The diagnosis of Alzheimer’s disease (AD) often occurs well after initial symptoms of memory loss. Evaluation of patients’ memory and functioning using diagnostic tests such as the mini-mental state exam (MMSE) could assist physicians and specialists in arriving at a more timely diagnosis. However, the information about such diagnostic testing, including the findings, may only be reported in physician notes (free text) or discharge summaries (if such testing is conducted by specialists) or discharge letters (if testing is conducted in memory clinics). As a result, little is known about the extent to which such evaluations are conducted before providing a diagnosis of AD to patients with memory issues. In this study, we will review the information captured by physicians, either in their notes or from discharge summaries/letters, to identify the timing and results of diagnostic testing for memory-related issues in the years before a formal AD diagnosis is provided to patients in the UK. The results of this study will be used to inform a future large scale study to understand the effects of diagnostic testing before AD diagnosis on timing of AD diagnosis and associated health outcomes, if any.
The proposed study will assess the feasibility of using physician notes in combination with discharge letters and discharge summaries available to physicians to develop an algorithm to accurately identify the timing and results of cognitive and functional testing in CPRD data. Specifically, physician notes and select discharge letters/summaries will be requested for a random sample of 50 patients with confirmed AD diagnosis. Search algorithms will be developed to identify consultations containing information about whether a physician evaluated patient’s cognitive function either directly or through recording of information from the discharge summaries and/or reported the results of those assessments. The algorithm will be developed in two main steps. First, the physician notes as well as discharge letters/summaries will be reviewed manually and targeted key word searches will be conducted for a subset of 15 patients. Then, the findings of the preliminary analysis will be applied to the remaining patients. All text data will be reviewed for the entire sample to evaluate the rates of false positives and false negatives. In addition, the findings from the algorithm will be separately compared to those based on discharge letters/summaries and Read codes corresponding to cognitive assessments and symptoms of cognitive impairment.
Noam Kirson - Chief Investigator - Analysis Group, Inc.
Alan Lenox-Smith - Collaborator - Eli Lilly & Co - UK
Carlos Martinez - Collaborator - Institute for Epidemiology, Statistics and Informatics GmbH (Pharma Epi)
Catherine Reed - Collaborator - Eli Lilly & Co - UK
Grazia Dell'Agnello - Collaborator - Eli Lilly & Co Ltd - US Headquarters
Howard Birnbaum - Collaborator - Analysis Group, Inc.
Jill Rasmussen - Collaborator - Not from an Organisation
Jody Wen - Collaborator - Analysis Group, Inc.
Mark Belger - Collaborator - Eli Lilly & Co - UK
Mark Meiselbach - Collaborator - Analysis Group, Inc.
Sarah King - Collaborator - Analysis Group, Inc.
Urvi Desai - Collaborator - Analysis Group, Inc.
HES Admitted Patient Care;HES Outpatient