Doctors measure particular features of their patients, to diagnose conditions and to monitor progress of those conditions. Examples include laboratory tests, lung function measures, or questionnaires to assess pain, memory or depression. However, within the same patient, even if the patient’s condition has not changed, all measurements vary over time. This is partly due to chance. The variation is important, because chance variation in a measurement can lead to incorrect diagnosis or to incorrect judgement about the progress of the condition. Depression questionnaires are widely used to help diagnose and to monitor progress of depression. They are often used in people with other health conditions such as heart disease, to check if they may be suffering from depression. However there is almost no research on how scores from depression questionnaires vary within individuals. This means we may miss depression, overdiagnose depression, or incorrectly judge that a patient is either improving or getting worse.
This study aims to investigate how much depression scores vary within individuals. We will investigate the data from individuals with more than one depression score and assess how much their scores vary. We will compare the variation in depression scores in people who have not been diagnosed with depression and those who have been diagnosed with depression. We will also investigate whether the variation differs in men and women, older and younger patients, those of different ethnicities and income groups.
Our results will help give practical guidance to doctors on how to use and interpret depression scores.
Clinical measurements are undertaken to diagnose conditions, to monitor progress and to assess treatment effects. They include laboratory tests, clinical assessments and questionnaires.
Clinical measurements change when there are pathophysiological changes but even in the absence of pathophysiological change all measurements show within-individual variation. Some of this is due to cyclical factors but most is due to a combination of measurement error and chance variation in the parameter around a physiological homeostatic point. This is important, because chance variation can lead to incorrect diagnosis, incorrect judgement about progress or treatment effects. There is substantial research on within-individual variability of some measured parameters such as blood pressure.
Depression questionnaires (Hospital Anxiety and Depression Score, HADS; Becks Depression Inventory I and II, BDI; Patient Health Questionnaire-9, PHQ-9) are widely used for diagnosis and monitoring depression. There is little research on within-individual variation in depression scores and none on their variation in routine clinical practice. This means we have little understanding of the probability of misdiagnosis or of incorrect judgements about the effectiveness of treatment.
This study aims to quantify within-individual variation in commonly used depression scores (HADS, BDI, PHQ-9).
We will identify individuals with more than one measurement of a particular depression score. We will estimate the mean and the variance components as within and between-individual variance using a linear regression model. From this we calculate the within-individual coefficient of variation (CVi), as the square root of the variance divided by the mean. We will investigate how CVi differs between people with and without a diagnosis of depression or with or without drug treatment for depression, by sociodemographic characteristics and other patient characteristics such as chronic diseases, smoking, alcohol consumption, BMI and number of comorbidities.
Our results will help give practical guidance to doctors on the interpretation of depression scores.
Within-individual variation of Hospital Anxiety and Depression Score (HADS); Becks Depression Inventory I and II (BDI) and Patient Health Questionnaire-9 (PHQ-9)
Tom Marshall - Chief Investigator - University of Birmingham
Alex Gough - Corresponding Applicant - University of Birmingham
Alice Sitch - Collaborator - University of Birmingham