Validity of The Danish National Chronic Myeloid Neoplasia Registry
Dustin Andersen Patel1,2, Karina Kannik3, Iman Chanchiri4
1Section of General Practice and Research Unit for General Practice, Department of Public Health, University of Copenhagen, Copenhagen, Denmark.
Purpose:
Myeloproliferative neoplasms (MPN) are rare hematological malignancies with long survival, making real-world evidence essential to complement clinical trial data. The Danish National Chronic Myeloid Neoplasia Registry (DMR) captures nearly all MPN patients nationwide, but its validity has not been systematically evaluated. We aimed to validate key clinical variables in the DMR through a nationwide audit.
Methods:
We sampled DMR entries from 372 persons registered with essential thrombocythemia (ET), polycythemia vera (PV), primary myelofibrosis (PMF), prefibrotic myelofibrosis (preMF), or unclassifiable MPN (MPN-U) between 2017 and 2021 across all regions of Denmark. We validated eight main variables, including 27 subordinate variables of clinical importance from the DMR, against electronic health records using standardized audit procedures. Positive predictive value (PPV), negative predictive value (NPV), sensitivity, specificity, and accuracy were calculated.
Results:
Overall, the DMR showed mostly high PPV and intermediate/high NPV values. PPV was high for core diagnostic, treatment and comorbidity variables: ET (0.91), PV (0.95) and PMF (0.97), date of diagnosis (0.97), splenomegaly (0.98), JAK2V617F (0.97), CALR (0.92), hypertension (0.93), hyperlipidemia (0.90), diabetes (0.97), and common treatments (phlebotomy, hydroxyurea, interferon, ruxolitinib ≥.93). Intermediate PPV was observed for diagnoses of preMF (0.78) and MPN-U (0.77), and for the remaining comorbidity variables (0.75-0.85). Most misclassifications arose from false negatives, reflecting the automatic imputation of negative (no) for missing values for selected variables in the DMR which resulted in lower sensitivity and high PPV/specificity.
Conclusion:
The DMR demonstrates high validity for several clinically relevant variables and constitutes a reliable source of real-world data in MPN research and therefore, crosslinking registry variables with other data sources may enable robust epidemiological studies and unprecedented precision in real-world evidence for MPN patients. Nevertheless, variable documentation and data quality must be critically assessed to account for changes in data entry practices across time periods.
Insights
The Danish National Chronic Myeloid Neoplasia Registry (DMR) shows high validity for key myeloproliferative neoplasms (MPN) data, confirming its reliability for real-world evidence. This supports its use in epidemiological studies and improving MPN patient care.
Area of Science:
- Hematology
- Epidemiology
- Medical Informatics
Background:
- Myeloproliferative neoplasms (MPN) are rare, chronic hematological malignancies requiring long-term management.
- Real-world evidence (RWE) is crucial for understanding MPN patient outcomes beyond clinical trials.
- The Danish National Chronic Myeloid Neoplasia Registry (DMR) is a comprehensive national database for MPN patients.
Purpose of the Study:
- To systematically evaluate the validity of key clinical variables within the Danish National Chronic Myeloid Neoplasia Registry (DMR).
- To assess the reliability of the DMR for capturing essential thrombocythemia (ET), polycythemia vera (PV), and primary myelofibrosis (PMF) data.
- To determine the accuracy of DMR data for diagnoses, treatments, and comorbidities in MPN patients.
Main Methods:
- A nationwide audit was conducted, validating 8 main and 27 subordinate variables from the DMR.
- Data from 372 MPN patients diagnosed between 2017-2021 were compared against electronic health records.
- Key performance metrics including positive predictive value (PPV), negative predictive value (NPV), sensitivity, specificity, and accuracy were calculated.
Main Results:
- The DMR demonstrated high positive predictive values (PPV) for core diagnostic, treatment, and comorbidity variables (e.g., PV PPV=0.95, JAK2V617F PPV=0.97).
- Intermediate PPV was noted for diagnoses of prefibrotic myelofibrosis (preMF) and unclassifiable MPN (MPN-U).
- Most misclassifications were false negatives due to automatic imputation of missing values, leading to lower sensitivity but high PPV and specificity.
Conclusions:
- The DMR is a largely valid and reliable source for real-world data in MPN research.
- Crosslinking DMR data with other sources can facilitate robust epidemiological studies and enhance MPN RWE.
- Critical assessment of variable documentation and data quality is necessary, considering potential changes in data entry practices over time.
