Automated data cleaning of paediatric anthropometric data from longitudinal electronic health records: protocol and application to a large patient cohort

Abstract

'Big data' in healthcare encompass measurements collated from multiple sources with various degrees of data quality. These data require quality control assessment to optimise quality for clinical management and for robust large-scale data analysis in healthcare research. Height and weight data represent one of the most abundantly recorded health statistics. The shift to electronic recording of anthropometric measurements in electronic healthcare records, has rapidly inflated the number of measurements. WHO guidelines inform removal of population-based extreme outliers but an absence of tools limits cleaning of longitudinal anthropometric measurements. We developed and optimised a protocol for cleaning paediatric height and weight data that incorporates outlier detection using robust linear regression methodology using a manually curated set of 6,279 patients' longitudinal measurements. The protocol was then applied to a cohort of 200,000 patient records collected from 60,000 paediatric patients attending a regional teaching hospital in South England. WHO guidelines detected biologically implausible data in <1% of records. Additional error rates of 3% and 0.2% for height and weight respectively were detected using the protocol. Inflated error rates for height measurements were largely due to small but physiologically implausible decreases in height. Lowest error rates were observed when data was measured and digitally recorded by staff routinely required to do so. The protocol successfully automates the parsing of implausible and poor quality height and weight data from a voluminous longitudinal dataset and standardises the quality assessment of data for clinical and research applications.

Figure 1

Percentage of datapoints identified as true errors in the gold standard dataset stratified by year for weight and height, weight for height. Outliers were split into three types: height outlier flagging using linear regression (LR), height entry error with adult height check and height with height decrease check.

Figure 2

Manual outlier curation results of UHS gold standard paediatric height and weight data: (a) Percentage of outliers for each of the occupation categories for weight, height using LR, height with adult height check, and height with height decrease check. (b) Percentage of outliers for each of the department categories for weight, height using LR, height with adult height check, and height with height decrease check.

Figure 3

UHS age 2–20 years’ height and weight data (1932–2018) summary: (a) Number of patients and records of height and weight, broken down by number of datapoints per patients. (b) Total number of height, weight and BMI measurements over time from prior to 2008 to 2018 (c) Percentage of data flagged by WHO guidelines over time. (d) Number of patients within groups of patients defined by their number of longitudinal datapoints for height and weight. (e) Number of height and weight records per group of patients binned by number of datapoints per patient.

Figure 4

One decimal place digit distribution for height and weight measurements, demonstrating the bias in recording height and weight measurements, rounding to the precision of kg for weight and the precision of cm or 0.5 cm for height. This bias is reflected in the Myers’ index of height and weight measurements.

Figure 5

UHS data characterisation by occupation and by department of staff entering the data (a) Weight records by occupation (b) Height records by occupation (c) Percentage of height and weight data flagged by WHO rules by occupation (d) Weight records by department (e) Height records by department (f) Percentage of height and weight data flagged by WHO rules by department.