Unrecognised Comorbidity Detection in Hospitalised Patients (CODETECT)

Unrecognised Comorbidity Detection in Hospitalised Patients (CODETECT)

Over two million people in the UK are unaware that they're living with a long-term (chronic) health condition, such as diabetes or a heart problem. These chronic conditions can lead to serious complications such as heart attacks, strokes, and kidney problems. By diagnosing these conditions earlier, effective treatments can be started sooner which will reduce the risk of harm. However, diagnosis relies on people having symptoms and contacting their doctor or attending NHS Health Checks.

There are over 16 million admissions to English hospitals each year. Hospitals collect a lot of information during a hospital stay including patients' age, blood test results and blood pressure measurements. Research has shown that this information can be helpful in spotting people with chronic conditions.

This study aims to design and test a digital platform to find the patients in hospital who are most likely to have a chronic disease or develop one in the near future.

To do this, the investigators will:

• Use information from earlier research studies and experts to pinpoint which patient information (for example, certain blood tests) would be most useful to spot people with chronic conditions.
• Extract relevant information from historical patient records, looking at who has these risk factors and which patients developed chronic conditions. The investigators will use information from hospital and general practitioner records.
• Build tools to combine this information to predict which people have, or will develop, chronic conditions.
• Implement these tools into a "real-time" digital platform that could be used to find which people should undergo further testing for a chronic condition.
• Test the platform usability with clinical stake holders.

Study Overview

• Status: Active, not recruiting
• Conditions: Diabetes; Cardiac Disease; Atrial Fibrillation (AF)

Detailed Description

This is a multi-centre observational cohort study of adult patients admitted to acute hospitals. Data will be collected from hospital systems sourcing data from both hospital and primary care electronic health record systems. The study will then use retrospective data to develop and validate tools to identify patients with undiagnosed long-term conditions.

These diagnostic tools will be implemented into a real-time digital platform and further validated on prospectively collected data. Once developed and validated, the digital platform could be used to identify patients who likely have undiagnosed long-term conditions and should undergo further investigation and preventative intervention.

The investigators will initially focus on two long-term conditions (diabetes and atrial fibrillation) and aim to expand this to others within the study period.

Why Diabetes and Atrial Fibrillation? Diabetes Diabetes is a major contributor to multimorbidity. More than 4.3 million people in the UK are living with this condition, with a further one million thought to be undiagnosed. Diabetes increases cardiovascular risk and can lead to chronic kidney disease and debilitating neuropathy. Current diabetes screening occurs through the NHS Health Checks and when people seek healthcare for unrelated symptoms. Early intervention can reduce the risk of long-term complications, including myocardial infarctions and death. However, diagnosing diabetes can be challenging when people are asymptomatic yet already have complications from their diabetes.

There are a range of well-established risk factors including non-white ethnicity, obesity, hypertension, family history, socioeconomic deprivation and increasing age. Recent systematic reviews of existing diabetes screening tools highlight poor or limited external validation, methodological weaknesses, and heterogenous definitions of diabetes that limit comparison between tools.

Atrial Fibrillation (AF) Atrial fibrillation (AF) is a common cardiac arrythmia, affecting 2.5 million people in England alone. Of these, 30% are undiagnosed. AF increases the risk of stroke five-fold, leading to decreased mobility and vascular dementia. There is currently no UK screening programme.

AF is a common complication of critical illness, associated with prolonged intensive care treatment and higher mortality. Lifestyle factors, such as obesity, smoking and high alcohol consumption also increase AF risk. People with AF are often prescribed anticoagulation to reduce stroke risk. However, the benefits of anticoagulation must be carefully balanced with the risk of bleeding, emphasising the need for more accurate prognostic models.

Study Activities

The investigators will reach our objectives by completing the following study activities:

• Use expert panels to agree existing diagnostic definitions for at least 2 long-term health conditions that can be defined from electronic health records.
• Identify risk factors for long-term health conditions through a literature review, expert panel, and machine learning methods (using retrospective data).
• Develop and validate diagnostic models (using retrospective data) to identify patients with previously undiagnosed long-term health conditions.
• Develop a real-time digital platform in at least one hospital to collect data to prospectively validate the diagnostic models.

• Study Type: Observational
• Enrollment (Estimated): 4500000

Contacts and Locations

Study Locations

• United Kingdom
  • Oxfordshire
    • Oxford, Oxfordshire, United Kingdom, OX3 9DU
      • Critical Care Research Group, Nuffield Department of Clinical Neurosciences, University of Oxford

Participation Criteria

Eligibility Criteria

• Ages Eligible for Study: Adult; Older Adult
• Accepts Healthy Volunteers: Yes

• Sampling Method: Non-Probability Sample

Study Population

The analysis population will be based on all hospital admissions in both the retrospective and prospective cohorts. Any subgroup analysis will be prespecified in the statistical analysis plan with justification.

Description

Inclusion Criteria:

• Adults aged 18 years or above.
• Admitted to a participating NHS hospital
• Registered with a primary care practice

Exclusion Criteria:

• Has "opted-out" of having their data used for research purposes using the national data opt-out service

Study Plan

How is the study designed?

Number of groups / cohorts

2

Cohorts and Interventions

Group / Cohort
Retrospective cohort; Retrospective Cohort: Around 3,600,000 hospital admissions from 3 sites over 12 years* *Study will begin as single site, aiming for 3 participating Trusts Retrospective sub-study data collection period: 1st December 2015 to 31st August 2027 (retrospective cohort 1st December 2015 to 30th June 2024, with rolling follow-up to include data to 31st August 2027.
Prospective Cohort; Prospective Cohort: Around 900,000 hospital admissions from 3 sites over 3 years* *Study will begin as single site, aiming for 3 participating Trusts Prospective sub-study data collection period: 1st July 2024 to 31st August 2027

What is the study measuring?

Primary Outcome Measures

Outcome Measure	Measure Description	Time Frame
Use data to design and use a real-time, digital platform to prospectively validate prediction models to identify hospitalised patients with potentially undiagnosed chronic health problems for at least 2 chronic health problems.	Measure #1 Discrimination (c-statistic) and calibration (intercept and slope) of model predicting diagnosis of a new chronic health problem Measure #2 Positive and negative predictive values, sensitivity, and specificity	Primary timepoint Within five years of hospital discharge. Secondary timepoints • Within three years of hospital discharge • Within two years of hospital discharge • Within one year of hospital discharge • Within six months of hospital discharge

Secondary Outcome Measures

Outcome Measure	Measure Description	Time Frame
The implementation of externally validated prediction models into a novel digital platform to identify undiagnosed chronic health problems (comorbidities) in hospitalised patients		Up to five years post-hospital discharge
The generation of an intuitive usable digital platform ready for clinical use		Up to five years post-hospital discharge
Association of risk factors that would be available at hospital discharge with at least 2 chronic health problems	Statistical measures of association including odds ratio with 95% confidence intervals.	Up to five years post-hospital discharge

Collaborators and Investigators

• Sponsor: University of Oxford
• Investigators: Principal Investigator: Peter Watkinson, University of Oxford

Study record dates

Study Major Dates

• Study Start (Actual): July 1, 2024
• Primary Completion (Estimated): November 30, 2027
• Study Completion (Estimated): November 30, 2027

Study Registration Dates

• First Submitted: February 17, 2025
• First Submitted That Met QC Criteria: March 11, 2025
• First Posted (Actual): March 18, 2025

Study Record Updates

• Last Update Posted (Actual): May 19, 2026
• Last Update Submitted That Met QC Criteria: May 15, 2026
• Last Verified: February 1, 2026

More Information

Terms related to this study

• Keywords: digital platform; Machine learning; Statistical modeling
• Additional Relevant MeSH Terms: Endocrine System Diseases; Cardiovascular Diseases; Pathologic Processes; Metabolic Diseases; Arrhythmias, Cardiac; Glucose Metabolism Disorders; Pathological Conditions, Signs and Symptoms; Nutritional and Metabolic Diseases; Heart Diseases; Diabetes Mellitus; Atrial Fibrillation
• Other Study ID Numbers: PID 17881

Plan for Individual participant data (IPD)

• Plan to Share Individual Participant Data (IPD)?: NO

Drug and device information, study documents

• Studies a U.S. FDA-regulated drug product: No
• Studies a U.S. FDA-regulated device product: No