Within-individual variation of selected clinical measurements

Gough, Alexander Gerard ORCID: 0000-0001-6954-7408 (2024). Within-individual variation of selected clinical measurements. University of Birmingham. Ph.D.

Preview

Gough2024PhD.pdf
Text
Available under License All rights reserved.
Download (8MB) | Preview

Abstract

Background
Within-individual measured variation of a clinical test is the variation over time of an individual’s clinical test results. Total measured within-individual variation includes biological variation and pre-analytic and analytic variation. Biological variation includes systematic variation, for example circadian or seasonal, and chance fluctuation around a homeostatic set point. Pre-analytic variation includes environmental and patient pre-measurement factors that affect the test result and analytic variation is the variation caused by the inherent accuracy of the test.
Within-individual variation is important in clinical decision-making because clinical decisions are informed by clinical test results and the existence of within-individual variation means that a single, or even multiple clinical measurements may not truly reflect the average status of the patient, which can lead to incorrect diagnostic, prognostic and therapeutic decisions.

Aims
This thesis aims to explore the existing data on within-individual variation on laboratory tests, functional measures and questionnaire-based tests and use large primary care databases to estimate this variation in these selected tests.

Methods
Four systematic reviews of the existing literature on within-individual variation of seven tests were performed according to PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines. Three tests were laboratory-based: HbA1c (Haemoglobin A1c), CRP (C-Reactive Protein) and hsCRP (high-sensitivity C-Reactive Protein); two were functional tests, the spirometry measures FEV1 (Forced Expiratory Volume in one second) and FVC (Forced vital capacity); three were questionnaire-based, the depression symptom measurement instruments PHQ-9 (Patient Health Questionnaire-9), BDI (Becks Depression Inventory) and HADS (Hospital Anxiety and Depression Scale). Four large retrospective cohort studies were then conducted using the large primary care databases, The IQVIA Medical Research Database (IMRD) and The Clinical Practice Research Datalink (CPRD), to estimate the real-world within-individual measured variation of seven tests, HbA1c, CRP, FEV1, FVC, PHQ-9, HADS and BDI.

Results
For the four systematic reviews, in total, 8,724 abstracts were screened, 873 full texts were examined and 316 studies were included.
For the HbA1c systematic review, median within-individual coefficient of variation (CV) of HbA1c was 0.070 (IQR 0.034 to 0.09). For the CRP/hsCRP systematic review, median CV for CRP was 0.41 (range 0.11 to 0.89), and for hsCRP, median CV was 0.44 (range 0.27 to 0.76). For the spirometry systematic review, the median CV of FEV1 was 0.034 (IQR 0.025 to 0.067) and of FVC was 0.031 (0.02 to 0.048). For the depression symptom measurement instruments review, the pooled estimates of ICCs (Intraclass Correlations) and 95% confidence intervals for BDI, HADS, PHQ-9 and PHQ-2 were 0.89 (0.84-0.93), 0.89 (0.85-0.93), 0.84 (0.81-0.88) and 0.75 (0.56-0.93), respectively.
For the HbA1c retrospective cohort study, the overall measured within-individual coefficient of variation for HbA1c (CVT) was 0.20 (95%CI 0.20 to 0.20). For the CRP retrospective cohort study, the overall measured coefficient of variation for CRP was 1.148 (95% CI 1.147 to 1.149). For the spirometry retrospective cohort study, the overall coefficient of variation for FEV1 was 0.224 (95% CI 0.221 to 0.228) and for FVC was 0.152 (0.150 to 0.155). For the depression symptom measurement instrument retrospective cohort study, the overall measured coefficient of variation (95% CI) was 0.412 (0.411 to 0.413) for PHQ-9, 0.370 (0.365 to 0.374) for HADS and 0.763 (0.701 to 0.826) for BDI. Measured variation tended to increase with worsening disease severity except for the depression symptom measurement instruments.

Discussion
The systematic reviews in this thesis showed that there was a lack of good quality, generalisable data on the within-individual measured variation of a variety of clinical measurements, and this was particularly pronounced for non-laboratory tests. The retrospective cohort studies showed that that the estimated measured within-individual variation for the selected tests was much higher than previously reported. They also showed that with the exception of the depression symptom measurement instruments, variation increased with worsening disease severity.
Limitations of the systematic reviews in this thesis include difficulties in identifying data on variation in studies which were not primarily designed to report variation. Limitations with the cohort studies include the difficulty of accounting for confounding by indication, and inaccuracies in the database.

Conclusion
This thesis demonstrates that the current published amount and quality of data on within-individual measured variation in real-world settings is limited, especially for non-laboratory tests. The estimates of within-individual measured variation from the analyses of primary care records in this thesis are higher than published estimates of within-individual variability. This may have important consequences for the accuracy of clinical decision-making with respect to diagnosis, prognosis and management. The extent of measured within-individual variation should be considered by clinicians, clinical trial authors and the designers of drug trials. Further research is needed on the extent of measured within-individual variation in a wide range of tests in real-world settings and its effect on clinical decision-making. Research into methods to account for confounding by indication would be helpful.

Type of Work:

Thesis (Doctorates > Ph.D.)

Award Type:

Doctorates > Ph.D.

Supervisor(s):