A multi-site randomized trial of a clinical decision support intervention to improve problem list completeness

OBJECTIVE

To improve problem list documentation and care quality.

MATERIALS AND METHODS

We developed algorithms to infer clinical problems a patient has that are not recorded on the coded problem list using structured data in the electronic health record (EHR) for 12 clinically significant heart, lung, and blood diseases. We also developed a clinical decision support (CDS) intervention which suggests adding missing problems to the problem list. We evaluated the intervention at 4 diverse healthcare systems using 3 different EHRs in a randomized trial using 3 predetermined outcome measures: alert acceptance, problem addition, and National Committee for Quality Assurance Healthcare Effectiveness Data and Information Set (NCQA HEDIS) clinical quality measures.

RESULTS

There were 288 832 opportunities to add a problem in the intervention arm and the problem was added 63 777 times (acceptance rate 22.1%). The intervention arm had 4.6 times as many problems added as the control arm. There were no significant differences in any of the clinical quality measures.

DISCUSSION

The CDS intervention was highly effective at improving problem list completeness. However, the improvement in problem list utilization was not associated with improvement in the quality measures. The lack of effect on quality measures suggests that problem list documentation is not directly associated with improvements in quality measured by National Committee for Quality Assurance Healthcare Effectiveness Data and Information Set (NCQA HEDIS) quality measures. However, improved problem list accuracy has other benefits, including clinical care, patient comprehension of health conditions, accurate CDS and population health, and for research.

CONCLUSION

An EHR-embedded CDS intervention was effective at improving problem list completeness but was not associated with improvement in quality measures.

Trends in buprenorphine treatment disparities during the COVID pandemic in Massachusetts

Background: Racial, sex, and age disparities in buprenorphine treatment have previously been demonstrated. We evaluated trends in buprenorphine treatment disparities before and after the onset of the COVID pandemic in Massachusetts. Methods: This cross-sectional study used data from an integrated health system comparing 12-months before and after the March 2020 Massachusetts COVID state of emergency declaration, excluding March as a washout period. Among patients with a clinical encounter during the study periods with a diagnosis of opioid use disorder or opioid poisoning, we extracted outpatient buprenorphine prescription rates by age, sex, race and ethnicity, and language. Generating univariable and multivariable Poisson regression models, we calculated the probability of receiving buprenorphine. Results: Among 4,530 patients seen in the period before the COVID emergency declaration, 57.9% received buprenorphine. Among 3,653 patients seen in the second time period, 55.1% received buprenorphine. Younger patients (<24) had a lower likelihood of receiving buprenorphine in both time periods (adjusted prevalence ratio (aPR), 0.56; 95% CI, 0.42-0.75 before vs. aPR, 0.76; 95% CI, 0.60-0.96 after). Male patients had a greater likelihood of receiving buprenorphine compared to female patients in both time periods (aPR: 1.05; 95% CI, 1.00-1.11 vs. aPR: 1.09; 95% CI, 1.02-1.16). Racial disparities emerged in the time period following the COVID pandemic, with non-Hispanic Black patients having a lower likelihood of receiving buprenorphine compared to non-Hispanic white patients in the second time period (aPR, 0.85; 95% CI, 0.72-0.99). Conclusions: Following the onset of the COVID pandemic in Massachusetts, ongoing racial, age, and gender disparities were evident in buprenorphine treatment with younger, Black, and female patients less likely to be treated with buprenorphine across an integrated health system.

Using statistical anomaly detection models to find clinical decision support malfunctions

Objective

Malfunctions in Clinical Decision Support (CDS) systems occur due to a multitude of reasons, and often go unnoticed, leading to potentially poor outcomes. Our goal was to identify malfunctions within CDS systems.

Methods

We evaluated 6 anomaly detection models: (1) Poisson Changepoint Model, (2) Autoregressive Integrated Moving Average (ARIMA) Model, (3) Hierarchical Divisive Changepoint (HDC) Model, (4) Bayesian Changepoint Model, (5) Seasonal Hybrid Extreme Studentized Deviate (SHESD) Model, and (6) E-Divisive with Median (EDM) Model and characterized their ability to find known anomalies. We analyzed 4 CDS alerts with known malfunctions from the Longitudinal Medical Record (LMR) and Epic® (Epic Systems Corporation, Madison, WI, USA) at Brigham and Women's Hospital, Boston, MA. The 4 rules recommend lead testing in children, aspirin therapy in patients with coronary artery disease, pneumococcal vaccination in immunocompromised adults and thyroid testing in patients taking amiodarone.

Results

Poisson changepoint, ARIMA, HDC, Bayesian changepoint and the SHESD model were able to detect anomalies in an alert for lead screening in children and in an alert for pneumococcal conjugate vaccine in immunocompromised adults. EDM was able to detect anomalies in an alert for monitoring thyroid function in patients on amiodarone.

Conclusions

Malfunctions/anomalies occur frequently in CDS alert systems. It is important to be able to detect such anomalies promptly. Anomaly detection models are useful tools to aid such detections.

Variation in high-priority drug-drug interaction alerts across institutions and electronic health records

Objective: The United States Office of the National Coordinator for Health Information Technology sponsored the development of a “high-priority” list of drug-drug interactions (DDIs) to be used for clinical decision support. We assessed current adoption of this list and current alerting practice for these DDIs with regard to alert implementation (presence or absence of an alert) and display (alert appearance as interruptive or passive).

Materials and methods: We conducted evaluations of electronic health records (EHRs) at a convenience sample of health care organizations across the United States using a standardized testing protocol with simulated orders.

Results: Evaluations of 19 systems were conducted at 13 sites using 14 different EHRs. Across systems, 69% of the high-priority DDI pairs produced alerts. Implementation and display of the DDI alerts tested varied between systems, even when the same EHR vendor was used. Across the drug pairs evaluated, implementation and display of DDI alerts differed, ranging from 27% (4/15) to 93% (14/15) implementation.

Discussion: Currently, there is no standard of care covering which DDI alerts to implement or how to display them to providers. Opportunities to improve DDI alerting include using differential displays based on DDI severity, establishing improved lists of clinically significant DDIs, and thoroughly reviewing organizational implementation decisions regarding DDIs.

Conclusion: DDI alerting is clinically important but not standardized. There is significant room for improvement and standardization around evidence-based DDIs.

A Picture is Worth 1,000 Words. The Use of Clinical Images in Electronic Medical Records

OBJECTIVE

To understand how clinicians utilize image uploading tools in a home grown electronic health records (EHR) system.

METHODS

A content analysis of patient notes containing non-radiological images from the EHR was conducted. Images from 4,000 random notes from July 1, 2009 - June 30, 2010 were reviewed and manually coded. Codes were assigned to four properties of the image: (1) image type, (2) role of image uploader (e.g. MD, NP, PA, RN), (3) practice type (e.g. internal medicine, dermatology, ophthalmology), and (4) image subject.

RESULTS

3,815 images from image-containing notes stored in the EHR were reviewed and manually coded. Of those images, 32.8% were clinical and 66.2% were non-clinical. The most common types of the clinical images were photographs (38.0%), diagrams (19.1%), and scanned documents (14.4%). MDs uploaded 67.9% of clinical images, followed by RNs with 10.2%, and genetic counselors with 6.8%. Dermatology (34.9%), ophthalmology (16.1%), and general surgery (10.8%) uploaded the most clinical images. The content of clinical images referencing body parts varied, with 49.8% of those images focusing on the head and neck region, 15.3% focusing on the thorax, and 13.8% focusing on the lower extremities.

CONCLUSION

The diversity of image types, content, and uploaders within a home grown EHR system reflected the versatility and importance of the image uploading tool. Understanding how users utilize image uploading tools in a clinical setting highlights important considerations for designing better EHR tools and the importance of interoperability between EHR systems and other health technology.

Association between shunt-responsive idiopathic normal pressure hydrocephalus and alcohol

OBJECTIVE Idiopathic normal pressure hydrocephalus (iNPH) is characterized by ventriculomegaly, gait difficulty, incontinence, and dementia. The symptoms can be ameliorated by CSF drainage. The object of this study was to identify factors associated with shunt-responsive iNPH. METHODS The authors reviewed the medical records of 529 patients who underwent shunt placement for iNPH at their institution between July 2001 and March 2015. Variables associated with shunt-responsive iNPH were identified using bivariate and multivariate analyses. Detailed alcohol consumption information was obtained for 328 patients and was used to examine the relationship between alcohol and shunt-responsive iNPH. A computerized patient registry from 2 academic medical centers was queried to determine the prevalence of alcohol abuse among 1665 iNPH patients. RESULTS Bivariate analysis identified associations between shunt-responsive iNPH and gait difficulty (OR 4.59, 95% CI 2.32-9.09; p < 0.0001), dementia (OR 1.79, 95% CI 1.14-2.80; p = 0.01), incontinence (OR 1.77, 95% CI 1.13-2.76; p = 0.01), and alcohol use (OR 1.98, 95% CI 1.23-3.16; p = 0.03). Borderline significance was observed for hyperlipidemia (OR 1.56, 95% CI 0.99-2.45; p = 0.054), a family history of hyperlipidemia (OR 3.09, 95% CI 0.93-10.26, p = 0.054), and diabetes (OR 1.83, 95% CI 0.96-3.51; p = 0.064). Multivariate analysis identified associations with gait difficulty (OR 3.98, 95% CI 1.81-8.77; p = 0.0006) and alcohol (OR 1.94, 95% CI 1.10-3.39; p = 0.04). Increased alcohol intake correlated with greater improvement after CSF drainage. Alcohol abuse was 2.5 times more prevalent among iNPH patients than matched controls. CONCLUSIONS Alcohol consumption is associated with the development of shunt-responsive iNPH.

Quantitative evaluation of changes in gait after extended cerebrospinal fluid drainage for normal pressure hydrocephalus

Idiopathic normal pressure hydrocephalus (iNPH) is characterized by gait instability, urinary incontinence and cognitive dysfunction. These symptoms can be relieved by cerebrospinal fluid (CSF) drainage, but the time course and nature of the improvements are poorly characterized. Attempts to prospectively identify iNPH patients responsive to CSF drainage by evaluating presenting gait quality or via extended lumbar cerebrospinal fluid drainage (eLCD) trials are common, but the reliability of such approaches is unclear. Here we combine eLCD trials with computerized quantitative gait measurements to predict shunt responsiveness in patients undergoing evaluation for possible iNPH. In this prospective cohort study, 50 patients presenting with enlarged cerebral ventricles and gait, urinary, and/or cognitive difficulties were evaluated for iNPH using a computerized gait analysis system during a 3day trial of eLCD. Gait speed, stride length, cadence, and the Timed Up and Go test were quantified before and during eLCD. Qualitative assessments of incontinence and cognition were obtained throughout the eLCD trial. Patients who improved after eLCD underwent ventriculoperitoneal shunt placement, and symptoms were reassessed serially over the next 3 to 15months. There was no significant difference in presenting gait characteristics between patients who improved after drainage and those who did not. Gait improvement was not observed until 2 or more days of continuous drainage in most cases. Symptoms improved after eLCD in 60% of patients, and all patients who improved after eLCD also improved after shunt placement. The degree of improvement after eLCD correlated closely with that observed after shunt placement.