Protocol for a systematic review and meta-analysis of long influenza following the last five influenza pandemics

Abstract

Introduction

Influenza pandemics have been evaluated in terms of acute disease, while giving limited attention to long-term consequences. Renewed interest in postviral conditions following the COVID-19 pandemic highlights the need to reassess earlier influenza pandemics from a long-term perspective. This protocol describes a forthcoming systematic review and meta-analysis examining long-term health outcomes following exposure to five influenza pandemics. The aim is to assess the magnitude, consistency, mechanisms and determinants of postpandemic health outcomes.

Methods and analysis

This protocol followed the Preferred Reporting Items for Systematic Review and Meta-Analysis Protocols (PRISMA-P) 2015 checklist. The planned systematic review and meta-analysis resulting from the protocol will follow the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) 2020 checklist. Nine databases were searched using the PEO (Population, Exposure and Outcome) framework. Eligible studies include peer-reviewed quantitative observational and clinical follow-up studies reporting long-term outcomes. Two investigators will independently screen and select studies, extract data and assess quality. Study quality and risk of bias will be evaluated using the Newcastle-Ottawa Scale (NOS) and publication bias will be assessed using Egger’s and Begg’s tests. Narrative synthesis will be conducted before quantitative pooling. Fixed- and random-effects meta-analysis, meta-regression and Bayesian models will be applied to examine heterogeneity across populations, pandemics and outcomes.

Ethics and dissemination

This systematic review and meta-analysis do not require ethical approval, as it uses data from previously published studies (Head of department e-mail: elno@oslomet.no). The results will be published as a peer-reviewed article.

Systematic review registration

This systematic review protocol was registered in PROSPERO under registration number CRD420251267816.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• This systematic review and meta-analysis draw on comprehensive bibliographic databases to identify evidence on prolonged health consequences following the last five influenza pandemics.

• The review considers multiple exposure pathways including influenza infection and other pandemic-related exposures.

• The language restriction may result in an important loss of evidence, particularly for earlier pandemics.

• Heterogeneous evidence from earlier pandemics may limit the comparability across the influenza pandemics studied.

Introduction and aims

Influenza pandemics have shaped population health for centuries, yet research has largely focused on acute morbidity and mortality rather than possible long-term consequences.¹ The COVID-19 pandemic renewed scientific and public interest in persistent postviral conditions, underscoring the importance of reassessing earlier pandemics from a longer-term perspective.^{2 3} Despite the historical frequency and global reach of influenza pandemics, the nature, determinants and duration of prolonged postinfluenza outcomes—here referred to as ‘long influenza’—remain insufficiently studied.

Evidence from the 2009 influenza pandemic indicates that clinically significant impairments may persist beyond the acute phase of infection. Follow-up studies have documented reduced pulmonary function, sometimes independent of initial severity,^4–6 exertion dyspnoea & decreased diffusion capacity across the blood-gas barrier.⁵ Neurological sequelae have also been observed, particularly among hospitalised children, with residual symptoms reported up to 1 year following infection.⁷ These findings raise broader questions about whether similar long-term patterns occurred following earlier pandemics.

Historical investigations of the 1918 pandemic suggest potential association with later-life health outcomes, including elevated mortality risks and possible links to chronic disease, although results have been heterogeneous and occasionally contradictory.^8–11 Meanwhile, systematic evidence concerning the influenza pandemics of 1889, 1957 and 1968 remains comparatively sparse. Variability in diagnostic frameworks, outcome definitions and study designs further complicates interpretation across pandemics and time periods.^{3 12}

Long-term consequences of pandemics may arise through multiple mechanisms. Beyond direct biological effects of infection—such as persistent pulmonary impairment, neurological sequelae and later-life health impacts on cardiovascular and developmental outcomes from the 1918 pandemic cohorts¹³—indirect influences including social disruption, economic hardship, bereavement and public health interventions may also contribute to persistent symptoms and functional limitation. Evidence from systematic reviews and historical analyses indicates that both infection-related pathology and broader socio-economic conditions are associated with longer-term morbidity following influenza pandemics.^{1 5 14} Disentangling these pathways is essential for understanding postpandemic health dynamics and for informing preparedness planning.

This study presents a protocol for a systematic narrative review and meta-analysis of long-term health outcomes associated with the last five major influenza pandemics (1889, 1918, 1957, 1968 and 2009). The review aims to assess the magnitude, consistency, mechanisms and determinants (eg, age, sex, comorbidities, acute disease length and severity) of persistent postinfluenza outcomes while applying transparent selection criteria and standardised quality appraisal to minimise bias and improve comparability.

Methods

We will conduct a systematic review and meta-analysis. The following protocol is reported using guidance from the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P 2015)^{15 16} (see checklist in online supplemental material 1). The protocol has been registered with the International Prospective Register of Systematic Reviews (PROSPERO) with the registration number CRD42021253307. The planned systematic review and meta-analysis resulting from the protocol will be reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020 checklist).¹⁷ The review will examine long-term health outcomes associated with exposure to the five major influenza pandemics—1889, 1918, 1957, 1968 and 2009—focusing on both direct effects of infection and indirect or broader societal effects arising from pandemic conditions and context. Because no consensus definition of ‘long influenza’ exists, the review will adopt an inclusive approach, considering any health outcome measured at least several weeks after acute infection and/or indirect pandemic exposure, response and context, provided that authors describe it as a longer-term or delayed sequela.

Systematic literature search

Our final literature search on 11 November 2025 used nine medical, sociological and multi-disciplinary databases: MEDLINE (Ovid), Embase (Ovid), PsycInfo (Ovid), CINAHL (Ebsco), SocIndex (Ebsco), Academic Search Ultimate (Ebsco), ASSIA: Applied Social Sciences Index and Abstracts, Scopus (Elsevier) and Web of Science (Clarivate). Filters excluding animal studies were applied in MEDLINE and Embase.

The search strategy was developed and piloted in collaboration with librarians from 17 September to 11 November 2025. The search was conducted according to PEO (Population, Exposure and Outcome) criteria. Key PEO terms in our study were as follows: individuals or populations (P), the last five influenza pandemics (E), and long-term health outcomes (O).

Specific conditions were not included in the strategy due to the large variety of both known and possible unknown outcomes, syndromes, conditions and diseases and not least that the purpose of this review is to identify also conditions not known prior to this review.

The online supplemental material 2 provides an example of search strategy for one database.

The librarians used the application ASySD (Automated Systematic Search De-duplicator) to remove duplicates before we received the EndNote library.¹⁸

Two reviewers will screen the abstracts and full-text studies in Covidence. Expanded set of studies, including the ones identified in the reference lists, will be screened.

While our study started developing and piloting the search strategy in September 2025, we expect to have screened the abstracts and full-text studies, extracted the data and finished a narrative review and meta-analysis for journal submission at the end of 2027.

Eligibility criteria

Inclusion criteria

Studies will be eligible if they meet the following criteria:

1. Population and Exposure: individuals or populations exposed to one of the last five influenza pandemics (1889, 1918, 1957, 1968, 2009). This includes individuals exposed either through laboratory confirmed infection, clinical diagnosis, hospitalisation, Intensive Care Unit (ICU) care or death, and/or through documented pandemic responses (antivirals and vaccination can reduce the impact of acute disease and long-influenza; public health measures can reduce exposure to acute and long-term disease outcomes but can also lead to social isolation and a drop in social integration, unemployment and postponed treatments of other health conditions), indirect pandemic exposure (eg, fear of infection, bereavement); 1914–1918 World War One (WWI) or configurative exposure mechanisms (eg, 1914–1918 WWI disruptions, proximity to a military camp in 1918 etc; financial crisis in 2008–2009).

2. Outcome: quantitative measurement for long-term health outcomes occurring beyond the acute phase—including but not limited to—pulmonary function deficits, neurological symptoms, mental health conditions, cardiovascular outcomes, functional impairment.

3. Study design: quantitative observational studies (cohort, register-based, case-control, cross-sectional) and clinical follow-up studies.

4. Publication: peer-reviewed full-text articles.

5. Language: due to the language proficiency of the research team and the feasibility of the screening and data extraction process, we will include English or Scandinavian language papers. However, as a substantial number of articles published in other languages probably have their titles and/or abstracts available in English (either through the information provided to the bibliographic databases by the authors or through automated/machine translation), we intend to retain potentially relevant records in other languages during the title/abstract screening process in cases where sufficient information is available to assess their relevance for our study. When such records are considered for the full-text review, we will use translation support to determine their eligibility and extract relevant information.

Exclusion criteria

Studies will be excluded if they fall into one or more of the following categories:

1. Reviews, editorials, theoretical papers or policy briefs without original empirical data.

2. Studies focusing solely on seasonal influenza without distinguishing pandemic periods.

3. Studies combining seasonal and pandemic influenza without separate reporting.

4. Studies on diseases other than influenza (eg, COVID-19).

5. Case reports, case series or qualitative studies.

6. Studies examining antiviral or vaccine uptake, vaccination attitudes or Non-pharmaceutical interventions (NPI) compliance without reporting long-term outcomes.

7. Studies focusing exclusively on in utero exposure to influenza pandemics (except those focused on mental conditions, eg, schizophrenia).

Data selection and extraction

Data will be extracted independently by two reviewers using a structured extraction form. Any discrepancies will be solved through discussion. The following information will be collected from each included study:

1. Article info

   1. Authors.

   2. Journal.

   3. Year.

2. Study setting

   1. Country/region.

   2. Pandemic period covered (1889, 1918, 1957, 1968, 2009).

   3. Historical and contemporary context.

   4. Data source.

3. Population characteristics

   1. Age, sex, comorbidities, disease length and severity, civilian/military status and sample inclusion criteria.

4. Exposure definition

   1. Laboratory-confirmed infection.

   2. Clinical diagnosis.

   3. Hospitalisation/ICU care/death.

   4. Indirect exposure (eg, fear of infection, bereavement due to the pandemic).

   5. Exposure to configurative conditions (eg, 1914–1918 WWI and proximity to military camps; financial crisis in 2008–2009).

   6. Exposure to pandemic responses (eg, antiviral and vaccine programmes, public health interventions).

5. Outcome definitions

   1. Pulmonary outcomes (eg, diffusion capacity impairment, reduced vital capacity).

   2. Neurological outcomes (eg, seizures, cognitive deficits, encephalopathy).

   3. Cardiac outcomes.

   4. Mental health outcomes (eg, psychosis, suicide (attempts), long-term distress).

   5. Functional outcomes (eg, balance, walking, grip strength, hearing).

6. Comparator groups

   1. Healthy population.

   2. Uninfected population.

   3. Seasonal influenza.

   4. Non-ICU hospitalisations, etc.

7. Study methodology

   1. Study design: statistical models (eg, logistic regression, Cox models, Bayesian models).

8. Effect estimates: point estimates, CIs, standard errors and measures of precision.

9. Mechanisms explored: direct biological effects, indirect social impacts, configurative elements or pandemic responses.

Data on all quantitative results will be stored in a separate spreadsheet for use in meta-analysis.

Data synthesis

We plan to conduct narrative synthesis and meta-analysis. Studies will be grouped by pandemic period, exposure type and outcome. Indirect social impacts, pandemic responses and configurative elements will also be documented.

Study quality and risk of bias will be evaluated using the Newcastle-Ottawa Scale (NOS).¹⁹ If this tool proves insufficient for certain study designs or historical datasets, an alternative quality assessment tool will be chosen (eg, Risk Of Bias In Non-Randomized Studies - of Interventions (ROBINS-I)). Two reviewers will assess each study independently, and discrepancies will be solved through consensus.

Meta-analysis

Three complementary meta-analytic approaches will be used where data are sufficiently comparable:

1. Fixed- and random-effects meta-analysis

   These models will be applied to sets of studies examining similar outcomes within the same pandemic period. Random-effects models will be preferred due to expected heterogeneity in study design, population characteristics and exposure measurement.

2. Meta regression

   Meta-regression will be conducted on studies with harmonised outcome measures, evaluating study-level covariates such as age and sex distribution, comorbidity burden, pandemic year, exposure type and analytical methods to explore sources of heterogeneity.

3. Bayesian meta-analysis

   A Bayesian hierarchical model will synthesise all available effect estimates across pandemics, outcomes and exposure types. This approach will allow estimation of underlying susceptibility patterns related to demographic and clinical factors (age, sex, comorbidities), mechanisms (direct, indirect, configurative, interventions), pandemic cases (eg, exposed, lab-confirmed infections and hospitalisation, ICU), and controls (eg, healthy population, Influenza-like Illness (ILI) for seasonal influenza, non-ICU hospitalisations).

Results will be presented visually using forest plots. Publication bias will be assessed using Egger’s and Begg’s tests implemented via the regtest and ranktest functions in the metafor package in R.¹⁶

Discussion

This study is expected to provide the first comprehensive synthesis of evidence on long-term health consequences following influenza pandemics across more than a century. By integrating findings from the pandemics of 1889, 1918, 1957, 1968 and 2009, the review aims to clarify whether long influenza represents a recurring phenomenon rather than an isolated observation. The results may help reposition influenza pandemics within the broader framework of postviral illness research, which has historically focused more on acute morbidity and mortality and less on prolonged morbidity, functional limitations and mortality.^{2 3 12}

An important feature of this review is the evaluation of different pathways that may underlie long-term outcomes. In addition to direct effects of infection, the study considers indirect influences linked to social disruption, public health responses and economic stressors. It is essential to differentiate between these mechanisms for interpretation, as prolonged symptoms may reflect interacting biological and contextual factors.^{3 12}

The findings may inform pandemic preparedness and health system planning. Consistent evidence of long-term sequelae would support incorporating longer follow-up, rehabilitation strategies and monitoring of chronic outcomes into response frameworks. Evidence that non-infectious mechanisms contribute meaningfully to long-term outcomes would further support policies aimed at mitigating social and economic disruption during pandemics. Conversely, if long-term effects appear limited, this will highlight the need for more precise definitions, improved study designs and cautious interpretation of historical data.^4–6

The review is expected to expose important knowledge gaps. Previous research has concentrated heavily on the 1918 and 2009 pandemics, while evidence from mid-20th-century pandemics and the 1889 pandemic remains sparse.^{1 14 20} Moreover, studies examining the long-term consequences of the 1918 pandemic have produced divergent findings. Some investigations report elevated later-life mortality and cardiovascular risks among exposed cohorts,^{8 9 21 22} whereas others find no consistent effects on survival and later-life mortality.^{10 11}

Collectively, these challenges underscore the need for clearer conceptual frameworks and greater methodological standardisation in the study of long-term health consequences of respiratory virus pandemics. Addressing these gaps may strengthen future research on the long-term health consequences of pandemic influenza and related respiratory viruses.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.