April Coombe
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
April Coombe has a total of 51 co-authors. Here are the top co-authors by number of shared publications:
- Amanda Jane Daley
Centre for Lifestyle Medicine and Behaviour, School of Sport, Exercise and Health Sciences, Loughborough University, Loughborough, UK
3 shared publications - David Moore
University of Birmingham, Birmingham, UK
3 shared publications - Jacqueline Dinnes
Department of Applied Health Sciences, College of Medicine and Health, University of Birmingham, Birmingham, UK
3 shared publications - Katie Scandrett
Department of Applied Health Sciences, College of Medicine and Health, University of Birmingham, Birmingham, UK
3 shared publications - Yemisi Takwoingi
Department of Applied Health Sciences, University of Birmingham, and National Institute for Health and Care Research (NIHR) Birmingham Biomedical Research Centre, Birmingham, United Kingdom (C.F.D., Y.T., J.J.D.)
3 shared publications - Aditya Uday Kale
College of Medicine and Health, University of Birmingham , Birmingham, B15 2TT
2 shared publications
Latest Publications
- Open Access
- 1 January 2026
- Published by National Institute for Health and Care Research in Health Technology Assessment
- p. 1-48
Methods and mechanisms for measuring and monitoring outcomes from newborn bloodspot screening: a scoping review
Background: Newborn bloodspot screening offers the potential to detect rare diseases early, enabling timely treatment that can reduce mortality and morbidity. Generating evidence for rare diseases often depends on observational data, making it challenging to formulate recommendations for new screening programmes and evaluate the effectiveness of existing ones. Objective(s): To identify the range of methods and mechanisms used to measure and monitor outcomes from newborn screening programmes using a scoping review. Methods: We included studies published between 2019 and 2024, which evaluated a current or candidate newborn screening programme, or which reported outcomes in screen-detected cases. Studies were categorised into four groups: group 1 reported a comparison and follow-up; group 2 reported a comparison but no follow-up; group 3 reported no comparison with follow-up; and group 4 reported no comparison or follow-up. Data were extracted from a random sample of studies within each group; studies in group 1 were prioritised. Results were reported narratively according to study group. The review was conducted and reported according to current guidance for scoping reviews. Data sources: EMBASE (Ovid), MEDLINE (Ovid) and Science Citation Index (Web of Science – Clarivate). Results: We included 574 primary studies and extracted data from 178. Of the 75 studies in group 1, most compared screen-detected cases with controls (74%). Studies in this group used newborn bloodspot programme databases, registries or record review to identify participants and outcomes; only six (8%) reported use of record linkage. Studies in group 2 ( n = 31) mostly reported comparisons of screening tests (25, 81%). Over half of studies in group 3 ( n = 34) used newborn bloodspot programme databases to identify participants (53%) and outcomes (65%). A similar pattern was seen in the group 4 ( n = 38). Studies reporting follow-up typically relied on retrospective record review or were not well reported. Across all study groups, data on accuracy, epidemiology and genetic variants were common. Studies in group 1 also reported on the effectiveness of newborn bloodspot screening (32/75, 43%), treatment effectiveness (20%) or harms of newborn bloodspot screening (3%). Limitations: Restricting data extraction to a random sample of studies risks missing novel methods or mechanisms. Conclusions: Many studies reported test accuracy metrics and genetic variants in newborn screening. Some data on programme effectiveness were identified, but assessment of potential harms remains limited, and methods for follow-up were poorly reported. Assessment of harms, including overdiagnosis and psychological impact, is crucial to ensuring a net benefit at the population level. Future work: In a second phase of work, an in-depth assessment of studies using different methods and mechanisms will be conducted to identify the extent to which they can provide outcome data to inform the evaluation of ongoing and candidate screening programmes. Funding: This article presents independent research funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme as award number NIHR167910.
- 1 January 2026
- Published by National Institute for Health and Care Research in Health Technology Assessment
- p. 1-48
Open AccessMethods and mechanisms for measuring and monitoring outcomes from newborn bloodspot screening: a scoping review
Background: Newborn bloodspot screening offers the potential to detect rare diseases early, enabling timely treatment that can reduce mortality and morbidity. Generating evidence for rare diseases often depends on observational data, making it challenging to formulate recommendations for new screening programmes and evaluate the effectiveness of existing ones. Objective(s): To identify the range of methods and mechanisms used to measure and monitor outcomes from newborn screening programmes using a scoping review. Methods: We included studies published between 2019 and 2024, which evaluated a current or candidate newborn screening programme, or which reported outcomes in screen-detected cases. Studies were categorised into four groups: group 1 reported a comparison and follow-up; group 2 reported a comparison but no follow-up; group 3 reported no comparison with follow-up; and group 4 reported no comparison or follow-up. Data were extracted from a random sample of studies within each group; studies in group 1 were prioritised. Results were reported narratively according to study group. The review was conducted and reported according to current guidance for scoping reviews. Data sources: EMBASE (Ovid), MEDLINE (Ovid) and Science Citation Index (Web of Science – Clarivate). Results: We included 574 primary studies and extracted data from 178. Of the 75 studies in group 1, most compared screen-detected cases with controls (74%). Studies in this group used newborn bloodspot programme databases, registries or record review to identify participants and outcomes; only six (8%) reported use of record linkage. Studies in group 2 ( n = 31) mostly reported comparisons of screening tests (25, 81%). Over half of studies in group 3 ( n = 34) used newborn bloodspot programme databases to identify participants (53%) and outcomes (65%). A similar pattern was seen in the group 4 ( n = 38). Studies reporting follow-up typically relied on retrospective record review or were not well reported. Across all study groups, data on accuracy, epidemiology and genetic variants were common. Studies in group 1 also reported on the effectiveness of newborn bloodspot screening (32/75, 43%), treatment effectiveness (20%) or harms of newborn bloodspot screening (3%). Limitations: Restricting data extraction to a random sample of studies risks missing novel methods or mechanisms. Conclusions: Many studies reported test accuracy metrics and genetic variants in newborn screening. Some data on programme effectiveness were identified, but assessment of potential harms remains limited, and methods for follow-up were poorly reported. Assessment of harms, including overdiagnosis and psychological impact, is crucial to ensuring a net benefit at the population level. Future work: In a second phase of work, an in-depth assessment of studies using different methods and mechanisms will be conducted to identify the extent to which they can provide outcome data to inform the evaluation of ongoing and candidate screening programmes. Funding: This article presents independent research funded by the National Institute for Health and Care Research (NIHR) Health Technology Assessment programme as award number NIHR167910.
- Open Access
- 26 November 2025
- Published by Springer Nature in Journal of Experimental Criminology
- p. 1-24
Information retrieval for a global systematic map on interventions to reduce gang violence: reflection and analysis from an update search using 55 resources
Background: Systematic searches for research within the field of crime interventions are challenging. There is a lack of examples to inform how and where to search, and which resources can help identify research from countries across the world. Updating a systematic map of interventions to reduce gang-related violence included searching 55 resources to identify research published since 2005. The search strategy used in the original review required revision and efforts were made to identify relevant research published outside scholarly journals, and research conducted outside of the USA. The updated search included conventional database searches, network graph searches (citation- and related-item-searching using the former Microsoft Academic Graph), and website searches. Objective: The objective of this study is to describe the design of the database update searches and to understand which information resources were useful in identifying relevant records, non-journal records, and records of research conducted outside of the USA. Methods: (1) We reflected on the search strategy design process, which was informed from analysing the records in the original systematic map and exploratory searches. (2) We focussed on 157 new records in the updated map that met the eligibility criteria of the original map and matched these to the search results obtained from each information resource. Findings: At the planning stage it appeared necessary to search many resources and to use multiple concepts to capture the diverse literature. The 157 relevant records were found across 37 of the 55 information resources searched. These resources spanned multiple research fields. Eleven resources each provided at least 10% of the records, collectively yielding 80% of the records. Citation and related-item searches provided 24 records not identified elsewhere. Just 20% (32/157) of records described studies conducted outside the USA. This included 11% (17) from the UK and 6% (10) from low- and middle-income countries (LMIC). There was a lack of resources that supported identification of studies from LMIC. Most of the non-US grey literature comprised reports conducted in the UK. Conclusions: Information resources change over time, though the principle of taking a multi-stranded approach to the choice of search terms and selection of information resources appears essential to find intervention research within the field of criminology. Boolean text-based searches can be strengthened by using network graph searching (citation and related-items searches). The findings emphasise challenges for locating research published outside of the USA. Newer resources are worth investigating for their potential in addressing this challenge. We recommend those funding, producing and disseminating research consider how to enhance its visibility and reduce research waste.
- 26 November 2025
- Published by Springer Nature in Journal of Experimental Criminology
- p. 1-24
Open AccessInformation retrieval for a global systematic map on interventions to reduce gang violence: reflection and analysis from an update search using 55 resources
Background: Systematic searches for research within the field of crime interventions are challenging. There is a lack of examples to inform how and where to search, and which resources can help identify research from countries across the world. Updating a systematic map of interventions to reduce gang-related violence included searching 55 resources to identify research published since 2005. The search strategy used in the original review required revision and efforts were made to identify relevant research published outside scholarly journals, and research conducted outside of the USA. The updated search included conventional database searches, network graph searches (citation- and related-item-searching using the former Microsoft Academic Graph), and website searches. Objective: The objective of this study is to describe the design of the database update searches and to understand which information resources were useful in identifying relevant records, non-journal records, and records of research conducted outside of the USA. Methods: (1) We reflected on the search strategy design process, which was informed from analysing the records in the original systematic map and exploratory searches. (2) We focussed on 157 new records in the updated map that met the eligibility criteria of the original map and matched these to the search results obtained from each information resource. Findings: At the planning stage it appeared necessary to search many resources and to use multiple concepts to capture the diverse literature. The 157 relevant records were found across 37 of the 55 information resources searched. These resources spanned multiple research fields. Eleven resources each provided at least 10% of the records, collectively yielding 80% of the records. Citation and related-item searches provided 24 records not identified elsewhere. Just 20% (32/157) of records described studies conducted outside the USA. This included 11% (17) from the UK and 6% (10) from low- and middle-income countries (LMIC). There was a lack of resources that supported identification of studies from LMIC. Most of the non-US grey literature comprised reports conducted in the UK. Conclusions: Information resources change over time, though the principle of taking a multi-stranded approach to the choice of search terms and selection of information resources appears essential to find intervention research within the field of criminology. Boolean text-based searches can be strengthened by using network graph searching (citation and related-items searches). The findings emphasise challenges for locating research published outside of the USA. Newer resources are worth investigating for their potential in addressing this challenge. We recommend those funding, producing and disseminating research consider how to enhance its visibility and reduce research waste.
- Open Access
- 28 June 2024
- Published by JMIR Publications Inc. in JMIR Research Protocols
- Vol. 13 (1) , e51614
Detecting Algorithmic Errors and Patient Harms for AI-Enabled Medical Devices in Randomized Controlled Trials: Protocol for a Systematic Review
Background: Artificial intelligence (AI) medical devices have the potential to transform existing clinical workflows and ultimately improve patient outcomes. AI medical devices have shown potential for a range of clinical tasks such as diagnostics, prognostics, and therapeutic decision-making such as drug dosing. There is, however, an urgent need to ensure that these technologies remain safe for all populations. Recent literature demonstrates the need for rigorous performance error analysis to identify issues such as algorithmic encoding of spurious correlations (eg, protected characteristics) or specific failure modes that may lead to patient harm. Guidelines for reporting on studies that evaluate AI medical devices require the mention of performance error analysis; however, there is still a lack of understanding around how performance errors should be analyzed in clinical studies, and what harms authors should aim to detect and report. Objective: This systematic review will assess the frequency and severity of AI errors and adverse events (AEs) in randomized controlled trials (RCTs) investigating AI medical devices as interventions in clinical settings. The review will also explore how performance errors are analyzed including whether the analysis includes the investigation of subgroup-level outcomes. Methods: This systematic review will identify and select RCTs assessing AI medical devices. Search strategies will be deployed in MEDLINE (Ovid), Embase (Ovid), Cochrane CENTRAL, and clinical trial registries to identify relevant papers. RCTs identified in bibliographic databases will be cross-referenced with clinical trial registries. The primary outcomes of interest are the frequency and severity of AI errors, patient harms, and reported AEs. Quality assessment of RCTs will be based on version 2 of the Cochrane risk-of-bias tool (RoB2). Data analysis will include a comparison of error rates and patient harms between study arms, and a meta-analysis of the rates of patient harm in control versus intervention arms will be conducted if appropriate. Results: The project was registered on PROSPERO in February 2023. Preliminary searches have been completed and the search strategy has been designed in consultation with an information specialist and methodologist. Title and abstract screening started in September 2023. Full-text screening is ongoing and data collection and analysis began in April 2024. Conclusions: Evaluations of AI medical devices have shown promising results; however, reporting of studies has been variable. Detection, analysis, and reporting of performance errors and patient harms is vital to robustly assess the safety of AI medical devices in RCTs. Scoping searches have illustrated that the reporting of harms is variable, often with no mention of AEs. The findings of this systematic review will identify the frequency and severity of AI performance errors and patient harms and generate insights into how errors should be analyzed to account for both overall and subgroup performance. Trial Registration: PROSPERO CRD42023387747; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=387747 International Registered Report Identifier (IRRID): PRR1-10.2196/51614
- 28 June 2024
- Published by JMIR Publications Inc. in JMIR Research Protocols
- Vol. 13 (1) , e51614
Open AccessDetecting Algorithmic Errors and Patient Harms for AI-Enabled Medical Devices in Randomized Controlled Trials: Protocol for a Systematic Review
Background: Artificial intelligence (AI) medical devices have the potential to transform existing clinical workflows and ultimately improve patient outcomes. AI medical devices have shown potential for a range of clinical tasks such as diagnostics, prognostics, and therapeutic decision-making such as drug dosing. There is, however, an urgent need to ensure that these technologies remain safe for all populations. Recent literature demonstrates the need for rigorous performance error analysis to identify issues such as algorithmic encoding of spurious correlations (eg, protected characteristics) or specific failure modes that may lead to patient harm. Guidelines for reporting on studies that evaluate AI medical devices require the mention of performance error analysis; however, there is still a lack of understanding around how performance errors should be analyzed in clinical studies, and what harms authors should aim to detect and report. Objective: This systematic review will assess the frequency and severity of AI errors and adverse events (AEs) in randomized controlled trials (RCTs) investigating AI medical devices as interventions in clinical settings. The review will also explore how performance errors are analyzed including whether the analysis includes the investigation of subgroup-level outcomes. Methods: This systematic review will identify and select RCTs assessing AI medical devices. Search strategies will be deployed in MEDLINE (Ovid), Embase (Ovid), Cochrane CENTRAL, and clinical trial registries to identify relevant papers. RCTs identified in bibliographic databases will be cross-referenced with clinical trial registries. The primary outcomes of interest are the frequency and severity of AI errors, patient harms, and reported AEs. Quality assessment of RCTs will be based on version 2 of the Cochrane risk-of-bias tool (RoB2). Data analysis will include a comparison of error rates and patient harms between study arms, and a meta-analysis of the rates of patient harm in control versus intervention arms will be conducted if appropriate. Results: The project was registered on PROSPERO in February 2023. Preliminary searches have been completed and the search strategy has been designed in consultation with an information specialist and methodologist. Title and abstract screening started in September 2023. Full-text screening is ongoing and data collection and analysis began in April 2024. Conclusions: Evaluations of AI medical devices have shown promising results; however, reporting of studies has been variable. Detection, analysis, and reporting of performance errors and patient harms is vital to robustly assess the safety of AI medical devices in RCTs. Scoping searches have illustrated that the reporting of harms is variable, often with no mention of AEs. The findings of this systematic review will identify the frequency and severity of AI performance errors and patient harms and generate insights into how errors should be analyzed to account for both overall and subgroup performance. Trial Registration: PROSPERO CRD42023387747; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=387747 International Registered Report Identifier (IRRID): PRR1-10.2196/51614
- Open Access
Risk prediction tools for pressure injury occurrence: An umbrella review of systematic reviews reporting model development and validation methods
Background Pressure injuries (PIs) place a substantial burden on healthcare systems worldwide. Risk stratification of those who are at risk of developing PIs allows preventive interventions to be focused on patients who are at the highest risk. The considerable number of risk assessment scales and prediction models available underscore the need for a thorough evaluation of their development, validation and clinical utility. Our objectives were to identify and describe available risk prediction tools for PI occurrence, their content and development and validation methods used. Methods The umbrella review was conducted according to Cochrane guidance. MEDLINE, Embase, CINAHL, EPISTEMONIKOS, Google Scholar and reference lists were searched to identify relevant systematic reviews. Risk of bias was assessed using adapted AMSTAR-2 criteria. Results were described narratively. All included reviews contributed to build a comprehensive list of risk prediction tools. Results We identified five systematic reviews describing the development and validation of risk prediction tools for pressure injuries, 16 that assessed the prognostic accuracy of the tools and 10 that assessed the clinical effectiveness. Of the five reviews of model development and validation, four included only machine learning models. One review included detail about external validation, and this was the only review to include model performance metrics. Where quality assessment was completed (3 out of 5 reviews), most prediction tools were assessed by review authors as being at high risk of bias and no tools were assessed as being at low risk of bias. Conclusions Available tools do not meet current standards for the development or reporting of risk prediction models. The majority of tools have not been externally validated. Standardised and rigorous approaches to risk prediction model development and validation are needed. Registration The protocol was registered on the Open Science Framework (https://osf.io/tepyk).
Open AccessRisk prediction tools for pressure injury occurrence: An umbrella review of systematic reviews reporting model development and validation methods
Background Pressure injuries (PIs) place a substantial burden on healthcare systems worldwide. Risk stratification of those who are at risk of developing PIs allows preventive interventions to be focused on patients who are at the highest risk. The considerable number of risk assessment scales and prediction models available underscore the need for a thorough evaluation of their development, validation and clinical utility. Our objectives were to identify and describe available risk prediction tools for PI occurrence, their content and development and validation methods used. Methods The umbrella review was conducted according to Cochrane guidance. MEDLINE, Embase, CINAHL, EPISTEMONIKOS, Google Scholar and reference lists were searched to identify relevant systematic reviews. Risk of bias was assessed using adapted AMSTAR-2 criteria. Results were described narratively. All included reviews contributed to build a comprehensive list of risk prediction tools. Results We identified five systematic reviews describing the development and validation of risk prediction tools for pressure injuries, 16 that assessed the prognostic accuracy of the tools and 10 that assessed the clinical effectiveness. Of the five reviews of model development and validation, four included only machine learning models. One review included detail about external validation, and this was the only review to include model performance metrics. Where quality assessment was completed (3 out of 5 reviews), most prediction tools were assessed by review authors as being at high risk of bias and no tools were assessed as being at low risk of bias. Conclusions Available tools do not meet current standards for the development or reporting of risk prediction models. The majority of tools have not been externally validated. Standardised and rigorous approaches to risk prediction model development and validation are needed. Registration The protocol was registered on the Open Science Framework (https://osf.io/tepyk).
- Open Access
Accuracy and clinical effectiveness of risk prediction tools for pressure injury occurrence: An umbrella review
Background Pressure injuries (PIs) pose a substantial healthcare burden and incur significant costs worldwide. Several risk prediction models to allow timely implementation of preventive measures and potentially reduce healthcare system burden are available and in use. The ability of risk prediction tools to correctly identify those at high risk of PI (prognostic accuracy) and to have a clinically significant impact on patient management and outcomes (effectiveness) is not clear. We aimed to evaluate the prognostic accuracy and clinical effectiveness of risk prediction tools for PI, and to identify gaps in the literature. Methods and Findings The umbrella review was conducted according to Cochrane guidance. MEDLINE, Embase, CINAHL, EPISTEMONIKOS, Google Scholar and reference lists were searched to identify relevant systematic reviews. Risk of bias was assessed using adapted AMSTAR-2 criteria. Results were described narratively. We identified 16 reviews that assessed prognostic accuracy and 10 that assessed clinical effectiveness of risk prediction tools for PI. The 16 reviews of prognostic accuracy evaluated 63 tools (39 scales and 24 machine learning models), with the Braden, Norton, Waterlow, Cubbin-Jackson scales (and modifications thereof) the most evaluated tools. Meta-analyses from a focused set of included reviews showed that the scales had sensitivities and specificities ranging from 53%-97% and 46%-84%, respectively. Only 2/16 reviews performed appropriate statistical synthesis and quality assessment. One review assessing machine learning based algorithms reported high prognostic accuracy estimates, but some of which were sourced from the same data within which the models were developed, leading to potentially overoptimistic results. Two randomised trials assessing the effect of PI risk assessment tools on incidence of PIs were identified from the 10 systematic reviews of clinical effectiveness; both were included in a Cochrane review and assessed as high risk of bias. Both trials found no evidence of an effect on PI incidence. Conclusions Our findings underscore the lack of high-quality evidence for the accuracy of risk prediction tools for PI. There is no reliable evidence to suggest that using existing risk prediction tools effectively reduces the incidence of PIs. Further research is needed on their clinical effectiveness, but only once promising prediction tools have been developed and appropriately validated.
Open AccessAccuracy and clinical effectiveness of risk prediction tools for pressure injury occurrence: An umbrella review
Background Pressure injuries (PIs) pose a substantial healthcare burden and incur significant costs worldwide. Several risk prediction models to allow timely implementation of preventive measures and potentially reduce healthcare system burden are available and in use. The ability of risk prediction tools to correctly identify those at high risk of PI (prognostic accuracy) and to have a clinically significant impact on patient management and outcomes (effectiveness) is not clear. We aimed to evaluate the prognostic accuracy and clinical effectiveness of risk prediction tools for PI, and to identify gaps in the literature. Methods and Findings The umbrella review was conducted according to Cochrane guidance. MEDLINE, Embase, CINAHL, EPISTEMONIKOS, Google Scholar and reference lists were searched to identify relevant systematic reviews. Risk of bias was assessed using adapted AMSTAR-2 criteria. Results were described narratively. We identified 16 reviews that assessed prognostic accuracy and 10 that assessed clinical effectiveness of risk prediction tools for PI. The 16 reviews of prognostic accuracy evaluated 63 tools (39 scales and 24 machine learning models), with the Braden, Norton, Waterlow, Cubbin-Jackson scales (and modifications thereof) the most evaluated tools. Meta-analyses from a focused set of included reviews showed that the scales had sensitivities and specificities ranging from 53%-97% and 46%-84%, respectively. Only 2/16 reviews performed appropriate statistical synthesis and quality assessment. One review assessing machine learning based algorithms reported high prognostic accuracy estimates, but some of which were sourced from the same data within which the models were developed, leading to potentially overoptimistic results. Two randomised trials assessing the effect of PI risk assessment tools on incidence of PIs were identified from the 10 systematic reviews of clinical effectiveness; both were included in a Cochrane review and assessed as high risk of bias. Both trials found no evidence of an effect on PI incidence. Conclusions Our findings underscore the lack of high-quality evidence for the accuracy of risk prediction tools for PI. There is no reliable evidence to suggest that using existing risk prediction tools effectively reduces the incidence of PIs. Further research is needed on their clinical effectiveness, but only once promising prediction tools have been developed and appropriately validated.
- Open Access
- 23 August 2023
- Published by JMIR Publications Inc. in JMIR Preprints
Detecting Algorithmic Errors and Patient Harms for AI-Enabled Medical Devices in Randomized Controlled Trials: Protocol for a Systematic Review (Preprint)
BACKGROUND Artificial intelligence (AI) medical devices have the potential to transform existing clinical workflows and ultimately improve patient outcomes. AI medical devices have shown potential for a range of clinical tasks such as diagnostics, prognostics, and therapeutic decision-making such as drug dosing. There is, however, an urgent need to ensure that these technologies remain safe for all populations. Recent literature demonstrates the need for rigorous performance error analysis to identify issues such as algorithmic encoding of spurious correlations (eg, protected characteristics) or specific failure modes that may lead to patient harm. Guidelines for reporting on studies that evaluate AI medical devices require the mention of performance error analysis; however, there is still a lack of understanding around how performance errors should be analyzed in clinical studies, and what harms authors should aim to detect and report. OBJECTIVE This systematic review will assess the frequency and severity of AI errors and adverse events (AEs) in randomized controlled trials (RCTs) investigating AI medical devices as interventions in clinical settings. The review will also explore how performance errors are analyzed including whether the analysis includes the investigation of subgroup-level outcomes. METHODS This systematic review will identify and select RCTs assessing AI medical devices. Search strategies will be deployed in MEDLINE (Ovid), Embase (Ovid), Cochrane CENTRAL, and clinical trial registries to identify relevant papers. RCTs identified in bibliographic databases will be cross-referenced with clinical trial registries. The primary outcomes of interest are the frequency and severity of AI errors, patient harms, and reported AEs. Quality assessment of RCTs will be based on version 2 of the Cochrane risk-of-bias tool (RoB2). Data analysis will include a comparison of error rates and patient harms between study arms, and a meta-analysis of the rates of patient harm in control versus intervention arms will be conducted if appropriate. RESULTS The project was registered on PROSPERO in February 2023. Preliminary searches have been completed and the search strategy has been designed in consultation with an information specialist and methodologist. Title and abstract screening started in September 2023. Full-text screening is ongoing and data collection and analysis began in April 2024. CONCLUSIONS Evaluations of AI medical devices have shown promising results; however, reporting of studies has been variable. Detection, analysis, and reporting of performance errors and patient harms is vital to robustly assess the safety of AI medical devices in RCTs. Scoping searches have illustrated that the reporting of harms is variable, often with no mention of AEs. The findings of this systematic review will identify the frequency and severity of AI performance errors and patient harms and generate insights into how errors should be analyzed to account for both overall and subgroup performance. CLINICALTRIAL PROSPERO CRD42023387747; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=387747 INTERNATIONAL REGISTERED REPORT PRR1-10.2196/51614
- 23 August 2023
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessDetecting Algorithmic Errors and Patient Harms for AI-Enabled Medical Devices in Randomized Controlled Trials: Protocol for a Systematic Review (Preprint)
BACKGROUND Artificial intelligence (AI) medical devices have the potential to transform existing clinical workflows and ultimately improve patient outcomes. AI medical devices have shown potential for a range of clinical tasks such as diagnostics, prognostics, and therapeutic decision-making such as drug dosing. There is, however, an urgent need to ensure that these technologies remain safe for all populations. Recent literature demonstrates the need for rigorous performance error analysis to identify issues such as algorithmic encoding of spurious correlations (eg, protected characteristics) or specific failure modes that may lead to patient harm. Guidelines for reporting on studies that evaluate AI medical devices require the mention of performance error analysis; however, there is still a lack of understanding around how performance errors should be analyzed in clinical studies, and what harms authors should aim to detect and report. OBJECTIVE This systematic review will assess the frequency and severity of AI errors and adverse events (AEs) in randomized controlled trials (RCTs) investigating AI medical devices as interventions in clinical settings. The review will also explore how performance errors are analyzed including whether the analysis includes the investigation of subgroup-level outcomes. METHODS This systematic review will identify and select RCTs assessing AI medical devices. Search strategies will be deployed in MEDLINE (Ovid), Embase (Ovid), Cochrane CENTRAL, and clinical trial registries to identify relevant papers. RCTs identified in bibliographic databases will be cross-referenced with clinical trial registries. The primary outcomes of interest are the frequency and severity of AI errors, patient harms, and reported AEs. Quality assessment of RCTs will be based on version 2 of the Cochrane risk-of-bias tool (RoB2). Data analysis will include a comparison of error rates and patient harms between study arms, and a meta-analysis of the rates of patient harm in control versus intervention arms will be conducted if appropriate. RESULTS The project was registered on PROSPERO in February 2023. Preliminary searches have been completed and the search strategy has been designed in consultation with an information specialist and methodologist. Title and abstract screening started in September 2023. Full-text screening is ongoing and data collection and analysis began in April 2024. CONCLUSIONS Evaluations of AI medical devices have shown promising results; however, reporting of studies has been variable. Detection, analysis, and reporting of performance errors and patient harms is vital to robustly assess the safety of AI medical devices in RCTs. Scoping searches have illustrated that the reporting of harms is variable, often with no mention of AEs. The findings of this systematic review will identify the frequency and severity of AI performance errors and patient harms and generate insights into how errors should be analyzed to account for both overall and subgroup performance. CLINICALTRIAL PROSPERO CRD42023387747; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=387747 INTERNATIONAL REGISTERED REPORT PRR1-10.2196/51614
- Open Access
- 19 July 2023
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 25, e45599
Accelerometer-Measured Physical Activity Data Sets (Global Physical Activity Data Set Catalogue) That Include Markers of Cardiometabolic Health: Systematic Scoping Review
Background: Cardiovascular disease accounts for 17.9 million deaths globally each year. Many research study data sets have been collected to answer questions regarding the relationship between cardiometabolic health and accelerometer-measured physical activity. This scoping review aimed to map the available data sets that have collected accelerometer-measured physical activity and cardiometabolic health markers. These data were then used to inform the development of a publicly available resource, the Global Physical Activity Data set (GPAD) catalogue. Objective: This review aimed to systematically identify data sets that have measured physical activity using accelerometers and cardiometabolic health markers using either an observational or interventional study design. Methods: Databases, trial registries, and gray literature (inception until February 2021; updated search from February 2021 to September 2022) were systematically searched to identify studies that analyzed data sets of physical activity and cardiometabolic health outcomes. To be eligible for inclusion, data sets must have measured physical activity using an accelerometric device in adults aged ≥18 years; a sample size >400 participants (unless recruited participants in a low- and middle-income country where a sample size threshold was reduced to 100); used an observational, longitudinal, or trial-based study design; and collected at least 1 cardiometabolic health marker (unless only body mass was measured). Two reviewers screened the search results to identify eligible studies, and from these, the unique names of each data set were recorded, and characteristics about each data set were extracted from several sources. Results: A total of 17,391 study reports were identified, and after screening, 319 were eligible, with 122 unique data sets in these study reports meeting the review inclusion criteria. Data sets were found in 49 countries across 5 continents, with the most developed in Europe (n=53) and the least in Africa and Oceania (n=4 and n=3, respectively). The most common accelerometric brand and device wear location was Actigraph and the waist, respectively. Height and body mass were the most frequently measured cardiometabolic health markers in the data sets (119/122, 97.5% data sets), followed by blood pressure (82/122, 67.2% data sets). The number of participants in the included data sets ranged from 103,712 to 120. Once the review processes had been completed, the GPAD catalogue was developed to house all the identified data sets. Conclusions: This review identified and mapped the contents of data sets from around the world that have collected potentially harmonizable accelerometer-measured physical activity and cardiometabolic health markers. The GPAD catalogue is a web-based open-source resource developed from the results of this review, which aims to facilitate the harmonization of data sets to produce evidence that will reduce the burden of disease from physical inactivity.
- 19 July 2023
- Published by JMIR Publications Inc. in Journal of Medical Internet Research
- Vol. 25, e45599
Open AccessAccelerometer-Measured Physical Activity Data Sets (Global Physical Activity Data Set Catalogue) That Include Markers of Cardiometabolic Health: Systematic Scoping Review
Background: Cardiovascular disease accounts for 17.9 million deaths globally each year. Many research study data sets have been collected to answer questions regarding the relationship between cardiometabolic health and accelerometer-measured physical activity. This scoping review aimed to map the available data sets that have collected accelerometer-measured physical activity and cardiometabolic health markers. These data were then used to inform the development of a publicly available resource, the Global Physical Activity Data set (GPAD) catalogue. Objective: This review aimed to systematically identify data sets that have measured physical activity using accelerometers and cardiometabolic health markers using either an observational or interventional study design. Methods: Databases, trial registries, and gray literature (inception until February 2021; updated search from February 2021 to September 2022) were systematically searched to identify studies that analyzed data sets of physical activity and cardiometabolic health outcomes. To be eligible for inclusion, data sets must have measured physical activity using an accelerometric device in adults aged ≥18 years; a sample size >400 participants (unless recruited participants in a low- and middle-income country where a sample size threshold was reduced to 100); used an observational, longitudinal, or trial-based study design; and collected at least 1 cardiometabolic health marker (unless only body mass was measured). Two reviewers screened the search results to identify eligible studies, and from these, the unique names of each data set were recorded, and characteristics about each data set were extracted from several sources. Results: A total of 17,391 study reports were identified, and after screening, 319 were eligible, with 122 unique data sets in these study reports meeting the review inclusion criteria. Data sets were found in 49 countries across 5 continents, with the most developed in Europe (n=53) and the least in Africa and Oceania (n=4 and n=3, respectively). The most common accelerometric brand and device wear location was Actigraph and the waist, respectively. Height and body mass were the most frequently measured cardiometabolic health markers in the data sets (119/122, 97.5% data sets), followed by blood pressure (82/122, 67.2% data sets). The number of participants in the included data sets ranged from 103,712 to 120. Once the review processes had been completed, the GPAD catalogue was developed to house all the identified data sets. Conclusions: This review identified and mapped the contents of data sets from around the world that have collected potentially harmonizable accelerometer-measured physical activity and cardiometabolic health markers. The GPAD catalogue is a web-based open-source resource developed from the results of this review, which aims to facilitate the harmonization of data sets to produce evidence that will reduce the burden of disease from physical inactivity.
- Open Access
- 12 January 2023
- Published by JMIR Publications Inc. in JMIR Preprints
Accelerometer-Measured Physical Activity Data Sets (Global Physical Activity Data Set Catalogue) That Include Markers of Cardiometabolic Health: Systematic Scoping Review (Preprint)
BACKGROUND Cardiovascular disease accounts for 17.9 million deaths globally each year. Many research study data sets have been collected to answer questions regarding the relationship between cardiometabolic health and accelerometer-measured physical activity. This scoping review aimed to map the available data sets that have collected accelerometer-measured physical activity and cardiometabolic health markers. These data were then used to inform the development of a publicly available resource, the Global Physical Activity Data set (GPAD) catalogue. OBJECTIVE This review aimed to systematically identify data sets that have measured physical activity using accelerometers and cardiometabolic health markers using either an observational or interventional study design. METHODS Databases, trial registries, and gray literature (inception until February 2021; updated search from February 2021 to September 2022) were systematically searched to identify studies that analyzed data sets of physical activity and cardiometabolic health outcomes. To be eligible for inclusion, data sets must have measured physical activity using an accelerometric device in adults aged ≥18 years; a sample size >400 participants (unless recruited participants in a low- and middle-income country where a sample size threshold was reduced to 100); used an observational, longitudinal, or trial-based study design; and collected at least 1 cardiometabolic health marker (unless only body mass was measured). Two reviewers screened the search results to identify eligible studies, and from these, the unique names of each data set were recorded, and characteristics about each data set were extracted from several sources. RESULTS A total of 17,391 study reports were identified, and after screening, 319 were eligible, with 122 unique data sets in these study reports meeting the review inclusion criteria. Data sets were found in 49 countries across 5 continents, with the most developed in Europe (n=53) and the least in Africa and Oceania (n=4 and n=3, respectively). The most common accelerometric brand and device wear location was Actigraph and the waist, respectively. Height and body mass were the most frequently measured cardiometabolic health markers in the data sets (119/122, 97.5% data sets), followed by blood pressure (82/122, 67.2% data sets). The number of participants in the included data sets ranged from 103,712 to 120. Once the review processes had been completed, the GPAD catalogue was developed to house all the identified data sets. CONCLUSIONS This review identified and mapped the contents of data sets from around the world that have collected potentially harmonizable accelerometer-measured physical activity and cardiometabolic health markers. The GPAD catalogue is a web-based open-source resource developed from the results of this review, which aims to facilitate the harmonization of data sets to produce evidence that will reduce the burden of disease from physical inactivity.
- 12 January 2023
- Published by JMIR Publications Inc. in JMIR Preprints
Open AccessAccelerometer-Measured Physical Activity Data Sets (Global Physical Activity Data Set Catalogue) That Include Markers of Cardiometabolic Health: Systematic Scoping Review (Preprint)
BACKGROUND Cardiovascular disease accounts for 17.9 million deaths globally each year. Many research study data sets have been collected to answer questions regarding the relationship between cardiometabolic health and accelerometer-measured physical activity. This scoping review aimed to map the available data sets that have collected accelerometer-measured physical activity and cardiometabolic health markers. These data were then used to inform the development of a publicly available resource, the Global Physical Activity Data set (GPAD) catalogue. OBJECTIVE This review aimed to systematically identify data sets that have measured physical activity using accelerometers and cardiometabolic health markers using either an observational or interventional study design. METHODS Databases, trial registries, and gray literature (inception until February 2021; updated search from February 2021 to September 2022) were systematically searched to identify studies that analyzed data sets of physical activity and cardiometabolic health outcomes. To be eligible for inclusion, data sets must have measured physical activity using an accelerometric device in adults aged ≥18 years; a sample size >400 participants (unless recruited participants in a low- and middle-income country where a sample size threshold was reduced to 100); used an observational, longitudinal, or trial-based study design; and collected at least 1 cardiometabolic health marker (unless only body mass was measured). Two reviewers screened the search results to identify eligible studies, and from these, the unique names of each data set were recorded, and characteristics about each data set were extracted from several sources. RESULTS A total of 17,391 study reports were identified, and after screening, 319 were eligible, with 122 unique data sets in these study reports meeting the review inclusion criteria. Data sets were found in 49 countries across 5 continents, with the most developed in Europe (n=53) and the least in Africa and Oceania (n=4 and n=3, respectively). The most common accelerometric brand and device wear location was Actigraph and the waist, respectively. Height and body mass were the most frequently measured cardiometabolic health markers in the data sets (119/122, 97.5% data sets), followed by blood pressure (82/122, 67.2% data sets). The number of participants in the included data sets ranged from 103,712 to 120. Once the review processes had been completed, the GPAD catalogue was developed to house all the identified data sets. CONCLUSIONS This review identified and mapped the contents of data sets from around the world that have collected potentially harmonizable accelerometer-measured physical activity and cardiometabolic health markers. The GPAD catalogue is a web-based open-source resource developed from the results of this review, which aims to facilitate the harmonization of data sets to produce evidence that will reduce the burden of disease from physical inactivity.
- Open Access
- 20 January 2021
- Published by Springer Nature in Systematic Reviews
- Vol. 10 (1) , 1-7
Effectiveness of interventions to reduce household air pollution from solid biomass fuels and improve maternal and child health outcomes in low- and middle-income countries: a systematic review protocol
Background: A variety of public health interventions have been undertaken in low- and middle-income countries (LMICs) to prevent morbidity and mortality associated with household air pollution (HAP) due to cooking, heating and lighting with solid biomass fuels. Pregnant women and children under five are particularly vulnerable to the effects of HAP, due to biological susceptibility and typically higher exposure levels. However, the relative health benefits of interventions to reduce HAP exposure among these groups remain unclear. This systematic review aims to assess, among pregnant women, infants and children (under 5 years) in LMIC settings, the effectiveness of interventions which aim to reduce household air pollutant emissions due to household solid biomass fuel combustion, compared to usual cooking practices, in terms of health outcomes associated with HAP exposure. Methods: This protocol follows standard systematic review processes and abides by the PRISMA-P reporting guidelines. Searches will be undertaken in MEDLINE, EMBASE, CENTRAL, WHO International Clinical Trials Registry Platform (ICTRP), The Global Index Medicus (GIM), ClinicalTrials.gov and Greenfile, combining terms for pregnant women and children with interventions or policy approaches to reduce HAP from biomass fuels or HAP terms and LMIC countries. Included studies will be those reporting (i) pregnant women and children under 5 years; (ii) fuel transition, structural, educational or policy interventions; and (iii) health events associated with HAP exposure which occur among pregnant women or among children within the perinatal period, infancy and up to 5 years of age. A narrative synthesis will be undertaken for each population-intervention-outcome triad stratified by study design. Clinical and methodological homogeneity within each triad will be used to determine the feasibility for undertaking meta-analyses to give a summary estimate of the effect for each outcome. Discussion: This systematic review will identify the effectiveness of existing HAP intervention measures in LMIC contexts, with discussion on the context of implementation and adoption, and summarise current literature of relevance to maternal and child health. This assessment reflects the need for HAP interventions which achieve measurable health benefits, which would need to be supported by policies that are socially and economically acceptable in LMIC settings worldwide. Systematic review registration: PROSPERO CRD42020164998
- 20 January 2021
- Published by Springer Nature in Systematic Reviews
- Vol. 10 (1) , 1-7
Open AccessEffectiveness of interventions to reduce household air pollution from solid biomass fuels and improve maternal and child health outcomes in low- and middle-income countries: a systematic review protocol
Background: A variety of public health interventions have been undertaken in low- and middle-income countries (LMICs) to prevent morbidity and mortality associated with household air pollution (HAP) due to cooking, heating and lighting with solid biomass fuels. Pregnant women and children under five are particularly vulnerable to the effects of HAP, due to biological susceptibility and typically higher exposure levels. However, the relative health benefits of interventions to reduce HAP exposure among these groups remain unclear. This systematic review aims to assess, among pregnant women, infants and children (under 5 years) in LMIC settings, the effectiveness of interventions which aim to reduce household air pollutant emissions due to household solid biomass fuel combustion, compared to usual cooking practices, in terms of health outcomes associated with HAP exposure. Methods: This protocol follows standard systematic review processes and abides by the PRISMA-P reporting guidelines. Searches will be undertaken in MEDLINE, EMBASE, CENTRAL, WHO International Clinical Trials Registry Platform (ICTRP), The Global Index Medicus (GIM), ClinicalTrials.gov and Greenfile, combining terms for pregnant women and children with interventions or policy approaches to reduce HAP from biomass fuels or HAP terms and LMIC countries. Included studies will be those reporting (i) pregnant women and children under 5 years; (ii) fuel transition, structural, educational or policy interventions; and (iii) health events associated with HAP exposure which occur among pregnant women or among children within the perinatal period, infancy and up to 5 years of age. A narrative synthesis will be undertaken for each population-intervention-outcome triad stratified by study design. Clinical and methodological homogeneity within each triad will be used to determine the feasibility for undertaking meta-analyses to give a summary estimate of the effect for each outcome. Discussion: This systematic review will identify the effectiveness of existing HAP intervention measures in LMIC contexts, with discussion on the context of implementation and adoption, and summarise current literature of relevance to maternal and child health. This assessment reflects the need for HAP interventions which achieve measurable health benefits, which would need to be supported by policies that are socially and economically acceptable in LMIC settings worldwide. Systematic review registration: PROSPERO CRD42020164998
- 20 December 2019
- Published by BMJ in Journal of Epidemiology and Community Health
- Vol. 74 (3) , 269-275
Effects of physical activity calorie equivalent food labelling to reduce food selection and consumption: systematic review and meta-analysis of randomised controlled studies
Background There is limited evidence that nutritional labelling on food/drinks is changing eating behaviours. Physical activity calorie equivalent (PACE) food labelling aims to provide the public with information about the amount of physical activity required to expend the number of kilocalories in food/drinks (eg, calories in this pizza requires 45 min of running to burn), to encourage healthier food choices and reduce disease. Objective We aimed to systematically search for randomised controlled trials and experimental studies of the effects of PACE food labelling on the selection, purchase or consumption of food/drinks. Methods PACE food labelling was compared with any other type of food labelling or no labelling (comparator). Reports were identified by searching electronic databases, websites and social media platforms. Inverse variance meta-analysis was used to summarise evidence. Weighted mean differences (WMD) and 95% CIs were used to describe between-group differences using a random effects model. Results 15 studies were eligible for inclusion. When PACE labelling was displayed on food/drinks and menus, significantly fewer calories were selected, relative to comparator labelling (WMD=−64.9 kcal, 95% CI −103.2 to −26.6, p=0.009, n=4606). Presenting participants with PACE food labelling results in the consumption of significantly fewer calories (WMD=−80.4 kcal, 95% CI−136.7 to −24.2, p=0.005, n=486) relative to comparator food labelling. Conclusion Based on current evidence PACE food labelling may reduce the number of kilocalories selected from menus and decrease the number of kilocalories/grams of food consumed by the public, compared with other types of food labelling/no labelling. Trial registration number CRD42018088567.
- 20 December 2019
- Published by BMJ in Journal of Epidemiology and Community Health
- Vol. 74 (3) , 269-275
Effects of physical activity calorie equivalent food labelling to reduce food selection and consumption: systematic review and meta-analysis of randomised controlled studies
Background There is limited evidence that nutritional labelling on food/drinks is changing eating behaviours. Physical activity calorie equivalent (PACE) food labelling aims to provide the public with information about the amount of physical activity required to expend the number of kilocalories in food/drinks (eg, calories in this pizza requires 45 min of running to burn), to encourage healthier food choices and reduce disease. Objective We aimed to systematically search for randomised controlled trials and experimental studies of the effects of PACE food labelling on the selection, purchase or consumption of food/drinks. Methods PACE food labelling was compared with any other type of food labelling or no labelling (comparator). Reports were identified by searching electronic databases, websites and social media platforms. Inverse variance meta-analysis was used to summarise evidence. Weighted mean differences (WMD) and 95% CIs were used to describe between-group differences using a random effects model. Results 15 studies were eligible for inclusion. When PACE labelling was displayed on food/drinks and menus, significantly fewer calories were selected, relative to comparator labelling (WMD=−64.9 kcal, 95% CI −103.2 to −26.6, p=0.009, n=4606). Presenting participants with PACE food labelling results in the consumption of significantly fewer calories (WMD=−80.4 kcal, 95% CI−136.7 to −24.2, p=0.005, n=486) relative to comparator food labelling. Conclusion Based on current evidence PACE food labelling may reduce the number of kilocalories selected from menus and decrease the number of kilocalories/grams of food consumed by the public, compared with other types of food labelling/no labelling. Trial registration number CRD42018088567.