Breaking Barriers in International Psoriasis Research Through Data Standardisation
About
Europe consists of many different countries and nearly all of them have their own unique healthcare system. The data is structured and coded based on local and countrywide standards. This makes it very challenging to do international European research based on data analytics. To enable optimal use of data, it is critical to make sure the relevant data sets are harmonised as much as possible. This means that all data points measure the same thing (ensuring the data sets are standardised) and that the data speaks the same ‘language’.
The Observational Health Data Sciences and Informatics (OHDSI) program, an international multi-stakeholder collaborative, has developed The Observational Medical Outcomes Partnership, or the OMOP Common Data Model to tackle this issue. OMOP is defined as an open community data standard, designed to standardise the structure and content of observational data and to enable efficient analyses that can produce reliable evidence.
Currently, OMOP use is driven by an enthusiastic community, but coverage is still low, which seriously affects its current usefulness. Each (local) data point needs to be mapped and restructured to align with OMOP. Doing this manually is extremely time and labour intensive. Automating this process increases the applicability of OMOP significantly. To address this, LOGEX embarked on a project to automate the mapping, restructuring, and recoding process for healthcare data of psoriasis patients from England, Sweden, and the Netherlands.
The challenge
Psoriasis is a condition that approximately 1.5% of the European population suffers from. The condition can have a profound negative effect on the life of patients. To remedy this, a significant number of new medicines have entered the market in the last few years. Currently, there are more than 15 different medicines to choose from for the treatment of advanced psoriasis. Keeping in mind that these medicines can be prescribed in combination with one another, or in many consecutive orders, it is clear that there is an almost unlimited number of potential care paths.
Having so many options is a recipe for uncertainty. It is unclear what exact care path is most effective for specific patient cohorts. A real-world data evaluation is a good tactic for closing this knowledge gap. But with the high number of care path options, it is unlikely that all possible care paths are used in a single country. Including data of several countries will add significant value to such a project.
The European Psoriasis Observatory does precisely this. It gathers diagnostics, treatment and outcome data from psoriasis patients from five different countries to analyse care paths and to find patterns of effectiveness for different patient cohorts.
However, since the participating countries use different data structures, different data points and different data languages, it is not possible to simply put the data together for analytics. Instead, the data first needs to be standardised and harmonised so it is comparable and analysable.
The solution
The first step to solve this problem was to scan what information was relevant. The team determined which data points from the available data sources were indicative of the followed care path, of the reasons why this path was chosen (patient characteristics) and of what the clinical outcome was. In other words: a standard data set was developed.
After the data set was defined, it was necessary to look through the different national data sets to see where the relevant data points were hiding. Our team started by studying the English, Swedish and Dutch data. After having completed the search for all relevant data points, a detailed data map that points to the precise location of each relevant data point within the gathered data was created.
However, this inventory did bring to light some inconsistencies between the different national data sets. One specific example of this the categorisation of light therapy, which differs in each country (see table below for details).
Country Code System Code Description NL NZa 039992 Light therapy, whether or not supported by drug-based photosensitizing therapy, per treatment. GB OPCS S12.1 Ultraviolet A light therapy to skin GB OPCS S12.2 Ultraviolet B light therapy to skin SE KVÅ DQ012 Light treatment, UVA SE KVÅ DQ014 Light treatment, UVB
To solve this issue, the team resorted to using SNOMED CT, which stands for Systematised Nomenclature of Medicine — Clinical Terms. This is a comprehensive, multilingual healthcare terminology system designed to provide a standardised way to represent medical information. The different light therapy codes were mapped to the SNOMED CT code 31394004 – Light therapy (procedure). This way, the analyst will not need to know the in-depth details per country and make comparisons based on a slightly higher level of information.
For some countries the mapping is provided by the health authorities, but for other countries LOGEX has created its own mappings, in some cases making use of AI Large Language Models to assist in the process.
But, even with these psoriasis data maps created, there would still be the need to transform the national data points to the international OMOP format. Rather than doing this manually per data point, our data engineers further developed the ‘transformer machine’ they previously worked on for past real-world data projects. They ensured it could automatically find the right data points in the national data, guided by the psoriasis data maps, and subsequently transform those data to OMOP. With this ‘transformer machine’, the Psoriasis Observatory has been future-proofed: any future data is automatically standardised and transformed, so it can be used for analysis. The methodology behind the transformer machine can even be adapted and for other therapeutic areas.
“By creating this solution, using OMOP and SNOMED CT, and thus harmonising the data, our data scientists can now truly perform scalable analytics across multiple countries. This enables them to focus on the value in the data in order to help patients get the best possible treatment in the future.”
Laurens Krüger, Senior Product Manager at LOGEX.
The impact
“Now that this process has been finalised for the English, Swedish and Dutch data, with Finland and Spain up next, it is possible to process these data much faster. In practice, with the automation of the data mapping and transformation of the data to OMOP, about 80% of the work is already done. This means five times more time and resources to do relevant analytics on the international data sets,”
Laurens Krüger, Senior Product Manager at LOGEX
With the knowledge that can be gained from this research, the effectiveness of pathways can be quantified, and uncertainty can be reduced. Rather than clinicians having to take their patients through as process of trial and error, they now have a much more informed way to reach a beneficial clinical outcome quickly. This is better for the patient, better for the clinician, and better for society.
Would you like to know how we can help you derive richer insights from Real-World data?