Electronic Health Record (EHR) data are complex, messy and large. Data are recorded over time for the purposes of clinical care. As such, they raise complex analytical challenges when used for health research. This module provides the key analytic skills needed to analyse and interpret electronic health record data. 

This module will tackle key analytic concepts and techniques required to analyse EHR data. This will include: creating data sets for analysis from EHRs using various study designs; code list creation and use; measurement in EHR (exposures, outcomes, covariates) including drug exposure data; a review of the causal inference framework and important causal analytic methods, including regression, propensity score analysis, and survival analysis. 

The module includes an in-depth exploration of survival analysis. These methods are widely used in health data science and particularly in the analysis of EHR data. The methods covered also apply to other data sources, including longitudinal cohort data, patient disease registries, and randomised trials.

The overall module aim is to equip students with the necessary skills to analyse data arising from Electronic Health Records.

Intended learning outcomes

Upon successful completion of the module a student will be able to: 

1. Demonstrate knowledge and understanding of the unique features of Electronic Health Record (EHR) data and why these features lead to analytic complexities.
2. Critically apply a range of analytic methods to EHR data and demonstrate an understanding of their purposes.
3. Demonstrate an understanding of the assumptions underlying a range of relevant methods to handle confounding, and critically apply the methods to EHR using R.
4. Analyse and interpret EHR data using the most appropriate method of analysis and evaluate its effectiveness.

Session content

The module is expected to cover the following topics: 

• Measurement in Electronic Health Record (EHR) data, including use of codelists and obtaining drug exposure data.
• Causal analysis, review of basic concepts.
• Propensity score analysis, regression, and related approaches.
• Functions used in the description and analysis of survival data, including hazard and survivor functions.
• The theory and use of proportional hazard models, including parametric models and the Cox model.#

Mode of delivery

This module is delivered predominantly face-to-face. Where specific teaching methods (lectures, seminars, discussion groups) are noted in this module specification these will be delivered by predominantly face-to-face sessions. There will be a combination of live and interactive activities (synchronous learning) as well as self-directed study (asynchronous learning).

Assessment

The summative assessment will consist of an analysis of electronic health record data to answer a given health data science question. Students will select appropriate techniques from the module to address the question, apply them to the data, and write a short report on the results and interpretation. 

Credits

• CATS: 15
• ECTS: 7.5

Module specification

For full information regarding this module please see the module specification.