Using advanced quantitative analyses on big data to determine the role of depression in outcomes of hip fracture: Interdisciplinary investigation

Lead Supervisor
Dr Salma Ayis
Senior Lecturer in Medical Statistics
School of Population Health and Environmental Sciences, King’s College London

Dr Katie Sheehan
Dept. Population Health Sciences, King’s College London

Professor Catherine Sackley will guide the interpretation of modelling in context of rehabilitation and health services research. 

Project Details


Depressive symptoms after hip fracture are common, as 1 in 4 have such symptoms. The duration of depressive symptoms varies due to patient, social, and environmental factors. However, there is limited research on the role of these factors in the continuity of depressive symptoms. Further, the extent to which factors interact with depressive symptoms to contribute to healthcare dependence and poor outcomes is unknown. 

Proposed methods

This PhD focuses on the development and application of advanced and novel statistical techniques, including Growth Mixture Models (GMM), and Group Based Trajectory Models (GBTM) to address patients’ heterogeneity and establish better causal inference between depression and outcomes of hip fracture. The methods were embraced for their capacity to map closely how growth and development of disorders are conceptualised by clinicians and researchers and provide empirical means of identifying clusters of individuals following both typical and atypical courses of development. Prognostic models would then be used to identify predictors of outcomes for each cluster.  

This project will capitalise on data on 18,000 patients, from the Patient Episode Database for Wales (PEDW) to determine the role of prognostic factors of depression diagnosis, in outcomes of hip fracture. The cohort was already used in an ongoing investigation on survival after hip fracture. The study will initially conduct a systematic review of prognostic factors for depression after hip fracture. The review will be reported in adherence to the Preferred Reporting Items for Systematic Reviews and Meta-analysis. Quality appraisal will be completed using the Quality in Prognostic Studies Tool, Cochrane Risk of Bias Tool and/or the Quality in Qualitative Evaluation as required. Narrative approach and Meta-analysis may be completed depending on homogeneity of studies.

Skills development

The project is an interdisciplinary, at the interface of data management, statistics, epidemiology, health services research and quality improvement.  The student will engage with experts in each of these disciplines to ensure their project is relevant, rigorous, and has potential for impact across disciplines. The student will build on research undertaken in the Department of Population Health Sciences by experts in medical statistics and hip fracture research. The project can be tailored to the student’s interests and could be more oriented towards epidemiology and statistics if appropriate. 

Specific objectives are to:

  1. Conduct a systematic review of prognostic factors for depression after hip fracture; to synthesize the evidence within a systematic review framework.
  2. Develop prognostic models for depressive symptoms after hip fracture; to identify factors which are modifiable, and which are immutable. 
  3. Detect heterogenous trajectories/clusters of depression after hip fracture and investigate their determinants.
  4. Investigate differential healthcare dependence among identified clusters and association with adverse outcomes including mortality. 
  5. Investigate and develop strategies for dealing with data missingness (e.g. simulation).


Data from the Patient Episode Database for Wales (PEDW) will be used.  A standardised “Data Request Specification Form” will be used to request a specific number of variable and data for patients who undergone hip fracture between 2014-2016, and follow up on diagnosis and prognostic factors to 2020/2021. 


Depression; Hip Fracture Prognosis;rowth Mixture Models (GMM); Group Based Trajectory Models (GBTM);  Patient Episode Database for Wales (PEDW)