Objective Disease Control and Treatment Strategies in Ulcerative Colitis

Background and Rationale

Ulcerative colitis is a chronic inflammatory disease characterized by fluctuating periods of inflammatory activity and disease control. Conventional real-world effectiveness studies frequently evaluate isolated outcomes, such as treatment persistence, hospitalization, colectomy, or disease activity at individual assessment time points. These approaches may not adequately represent the cumulative inflammatory burden experienced by patients over time.

This study will use longitudinal multimodal clinical data to reconstruct dynamic disease states and quantify the amount of follow-up time during which disease is objectively controlled, objectively uncontrolled, or insufficiently observed.

The principal longitudinal disease-burden measure will be time under objective uncontrolled disease (TUOD). A complementary measure, objective disease-control time (ODCT), will quantify the proportion of evaluable follow-up time spent in objective disease control.

Objective inflammatory status will be determined using a prespecified hierarchical algorithm incorporating available endoscopic assessments, fecal calprotectin, intestinal ultrasound, C-reactive protein, and other clinically appropriate objective assessments. Clinical symptoms, corticosteroid exposure, treatment escalation, hospitalization, and surgery will be incorporated as complementary disease-state and clinical-event information. Periods lacking sufficient objective information will not automatically be classified as disease control.

The study has three complementary analytical objectives.

Objective 1: Longitudinal Disease-Control Trajectories

To characterize longitudinal patterns of objective disease control and cumulative objectively uncontrolled disease burden in patients with ulcerative colitis and to evaluate their associations with subsequent treatment escalation, hospitalization, colectomy, corticosteroid exposure, and other clinically relevant outcomes.

Longitudinal states will broadly distinguish:

Objective disease control;

Objective disease non-control;

Clinical or therapeutic escalation states;

Major disease-related events, including hospitalization and surgery; and

Unknown or insufficiently observed disease status.

Detailed operational definitions and state-assignment rules will be prespecified in the protocol and Statistical Analysis Plan.

Objective 2: Initial Advanced-Therapy Strategy

A nested analytical cohort will include eligible patients initiating an advanced therapy.

The index date will be the date of initiation of the first eligible advanced therapy.

Alternative initial treatment strategies may include tumor necrosis factor antagonists, anti-integrin therapy, interleukin-12/23 or interleukin-23 pathway therapy, Janus kinase inhibitors, sphingosine-1-phosphate receptor modulators, or other eligible advanced therapies used during the study period.

Comparative analyses will only be conducted for treatment strategies with adequate sample size, clinical overlap, and empirical positivity.

The principal estimand will represent the difference in long-term objective disease-control time under alternative initial treatment strategies.

Subsequent treatment optimization, discontinuation, or switching will not change the original treatment-strategy classification for the primary treatment-policy analysis.

Baseline confounding will be addressed using prespecified causal adjustment methods, which may include propensity-score weighting, overlap weighting, standardization, and outcome regression.

Objective 3: Dynamic Treatment Adaptation After Objective Disease Non-Control

A second nested analytical cohort will include patients who experience a first confirmed episode of objective disease non-control while receiving an eligible advanced therapy.

The index date for this analysis will be the first qualifying objective non-control assessment.

Alternative dynamic strategies may include:

Timely within-mechanism treatment optimization; and

Timely switching to an alternative therapeutic mechanism.

A prespecified implementation grace period will be used where appropriate.

Because disease activity influences subsequent treatment decisions and may itself be affected by previous treatment, analyses of dynamic strategies will account for time-varying confounding.

Where supported by data completeness and strategy definitions, target-trial emulation methods, including clone-censor-weight approaches or marginal structural models, may be used.

The primary dynamic-strategy estimand will represent the difference in subsequent objective disease-control time under the alternative treatment-adaptation strategies.

Data Sources

Data will be derived from routinely collected clinical information at participating centers and may include:

Demographic characteristics;

Ulcerative colitis phenotype and disease history;

Clinical disease activity;

Laboratory measurements;

C-reactive protein;

Fecal calprotectin;

Hemoglobin and albumin;

Endoscopic findings;

Histopathology where clinically available;

Intestinal ultrasound where clinically available;

Corticosteroid exposure;

Conventional therapies;

Advanced therapies;

Treatment optimization and dose modification;

Therapeutic drug monitoring where available;

Treatment discontinuation and switching;

Emergency visits and hospitalization;

Surgery and colectomy;

Other clinically relevant longitudinal outcomes.

No diagnostic examination, treatment, or treatment modification will be mandated solely for research purposes.

Handling of Observation Gaps

Absence of a recorded objective assessment will not be interpreted as absence of inflammation.

Follow-up intervals will therefore be classified as objectively controlled, objectively uncontrolled, or unknown/unobservable according to prespecified temporal reconstruction rules.

The duration and pattern of unknown observation time will be explicitly quantified.

Sensitivity analyses will assess the robustness of findings to alternative disease-state reconstruction rules, carry-forward windows, and observation-density assumptions.