Association between type of personal protective equipment worn and clinician self-reported COVID-19 infection

Clinical Implications.

• •Among clinicians responding to a survey regarding personal protective equipment (PPE) use, universal patient masking during asymptomatic encounters and higher efficiency PPE worn in symptomatic encounters were significantly associated with lower odds of clinicians self-reporting COVID-19 infection.

With more than 90,000 health care workers, including clinicians, around the world having been infected with SARS-CoV-2 virus,¹ further knowledge is needed about personal protective equipment (PPE) to reduce COVID-19-related morbidity and mortality. It remains unclear if all types of PPE are beneficial and what benefit eye protection and universal patient masking offer clinicians. The US Centers for Disease Control and Prevention (CDC) reports that PPE shortages currently pose a challenge to the nation's health care system. Clarifying PPE efficacy could help divert resources toward maximizing production of these types. The CDC recommends an n95 or higher respirator, face shield or goggles, gown, and gloves for confirmed/suspected COVID-19 cases. If unavailable, the CDC mentions that a facemask is an acceptable alternative.² The CDC allows reusable elastomeric particulate respirators as an alternative to National Institute for Occupational Safety and Health-approved n95 respirators, which are commonly used by workers outside the health care setting.³ In this small exploratory study, we attempt to determine if certain PPE types are associated with a risk of self-reported COVID-19 infection in clinicians.

To better quantify the association between PPE type and self-reported COVID-19 infection, an online survey of clinicians was created using Qualtrics. Nonrandomized convenience samples from self-identifying clinicians were obtained using social media posts on Facebook clinician groups, soliciting participation in a brief survey regarding PPE use and if they contracted COVID-19, collected from June 15, 2020, to July 9, 2020. Electronic informed consent was obtained. Items queried were PPE type worn during both asymptomatic and symptomatic patient encounters, type of eye protection worn, if their patients routinely wore a mask, and if the participants contracted COVID-19. Responses were all deidentified, and medical specialty and respondent location were not recorded, as such.

Data were analyzed for descriptive frequencies and for inferential associations using logistic regression using Stata 15 SE (College Station, Tex). An adjusted logistic regression was created to explore the surveyed factors predictive of the outcome of clinician self-reported COVID-19 infection, with exploration of predictive probabilities using the Stata margins command. Variance inflation factor was examined to assess multicollinearity of predictors. The study was approved as exempt from ongoing review by the Institutional Review Board at California State University, Sacramento.

Of the 801 respondents, 665 (83%) reported seeing patients with confirmed or suspected COVID-19. Among these respondents, 43 (5.4%) reported infection, including 3 of 136 (2.2%) who did not see symptomatic encounters. Among those seeing patients with confirmed or suspected COVID-19 (eg, symptomatic encounters), the following rates of self-reported COVID-19 infections by PPE type were noted: 7 of 69 (10.1%) who wore a surgical mask, 31 of 490 (6.3%) who wore a disposable n95-100, 2 of 68 (2.9%) who wore a reusable n95-100, and 0 of 38 (0%) who wore a powered air purifying respirator (PAPR). In asymptomatic patient encounters, approximately 95% of the respondents wore a surgical mask or disposable n95-100. Figure 1 , A, compares the rates of reported infections if surgical masks versus disposable n95-100 level respirators were worn during asymptomatic encounters, showing a nonsignificant trend (P = .08) of lower reported infection associated with the use of the n95-100 mask. Respondents were also asked if their patients wore a mask during asymptomatic encounters. The self-reported COVID-19 diagnosis history was 21 of 600 (3.5%) among clinicians reporting their patients wore a mask, and 22 of 194 (11.3%) among clinicians whose patient sometimes or do not wear a mask during asymptomatic encounters. Table E1 (available in this article's Online Repository at www.jaci-inpractice.org) details the type of reported eye protection worn along with the PPE type in both symptomatic and asymptomatic encounters. In an adjusted regression model (Table I ), there were 2 distinct findings. First, the odds of clinician-reported COVID-19 infection were significantly higher among those reporting their patients only sometimes or do not wear masks at all during asymptomatic encounters. Second, clinicians had significantly lower odds of having been diagnosed with COVID-19 with reusable PPE with a higher protective rating versus disposable surgical mask. There was no association between eye protection being worn and clinician history of having COVID-19. No significant interaction effects were noted among these variables. These relationships are shown in Figure 1, B and C.

Figure 1.

Predictive associations with PPE and self-reported COVID-19 infection. A, Relationship between reported infection rates for a surgical vs n95 level mask. B, Probability of reported infection if patients wear a mask. C, Relationship between type of clinician PPE worn and reported infection risk. Gray shading indicates the 95% confidence interval and the dot the point estimate. PPE, Personal protective equipment.

Table I.

Adjusted predictors of clinician self-reported COVID-19 infection

Clinician self-reported COVID-19 infection	Unadjusted odds ratio (95% CI, P value)	Adjusted odds ratio	Standard error	P value	95% CI
Symptomatic encounter model (ROC 0.68, P = .003)
Asymptomatic patients do not wear masks	3.5 (1.9-6.5, P < .001)	3.5	1.13	<.001	1.9-6.6
PPE worn in a symptomatic encounter (surgical mask reference)
Disposable respirator	0.51 (0.2-1.2, P = .12)	0.54	0.24	.17	0.2-1.3
Reusable respirator	0.17 (0.03-0.8, P = .03)	0.17	0.14	.03	0.03-0.9
Eye protection worn	0.78 (0.3-1.9, P = .59)	0.98	0.46	.96	0.4-2.4

All variables used in the model are shown. Disposable respirators included n95/100 intended for single use under pre-COVID-19 indications. Reusable respirators included elastomeric n95/100 and PAPR that were intended for reuse. Eyewear included any use of goggles, face shield, or PAPR hood.

CI, Confidence interval; PAPR, powered air purifying respirator; PPE, personal protective equipment; ROC, receiver operating characteristic.

There are many confounding risk factors during the pandemic, and it would be difficult to account for all of them. Our study was purely exploratory, to see if any crude associations are present, and to determine other potential moderators. Indeed, patient mask wearing during asymptomatic encounters and higher filtration rating on clinician PPE during symptomatic encounters were both significant predictors of the clinician self-reporting COVID-19 infection. These findings help substantiate that there may be a likelihood of a protective association for the clinician from both universal masking of all parties in an encounter and for optimizing the viral filtering capabilities of PPE, though these need further study.

Figure E1 (available in this article's Online Repository at www.jaci-inpractice.org) provides a picture of the types of PPE. Logically, as the PPE assigned protection factor (APF) increases, the likelihood of a viral particle exposure should decrease (Table I, Figure 1, A and C). Surgical masks are not respirators, though CDC recommendations regarding surgical masks are confusing and at times conflicting. This recommendation was intended to help preserve critical PPE supply and offer some protection in a difficult crisis.⁴ The CDC says that surgical masks can be used if respirators are not available,² but on another CDC webpage, it says: “surgical masks should never be used for protection against infectious aerosolized particles.”⁵ When considering the difference in APF relative to a surgical mask (APF of 1), disposable n95-100 respirators offer 10-fold, reusable n95-100 10- to 50-fold, and PAPRs 25- to 1000-fold (fit dependent) higher protection.⁶ Our trends reflect this protection gradient and suggest that the current approach of matching PPE APF is warranted. This trended very close to significance even for PPE choice in asymptomatic encounters (P = .08), likely due to limited power. There was no association between self-reported COVID-19 and eye protection in the logistic regression model.

Figure E1.

Types of personal protective equipment. A, PAPR. B, Standard surgical mask. C, Disposable n95 mask. D, Reusable half-face respirator, which can accommodate n or p rated 95/100 fiters. PAPR, Powered air purifying respirator.

The study has limitations in terms of use of a nonrandom convenience sample, unclear response rate (open posting), use of self-reported COVID-19 status, and did not attempt to stratify by specialty, location, care volume, or out of work exposure risk, given this was purely exploratory and we were limited in the number of items we could ask. This was not intended to be representative. The small sample size limited the power to look for fairly nuanced interactions between patients wearing masks and the type of PPE worn by the clinician. Also, within the sample, there was a low rate of self-reported infection, and it is unclear if this is because of the PPE worn, or other factors. However, the exploratory data suggest that there is value in pursuing a larger, more representative and powered sample to investigate these associations.

In conclusion, in this small exploratory study, both having patients always wear masks and higher PPE APF worn in symptomatic encounters were significantly associated with lower odds of clinician self-reporting COVID-19 infection. Because of the associations noted, and similar recent reports from a large health care system,⁷ a more nuanced study is indicated in a larger sample that is stratified by specialty, volume, and location. However, these findings do support a logical association that there are limited benefits and some defined risks to the use of surgical masks (even perhaps in asymptomatic encounters), which are not a form of respiratory protection, during a pandemic with a high probability of respiratory spread, which can help practices make more informed decisions.^{2 , 3}

Online Repository

Table E1.

Reported eye protection and PPE worn during symptomatic and asymptomatic encounters

	No mask	Surgical	Disposable respirator	Reusable respirator	PAPR	Total
Symptomatic encounters
No eyewear		13 (25.5%)	35 (68.6%)	2 (3.9%)	1 (1.9%)	51
Goggles		24 (11.5%)	151 (72.2%)	24 (11.5%)	10 (4.8%)	209
Face shield		28 (9.5%)	232 (78.9%)	22 (7.5%)	12 (4.1%)	294
Face shield/goggles		3 (3.4%)	68 (78.2%)	15 (17.2%)	1 (1.1%)	87
Full facemask		0	3 (13.6%)	5 (22.7%)	14 (63.6%)	22
Total		68 (10.3%)	489 (73.7%)	68 (10.2%)	38 (5.7%)	663
Asymptomatic encounters
No eyewear	1 (1.1%)	63 (69.2%)	26 (28.6%)	0	1 (1.1%)	91
Goggles	2 (0.8%)	161 (64.7%)	76 (30.5%)	10 (4%)	0	249
Face shield	1 (0.3%)	217 (65%)	107 (32%)	9 (2.7)	0	334
Face shield/goggles	0	48 (52.7%)	36 (39.6%)	7 (7.7%)	0	91
Full facemask	0	13 (59.1%)	4 (18.2%)	1 (4.5%)	4 (18.2%)	22
Total	4 (0.5%)	502 (63.8%)	249 (31.6%)	27 (3.4%)	5 (0.6%)	787

PAPR, Powered air purifying respirator; PPE, personal protective equipment.