CaHRU/LIH seminar: Causal models and Directed Acyclic Graphs – Professor Graham Law

Graham LawProfessor Graham Law, who recently joined the university and CaHRU as Professor in Medical Statitics, delivered the latest of CaHRU/LIH’s Implementation Science and Research Methods seminar series – on Causal Models and the use of Directed Acyclic Graphs. Professor Law set out the epidemiological context for the seminar. Epidemiology is the study of disease. The scientific methods used within epidemiology aim to discover the determinants of disease. The state of having the disease is the outcome. Causal models seek to examine the factors that contribute to this outcome. Causation is not usually dealt with by statistics. Instead, statistics concerns itself more with associations and relationships between variables.

dagA Directed Acyclic Graph (DAG) comprises variables (nodes) and arrows between nodes (directed edges) such that the graph is acyclic, i.e., it not being possible to start at any node, follow the directed edges in the arrowhead direction, and end up back at the same node. In seeking to represent causation, DAGs typically display a series of factors, mediators and outcomes. Having explained this, Graham split the audience into two groups and challenged them to create their own DAGs to display all the possible factors involved in whether sunlight causes lymphoma. The interactive nature of the seminar engaged the audience. The fun nature of the concluding task made a potentially complicated subject easier for the audience to understand and apply.

By Viet-Hai Phung

CaHRU/LIH seminar: Scoping reviews – Viet-Hai Phung

The latest CaHRU/LIH Improvement Science and Research Methods seminar was on scoping reviews, delivered on 28th February 2017 by Viet-Hai Phung, Research Assistant at CaHRU. Viet-Hai began the seminar by stating the aims of scoping reviews: to “map rapidly the key concepts underpinning a research area and the main sources and types of evidence available, and can be undertaken as stand-alone projects in their own right, especially where an area is complex or has not been reviewed comprehensively before” (Mays et al., 2001: 194).Viet Hai Scoping Reviews

Viet-Hai who is currently conducting a scoping review as the first phase of his PhD, explained that he is doing this because the extent of knowledge on the evidence relating to his research area in ‘healthcare experiences of migrants in Europe’ is limited. Secondly, the scoping review will help to establish themes or research questions for further exploration using appropriate research methods.

He went on to outline the features and steps to conducting a scoping review and took time to explain each in detail, placing emphasis on the need to have a well-defined search strategy for capturing all potentially relevant studies. He also recommended the use of EndNote and Excel for managing the studies or data generated from the search, which can facilitate the identification and inclusion or exclusion of studies from the review. Constructing a flow diagram can also help in presenting a summary of the study identification process.

vhp2For the synthesis of the data generated, Viet-Hai suggested that reviewers may summarise each of the included studies according to relevant headings: population, methods, results etc, and thematically summarise the evidence from the review using a narrative, reflective approach.

He concluded the session by highlighting a number of caveats which should be noted when planning to conduct a scoping review. He said, for example, that some search terms may need to be adapted or modified for different databases, and that only a limited number of publications can be imported to Endnote at a time. In addition, many publications have copyright statements that need to be removed in Excel, and that data extraction can be time consuming.

By Joseph Akanuwe

[su_document url=”https://communityandhealth.dev.lincoln.ac.uk/files/2017/03/Scoping-review-28-February-2017.pdf” responsive=”no”]Multi-morbidity, goal-oriented care, the community and equity[/su_document]

CaHRU/LIH seminar: Consensus methods – Dr Paul Leighton

WP_20170124_11_59_23_RichThe latest CaHRU/LIH Improvement Science and Research Methods seminar was given on 24th January 2017 by Dr. Paul Leighton, Deputy Director at the NIHR Research Design Service for the East Midlands. Paul’s talk, entitled ‘The Delphi technique and other methods of consensus development in applied health settings’, focussed on useful tools for achieving consensus among experts.

WP_20170124_12_05_46_RichPaul began the seminar with an overview of the background of consensus methods and gave examples of how such techniques have been used in the past. He then presented the three most commonly used approaches to reaching consensus: the Delphi technique, the nominal group technique, and the consensus conference. He explained the different techniques, discussed issues around using them and conducting consensus events, and finished by comparing the three methods.

Paul also offered guidance on when it is best to use each technique; this covered when it was appropriate to use a particular method, choosing the correct number and most suitable members/experts for panels, agreeing on the threshold for consensus, and the proper reporting standards. At the end of the seminar, Paul reminded everyone that the key enemy of consensus methods is always dropout rates!

The seminar was very well received and raised some very important issues around consensus methods. The next Improvement Science and Research Methods seminar will be presented by Viet-Hai Phung, researcher at CaHRU, on Tuesday 28th February 2017 on the topic of scoping reviews.

By Despina Laparidou

Implementation science and research methods seminar: statistical modelling of time series data

Dr Murray SmithThe latest CaHRU Implementation Science and Research Methods seminar was presented by Dr Murray Smith on 25th October 2016, on the topic of statistical modelling of time series data. Dr Smith, research fellow at the Community and Health Research Unit (CaHRU), is an econometrician with a wide range of experience and research interests in micro-econometric modelling, health informatics, health economic modelling, mathematical statistics and computer algebra. Murray has previously worked as an Associate Professor in the Department of Econometrics and Business Statistics at the University of Sydney, Australia, and as an Associate Professor at the School of Pharmacy, University of Nottingham before coming to the University of Lincoln. His research has focuses on the quality of use of pharmaceutical medicines in chronic disease, with analyses that utilise prescribing data drawn from a large database of general practice patient records.

segmentedregressionHe introduced his topic by defining (and giving examples) of data, and related these to how researchers can statistically model time series data in different ways. Murray described four fundamental components of times series including: time (Tt); Cycle (Ct); Seasonal (St) and Irregular (It). He also gave examples of Time Series Analysis Curves such as survival data, Engle’s AutoRegressive Conditional Heteroscedasticity (ARCH) and segmented regression. He showed how modelling time series data can be useful for quality improvement research, as this can facilitate the measurement of outcomes or impact over time and extrapolate or estimate future values or outcomes. A time series method (using segmented regression) is being used in a paper currently being prepared by Dr Murray Smith and Professor Niro Siriwardena for publication.

QIlogoDr Smith concluded his interesting presentation by recommending three key books for further reading on Statistical Modelling of Time Series Data: Time Series Analysis by Jame D Hamilton, Time Series Analysis: Forecasting and Control by George E. P. Box et al. and Introduction to Time Series and Forecasting by Peter J. Brockwell & Richard A. Davis. He made the session so interesting that participants asked for further presentations on time series analysis. For his next seminar, Murray will focus more on use of the method for current research at CaHRU to enable more researchers to apply this method to their research.

By Joseph Akanuwe

CAHRU/LIH improvement science and research methods seminar: Mixed methods

QIlogoThe first of this year’s series of CaHRU/LIH (Community and Health Research Unit/Lincoln Institute for Health) improvement science and research methods seminars was given by Prof Siriwardena on mixed methods on 16 February 2016. Improvement and implementation science benefits from the use of mixed research designs which combine quantitative and qualitative methods to show not only what happened but also why and how this might have occurred. Mixed methods approaches are a subset of multiple methods which involve more than one type of qualitative or quantitative method.

ascqilogoThe seminar covered principles such as definitions, theoretical approaches (such as pragmatism and transformation), basic and advanced (including case study) designs and approaches to data integration and transformation. This was then applied to examples of mixed methods designs used by CaHRU in a previous programme of research: the Ambulance Services Cardiovascular Quality Initiative (ASCQI). ASCQI was a national project, led by CaHRU and East Midlands Ambulance Service NHS Trust, designed to improve care for people presenting to ambulance services with heart attack or stroke using a large-scale quality improvement collaborative (QIC), evaluated using a multiple case-study design.

NWASami170112ASCQI involved gathering quantitative and qualitative data to describe what effect the QIC had, and how improvements, if they did occur, were brought about. Integration of data was carried out using techniques such annotated control charts showing time series data together with what was implemented, pattern matching comparing what services did and whether improvements occurred (doi:  10.1186/1748-5908-9-17), and comparison of quantitative and qualitative data from an online questionnaire (doi: 10.1111/jep.12438). Attendees were finally asked to consider a mixed methods question and think about research designs which they might use to answer it.

Thank you to all those staff and students who attended. Details of future seminars will be posted on the CaHRU and the LIH sites shortly.

[su_document url=”https://communityandhealth.dev.lincoln.ac.uk/files/2016/02/Mixed-methods_Siriwardena.pdf” responsive=”no”]Multi-morbidity, goal-oriented care, the community and equity[/su_document]