ANALYSIS of REAL-WORLD DATA to IDENTIFY PREDICTORS for EARLIER DIAGNOSIS of APLASTIC ANEMIA.
No Thumbnail Available
All Authors
Scheinberg, P.
Bruemmendorf, T.H.
De Latour, R.P.
Dufour, C.
Gokani, R.
Griffin, M.
Hey-Hadavi, J.H.
Kanda, Y.
Mitchell, E.
Mozejko-Pastewka, B.
LTHT Author
Griffin, Morag
LTHT Department
Oncology
Haematology
Haematology
Non Medic
Publication Date
2024
Item Type
Conference Abstract
Language
Subject
ACUTE KIDNEY INJURY , ADOLESCENT , ADULT , AGED , ANAEMIA , ANAEMIA, APLASTIC , BONE MARROW EXAMINATION , CHILD , COMORBIDITY , CONTROLLED CLINICAL TRIALS AS TOPIC , DIAGNOSIS , DIAGNOSIS, DIFFERENTIAL , DRUG THERAPY , ENCEPHALITIS , WOMEN , HEART FAILURE , LIVER FAILURE , MEN , MIDDLE AGED , PRESCRIPTIONS , MALNUTRITION , VASCULITIS
Subject Headings
Abstract
Background: Aplastic Anemia (AA) is a rare and life-threatening disorder. Diagnosis is challenging and can be delayed. Aim(s): To identify predictors of AA and whether a shorter time to diagnosis is feasible for earlier intervention and treatment. Method(s): Data Source: All-payor US medical and prescription claims data (Veeva Compass) of 330M people from January 2017 to September 2023. Population At Risk (PAR) and Incident Cases: The PAR for undiagnosed AA and its differential diagnosis (DDx) included patients with peripheral cytopenia of 2 blood cell types who had 1 AA and/or DDx related claim(s) from any clinical setting. To be classified as likely incident cases, patients required a bone marrow biopsy before the first AA or DDx claim. Patients were further categorized as cases of AA or DDx based on the diagnosis with most claims. Modeling: Predictors of undiagnosed AA were identified via logistic regression with DDx as a reference group. Then, T0-TDx was estimated using a moving average to calculate claims volumes over time, defining T0 as the start of accelerating claims volumes, and defining TDx as the time of the first AA or DDx claim. Result(s): From the identified >850K patients at risk, there were an estimated 1,937 incident cases of all AAs (Fig. 1a) with a bimodal distribution that skews towards the elderly (Fig. 1b) and females (55%). The median T0-TDx for AA was 95 days. AA patients had significantly more claims than healthy controls for 3 years before TDx and claims volume increased sharply at one-year pre-diagnosis. Patients exhibited notable heterogeneity in claims volume. Four distinct clusters were identified with a between-cluster difference in T0-TDx (Kruskal-Wallis, p<0.001). Clusters 1 and 3 had less intense management, the shortest T0-TDx, and the longest time to treatment (TRx), suggesting less severe AA. By contrast, clusters 2 and 0 had a longer T0-TDx, with more intense management and hepatic failure (OR=11.8, p<0.001), respectively. The latter had the longest T0-TDx. Cluster 2 had the 2nd longest T0-TDx but the shortest TDx-TRx. The probability of undiagnosed AA was lower in patients with symptomatic monocytosis and relapsing leukemia, and higher in those with nutritional marasmus, congenital anemias, diffuse eosinophilic fasciitis, PNH and reactive hepatitis. AA was diagnosed earlier in patients with R-sided heart failure, hepatitis, fluid overload, acute kidney failure and severe AA, and later in patients with transfusion-associated fluid overload, alkalosis, encephalitis and cutaneous vasculitis. Summary/Conclusion: This is the largest epidemiological analysis of incident AA across all real-world clinical settings, making it more representative of the true epidemiology of AA. An older cohort of patients was identified than in classical literature1, reflecting a potential referral bias against the elderly for care at tertiary or trial centres. Further research is needed to confirm this finding. This study also identifies variables that may increase the index of suspicion for AA, potentially enabling earlier diagnosis. Delays in diagnosis may be explained by certain co-morbidities confounding the clinical picture. Further validation of this model is required using a dataset of adjudicated AA patients. (Figure present).
Journal
HemaSphere