Development and Initial Psychometric Validation of the COVID-19 Pandemic Burden Index for Healthcare Workers

Abstract

Background

The burden of COVID-19 on healthcare workers (HCWs) is reported to be increasing, yet the psychometric scales now in use evaluate only single aspects; few measure the pandemic-specific burden on HCWs comprehensively.

Objective

To develop a scale to quantify the physical, mental, and socioeconomic burden of the COVID-19 pandemic on HCWs.

Design

Scale development and cross-sectional survey.

Participants

Consenting HCWs aged ≥20.

Main Measures

Development of an item-list based on literature reviews and HCW panel input, evaluation of content validity and item selection using the Delphi method, psychometric testing conducted on HCWs, validity assessment by factor analyses and hypothesis verification, internal consistency evaluation by Cronbach’s alpha, test-retest analysis, and interpretability assessment.

Key Results

Through the Delphi process, a 29-item pilot scale was generated. In psychometric testing, data from 863 HCWs contributed to the development of the final version of this scale, called Pandemic Burden Index twenty for HCWs (PBI-20), a 20-item scale to measure six domains: fatigue, fear of infection, inadequacy as a medical professional, mental health concerns, prejudice or discrimination, and anxiety about one’s livelihood and daily life. Factor analysis showed each factor corresponded to the six domains of this scale. Hypothesis verification showed the PBI-20 total score to be moderately to highly correlated with the Short Form 36 vitality score and mental health score and with intention of turnover. The PBI-20 had good internal consistency (Cronbach’s alpha 0.92). Test-retest analysis showed the intraclass correlation coefficient to be 0.70 and the minimal important change to be −7.0.

Conclusions

The psychometrically sound questionnaire we developed to measure pandemic-specific burdens for HCWs provides an understanding of comprehensive burdens on HCWs and may serve to evaluate interventions to reduce the burdens.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11606-023-08028-3.

INTRODUCTION

The pandemic of COVID-19, a highly infectious, rapidly spreading, life-threatening respiratory infection, has caused changes in the healthcare working environment and heavier workloads for healthcare workers (HCWs).^1,2 Vaccines and treatments for COVID-19 have been developed, but the burden on HCWs is still inadequately studied. HCWs involved in the care of COVID-19 patients, especially in the early stage of the COVID-19 pandemic, faced and still face unprecedentedly diverse and unique burdens.² Frequent donning and doffing of personal protective equipment lead to physical burdens such as fatigue, skin injuries, and headaches.^3,4 In addition, HCWs fear that they may infect family members and friends, becoming infected themselves,⁵ and suffer traumatic stress,⁶ depression, anxiety,^7,8 burnout, and sleep problems.^9,10 The burden of COVID-19 has several aspects and can lead to HCW turnover and worse mental health.¹¹ To investigate the burdens of the COVID-19 pandemic on HCWs in all of their several aspects, a comprehensive instrument to measure the specific burdens on HCWs is essential.

Previous studies developed new scales for HCWs to measure fear, anxiety, and stress due to COVID-19,^12–16 but important aspects of pandemic-specific burdens such as fatigue,^3,4,17 discrimination,^18,19 and inadequacy as a medical professional were lacking.^20–22 To measure the burden of the COVID-19 pandemic adequately, a comprehensive scale focused on the pandemic is needed.

The purpose of this study was to develop a scale to quantify the physical, mental, and socioeconomic burdens faced by HCWs during the COVID-19 pandemic and to evaluate the validity and reliability of this scale.

METHODS

Overview

This study was carried out in two phases: the first phase comprised item-list development by the HCW panel using the Delphi method; the second phase was psychometric testing of HCWs.

Phase 1: Item List Development Using the Delphi Method

Phase 1 was conducted from March 2020 to October 2020. Three clinicians and epidemiologists (the “development panel”) discussed the concept of burden on HCWs and defined it as the physical, mental, and socioeconomic stressed conditions that result from engaging in healthcare during the COVID-19 pandemic. Four steps were taken to develop an item list to measure burdens on HCWs.

The first step was to extract candidate domains caused by the pandemic burden. One reviewer (R.Y.) reviewed the literature that evaluated the burden on HCWs of recent coronavirus pandemics (severe acute respiratory syndrome (SARS), Middle East respiratory syndrome (MERS), and COVID-19) (Supplementary Methods, Supplementary Table 1).

In the second step, the development panel discussed and identified additional candidate domains. As a result, an item list including six domains and 20 candidate items was developed.

The third step was the selection of an HCW panel. Because the burden may vary depending on the nature of the HCW’s work, we contacted multidisciplinary HCWs who work in healthcare facilities and are indirectly or directly involved in COVID-19 care. These HCWs were selected from among those who had conducted clinical research in collaboration with the development panel in the past. We sent an email to eight HCWs asking them to join the HCW panel. Eight multidisciplinary HCWs including physicians (in respiratory, rheumatology, general internal medicine, general practice, and surgery), nurses, and physical therapists agreed to be panelists.

In the Fourth step, three Delphi rounds were used to select domains and items from the list.^23,24 Details of Delphi rounds are described in the Supplementary Methods. As a result of three rounds, 19 items were removed and we developed a 29-item pilot scale with seven domains representing pandemic-specific burdens for HCWs: fatigue, problems with sleep, inadequacy as a medical professional, fear of infection, mental health concerns, prejudice or discrimination, and anxiety about one’s livelihood and daily life (Supplementary Table 2). Two of the 29 items (global items Q28 and 29) were intended to evaluate the burdens in a general, comprehensive way. All items were rated on a 5-point polytomous response scale where 1 = never, 2 = rarely, 3 = sometimes, 4 = frequently, and 5 = always. The scale items were worded to ensure that negative answers (indicating a higher burden) were always at the higher end of the scale, and higher scores indicated a greater burden.

Phase 2: Psychometric Testing

We conducted a cross-sectional, web-based self-report survey among Japanese HCWs between January 16 and March 17, 2021. This period coincides with the spread of COVID-19 in Japan and the government’s declaration of a second state of emergency (Supplementary Figure 1). Our study was approved by the institutional review board at the Institute for Health Outcomes and Process Evaluation Research and Kyoto University. Participants’ consent to participation in this study was signified by filling out the consent portion of the questionnaire. Snowball sampling was used. First, the development and HCW panelists emailed inquiries to potential participants for recruitment. After participants responded to the questionnaire, they were requested to forward the email to other referrals. Included were all consenting HCWs aged 20 and over who worked in healthcare facilities during the survey period: excluded were HCWs who worked only outside a healthcare facility, or who withdrew their consent. In cases where HCWs responded in duplicate, the older responses were excluded.

The participants answered the 29-item pilot questionnaire; designed to measure the burden of HCWs, it also solicited information on the participants’ characteristics, work environment, and degree of involvement in COVID-19 patients’ care. The Short Form-36 vitality (SF-36) and mental health,^25,26 and the Pittsburgh Sleep Quality Index (PSQI),²⁷ because they cover overlapping similar domains, were used as comparators to evaluate the relevance of participant reports.

Item Reduction

We used parallel analysis to define the number of domains (with the exception of the global domain (Q28, 29)). A correlation matrix was assessed using the Kaiser-Meyer-Olkin (KMO) test.²⁸ Items were assumed to be continuous. Explanatory factor analysis with oblique (oblimin) rotation was used.²⁹ To make the instrument shorter and more suitable for practical use, we discussed whether to remove items if warranted by the findings of the factor loading, item-total correlations, or a change in Cronbach’s alpha caused by dropping a particular item.

Validity

We performed confirmatory factor analysis assuming a second-order factor model for the remaining domains. In this model, we assumed the remaining domains comprised the first-order factors, which are influenced by the COVID-19 burden as the second-order factor. Standardized root-mean-square residuals (SRMR), root-mean-square error of approximation (RMSEA), and comparative fit index (CFI) were used to assess the goodness of fit.³⁰ We calculated a tentative total score and the correlation between the tentative total score and the total score of the two global items. If the correlation was high, a definitive total score was calculated next by summing the scores of all items, including global items. We summarized the data and the distribution of item scores. We also assessed ceiling or floor effects and the characteristics of missing data.

Construct validity was evaluated by hypothesis verification according to a priori formulated hypotheses (Table 1). Pearson’s correlation was used in the correlation analysis.³¹ Correlation coefficients from 0.1 to less than 0.243, from 0.243 to less than 0.371, and 0.371 or higher were defined as low, moderate, and high, respectively. Standardized mean difference (SMD) was used to compare the two groups; from 0.2 to less than 0.5, an SMD was considered small; from 0.5 to less than 0.8, it was considered medium; and at 0.8 or higher, it was considered large.³²

Table 1.

Hypotheses for the Cross-Sectional Construct Validity of PBI-20 Score

No.	Domain	Hypothesis	Correlation or SMD*	Accepted
1	Fatigue	The fatigue domain was expected to be moderately to highly correlated to the vitality score of SF-36.	0.66	Yes
2	Fear of infection	We expected a higher score for fear of infection domain in participants with COVID-19 comorbidity risk (answered yes to the question “In the past month, have you been treated for any disease known to be a risk for COVID-19?”) than in participants without COVID-19 comorbidity risk.	0.09*	No
3	Inadequacy as a medical professional	The inadequacy as a medical professional domain was expected to be moderately to highly correlated to the question “Has the care you provide in the past month been more limited than before the COVID-19 outbreak?”	0.41	Yes
4	Mental health concerns	The mental health concerns domain was expected to be moderately to highly correlated to the mental health score of the SF-36.	0.70	Yes
5	Prejudice or discrimination	The prejudice or discrimination domain was expected to be moderately to highly correlated to the question “In the past month, have you ever refrained from using a restaurant, commercial facility, public facility, or service because you are a medical worker?”	0.17	No
6	Prejudice or discrimination	The prejudice or discrimination domain was expected to be moderately to highly correlated to the question “In the past month, have you ever been refused permission to use a restaurant, commercial facility, public facility, or service because you are a medical worker?”	0.24	No
7	Anxiety about one’s livelihood and daily life	The anxiety about one’s livelihood and daily life domain was expected to be moderately to highly correlated to the question “How much do you expect your income in the current year (April 2020 to March 2021) to change from your income in the previous year (April 2019 to March 2020)?”	0.32	Yes
8	Total Score	We expected a higher PBI-20 total score in frontline compared to non-front-line jobs†	0.33*	Yes
9	Total Score	We expected a moderate to high correlation between the PBI-20 total score and the percentage directly engaged in the care of COVID-19.	0.19	No
10	Total Score	We expected a higher PBI-20 total score in women than in men.	0.32*	Yes
11	Total Score	We expected a higher PBI-20 total score in nurses than in physicians.	0.64*	Yes
12	Total Score	We expected higher PBI-20 total scores in areas with a declared national emergency than in areas without a declared national emergency.	0.11*	No
13	Total Score	We expected a moderate to high correlation between the PBI-20 total score and intention to leave the facility.	0.44	Yes
14	Total Score	We expected a moderate to high correlation between the PBI-20 total score and intention to leave the healthcare profession.	0.40	Yes
15	Total Score	We expected a moderate to high correlation between the PBI-20 total score and the PSQI score.	0.47	Yes
16	Total Score	We expected a moderate to high correlation between the PBI-20 total score and the SF-36 vitality score.	0.63	Yes
17	Total Score	We expected a moderate to high correlation between the PBI-20 total score and the SF-36 mental health score.	0.67	Yes
Number of hypotheses that were accepted			12/17 (70.6%)

PBI-20 Pandemic Burden Index twenty, SF-36 Short Form 36, COVID-19 Coronavirus Disease 2019, PSQI Pittsburgh Sleep Quality Index

*Standardized mean difference (SMD)

^†Non-frontline jobs were defined as those who answered “0% (not at all)” to the question “What percentage of your total work hours per week is related to care for patients with COVID-19?”

All correlation coefficients were computed using the Pearson correlation. Correlation coefficients from 0.1 to less than 0.243 = low, from 0.243 to less than 0.371 = moderate, 0.371 or higher = high. SMD from 0.2 to less than 0.5 = small, from 0.5 to less than 0.8 = medium, 0.8 or higher = large

Reliability

We calculated Cronbach’s α to determine the internal consistency of the items (interrelatedness among items). An alpha value above 0.7 is desirable. Test-retest analysis (consistency across time) was assessed in respondents who were administered the scale 6 months after the initial survey (from 3 August to 10 October 2021). To detect unchanged participants, we used answers to an anchor question on a 7-point global rating of change (GRC) scale.³³ In the case of participants whose answers to the anchor question indicated no change (GRC scale = 4), we evaluated the reproducibility of the total scores using the intraclass correlation coefficient (ICC). We quantified the measurement error by Bland-Altman plot analysis and by calculating the standard error of measurement (SEM) from the ICC formula.^34,35 The SEM refers to within-subject variability in repeated measures for evaluating individual change.

Interpretability

To estimate the minimal important change (MIC) in the total score of this scale, we used the mean change method³⁶; this method calculates the mean score of participants who answered the anchor question “slightly improved (GRC scale = 3).” The smallest detectable change (SDC), which is a measure of the variability of a scale owing to measurement error, was calculated using the formula: 1.96×$\sqrt{2}$×SEM. The SDC indicates the amount of change necessary to be significant on the reliable change index.³⁷

All data analyses were performed using R Statistical Software version 4.1.1 with the “psych,” “lavaan,” and “semPlot” R packages.^38–41

Translation

The original version of the questionnaire was forward-translated into English by two independent translators whose mother tongue is English. A consensus English translation was created before it was translated back into Japanese by another independent individual, who was blinded to the original Japanese version.

RESULTS

Participant Characteristics

During the study period, a total of 894 responses were received; of these, 31 were excluded (five withdrew their consent, 25 were duplicate responses, and one was not working in a healthcare facility). After these exclusions, 863 responses were analyzed. Participant characteristics are shown in Table 2.

Table 2.

Participant Characteristics

	Overall
Variables	n = 863
Age, median (IQR)	34 (29, 40)
Female, n (%)	450 (52)
Post-graduate year, median (IQR)	9 (5, 15)
Occupational category, n (%)
Physician	422 (49)
Nurses	370 (43)
Midwife	17 (2.0)
Pharmacist	16 (1.9)
Physical therapist	12 (1.4)
Others	26 (3.0)
Specialty category
Internal medicine	377 (44)
Surgery	132 (15)
Intensive care	78 (9.0)
Emergency medicine	153 (18)
Others	123 (14)
Employment
Full-time	783 (91)
Part-time	70 (8.1)
Other	10 (1.2)
Night shift (per month), median (IQR)	4.0 (1.0, 6.0)
Institutional Type
Community hospital	581 (67)
Academic hospital	214 (25)
Others	68 (7.9)
Designated hospital for infectious disease, n (%)	343 (40)
Percentage of daily care for patients with COVID-19, n (%)
None	214 (25)
1–20%	381 (44)
21–40%	90 (10)
41–60%	59 (6.8)
61–80%	43 (5.0)
81–100%	76 (8.8)

IQR interquartile range, COVID-19 Coronavirus Disease 2019

Item Reduction from the Pilot Scale

Nine items were removed from the pilot scale (Supplementary Table 2). Although no item was more than 90% biased toward a single answer, in the domain of problems with sleep (Q6 and Q7) about 50% of respondents answered “never” (Supplementary Table 3). After discussion among the development panelists, we concluded that these questions have a weak association with the burden, and removed them. The deletion of Q4, Q11, Q12, and Q16 did not decrease Cronbach’s alpha coefficients (Supplementary Table 4); we removed these items because we considered that they were similar to other questions and deleting them would not make the scale less reliable.

An exploratory factor analysis was performed on all but the global domain. The KMO measure of sampling adequacy was 0.93 (marvelous). Parallel analysis indicated a 6-factor solution. The factor analysis showed each factor corresponded to the six domains of this scale. We removed Q5 because of its small factor loadings (Supplementary Figure 2). Q17 and Q18 were also removed because they loaded on both the fear of infection and mental health concerns. After these item reductions, six domains and 18 items remained.

Confirmatory Factor Analysis

Confirmatory factor analysis assuming a second-order factor model for the remaining 18 items was performed (Fig. 1) and indicated acceptable goodness-of-fit: SRMR 0.06, RMSEA 0.06, CFI 0.96.

Figure 1.

Structural diagram of the second-order factor model. Brd: Burden, Ftg: Fatigue, Foi: Fear of infection, Ind: Inadequacy as a medical professional, Mnt: Mental health concerns, PrD: Prejudice or discrimination, Anx: Anxiety about one’s livelihood and daily life. The first-order factors are the six domains. Those are influenced by the COVID-19 burden as the single second-order factor. The standardized estimated coefficient is shown for each path.

We calculated a tentative total score from these 18 items that correlated well with the total score of the two global items (Pearson’s correlation coefficient 0.77). To measure the COVID-19 burden, which might not be captured by the six domains, we included the global items when calculating the total score.

The final version of the questionnaire, called Pandemic Burden Index twenty for HCWs (hereinafter PBI-20), included twenty items, all of which were scored from 1 to 5, giving a total score range of 20 to 100 (Table 3, Supplementary Figure 3). Data from 863 respondents showed scores ranging from 20 to 90 (mean 46.1 SD 13.3), as described in Fig. 2.

Table 3.

Items Used in the PBI-20 Questionnaire (final version)

Domain	No.	Item	Mean (SD)	Factor loading†	α
Fatigue	1	Q1	2.55 (1.05)	0.92	0.92
	2	Q2	2.26 (1.05)	0.77
	3	Q3	2.42 (1.07)	0.77
Inadequacy as a medical professional	4	Q8	2.69 (1.12)	0.82	0.81
	5	Q9	2.46 (1.13)	0.59
	6	Q10	2.70 (1.17)	0.76
Fear of infection	7	Q13	3.06 (1.07)	0.91	0.84
	8	Q14	1.93 (1.02)	0.55
	9	Q15	3.05 (1.18)	0.77
Mental health concerns	10	Q19	2.32 (1.13)	0.72	0.86
	11	Q20	2.19 (1.10)	0.87
	12	Q21	1.65 (0.94)	0.67
Prejudice or discrimination	13	Q22	2.15 (1.07)	0.55	0.85
	14	Q23	1.52 (0.82)	0.90
	15	Q24	1.50 (0.81)	0.94
Anxiety about one's livelihood and daily life	16	Q25	1.34 (0.72)	0.55	0.68
	17	Q26	1.87 (1.11)	0.96
	18	Q27	2.74 (1.24)	0.36
Global	19	Q28	2.74 (1.14)	–	0.90
	20	Q29	2.98 (1.11)	–

PBI-20 Pandemic Burden Index twenty, SD standard deviation, α: Cronbach’s alpha (calculated by each domain)

*The items were rated on a 5-point polytomous response scale ranging 1 = never, 2 = rarely, 3 = sometimes, 4 = frequently, and 5 = always

^†Factor loadings to other domains were all less than 0.3

There are no missing data. The 6-factor solution was responsible for the common variance constituting 65.8% of the total variance. The share of explained variance by individual factors was as follows: fatigue (27.3%), fear of infection (21.2%), inadequacy as a medical professional (18.1%), mental health concerns (28.3%), prejudice or discrimination (19.4%), and anxiety about one’s livelihood and daily life (13.5%)

Figure 2.

The distribution of PBI-20 total scores. Total scores can range from 20 to 100 points. Data from respondents ranged from 20 to 90 points (mean 46.1 points, standard deviation 13.3 points).

Hypothesis Verification

A total of 70.6% of the hypotheses were accepted (Table 1). The PBI-20 total score correlated well with both the SF-36 vitality score and the SF-36 mental health score. The correlation between the PBI-20 total score and intention to turnover was moderate.

Internal Consistency

Cronbach’s alpha, a coefficient estimating reliability, was 0.92 (95% CI 0.91 to 0.93). Each domain of Cronbach’s alpha was >0.7, except for anxiety about one’s livelihood and daily life, 0.68 (Table 3).

Test-Retest Analysis

Test-retest analysis was evaluated at roughly 6-month intervals after the initial survey. This was the period during which the COVID-19 outbreak was occurring, similar to the initial survey (Supplementary Figure 1). Of the 863 participants in the first survey, 427 responded in the second survey. Of these, 8 responses could not be linked to the first survey; thus, a total of 419 respondents (48.6%) underwent retest reliability analysis. There was no substantial difference between the background of participants who responded to the second survey and those who did not (Supplementary Table 5). Of the 419 retest respondents, 173 participants gave the answer “unchanged” to the anchor question. The ICC for test and retest PBI-20 total scores was 0.70 (95% CI 0.62 to 0.76). In the Bland-Altman plot, the mean difference was −2.51 (95% CI −3.91 to −1.10), and the 95% limits of agreement were −20.9 to 15.8 (Supplementary Figure 4). The SEM was 6.83.

Minimal Important Change and Smallest Detectable Change

The Pearson correlation coefficient measuring the correlation between the changes in the total PBI-20 scores (between the first survey and second survey) and the seven-point GRC scale was 0.40 (95% CI 0.32 to 0.48). The MIC defined by the mean change method was −7.0 points (Table 4). The SDC was 18.9.

Table 4.

Mean Differences of the PBI-20 Total Score by the GRC Scale

	GRC categories	Total score at baseline		Total score at 6 months		Change in scores
	n (%)	Mean	SD	Mean	SD	Mean	SD
Total group	419 (100)	45.4	12.7	42.4	12.9	-2.9	10.9
GRC scales
1. Very much improved	8 (1.9)	43.4	19.9	24.0	3.4	-19.4	20.9
2. Much improved	34 (8.1)	42.2	12.6	32.6	8.3	-9.6	12.6
3. Slightly improved	83 (19.8)	48.6	13.6	41.6	11.3	-7.0	10.0
4. Unchanged	173 (41.3)	43.5	12.2	41.0	12.5	-2.5	9.4
5. Slightly worse	76 (18.1)	44.8	10.3	45.6	10.9	0.8	8.1
6. Much worse	34 (8.1)	48.8	14.0	53.5	12.2	4.8	10.9
7. Vastly worsened	11 (2.6)	55.3	9.4	58.6	11.9	3.4	8.4

PBI-20 Pandemic Burden Index twenty, GRC global rating of change, SD standard deviation

DISCUSSION

Summary of Findings

We developed a comprehensive COVID-19 pandemic-specific measure of HCWs’ burden with six domains and two global items. Development was based on scientifically and psychometrically validated methods. The questionnaire’s robustness was ensured by a rigorous development process that included multidisciplinary and expert discussions. This process was conducted by a development panel with experience in developing psychometric scales and a panel of HCWs who worked in healthcare during the COVID-19 pandemic. PBI-20 showed good content validity, construct validity, and reliability.

Previous Measures for Burdens on HCWs

Evidence in the literature of comprehensive measuring of the burden on HCWs during the COVID-19 pandemic is limited.^3,9,12–15 One scale to measure the burden on HCWs, the Tokyo Metropolitan Distress Scale for Pandemic, measures anxiety about infection and social stress.⁴² However, the content validity of this scale has been insufficiently evaluated, and the reliability and interpretability of the scale has not been assessed. PBI-20, which was developed on a methodologically sound scale, consists of six domains (fatigue, fear of infection, inadequacy as a medical professional, mental health concerns, prejudice or discrimination, and anxiety about one’s livelihood and daily life), plus a global domain. These domains have been emphasized as burdens on HCWs in various countries.^{2–7,15,18,19,43} Although the application of the scale to populations other than Japanese HCWs has not been evaluated, our scale has the potential to provide a valid and comprehensive measure of burden. Further studies in various countries outside Japan are needed to verify our results.

PBI-20’s Total Score and Intention to Turnover

Participants who had a turnover intention had a higher PBI-20 total score, and this score may be used to identify conditions leading to turnover. Determinants of turnover in HCWs are varied^44,45; among them, stress, frustration, and fear of infection are reported to be important.^{11,44,46–48} The reason this scale was shown to be associated with turnover may be because it measures feelings of inadequacy as a medical professional and fear of infection.^44,46 Higher PBI-20 total scores were also associated with poorer SF-36 vitality and mental health scores. Several studies have shown that deterioration in quality of life, such as physical and psychological health, is associated with burnout and turnover in the workplace.^47,48 Because increased turnover in healthcare might lead to the healthcare system’s collapse⁴⁶ and have a huge impact on society, this risk underscores the need to quantify and monitor the burden on HCWs.

Implications of Study Findings

The PBI-20 scale is a self-administered rating scale consisting of 20 simple items. This scale is intended to help HCWs understand their burden status. It may also help employers and policymakers identify groups of workers at high burden and induce administration to put more resources where they work. Moreover, the phases of the pandemic might conceivably have influenced the development of the items and the subsequent utility of the measure. The items of prejudice or discrimination, for example, might be less burdensome in the present than in the early phases of the pandemic. Measuring the PBI-20 over time may reveal temporal changes in the pandemic burden. Furthermore, this scale could be used as an outcome measure to evaluate the effectiveness of interventions to reduce the burden on HCWs.⁴⁹ If verified by future studies, this scale could also be of use in other pandemics of other emerging infectious diseases that, like COVID-19, are highly transmissible, spread rapidly, and cause severe symptoms.

Limitations

This study has several limitations. First, most participants were HCWs working in acute care hospitals, and young. Second, there may be nonresponse bias. Some HCWs who received this questionnaire may not have responded because they might not be sufficiently interested in the survey or because they might be too burdened by COVID-19 to respond. We were unable to evaluate nonresponse bias because we did not obtain information to calculate response rates. Third, we included only participants who worked in Japan. Additional research is needed to verify whether the results are generalizable to groups with different languages and from other cultures. Fourth, the test-retest analysis was evaluated after about 6 months; although a target time lapse of 2 weeks is recommended,⁵⁰ the length of the most appropriate interval is unknown.³³ Our results showed good reliability after measurement twice, after a sixth-month interval.

CONCLUSION

PBI-20, the scale we developed to quantify the physical, mental, and socioeconomic burden of HCWs during the COVID-19 pandemic, can help to identify high-burden groups and can be utilized as an outcome measure to assess the effectiveness of interventions to reduce the burden on HCWs.

Declarations

Conflict of Interest

The authors declare that they have no conflict of interest.