Data mining of digitized health records in a resource-constrained setting reveals that timely immunophenotyping is associated with improved breast cancer outcomes

Abstract

BackgroundOrganizations that issue guidance on breast cancer recommend the use of immunohistochemistry (IHC) for providing appropriate and precise care. However, little focus has been directed to the identification of maximum allowable turnaround times for IHC, which is necessary given the diversity of hospital settings in the world. Much less effort has been committed to the development of digital tools that allow hospital administrators to monitor service utilization histories of their patients.MethodsIn this retrospective cohort study, we reviewed electronic and paper medical records of all suspected breast cancer patients treated at one secondary-care hospital of the Mexican Institute of Social Security (IMSS), located in western Mexico. We then followed three years of medical history of those patients with IHC testing.ResultsIn 2014, there were 402 breast cancer patients, of which 30 (7.4% of total) were tested for some IHC biomarker (ER, PR, HER2). The subtyping allowed doctors to adjust (56.7%) or confirm (43.3%) the initial therapeutic regimen. The average turnaround time was 56 days. Opportune IHC testing was found to be beneficial when it was available before or during the first rounds of chemotherapy.ConclusionsThe use of data mining tools applied to health record data revealed that there is an association between timely immunohistochemistry and improved outcomes in breast cancer patients. Based on this finding, inclusion of turnaround time in clinical guidelines is recommended. As much of the health data in the country becomes digitized, our visualization tools allow a digital dashboard of the hospital service utilization histories.Electronic supplementary materialThe online version of this article (10.1186/s12885-018-4833-4) contains supplementary material, which is available to authorized users.

A Competing Risk Analysis of Early COVID-19 Treatments

“…Furthermore, the emergence of virtual randomized clinical trials has shown the value to accelerate results and provide informed guidance to patients ( 37 ). The digitization of electronic health records is possible and should be an overall strategy in resource-constrained settings to provide a data-driven approach to public health concerns ( 38 ). Machine learning models and data science can be effectively applied to these settings augmenting the capabilities of the healthcare system.…”