Evaluation of routine data sources for ascertainment of hypospadias cases

1. Evaluation of routine data sources for ascertainment of hypospadias cases Congenital Anomaly Register for Oxfordshire, Berkshire, and Buckinghamshire (CAROBB)

2. Introduction • Background • What is hypospadias? • Why hypospadias? • Evaluation of data sources by validation of diagnostic information

3. Hypospadias Source: (Patient UK 2011)

4. Why is accurate ascertainment important? • Linked to environmental pollutants • Is it increasing in prevalence? • Wide variation (5–30 per 10,000 births, 2005-12) • Real difference in prevalence or ascertainment differences? • Multiple treatment options, more evidence needed re: effectiveness • Little long term follow up data

5. Ascertainment • Ascertainment difficulties well documented* • Postnatally diagnosed, not requiring any/immediate intervention • Variation in threshold for surgery • Wide variation in terminology used to describe hypospadias making coding difficult • CAROBB difficulties with validation • Uncertain which sources were accurate • Follow up information unreliable from hospital systems *EUROCAT Special Report: An Assessment and Analysis of Surveillance Data in Europe, 2003. (http://www.eurocat-network.eu) *Dolk H et al (2004) Toward the effective surveillance of hypospadias. Environ Health Perspect 112 (3), 398-402

6. Objectives • Ascertain all reports of babies with hypospadias, born within a defined region and timeframe (2005-2010). • Validate each report of hypospadias by comparison with medical records to discover the true diagnosis • Evaluate ascertainment sources (birth reports and paediatric admission lists) for accuracy and completeness of diagnostic information • Propose a future hypospadias ascertainment strategy

7. Search • Inclusion: DOB: 2005 -2010 Born in Oxfordshire • Search ICD 10 codes the two sources to be evaluated: birth reports and paediatric admissions lists: • Hypospadias: Q54.0 - 3 or Q54.8 - 9 • Other codes: Q54.4 or Q55.6 or Q55.8 or Q56* • Plus records on CAROBB notified from other sources • HES data +

8. Potential casesfrom search results Diaries, clinic letters, audits HES data

9. Validation • Find ‘True Diagnosis’ • Notes retrieved for every case (or not) • Many consultant notes had a diagram of the position of the opening – great! • Many different classifications (Hadidi, 2004)

10. Validation of all hypospadias records

11. Comparison of completeness of reporting from the evaluated sources by severity

12. Relationship between sources of notification and the true diagnosis of hypospadias * #from clinician notification *Not coded as having hypospadias so would not be found on standard search of ICD 10s or from clinicians only but excluding 36 true negatives ie not coded as hyp and did not have hyp

13. Results summary When searching for hypospadias cases using specific hypospadias ICD10 codes: • Sensitivity • Birth reports identified 75% of total cases • Paediatric lists identified 76% of total cases • Both sources combined identified 97% of total cases • Positive predictive value • Birth reports: 65% of records identified had hypospadias confirmed • Paediatric lists: 96% of records identified had hypospadias confirmed Also – coding accuracy of evaluated sources Coding of the severity of the hypospadias was either unspecified or inaccurate in both sources and could not be relied upon.

14. Specific accuracy of ICD-10 coding Paediatric lists Birth reports Specific correct: eg, coded as mild in the evaluated source and true diagnosis was mild Wrong specific: eg, coded as mild in the evaluated source and true diagnosis was severe

15. Conclusion • When both sources are combined, they offer adequate ascertainment of hypospadias • BUT urology / plastics clinical notes needed to confirm: • hypospadias presence / absence in birth reports • any detail on position of urethral opening • Identifying hypospadias via routine data sources is the first of many steps to good ascertainment

16. Thanks! • Jenny Kurinczuk, Patricia Boyd, Kay Randall and Jane Forrester-Barker at NPEU • Phil Harper at Oxford Brookes University • Staff at Oxford University Hospitals NHS Trust

18. Strategy for ascertainment of hypospadias in the region studied • From the moderately low positive predictive value of birth reports, it cannot be assumed that a baby has hypospadias if he is only included in a birth anomaly report with a hypospadias code and not on the paediatric list. The case must be validated. • From the high positive predictive value of paediatric admission lists, it is reasonable to assume that a baby has hypospadias if he is present on a paediatric admission list with a hypospadias code (provided a false positive rate of approximately 4% is acceptable). The case may be added to the register without validation but must be given an unspecified code until the position of the hypospadias has been validated. • Hypospadias ICD-10 codes should be searched for in both birth reports and paediatric admission lists to produce near complete ascertainment. • If resources allow, general genital ICD-10 codes should be investigated for hypospadias in both birth reports and paediatric admission lists. • When searching paediatric admission lists, the search should be extended to include children at least up to the age of four years to allow for delayed surgery. • To gain an accurate assessment of severity of hypospadias, all reports need validation via medical records and should be considered unspecified until they have been validated in this way. • Each case should contain a text description of the hypospadias position to help clarify the uncertain boundaries between categories and normality. • Each case should include an accurate record of where the case was ascertained and how it was validated, so that when the data are used for research, judgements can be made into the accuracy of the records.