Sian Taylor-Phillips

Sian Taylor-Phillips

Auto-generated profile, based on publicly available information

Sian Taylor-Phillips

University of Warwick

Research Focus & Metrics

Academic Career Length

19Years (2008-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
18 January 2016 in BMJ Open
Verified
Journal Article
Open access
PDF file
STSian Taylor-Phillips
Sian Taylor-Phillips
KFKaroline Freeman
Karoline Freeman
JGJulia Geppert
Julia Geppert
AAAdeola Agbebiyi
Adeola Agbebiyi
OUOlalekan Uthman
Olalekan Uthman
ACAileen Clarke
Aileen Clarke
Objective To measure test accuracy of non-invasive prenatal testing (NIPT) for Down, Edwards and Patau syndromes using cell-free fetal DNA and identify factors affecting accuracy. Design Systematic review and meta-analysis of published studies. Data sources PubMed, Ovid Medline, Ovid Embase and the Cochrane Library published from 1997 to 9 February 2015, followed by weekly autoalerts until 1 April 2015. Eligibility criteria for selecting studies English language journal articles describing case–control studies with ≥15 trisomy cases or cohort studies with ≥50 pregnant women who had been given NIPT and a reference standard. Results 41, 37 and 30 studies of 2012 publications retrieved were included in the review for Down, Edwards and Patau syndromes. Quality appraisal identified high risk of bias in included studies, funnel plots showed evidence of publication bias. Pooled sensitivity was 99.3% (95% CI 98.9% to 99.6%) for Down, 97.4% (95.8% to 98.4%) for Edwards, and 97.4% (86.1% to 99.6%) for Patau syndrome. The pooled specificity was 99.9% (99.9% to 100%) for all three trisomies. In 100 000 pregnancies in the general obstetric population we would expect 417, 89 and 40 cases of Downs, Edwards and Patau syndromes to be detected by NIPT, with 94, 154 and 42 false positive results. Sensitivity was lower in twin than singleton pregnancies, reduced by 9% for Down, 28% for Edwards and 22% for Patau syndrome. Pooled sensitivity was also lower in the first trimester of pregnancy, in studies in the general obstetric population, and in cohort studies with consecutive enrolment. Conclusions NIPT using cell-free fetal DNA has very high sensitivity and specificity for Down syndrome, with slightly lower sensitivity for Edwards and Patau syndrome. However, it is not 100% accurate and should not be used as a final diagnosis for positive cases. Trial registration number CRD42014014947.

1 September 2021 in BMJ
Verified
Journal Article
Open access
PDF file
KFKaroline Freeman
Karoline Freeman
JGJulia Geppert
Julia Geppert
CSChris Stinton
Chris Stinton
DTDaniel Todkill
Daniel Todkill
SJSamantha Johnson
Samantha Johnson
STSian Taylor-Phillips
Sian Taylor-Phillips
Objective To examine the accuracy of artificial intelligence (AI) for the detection of breast cancer in mammography screening practice. Design Systematic review of test accuracy studies. Data sources Medline, Embase, Web of Science, and Cochrane Database of Systematic Reviews from 1 January 2010 to 17 May 2021. Eligibility criteria Studies reporting test accuracy of AI algorithms, alone or in combination with radiologists, to detect cancer in women’s digital mammograms in screening practice, or in test sets. Reference standard was biopsy with histology or follow-up (for screen negative women). Outcomes included test accuracy and cancer type detected. Study selection and synthesis Two reviewers independently assessed articles for inclusion and assessed the methodological quality of included studies using the QUality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool. A single reviewer extracted data, which were checked by a second reviewer. Narrative data synthesis was performed. Results Twelve studies totalling 131 822 screened women were included. No prospective studies measuring test accuracy of AI in screening practice were found. Studies were of poor methodological quality. Three retrospective studies compared AI systems with the clinical decisions of the original radiologist, including 79 910 women, of whom 1878 had screen detected cancer or interval cancer within 12 months of screening. Thirty four (94%) of 36 AI systems evaluated in these studies were less accurate than a single radiologist, and all were less accurate than consensus of two or more radiologists. Five smaller studies (1086 women, 520 cancers) at high risk of bias and low generalisability to the clinical context reported that all five evaluated AI systems (as standalone to replace radiologist or as a reader aid) were more accurate than a single radiologist reading a test set in the laboratory. In three studies, AI used for triage screened out 53%, 45%, and 50% of women at low risk but also 10%, 4%, and 0% of cancers detected by radiologists. Conclusions Current evidence for AI does not yet allow judgement of its accuracy in breast cancer screening programmes, and it is unclear where on the clinical pathway AI might be of most benefit. AI systems are not sufficiently specific to replace radiologist double reading in screening programmes. Promising results in smaller studies are not replicated in larger studies. Prospective studies are required to measure the effect of AI in clinical practice. Such studies will require clear stopping rules to ensure that AI does not reduce programme specificity. Study registration Protocol registered as PROSPERO CRD42020213590.

30 November 2021 in BMJ Open
Verified
Journal Article
Open access
PDF file
RMRebecca Mottram
Rebecca Mottram
WKWendy Lynn Knerr
Wendy Lynn Knerr
DGDaniel Gallacher
Daniel Gallacher
HFHannah Fraser
Hannah Fraser
LALena Al-Khudairy
Lena Al-Khudairy
ACAileen Clarke
Aileen Clarke
Objective Attendance at population-based breast cancer (mammographic) screening varies. This comprehensive systematic review and meta-analysis assesses all identified patient-level factors associated with routine population breast screening attendance. Design CINAHL, Cochrane Library, Embase, Medline, OVID, PsycINFO and Web of Science were searched for studies of any design, published January 1987–June 2019, and reporting attendance in relation to at least one patient-level factor. Data synthesis Independent reviewers performed screening, data extraction and quality appraisal. OR and 95% CIs were calculated for attendance for each factor and random-effects meta-analysis was undertaken where possible. Results Of 19 776 studies, 335 were assessed at full text and 66 studies (n=22 150 922) were included. Risk of bias was generally low. In meta-analysis, increased attendance was associated with higher socioeconomic status (SES) (n=11 studies; OR 1.45, 95% CI: 1.20 to 1.75); higher income (n=5 studies; OR 1.96, 95% CI: 1.68 to 2.29); home ownership (n=3 studies; OR 2.16, 95% CI: 2.08 to 2.23); being non-immigrant (n=7 studies; OR 2.23, 95% CI: 2.00 to 2.48); being married/cohabiting (n=7 studies; OR 1.86, 95% CI: 1.58 to 2.19) and medium (vs low) level of education (n=6 studies; OR 1.24, 95% CI: 1.09 to 1.41). Women with previous false-positive results were less likely to reattend (n=6 studies; OR 0.77, 95% CI: 0.68 to 0.88). There were no differences by age group or by rural versus urban residence. Conclusions Attendance was lower in women with lower SES, those who were immigrants, non-homeowners and those with previous false-positive results. Variations in service delivery, screening programmes and study populations may influence findings. Our findings are of univariable associations. Underlying causes of lower uptake such as practical, physical, psychological or financial barriers should be investigated. Trial registration number CRD42016051597.

23 October 2017 in BMC Cancer
Verified
Journal Article
Open access
PDF file
FCFor the UK Early Cancer Detection Consortium
For the UK Early Cancer Detection Consortium
ICIan A. Cree
Ian A. Cree
LULesley Uttley
Lesley Uttley
HWHelen Buckley Woods
Helen Buckley Woods
HKHugh Kikuchi
Hugh Kikuchi
JSJacqui Shaw
Jacqui Shaw
The presence of circulating cell-free DNA from tumours in blood (ctDNA) is of major importance to those interested in early cancer detection, as well as to those wishing to monitor tumour progression or diagnose the presence of activating mutations to guide treatment. In 2014, the UK Early Cancer Detection Consortium undertook a systematic mapping review of the literature to identify blood-based biomarkers with potential for the development of a non-invasive blood test for cancer screening, and which identified this as a major area of interest. This review builds on the mapping review to expand the ctDNA dataset to examine the best options for the detection of multiple cancer types. The original mapping review was based on comprehensive searches of the electronic databases Medline, Embase, CINAHL, the Cochrane library, and Biosis to obtain relevant literature on blood-based biomarkers for cancer detection in humans (PROSPERO no. CRD42014010827). The abstracts for each paper were reviewed to determine whether validation data were reported, and then examined in full. Publications concentrating on monitoring of disease burden or mutations were excluded. The search identified 94 ctDNA studies meeting the criteria for review. All but 5 studies examined one cancer type, with breast, colorectal and lung cancers representing 60% of studies. The size and design of the studies varied widely. Controls were included in 77% of publications. The largest study included 640 patients, but the median study size was 65 cases and 35 controls, and the bulk of studies (71%) included less than 100 patients. Studies either estimated cfDNA levels non-specifically or tested for cancer-specific mutations or methylation changes (the majority using PCR-based methods). We have systematically reviewed ctDNA blood biomarkers for the early detection of cancer. Pre-analytical, analytical, and post-analytical considerations were identified which need to be addressed before such biomarkers enter clinical practice. The value of small studies with no comparison between methods, or even the inclusion of controls is highly questionable, and larger validation studies will be required before such methods can be considered for early cancer detection.

1 April 2017 in Health Technology Assessment
Verified
Journal Article
Open access
PDF file
KFKaroline Freeman
Karoline Freeman
HMHema Mistry
Hema Mistry
ATAlexander Tsertsvadze
Alexander Tsertsvadze
PRPam Royle
Pam Royle
NMNoel McCarthy
Noel McCarthy
JMJames Mason
James Mason
Background: Gastroenteritis is a common, transient disorder usually caused by infection and characterised by the acute onset of diarrhoea. Multiplex gastrointestinal pathogen panel (GPP) tests simultaneously identify common bacterial, viral and parasitic pathogens using molecular testing. By providing test results more rapidly than conventional testing methods, GPP tests might positively influence the treatment and management of patients presenting in hospital or in the community.Objective: To systematically review the evidence for GPP tests [xTAG®(Luminex, Toronto, ON, Canada), FilmArray (BioFire Diagnostics, Salt Lake City, UT, USA) and Faecal Pathogens B (AusDiagnostics, Beaconsfield, NSW, Australia)] and to develop a de novo economic model to compare the cost-effectiveness of GPP tests with conventional testing in England and Wales.Data sources: Multiple electronic databases including MEDLINE, EMBASE, Web of Science and the Cochrane Database were searched from inception to January 2016 (with supplementary searches of other online resources).Review methods: Eligible studies included patients with acute diarrhoea; comparing GPP tests with standard microbiology techniques; and patient, management, test accuracy or cost-effectiveness outcomes. Quality assessment of eligible studies used tailored Quality Assessment of Diagnostic Accuracy Studies-2, Consolidated Health Economic Evaluation Reporting Standards and Philips checklists. The meta-analysis included positive and negative agreement estimated for each pathogen. A de novo decision tree model compared patients managed with GPP testing or comparable coverage with patients managed using conventional tests, within the Public Health England pathway. Economic models included hospital and community management of patients with suspected gastroenteritis. The model estimated costs (in 2014/15 prices) and quality-adjusted life-year losses from a NHS and Personal Social Services perspective.Results: Twenty-three studies informed the review of clinical evidence (17 xTAG, four FilmArray, two xTAG and FilmArray, 0 Faecal Pathogens B). No study provided an adequate reference standard with which to compare the test accuracy of GPP with conventional tests. A meta-analysis (of 10 studies) found considerable heterogeneity; however, GPP testing produces a greater number of pathogen-positive findings than conventional testing. It is unclear whether or not these additional ‘positives’ are clinically important. The review identified no robust evidence to inform consequent clinical management of patients. There is considerable uncertainty about the cost-effectiveness of GPP panels used to test for suspected infectious gastroenteritis in hospital and community settings. Uncertainties in the model include length of stay, assumptions about false-positive findings and the costs of tests. Although there is potential for cost-effectiveness in both settings, key modelling assumptions need to be verified and model findings remain tentative.Limitations: No test–treat trials were retrieved. The economic model reflects one pattern of care, which will vary across the NHS.Conclusions: The systematic review and cost-effectiveness model identify uncertainties about the adoption of GPP tests within the NHS. GPP testing will generally correctly identify pathogens identified by conventional testing; however, these tests also generate considerable additional positive results of uncertain clinical importance.Future work: An independent reference standard may not exist to evaluate alternative approaches to testing. A test–treat trial might ascertain whether or not additional GPP ‘positives’ are clinically important or result in overdiagnoses, whether or not earlier diagnosis leads to earlier discharge in patients and what the health consequences of earlier intervention are. Future work might also consider the public health impact of different testing treatments, as test results form the basis for public health surveillance.Study registration: This study is registered as PROSPERO CRD2016033320.Funding: The National Institute for Health Research Health Technology Assessment programme.

1 August 2013 in Health Technology Assessment
Verified
Journal Article
Closed access
PDF file
NWNr Waugh
Nr Waugh
DSD Shyangdan
D Shyangdan
STS Taylor-Phillips
S Taylor-Phillips
GSG Suri
G Suri
BHB Hall
B Hall
Background The prevalence of type 2 diabetes mellitus (T2DM) has been increasing, owing to increases in overweight and obesity, decreasing physical activity and the changing demographic structure of the population. People can develop T2DM without symptoms and up to 20% may be undiagnosed. They may have diabetic complications, such as retinopathy, by the time they are diagnosed, or may suffer a heart attack, without warning. Undiagnosed diabetes can be detected by raised blood glucose levels. Aim The aim of this review was to provide an update for the UK National Screening Committee (NSC) on screening for T2DM. Methods As this review was undertaken to update a previous Health Technology Assessment review published in 2007, and a more recent Scottish Public Health Network review, searches for evidence were restricted from 2009 to end of January 2012, with selected later studies added. The databases searched were MEDLINE, EMBASE, MEDLINE-in-Process & Other Non-Indexed Citations, Science Citation Index and Conference Proceedings Citation Index. The case for screening was considered against the criteria used by the NSC to assess proposed population screening programmes. Results Population screening for T2DM does not meet all of the NSC criteria. Criterion 12, on optimisation of existing management, has not been met. A report by the National Audit Office (NAO) gives details of shortcomings. Criterion 13 requires evidence from high-quality randomised controlled trials that screening is beneficial. This has not been met. The Ely trial of screening showed no benefit. The ADDITION trial was not a trial of screening, but showed no benefit in cardiovascular outcomes from intensive management in people with screen-detected T2DM. Criterion 18 on staffing and facilities does not appear to have been met, according to the NAO report. Criterion 19 requires that all other options, including prevention, should have been considered. A large proportion of cases of T2DM could be prevented if people avoided becoming overweight or obese. The first stage of selection would use risk factors, using data held on general practitioner computer systems, using the QDiabetes Risk Score, or by sending out questionnaires, using the Finnish Diabetes Risk Score (FINDRISC). Those at high risk would have a measure of blood glucose. There is no perfect screening test. Glycated haemoglobin (HbA1c) testing has advantages in not requiring a fasting sample, and because it is a predictor of vascular disease across a wider range than just the diabetic one. However, it lacks sensitivity and would miss some people with diabetes. Absolute values of HbA1c may be more useful as part of overall risk assessment than a dichotomous ‘diabetes or not diabetes’ diagnosis. The oral glucose tolerance test is more sensitive, but inconvenient, more costly, has imperfect reproducibility and is less popular, meaning that uptake would be lower. Conclusions When considered against the NSC criteria, the case for screening is less strong than it was in the 2007 review. The main reason is the absence of cardiovascular benefit in the two trials published since the previous review. There is a case for selective screening as part of overall vascular risk assessment. Population screening for T2DM does not meet all of the NSC criteria. Funding The National Institute for Health Research Health Technology Assessment programme.