Addressing the Costs and Care for Long COVID

2. Long COVID impacts health systems and the economies

Abstract

This chapter provides an overview of existing evidence on the economic burden of long COVID, focussing on both direct medical costs and indirect costs linked to productivity losses, including absenteeism, reduced working hours and labour-market withdrawal. It then introduces the OECD microsimulation model developed to produce retrospective and prospective estimates of these costs across OECD and EU Member countries, drawing on the inputs identified through the literature review.

2.1. The economic impact of long COVID is substantial, and mainly stems from the indirect costs from reduced productivity and participation in the workforce

Although its clinical features are now better understood, the economic and social consequences of long COVID are only beginning to be systematically measured. This section reviews and synthesises the available empirical research and modelling analyses quantifying the socio‑economic burden of long COVID. These sources will serve to feed certain parameters of the OECD model presented in the next section.

2.1.1. Long COVID generates substantial direct medical costs

Evidence from multiple countries shows that long COVID has become a persistent source of health-system demand and spending, extending well beyond the acute phase of infection. Rather than a short-lived recovery period, it generates continuing use of general practice, diagnostics, outpatient and hospital services.

In the United Kingdom, a linked-records study of over 280 000 adults with long COVID found that higher use of general practice, outpatient, inpatient and emergency care persisted or even increased during two years of follow-up compared with several matched control groups (Mu et al., 2024[1]). Primary care data similarly show that consultation costs for adults with long COVID were around 40% higher than for COVID‑19 patients without persistent symptoms – equivalent to about GBP 23 million in additional National Health Service spending in 2020-2021 (Tufts et al., 2023[2]). These results confirm that long COVID imposes a structural, not transient, load on frontline services.

Comparable patterns emerge elsewhere. In the United States, analysis of 277 000 insurance claims by adults showed medical spending in the year after infection averaging USD 30 400, compared with USD 21 000 in matched controls – an incremental USD 9 400 per patient-year (+45%), driven mainly by hospital and outpatient care (Scott et al., 2024[3]). In France, community-managed adults meeting the World Health Organization (WHO) definition of long COVID had sustained excess use of primary care, specialist consultations and laboratory tests for up to two years (Yang et al., 2025[4]). Israeli data show 70‑90% higher monthly healthcare costs among long COVID patients at 4‑12 months, with inpatient spending nearly doubled (Wolff Sagy et al., 2023[5]). Synthesising results from multiple OECD and EU countries, Łukomska et al. (2025[6]) estimate direct medical costs of around GBP 3 000‑3 500 or EUR 4 000‑5 000 per affected patient per year (2‑3 times higher than for comparable individuals without long COVID) and national totals already in the hundreds of millions of euros annually.

Beyond the overall rise in expenditure, the evidence reveals where the extra costs occur. Across OECD countries, hospital and specialist outpatient care account for roughly half to two‑thirds of the total incremental spending, reflecting recurrent admissions and follow-up for cardiometabolic, respiratory and mental health complications. Primary care contributes about 20‑30% – particularly among community-managed cases – while diagnostic tests, imaging and prescriptions represent the remaining 10‑20%.

Together, these patterns show that long COVID imposes both a financial burden on hospital budgets and a sustained operational strain on general practice and diagnostic services, underscoring the need for co‑ordinated responses across levels of care.

Across this evidence base, long COVID emerges as a chronic budgetary and capacity challenge for health systems, irrespective of financing model. Medical costs per patient typically rise by 40‑100% relative to controls, and persist for 12‑24 months. Direct healthcare spending absorbs scarce resources and underscores the need for integrated long-term management, rehabilitation and prevention strategies to mitigate its financial and service impacts.

2.1.2. The indirect costs of long COVID are a major burden for OECD and EU Member countries

The findings above only focus on the direct medical costs generated by long COVID, but this condition also has a much broader socio‑economic toll that shapes labour markets, household incomes and public finances. A growing body of micro-level evidence shows that long COVID imposes a measurable drag on labour-market participation, working hours and earnings. Despite variation in study design, case definition and follow-up, the findings converge on a consistent conclusion: individuals affected by long COVID are significantly more likely to exit employment – temporarily or permanently – reduce their working hours, or experience sustained declines in work ability that translate into productivity losses, with significant economic consequences.

Early international evidence underscored the severe and prolonged impact of long COVID on work ability. In a global cohort of over 3 700 participants, Davis et al. found that nearly half had reduced work schedules or were unable to work seven months after infection, with fatigue and cognitive dysfunction cited as primary barriers to return. These early findings foreshadowed the employment and productivity losses later documented in national longitudinal studies (Davis et al., 2021[7]).

Rigorous evidence from longitudinal population data in the United Kingdom confirms a sustained impact of long COVID on employment outcomes. Using linked administrative records that allow within-person comparisons, Ayoubkhani et al. reported that the odds of economic inactivity were 45% higher 30‑40 weeks after infection and 34% higher 40‑52 weeks after infection among individuals with long COVID, relative to their pre‑infection status – equivalent to roughly 27 000 inactive working-age adults in mid-2022 (Ayoubkhani et al., 2024[8]). Complementary panel analysis by Reuschke et al. similarly found an elevated risk of employment exit and episodes of zero-hours work (a proxy for extended sick leave) among individuals with symptoms lasting 29 weeks or more (Reuschke and Houston, 2022[9]). When aggregated, these labour market disruptions represent a significant economic burden; the United Kingdom estimated that the annual national productivity loss from long COVID amounted to GBP 5.7 billion (Kwon et al., 2024[10]). Evidence from smaller occupational cohorts, including co-produced online surveys (Ziauddeen et al., 2023[11]) and healthcare worker panels (Grant et al., 2024[12]), corroborates these patterns, documenting increased inactivity, recurrent sick leave and reduced hours, as well as the associated psychosocial and financial strain of prolonged illness.

Outside the United Kingdom, results are consistent. In Belgium, a national cohort survey (Smith et al., 2022[13]) found that 14% of previously employed adults had not returned to work six months after infection. Analyses of US national surveys (Bonner and Ghouralal, 2024[14]; Ford et al., 2025[15]; Ford et al., 2023[16]) further show a 1.4 times higher likelihood of sickness absenteeism and significant activity limitations among roughly one‑fifth of adults with long COVID, with a three‑fold higher risk of disability among those with comorbidities. Cohen and Rodgers likewise identify elevated rates of work-related disability and accommodation requests – particularly among women and minority groups (Cohen and Rodgers, 2024[17]).

Complementing these survey-based findings, a recent nationwide administrative study of over 150 million US workers by Dennett et al. tracked labour-market outcomes through 2024 (Dennett et al., 2025[18]). It found a 12.9% rise in health-related absences and a 13.1% increase in labour-force exits compared with pre‑pandemic levels, confirming at scale that post-COVID health sequelae continue to depress labour supply. The study also revealed marked occupational differences, with absences and exits concentrated in the healthcare, education and service sectors, underscoring the unequal labour-market impact of long COVID across industries (Dennett et al., 2025[18]).

Collectively, these studies suggest that long COVID leads to employment disruption in around one in five affected workers – equivalent to a 5‑10% loss of labour input per affected individual during the first year after infection. Building on this evidence of reduced employment and working hours, a growing set of cost-of-illness studies using micro-level or bottom-up data have begun to monetise the productivity losses associated with long COVID, providing sharper estimates of their fiscal and societal magnitude (Box 2.1).

Box 2.1. Cost-of-illness studies have monetised the productivity losses associated with long COVID

In the United Kingdom, the Institute for Fiscal Studies estimated an average reduction of 2.4‑2.5 hours worked per week and of GBP 65 per month (6%) in earnings, implying an aggregate annual loss of about GBP 1.5 billion at prevailing prevalence rates (Waters and Wernham, 2022[19]). In Canada, Naik et al. employed validated valuation-of-lost-productivity questionnaires to capture both absenteeism and presenteeism over a two‑year period, estimating an additional 99 hours of lost productivity per quarter and an annualised cost of CAD 13 700 (EUR 9 200) per worker (Naik et al., 2025[20]).

A study in Japan applied the WHO Health and Work Performance Questionnaire in a 12‑month cohort, finding annual productivity losses of USD 21 700 among individuals with persistent symptoms – more than double those who recovered. This highlights the steep economic gradient associated with symptom duration (Konishi et al., 2025[21]). In Europe, Fischer, Reade and Schmal analysed objective productivity data from professional footballers, and identified a sustained 5% output deficit eight months after infection (EUR 3 700 per year), providing independent confirmation of persistent performance impairment in affected workers (Fischer, Reade and Schmal, 2022[22]).

Across these diverse contexts, findings converge on 5‑10% annual earnings losses per affected worker – a magnitude consistent with earlier evidence, now expressed in monetary terms. By converting reductions in hours or performance into currency values, the monetised studies add fiscal clarity, and extend coverage to presenteeism and unpaid work – dimensions often omitted from previous analyses, and which may further enlarge the socio‑economic footprint of long COVID (Box 2.2).

While the direction of effect is remarkably aligned across countries, the breadth of reported magnitudes reflects methodological diversity rather than actual differences in impact. Differences in data sources, case definitions, follow-up duration and valuation methods – from administrative longitudinal panels to self-reported cross-sectional surveys – shape the comparability of results. Recognising these design contrasts is essential for interpreting existing estimates and harmonising future cost-of-illness assessments across countries.

Box 2.2. Presenteeism and unpaid work are the hidden productivity losses of long COVID

While sickness absence and employment exits are visible in administrative data, a large share of long COVID’s productivity burden occurs below the radar of formal labour statistics. Two mechanisms – presenteeism and unpaid work impairment – capture this “hidden” component.

Presenteeism refers to reduced productivity while working. Individuals who remain employed but continue to experience fatigue, cognitive dysfunction or post-exertional malaise often perform below their usual capacity.

• In Japan, Konishi et al. (2025[21]) found that presenteeism accounted for more than two‑thirds of the total productivity cost among workers with persistent symptoms.

• The Canadian Naik et al. (2025[23]) study likewise reported substantial self-assessed productivity loss during working hours, even among those not absent from work.

Traditional sickness-absence metrics fail to capture these within-work deficits, leading to systematic underestimation of the economic burden when only days off work are counted.

Unpaid work refers to the invisible economy. Long COVID also constrains unpaid domestic, caregiving and community work – activities essential to household welfare and indirectly to the formal economy. Naik et al. (2025[23]) included these hours within the valuation-of-lost-productivity framework, valuing unpaid work at local replacement wages. Doing so increased total productivity losses by roughly 15‑20%, underscoring the importance of incorporating non-market work in societal-perspective analyses.

These mechanisms have an important implication for policy. Because presenteeism and unpaid work are largely excluded from administrative datasets, productivity losses estimated from labour-force or employer data likely represent a conservative estimate. Incorporating validated instruments (e.g. valuation-of-lost-productivity questionnaires and the WHO Health and Work Performance Questionnaire) into national health and labour surveys would enable more accurate tracking of the true economic footprint of long COVID.

2.1.3. The true costs of long COVID are best measured in macroeconomic terms

While micro-level evidence has documented the medical and productivity costs of long COVID, national-level modelling shows how these individual impacts aggregate into a broader macroeconomic drag on growth, labour supply and fiscal stability. Across high-income countries (where the macro-level data are available), the message is consistent: persistent post-infection symptoms are not only a health challenge but also a structural brake on economic output.

In the United States, Cutler first estimated a total societal cost of USD 3.7 trillion (equivalent to around 17% of GDP), comprising USD 1 trillion in lost earnings (27%), USD 528 billion in medical spending (14%) and USD 2.2 trillion in quality-of-life and disability losses (59%) (Cutler, 2022a[24]; Cutler, 2022b[25]). More recent modelling by Bartsch et al. refines these projections using a population health-economic framework that integrates surveillance, labour and expenditure data. It estimates an annual burden of USD 218 billion in 2024, with 43% from medical care and 57% from productivity losses. Projected forward, cumulative costs between 2025 and 2050 could reach USD 7 trillion (0.3% of GDP per year), and exceed USD 10 trillion (0.5% of GDP) under higher-prevalence or slower-recovery scenarios (Bartsch et al., 2025[26]).

In the United Kingdom, macroeconomic simulations by Cambridge Econometrics (2024) using the E3ME model1 show a similar pattern. Long COVID is projected to reduce GDP by 0.05‑0.10% annually by 2030 – equivalent to GBP 1.5‑2.7 billion per year, depending on prevalence and recovery rates. The corresponding employment shortfall could reach over 300 000 jobs, while healthcare costs per person average GBP 3 300‑5 000 per year. As elsewhere, the dominant macroeconomic channel is the sustained loss of labour supply, not direct medical spending (Cambridge Econometrics, 2024[27]).

In Australia, the aggregate scale appears smaller but remains economically significant. The economic impact on the Australian economy was estimated to be USD 9.6 billion, or 0.5% of GDP, during the peak years of the pandemic 2020-2021 (Costantino et al., 2024[28]). Angeles et al. estimate that the burden of long COVID in 2022 ranged between AUD 1.7 billion and AUD 6.3 billion, or 0.07‑0.26% of GDP, with 25 000‑100 000 working-age adults absent from the labour force in 2022 (Angeles et al., 2024[29]).

These cross-country differences reflect variations in prevalence, data quality and labour-market resilience, but the direction of impact is uniform. Together, this evidence demonstrates that long COVID has evolved from a clinical condition into a macroeconomic headwind. Across all settings, indirect costs – lost work capacity, early exits and prolonged functioning limitation – account for the majority of the total burden. Addressing this drag will require sustained investments in rehabilitation, workplace accommodation and preventive measures that protect both population health and long-term economic growth. For governments, the question is no longer simply “What does long COVID cost the health system?” but rather “What is the cost of not restoring people’s capacity to work?”

A Nature review of the literature recognises the discrepancies in measures and methodologies used to estimate the economic burden of long COVID (Bansal, 2025[30]). In the absence of a standardised approach, certain epidemiological and economic assumptions are required to account for the divergences in cases definitions, the inclusion criteria for economic impact components, valuation methods, and the timelines under study.

2.2. Estimating the socio‑economic costs of long COVID in OECD and EU countries requires certain modelling assumptions

To propose an estimate of the costs of long COVID in OECD and EU countries, the OECD Secretariat developed a dynamic epidemiological-economic model to project the prevalence of long COVID and estimate its impact on healthcare costs, labour-force participation and GDP from 2020 to 2035. The model integrates COVID‑19 mortality data from OECD Member countries covering the period 2020-2023 to establish baseline long COVID prevalence estimates. The detailed methodology and parameters used in the model are described below in Box 2.3.

Box 2.3. The OECD SPHeP framework: A tool to assess the medium- and long-term effects of top public health threats, including long COVID

The OECD SPHeP framework model is an advanced systems modelling tool for public health policy and strategic planning. The model is used to predict the health and economic outcomes of the population of a country or a region up to 2050. The framework was updated to model the health and the economic consequences of long COVID.

The model focussing on long COVID currently covers 42 countries. These countries include OECD Member States as well as EU27 countries not part of the OECD. For each of the 42 countries, the model uses demographic and risk factor characteristics by age and gender-specific population groups from international databases. These inputs are used to generate synthetic populations, in which each individual is assigned demographic characteristics and a risk factor profile – smoking and obesity. Epidemiological assumptions described in Section 2.2.1, were used to model the incidence, the severity and the duration of long COVID and its consequences on the quality of life.

For each year, a cross-sectional representation of the population can be obtained, to calculate health status indicators such as healthy life expectancy, disease prevalence and disability-adjusted life years using disability weights. Extra healthcare costs associated with long COVID are estimated based on a per-case annual cost using assumptions described in Section 2.2.2.

The labour market module uses estimates described in Section 2.2.3. to relate long COVID status and severity to the risk of absenteeism and employment. These changes in employment and productivity are estimated in number of full-time equivalent workers. The output of the labour market module is also the main input for calculating the impact on GDP. Consistent with the approach used in the OECD’s long-term economic forecasting mode (Guillemette and Turner, 2017[31]), and other established long-term models, such as the World Bank’s long-term growth model, a Cobb-Douglas production function is used (Loayza and Pennings, 2022[32]). In practical terms, the labour component of the working age population (i.e. those aged 15‑74) is modified based on the change of full-time equivalent workers.

For more information on the OECD SPHeP Framework model, see the SPHeP-NCDs Technical Documentation, available at: http://oecdpublichealthexplorer.org/ncd-doc.

2.2.1. Epidemiological assumptions

The epidemiology of long COVID in terms of its incidence, duration and recovery is still evolving. Several research studies and surveys have reported on the prevalence and the severity of long COVID over time, which have been used to inform the model for predicting future trends in this condition. For modelling purposes and to account for uncertainties, the following epidemiological assumptions were made.

The prevalence of long COVID from 2020 to 2023 is derived from COVID‑19 mortality data

The initial long COVID prevalence estimates are derived from reported COVID‑19 mortality data across countries during 2020-2023, which serves as a proxy for COVID‑19 infections and subsequent long COVID incidence. Case fatality ratios for COVID‑19 were used to infer incidence of COVID‑19 infections. Depending on the study population, the risk of developing long COVID among those with SARS‑CoV‑2 infection ranges from 5% to 15% in adults (Subramanian et al., 2022[33]; Ballering et al., 2022[34]; OECD, 2025[35]; Perlis et al., 2023[36]). Based on these estimates, in the model the baseline proportion of people infected by COVID‑19 who developed long COVID was set at 10%.

The risk of developing long COVID decreases with successive variants of SARS‑CoV‑2

The risk of developing long COVID is lower with more recent omicron variants compared to the original wild type and alpha variants of SARS‑CoV‑2. This work assumes a decreasing risk of developing long COVID with successive variants over time, based on reported relative risks (Xie, Choi and Al-Aly, 2024[37]).

The prevalence of long COVID remains constant after 2023

The conversion method using case fatality ratios could no longer be used to calculate long COVID cases after 2023, owing to the lack of a reliable data source on COVID‑19 deaths reported by OECD and EU Member countries. Instead, the model produced two scenarios that assume a stable low or moderate COVID‑19 incidence fixed at 5% and 10% respectively from 2024 onwards, to project long COVID prevalence from 2024 to 2035. These incidence levels are based on reported circulation of influenza virus (flu) affecting 3 to 11% of the population depending on the year (Tokars, Olsen and Reed, 2018[38]). This fixed COVID‑19 incidence and therefore fixed long COVID prevalence assumption is informed by the findings of two population surveys from the United States, which reported a stable 3.4% point prevalence of long COVID over time in 2022 (Adjaye-Gbewonyo et al., 2023[39]) and again in 2024 (Selden, 2025[40]).

The average duration of long COVID symptoms is set at two years

Few longitudinal cohort studies exist that report the duration of symptoms for patients living with long COVID. Nonetheless, a sizeable proportion can expect to recover within 6 months, while others who experience symptoms beyond 24 months are less likely to recover (RIVM, 2025[41]; Ballouz et al., 2023[42]) (Ballouz et al., 2023[42]; Servier et al., 2023[43]). An economic modelling survey from the United States fixed the duration at 1 year (Bartsch et al., 2025[26]). The OECD model assumes an average duration of 2 years for all long COVID cases to account for this variation.

The severity of long COVID is weighted towards mild and moderate cases in the population

Long COVID varies in terms of symptoms and their severity. Owing to a lack of recognition and detection, mild cases are likely to be underestimated and widely under-reported, while moderate and severe cases are over-represented in research studies and surveys. The OECD model assumes the following distribution of long COVID cases by severity: mild (88%), moderate (11%) and severe (1%), based on estimates reported in previous modelling study (Zhu et al., 2024[44]).

2.2.2. Direct healthcare costs assumptions

Healthcare costs are modelled as incremental expenses above baseline per capita health expenditures, differentiated by severity level. Severe cases are associated with a 67% increase in annual per capita healthcare costs (Wolff Sagy et al., 2023[5]),2 moderate cases with a 35% increase (Tufts et al., 2023[2]; Mu et al., 2024[1]), and mild cases with a 7% increase. These multipliers are applied to baseline healthcare expenditure data to calculate the total direct medical costs attributable to long COVID.

2.2.3. Labour-force and productivity assumptions

The model captures both workforce exit and productivity reduction among those who remain employed.

All long COVID cases have an initial minimum six‑week absence from the workforce

Long COVID is defined as persistence of symptoms for at least 3 months following probable or confirmed COVID‑19 illness. Due to persisting symptoms that may impair cognitive function and mental and physical health, the model assumes that all long COVID cases incur an initial period of absence from work of 6 weeks, regardless of their severity. This assumption is based on the average period of short-term sick leave afforded in OECD countries.

The risk of unemployment increases with severity of long COVID

Regarding workforce exit, 18% of severe long COVID cases (Stelson et al., 2023[45]; Ziauddeen et al., 2022[46]; Davis et al., 2021[7]) and 14% of moderate long COVID cases (Venkatesh, 2024[47]; Ayoubkhani et al., 2024[8]; Smith et al., 2022[13]; Dennett et al., 2025[18]) are assumed to exit the labour force entirely, while mild cases are assumed to remain in the workforce.

For individuals who remain employed, the model incorporates two additional components of productivity loss, with a productivity reduction due to absenteeism and presenteeism ranging from 6% of total working time for mild cases to 50% for severe cases (Bonner and Ghouralal, 2024[14]; Ayoubkhani et al., 2024[8]; Davis et al., 2021[7]; Dennett et al., 2025[18]).

The combined effect of workforce exits and reduced productivity among remaining workers is then converted into full-time equivalent worker losses, providing a standardised metric for labour-force reduction.

The macroeconomic impact is estimated by translating labour-force losses into GDP effects

The reduction in available full-time equivalent worker numbers is applied to aggregate economic output measures, assuming a direct relationship between labour input and economic production. A human capital valuation method was used to measure productivity losses.

2.2.4. Sensitivity analyses

Three scenarios are considered to model the future circulation of SARS‑CoV‑2

To address uncertainty in possible dynamics of COVID‑19 infections, three scenarios were developed. The baseline scenario assumes no new COVID‑19 cases from 2024 onwards, resulting in a progressive decline of long COVID prevalence to zero by 2025, accounting for the 2‑year disease duration assumption set in the model. The low residual transmission scenario assumes a 5% annual COVID‑19 incidence rate from 2024 onwards, generating ongoing long COVID cases. The moderate residual transmission scenario assumes a 10% annual COVID‑19 incidence rate from 2024 onwards, generating a higher ongoing long COVID burden. These scenarios provide a range of plausible futures spanning best-case disease elimination to worst-case endemic circulation.

Variations on sick leave duration are included in the analysis

Given uncertainty in the appropriate sick leave parameter, this assumption was integrated to another sensitivity analysis. The default assumption is 6 weeks of sick leave, and a sensitivity range from 0 weeks (representing no formal sick leave) to 12 weeks was tested. This analysis examines how different sick leave policies or work accommodation practices affect the overall productivity loss estimates and economic impact projections.

Model outputs

Based on the above assumptions, the model generates annual estimates for 2020-2035 for long COVID prevalence, incremental direct healthcare costs, labour-force reductions and reductions in GDP. All outputs are provided for the baseline scenario and relevant sensitivity analysis variants, enabling comprehensive assessment of both central estimates and plausible ranges under alternative assumptions.

References

[39] Adjaye-Gbewonyo, D. et al. (2023), Long COVID in Adults: United States, 2022, Centers for Disease Control and Prevention, Atlanta, GA, https://doi.org/10.15620/cdc:132417.

[29] Angeles, M. et al. (2024), “The economic burden of long COVID in Australia: more noise than signal?”, Medical Journal of Australia, Vol. 221/S9, https://doi.org/10.5694/mja2.52468.

[8] Ayoubkhani, D. et al. (2024), “Employment outcomes of people with Long Covid symptoms: community-based cohort study”, European Journal of Public Health, Vol. 34/3, pp. 489-496, https://doi.org/10.1093/eurpub/ckae034.

[47] Balasubramani, G. (ed.) (2024), “The association between prolonged SARS-CoV-2 symptoms and work outcomes”, PLOS ONE, Vol. 19/7, p. e0300947, https://doi.org/10.1371/journal.pone.0300947.

[34] Ballering, A. et al. (2022), “Persistence of somatic symptoms after COVID-19 in the Netherlands: an observational cohort study”, The Lancet, Vol. 400/10350, pp. 452-461, https://doi.org/10.1016/s0140-6736(22)01214-4.

[42] Ballouz, T. et al. (2023), “Recovery and symptom trajectories up to two years after SARS-CoV-2 infection: population based, longitudinal cohort study”, BMJ, p. e074425, https://doi.org/10.1136/bmj-2022-074425.

[30] Bansal, A. (2025), “Economic burden of long COVID: macroeconomic, cost-of-illness and microeconomic impacts”, npj Primary Care Respiratory Medicine, Vol. 35/1, p. 53, https://doi.org/10.1038/s41533-025-00460-8.

[26] Bartsch, S. et al. (2025), “The Current and Future Burden of Long COVID in the United States”, The Journal of Infectious Diseases, Vol. 231/6, pp. 1581-1590, https://doi.org/10.1093/infdis/jiaf030.

[14] Bonner, C. and S. Ghouralal (2024), “Long COVID and Chronic Conditions in the US Workforce”, Journal of Occupational & Environmental Medicine, Vol. 66/3, pp. e80-e86, https://doi.org/10.1097/jom.0000000000003026.

[27] Cambridge Econometrics (2024), “The Economic Burden of Long Covid in the UK”, Cambridge Econometrics, https://www.camecon.com/hubfs/145725293/The-Economic-Burden-of-Long-Covid-in-the-UK_Cambridge-Econometrics_V1.1_March2024.pdf.

[17] Cohen, J. and Y. Rodgers (2024), “Long COVID Prevalence, Disability, and Accommodations: Analysis Across Demographic Groups”, Journal of Occupational Rehabilitation, Vol. 34/2, pp. 335-349, https://doi.org/10.1007/s10926-024-10173-3.

[28] Costantino, V. et al. (2024), “The public health and economic burden of long COVID in Australia, 2022–24: a modelling study”, Medical Journal of Australia, Vol. 221/4, pp. 217-223, https://doi.org/10.5694/mja2.52400.

[24] Cutler, D. (2022a), The Costs of Long COVID, American Medical Association, https://doi.org/10.1001/jamahealthforum.2022.1809.

[25] Cutler, D. (2022b), The Economic Cost of Long COVID: An Update, Harvard Kennedy School, https://www.hks.harvard.edu/centers/mrcbg/programs/growthpolicy/economic-cost-long-covid-update-david-cutler.

[7] Davis, H. et al. (2021), “Characterizing long COVID in an international cohort: 7 months of symptoms and their impact”, eClinicalMedicine, Vol. 38, https://doi.org/10.1016/j.eclinm.2021.101019.

[18] Dennett, J. et al. (2025), “Enduring Outcomes of COVID-19 Work Absences on the US Labor Market”, JAMA Network Open, Vol. 8/10, p. e2536635, https://doi.org/10.1001/jamanetworkopen.2025.36635.

[22] Fischer, K., J. Reade and W. Schmal (2022), “What cannot be cured must be endured: The long-lasting effect of a COVID-19 infection on workplace productivity: The long-lasting effect of a COVID-19 infection on productivity”, Labour Economics, Vol. 79, https://doi.org/10.1016/j.labeco.2022.102281.

[15] Ford, N. et al. (2025), “Employment Status, Work Limitations, Cognitive Dysfunction, and Sickness Absenteeism Among US Adults With and Without Long COVID”, American Journal of Industrial Medicine, Vol. 68/10, pp. 909-919, https://doi.org/10.1002/ajim.70014.

[16] Ford, N. et al. (2023), Long COVID and Significant Activity Limitation Among Adults, by Age — United States, June 1–13, 2022, to June 7–19, 2023, https://www2.census.gov/programs-surveys/demo/technical-documentation/.

[12] Grant, A. et al. (2024), “Long COVID in healthcare workers: longitudinal mixed-methods study”, Occupational Medicine, Vol. 75/3-4, pp. 171-178, https://doi.org/10.1093/occmed/kqae113.

[31] Guillemette, Y. and D. Turner (2017), “The fiscal projection framework in long-term scenarios”, OECD Economics Department Working Papers, No. 1440, OECD Publishing, Paris, https://doi.org/10.1787/8eddfa18-en.

[21] Konishi, S. et al. (2025), “The relationship between long COVID, labor productivity, and socioeconomic losses in Japan: A cohort study”, IJID Regions, Vol. 14, https://doi.org/10.1016/j.ijregi.2024.100495.

[10] Kwon, J. et al. (2024), “Impact of Long COVID on productivity and informal caregiving”, The European Journal of Health Economics, Vol. 25, pp. 1095–1115, https://doi.org/10.1007/s10198-023-01653-z.

[32] Loayza, N. and S. Pennings (2022), The Long Term Growth Model: Fundamentals, Extensions, and Applications, World Bank Group, Washington.

[6] Łukomska, E. et al. (2025), “Healthcare Resource Utilization (HCRU) and Direct Medical Costs Associated with Long COVID or Post-COVID-19 Conditions: Findings from a Literature Review”, Journal of Market Access & Health Policy, Vol. 13/1, p. 7, https://doi.org/10.3390/jmahp13010007.

[1] Mu, Y. et al. (2024), “Healthcare utilisation of 282,080 individuals with long COVID over two years: a multiple matched control, longitudinal cohort analysis”, Journal of the Royal Society of Medicine, Vol. 117/11, pp. 369-381, https://doi.org/10.1177/01410768241288345.

[20] Naik, H. et al. (2025), “Health-related adverse work outcomes associated with post COVID-19 condition: a cross-sectional study”, BMJ Public Health, Vol. 3/1, p. e001801, https://doi.org/10.1136/bmjph-2024-001801.

[23] Naik, H. et al. (2025), “Work Productivity Loss in People Living With Long COVID Symptoms Over 2 Years From Infection”, Journal of Occupational & Environmental Medicine, Vol. 67/8, pp. 588-594, https://doi.org/10.1097/jom.0000000000003440.

[35] OECD (2025), “The prevalence and impact of Long COVID in the primary care population: Findings from the OECD PaRIS survey”, OECD Publishing, Paris, https://doi.org/10.1787/119b0e8f-en.

[36] Perlis, R. et al. (2023), “Association of Post–COVID-19 Condition Symptoms and Employment Status”, JAMA Network Open, Vol. 6/2, pp. e2256152-e2256152, https://doi.org/10.1001/jamanetworkopen.2022.56152.

[9] Reuschke, D. and D. Houston (2022), “The impact of Long COVID on the UK workforce”, Applied Economics Letters, Vol. 30/18, pp. 2510-2514, https://doi.org/10.1080/13504851.2022.2098239.

[41] RIVM (2025), Post-covid, https://www.rivm.nl/gezondheidsonderzoek-covid-19/kwartaalonderzoek-volwassenen/post-covid.

[3] Scott, A. et al. (2024), “Substantial health and economic burden of COVID-19 during the year after acute illness among US adults at high risk of severe COVID-19”, BMC Medicine, Vol. 22/1, https://doi.org/10.1186/s12916-023-03234-6.

[40] Selden, T. (2025), Sources of Health Insurance among Adults with Long COVID: Estimates from the Medical Expenditure Panel Survey, Agency for Healthcare Research and Quality.

[43] Servier, C. et al. (2023), “Trajectories of the evolution of post-COVID-19 condition, up to two years after symptoms onset”, International Journal of Infectious Diseases, Vol. 133, pp. 67-74, https://doi.org/10.1016/j.ijid.2023.05.007.

[13] Smith, P. et al. (2022), “Post COVID-19 condition and its physical, mental and social implications: protocol of a 2-year longitudinal cohort study in the Belgian adult population”, Archives of Public Health, Vol. 80/1, https://doi.org/10.1186/s13690-022-00906-2.

[45] Stelson, E. et al. (2023), “Return-to-work with long COVID: An Episodic Disability and Total Worker Health® analysis”, Social Science & Medicine, Vol. 338, p. 116336, https://doi.org/10.1016/j.socscimed.2023.116336.

[33] Subramanian, A. et al. (2022), “Symptoms and risk factors for long COVID in non-hospitalized adults”, Nature Medicine, Vol. 28/8, pp. 1706-1714, https://doi.org/10.1038/s41591-022-01909-w.

[46] Sutcliffe, C. (ed.) (2022), “Characteristics and impact of Long Covid: Findings from an online survey”, PLOS ONE, Vol. 17/3, p. e0264331, https://doi.org/10.1371/journal.pone.0264331.

[38] Tokars, J., S. Olsen and C. Reed (2018), “Seasonal Incidence of Symptomatic Influenza in the United States”, Clinical Infectious Diseases, Vol. 66/10, pp. 1511-1518, https://doi.org/10.1093/cid/cix1060.

[2] Tufts, J. et al. (2023), “The cost of primary care consultations associated with long COVID in non-hospitalised adults: a retrospective cohort study using UK primary care data”, BMC Primary Care, Vol. 24/1, https://doi.org/10.1186/s12875-023-02196-1.

[19] Waters, T. and T. Wernham (2022), Long COVID and the labour market, The Institute for Fiscal Studies, https://doi.org/10.1920/BN.IFS.2022.BN0346.

[5] Wolff Sagy, Y. et al. (2023), “Estimating the economic burden of long-Covid: the additive cost of healthcare utilisation among COVID-19 recoverees in Israel”, BMJ Global Health, Vol. 8/7, https://doi.org/10.1136/bmjgh-2023-012588.

[37] Xie, Y., T. Choi and Z. Al-Aly (2024), “Postacute Sequelae of SARS-CoV-2 Infection in the Pre-Delta, Delta, and Omicron Eras”, New England Journal of Medicine, Vol. 391/6, pp. 515-525, https://doi.org/10.1056/NEJMoa2403211.

[4] Yang, J. et al. (2025), “Quantifying all-cause healthcare resource utilization and costs of children with mild-to-moderate long COVID in France”, Journal of Medical Economics, Vol. 28/1, pp. 1002-1013, https://doi.org/10.1080/13696998.2025.2525002.

[44] Zhu, S. et al. (2024), “Modeling the burden of long COVID in California with quality adjusted life-years (QALYS)”, Scientific Reports, Vol. 14/1, p. 22663, https://doi.org/10.1038/s41598-024-73160-x.

[11] Ziauddeen, N. et al. (2023), “Impact of long COVID-19 on work: a co-produced survey”, The Lancet, Vol. 402, p. S98, https://doi.org/10.1016/s0140-6736(23)02157-8.

Notes

← 1.  A macro‑econometric model linking economic activity, labour markets and health-system spending developed by Cambridge Econometrics.

← 2.  This estimate is derived from a study in Israel with a substantial sample and the longest follow-up (up to 15 months post-infection) of severe (or hospitalised) cases.