Comparing apples and oranges: a retrospective linked data study assessing the concordance between hospital discharge data, electronic health records and register books for diagnosis of inpatient admissions of miscarriage
Indra San Lazaro Campillo MPh (1,2), Sarah Meaney PhD (1,2), Maria Harrington (3), Karen McNamara FRCOG (1,4), Anna Maria Verling MA (1,4), Paul Corcoran PhD (2), Keelin O’Donoghue, PhD FRCOG (1,4)
(1) Pregnancy Loss Research Group, The Irish Centre for Fetal and Neonatal Translational Research University College Cork, Ireland; (2) National Perinatal Epidemiology Centre, Department of Obstetrics and Gynaecology, University College Cork, Ireland; (3) Cork University Hospital, Wilton, Cork Ireland; (4) Cork University Maternity Hospital, Wilton, Cork Ireland
Despite the high occurrence of miscarriage, there is a limited research assessing the validity of the outcome of a diagnosis of miscarriage at hospital settings. The aim of this study was to determine agreement between the Hospital Inpatient-Enquiry (HIPE) system, the electronic health records (EHR) and register books for the diagnosis of miscarriage.
Retrospective chart review comparing agreement of diagnosis of inpatient admissions of miscarriage between three data sources from January to June 2017 at a single, tertiary maternity hospital in Ireland. Kappa (k), sensitivity, specificity, positive and negative predictive value (PPV & NPV) were calculated using Stata.
After excluding missing data, this study identified 304 diagnoses of miscarriage out of 370 records in the EHR, 291 out of 360 records in HIPE, and 219 out of 255 records in register books. Using the EHR as a gold standard, HIPE had a sensitivity of 96.3%, specificity of 97.3%, PPV of 98.3%, NPV of 90.0%, with a very good strength of agreement (k=0.92; p-value < 0.001). Using the EHR as a gold standard, register books had a sensitivity of 97.2%, specificity of 80.5%, PPV of 96.3%, NPV of 84.6%, with a good strength level of agreement (k=0.79; p-value < 0.001). Only 4.2% (n=16) were classified as missed miscarriage according to HIPE, whereas more than 40% of admissions were classified as missed miscarriage by EHR and register books (n=173, 44.9% and n=150, 39.9%). Almost 60% of admissions were incomplete miscarriage according to HIPE (n=231).
Conclusion: Our findings indicate that HIPE and EHR are reliable and valid databases for monitoring and reporting prevalence of miscarriage in Ireland. However, discrepancies were found when classifying miscarriage between the three data sources Standardisation of these classifications are essential in order to ensure hospitals are appropriately resourced for women who miscarry.
This study received ethical approval from the Clinical Research Ethics Committee of the Cork Teaching Hospital on ECM 4 (I) 17/10/2017. A patient consent form was not required by the Ethics Committee because this was an observational study which did not include any intervention and which examined routinely collected data. Privacy was of the utmost importance precaution in this study. To assure this, all data were collected and stored within the hospital in strict confidence. In addition, only anonymous data were extracted and coded with a Medical Record Number (MRN).