Evolving virulence? Decreasing COVID-19 complications among Massachusetts healthcare workers: a cohort study

Limited research has investigated possible changes in SARS-CoV-2 virulence. The United States case-severity and case-fatality rates have decreased [1] but may be biased by improved clinical acumen and increased testing detecting more asymptomatic cases. An Italian study has demonstrated decreasing mortality among hospitalized COVID-19 patients over time, after accounting for covariates [2]. Therefore, we investigated the temporal distribution of COVID-19 complications among healthcare workers (HCWs) to determine if complication rates are decreasing.

Our cohort study examined HCWs from a Massachusetts healthcare system using a uniform COVID-19 protocol to screen/test employees since the initial outbreak in March 2020. Inclusion criteria were (1) COVID-19 triage via the occupational health ‘hotline’ and a documented positive SARS-CoV-2 viral assay between 9 March and 8 July 2020, or (2) any untriaged employee with a COVID-19 complication subsequently confirmed by the occupational health department. Since March, the ‘hotline’ phone-interviewed HCWs with any COVID-19-related concerns and referred HCWs for a nasopharyngeal SARS-CoV-2 RT-PCR assay when clinically indicated. Outside PCR results were confirmed. All cases (early and late) were followed regularly until 31 July 2020 for any complication, including ER (Emergency Room) visit, hospitalization, intubation/code, and/or death. HCWs with COVID-19 were followed by occupational health every 48 hours until they returned to work, but specific treatment was deferred to the HCWs’ treating physicians. Detailed data acquisition, de-identification and ethical statement have been published previously [3].

Because Massachusetts has used 15 April as the date to measure improvements from the states’ pandemic peak (Figure 1) [4], we defined the cases triaged up to 15 April 2020, as ‘early’, and those afterward as ‘late’. We also included as early, any untriaged death occurring within 18 days of 15 April 2020 based on the median time from infection onset to death [5,6]. Sensitivity analysis was done by using only triaged COVID-19 cases.

Figure 1.

Temporal distribution of incident confirmed COVID-19 cases from 5 March to 9 July 2020 in (A) the Study Population and (B) Massachusetts statewide

Bars indicate the absolute number of daily cases, and the red line denotes the 7-day average of new cases.

We used the t-test or Wilcoxon rank sum test, as appropriate, to compare continuous variables, and chi-square test or the Fisher’s exact test, as appropriate, for categorical variables. Logistic regression models were built for multivariate adjustment. The analyses were performed using R software (version 3.6.3), two-sided and with a P-value <0.05 considered statistically significant.

The temporal distribution of the 166 incident COVID-19 cases in the study population was similar to the statewide pattern [4] (Figure 1). There were no significant differences between the 98 early and 68 late cases regarding age, sex, race, residential community attack rates [4], and number of presenting symptoms. Cough was more frequently reported among early cases (66% vs. 50%, P = 0.04), while headache was more common among late cases (42% vs. 60%, P = 0.03). Thirteen cases developed any complication: 12 early cases and one late case (P = 0.02). Ten cases experienced hospitalization or death (9 early vs. 1 late, P < 0.05). Three cases underwent intubation or a code (all early cases) and two of these HCWs died (Table 1). After adjusting for age, sex, and race, early cases had increased odds ratios (ORs) of any complication(s) (10.3, 95%CI: 1.7–199.9), and hospitalization/death (6.2, 95%CI: 1.0–122.5).

Table 1.

Early and late COVID-19 cases’ clinical characteristics and complication rates

	Early (thru 15 April) (n = 98)	Late (16 April–9 July) (n = 68)	P-value
Age (n = 166)	43.1 (12.9)	41.5 (11.9)	0.440
Sex (n = 159)			0.944
Female	70 (71.4%)	46 (67.6%)
Male	25 (25.5%)	18 (26.5%)
Race (n = 156)			0.517
Non–Hispanic white	38 (38.8%)	18 (26.5%)
African American	33 (33.7%)	27 (39.7%)
Hispanic	18 (18.4%)	14 (20.6%)
Others	5 (5.1%)	3 (4.4%)
Cumulative community attack rate (per 100,000)* (as of 3 June) (n = 156)	1723.7 (1214.2–3393.3)	1802.8 (1634.5–2330.2)	0.883†
Total number of symptoms (n = 165)	4 (3-5)	4 (2-5)	0.529†
No symptom	4 (4.1%)	4 (5.9%)	0.718‡
Fever	50 (51.0%)	25 (36.8%)	0.086
Cough	65 (66.3%)	34 (50.0%)	0.042
Dyspnea	13 (13.3%)	12 (17.6%)	0.598
Myalgia	48 (49.0%)	34 (50.0%)	0.999
Malaise	51 (52.0%)	29 (42.6%)	0.272
Nasal symptoms	33 (33.7%)	28 (41.2%)	0.439
GI symptoms	26 (26.5%)	15 (22.1%)	0.609
Rash	3 (3.1%)	1 (1.5%)	0.644‡
Anosmia/Ageusia	14 (14.3%)	18 (26.5%)	0.085
Headache	41 (41.8%)	41 (60.3%)	0.034
Sore throat	41 (41.8%)	20 (29.4%)	0.129
Any complication(s) (n = 166)	12 (12.2%)	1 (1.5%)	0.016‡
ER visit	12 (12.2%)	1 (1.5%)	0.016‡
Hospitalization/death§	9 (9.2%)	1 (1.5%)	0.049‡
Intubation/coded§	3 (3.1%)	0	0.270‡
Death	2 (2.0%)	0	0.513‡

Mean (SD) for age. Median (Q1-Q3) for cumulative community attack rate and total number of symptoms. Count (%) for other variables.

Nasal symptoms include runny, sneezing, congestion, and sinus symptoms. GI (Gastrointestinal) symptoms include nausea/vomiting/diarrhea. ER, Emergency Room.

*Community rates based on Massachusetts data for each healthcare worker’s town of residence.

†Wilcoxon rank sum test.

‡Fisher’s exact test.

§Patient coded and died in an Emergency Department.

In the sensitivity analysis, we excluded one untriaged COVID-19 death from 29 April with an unknown date of symptom onset. The results remained robust regarding early versus late total complications: 11.3% (11/97) and 1.5% (1/68) (P = 0.02).

Our study provides additional epidemiologic support for the hypothesis of possible attenuated viral virulence [2], where cases infected later in the pandemic are on average less severe. In fact, attenuated SARS-CoV-2 variants and their related mutations have been reported within the first few months after the initial outbreaks [7]. We also found later cases reported more headaches, which may portend a better course [8]. The current study has several strengths: uniform screening protocols and testing criteria for all employees throughout the study period, which minimized bias from younger/paucisymptomatic cases as testing became more widespread. In fact, the age, sociodemographic characteristics and presenting symptoms of early and late cases were similar, and the results remained significant after adjusting for potential confounders. Second, the similar temporal epidemic pattern between the HCWs and the general population implies possible generalizability to other working populations. Third, although improving clinical management of hospitalized patients could confound complication rates after cases were hospitalized, we used the difference in proportion of ER visit/hospitalization versus solely outpatient management, which is independent of changing inpatient treatment protocols and the results remained robust. Nonetheless, the present study had limited sample size and the HCWs’ underlying medical conditions were unavailable. Finally, it is possible that better distancing and the increased use of masks may have led to lower viral inoculation/viral loads as the pandemic progressed, and therefore, less severe cases [9]. Additional investigations are warranted in larger populations with adjustment for underlying comorbidities, other factors predisposing to complications and clinical management.

Highlights

• The hypothesis of attenuating SARS-CoV-2 virulence has been raised.

• COVID-19 complications among healthcare workers are decreasing.

• The virulence of SARS-CoV-2 is likely evolving.

Disclosure statement

S.N.K. has received COVID-19-related consulting fees from Open Health. All other authors declare no competing interests.