-10 (Elementary school) | 565 (41) | 0 | 0 |
| 11-14 (Middle school) | 308 (22) | 0 | 0 |
| 15-18 (High school) | 510 (37) | 0 | 1 (3) |
| 19-49 | 0 | 96 (66) | 26 (87) |
| 50-64 | 0 | 47 (32) | 2 (7) |
| ≥65 | 0 | 2 (1) | 1 (3) |
| Gender | |||
| Female | 738 (53) | 117 (81) | 15 (50) |
| Male | 629 (46) | 27 (19) | 14 (47) |
| Nonbinary | 4 (0) | 0 | 0 |
| Transgender male | 1 (0) | 0 | 0 |
| Not disclosed | 11 (1.0) | 1 (1) | 1 (3) |
| Race/ethnicitya | |||
| American Indian or Alaska Native | 6 (0) | 0 | 0 |
| Asian | 381 (28) | 18 (12) | 8 (27) |
| Black | 28 (2.0) | 4 (3) | 1 (3) |
| Hispanic, all races | 205 (15) | 23 (16) | 6 (20) |
| Multiracial, non-Hispanic | 119 (9) | 6 (4) | 1 (3) |
| Native Hawaiian or Other Pacific Islander | 4 (0) | 0 | 0 |
| White | 471 (34) | 68 (47) | 12 (40) |
| Otherb | 77 (6) | 1 (1) | 0 |
| Not disclosed | 92 (7) | 25 (17) | 2 (7) |
| School type | |||
| Private | 455 (33) | 16 (11.) | 0 |
| Public | 928 (67) | 129 (89) | 30 (100) |
| No. of dog screenings | |||
| 1 | 450 (33) | 55 (38) | 18 (60) |
| 2 | 565 (41) | 64 (44) | 12 (40) |
| 3 | 53 (4) | 11 (8) | 0 |
| ≥4 | 315 (23) | 15 (10) | 0 |
Self-reported by participants.
No additional details were provided.
Table 2. Dog Performance for Detecting COVID-19–Infected Individuals in K-12 Schools.
| Dogs’ screening result | Rapid antigen test result | Total | |
|---|---|---|---|
| Positive | Negative | ||
| Positive | 85 | 383 | 468 |
| Negative | 18 | 3411 | 3429 |
| Total | 103 | 3794 | 3897 |
| Sensitivity, % (95% CI) | 83 (75-90) | NA | NA |
| Specificity, % (95% CI) | 90 (89-91) | NA | NA |
Abbreviation: NA, not applicable.
Discussion
Studies have demonstrated dogs’ impressive ability for detecting VOCs associated with COVID-19 infection using specimens collected from SARS-CoV-2–infected and uninfected individuals.2,3,5,6 After training in the laboratory, our dogs were field tested and, in more than 3500 screenings, correctly determined COVID-19 status in most instances. Unlike most other studies,2,3,5,6 our dogs directly screened people in the field, rather than specimens. Our method was associated with improved testing efficiency but had a modest decrease in sensitivity and specificity compared with laboratory results.
Dog screening for COVID-19 infection can be completed in a matter of seconds. However, dog screening directly on individuals introduced variables, such as distractions (eg, noises, young children) and environmental factors (eg, wind, smells), that likely contributed to decreased sensitivity and specificity. We considered other options, including a sample collection strategy used by other investigators2,3,5,6; however, those options would sacrifice cost and time efficiency. Study limitations included the low prevalence of SARS-CoV-2 during the study period and the consequently low number of COVID-19 infections.
The goal is for dogs to perform large-scale VOC screening with antigen testing being performed only on persons with positive dog screening results, thereby reducing antigen tests performed by approximately 85%. While modifications are needed before widespread implementation, this study supports use of dogs for efficient and noninvasive COVID-19 screening and could be used for other pathogens.
eFigure 1. Dogs used in the COVID-19 Dog Screening Program
eFigure 2. COVID-19 Dog Screening at a School Site
Data Sharing Statement.
References
- 1.Edwards TL, Browne CM, Schoon A, Cox C, Poling A. Animal olfactory detection of human diseases: guidelines and systematic review. J Vet Behav. 2017;20:59-73. doi: 10.1016/j.jveb.2017.05.002 [DOI] [Google Scholar]
- 2.Guest C, Dewhirst SY, Lindsay SW, et al. ; COVID Dogs Research Team . Using trained dogs and organic semi-conducting sensors to identify asymptomatic and mild SARS-CoV-2 infections: an observational study. J Travel Med. 2022;29(3):taac043. doi: 10.1093/jtm/taac043 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Maurer M, Seto T, Guest C, Somal A, Julian C. Detection of SARS-CoV-2 by canine olfaction: a pilot study. Open Forum Infect Dis. 2022;9(7):ofac226. doi: 10.1093/ofid/ofac226 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Petersen I, Crozier A, Buchan I, Mina MJ, Bartlett JW. Recalibrating SARS-CoV-2 antigen rapid lateral flow test relative sensitivity from validation studies to absolute sensitivity for indicating individuals shedding transmissible virus. Clin Epidemiol. 2021;13:935-940. doi: 10.2147/CLEP.S311977 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Jendrny P, Twele F, Meller S, et al. Scent dog identification of SARS-CoV-2 infections in different body fluids. BMC Infect Dis. 2021;21(1):707. doi: 10.1186/s12879-021-06411-1 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Ten Hagen NA, Twele F, Meller S, et al. Canine real-time detection of SARS-CoV-2 infections in the context of a mass screening event. BMJ Glob Health. 2022;7(11):e010276. doi: 10.1136/bmjgh-2022-010276 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eFigure 1. Dogs used in the COVID-19 Dog Screening Program
eFigure 2. COVID-19 Dog Screening at a School Site
Data Sharing Statement.