TABLE OF CONTENTS

Introduction

All data from Cortex can be delivered to your organisation's business intelligence team. This includes: 

  • Activity data - anything that gets done in Cortex such as creating a clinical document, looking at lab results etc.
  • Form data - data fields from within forms.


Therefore Cortex can be a useful tool for collecting data to understand the patient journey and improve processes and patient care. 


Types of Data


Categorical data 

  • Options component 
  • Boolean component 
  • Tokens component 


Data in Cortex can also be categorized at point-of-care using SNOMED.  This is a system of concepts and terms developed by clinicians and researchers to accurately and consistently represent clinical information. 

The NZ Ministry of Health page on SNOMED is here


The Cortex notes engine allows SNOMED coding of any piece of categorical data (i.e. an answer to a question collected a checkbox, radio button, drop down menu, yes/no).  


The SNOMED browser (http://browser.ihtsdotools.org/) allows searching to identify the codes that may fit specific observations. For examples, search for ‘RIF tenderness’ produces the code 301410009.



There are two methods of capturing SNOMED data in Cortex 

  • Tokens that use SNOMED subsets 
  • Manually adding SNOMED codes to the answers to options components.  For example: 


A screenshot of a cell phone

Description automatically generated



When considering collecting SNOMED data via tokens, ask what the benefit to the clinician might be.  It is more time consuming for the clinician to choose a defined diagnosis that simply typing their impression - the payback may be in returning useful data to the department on diagnoses.


Ordinal data 

  • Options component to create ordered categories 


Numerical data 

  • Number component 
  • Options component that uses numbers 



Consistency of data 


Quality of the data collected in Cortex is key for both clinical care and research / audits. As most clinical data is based on the clinician’s interpretation of their observations, be it an examination finding or interpretation of a patient statement, it is helpful to provide guidance as to how a clinician should categorise these observations. 


In Cortex this is primarily done with ‘Definitions’. 


In the example below, the definitions help the clinician decide how to categorise the patient’s ‘Smoking history’ into either ‘Current smoker’, ‘Ex-smoker’ or ‘Never smoked’.  In this case, the information is based on New Zealand Ministry of Health guidelines. 


A screenshot of a social media post

Description automatically generated


 


Data Analytics perspective 


The following is provided from the perspective of those processing the data to then create reports and dashboards for clinicians


  • Boolean-components are much easier to report than options-components. So use Boolean where there can be just two options.
  • From a data quality perspective, there isn't much point in making a field mandatory if there is a meaningless easy-out option like 'unknown'. If 'unknown' is likely to be selected a lot, then it would be worth considering not making it mandatory or not having the 'unknown' option available.  There is an added bonus here, in that this increases the possibility of making the field a boolean-component instead of an options-component.   
  • For data reporting, the data key field for each component is essential. This is the field name that BIDA can see. Make this key meaningful so it is easy to identify.
  • Consider only starting the data analytics request from the point when the form design is settled.  It is difficult to create reports from multiple versions of forms where there have been significant changes in structure.
  • Ask questions of the data analytics experts early, particularly if you have in mind the type of report or dashboard you would like to generate from the Cortex data.