The COVID-19 pandemic has substantially reshaped the global health burden by producing not only acute morbidity and mortality but also a spectrum of persistent postinfectious sequelae that continue to affect large numbers of survivors [1-3]. One of the most clinically and socioeconomically consequential of these sequelae is Long COVID, frequently termed postacute sequelae of SARS-CoV-2 infection (PASC), a heterogeneous syndrome comprising persistent, relapsing, or newly emergent symptoms after the acute phase of infection [4-6]. Population and cohort analyses show appreciable heterogeneity in estimates of prevalence across definitions, follow-up intervals, and dominant variants, with many studies reporting persistent symptoms in the approximate 10%-30% range among infected persons depending on case definition and follow-up timing [7-9]. Typical symptom clusters include debilitating fatigue, dyspnea/exertional intolerance, and cognitive dysfunction (often described as “brain fog”), and these complaints occur in individuals across the spectrum of acute disease severity, including those who had mild acute illness [2,8,9].

Mechanistic concepts for PASC are multifactorial and emphasize an interplay of viral, immune, and host-environment factors, including potential viral persistence or residual antigen, dysregulated or protracted immune activation, chronic systemic and tissue-specific inflammation, perturbations of metabolic pathways, and, in some patients, autoimmune sequelae [5,10,11]. Proteomic and metabolomic studies and longitudinal immune profiling have reported persistent alterations in circulating proteins and immune mediators—including reported associations with proinflammatory cytokines such as interleukin-6 (IL-6) and tumor necrosis factor-α—although results vary across cohorts and time points [10,12,13]. At the same time, psychosocial determinants (eg, preexisting psychiatric morbidity, chronic stress, and social isolation) and baseline comorbidity burden (eg, obesity, cardiopulmonary disease, and chronic kidney disease) have been associated with greater acute severity and may influence recovery trajectories, symptom perception, and care seeking after SARS-CoV-2 infection [14-17].

Despite rapid expansion of descriptive and mechanistic work, the extant literature remains fragmented: many investigations target single domains (immunologic, imaging, symptom inventories, or psychosocial measures) or selected subpopulations (hospitalized cases, clinic-referred patients, or pediatric cohorts), limiting integrative inference and generalizability [5,18]. Large electronic health record–driven and population-based studies have mapped cooccurrence patterns and identified candidate risk factors, but frequently lack prospectively collected multimodal biomarker and psychosocial data required to link biological signals to clinically meaningful longitudinal trajectories [6,19]. Neurocognitive sequelae exemplify this gap: observational and translational work implicates persistent neuroinflammation and systemic immune perturbations as plausible contributors to post–COVID-19 cognitive complaints and to accelerated decline among susceptible individuals, but definitive, integrative longitudinal evidence is limited [20-22]. Taken together, these observations emphasize the need for prospective, multimodal cohorts that combine standardized clinical phenotyping, serial biosampling, and structured psychosocial measurement to permit hypothesis-driven biomarker discovery and etiologic modeling of PASC [3,23].

Demographic factors and baseline comorbidities modulate both acute disease course and postacute outcomes: older age and multimorbidity are consistently associated with a higher risk of persistent symptoms and adverse long-term outcomes in many cohort studies, and prior cognitive impairment has been identified as a vulnerability factor for prolonged neuropsychiatric sequelae [9,24-26]. Obesity and chronic respiratory disease have been linked to greater acute severity and higher likelihood of prolonged symptoms, and these associations have implications for targeted surveillance and rehabilitation strategies [17,27,28]. Collectively, these epidemiologic and mechanistic observations underscore the need for health systems to develop evidence-based approaches to identify at-risk populations early and to tailor prevention, monitoring, and rehabilitation resources accordingly [3,23,29].

In summary, although the body of research on Long COVID has expanded quickly, there remains an urgent requirement for coordinated, integrative studies that synthesize clinical phenotyping, longitudinal immunology, objective physiologic assessment, and psychosocial characterization, with the ultimate goal of deriving reproducible biomarkers, mechanism-informed subtypes, and actionable interventions for PASC [3,5,6,12,23].

We hypothesize that antecedent and periacute factors (eg, sociodemographic, preinfection comorbidities, and acute COVID-19 severity indicators) are associated with Long COVID status, and that baseline postacute measures collected at T0 (eg, patient-reported outcome measures and biochemical/immunological markers) primarily differentiate Long COVID from recovered controls as concurrent phenotypic correlates rather than implying causality. By comparing individuals with Long COVID to carefully matched recovered controls and repeating standardized assessments longitudinally, we aim to identify differential factors across clinical, immunological, and lifestyle domains. These findings are expected to support biomarker discovery, mechanism-informed subtyping, and clinically actionable risk stratification in future studies.

This study is a longitudinal matched case-control study design combining a retrospective cohort with a prospective cohort. The retrospective cohort comprises individuals diagnosed during the early stages of the pandemic and provides long-term follow-up. In parallel, the prospective cohort will enroll newly identified COVID-19 cases, with follow-up assessments conducted after the baseline evaluation. This combined approach allows examination of symptom persistence for up to 3 years in the retrospective cohort while also enabling real-time tracking of changes over time in the prospectively enrolled participants. The study is registered in the International Standard Randomized Controlled Trial Number (ISRCTN) Registry (ISRCTN27312680), and reporting will adhere to the STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines.

The study will be conducted in Aragón, Spain, leveraging the region’s established primary care network and health care infrastructure. Eligible participants will be adults (≥18 years of age) with a confirmed diagnosis of COVID-19. Participants will be identified through three complementary sources: (1) the Long COVID Aragón Patient Association (patients already registered with the association), (2) the Aragón Health Service database (individuals identified from electronic health records), and (3) primary care consultations (direct referrals from general practitioners and other health care professionals).

The Long COVID Aragón Patient Association is a regional, patient-led organization based in Aragón, Spain, that provides peer support, advocacy, and information for individuals experiencing persistent symptoms following SARS-CoV-2 infection. The association was established in 2020 in response to the emergence of postacute sequelae of COVID-19 and maintains a secure registry of members who voluntarily enroll and provide contact information and basic clinical descriptors (eg, date of acute infection, persistence and type of symptoms, and care pathways). Members who indicate willingness to be contacted for research constitute an important recruitment frame for this study, enabling identification of community-dwelling individuals with persistent symptoms who may not be fully captured through health care databases alone. For eligible individuals recruited through the association, study participation will proceed via the same standardized eligibility assessment and consent procedures used for all recruitment sources, and (where applicable and permitted) self-reported information will be verified or complemented using clinical records.

Participants will be included if they have confirmed SARS-CoV-2 infection by polymerase chain reaction, antigen testing, or serology, and are aged ≥18 years. For the Long COVID group, eligibility will require persistence of symptoms for ≥12 weeks after infection onset, consistent with the World Health Organization (WHO) definition. For the control group, complete recovery will be defined as the absence of persistent or new symptoms after COVID-19. Individuals will be excluded if they refuse to participate, report preexisting symptoms similar to Long COVID prior to SARS-CoV-2 infection, or present severe psychiatric disorders or cognitive impairment that preclude informed participation.

Because the primary aim of this study is hypothesis-generating—that is, to identify differential factors associated with Long COVID status rather than to test a single prespecified end point—sample size was chosen to provide adequate power to detect moderate associations between candidate predictors and case status within a matched case-control framework. Power was evaluated using the standard normal approximation for comparing 2 independent proportions (Fleiss et al [30]), with a 2-sided significance level of α=.05 and 90% power. Under a conservative scenario in which a candidate factor has an expected prevalence of 20% among controls, we considered an association of odds ratio 2.3 to be clinically meaningful; this corresponds to an expected prevalence of approximately 36.5% among cases. Under these assumptions, the required analyzable sample size is approximately 154 participants per group, rounded to 155 per group. To accommodate incomplete matching, potential attrition between T0 and T1, and planned multivariable and subgroup analyses, the overall recruitment target will be increased to approximately 400 participants [30].

Recovered controls will be individually matched 1:1 to Long COVID cases without replacement to improve baseline comparability. Matching will require exact matching on sex/gender, with flexibility of ±3 years for age and ±30 days for date of acute COVID-19 diagnosis. When multiple eligible controls are available for a given case, the selected control will be the one minimizing the absolute age difference; if ties remain, the absolute diagnosis date difference will be minimized, with random selection used only if ties persist. Each matched set will be assigned a unique pair identifier (pair_id) and retained in the analytical dataset; participants who cannot be matched within the prespecified calipers will be retained in the study database and considered in prespecified sensitivity analyses.

After written informed consent is obtained, each participant will be assigned a unique internal study code to ensure confidentiality. The code key linking personal identifiers to the study code will be accessible only to the principal investigator and will be stored separately from study data, in accordance with the General Data Protection Regulation (GDPR) and applicable Spanish data protection legislation. All study information will be recorded and managed using secure electronic data capture procedures; self-administered questionnaires will be completed via REDCap (Research Electronic Data Capture; Vanderbilt University) software [31]. Study datasets will be deidentified/anonymized for analysis and will be stored securely, with participants linked only through the assigned study code.

Data collection will be organized into 2 study appointments. During the first appointment, fasting blood samples will be collected for routine laboratory analyses, and participants will sign informed consent covering both study participation and biobank sample storage. Participants will also receive standardized instructions for completing the self-administered questionnaires online through REDCap [31]. The second appointment will include completion of exploratory tests and the collection of any additional clinical or sociodemographic data not captured at the first visit. Standardized procedures will be used at each visit to minimize measurement variability and ensure comparability across participants and across time points. The participant flow and assessment schedule are summarized in Figure 1.

The study was approved by the Clinical Research Ethics Committee of Aragón, Spain (Comité de Ética de la Investigación Clínica de Aragón; PI21/278) and will be conducted in accordance with the Declaration of Helsinki and applicable local regulations. The collection and processing of personal data will comply with the GDPR (European Union) 2016/679 and the Spanish Organic Law 3/2018 of December 5 on the protection of personal data and guarantee of digital rights.

Biological samples and associated data will be managed through the Biobank of the Aragón Health System (National Biobank Register B.0000873; PT20/00112), integrated within the Instituto de Salud Carlos III (ISCIII) Biobanks and Biomodels Platform. Samples and data will be handled according to approved standard operating procedures and under the oversight and authorization of the relevant ethics and scientific committees and will be used exclusively for the purposes described in the approved protocol.

All participants will provide written informed consent prior to any study procedures, including specific consent for biobanking where applicable. Signed consent forms will be securely archived by the principal investigator. No participant compensation is planned due to limited funding.

Participants will be assigned a unique internal study code at enrollment. Direct identifiers will be stored separately from research data, and the reidentification key linking identifiers to study codes will be accessible only to the principal investigator and stored in a restricted, encrypted location. Analytical datasets will be deidentified (coded) and will not include direct identifiers. Data will be collected and managed using secure electronic data capture systems with role-based access controls and audit trails and stored on secure institutional servers with restricted access. Data transfers, if required, will occur only via encrypted channels. Study results will be reported in aggregate form to minimize the risk of reidentification.

ARALongCOV is a registered observational study (ISRCTN27312680) with ethics approval from the Clinical Research Ethics Committee of Aragón (PI21/278). The study is conducted under GDPR and applicable Spanish data protection legislation, with biospecimens and associated data managed through the Biobank of the Aragón Health System (PT20/00112) following approved standard operating procedures. Funding was provided by the ISCIII through project PI22/01070 (cofounded by the European Union) for the period 2023-2027.

The study combines (1) a retrospective cohort providing long-term follow-up (enabling evaluation of symptom persistence for up to 3 years) and (2) a prospective cohort enrolling newly identified Long COVID cases with follow-up assessments after baseline. Baseline assessments (T0) were initiated in late 2022/early 2023, and follow-up assessments (T1) are scheduled for late 2025/early 2026 for the earliest enrolled participants. Study assessments are conducted at 2 time points (baseline [T0] and follow-up [T1]) using the same instruments and standardized procedures to support longitudinal comparisons. Data collection is organized into 2 appointments per assessment wave, including fasting blood sampling for routine laboratory analyses and completion of online questionnaires via REDCap.

The planned recruitment target is approximately 400 participants, exceeding the minimum analyzable sample size of 155 per group derived from power considerations for detecting moderate associations in the matched case-control framework. As of February 2026, a total of 200 participants have been enrolled (n=100 Long COVID; n=100 recovered controls) and have completed T0 assessments; T1 assessments are scheduled for late 2025/early 2026.

Primary analyses will compare matched Long COVID and recovered participants using methods that account for the paired design (eg, conditional logistic regression with pair strata) and will evaluate longitudinal change using mixed-effects models including group, time, and group × time interaction. The first manuscripts reporting baseline comparisons and longitudinal outcomes are planned for submission in the second half of 2026, followed by mechanistic/biomarker and predictive modeling outputs.

In this protocol, we describe a longitudinal matched case-control study design that combines retrospective and prospective data collection to investigate the multifaceted determinants of Long COVID in adult populations; such mixed longitudinal approaches have proven useful for characterizing recovery trajectories and for linking clinical phenotypes with immune and proteomic signatures in prior cohort work [3,48,49]. Our design explicitly aims to overcome limitations of single-dimension studies by integrating clinical, functional, and laboratory domains within a single, longitudinal framework to delineate temporal relationships among symptoms, objective physiologic measures, and circulating biomarkers and thereby to identify candidate mechanistic pathways and potential intervention targets [10,14,18].

This is a longitudinal matched case-control (1:1 matching) design. The prospective 3-year follow-up combined with retrospective characterization permits within-participant trajectory analysis and between-group comparisons that are less susceptible to the inferential constraints of purely cross-sectional or clinic-based samples [50].

The multimodal phenotyping and biospecimen through systematic collection of validated patient-reported instruments, functional measures, and comprehensive biomarker panels enhance the potential to identify biologically coherent PASC subtypes and to relate subjective symptoms to measurable biological perturbations [10,14,49,51].

In particular, including a matched recovered control group reduces confounding by age, sex, and calendar time (including variant and vaccination epoch) and helps distinguish sequelae specifically attributable to prior SARS-CoV-2 infection from those related to general post-illness or secular effects [5,18,24].

The use of machine learning and multivariable modeling (eg, random forests for variable importance and receiver operating characteristic analyses for predictive performance) enables exploration of high-dimensional interactions [52].

Despite these strengths, several important limitations merit emphasis. First, the targeted sample size of roughly 400 participants may limit power to detect associations for low-prevalence outcomes and to perform highly stratified subgroup analyses or fully corrected high-dimensional biomarker discovery without external validation, a limitation recognized in therapeutic and mechanistic PASC research [3,23]. Second, geographic confinement to a single region (Aragón, Spain) may reduce external validity because regional differences in variant circulation, vaccination patterns, sociodemographic composition, and health care delivery can influence both exposure and postacute outcomes; this limitation underscores the rationale for multicenter and multinational replication [29]. Third, recruitment through patient associations, health-service registries, and clinical referrals risks selection bias because individuals with persistent symptoms or heightened health care engagement may be overrepresented, an issue documented across online and clinic-recruited PASC cohorts [53]. Fourth, retrospective self-report carries the potential for recall and misclassification bias, and longitudinal attrition can induce informative censoring if participants who drop out differ systematically from those retained, both of which are well-recognized challenges in longitudinal PASC investigations [4,48,50,54].

This protocol serves as a foundation for broader etiologic and interventional efforts. Key priorities for subsequent work include (1) expansion to multicenter/multinational networks to improve power and generalizability and to permit pooled analyses across variable systems and variant epochs [29], (2) harmonization of phenotyping and biospecimen protocols across cohorts to enable external validation and meta-analytic biomarker discovery [3,12,19], (3) incorporation of objective physiologic and imaging modalities (eg, cardiac and pulmonary imaging, cardiopulmonary exercise testing, and ambulatory wearable sensors) to corroborate and contextualize self-reported symptoms and to identify organ-level dysfunctions amenable to targeted therapies [55], and (4) prolonged longitudinal follow-up beyond 1 year to capture late-emerging sequelae and to quantify long-term risks of incident cardiometabolic or autoimmune conditions that have been reported in some cohorts [13,56].

Systematic identification of clinical, biological, and psychosocial determinants of Long COVID has immediate implications for clinical care and public health. Clinically, evidence of differential risk and mechanism-specific subtypes can inform stratified follow-up algorithms, personalized rehabilitation plans, and selection criteria for interventional trials [3,57]. From the public health and systems perspective, robust longitudinal data are necessary to develop standardized case definitions, to guide resource allocation for post–COVID-19 care pathways and rehabilitation services, and to underpin evidence-based policy and clinical guidance [23,29].

This longitudinal matched case-control study, with mixed retrospective-prospective protocol, is designed to generate multimodal, longitudinal evidence on clinical, biological, and psychosocial correlates of Long COVID, leveraging matched controls, standardized phenotyping, and modern analytic methods to identify reproducible markers of symptom persistence and potential targets for intervention [3-5,18,23,52]. The study acknowledges limitations related to sample size, geographic scope, and selection or information biases and therefore positions its findings as a basis for validation in larger, multicenter consortia and for informing longer-term follow-up studies and therapeutic trials [29]. Ultimately, integrated longitudinal investigations of the type described here are essential to reduce the long-term burden of SARS-CoV-2 infection through improved prevention, diagnosis, and management of PASC [3,5,6,58].