INTRODUCTION:Applying Fair Information Practice principles to electronic health records (EHRs) requires allowing patient control over who views their data. METHODS: We designed a program that captures patients' preferences for provider access to an urban health system's EHR. Patients could allow or restrict providers' access to all data (diagnoses, medications, test results, reports, etc.) or only highly sensitive data (sexually transmitted infections, HIV/AIDS, drugs/alcohol, mental or reproductive health). Except for information in free-text reports, we redacted EHR data shown to providers according to patients' preferences. Providers could "break the glass" to display redacted information. We prospectively studied this system in one primary care clinic, noting redactions and when users "broke the glass," and surveyed providers about their experiences and opinions. RESULTS: Eight of nine eligible clinic physicians and all 23 clinic staff participated. All 105 patients who enrolled completed the preference program. Providers did not know which of their patients were enrolled, nor their preferences for accessing their EHRs. During the 6-month prospective study, 92 study patients (88 %) returned 261 times, during which providers viewed their EHRs 126 times (48 %). Providers "broke the glass" 102 times, 92 times for patients not in the study and ten times for six returning study patients, all of whom had restricted EHR access. Providers "broke the glass" for six (14 %) of 43 returning study patients with redacted data vs. zero among 49 study patients without redactions (p=0.01). Although 54 % of providers agreed that patients should have control over who sees their EHR information, 58 % believed restricting EHR access could harm provider-patient relationships and 71 % felt quality of care would suffer. CONCLUSIONS: Patients frequently preferred restricting provider access to their EHRs. Providers infrequently overrode patients' preferences to view hidden data. Providers believed that restricting EHR access would adversely impact patient care. Applying Fair Information Practice principles to EHRs will require balancing patient preferences, providers' needs, and health care quality.KEY WORDS: fair information practices; electronic health records; patient preferences; quality of care.

Objective. Hypoglycemia occurs in 20% to 60% of patients with diabetes mellitus. Identifying at-risk patients can facilitate interventions to lower risk. We sought to develop a hypoglycemia prediction model. Methods. In this retrospective cohort study, urban adults prescribed a diabetes drug between 2004 and 2013 were identified. Demographic and clinical data were extracted from an electronic medical record (EMR). Laboratory tests, diagnostic codes, and natural language processing (NLP) identified hypoglycemia. We compared multiple logistic regression, classification and regression trees (CART), and Random Forest. Models were evaluated on an independent test set or through cross-validation. Results. 38,780 patients had mean age 57 years; 56% were female, 40% African-American, and 39% uninsured. Hypoglycemia occurred in 8,128 (539 identified only by NLP). In logistic regression, factors positively associated with hypoglycemia included infection, non-long-acting insulin, dementia, and recent hypoglycemia. Negatively associated factors included long-acting insulin plus sulfonylurea, and age 75 or older. Models' area under curve was similar (logistic regression, 89%; CART, 88%; Random Forest, 90%, with 10-fold cross-validation). Conclusions. NLP improved identification of hypoglycemia. Non-long-acting insulin was an important risk factor. Decreased risk with age may reflect treatment or diminished awareness of HG. More complex models did not improve prediction.

No abstract