From February to March 2020, we enrolled 2200 HCP, based on convenience sampling, in Thai hospitals: Bamrasnaradura Infectious Diseases Institute (BIDI, 300 beds referral hospital specializing in infectious diseases, total 819 HCP); Phaholpolpayuhasena hospital (PH, 540 beds tertiary hospital, total 1400 HCP); Phramongkutklao hospital (PMK, 1200 beds super tertiary and academic hospital, total 1428 HCP); and Rayong hospital (RY, 567 beds tertiary hospital, total 1600 HCP) for 1-year follow-up. We defined HCP as individuals providing services in a healthcare setting, and enrolled those aged ≥18 years, who had contact with ≥1 patient/day and worked ≥30 h/week. HCP employed <1 year, anticipated leaving the jobs in 1 year, or who did not have a telephone number for study contact were ineligible for enrollment. For recruitment, posters detailing briefed study information (including enrollment criteria, enrollment venues, and times) were posted in areas frequented by the HCP. Interested HCP were free to come forward for screening and enrolling during the scheduled dates; appointments were not needed. At enrollment, study staff interviewed participants using a structured questionnaire for work and health history, gross income information, frequency of appropriate personal protective equipment (PPE) use (defined as donning the PPE suitable for the procedure), and seasonal influenza vaccination history. Participants self-graded their risk from 0 (no risk) to 10 (highest possible risk) in different scenarios: (i) becoming infected with any respiratory virus; (ii) developing an acute respiratory illness (ARI); (iii) developing ARI and having to take leave from work; (iv) developing severe ARI; and (v) transmitting respiratory viruses to patients. Study staff taught participants how to self-collect midturbinate nasal specimens for laboratory testing (Suntarattiwong et al., 2021).

During the COVID-19 pandemic, non-pharmaceutical and pharmaceutical measures were implemented across all hospitals countrywide (e.g. triage of patients with ARI, installation of physical barriers or partitions between patients and the HCP, encouraging the use of face mask when caring for all patients and the use of respirator and other PPE when caring for patients suspected/confirmed of SARS-CoV-2 infection, isolation of SARS-CoV-2-infected HCP and quarantine of HCP with potential exposure to SARS-CoV-2 and COVID-19 vaccination among HCP). Furthermore, the HCP from non-critical departments (e.g. dental, ophthalmology) were transferred to the newly established COVID-19 wards or ARI clinics to meet the significant demand in services.

From July 2020 to June 2021 (Thailand’s expected 2020 influenza season), study staff called/texted participants weekly to ascertain an ILS (defined as new onset or sudden worsening of ≥1 of: muscle pain, cough, runny nose/nasal congestion, sore throat, and difficulty breathing) during the preceding 7 days. Participants were asked to contact the study staff if an ILS developed between weekly contacts. Symptomatic participants self-collected nasal specimens using a free of charge prelabeled study-provided kit (sterile polyester-tipped swab [MacroPur Swab P® regular size with plastic applicator, AMD Ritmed, China] and 3-ml cryovial tube of Universal Transport Medium [COPAN ITALIA SpA, Italy]), and kept them in a refrigerator (4–8°C). Specimens were transported to Armed Forces Research Institute of Medical Sciences for respiratory virus real-time reverse transcription-polymerase chain reaction (rRT-PCR) testing free of charge to participants.

Contact with ill participants was ceased upon participants’ reporting of an ILS and re-started at 7 days post-illness onsets to determine illness resolution. Participants returned to weekly contact at illness resolution defined as 48 h symptom-free of all five symptoms, or ≥13 days after ILS onset, whichever came first. Participants missing four consecutive weekly contacts were considered lost to follow-up. A re-enrollment was possible if a contact to study staff was made by the lost participants.

Study staff called ill participants every other day starting at 7 days after symptom onsets until they were no longer ill (or ≥13 days after illness onsets, whichever came first) for an illness interview. The interview was conducted using a structured questionnaire for illness management information and resolution date; all out-of-pocket and subsidized expenses associated with ILS (laboratory and examination tests, clinician fee, hospital fee, prescribed and over-the-counter medication fees, and transportation fee); monetary income loss due to ILS; and number of hour(s) participants and/or family members (who took care of the participants while ill) were absent from work, as well as family members’ income information.

All participants were compensated US$16 for the enrollment interview and US$0.3 for each successful weekly contact, but not for nasal specimen collection or illness interviews.

The quality of collected specimens was assessed by testing for human Ribonuclease P (RNase P) prior to laboratory testing. Influenza virus detection was performed using the U.S. Centers for Disease Control and Prevention’s (U.S. CDC’s) rRT-PCR protocol (WHO Collaborating Centre for Influenza, 2009). When additional funding became available, RSV detection was performed on specimens with remaining aliquots using commercial rRT-PCR assay (QIAstat-Dx Respiratory 2019-nCoV Panel, Qiagen, Germany). All participants had access to free-of-charge SARS-CoV-2 testing by rRT-PCR at their hospitals and were encouraged to share positive results with the study team.

We included only participants with ≥1 weekly contact for analysis. We used the Chi-square tests (categorical variables) or Student’s T-tests (log-transformed continuous variables) as appropriate to compare characteristics of (i) ILS cases (individuals meeting the ILS case definition) and non-cases (individuals who did not experience any ILS while in study) and (ii) ILS cases with and without hospitalization.

To calculate illness-associated expenses using the societal perspective, we summed all reported fees and monetary income loss (if any). We accounted for time lost due to missed work by multiplying the number of hours absent from work by self-reported hourly wages and added this calculated time lost to the overall expenses. Exchange rate used was 31 Thai Baht to US$1 (XE, 2022).

To calculate ILS incidence, the outcome was time to ILS onset (the number of days from 1 July 2020 to illness onset). The follow-up time of participants who did not experience any ILS ended on 30 June 2021. Participants who dropped out (n = 102; 4.7%) contributed follow-up time until their last successful contacts. Re-enrolled participants (n = 29; 1.3%) contributed a full follow-up time with the time not under observation deducted. We fitted population-averaged Poisson regression models (Neuhaus et al., 1991; Neuhaus, 1992) for the number of ILS events with person-time entered into the models as an offset. Adjusted ILS incidence per 100 person-years was calculated from Poisson parameters using least-squares means methods (Cameron and Trivedi, 2013). The 95% Poisson confidence intervals (95% CIs) were calculated using the exact method.

We calculated the economic impact due to ILS by applying number of days missed work, illness-associated expenses, etc. to Thailand’s HCP population (227 349 persons) (Strategy and Planning Division, 2020) and reported annual burden adjusted for sex, medical condition, frequency of PPE use, study site, and calendar month. We used a two-sided P-value of < 0.05 to indicate statistical significance and performed analysis using Stata software version 16.0 (StataCorp LP, College Station, TX, USA).

This study was approved by the Institutional Review Boards (IRBs) of: BIDI; PH; PMK; RY; Department of Disease Control of the Thai Ministry of Public Health (Thailand); and Walter Reed Army Institute of Research (USA). The U.S. CDC’s IRB (USA) issued a reliance on the determination of PMK. All participants had provided written informed consent to participate.

From February to March 2020, we enrolled 2200 participants (BIDI = 306, PH = 548, PMK = 900, and RY = 446). Of these, 2184 (99.3%) had ≥1 weekly contact with a median weekly response rate to symptom surveillance of 99.5% (interquartile range [IQR], 99.4–99.7%), and 2005 (91.8%) completed a 1-year follow-up. The median enrollment age was 38 years (IQR 29–47.5); most were female (1906; 87.3%; Table 1). All were salaried employees with paid sick leave (30 days) and covered by a health insurance provided by the government for both in- and out-patient expenditures. Nurses comprised the majority of the participants (1277; 58.5%), followed by administrative staff (572; 26%), and other professions involving patient contact including seven physicians (335; 15.3%). Three hundred and ninety-seven participants (18%) self-reported that they had ≥1 underlying medical condition (Supplementary Table 1, available at Annals of Work Exposures and Health online).

ILS cases and non-cases were similar at baseline with regards to age; sex; profession; career and service lengths; work location and associated department; numbers of working hours, patients encountered, and times handling potentially infectious materials; and self-reported medical condition (Table 1). However, non-cases were statistically more likely than cases to report always using PPE (P-value = 0.042), to receive annual seasonal influenza vaccination in the past 5 years (P-value = 0.004), to not experience an ARI in the past 12 months prior to study enrollment (P-value = 0.015), and to work at hospitals other than PMK (P-value < 0.001). Using a 10-point scale, perceived risks in various scenarios were comparable between ILS cases and non-cases except perceived risk of becoming infected with any respiratory virus which was higher among cases (median 7 points; IQR 5–8) than non-cases (median 6 points; IQR 5–8; P-value = 0.022).

Most participants (1911; 87.5%) self-reported a history of ever receiving seasonal influenza vaccination, 199 (9.1%) self-reported no history of vaccination, and 74 (3.4%) were unsure of their history. Among those with seasonal influenza vaccination history, 777 (40.7%) received an influenza vaccine annually during the past 5 years prior to study enrollment (Table 1).

Of 242 ILS episodes identified during the course of this study, 232 (95.9%) had nasal specimens collected. The median time from ILS onset to specimen collection was 3 days (IQR 2–4). All collected specimens were of good quality as indicated by RNaseP, but none tested positive for influenza viruses. Three (2.3%; 95% CI 0.4%–6.7%) of 128 specimens with remaining aliquots tested positive for RSV. Seven participants (0.3%; 95% CI 0.1%–0.6%) self-reported testing positive for SARS-CoV-2.

One hundred and ninety-nine participants (9.1%) reported 242 ILS episodes; 167 (83.9%) reported one, 24 (12.1%) reported two, and 8 (4.0%) reported ≥3 episodes. Of 242 episodes, 86 (35.5%) were associated with two symptoms. Of the five symptoms constituting the ILS case definition, runny nose/nasal congestion (174; 71.9%) was the most commonly reported symptom, followed by sore throat (163; 67.4%) and cough (138; 57.0%; Supplementary Table 2, available at Annals of Work Exposures and Health online). Additionally, 82 ILS episodes (33.9%) presented with fever, 40 (16.5%) with headache, and 29 (12.0%) with sputum production. Most ILS episodes (233; 96.3%) did not require hospitalization, and 200 (82.6%) resolved before illness interviews. Among 200 ILS with known resolution dates, illness durations were not statistically different between ILS episodes with and without hospitalization (median 5.5 days [IQR 4–7] versus median 5 days [IQR 3–6]; P-value = 0.407).

Among the nine hospitalized ILS cases, one tested negative for influenza, but positive for RSV, three tested negative for both influenza and RSV, and five self-reported SARS-CoV-2 infection. The median hospitalized duration was 7 days (IQR 3.5–12). All had missed work (median 7 days, IQR 5–9) and incurred illness-associated expenses (median US$271 [58.5% of median monthly income], IQR US$148–US$402). The highest proportion of the expenses was healthcare-related cost (61%), followed by indirect cost of self and family’s lost time (39%). There was no non-healthcare-related cost (e.g. travel expense) incurred. Seven (77.8%) reported presenting at work while ill (median 3 days, IQR 2–6).

Among the 233 non-hospitalized ILS cases, 43 (18.5%) had missed work (median 2 days, IQR 1–2) and 118 (51.0%) incurred illness-associated expenses (median US$14 [3.0% of median monthly income], IQR US$5–US$42). The highest proportion of the expenses was indirect cost of self and family’s lost time (60%), followed by healthcare-related cost (40%), and non-healthcare-related cost, e.g. travel expense (<1%). Two hundred and twenty-five (96.5%) reported presenting at work while ill (median 6 days, IQR 4–9). Numbers of days presenting at work while ill were not significantly different between ILS episodes with and without hospitalization (median 3 days [IQR 2–6] versus median 6 days [IQR 4–9]; P-value = 0.779); however, numbers of days of missed work (median 7 days [IQR 5–9] versus median 2 days [IQR 1–2]; P-value < 0.001), and illness-associated expenses (median US$271 [IQR US$148–US$402] versus median US$14 [IQR US$5–US$42]; P-value < 0.001) were greater among hospitalized ILS cases.

Factors independently associated with acquiring ≥1 ILS episodes included female sex, self-reported asthma as a medical condition, lower frequency of PPE use, study site, and calendar month (Table 2). Controlling for covariates, females were 2.1 times more likely than males to acquire ILS (95% CI 1.1–4.0). Those with asthma were 2.4 times more likely than those without the condition to acquire ILS (95% CI 1.1–5.2). Those who used PPE frequently but not always were 1.6 times more likely than those who always used PPE to acquire ILS (95% CI 1.2–2.3). Compared to BIDI, ILS episodes were 1.6 times more likely to be reported from PMK participants (95% CI 1.0–2.6) and 0.3 times less likely to be reported from RY participants (95% CI 0.2–0.7). Lastly, ILS episodes were 0.14 times less likely to be reported during month of June compared to January (95% CI 0.1–0.9).

Study participants contributed 2142 person-years of observation time during Thailand’s expected 2020 influenza season. The adjusted all-cause ILS incidence was 6.1 episodes per 100 person-years (95% CI 3.4–10.9; Table 3). With Thailand’s HCP population of 227 349 persons, 13 909 ILS episodes (95% CI 7798–24 781) were estimated to occur annually, resulting in 78 800 days of HCP presenting at work while ill (95% CI 44 178–140 392), 8823 days of HCP missed work (95% CI 4946–15 719), and 3621 days of hospital beds occupied (95% CI 2030–6451). In terms of economic burden, these episodes would result in US$235 135 economic loss when all fees, monetary income loss, and time loss were accounted for (95% CI US$131 826–US$418 922; Table 4).