Uncovering interpretable potential confounders in electronic medical records

AbstractIn medicine, randomized clinical trials (RCT) are the gold standard for informing treatment decisions. Observational comparative effectiveness research (CER) is often plagued by selection bias, and expert-selected covariates may not be sufficient to adjust for confounding. We explore how the unstructured clinical text in electronic medical records (EMR) can be used to reduce selection bias and improve medical practice. We develop a method based on natural language processing to uncover interpretable potential confounders from the clinical text. We validate our method by comparing the hazard ratio (HR) from survival analysis with and without the confounders against the results from established RCTs. We apply our method to four study cohorts built from localized prostate and lung cancer datasets from the Stanford Cancer Institute Research Database and show that our method adjusts the HR estimate towards the RCT results. We further confirm that the uncovered terms can be interpreted by an oncologist as potential confounders. This research helps enable more credible causal inference using data from EMRs, offers a transparent way to improve the design of observational CER, and could inform high-stake medical decisions. Our method can also be applied to studies within and beyond medicine to extract important information from observational data to support decisions.

Download Full-text

908-P: Predictive Model for Glycemic Level of People with Diabetes Mellitus Using Data Extracted from Electronic Medical Records in Primary Care

Diabetes ◽

10.2337/db20-908-p ◽

2020 ◽

Vol 69 (Supplement 1) ◽

pp. 908-P

Author(s):

SOSTENES MISTRO ◽

THALITA V.O. AGUIAR ◽

VANESSA V. CERQUEIRA ◽

KELLE O. SILVA ◽

JOSÉ A. LOUZADO ◽

...

Keyword(s):

Diabetes Mellitus ◽

Primary Care ◽

Predictive Model ◽

Electronic Medical Records ◽

Medical Records ◽

Glycemic Level ◽

Using Data

Download Full-text

Natural Language Processing and Electronic Medical Records—Reply

JAMA ◽

10.1001/jama.2011.1781 ◽

2011 ◽

Vol 306 (21) ◽

pp. 2325

Author(s):

Harvey J. Murff

Keyword(s):

Natural Language Processing ◽

Natural Language ◽

Electronic Medical Records ◽

Language Processing ◽

Medical Records

Download Full-text

Assessment of vitamin D deficiency and COVID-19 diagnosis in patients with breast or prostate cancer using electronic medical records.

Journal of Clinical Oncology ◽

10.1200/jco.2021.39.15_suppl.6589 ◽

2021 ◽

Vol 39 (15_suppl) ◽

pp. 6589-6589

Author(s):

Aaron Galaznik ◽

Emelly Rusli ◽

Vicki Wing ◽

Rahul Jain ◽

Sheila Diamond ◽

...

Keyword(s):

Breast Cancer ◽

Prostate Cancer ◽

Vitamin D ◽

Vitamin D Deficiency ◽

Cancer Patients ◽

Electronic Medical Records ◽

Medical Records ◽

Research Database ◽

Breast Cancer Cohort ◽

Icd 10

6589 Background: While patients with cancer are known to be at increased risk of infection in part due to the immunocompromising nature of cancer treatments, recent data indicate a particularly high risk for COVID-19 infection and poor outcomes (Wang et al., 2020). A recent study (Meltzer et al., 2020) demonstrated Vitamin D deficiency may increase risk of COVID-19 infection, and a small randomized controlled trial in Spain reported significant improvement in mortality among hospitalized patients treated with calcifediol. Vitamin D deficiency has been reported in two leading causes of cancer deaths: breast and prostate. In this study, we performed a retrospective cohort analysis on nationally representative electronic medical records (EMR) to assess whether Vitamin D deficiency affects risk of COVID-19 among these patients. Methods: Patients with breast (female) or prostate (male) cancer were identified between 3/1/2018 and 3/1/2020 from EMR data provided pro-bono by the COVID-19 Research Database ( covid19researchdatabase.org ). Patients with an ICD-10 code for Vitamin D deficiency or < 20ng/mL 20(OH)D laboratory result within 12 months prior to 3/1/2020 were classified as Vitamin D deficient. COVID-19 diagnosis was defined using ICD-10 codes and laboratory results for COVID-19 at any time after 3/1/2020. Logistic regressions, adjusting for baseline demographic and clinical characteristics, were conducted to estimate the effect of Vitamin D deficiency on COVID-19 incidence in each cancer cohort. Results: A total of 16,287 breast cancer and 14,919 prostate cancer patients were included in the study. The average age was 68.9 years in the breast cancer cohort and 73.6 years in the prostate cancer cohort. The breast cancer cohort consisted of 85% Whites, 13% Black or African Americans, and less than 5% of other races. A similar race distribution was observed in the prostate cancer cohort. Unadjusted analysis showed the risk of COVID-19 was higher among Vitamin D deficient patients compared to non-deficient patients in both cohorts (breast: OR = 1.60 [95% C.I.: 1.15, 2.20]; prostate: OR = 1.59 [95% C.I.: 1.08, 2.33]). Similar findings were observed when assessed in subgroups of patients with newly diagnosed cancer in the dataset, as well as after adjusting for baseline characteristics. Conclusions: Our study suggests breast and prostate cancer patients may have an elevated risk of COVID-19 infection if Vitamin D deficient. These results support findings by Meltzer et al., 2020 demonstrating a relationship between Vitamin D deficiency and COVID-19 infection. While a randomized clinical trial is warranted to confirm the role for Vitamin D supplementation in preventing COVID-19, our study underscores the importance of monitoring Vitamin D levels across and within cancer populations, particularly in the midst of the global COVID-19 pandemic.

Download Full-text

Comparing the Clinical Characteristics and Mortality of Residential and Non-Residential Older People with COVID-19: Retrospective Observational Study

International Journal of Environmental Research and Public Health ◽

10.3390/ijerph19010483 ◽

2022 ◽

Vol 19 (1) ◽

pp. 483

Author(s):

Francesc X. Marin-Gomez ◽

Jacobo Mendioroz-Peña ◽

Miguel-Angel Mayer ◽

Leonardo Méndez-Boo ◽

Núria Mora ◽

...

Keyword(s):

Risk Factors ◽

Electronic Medical Records ◽

Medical Records ◽

Clinical Characteristics ◽

Fold Increase ◽

The Elderly ◽

Health Administration ◽

Epidemic Outbreak ◽

Risk Of Death ◽

Using Data

Nursing homes have accounted for a significant part of SARS-CoV-2 mortality, causing great social alarm. Using data collected from electronic medical records of 1,319,839 institutionalised and non-institutionalised persons ≥ 65 years, the present study investigated the epidemiology and differential characteristics between these two population groups. Our results showed that the form of presentation of the epidemic outbreak, as well as some risk factors, are different among the elderly institutionalised population with respect to those who are not. In addition to a twenty-fold increase in the rate of adjusted mortality among institutionalised individuals, the peak incidence was delayed by approximately three weeks. Having dementia was shown to be a risk factor for death, and, unlike the non-institutionalised group, neither obesity nor age were shown to be significantly associated with the risk of death among the institutionalised. These differential characteristics should be able to guide the actions to be taken by the health administration in the event of a similar infectious situation among institutionalised elderly people.

Download Full-text