Disassociation for electronic health record privacy

The dissemination of Electronic Health Record (EHR) data, beyond the originating healthcare institutions, can enable large-scale, low-cost medical studies that have the potential to improve public health. Thus, funding bodies, such as the National Institutes of Health (NIH) in the U.S., encourage or require the dissemination of EHR data, and a growing number of innovative medical investigations are being performed using such data. However, simply disseminating EHR data, after removing identifying information, may risk privacy, as patients can still be linked with their record, based on diagnosis codes. This paper proposes the first approach that prevents this type of data linkage using disassociation, an operation that transforms records by splitting them into carefully selected subsets. Our approach preserves privacy with significantly lower data utility loss than existing methods and does not require data owners to specify diagnosis codes that may lead to identity disclosure, as these methods do. Consequently, it can be employed when data need to be shared broadly and be used in studies, beyond the intended ones. Through extensive experiments using EHR data, we demonstrate that our method can construct data that are highly useful for supporting various types of clinical case count studies and general medical analysis tasks.     

Publishing data from electronic health records while preserving privacy: A survey of algorithms

The dissemination of Electronic Health Records (EHRs) can be highly beneficial for a range of medical studies, spanning from clinical trials to epidemic control studies, but it must be performed in a way that preserves patients’ privacy. This is not straightforward, because the disseminated data need to be protected against several privacy threats, while remaining useful for subsequent analysis tasks. In this work, we present a survey of algorithms that have been proposed for publishing structured patient data, in a privacy-preserving way. We review more than 45 algorithms, derive insights on their operation, and highlight their advantages and disadvantages. We also provide a discussion of some promising directions for future research in this area.     

Strategies for handling missing clinical data for automated surgical site infection detection from the electronic health record

Proper handling of missing data is important for many secondary uses of electronic health record (EHR) data. Data imputation methods can be used to handle missing data, but their use for analyzing EHR data is limited and specific efficacy for postoperative complication detection is unclear. Several data imputation methods were used to develop data models for automated detection of three types (i.e., superficial, deep, and organ space) of surgical site infection (SSI) and overall SSI using American College of Surgeons National Surgical Quality Improvement Project (NSQIP) Registry 30-day SSI occurrence data as a reference standard. Overall, models with missing data imputation almost always outperformed reference models without imputation that included only cases with complete data for detection of SSI overall achieving very good average area under the curve values. Missing data imputation appears to be an effective means for improving postoperative SSI detection using EHR clinical data.     

Learning relational policies from electronic health record access logs

Modern healthcare organizations (HCOs) are composed of complex dynamic teams to ensure clinical operations are executed in a quick and competent manner. At the same time, the fluid nature of such environments hinders administrators' efforts to define access control policies that appropriately balance patient privacy and healthcare functions. Manual efforts to define these policies are labor-intensive and error-prone, often resulting in systems that endow certain care providers with overly broad access to patients' medical records while restricting other providers from legitimate and timely use. In this work, we propose an alternative method to generate these policies by automatically mining usage patterns from electronic health record (EHR) systems. EHR systems are increasingly being integrated into clinical environments and our approach is designed to be generalizable across HCOs, thus assisting in the design and evaluation of local access control policies. Our technique, which is grounded in data mining and social network analysis theory, extracts a statistical model of the organization from the access logs of its EHRs. In doing so, our approach enables the review of predefined policies, as well as the discovery of unknown behaviors. We evaluate our approach with 5 months of access logs from the Vanderbilt University Medical Center and confirm the existence of stable social structures and intuitive business operations. Additionally, we demonstrate that there is significant turnover in the interactions between users in the HCO and that policies learned at the department-level afford greater stability over time.     

Defining and measuring completeness of electronic health records for secondary use

We demonstrate the importance of explicit definitions of electronic health record (EHR) data completeness and how different conceptualizations of completeness may impact findings from EHR-derived datasets. This study has important repercussions for researchers and clinicians engaged in the secondary use of EHR data. We describe four prototypical definitions of EHR completeness: documentation, breadth, density, and predictive completeness. Each definition dictates a different approach to the measurement of completeness. These measures were applied to representative data from NewYork–Presbyterian Hospital’s clinical data warehouse. We found that according to any definition, the number of complete records in our clinical database is far lower than the nominal total. The proportion that meets criteria for completeness is heavily dependent on the definition of completeness used, and the different definitions generate different subsets of records. We conclude that the concept of completeness in EHR is contextual. We urge data consumers to be explicit in how they define a complete record and transparent about the limitations of their data.     

Contingency planning for electronic health record-based care continuity: A survey of recommended practices

BackgroundReliable health information technology (HIT) in general, and electronic health record systems (EHRs) in particular are essential to a high-performing healthcare system. When the availability of EHRs are disrupted, alternative methods must be used to maintain the continuity of healthcare.MethodsWe developed a survey to assess institutional practices to handle situations when EHRs were unavailable for use (downtime preparedness). We used literature reviews and expert opinion to develop items that assessed the implementation of potentially useful practices. We administered the survey to U.S.-based healthcare institutions that were members of a professional organization that focused on collaboration and sharing of HIT-related best practices among its members. All members were large integrated health systems.ResultsWe received responses from 50 of the 59 (84%) member institutions. Nearly all (96%) institutions reported at least one unplanned downtime (of any length) in the last 3 years and 70% had at least one unplanned downtime greater than 8 h in the last 3 years. Three institutions reported that one or more patients were injured as a result of either a planned or unplanned downtime. The majority of institutions (70–85%) had implemented a portion of the useful practices we identified, but very few practices were followed by all organizations.ConclusionsUnexpected downtimes related to EHRs appear to be fairly common among institutions in our survey. Most institutions had only partially implemented comprehensive contingency plans to maintain safe and effective healthcare during unexpected EHRs downtimes.     

Impact of integrating genomic data into the electronic health record on genetics care delivery

PurposeIntegrating genomic data into the electronic health record (EHR) is key for optimally delivering genomic medicine.MethodsThe PennChart Genomics Initiative (PGI) at the University of Pennsylvania is a multidisciplinary collaborative that has successfully linked orders and results from genetic testing laboratories with discrete genetic data in the EHR. We quantified the use of the genomic data within the EHR, performed a time study with genetic counselors, and conducted key informant interviews with PGI members to evaluate the effect of the PGI’s efforts on genetics care delivery.ResultsThe PGI has interfaced with 4 genetic testing laboratories, resulting in the creation of 420 unique computerized genetic testing orders that have been used 4073 times to date. In a time study of 96 genetic testing activities, EHR use was associated with significant reductions in time spent ordering (2 vs 8 minutes, P < .001) and managing (1 vs 5 minutes, P < .001) genetic results compared with the use of online laboratory-specific portals. In key informant interviews, multidisciplinary collaboration and institutional buy-in were identified as key ingredients for the PGI’s success.ConclusionThe PGI’s efforts to integrate genomic medicine into the EHR have substantially streamlined the delivery of genomic medicine.     

Learning from heterogeneous temporal data in electronic health records

Electronic health records contain large amounts of longitudinal data that are valuable for biomedical informatics research. The application of machine learning is a promising alternative to manual analysis of such data. However, the complex structure of the data, which includes clinical events that are unevenly distributed over time, poses a challenge for standard learning algorithms. Some approaches to modeling temporal data rely on extracting single values from time series; however, this leads to the loss of potentially valuable sequential information. How to better account for the temporality of clinical data, hence, remains an important research question. In this study, novel representations of temporal data in electronic health records are explored. These representations retain the sequential information, and are directly compatible with standard machine learning algorithms. The explored methods are based on symbolic sequence representations of time series data, which are utilized in a number of different ways. An empirical investigation, using 19 datasets comprising clinical measurements observed over time from a real database of electronic health records, shows that using a distance measure to random subsequences leads to substantial improvements in predictive performance compared to using the original sequences or clustering the sequences. Evidence is moreover provided on the quality of the symbolic sequence representation by comparing it to sequences that are generated using domain knowledge by clinical experts. The proposed method creates representations that better account for the temporality of clinical events, which is often key to prediction tasks in the biomedical domain.     

The Australian general public's perceptions of having a personally controlled electronic health record (PCEHR)

ObjectiveThe move internationally by Governments and other health providers to encourage patients to have their own electronic personal health record (e-PHRs) is growing exponentially. In Australia the initiative for a personally controlled electronic health record (known as PCEHR) is directed towards the public at large. The first objective of this study then, is to examine how individuals in the general population perceive the promoted idea of having a PCEHR. The second objective is to extend research on applying a theoretically derived consumer technology acceptance model to guide the research.MethodAn online survey was conducted to capture the perceptions and beliefs about having a PCEHR identified from technology acceptance models and extant literature. The survey was completed by 750 Queensland respondents, 97% of whom did not have a PCEHR at that time. The model was examined using exploratory factor analysis, regressions and mediation tests.ResultsFindings support eight of the 11 hypothesised relationships in the model. Perceived value and perceived risk were the two most important variables explaining attitude, with perceived usefulness and compatibility being weak but significant. The perception of risk was reduced through partial mediation from trust and privacy concerns. Additionally, web-self efficacy and ease of use partially mediate the relationship between attitude and intentions.ConclusionsThe findings represent a snapshot of the early stages of implementing this Australian initiative and capture the perceptions of Queenslanders who at present do not have a PCEHR. Findings show that while individuals appreciate the value of having this record, they do not appear to regard it as particularly useful at present, nor is it particularly compatible with their current engagement with e-services. Moreover, they will need to have any concerns about the risks alleviated, particularly through an increased sense of trust and reduction of privacy concerns. It is noted that although the respondents are non-adopters, they do not feel that they lack the necessary web skills to set up and use a PCEHR. To the best of our knowledge this is one of a very limited number of studies that examines a national level implementation of an e-PHR system, where take-up of the PCEHR is optional rather than a centralised, mandated requirement.     

Rates,levels, and determinants of electronic health record system adoption: A study of hospitals in Riyadh,Saudi Arabia

ObjectiveOutside a small number of OECD countries, little information exists regarding the rates, levels, and determinants of hospital electronic health record (EHR) system adoption. This study examines EHR system adoption in Riyadh, Saudi Arabia.Materials and methodsRespondents from 22 hospitals were surveyed regarding the implementation, maintenance, and improvement phases of EHR system adoption. Thirty-seven items were graded on a three-point scale of preparedness/completion. Measured determinants included hospital size, level of care, ownership, and EHR system development team composition.ResultsEleven of the hospitals had implemented fully functioning EHR systems, eight had systems in progress, and three had not adopted a system. Sixteen different systems were being used across the 19 adopting hospitals. Differential adoption levels were positively related to hospital size and negatively to the level of care (secondary versus tertiary). Hospital ownership (nonprofit versus private) and development team composition showed mixed effects depending on the particular adoption phase being considered.DiscussionAdoption rates compare favourably with those reported from other countries and other districts in Saudi Arabia, but wide variations exist among hospitals in the levels of adoption of individual items. General weaknesses in the implementation phase concern the legacy of paper data systems, including document scanning and data conversion; in the maintenance phase concern updating/maintaining software; and in the improvement phase concern the communication and exchange of health information.ConclusionThis study is the first to investigate the level and determinants of EHR system adoption for public, other nonprofit, and private hospitals in Saudi Arabia. Wide interhospital variations in adoption bear implications for policy-making and funding intervention. Identified areas of weakness require action to increase the degree of adoption and usefulness of EHR systems.     

Sleep health,diseases, and pain syndromes: findings from an electronic health record biobank

Study ObjectivesImplementation of electronic health record biobanks has facilitated linkage between clinical and questionnaire data and enabled assessments of relationships between sleep health and diseases in phenome-wide association studies (PheWAS). In the Mass General Brigham Biobank, a large health system-based study, we aimed to systematically catalog associations between time in bed, sleep timing, and weekly variability with clinical phenotypes derived from ICD-9/10 codes.MethodsSelf-reported habitual bed and wake times were used to derive variables: short (<7 hours) and long (≥9 hours) time in bed, sleep midpoint, social jetlag, and sleep debt. Logistic regression and Cox proportional hazards models were used to test cross-sectional and prospective associations, respectively, adjusted for age, gender, race/ethnicity, and employment status and further adjusted for body mass index.ResultsIn cross-sectional analysis (n = 34,651), sleep variable associations were most notable for circulatory system, mental disorders, and endocrine/metabolic phenotypes. We observed the strongest associations for short time in bed with obesity, for long time in bed and sleep midpoint with major depressive disorder, for social jetlag with hypercholesterolemia, and for sleep debt with acne. In prospective analysis (n = 24,065), we observed short time in bed associations with higher incidence of acute pain and later sleep midpoint and higher sleep debt and social jetlag associations with higher incidence of major depressive disorder.ConclusionsOur analysis reinforced that sleep health is a multidimensional construct, corroborated robust known findings from traditional cohort studies, and supported the application of PheWAS as a promising tool for advancing sleep research. Considering the exploratory nature of PheWAS, careful interrogation of novel findings is imperative.     

Semi-supervised learning of the electronic health record for phenotype stratification

Patient interactions with health care providers result in entries to electronic health records (EHRs). EHRs were built for clinical and billing purposes but contain many data points about an individual. Mining these records provides opportunities to extract electronic phenotypes, which can be paired with genetic data to identify genes underlying common human diseases. This task remains challenging: high quality phenotyping is costly and requires physician review; many fields in the records are sparsely filled; and our definitions of diseases are continuing to improve over time. Here we develop and evaluate a semi-supervised learning method for EHR phenotype extraction using denoising autoencoders for phenotype stratification. By combining denoising autoencoders with random forests we find classification improvements across multiple simulation models and improved survival prediction in ALS clinical trial data. This is particularly evident in cases where only a small number of patients have high quality phenotypes, a common scenario in EHR-based research. Denoising autoencoders perform dimensionality reduction enabling visualization and clustering for the discovery of new subtypes of disease. This method represents a promising approach to clarify disease subtypes and improve genotype-phenotype association studies that leverage EHRs.     

Transitioning from a computerized provider order entry and paper documentation system to an electronic health record: Expectations and experiences of hospital staff

ObjectivesTo examine healthcare worker's perceptions, expectations, and experiences regarding how work processes, patient-related safety, and care were affected when a quaternary care center transitioned from one computerized provider order entry (CPOE) system to a full electronic health record (EHR).MethodsThe I-SEE survey was administered prior to and 1-year after transition in systems. The construct validity and reliability of the survey was assessed within the current population and also compared to previously published results. Pre- and 1-year post-implementation scale means were compared within and across time periods.ResultsThe majority of respondents were nurses and personnel working in the acute care setting. Because a confirmatory factor analysis indicated a lack of fit of our data to the I-SEE survey's 5-factor structure, we conducted an exploratory factor analysis that resulted in a 7-factor structure which showed better reliability and validity. Mean scores for each factor indicated that attitudes and expectations were mostly positive and score trends over time were positive or neutral. Nurses generally had less positive attitudes about the transition than non-nursing respondents, although the difference diminished after implementation.ConclusionsFindings demonstrate that the majority of responding staff were generally positive about transitioning from CPOE system to a full electronic health record (EHR) and understood the goals of doing so, with overall improved ratings over time. In addition, the I-SEE survey, when modified based on our population, was useful for assessing patient care and safety related expectations and experiences during the transition from one CPOE system to an EHR.     

Provider interaction with the electronic health record: The effects on patient-centered communication in medical encounters

Objective

The computer with the electronic health record (EHR) is an additional ‘interactant’ in the medical consultation, as clinicians must simultaneously or in alternation engage patient and computer to provide medical care. Few studies have examined how clinicians’ EHR workflow (e.g., gaze, keyboard activity, and silence) influences the quality of their communication, the patient's involvement in the encounter, and conversational control of the visit.

Methods

Twenty-three primary care providers (PCPs) from USA Veterans Administration (VA) primary care clinics participated in the study. Up to 6 patients per PCP were recruited. The proportion of time PCPs spent gazing at the computer was captured in real time via video-recording. Mouse click/scrolling activity was captured through Morae, a usability software that logs mouse clicks and scrolling activity. Conversational silence was coded as the proportion of time in the visit when PCP and patient were not talking. After the visit, patients completed patient satisfaction measures. Trained coders independently viewed videos of the interactions and rated the degree to which PCPs were patient-centered (informative, supportive, partnering) and patients were involved in the consultation. Conversational control was measured as the proportion of time the PCP held the floor compared to the patient.

Results

The final sample included 125 consultations. PCPs who spent more time in the consultation gazing at the computer and whose visits had more conversational silence were rated lower in patient-centeredness. PCPs controlled more of the talk time in the visits that also had longer periods of mutual silence.

Conclusions

PCPs were rated as having less effective communication when they spent more time looking at the computer and when there was more periods of silence in the consultation. Because PCPs increasingly are using the EHR in their consultations, more research is needed to determine effective ways that they can verbally engage patients while simultaneously managing data in the EHR.

Practice implications

EHR activity consumes an increasing proportion of clinicians’ time during consultations. To ensure effective communication with their patients, clinicians may benefit from using communication strategies that maintain the flow of conversation when working with the computer, as well as from learning EHR management skills that prevent extended periods of gaze at computer and long periods of silence. Next-generation EHR design must address better usability and clinical workflow integration, including facilitating patient-clinician communication.     

Advancing cognitive engineering methods to support user interface design for electronic health records

BackgroundDespite many decades of research on the effective development of clinical systems in medicine, the adoption of health information technology to improve patient care continues to be slow, especially in ambulatory settings. This applies to dentistry as well, a primary care discipline with approximately 137,000 practitioners in the United States. A critical reason for slow adoption is the poor usability of clinical systems, which makes it difficult for providers to navigate through the information and obtain an integrated view of patient data.ObjectiveIn this study, we documented the cognitive processes and information management strategies used by dentists during a typical patient examination. The results will inform the design of a novel electronic dental record interface.MethodsWe conducted a cognitive task analysis (CTA) study to observe ten general dentists (five general dentists and five general dental faculty members, each with more than two years of clinical experience) examining three simulated patient cases using a think-aloud protocol.ResultsDentists first reviewed the patient's demographics, chief complaint, medical history and dental history to determine the general status of the patient. Subsequently, they proceeded to examine the patient's intraoral status using radiographs, intraoral images, hard tissue and periodontal tissue information. The results also identified dentists’ patterns of navigation through patient's information and additional information needs during a typical clinician–patient encounter.ConclusionThis study reinforced the significance of applying cognitive engineering methods to inform the design of a clinical system. Second, applying CTA to a scenario closely simulating an actual patient encounter helped with capturing participants’ knowledge states and decision-making when diagnosing and treating a patient. The resultant knowledge of dentists’ patterns of information retrieval and review will significantly contribute to designing flexible and task-appropriate information presentation in electronic dental records.     

Giving patients granular control of personal health information: Using an ethics ‘Points to Consider’ to inform informatics system designers

ObjectiveThere are benefits and risks of giving patients more granular control of their personal health information in electronic health record (EHR) systems. When designing EHR systems and policies, informaticists and system developers must balance these benefits and risks. Ethical considerations should be an explicit part of this balancing. Our objective was to develop a structured ethics framework to accomplish this.MethodsWe reviewed existing literature on the ethical and policy issues, developed an ethics framework called a “Points to Consider” (P2C) document, and convened a national expert panel to review and critique the P2C.ResultsWe developed the P2C to aid informaticists designing an advanced query tool for an electronic health record (EHR) system in Indianapolis. The P2C consists of six questions (“Points”) that frame important ethical issues, apply accepted principles of bioethics and Fair Information Practices, comment on how questions might be answered, and address implications for patient care.DiscussionThe P2C is intended to clarify what is at stake when designers try to accommodate potentially competing ethical commitments and logistical realities. The P2C was developed to guide informaticists who were designing a query tool in an existing EHR that would permit patient granular control. While consideration of ethical issues is coming to the forefront of medical informatics design and development practices, more reflection is needed to facilitate optimal collaboration between designers and ethicists. This report contributes to that discussion.     

Exploring the persistence of paper with the electronic health record

Objective

Healthcare organizations are increasingly implementing electronic health records (EHRs) and other related health information technology (IT). Even in institutions which have long adopted these computerized systems, employees continue to rely on paper to complete their work. The objective of this study was to explore and understand human-technology integration factors that may be causing employees to rely on paper alternatives to the EHR.

Methods

We conducted semi-structured interviews with 20 key-informants in a large Veterans Affairs Medical Center (VAMC), with a fully implemented EHR, to understand the use of paper-based alternatives. Participants included clinicians, administrators, and IT specialists across several service areas in the medical center.

Results

We found 11 distinct categories of paper-based workarounds to the use of the EHR. Paper use related to the following: (1) efficiency; (2) knowledge/skill/ease of use; (3) memory; (4) sensorimotor preferences; (5) awareness; (6) task specificity; (7) task complexity; (8) data organization; (9) longitudinal data processes; (10) trust; and (11) security. We define each of these and provide examples that demonstrate how these categories promoted paper use in spite of a fully implemented EHR.

Conclusions

In several cases, paper served as an important tool and assisted healthcare employees in their work. In other cases, paper use circumvented the intended EHR design, introduced potential gaps in documentation, and generated possible paths to medical error. We discuss implications of these findings for EHR design and implementation.