Predicting the occurrence and mortality of upper gastrointestinal haemorrhage, a population based study

Lead Research Organisation: University of Nottingham
Department Name: Sch of Biomedical Sciences

Abstract

Increasing numbers of elderly patients are presenting with bleeding from the stomach. I will use a general practitioner database to investigate the reasons underlying this. It could be because of higher rates of coexisting illnesses or medications. If the former, an episode of bleeding could indicate a general deterioration in health.

The increase in occurrence could be related to guidelines advising a combination of medications to be prescribed to all people following a heart attack. I will use a database of all patients presenting with chest pain in Nottingham since 1980 to investigate whether in some patients the increased risk of bleeding from these medications offsets their benefit.

Stomach bleeding is associated with an increased long-term mortality. Determining in the general practitioner database whether long-term mortality is caused by the bleeding itself or coexisting illness could explain why the mortality rate has not improved over the last decade.

It is also debated whether service provision influences mortality. There are limited facilities for urgent investigation and treatment of bleeding at weekends. I will use a national hospital database to compare the outcomes of patients presenting at weekends with those presenting on weekdays to assess the impact of service provision.

Technical Summary

Upper gastrointestinal haemorrhage is the commonest emergency medical admission for gastroenterology in the UK with an overall incidence between 47-172/100,000, and inpatient mortality of 10-14%. Despite improvements in management throughout the last decade the incidence in the elderly has increased by over 30% with minimal improvement in mortality.
Aim
I will investigate the complex interactions underlying the rising incidence of upper gastrointestinal haemorrhage in the elderly and examine the mortality in this population.
Objectives
In large population based studies I will investigate the following:
1. Do co-morbidities predict the occurrence of upper gastrointestinal haemorrhage independently of medications?
2. Do co-morbidities predict long-term mortality independently of initial haemorrhage?
3. In a single co-morbidity of acute coronary syndrome, what patient factors and treatments predict subsequent upper gastrointestinal haemorrhage?
4. Do service provisions impact on the mortality from upper gastrointestinal haemorrhage independently of co-morbidities?
Methodology
1. Using the General Practice Research Database in a nested case control study I will investigate the complex interactions between the co-morbidities and medications in predicting upper gastrointestinal haemorrhage.
2. Using the General Practice Research Database in a cohort study I will investigate the subsequent long-term mortality and its predictors, and its independence of the initial upper gastrointestinal haemorrhage.
3. Using the Nottingham Heart Attack Register in a case control study I will model the predictors of upper gastrointestinal haemorrhage following an acute coronary syndrome.
4. Using the UK Hospital Episode Statistics database, mortality from weekday admissions will be compared to weekend admissions, when facilities for investigation and treatment are not widely available.
Scientific and Medical Opportunities
This fellowship provides excellent scientific opportunities for training in clinical epidemiology using a range of databases and study designs, and it provides clinically important medical opportunities.
If co-morbidities predict long-term mortality in the elderly independently of bleeding then it will explain why efforts to improve management of bleeding have failed to improve mortality. An episode of upper gastrointestinal haemorrhage would therefore be similar to a hip fracture in the elderly, which indicates a general deterioration in the health of the patient requiring a more holistic approach to care.
Similarly the bleeding risk in vulnerable patients following an acute coronary syndrome needs to be established to inform evidence based prescribing. Finally, by investigating in a large study whether mortality increases at weekends compared to weekdays we will help clarify the clinical importance of pursuing national out of hours endoscopy cover.

Publications

10 25 50