Structured data is the key for advanced research finding normative valuesWeb 1400 x 480 px brez

Structured Data is the Key for Advanced Research: Finding Normative Values

Published December 09, 2020
Emil Plesnik
Emil Plesnik

Recently, I had the honour of taking part in a paediatric cardiology research project which was led by: University Children’s Hospital Ljubljana (UCHL), Department of Endocrinology, Diabetes and Metabolism; University Medical Centre Ljubljana (UMCL); and, the Faculty of Medicine, University of Ljubljana. The findings of the research were revealed in an original research article, titled “Carotid Intima-Media Thickness in Healthy Children and Adolescents: Normative Data and Systematic Literature Review” which was recently published in the journal Frontiers in Cardiovascular Medicine [1]. The main objective of this research was to determine carotid intima-media thickness (cIMT) normative values in a healthy population of children, which would then enable a comparison and assessment of cIMT measurements, and enhance their clinical usefulness as an atherosclerosis indicator. 

I believe that the research will add value to clinical practice and improve preventive care. My primary role in the research was to carry out all of the statistical analyses, and provide statistical advice and visual presentations of the results. For that reason, in this blog post I would like to highlight the significance of structured data in advanced medical research. 

Normative values, or normative data, is comprised of observations which describe what is usual or expected in a defined reference population, and at a specific point or period of time [2,3]. This information is of enormous importance to primary care physicians, and is essential for describing clinical conditions, developing standards of care, and establishing illness classifications. Normative data are used to compare the characteristics of specific conditions within a group of people, or an individual person, with data for the average person of a reference population. In this way, the reference population indicates what is normal, and the normative information helps to identify deviations from these norms. 

In medicine, normative values have most often been used in paediatrics to assess child growth. It has also become a popular method of interpreting several clinical or biological parameters, such as spirometric values, serum creatinine, cognitive states, and health-related quality of life. Norm-based comparisons have become a well-accepted and widely used interpretation strategy, as no gold standard is available to interpret complex and multidimensional health outcomes.  

To assure a meaningful interpretation using normative data, it is important to clearly indicate the population for which the norms have been collected, as the interpretation is only meaningful to individuals or groups that belong to the reference population. So, a representative population sample is the key to quality normative data. Representativeness can be achieved at different levels (e.g. national, regional, or local) and for different purposes (disease-specific), or institutions (primary, secondary, or tertiary care). Therefore, an exact definition of the purpose and level of the normative data is crucial for correctly collecting and selecting the data. 

The last step is reporting the normative data. This includes data exploration and analysis, which results in a summary of descriptive information for the total population, which is then separated for age and gender groups. The descriptive information is given in norm tables and figures that are used to visualise the distribution of the assessed characteristics in the reference population. Most often, descriptive information about a sample includes summaries of distribution, such as mean value and standard deviation and percentiles, which are a base for acquiring the median, interquartile range, etc. Percentiles are easy to use and interpret, and also allow for quantifying the score difference for an individual over time, or between two different individuals or groups. There are several established methods to estimate percentiles for a population sample, for example, the LMS and BCPE methods [4]. 

At Better Innovation, we have taken part in research with the goal of determining normative values for a disease-specific clinical case in paediatrics, and this has just recently been published [1]. The main objective of this research was to determine cIMT normative values in a healthy population of children in order to enable a comparison and assessment of cIMT measurements and enhance this information’s clinical usefulness as a non-invasive atherosclerosis biomarker. Early identification of children at risk of atherosclerosis is of paramount importance for implementing primary preventive measures which address vascular health. The cIMT is a thickness measurement of the innermost two layers of the wall of the head and neck’s major artery, and it can be measured using an external ultrasound, and is thus non-invasive, comparatively low in cost, and convenient. The availability of normative values for a clearly defined reference population further strengthens the benefits of this non-invasive biomarker. 

The collection and creation of normative data is a complex, demanding task, and researchers who tackle it require the proper tools to do it. Better Platform is designed to store, manage, query, retrieve, and exchange structured electronic health record data for different levels of healthcare. It enables the agile and transparent formation of representative population samples, which is fundamental for determining normative values. 

[1] Drole Torkar A, Plesnik E, Groselj U, Battelino T, Kotnik P. Carotid Intima-Media Thickness in Healthy Children and Adolescents: Normative Data and Systematic Literature Review. Front Cardiovasc Med 2020;7.

[2] O’Connor PJ. Normative data: their definition, interpretation, and importance for primary care physicians. Fam Med 1990;22:307–11.

[3] Schmidt S, Pardo Y. Normative Data. In: Michalos AC, editor. Encycl. Qual. Life Well- Res., Dordrecht: Springer Netherlands; 2014, p. 4375–9.

[4] Indrayan A. Demystifying LMS and BCPE methods of centile estimation for growth and other health parameters. Indian Pediatrics 2014;51:37–43.

Emil Plesnik

Written by Emil Plesnik

Data scientist, holding a PhD degree in biomedical data analysis. Curious about data analytics, statistics, machine learning and programming and their use to develop data-driven solutions to challenges in healthcare.

Subscribe to a group that wants to know what’s better

Join the discussion