Assessment of health economics in Alzheimer’s disease (AHEAD) based on need for full-time care

Abstract

The need for a generally applicable tool to forecast long-term outcomes based on relatively short-term data is becoming increasingly acute with the advent of new therapies for AD. The AHEAD model provides a relatively simple framework for the prediction of time to FTC requirement based on short-term observed data such as those from clinical trials. Although subject to the uncertainties inherent in modeling, the model nevertheless provides a standard estimation technique that may facilitate comparisons between existing and emerging therapies.

The clinical and cost-effectiveness of donepezil, rivastigmine, galantamine and memantine for Alzheimer's disease

“…131 In line with the AHEAD model, all studies report analyses based on a short-term module covering an initial 6-month (trial) period, followed by a long-term module over a 10-year time horizon.…”

“…The same process is undertaken for the predictive equations for death. Caro and colleagues, 131 in the development of the AHEAD model, use the methods from Stern and colleagues but undertake additional analysis to broaden the scope of the predictions, producing regression equations of a continuous nature for patients aged ≤ 73 years and for patients aged >73 years (this age divide is due to data stratification in the original publication). For further information on the statistical methods applied, see Caro and colleagues.…”

“…For further information on the statistical methods applied, see Caro and colleagues. 131 The predictive equation for 'requiring FTC' has parameter values included in the index for age, the presence of extrapyramidal symptoms (EPS), the presence of psychotic symptoms (e.g. delusions, hallucinations), age at onset, duration of illness and cognitive score as measured by the modified MMSE (mMMS).…”

“…statistical significance, generalisability) and in the overall approach to characterise disease severity, given the growing literature that indicates that cognitive function is an unreliable approach to predicting disease progression [123][124][125] and outcomes such as time to institutionalisation. 43,125,178 Others have used datasets and statistical techniques to model disease progression over time, 91,128,131,179 and these methods are also subject to concerns over validity and generalisability. In the datasets used to model AD progression, all have limitations, be it the use of trial-specific transition probabilities or the frailties which exist in observational data sets reporting on AD patients over time.…”

“…The approach presented by Caro and colleagues 131 offers an opportunity to consider disease progression across a broader description of AD than cognitive function, using patients' characteristics and other non-cognitive AD variables. The approach of Caro and colleagues (the AHEAD model) is described in some detail above, and it is adapted in an illustrative model below, yet it still remains limited given the crude representation of disease progression (i.e.…”