Matthew Williams
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Matthew Williams has a total of 574 co-authors. Here are the top co-authors by number of shared publications:
Amanda Johnston
9 shared publicationsBor-Wen Wu
9 shared publicationsFiona P. Harris
9 shared publicationsJane Evanson
9 shared publicationsJim A. Thomson
9 shared publicationsJohn S. Bomalaski
9 shared publications
Latest Publications
- Open Access
- 19 June 2026
- Published by Elsevier in Clinical Oncology
- Vol. 57, 104212
Patterns and Predictors of Dying at Home for Adults With Primary Malignant Brain Tumours: A Comprehensive National Cohort Analysis
- 19 June 2026
- Published by Elsevier in Clinical Oncology
- Vol. 57, 104212
Open AccessPatterns and Predictors of Dying at Home for Adults With Primary Malignant Brain Tumours: A Comprehensive National Cohort Analysis
- Open Access
- 25 April 2026
- Published by Springer Nature in Journal of Neuro-Oncology
- Vol. 177 (3)
Variation in rates of post-operative oncological treatment for patients with glioblastoma in England: a comprehensive multi-year National cohort study from the GlioCova project
Purpose: Although there are consistent guidelines of standardised glioblastoma patient care, there may be variation in treatments delivered due to differences in patient characteristics or adherence to guidelines across centres. This study explores the variation in treatments for glioblastoma patients across centres in a comprehensive multi-year brain tumour patient cohort in England. Methods: We analysed data from the Gliocova project that contains over 50,000 adult brain tumour patients diagnosed between 2013 and 2018 in England. We selected a glioblastoma patient cohort and explored treatment rates, systemic predictors of treatment, variation in treatment across surgical centres and survival rates for different treatment groups. Results: We analysed data from 11 359 patients with a histological diagnosis of GBM. Almost 80% of glioblastoma patients received at least some treatment after brain surgery, with 40% receiving the guideline recommended aggressive treatment (47% in the under-70s cohort). Age, sex, deprivation status, comorbidities, surgery type (resection versus biopsy) and ethnicity were identified as systemic predictors of receipt of post-operative treatment. There was also variation in receipt of post-operative treatment (64%-86%) and receipt of aggressive treatment (16%-65%) across centres. Patients receiving aggressive treatment had highest survival (15.6 months); in patients receiving any post operative treatment survival was 12.5 months and in patients receiving surgery only survival was 2.5 months. Conclusion: 22% of glioblastoma patients undergoing surgery receive no further treatment and there is variation in post-surgical treatment both at centre and individual patient level. We suggest that centres should measure and report rates of no further treatment as part of standard clinical governance.
- 25 April 2026
- Published by Springer Nature in Journal of Neuro-Oncology
- Vol. 177 (3)
Open AccessVariation in rates of post-operative oncological treatment for patients with glioblastoma in England: a comprehensive multi-year National cohort study from the GlioCova project
Purpose: Although there are consistent guidelines of standardised glioblastoma patient care, there may be variation in treatments delivered due to differences in patient characteristics or adherence to guidelines across centres. This study explores the variation in treatments for glioblastoma patients across centres in a comprehensive multi-year brain tumour patient cohort in England. Methods: We analysed data from the Gliocova project that contains over 50,000 adult brain tumour patients diagnosed between 2013 and 2018 in England. We selected a glioblastoma patient cohort and explored treatment rates, systemic predictors of treatment, variation in treatment across surgical centres and survival rates for different treatment groups. Results: We analysed data from 11 359 patients with a histological diagnosis of GBM. Almost 80% of glioblastoma patients received at least some treatment after brain surgery, with 40% receiving the guideline recommended aggressive treatment (47% in the under-70s cohort). Age, sex, deprivation status, comorbidities, surgery type (resection versus biopsy) and ethnicity were identified as systemic predictors of receipt of post-operative treatment. There was also variation in receipt of post-operative treatment (64%-86%) and receipt of aggressive treatment (16%-65%) across centres. Patients receiving aggressive treatment had highest survival (15.6 months); in patients receiving any post operative treatment survival was 12.5 months and in patients receiving surgery only survival was 2.5 months. Conclusion: 22% of glioblastoma patients undergoing surgery receive no further treatment and there is variation in post-surgical treatment both at centre and individual patient level. We suggest that centres should measure and report rates of no further treatment as part of standard clinical governance.
- Open Access
- 20 April 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Investigating Digital Outcomes (Indigo) A Randomised Observational Digital Trial: Results of Pilot Study (Preprint)
BACKGROUND There are more than 3M people in the UK living with a diagnosis of cancer. These are a varied group of people, ranging from those who had curative surgery 20 years ago to those who have exhausted all options and are under palliative care. However, the prevalent population is much larger than the incident population, and is under-researched, and we know very little about their healthcare utilisation or quality of life. In addition, there are open methodological questions about which questionnaires are best to use and how best to ask people for this information. OBJECTIVE To assess the feasibility of implementing, measuring and assessing the impact of timing on completion of questions about data linkage and the performance of different approaches to collecting quality of life and healthcare utilisation data in cancer survivors in a scalable robust manner via a pilot clinical trial. METHODS We designed and implemented an online randomised observational clinical study using an electronic survey platform. We provided trial information and obtained informed consent from all participants online and used randomisation to assess the acceptability of different quality of life tool and to assess the impact of timing on completion of questions about data linkage. We piloted the study in Northwest London and compared our trial population to the underlying patient population from the WSIC dataset. RESULTS 135/345 GP practices in NWL participated, and sent SMS messages to 25044 people, of whom 2518 accessed the survey (9%); in addition, 761 accessed the survey via email and a further 15 participants via social media. 1600 provided informed consent and completed the EQ-5D-5L questionnaire and these formed our analytical cohort. The median age of participants was 68 years; 46% were female, 72% heterosexual. 15% from an ethnic minority. The commonest tumour sites were breast, prostate and colorectal cancer (respectively 35%, 15% and 11%). 67% of participants had been diagnosed more than 5 years ago. 90% described themselves as cured. 46% had accessed healthcare services in the last 12 months. Of those working, 33% had reduced their hours because of their diagnosis. 22% were disabled and 33% reported that the standard QoL questionnaires did not cover their cancer-related symptoms and participant rating of additional QoL questionnaires was broadly similar. 74% agreed to linkage with national cancer data. CONCLUSIONS The implementation, and delivery of an online observational trial are feasible and enable us to recruit a prevalent cancer population, 2/3rds of whom had been diagnosed > 5 years ago. Most participants are happy to allow linkage to national cancer data. We used this pilot phase to refine some of search criteria, and the trial is currently recruiting nationally. CLINICALTRIAL NCT06095024: Investigating Digital Outcomes for Cancer Survivors in the Community. INTERNATIONAL REGISTERED REPORT RR2-https://doi.org/10.1136/bmjopen-2025-104336
- 20 April 2026
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessInvestigating Digital Outcomes (Indigo) A Randomised Observational Digital Trial: Results of Pilot Study (Preprint)
BACKGROUND There are more than 3M people in the UK living with a diagnosis of cancer. These are a varied group of people, ranging from those who had curative surgery 20 years ago to those who have exhausted all options and are under palliative care. However, the prevalent population is much larger than the incident population, and is under-researched, and we know very little about their healthcare utilisation or quality of life. In addition, there are open methodological questions about which questionnaires are best to use and how best to ask people for this information. OBJECTIVE To assess the feasibility of implementing, measuring and assessing the impact of timing on completion of questions about data linkage and the performance of different approaches to collecting quality of life and healthcare utilisation data in cancer survivors in a scalable robust manner via a pilot clinical trial. METHODS We designed and implemented an online randomised observational clinical study using an electronic survey platform. We provided trial information and obtained informed consent from all participants online and used randomisation to assess the acceptability of different quality of life tool and to assess the impact of timing on completion of questions about data linkage. We piloted the study in Northwest London and compared our trial population to the underlying patient population from the WSIC dataset. RESULTS 135/345 GP practices in NWL participated, and sent SMS messages to 25044 people, of whom 2518 accessed the survey (9%); in addition, 761 accessed the survey via email and a further 15 participants via social media. 1600 provided informed consent and completed the EQ-5D-5L questionnaire and these formed our analytical cohort. The median age of participants was 68 years; 46% were female, 72% heterosexual. 15% from an ethnic minority. The commonest tumour sites were breast, prostate and colorectal cancer (respectively 35%, 15% and 11%). 67% of participants had been diagnosed more than 5 years ago. 90% described themselves as cured. 46% had accessed healthcare services in the last 12 months. Of those working, 33% had reduced their hours because of their diagnosis. 22% were disabled and 33% reported that the standard QoL questionnaires did not cover their cancer-related symptoms and participant rating of additional QoL questionnaires was broadly similar. 74% agreed to linkage with national cancer data. CONCLUSIONS The implementation, and delivery of an online observational trial are feasible and enable us to recruit a prevalent cancer population, 2/3rds of whom had been diagnosed > 5 years ago. Most participants are happy to allow linkage to national cancer data. We used this pilot phase to refine some of search criteria, and the trial is currently recruiting nationally. CLINICALTRIAL NCT06095024: Investigating Digital Outcomes for Cancer Survivors in the Community. INTERNATIONAL REGISTERED REPORT RR2-https://doi.org/10.1136/bmjopen-2025-104336
- Open Access
- 19 February 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (2) , e0342011
Estimating post-operative complication rates in patients with primary brain tumours from routine administrative data: A national cohort study
Introduction: Neurosurgery is an important element of brain tumour treatment but carries with it the risk of complications. Previous work has defined a narrow set of general post-operative complications which are used as Patient Safety Indicators (PSIs), but these are not brain tumour specific and do not capture the full range of complications. As a result, there is no way of measuring post-operative complications in neurosurgery at a national level. Methods: We conducted a retrospective, observational cohort study using a comprehensive national administrative dataset from England on adult brain or spinal tumour patients to better define post-operative complications. We generated and validated a new list of post-operative complications – ICL list. The ICL list contains codes selected specifically from our Gliocova dataset combined with general OECD-defined PSI list. The ICL list is novel as it can assess brain tumour patient complications using an administrative dataset at a national level and captures more specific brain tumour related complications. Results: In our study, 30-day readmission after surgery was 12.7% and 30-day mortality was 2.3%. The ICL list of complications identified many more patients with complications (N = 3,274 (11.3%)) compared to OECD-defined PSI list (N = 568 (2.0%)) without reducing model performance. 30-day mortality was 6.5% in those with complications and 1.8% in those without. Discussion: We have identified a much wider set of complications than the OECD-defined PSIs and shown that patients developing these have worse outcomes than those without complications. This enables us to estimate the risk of post-operative complications in brain tumour patients using national administrative data. It forms the basis for planned further work, allowing us to explore the predictors of and consequences of post-operative complications.
- 19 February 2026
- Published by Public Library of Science (PLoS) in PLOS ONE
- Vol. 21 (2) , e0342011
Open AccessEstimating post-operative complication rates in patients with primary brain tumours from routine administrative data: A national cohort study
Introduction: Neurosurgery is an important element of brain tumour treatment but carries with it the risk of complications. Previous work has defined a narrow set of general post-operative complications which are used as Patient Safety Indicators (PSIs), but these are not brain tumour specific and do not capture the full range of complications. As a result, there is no way of measuring post-operative complications in neurosurgery at a national level. Methods: We conducted a retrospective, observational cohort study using a comprehensive national administrative dataset from England on adult brain or spinal tumour patients to better define post-operative complications. We generated and validated a new list of post-operative complications – ICL list. The ICL list contains codes selected specifically from our Gliocova dataset combined with general OECD-defined PSI list. The ICL list is novel as it can assess brain tumour patient complications using an administrative dataset at a national level and captures more specific brain tumour related complications. Results: In our study, 30-day readmission after surgery was 12.7% and 30-day mortality was 2.3%. The ICL list of complications identified many more patients with complications (N = 3,274 (11.3%)) compared to OECD-defined PSI list (N = 568 (2.0%)) without reducing model performance. 30-day mortality was 6.5% in those with complications and 1.8% in those without. Discussion: We have identified a much wider set of complications than the OECD-defined PSIs and shown that patients developing these have worse outcomes than those without complications. This enables us to estimate the risk of post-operative complications in brain tumour patients using national administrative data. It forms the basis for planned further work, allowing us to explore the predictors of and consequences of post-operative complications.
- Open Access
- 8 January 2026
- Published by Springer Nature in Journal of Neuro-Oncology
- Vol. 176 (2) , 157
Understanding the difference in symptoms and outcomes between glioblastoma patients diagnosed based on histological or molecular criteria: a retrospective cohort analysis from the Histo-Mol GBM collaborative
Purpose: Since the 2021 World Health Organisation (WHO) classification, glioblastoma could be diagnosed based on classical histological features (hGBM) or molecular criteria (mGBM). However, prior studies included patients who required reclassification as a mGBM, potentially biasing survival analyses. The Histo-Mol GBM collaborative performed an international multicentre retrospective real-world cohort study of glioblastoma patients diagnosed according to WHO CNS 5. Methods: We identified consecutive patients diagnosed in 2021 with IDH wildtype glioblastoma according to WHO CNS 5. Clinicopathological, treatment, and survival data were collected and compared between mGBM and hGBM. Results: 1828 patients diagnosed with glioblastoma were included. 75 mGBM patients (8.4% of tested patients) were identified, with no difference in age (median 61 vs 64, p = 0.057), gender ( p = 0.937), or proportion with performance status 0–1 (82.7% vs 68.3%, p = 0.052) compared to hGBM. mGBM patients had an extended interval from MRI to surgery (median 23 vs 14 days, p < 0.001) and more frequently underwent biopsy (69.3% vs 30.3%, p < 0.001), but equivalent proportions received oncological treatment (80.0% vs 78.7%, p = 0.784). Overall survival (OS) from surgery was not different ( p = 0.063). However, OS from initial MRI, stratified by surgical extent, demonstrated improved OS for mGBM patients (hazard ratio (HR) 0.56, 95% confidence interval (CI): 0.43–0.73). Propensity score matching identified improved survival following resection (HR 0.48, 95% CI: 0.24–0.95; median OS: 26.0 versus 14.0 months, p = 0.031) but not biopsy (HR 1.10, 95% CI: 0.71–1.72). Conclusion: In this large real-world cohort, mGBMs had longer OS than hGBMs following resection with implications for prognostication and clinical decision making.
- 8 January 2026
- Published by Springer Nature in Journal of Neuro-Oncology
- Vol. 176 (2) , 157
Open AccessUnderstanding the difference in symptoms and outcomes between glioblastoma patients diagnosed based on histological or molecular criteria: a retrospective cohort analysis from the Histo-Mol GBM collaborative
Purpose: Since the 2021 World Health Organisation (WHO) classification, glioblastoma could be diagnosed based on classical histological features (hGBM) or molecular criteria (mGBM). However, prior studies included patients who required reclassification as a mGBM, potentially biasing survival analyses. The Histo-Mol GBM collaborative performed an international multicentre retrospective real-world cohort study of glioblastoma patients diagnosed according to WHO CNS 5. Methods: We identified consecutive patients diagnosed in 2021 with IDH wildtype glioblastoma according to WHO CNS 5. Clinicopathological, treatment, and survival data were collected and compared between mGBM and hGBM. Results: 1828 patients diagnosed with glioblastoma were included. 75 mGBM patients (8.4% of tested patients) were identified, with no difference in age (median 61 vs 64, p = 0.057), gender ( p = 0.937), or proportion with performance status 0–1 (82.7% vs 68.3%, p = 0.052) compared to hGBM. mGBM patients had an extended interval from MRI to surgery (median 23 vs 14 days, p < 0.001) and more frequently underwent biopsy (69.3% vs 30.3%, p < 0.001), but equivalent proportions received oncological treatment (80.0% vs 78.7%, p = 0.784). Overall survival (OS) from surgery was not different ( p = 0.063). However, OS from initial MRI, stratified by surgical extent, demonstrated improved OS for mGBM patients (hazard ratio (HR) 0.56, 95% confidence interval (CI): 0.43–0.73). Propensity score matching identified improved survival following resection (HR 0.48, 95% CI: 0.24–0.95; median OS: 26.0 versus 14.0 months, p = 0.031) but not biopsy (HR 1.10, 95% CI: 0.71–1.72). Conclusion: In this large real-world cohort, mGBMs had longer OS than hGBMs following resection with implications for prognostication and clinical decision making.
- 12 December 2025
- Published by Elsevier in Clinical Oncology
- Vol. 50, 104000
Access to Stereotactic Radiosurgery Centres Across the United Kingdom—An Investigation of Travel Time
- 12 December 2025
- Published by Elsevier in Clinical Oncology
- Vol. 50, 104000
Access to Stereotactic Radiosurgery Centres Across the United Kingdom—An Investigation of Travel Time
- Open Access
- 6 November 2025
- Published by Springer Nature in Research Square
Understanding the difference in symptoms and outcomes between glioblastoma patients diagnosed based on histological or molecular criteria: a retrospective cohort analysis from the Histo-Mol GBM collaborative
Purpose: Since the 2021 World Health Organisation (WHO) classification, glioblastoma could be diagnosed based on classical histological features (hGBM) or molecular criteria (mGBM). However, prior studies included patients who required reclassification as a mGBM, potentially biasing survival analyses. The Histo-Mol GBM collaborative performed an international multicentre retrospective real-world cohort study of glioblastoma patients diagnosed according to WHO CNS 5. Methods: We identified consecutive patients diagnosed in 2021 with IDH wildtype glioblastoma according to WHO CNS 5. Clinicopathological, treatment, and survival data were collected and compared between mGBM and hGBM. Results: 1828 patients diagnosed with glioblastoma were included. 75 mGBM patients (8.4% of tested patients) were identified, with no difference in age (median 61 vs 64, p=0.057), gender (p=0.937), or proportion with performance status 0-1 (82.7% vs 68.3%, p=0.052) compared to hGBM. mGBM patients had an extended interval from MRI to surgery (median 23 vs 14 days, p<0.001) and more frequently underwent biopsy (69.3% vs 30.3%, p<0.001), but equivalent proportions received oncological treatment (80.0% vs 78.7%, p=0.784). Overall survival (OS) from surgery was not different (p=0.063). However, OS from initial MRI, stratified by surgical extent, demonstrated improved OS for mGBM patients (hazard ratio (HR) 0.56, 95% confidence interval (CI): 0.43-0.73). Propensity score matching identified improved survival following resection (HR 0.48, 95% CI: 0.24-0.95; median OS: 26.0 versus 14.0 months, p=0.031) but not biopsy (HR 1.10, 95% CI: 0.71-1.72). Conclusion: In this large real-world cohort, mGBMs had longer OS than hGBMs following resection with implications for prognostication and clinical decision making.
- 6 November 2025
- Published by Springer Nature in Research Square
Open AccessUnderstanding the difference in symptoms and outcomes between glioblastoma patients diagnosed based on histological or molecular criteria: a retrospective cohort analysis from the Histo-Mol GBM collaborative
Purpose: Since the 2021 World Health Organisation (WHO) classification, glioblastoma could be diagnosed based on classical histological features (hGBM) or molecular criteria (mGBM). However, prior studies included patients who required reclassification as a mGBM, potentially biasing survival analyses. The Histo-Mol GBM collaborative performed an international multicentre retrospective real-world cohort study of glioblastoma patients diagnosed according to WHO CNS 5. Methods: We identified consecutive patients diagnosed in 2021 with IDH wildtype glioblastoma according to WHO CNS 5. Clinicopathological, treatment, and survival data were collected and compared between mGBM and hGBM. Results: 1828 patients diagnosed with glioblastoma were included. 75 mGBM patients (8.4% of tested patients) were identified, with no difference in age (median 61 vs 64, p=0.057), gender (p=0.937), or proportion with performance status 0-1 (82.7% vs 68.3%, p=0.052) compared to hGBM. mGBM patients had an extended interval from MRI to surgery (median 23 vs 14 days, p<0.001) and more frequently underwent biopsy (69.3% vs 30.3%, p<0.001), but equivalent proportions received oncological treatment (80.0% vs 78.7%, p=0.784). Overall survival (OS) from surgery was not different (p=0.063). However, OS from initial MRI, stratified by surgical extent, demonstrated improved OS for mGBM patients (hazard ratio (HR) 0.56, 95% confidence interval (CI): 0.43-0.73). Propensity score matching identified improved survival following resection (HR 0.48, 95% CI: 0.24-0.95; median OS: 26.0 versus 14.0 months, p=0.031) but not biopsy (HR 1.10, 95% CI: 0.71-1.72). Conclusion: In this large real-world cohort, mGBMs had longer OS than hGBMs following resection with implications for prognostication and clinical decision making.
- Open Access
- 23 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 1498-1542
Hidden Conflicts in Neural Networks and their Implications for Explainability
- 23 June 2025
- Published by Association for Computing Machinery (ACM)
- p. 1498-1542
Open AccessHidden Conflicts in Neural Networks and their Implications for Explainability
- Open Access
- 9 April 2025
- Published by Elsevier in Cancer Epidemiology
- Vol. 97, 102811
Adult glioblastoma in England: Incidence, treatment, and outcomes with novel population-based strata
- 9 April 2025
- Published by Elsevier in Cancer Epidemiology
- Vol. 97, 102811
Open AccessAdult glioblastoma in England: Incidence, treatment, and outcomes with novel population-based strata