Type 2 diabetes mellitus (T2DM) is a lifelong condition where patients are unable to regulate their blood sugar levels. National treatment guidelines provide a framework for clinicians to prescribe treatments to that are safe, effective and good value for money for both newly diagnosed and long-term patients who may require a change in their anti-diabetic medicines.
UK treatment guidelines recommend targets for blood sugar control (measured by haemoglobin A1c (HbA1c)) of between 6.5%7.5% (4858 mmol/mol). However, as recommended in recent European treatment guidelines, clinicians must carefully consider treatment options based on the characteristics of the patient and therefore it is likely that the decision of choosing an alternative treatment is not reflected just in terms of a generic blanket use of UK guidelines.
The aim of this study is to explore relationships between alternative HbA1c targets, individualised patient factors and both short- and long-term health outcomes such as complications which occur due to poor blood sugar control in order to provide evidence to inform an individualised approach to Hba1c target setting. Based on potential alternate target setting proposed as a result of the initial aim, an additional aim to understand the health economic consequences of alternate strategies will be explored.
A retrospective database study of routinely collected UK primary and secondary care data will inform the development of predictive models (risk equations) relating the risk of T2DM-related events to patient profiles.
Data from the CPRD GOLD dataset in addition to linked hospital episode statistics (HES) admitted patient care (APC) will be extracted for all T2DM patients over the study period 01/01/199831/12/2017 based on a 5-year pre-index look back period and a 15-year post-index period.
Patient data will follow all observations up to and including the first occurrence of death, loss to follow up, or the end of the index period (31/12/2017).
Study outcomes include an individual occurrence of:
a) microvascular complications;
b) macrovascular complications;
c) hypoglycaemia;
d) admission to hospital;
e) death.
Explanatory variables will include factors relating to patients:
a) Demographic profile
b) Socioeconomic profile
c) Clinical profile
Multivariate regression equations relating outcome risk to patient profiles will be used to generate risk profiles for alternative HbA1c targets. Models will be developed concurrently based both on a single timepoint (i.e. baseline) and across multiple timepoints to consider not only the hazard at a single time-point but also how that hazard changes as a factor of time-varying factors (such as the effect of age, body mass index, etc.)
Within the model development process, any models developed will be assessed to understand the generalizability related to our dataset and model performance in relation to existing research. Assessments between the two models in relation to predicting complications will be compared utilizing receiver-operating characteristic (ROC) curves to determine generalizability and applicability.
The health economic consequences associated with an individualized approach to HbA1c target setting will be explored via analyses relating hospitalization and clinical events to HbA1c levels after adjusting for a range of patient characteristics.
Clinical outcomes:
The following clinical outcomes will be derived based on the earliest recording of the relevant diagnosis in either CRPD GOLD or HES APC using READ and ICD-10 codes respectively:
an occurrence of hypoglycaemia, defined as a composite outcome of the first recording of any of:
an occurrence of microvascular complications, i.e. diabetic nephropathy, diabetic neuropathy, retinopathy
an occurrence of macrovascular complications, i.e. congestive cardiac failure, ischaemic heart disease, myocardial infarction, stroke;
any admission to hospital;
death.
In cases where there are multiple records of an event across the combined data, the earliest record of an event (such as a myocardial infarction (MI)) will represent the index event, with other records defined as representing the same event (MI) if they were dated within 30 days of the index event (MI)).
Health and economic outcomes
The health and economic consequences associated with HbA1c target setting will be based on regression analysis of rates of clinical events and hospitalisations the event rate profiles estimated from the developed risk equations relating these outcomes to patient-level characteristics and HbA1c levels:
Event rates: micro- and macrovascular, hypoglycaemia, death;
Event costs: event costs in first and subsequent years, based on published sources;
Life expectancy: derived from life tables;
Quality -adjusted life expectancy: age, gender and event-based disutility profiles will be applied to life expectancy profiles.
Iskandar Idris - Chief Investigator - University of Nottingham
Jason Gordon - Corresponding Applicant - Health Economics & Outcomes Research Ltd ( HEOR Ltd )
Nadeem Qureshi - Collaborator - University of Nottingham
Stephen Weng - Collaborator - University of Nottingham
Yana Vinogradova - Collaborator - University of Nottingham
Michael Hurst - Collaborator - Health Economics & Outcomes Research Ltd ( HEOR Ltd )
Phil McEwan - Collaborator - Health Economics & Outcomes Research Ltd ( HEOR Ltd )
Thomas Mason - Collaborator - Health Economics & Outcomes Research Ltd ( HEOR Ltd )
HES Admitted Patient Care;ONS Death Registration Data;Patient Level Townsend Score