Getting back to work: risk assessing personal vulnerabilities to COVID-19

With lockdown restrictions lifting across many countries, organisations are considering how to modify their premises to reduce the risk of COVID-19 transmission amongst their workforces and third parties to as low as reasonably practical. Whilst the specific interventions may differ according to industry type, they can be summarised into those that reduce spread via direct personal contact or fomites, physical distancing measures and consideration of the environment such as ventilation patterns. SARS-COV-2 can be effectively destroyed by several cleaning agents, and thus stringent disinfection processes with attention to areas of potential surface contact is paramount. Increasing air change frequency and flow rates such as opening doors, windows or by moving activities outdoors are established mitigation strategies for similar transmissible diseases. Interpersonal contact may be reduced by home working, staggered usage or closure of communal areas, and behavioural changes such as avoidance of handshakes and close embraces. The role of personal protective equipment has been established where close contact cannot be avoided, with several countries recommending or mandating their use in crowded indoor buildings, as well as public transport. Whilst such public health measures are well established, implementation may be challenging for several reasons including limited space and possible impact on productivity. Instilling the behavioural changes such that individuals take account of the risks of COVID-19 transmission in the workplace may require time. Staggering timings such that the most essential workers return first, followed by less essential and finally medically vulnerable staff are pertinent considerations (Fig. 1 ).

Fig. 1.

Risk assessment matrix for COVID-19 vulnerability.

Such public health measures are well established, but less is known about how organisations can risk assess their staff according to underlying demographic features and personal vulnerabilities to COVID-19 infection and mortality. Age and gender-related disparities have been identified, with an analysis of 3665 cases and deaths in mainland China estimating that 1% of the population younger than 30 years would require hospital care, whereas this rose to 11.8% for those aged 60 years or older.¹ Male gender also appears to confer an excess risk. One study from Italy found a 4:1 ratio of male to female hospital deaths in more than 12,000 cases, whereas another from the United States noted an association but lower relative risk.^{2 , 3}

Differences in case fatality ratios according to deprivation status and ethnicity have received considerable public and media interest in the United Kingdom. Data from intensive care units indicate that one-third of 6574 patients were from black and minority ethnic backgrounds as of the end of April 2020.⁴ Chronic medical conditions such as obesity, cardiovascular illness such as diabetes and hypertension, asthma and cancer are also implicated.⁵ More recently, disaggregated data regarding the risk factors for having a positive PCR swab test and for COVID-19–related mortality have become available. In studying the odds of a positive test of 3802 UK primary care patients, adults aged 40–64 years had more than 5 times that of children, whereas those in more deprived areas and with chronic kidney disease had approximately twice the odds of their counterparts.⁶ A larger study analysed the factors associated with COVID-19 hospital death through examining the linked electronic health records of 17 million adult patients in the United Kingdom.⁷ The calculated hazard ratios provide an opportunity to inform risk matrices assessing workers’ vulnerability to COVID-19 in a systematic way, whilst recognising the limitations in applying general population data to the working population and those of sampling bias. In our method, logarithms were calculated for relevant hazard ratios and multiplied by 10 to produce a weighting factor which was subsequently imputed into a final matrix. We have suggested cut-offs for total scores for three levels of risk which we believe will apply to most organisations.

Certain variables were excluded from the final matrix for a variety of reasons. The hazard ratios for hypertension and splenic disease were non-significant in the original study. Use of hazard ratio for deprivation status at individual level may invoke the ecological fallacy. A counterintuitive protective association between smoking and critical hospital illness in COVID-19 was identified which may have resulted from selection pressures (collider bias) in the underlying data. Specific medical recommendations are provided elsewhere for supporting pregnant workers, although an excess risk of COVID-19–related complications has not been identified. Finally, in haematological malignancies, liver disease, cancers and other chronic diseases, treatment regimens may be complex and variable. We expect that such cases will be relatively rare amongst most workforces but merit specific consideration. We believe this approach is a useful additional tool to support organisations, with broad generalisability. In addition to acting as a stand-alone tool, it may complement existing risk assessment templates as an adjunct for decision-making. Concern regarding vulnerable cohabitants and feelings regarding readiness for return are difficulty to quantify but may influence return to work. Risk assessment templates may be developed to take such considerations into account.