The prevalence and impact of Long COVID in the primary care population

1. Long COVID has emerged as a serious long-term consequences of the pandemic

Long COVID (Post COVID‑19 condition) presents considerable challenges for patients, healthcare providers and policy makers, with millions of people still suffering from this condition five years after the pandemic began (Espinosa Gonzalez and Suzuki, 2024[1]). Patients with Long COVID experience a wide range of persistent symptoms that impact their physical and mental health, ability to work, care for their families, and engage in daily activities (Ely, Brown and Fineberg, 2024[2]). Many also encounter frustration and inadequate care pathways within the healthcare system, complicated by a lack of effective diagnostic tools or biomarkers leading to difficulties in clinically recognising and diagnosing Long COVID. The lack of a standardised case definition for Long COVID also negatively impacts recognition and diagnosis for patients and physicians. Although there has been increasing consensus on the case definition, implementing this in clinical practice across health systems may prove challenging (see Box 1).

This lack of recognition of Long COVID and organised care models exacerbates the already significant challenges patients face in managing their complex and disabling conditions. Planning for the management and long-term impacts of Long COVID requires reliable estimates of the prevalence of the condition, its persistence over time, costs to patients, health systems, and economies. However, this is challenging due to the epidemiological uncertainties, such as the still-unknown duration, long-term effects of, and future incidence of the condition. While some patients seem to recover over months, others suffer relapses or endure persistent symptoms for months or even years (Cai et al., 2024[3]).

Given these uncertainties, early findings paint a concerning picture. If Long COVID symptoms persist long-term as a chronic condition, its societal impact will be immense. It also presents clear equity challenges, disproportionately affecting women, essential workers, and individuals in more socio‑economically deprived areas (Subramanian et al., 2022[4]). Shedding light on the clinical, epidemiological, and economic implications of Long COVID through robust data collection and economic analysis is essential to inform effective policy responses.

Up-to-date, population-level estimates of Long COVID prevalence are required for OECD countries to budget and plan the allocation of healthcare resources. However, most countries currently lack routine surveillance for estimating country-level prevalence of Long COVID. Currently, most OECD countries are reliant on clinical coding of hospitalised patients or COVID‑19 research studies to inform estimates, which may not be generalisable to the primary care or general population.

Primary care‑based surveys such as PaRIS (see Box 2) may provide more representative estimates of the impact of Long COVID on populations. The patient perspective of living with Long COVID is important to capture. Patient-reported health outcome (PROMs) and healthcare experience measures (PREMs) can be used to inform more effective and people‑centred heath policy and care pathways. An understanding of the health and social impact on patients in terms of their reported well-being and functioning, and their experience of navigating the healthcare system, can help inform the organisation of care pathways and optimise the detection, diagnosis and management of patients living with Long COVID.

Box 1. Recognising, diagnosing and managing Long COVID in OECD countries in 2024

OECD countries are adopting consensus around a common case definition for Long COVID. A majority (64%) of OECD countries surveyed in November 2024 reported use of either the World Health Organization (WHO) or National Academies of Sciences, Engineering and Medicine (NASEM) case definition (see Figure 1). These two definitions have emerged as the consensus to define and detect Long COVID – which applies to patients presenting with new or persistent symptoms for at least two to three months, after an initial SARS‑CoV‑2 infection (COVID‑19). They provide a basis for detecting and referring suspected cases in a standardised way for healthcare systems, however the relative recency of their development (WHO in 2022 and NASEM in 2024), and gradual adoption by healthcare systems, limited the recognition of earlier cases of Long COVID prior to this, likely leading to an underestimation of Long COVID burden. Five surveyed OECD countries reported either use of a national case definition or lacking a standardised case definition depending on patient and healthcare provider.

Figure 1. Use of a standardised case definition to diagnose Long COVID

Source: OECD Survey on the Assessment of the Adoption of Systems and Interventions to Improve Diagnostic Safety, 2024.

Among OECD countries surveyed in November 2024, six (Belgium, France, Germany, Japan, Latvia and Luxembourg) reported having a national standard care pathway for referral and management of patients with suspected Long COVID (see Figure 2). Pathways typically involve organisation of care for patients to either specialist or primary care for standardised assessment to confirm or exclude Long COVID diagnosis and offer multidisciplinary management. Three countries surveyed reported a national care pathway being developed, but not yet implemented in practice. Lack of implementation of a standard care model is a barrier to timely detection and management of Long COVID and represents a challenge for patients and healthcare providers in OECD countries. For many, implementation of a care model is still in progress as of 2024 with organisation of care pathways and drafting of clinical guidelines.

Figure 2. Standard care pathways for referral and management of patients with Long COVID

Source: OECD Survey on the Assessment of the Adoption of Systems and Interventions to Improve Diagnostic Safety, 2024.

Box 2. Using the PaRIS results to shed light on Long COVID

PaRIS and recruitment of participants from primary care

The OECD Patient-Reported Indicator Surveys (PaRIS) initiative conducted a cross-sectional survey of 107 011 primary care patients aged 45 years and older, who had a primary care contact within the 6 months preceding the sampling (de Boer et al., 2022[5]; Valderas et al., 2024[6]). The PaRIS study design, sampling method of primary care practices and patients, and participant characteristics are detailed in Chapter 7, “PaRIS data and methods” in Does Healthcare Deliver?: Results from the Patient-Reported Indicator Surveys (PaRIS) (OECD, 2025[7]). Data were collected from May 2023 to January 2024. PaRIS surveyed COVID‑19 related outcomes in patients from 16 OECD countries: Australia, Belgium, Canada, Czechia, France, Greece, Iceland, Italy, Luxembourg, the Netherlands, Norway, Portugal, Slovenia, Spain, Switzerland and Wales (United Kingdom), and two non-OECD countries: Romania and Saudi Arabia. PaRIS patients from the United States were aged 65 years and older and were not asked the questions related to Long COVID. Patients were recruited from 1 816 primary care practices. This period corresponds to the fourth year of the COVID‑19 pandemic, by which time a substantial proportion of the population had contracted COVID‑19. PaRIS surveyed primary care patient reported sociodemographic status, chronic conditions, healthcare experiences, health outcomes and Long COVID symptoms. To estimate the disease burden of Long COVID in OECD populations, we used the PaRIS dataset to provide estimates of its prevalence, as well as its health and social impact.

Defining and measuring Long COVID using PaRIS

For analysis purposes, Long COVID was defined as persistence or onset of new symptoms for more than three months after initial COVID‑19, aligning with the WHO (2022)1 and NASEM (2024)2 case definitions. Of note, the outcome was measured as patient-reported persistence of symptoms, without confirmation based on medical records or diagnosis. We defined Long COVID prevalence as the percentage of participants in the total primary care population of PaRIS who reported ever having Long COVID according to the case definition. The COVID-related questions were based on those of the US Pulse household survey.3 The PaRIS patient questionnaire is available on the OECD PaRIS website.4

Firstly, crude estimates of COVID‑19 and Long COVID prevalence were calculated for the overall PaRIS survey population. Prevalence of Long COVID was calculated as the number of patients who fulfilled the case definition divided by the total primary care population of PaRIS. A multilevel model provided adjusted country-specific estimates of Long COVID prevalence and an overall prevalence estimate, with comparative intervals (CI).5 The multilevel model accounts for variation arising from different countries and primary care providers, which could influence the characteristics of the patients included in the survey from each location.

Estimates of Long COVID prevalence were calculated for the following sociodemographic variables: sex, age, income level, educational level, employment status, and presence of chronic conditions, and analysed for any statistical difference in prevalence for each variable. These prevalence estimates were standardised by age and sex using the country reference populations and included comparative intervals. Given the impact of pre‑existing chronic conditions on the risk developing of Long COVID, the crude analysis of prevalence was stratified by the number of patient chronic conditions. Participants with missing data for variables of interest were excluded from the sub-analysis.

Measuring patient reported outcomes and experiences using the PaRIS survey

Adjusted estimates of patient-reported health and social outcomes (severe fatigue, being on sick leave or unemployed, self-rated physical health and self-rated mental health) and patient-reported healthcare experience (having to repeat information on medical record and having trust in the healthcare system), according to Long COVID status (presence or absence of Long COVID), were produced using a different multilevel model. Finally, crude estimates of the prevalence of chronic conditions by Long COVID status were provided to examine for clustering of certain chronic conditions, and to profile the chronic disease pattern of patients with Long COVID.

1. World Health Organization (2022), Post COVID‑19 condition (Long COVID), https://www.who.int/europe/news-room/fact-sheets/item/post-covid-19-condition.

2. National Academies of Sciences Engineering and Medicine (2024), A Long COVID definition: A chronic, systemic disease state with profound consequences, https://nap.nationalacademies.org/read/27768/chapter/3#28.

3. United States Census Bureau (2023), Phase 3.7 Household Pulse Survey, https://www2.census.gov/programs-surveys/demo/technical-documentation/hhp/Phase_3-7_Household_Pulse_Survey_ENGLISH.pdf.

4. www.oecd.org/health/PaRIS.

5. The resulting comparative interval is equivalent to 84% confidence interval, which widens the intervals slightly beyond traditional 95% confidence limits to account for multiple comparisons, so their overlap effectively implies no statistically significant difference at the 5% error level (Goldstein and Healy, 1995[8]).