Detecting Miscoded Diabetes Diagnosis Codes in Electronic Health Records for Quality Improvement: Temporal Deep Learning Approach

Can telehealth expansion boost health care utilization specifically for patients with substance use disorders relative to patients with other types of chronic disease?

OBJECTIVE

Patients with substance use disorders (SUDs) exhibit low healthcare utilization despite high risk of poor outcomes. Telehealth expansion may boost utilization, but it is unclear whether telehealth can increase utilization for patients with SUDs beyond that expected for other chronic diseases amenable to remote treatment, like type 2 diabetes. This information is needed by health systems striving to improve SUD outcomes, specifically. This study compared the impact of telehealth expansion during the COVID-19 public health emergency (PHE) on utilization for patients with SUDs and diabetes.

METHODS

Using Wisconsin Medicaid administrative, enrollment and claims data 12/1/2018-12/31/2020, this cohort study included nonpregnant, nondisabled adults 19-64 years with SUDs (N = 17,336) or diabetes (N = 8,499). Outcomes included having a primary care visit in the week (any, and telehealth) for any diagnosis, or a SUD or diabetes diagnosis; and the weekly fraction of visits completed by telehealth. Logistic and fractional regression examined outcomes pre- and post-PHE. Covariates included age, sex, race, ethnicity, income, geography, and comorbid medical and psychotic disorders.

RESULTS

Post-PHE, patients with SUDs exhibited greater likelihood of telehealth utilization (percentage point difference (PPD) per person-week: 0.2; 95% CI: 0.001-0.003; p<0.001) and greater fractional telehealth use (PPD: 1.8; 95%CI: 0.002-0.033; p = 0.025) than patients with diabetes despite a larger overall drop in visits (PPD: -0.5; 95%CI: -0.007- -0.003; p<0.001).

CONCLUSIONS

Following telehealth expansion, patients with SUDs exhibited greater likelihood of telehealth utilization than patients with diabetes. This advantage lessened the substantial PHE-induced healthcare disruption experienced by patients with SUDs. Telehealth may boost utilization for patients with SUDs.

Developing an LSTM Model to Identify Surgical Site Infections using Electronic Healthcare Records

Recently, hospitals and healthcare providers have made efforts to reduce surgical site infections as they are a major cause of surgical complications, a prominent reason for hospital readmission, and associated with significantly increased healthcare costs. Traditional surveillance methods for SSI rely on manual chart review, which can be laborious and costly. To assist the chart review process, we developed a long short-term memory (LSTM) model using structured electronic health record data to identify SSI. The top LSTM model resulted in an average precision (AP) of 0.570 [95% CI 0.567, 0.573] and area under the receiver operating characteristic curve (AUROC) of 0.905 [95% CI 0.904, 0.906] compared to the top traditional machine learning model, a random forest, which achieved 0.552 [95% CI 0.549, 0.555] AP and 0.899 [95% CI 0.898, 0.900] AUROC. Our LSTM model represents a step toward automated surveillance of SSIs, a critical component of quality improvement mechanisms.