Prospective chest CT cohort with opportunistic multidimensional phenotype capture in routine care at a tertiary hospital in western China: Protocol

Abstract

Introduction

Routine chest CT contains clinically relevant multidomain information beyond the immediate examination indication, but translation into routine care requires more than identifying individual biomarkers. This protocol describes a prospective observational cohort designed to evaluate whether technical assessability, standardised multidomain research phenotyping, routine radiology reporting and documented care can be linked reproducibly within routine clinical chest CT at a tertiary hospital in western China.

Methods and analysis

This is a single-centre, prospective, consecutive-enrolment observational imaging cohort embedded in routine outpatient and inpatient chest CT practice rather than a dedicated screening programme. The study is organised around a scan-to-phenotype-to-report-to-care framework while retaining a phase 1 feasibility- and implementation-focused primary objective. The initial cohort target is approximately 3000 prospectively consented adults. At the time of this revision, formal recruitment has not commenced and no participant-level study data have been collected or analysed; recruitment will begin after funding and operational preparations are completed and any required ethics notification or amendment has been addressed. The core clinical-reporting layer includes pulmonary nodules, emphysema, interstitial lung abnormalities, coronary artery calcification and pulmonary artery/ascending aorta metrics. The extended measurement layer includes vertebral attenuation, vertebral deformity, thoracic muscle metrics and liver attenuation-based variables when technically assessable. Research phenotyping will be completed independently before routine-report abstraction. Primary phase 1 analyses will assess screening and enrolment flow, baseline data completeness, core phenotype assessability, report-quality abstraction completion and availability of within-system 6-month follow-up data. Secondary analyses will describe phenotype frequencies and distributions, cross-domain co-occurrence, report-quality indicators and early care-pathway documentation using prespecified operational definitions and variable-specific denominators. Descriptive estimates will be reported with 95% confidence intervals where appropriate. Follow-up will occur at 6 months, 12 months and annually thereafter using routine hospital systems. No additional research imaging is planned, and repeat chest CT is not mandated by the protocol.

Ethics and dissemination

The study was approved by the Medical Ethics Committee of Dazhou Central Hospital, Dazhou, China (approval no. 2026审(084)号; approved 27 April 2026), and written informed consent will be obtained before cohort enrolment and research data abstraction. Potentially actionable research-identified findings will undergo senior radiologist confirmation and, when confirmed and not already adequately addressed in routine care, will be communicated through the institutional clinical pathway. Planned outputs include a cohort profile or baseline resource paper and subsequent phenotype-specific analyses.

Trial registration

Not applicable. This is an observational cohort protocol rather than a clinical trial.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• A distinguishing feature is participant-level linkage of technical assessability, standardised multidomain research phenotyping, independent routine radiology report abstraction and documented care pathways within a single prospective routine-care chest CT cohort.

• The cohort uses prespecified variable dictionaries, case report forms (CRFs) and standard operating procedures (SOPs), and explicitly records routine acquisition metadata and phenotype-specific assessability so that technical non-assessability is not conflated with phenotype absence.

• Phase 1 is implementation-focused and incorporates consecutive screening logs, prespecified implementation benchmarks and transparent denominator conventions for feasibility and ascertainment assessment.

• Phase 1 is single-centre, so the generalisability of the findings will require validation in other settings.

• Consent-dependent enrolment, heterogeneous routine-care acquisition, contrast-related measurement constraints and reliance on within-system follow-up may introduce selection effects, variable-specific non-assessability and incomplete off-site outcome ascertainment; phenotype frequencies will therefore be interpreted within the enrolled cohort rather than as population prevalence estimates.

Introduction

Routine chest CT contains clinically relevant information beyond the immediate examination indication. Established opportunistic biomarkers include pulmonary, cardiovascular, skeletal and body-composition features that can be evaluated from images already acquired for clinical care.^{1 2}

Much of the existing evidence has been generated in dedicated research or large imaging cohorts. SubPopulations and InteRmediate Outcome Measures In COPD Study (SPIROMICS) uses harmonised, protocol-driven CT for deep pulmonary phenotyping, while the UK Biobank imaging enhancement provides highly standardised multimodal imaging at scale.^{3 4} These resources demonstrate the value of rigorous phenotyping, but they do not directly test how a multidomain framework performs when embedded prospectively in heterogeneous routine-care chest CT.

The implementation challenge is therefore not simply whether additional CT phenotypes can be measured, but whether clinically relevant multidomain information can be recovered reproducibly from heterogeneous routine acquisitions, represented consistently in research data, communicated appropriately in clinical reports and linked to documented subsequent care. Contemporary opportunistic-screening guidance emphasises workflow integration, technical heterogeneity, reporting pathways and clinical action as important implementation considerations.^{5 6}

Against this background, we developed a prospective observational chest CT cohort at Dazhou Central Hospital, a tertiary hospital in western China. The primary phase 1 objective remains feasibility and implementation: to establish and operationalise a routine-care cohort capable of standardised baseline multidomain phenotype capture, independent report-quality abstraction and within-system follow-up ascertainment. The distinctive contribution is not a new single CT biomarker, but a participant-level scan-to-phenotype-to-report-to-care framework that prospectively separates four linked stages: technical assessability, standardised research phenotyping, routine radiology reporting and documented care. This design allows technical non-assessability, phenotype presence, report communication and downstream-care documentation to be examined as distinct components of the same information pathway. Secondary objectives are to describe phenotype frequencies and cross-domain cooccurrence among enrolled participants, characterise reporting indicators and summarise early care-pathway documentation. The cohort is intended to provide an auditable resource for subsequent phenotype-specific studies and later evaluation of whether the same framework is transportable beyond a single centre.

Methods and analysis

Study design

This is a single-centre, prospective, consecutive-enrolment observational imaging cohort embedded in routine clinical chest CT practice rather than a dedicated screening programme. Phase 1 is implementation-focused and is intended to establish a reproducible framework for multidomain phenotype capture rather than to evaluate screening effectiveness. A staged prelaunch calibration phase will refine eligibility procedures, field definitions, case report form (CRF) usability, measurement reproducibility and practical workflow implementation. The protocol and manuscript structure follow relevant observational reporting principles, including Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) and REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) where applicable.^{7 8} An overview of the phase 1 study design and workflow is shown in figure 1.

Figure 1. Study design and workflow This flow diagram summarises the phase 1 cohort workflow. Adults undergoing routine outpatient or inpatient chest CT are screened for eligibility and provide written informed consent before enrolment. Baseline clinical and examination metadata are then captured, followed by standardised research image review and multidomain phenotyping. The original routine radiology report is abstracted independently after research phenotyping. Longitudinal within-system follow-up is planned at 6 months, 12 months and annually thereafter, with process, clinical and imaging outcomes recorded from routine-care data sources. No additional research CT or protocol-mandated repeat imaging is required.

Study setting

The study will be conducted at Dazhou Central Hospital, a tertiary hospital in western China. Phase 1 is restricted to the study centre so that the cohort framework, imaging measurement rules, data dictionary, CRFs, standard operating procedures (SOPs) and follow-up procedures can be implemented and evaluated under a common governance structure before any wider expansion.

Objectives

The primary phase 1 objective is to establish and operationalise a feasible prospective chest CT cohort embedded in routine care, with implementation-focused evaluation of standardised baseline multidomain phenotype capture, report-quality abstraction and within-system follow-up ascertainment.

Secondary phase 1 objectives are: (1) to describe the baseline distribution of prespecified core and extended chest CT phenotypes; (2) to characterise cross-domain co-occurrence of pulmonary, cardiovascular, musculoskeletal and body-composition phenotypes; (3) to characterise baseline report-quality indicators under existing local reporting practice and (4) to describe early care-pathway documentation and selected process outcomes within routine hospital systems.

Exploratory objectives are to establish a standardised cohort resource for future phenotype-specific analyses and possible multicentre or regional expansion, while recognising that confirmatory clinical-effectiveness questions are beyond the scope of the initial phase 1 implementation.

Conceptual framework

The cohort is organised around a scan-to-phenotype-to-report-to-care information pathway (figure 2). At participant level, routine chest CT first yields a phenotype-specific technical assessability state; assessable examinations then undergo standardised research phenotyping; the original routine radiology report is abstracted independently after phenotyping and longitudinal records are used to document subsequent care. These stages are analytically distinct: technical non-assessability is not phenotype absence, phenotype presence does not by itself establish that routine reporting is expected, non-mention outside a prespecified report-eligible subset is not automatically a reporting error and care-pathway analyses are descriptive rather than causal. Phase 1 feasibility outcomes quantify whether this linked pathway can be implemented reproducibly within ordinary clinical imaging rather than testing the effectiveness of a screening intervention.

Figure 2. Scan-to-phenotype-to-report-to-care framework. The conceptual framework links routine clinically indicated chest CT to four prospectively recorded information stages: phenotype-specific technical assessability, standardised multidomain research phenotyping, independent routine radiology report abstraction and documented care pathways. The original core clinical-reporting and extended measurement phenotype layers are retained, together with the clinical/metadata and follow-up/outcome layers, but are organised around this directional information pathway. Technical non-assessability is distinguished from phenotype absence, phenotype presence does not by itself establish a routine reporting requirement and reporting-to-care analyses are descriptive rather than causal. No study-mandated CT or repeat imaging is required.

Participants and recruitment

Participants will be screened sequentially from adults undergoing routine outpatient or inpatient chest CT at Dazhou Central Hospital. In this study, opportunistic phenotype capture refers to systematic recording of prespecified findings visible on routine-care chest CT, irrespective of the original clinical indication, without additional research imaging. For each enrolled participant, the first eligible chest CT linked to written study consent will define the baseline imaging record; subsequent routine chest CT examinations will be treated as follow-up rather than new enrolments. Formal recruitment has not yet commenced. The initial target remains approximately 3000 prospectively consented participants.

Potentially eligible individuals will be identified from routine scheduling lists or same-day attendance pathways, depending on local workflow and clinical appropriateness. Authorised study personnel will verify eligibility, provide study information and obtain written informed consent before research data abstraction and cohort registration. The approved protocol provides for continuous screening of routine chest CT candidates and identifies refusal to participate as a non-enrolment criterion. Accordingly, for individuals who do not enrol, the study will retain only aggregate process-level screening information needed to document recruitment: counts screened, eligible or ineligible, approached, consented, enrolled, deferred and declined, together with broad non-enrolment categories corresponding to eligibility, deferral or refusal status. No research phenotype abstraction, detailed clinical covariate dataset or participant-level research dataset will be created for non-consenters.

Because enrolment requires prospective written informed consent, selection cannot be fully characterised using participant-level data from non-consenters. The magnitude of attrition across the recruitment funnel can be described from aggregate screening counts and broad non-enrolment categories, but the direction of any residual selection effect cannot be estimated empirically from non-consenter phenotypes or clinical characteristics. Phenotype estimates will therefore be reported as frequencies and distributions among enrolled, consented participants using variable-specific assessable denominators, rather than as population prevalence estimates for all chest CT examinations at the study centre.

Eligibility criteria

Inclusion criteria are: (1) age 18 years or older; (2) routine outpatient or inpatient chest CT performed at Dazhou Central Hospital; (3) basic image quality sufficient for principal study review and measurement; (4) basic demographic and clinical information available through the approved data pathway and (5) written informed consent from the participant or legally authorised representative. Exclusion criteria are: (1) age younger than 18 years; (2) non-chest CT examinations or scan coverage clearly insufficient for the principal imaging assessment; (3) severe artefact, missing reconstruction data or inability to evaluate key thoracic structures; (4) inability to complete the study identifier linkage or link the examination to the required clinical record and (5) refusal to participate. After cohort entry, phenotype-specific assessability will be coded separately so that an unavailable extended measurement does not automatically exclude an otherwise eligible participant from analyses of other assessable domains.

Handling of major baseline clinical conditions

Participants with known malignancy or other major baseline clinical conditions will not be excluded solely on that basis unless they fail the prespecified eligibility criteria. These conditions remain part of the intended routine-care case mix but may influence phenotype frequencies, report content, downstream management and follow-up intensity. They will therefore be recorded as prespecified baseline stratification variables where available.

Handling of source-population heterogeneity

The source population comprises outpatient and inpatient adults undergoing chest CT for heterogeneous clinical indications. Phase 1 analyses will therefore report descriptive summaries of the enrolled routine-care cohort rather than pooled population prevalence. Prespecified subgroup reporting will prioritise care pathway, contrast status, known malignancy status and, where feasible, major indication strata so that phenotype and implementation findings can be interpreted within relevant source groups.

Pilot phase and cohort launch

Before formal recruitment, the study team will complete staged calibration using test cases permitted under the institutional workflow. Initial training will include SOP familiarisation, joint review of at least three demonstration cases and 3–5 individual test cases before independent study reading. Calibration will assess screening logic, data availability, form burden, phenotype measurability, coding consistency and practical workflow issues. These cases will be used for training and process refinement and will not be treated as confirmatory scientific results.

Baseline data collection

Baseline data collection is organised into seven domains: (1) identifiers and time variables; (2) clinical baseline variables; (3) examination metadata; (4) imaging phenotypes; (5) report-quality variables; (6) follow-up and outcome variables and (7) governance and quality-control variables. Clinical baseline variables include age, sex, anthropometric measures where available, smoking-related variables, selected comorbidities, outpatient/inpatient care pathway, CT indication and selected healthcare-access variables captured from structured hospital systems or supported manual abstraction. Known malignancy status and other major baseline conditions will be retained as potential stratification variables. Examination metadata include scanner manufacturer/model, examination date, contrast status and phase where relevant, computed tomography pulmonary angiography (CTPA) status, tube voltage, slice thickness, reconstruction kernel and coverage adequacy. Further operational details for these domains and related measurement and reporting procedures are provided in online supplemental tables S1–S5.

Routine acquisition and technical assessability

The study does not impose a harmonised research acquisition protocol. Routine technical heterogeneity is therefore treated as a measured design feature rather than unstructured noise. Scanner platform, tube voltage, contrast status and phase, slice thickness, reconstruction kernel, scan coverage, motion and other major technical limitations will be retained as method metadata where available. Phenotype-specific states of assessable, limited, not assessable and, for interval analyses, not comparable will be recorded explicitly. This structure is intended to separate technical non-assessability from biological absence of a phenotype and to identify the routine acquisition conditions under which each domain remains usable.

Imaging phenotypes

The imaging phenotype framework has two layers. Opportunistic phenotype capture, as defined above, does not imply that every recorded phenotype is incidental in the strict clinical sense; rather, the cohort captures prespecified findings visible on routine-care chest CT within a standardised observational framework.

The core clinical-reporting layer includes pulmonary nodules, emphysema, interstitial lung abnormalities (ILAs), coronary artery calcification (CAC) and pulmonary artery/ascending aorta metrics. Pulmonary terminology and nodule recording follow current Fleischner terminology and nodule guidance.^{9 10} Emphysema and ILA assessment use established visual/clinical-statement frameworks.^{11 12} CAC is recorded using a simple patient-level visual category on non-gated thoracic CT, and pulmonary artery/aortic measurements use prespecified axial measurement rules.^{13 14} The extended layer includes vertebral attenuation and deformity, pectoralis muscle metrics and liver attenuation-based variables when technically assessable; phenotype-specific technical rules are described below and in online supplemental material.

Scope of additional thoracic findings

Phase 1 intentionally prioritises prespecified domains for which operational definitions, CRF fields and measurement SOPs have been established. Thyroid abnormalities, oesophageal pathology, mediastinal lymph-node assessment and other thoracic incidental findings are therefore not systematic analytic domains in the initial phase. This focused scope is intended to preserve feasibility and reproducibility rather than imply that these structures lack clinical importance. Additional domains may be incorporated in later protocol extensions after definition, training and governance review. Importantly, any other potentially actionable abnormality encountered during research review will still enter the safety-feedback pathway described below.

Image review and measurement workflow

Cases will first undergo image-usability grading under the general imaging-measurement SOP. Core phenotypes will be recorded before extended measurements. Standard window settings, measurement levels, minimum recording requirements and escalation rules are prespecified. For each phenotype, assessable indicates that the variable can be recorded reliably on the available examination; limited indicates that an observation can be made but technical constraints affect interpretation; not assessable indicates that reliable recording is not possible and not comparable indicates that baseline and repeat examinations do not support valid interval comparison for that variable.

Standardised research image interpretation will be performed independently by two primary radiologists: an associate chief radiologist with 12 years of chest CT interpretation experience and an attending radiologist with 8 years of experience. Before formal recruitment, both readers will complete protocol- and SOP-based training, joint demonstration-case review and pilot calibration to harmonise phenotype definitions, measurement procedures, assessability rules and CRF coding. Research phenotype classification will be undertaken without consulting the original routine radiology report, except when access is required for an urgent safety process.

Formal inter-reader agreement will be evaluated in a minimum of 50 examinations drawn from the calibration and early implementation phases. Agreement for binary or nominal categorical variables will be assessed using kappa statistics, ordinal variables using weighted kappa and continuous measurements using intraclass correlation coefficients (ICCs). Prespecified minimum targets are κ≥0.60 for categorical/ordinal measures and ICC≥0.75 for continuous measures. If a major domain does not meet its target, targeted retraining and recalibration will be undertaken before repeat reliability assessment. Discrepant or uncertain cases will first undergo independent re-review; persistent disagreement will be adjudicated by a senior chief radiologist with 32 years of imaging experience, and the adjudicated value will constitute the final study record.

Contrast-specific cross-sectional handling

Cross-sectional analyses will not pool contrast-enhanced and non-contrast attenuation values as if they were technically equivalent. For vertebral attenuation, non-contrast examinations will be prioritised for threshold-based interpretation; vertebral level, tube voltage, contrast status and reconstruction characteristics will be retained because attenuation varies with acquisition conditions.^{15 16} Where sagittal image quality supports grading, vertebral deformity will be assessed separately using a semiquantitative morphology-based approach with established reproducibility on chest CT.^{17 18} Pectoralis muscle area and attenuation are established CT-derived measures.^{19 20} In this cohort, cross-sectional area may be described whenever segmentation is technically valid, while muscle attenuation will be stratified by contrast status/phase and no single low-density threshold will be applied across mixed phases. Liver attenuation threshold-based primary interpretation will be restricted to non-contrast CT; contrast-enhanced liver values will be analysed separately and will not be assigned non-contrast cut-offs.^{21 22} Portal-venous liver attenuation may be retained as a separate phase-specific descriptive stratum when technically appropriate.²³ Visual CAC grades will be reported with contrast status and will not be treated as Agatston-equivalent quantitative scores.¹³ Detailed rules are summarised in online supplemental table S6.

Handling and return of clinically actionable research-identified findings

If a research reader identifies a potentially clinically actionable finding, the case will be flagged for prompt review by the senior adjudicating radiologist. Triggers include findings that, in the reader’s judgement under the study SOPs, may warrant timely clinical assessment, such as a suspicious pulmonary lesion, substantial or fibrotic interstitial abnormality, clinically important CAC, vertebral fracture or another unexpected potentially serious abnormality. Senior confirmation of a finding that is not already adequately addressed in the routine report or an existing care pathway will trigger notification. The principal investigator or a delegated study radiologist will communicate the confirmed finding through the hospital’s established clinical workflow to the responsible clinical team. Study readers will not independently provide treatment advice to participants; communication with the patient and management decisions remain the responsibility of the treating clinical team. A dedicated findings-return log will document the finding category, date flagged, confirming reviewer, notifying study clinician, notification recipient/date and documented disposition or action. This process operationalises the safety pathway already specified in the approved study protocol.

Report-quality variables

Within the scan-to-phenotype-to-report-to-care framework, routine-report abstraction is treated as an independent information layer and will occur only after research phenotype classification is completed. Report-quality variables include whether a phenotype was mentioned, whether prespecified minimum descriptive elements were present, whether the finding appeared in the conclusion and whether a management recommendation was provided. The analysis will distinguish descriptive mention rates for all research-detected phenotypes from a prespecified report-eligible subset; absence of a mild or research-only phenotype from the clinical report will not automatically be classified as a reporting failure. Where locally approved structured-reporting templates exist, compliance will be recorded as an observational variable rather than introduced as a new intervention.²⁴

Follow-up design

The cohort follows a unified schedule with baseline, 6-month, 12-month and annual follow-up time points. Because formal recruitment has not yet commenced, calendar dates for the 6- and 12-month windows will be participant-specific and determined from the eventual enrolment date. These cohort-level ascertainment windows do not mandate phenotype-specific repeat imaging; repeat CT will depend on routine clinical care. The planned schedule of assessments is summarised in table 1.

Table 1. Schedule of assessments.

Assessment domain	Screening/enrolment	Baseline	6 months	12 months	Annual updates
Eligibility check and consent	X	—	—	—	—
Identifiers and time variables	X	X	Update	Update	Update
Clinical baseline variables	—	X	Selected variable updates	Selected variable updates	Selected variable updates
Examination metadata	—	X	If repeat CT	If repeat CT	If repeat CT
Core imaging phenotypes	—	X	If repeat CT	If repeat CT	If repeat CT
Extended measurements	—	When assessable	If repeat CT and assessable	If repeat CT and assessable	If repeat CT and assessable
Report-quality variables	—	X	If repeat report	If repeat report	If repeat report
Process outcomes	—	—	X	X	X
Clinical outcomes	—	—	Selected event capture	X	X
Governance and quality control	X	X	Update	Update	Update

Planned assessment schedule across screening/enrolment, baseline, 6-month follow-up, 12-month follow-up and annual updates. ‘X’ indicates planned collection at that time point. ‘Update’ indicates updating of longitudinal administrative or governance records. ‘Selected variable updates’ indicates planned updating of prespecified baseline variables that may change over time. ‘Selected event capture’ indicates early ascertainment of prespecified clinical events within routine hospital systems. ‘If repeat CT’ indicates collection only when additional routine imaging is available and does not imply protocol-mandated repeat CT for all participants. ‘If repeat CT and assessable’ indicates collection only when additional routine imaging is available and the relevant variable can be assessed on that scan. ‘If repeat report’ indicates collection only when an additional routine radiology report is available.

Follow-up information will be obtained preferentially from routine hospital systems, including repeat imaging, outpatient and inpatient encounters, referrals, multidisciplinary team discussions, pathology records and relevant downstream investigations, supplemented by telephone contact when necessary under the approved protocol. Care-pathway documentation will be linked at participant level to the research phenotype and routine-report layers, but comparisons involving reporting status and subsequent care will remain descriptive/associational and will not be interpreted causally. No defensible pre-study numerical estimate of 6-month within-system completeness is available because formal recruitment has not begun and no pilot cohort outcome data exist. Six-month follow-up-data availability is therefore retained as a primary feasibility outcome and will be reported as an observed proportion with 95% CIs overall and, where informative, by care pathway. Longitudinal outcomes will be interpreted as documented within-system care rather than complete capture of all care received elsewhere.

Outcomes

Outcomes are prespecified in three tiers for phase 1 implementation and are summarised in table 2. Primary phase 1 outcomes are feasibility- and implementation-focused and include screening and enrolment flow, consent rate, baseline data completeness, core phenotype assessability, report-quality abstraction completion and availability of within-system 6-month follow-up data within the prespecified cohort window. For phase 1 feasibility analyses, availability of 6-month follow-up data will be defined as documentation of at least one prespecified in-hospital follow-up data source within the participant-specific 6-month cohort window.

Table 2. Operational definitions of key phase 1 outcomes.

Outcome tier	Outcome	Operational definition	Denominator/time window	Primary source
Primary	Screening and enrolment flow	Counts at each stage: screened, eligible, approached, consented, enrolled and not enrolled or deferred.	All routine chest CT examinations assessed during phase 1/recruitment period	Screening log
Primary	Consent rate	Number consented divided by the number approached and eligible to consent.	Eligible approached individuals/recruitment period	Screening log
Primary	Baseline data completeness	Participant-level completion of the prespecified minimum baseline variables, with variable-level completeness summaries.	All enrolled participants/baseline	CRF and hospital systems
Primary	Core phenotype assessability	Whether each core domain can be assessed, reported overall and by domain.	Enrolled participants with baseline image retrieval/baseline	Image review workflow
Primary	Report-quality abstraction completion	Completion of the prespecified baseline report-quality review fields.	Enrolled participants with baseline radiology report available/baseline	Report review
Primary	Availability of 6-month follow-up data	Documentation of at least one prespecified in-hospital follow-up data source (eg, repeat imaging, outpatient/inpatient encounter, pathology, MDT, referral or downstream investigation) within the participant-specific 6-month cohort window.	Participants reaching the participant-specific 6-month window/6 months	Routine hospital systems
Secondary	Phenotype frequency and distribution	Frequency and descriptive distribution of core and extended phenotypes using variable-specific assessable denominators.	Assessable participants/baseline	Image review
Secondary	Cross-domain phenotype cooccurrence	Participant-level cooccurrence of prespecified pulmonary, cardiovascular, musculoskeletal and body-composition phenotypes, described using pairwise summaries and the number of affected domains.	Participants assessable for the relevant domain pair or domain-count definition/baseline	Image review
Secondary	Report-quality indicators	Report mention among all research-detected phenotypes; minimum descriptive elements, conclusion prominence, management recommendation and structured-reporting compliance evaluated primarily within prespecified report-eligible subsets.	All research-detected phenotype/report pairs for mention; report-eligible pairs for completeness/baseline	Report review
Secondary	Early care-pathway and process outcomes	Documented repeat CT, specialist referral, MDT discussion, downstream investigation and implementation of management recommendations after baseline; interpreted descriptively without causal attribution to research phenotype or reporting status.	Participants eligible for each process outcome/6 months, 12 months and annual updates	Routine hospital systems
Exploratory	Selected clinical outcomes	Prespecified diagnosis categories, hospital admission, readmission and pathology-linked confirmation where relevant.	Participants reaching each follow-up window/6 months onward	Routine hospital systems and pathology
Exploratory	Interval imaging change	Prespecified interval change for selected phenotypes when repeat imaging exists and is technically comparable.	Participants with comparable repeat CT/follow-up windows	Baseline and repeat imaging

CRF, case report form; MDT, multidisciplinary team.

Secondary phase 1 outcomes include the frequency and descriptive distribution of selected core and extended phenotypes among enrolled participants, cross-domain phenotype cooccurrence, report-quality indicators and early care-pathway/process outcomes, including repeat CT completion, specialist referral completion, multidisciplinary team participation, downstream investigation completion and documented implementation of management recommendations. Exploratory outcomes include selected diagnosis categories identifiable within routine systems, hospital admission, readmission, pathology-linked confirmation where relevant and interval imaging change when repeat imaging is available and technically comparable.

Denominators will be defined separately for all enrolled participants, variable-specific assessable subsets and follow-up-eligible subsets, as appropriate. Feasibility outcomes will be interpreted against prespecified implementation benchmarks intended to guide implementation review and protocol refinement rather than formal hypothesis-testing thresholds or rigid stop-go rules. Detailed coding and denominator conventions are prespecified in table 2, the study variable dictionary, the SOP package and online supplemental tables S1–S6.

Data management and quality assurance

Each participant will be assigned a study-specific identifier. Data capture is structured around the study variable dictionary, CRFs and SOP package. Data extracted from hospital systems will be distinguished from manually abstracted variables, and data provenance will be retained where feasible. Version control will be maintained for the protocol, variable dictionary, CRFs, issue logs and SOPs. Quality assurance is built into cohort launch through screening checks, structured measurement rules, CRF completion rules, issue logging, calibration review and layered re-check procedures.

Identifiable information required for recruitment, linkage and operational follow-up will be accessible only to authorised personnel under approved conditions. Research analysis datasets will be de-identified wherever feasible before analysis. The re-identification key, where required for approved follow-up procedures, will be stored separately from analytical datasets with controlled access.

Consistency assessment

Agreement analyses will use the prespecified minimum set of 50 examinations described above. Kappa or weighted kappa will be reported with 95% CIs for categorical/ordinal variables and ICCs with 95% CIs for continuous measurements. The targets κ≥0.60 and ICC≥0.75 are implementation thresholds for triggering recalibration, not claims of universal clinical acceptability. Agreement will be evaluated by major phenotype domain, and any retraining, repeat assessment or senior adjudication will be documented in the quality-control log.

Sample size considerations

This study is conceived as a platform cohort rather than a single-effect confirmatory study with one primary hypothesis requiring a fixed sample-size calculation at launch. The initial target of approximately 3000 participants reflects expected study-centre throughput, operational feasibility and the need for a sufficiently large resource to estimate multidomain phenotype frequencies with useful precision and support prespecified stratified summaries by care pathway and technical factors such as contrast status. At n=3000, an estimated proportion near 50% has an approximate 95% confidence-interval half-width of 1.8 percentage points, while a proportion near 10% has a half-width of about 1.1 percentage points. The target therefore supports both implementation evaluation and stable descriptive resource characterisation.

Statistical analysis

Analyses for phase 1 will be primarily feasibility-focused and descriptive. Screening, eligibility, approach, consent and enrolment counts will be summarised in a participant-flow diagram, including broad aggregate non-enrolment categories corresponding to eligibility, deferral or refusal status. No participant-level comparison of consenters and non-consenters is planned because research covariates and phenotypes will not be abstracted for non-consenters. Key feasibility measures, including consent rate, baseline data completeness, core phenotype assessability, report-quality abstraction completion and availability of within-system 6-month follow-up data, will be reported with 95% CIs using clearly defined numerators and denominators. These analyses are intended to describe implementation performance and ascertainment feasibility rather than screening effectiveness or downstream clinical benefit.

Baseline characteristics and phenotype summaries will be presented overall and by prespecified subgroups including outpatient/inpatient care pathway, sex, age group, contrast status, known malignancy status and, where feasible, major indication strata. Continuous variables will be summarised using means with SD or medians with IQRs, as appropriate, and categorical variables using counts and percentages. Phenotype frequencies will use variable-specific assessable denominators and 95% CIs where appropriate and will not be interpreted as population prevalence estimates. Cross-domain cooccurrence will be described using pairwise phenotype summaries and the distribution of the number of affected phenotype domains per participant; these analyses are descriptive and will not be used to define data-driven disease clusters in phase 1. Missingness, technical non-assessability and non-comparability will be reported separately rather than combined as generic missing data.

Exploratory adjusted association analyses will be limited to prespecified outcomes and covariates rather than data-driven variable selection. For selected fixed-window binary outcomes, unadjusted proportions will be reported first, followed where informative by regression models estimating adjusted associations with prespecified baseline covariates. Contrast-specific handling of attenuation variables will follow the rules described in the imaging methods and online supplemental table S6. Interval imaging analyses will be restricted to participants with repeat routine imaging that is technically comparable under the imaging SOPs. No formal adjustment for multiple testing is planned for exploratory analyses, which will be interpreted as hypothesis-generating rather than confirmatory.

Study status

As of August 2026, formal recruitment has not commenced and no participant-level study data have been collected or analysed. The originally planned launch has been deferred while funding and final operational preparations are being completed. Recruitment will commence only after these preparations and any required ethics notification or amendment are complete. The target remains approximately 3000 prospectively consented participants.

Patient and public involvement

Patients and members of the public were not formally involved as research partners in the development of the research question, outcome selection or initial study design. The study protocol and informed consent materials underwent independent review by the institutional Medical Ethics Committee, whose multidisciplinary membership includes lay/community representation. This review formed part of the institutional ethics and governance process and is not considered formal patient and public involvement in the research design.

Ethics and dissemination

Ethics approval was granted by the Medical Ethics Committee of Dazhou Central Hospital, Dazhou, China (approval no. 2026审(084)号; approved 27 April 2026). The approval covers the prospective observational cohort framework, written informed consent and the principle that clinically meaningful research-identified imaging findings are returned through the institutional clinical pathway rather than being independently interpreted therapeutically to participants by study researchers. Recruitment and data collection will proceed in accordance with the approved protocol and any subsequent required notification or amendment.

Written informed consent will be obtained from all prospectively enrolled participants before cohort inclusion and research data capture. Privacy and confidentiality will be protected through study-specific identifiers, controlled access to identifiable information, separation of the re-identification key from research datasets and de-identified data export wherever feasible. Only authorised study personnel, monitors or regulators operating under approved conditions will have access to study records.

Findings from this work will be disseminated in stages. This protocol paper will describe the cohort rationale, design and operational framework. Once enrolment and data quality are sufficiently stable, we plan to prepare a cohort profile or baseline resource paper. Phenotype-specific analyses will then be reported for pulmonary, cardiovascular, musculoskeletal, body-composition and pathway-related questions. Future substudies that go beyond the scope of the approved cohort protocol will be submitted for amendment or separate ethics review as required.

Scope and limitations

The deliberately focused phase 1 phenotype framework does not systematically capture every potentially relevant thoracic structure. Thyroid, oesophageal and mediastinal lymph-node abnormalities are examples of clinically relevant findings that may be incorporated in future extensions but are not prespecified analytic domains in this initial phase. Extended domains also differ in technical robustness: liver attenuation has greater expected non-assessability because upper-abdominal coverage and contrast phase vary, muscle attenuation is contrast-sensitive and visual CAC on routine non-gated or contrast-enhanced CT is a coarse ordinal measure rather than a quantitative calcium score. These limitations will be reported by domain and technical stratum rather than obscured within pooled estimates.

The same routine-care constraints are also part of the implementation question. Unlike tightly harmonised research imaging, this cohort deliberately retains ordinary acquisition and referral heterogeneity while measuring it explicitly through technical metadata, assessability states and stratified denominators. If the framework proves stable, the cohort can support phenotype-specific longitudinal studies and later multicentre evaluation of whether the scan-to-phenotype-to-report-to-care architecture is transportable across institutions.