Small for Gestational Age Coded Diagnoses in Aotearoa New Zealand's Administrative Health Datasets: A Validation Study
ABSTRACT Background and Aims Inaccurate coding of small for gestational age (SGA) infants in routinely collected health data has implications for research based on those data. We aimed to estimate the sensitivity and specificity of coded SGA diagnoses in New Zealand's routinely collected hospit...
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Wiley
2025-05-01
|
| Series: | Health Science Reports |
| Subjects: | |
| Online Access: | https://doi.org/10.1002/hsr2.70610 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | ABSTRACT Background and Aims Inaccurate coding of small for gestational age (SGA) infants in routinely collected health data has implications for research based on those data. We aimed to estimate the sensitivity and specificity of coded SGA diagnoses in New Zealand's routinely collected hospitalisation and mortality data, and determine whether sensitivity and specificity varied by infant, pregnancy, and maternal characteristics. Methods We estimated birthweight centiles of live and stillborn infants delivered in New Zealand between 2005 and 2020 using the Fenton Population Reference Calculator and the GROW Customised Bulk Centile Calculator (New Zealand version); values of the relevant variables (including gestational age, birthweight, infant sex, and others) were sourced from routinely collected national health data. We compared the SGA status derived from the calculators with coded SGA diagnoses (ICD‐10‐AM P051) in hospitalisation and mortality data. We estimated sensitivity and specificity ratios comparing coded diagnoses with each of the birthweight calculators using a generalised linear model, adjusting for infant, pregnancy, and maternal characteristics. Results This analysis included 887,871 infants, with 15,850 (1.8%) having a coded SGA diagnosis. By contrast, the number and proportion of babies classified as SGA using the Fenton and GROW calculators were 80,541 (9.1%) and 138,866 (15.6%), respectively. Overall, compared with the Fenton calculator, the sensitivity of coded SGA diagnoses was 13.1% (specificity 99.3%). Compared with the GROW calculator, the sensitivity was 9.8% (specificity 99.7%). Conclusion In New Zealand, population‐level research involving SGA diagnoses should derive birthweight centiles using an appropriate calculator instead of using ICD‐10‐AM coded diagnoses. |
|---|---|
| ISSN: | 2398-8835 |