1. Home
  2. Publications

Publications

S

Weitzman, Adida, Kelemen, Mandl. Sharing data for public health research by members of an international online diabetes social network. PLoS One. 2011;6:e19256.
BACKGROUND: Surveillance and response to diabetes may be accelerated through engaging online diabetes social networks (SNs) in consented research. We tested the willingness of an online diabetes community to share data for public health research by providing members with a privacy-preserving social networking software application for rapid temporal-geographic surveillance of glycemic control. METHODS AND FINDINGS: SN-mediated collection of cross-sectional, member-reported data from an international online diabetes SN entered into a software application we made available in a "Facebook-like" environment to enable reporting, charting and optional sharing of recent hemoglobin A1c values through a geographic display. Self-enrollment by 17% (n = 1,136) of n = 6,500 active members representing 32 countries and 50 US states. Data were current with 83.1% of most recent A1c values reported obtained within the past 90 days. Sharing was high with 81.4% of users permitting data donation to the community display. 34.1% of users also displayed their A1cs on their SN profile page. Users selecting the most permissive sharing options had a lower average A1c (6.8%) than users not sharing with the community (7.1%, p = .038). 95% of users permitted re-contact. Unadjusted aggregate A1c reported by US users closely resembled aggregate 2007-2008 NHANES estimates (respectively, 6.9% and 6.9%, p = 0.85). CONCLUSIONS: Success within an early adopter community demonstrates that online SNs may comprise efficient platforms for bidirectional communication with and data acquisition from disease populations. Advancing this model for cohort and translational science and for use as a complementary surveillance approach will require understanding of inherent selection and publication (sharing) biases in the data and a technology model that supports autonomy, anonymity and privacy.
Nakamura M, Simons, Samuels, Daniel, Mandl. Service-oriented architecture for pediatric immunization decision support. AMIA Annu Symp Proc. 2007:1056.
We integrate a personally-controlled health record (PCHR) with a Department of Public Health (DPH) immunization registry and clinical decision support (CDS) tool, creating an immunization information system using a service-oriented architecture (SOA). This SOA enables the DPH, a trusted authority, to provide CDS to both clinicians and patients/parents.
Boutis, Komar, Jaramillo, Babyn, Alman, Snyder, Mandl, Schuh. Sensitivity of a clinical examination to predict need for radiography in children with ankle injuries: a prospective study. Lancet. 2001;358:2118–21.
BACKGROUND: Radiographs are ordered routinely for children with ankle trauma. We assessed the predictive value of a clinical examination to identify a predefined group of low-risk injuries, management of which would not be affected by absence of a radiograph. We aimed to show that no more than 1% of children with low-risk examinations (signs restricted to the distal fibula) would have high-risk fractures (all fractures except avulsion, buckle, and non-displaced Salter-Harris I and II fractures of the distal fibula), and to compare the potential reduction in radiography in children with low-risk examinations with that obtained by application of the Ottawa ankle rules (OAR). METHODS: Standard clinical examinations and subsequent radiographs were prospectively and independently evaluated in two tertiary-care paediatric emergency departments in North America. Eligible participants were healthy children aged 3-16 years with acute ankle injuries. Sample size, negative and positive predictive values, sensitivity, and specificity were calculated. McNemar's test was used to compare differences in the potential reduction in radiographs between the low-risk examination and the OAR. FINDINGS: 607 children were enrolled; 581 (95.7%) received follow-up. None of the 381 children with low-risk examinations had a high-risk fracture (negative predictive value 100% [95% CI 99.2-100]; sensitivity 100% [93.3-100]). Radiographs could be omitted in 62.8% of children with low-risk examinations, compared with only 12.0% reduction obtained by application of the OAR (p0.0001). INTERPRETATION: A low-risk clinical examination in children with ankle injuries identifies 100% of high-risk diagnoses and may result in greater reduction of radiographic referrals than the OAR.
Garcia Pena, Cook, Mandl. Selective imaging strategies for the diagnosis of appendicitis in children. Pediatrics. 2004;113:24–8.
BACKGROUND: We previously reported an appendiceal imaging protocol in which children with equivocal clinical presentations for acute appendicitis undergo ultrasonography (US) followed by computed tomography (CT). However, risk groups of children who would benefit most from imaging studies have not been established. OBJECTIVE: To define and test selective imaging guidelines to increase diagnostic accuracy and reduce unnecessary testing for children with suspected appendicitis. METHODS: We modeled outcomes under 3 different management guidelines. Patients were risk-stratified by a recursive partitioning analysis of a retrospective cohort. Subjects included children with equivocal presentations of acute appendicitis evaluated between January 1996 and December 1999. By using recursive partitioning, 3 risk groups were identified: low, medium, and high risk for acute appendicitis. Three imaging guidelines were defined. Under the first guideline, representing standard clinical practice at Children's Hospital Boston at the time of the study, all children with equivocal signs and symptoms for acute appendicitis undergo US first. If the US is positive, the child proceeds to appendectomy. If the US is negative, the child undergoes CT. Under guideline 2, low-risk children undergo US and, if negative, are discharged from the hospital. High-risk children undergo CT, and medium-risk children undergo US followed by CT. Under the third guideline, low-risk children undergo no imaging and are admitted for observation. High-risk children proceed directly to appendectomy without imaging studies. Medium-risk children undergo US followed by CT. Clinical outcomes and the number of imaging studies performed were modeled under current practice and under each guideline. RESULTS: Identified were 1401 cases of equivocal appendicitis; 958 (68.4%) with complete data. The mean age was 11 +/- 4.3 years. Of 958 children, 588 (61.4%) had acute appendicitis. One hundred forty-three patients were in the low-risk group, defined as neutrophils 5%, and no guarding on abdominal examination. Fifteen (10%) children had appendicitis. Two hundred twenty-five were high-risk for appendicitis defined as neutrophils >67%, white blood cell count >10,000/mm(3), guarding, and abdominal pain >13 hours. Of these, 202 (90%) had appendicitis. Under guideline 1, there were 22 negative appendectomies, 35 missed or delayed diagnoses, and 958 USs and 673 CT scans performed. Under guideline 2, there would have been 23 negative appendectomies, 36 missed or delayed diagnoses, and 733 USs and 637 CT scans performed. Under guideline 3, there would have been 36 negative appendectomies, 37 missed or delayed diagnoses, and 590 USs and 412 CT scans performed. CONCLUSIONS: Selective imaging guidelines can reduce the number of radiographic studies performed with minimal diminution in accuracy of diagnosis of pediatric appendicitis.
Bosl, Mandel, Jonikas, Ramoni, Kohane, Mandl. Scalable decision support at the point of care: a substitutable electronic health record app for monitoring medication adherence. Interact J Med Res. 2013;2:e13.
BACKGROUND: Non-adherence to prescribed medications is a serious health problem in the United States, costing an estimated $100 billion per year. While poor adherence should be addressable with point of care health information technology, integrating new solutions with existing electronic health records (EHR) systems require customization within each organization, which is difficult because of the monolithic software design of most EHR products. OBJECTIVE: The objective of this study was to create a published algorithm for predicting medication adherence problems easily accessible at the point of care through a Web application that runs on the Substitutable Medical Apps, Reusuable Technologies (SMART) platform. The SMART platform is an emerging framework that enables EHR systems to behave as "iPhone like platforms" by exhibiting an application programming interface for easy addition and deletion of third party apps. The app is presented as a point of care solution to monitoring medication adherence as well as a sufficiently general, modular application that may serve as an example and template for other SMART apps. METHODS: The widely used, open source Django framework was used together with the SMART platform to create the interoperable components of this app. Django uses Python as its core programming language. This allows statistical and mathematical modules to be created from a large array of Python numerical libraries and assembled together with the core app to create flexible and sophisticated EHR functionality. Algorithms that predict individual adherence are derived from a retrospective study of dispensed medication claims from a large private insurance plan. Patients' prescription fill information is accessed through the SMART framework and the embedded algorithms compute adherence information, including predicted adherence one year after the first prescription fill. Open source graphing software is used to display patient medication information and the results of statistical prediction of future adherence on a clinician-facing Web interface. RESULTS: The user interface allows the physician to quickly review all medications in a patient record for potential non-adherence problems. A gap-check and current medication possession ratio (MPR) threshold test are applied to all medications in the record to test for current non-adherence. Predictions of 1-year non-adherence are made for certain drug classes for which external data was available. Information is presented graphically to indicate present non-adherence, or predicted non-adherence at one year, based on early prescription fulfillment patterns. The MPR Monitor app is installed in the SMART reference container as the "MPR Monitor", where it is publically available for use and testing. MPR is an acronym for Medication Possession Ratio, a commonly used measure of adherence to a prescribed medication regime. This app may be used as an example for creating additional functionality by replacing statistical and display algorithms with new code in a cycle of rapid prototyping and implementation or as a framework for a new SMART app. CONCLUSIONS: The MPR Monitor app is a useful pilot project for monitoring medication adherence. It also provides an example that integrates several open source software components, including the Python-based Django Web framework and python-based graphics, to build a SMART app that allows complex decision support methods to be encapsulated to enhance EHR functionality.
Mandl, Kohane, McFadden, Weber, Natter, Mandel, Schneeweiss, Weiler, Klann, Bickel, et al. Scalable Collaborative Infrastructure for a Learning Healthcare System (SCILHS): Architecture. J Am Med Inform Assoc. 2014.

We describe the architecture of the Patient Centered Outcomes Research Institute (PCORI) funded Scalable Collaborative Infrastructure for a Learning Healthcare System (SCILHS, http://www.SCILHS.org) clinical data research network, which leverages the $48 billion dollar federal investment in health information technology (IT) to enable a queryable semantic data model across 10 health systems covering more than 8 million patients, plugging universally into the point of care, generating evidence and discovery, and thereby enabling clinician and patient participation in research during the patient encounter. Central to the success of SCILHS is development of innovative 'apps' to improve PCOR research methods and capacitate point of care functions such as consent, enrollment, randomization, and outreach for patient-reported outcomes. SCILHS adapts and extends an existing national research network formed on an advanced IT infrastructure built with open source, free, modular components.

Mohtashemi, Szolovits, Dunyak, Mandl. A susceptible-infected model of early detection of respiratory infection outbreaks on a background of influenza. J Theor Biol. 2006;241:954–63.
The threat of biological warfare and the emergence of new infectious agents spreading at a global scale have highlighted the need for major enhancements to the public health infrastructure. Early detection of epidemics of infectious diseases requires both real-time data and real-time interpretation of data. Despite moderate advancements in data acquisition, the state of the practice for real-time analysis of data remains inadequate. We present a nonlinear mathematical framework for modeling the transient dynamics of influenza, applied to historical data sets of patients with influenza-like illness. We estimate the vital time-varying epidemiological parameters of infections from historical data, representing normal epidemiological trends. We then introduce simulated outbreaks of different shapes and magnitudes into the historical data, and estimate the parameters representing the infection rates of anomalous deviations from normal trends. Finally, a dynamic threshold-based detection algorithm is devised to assess the timeliness and sensitivity of detecting the irregularities in the data, under a fixed low false-positive rate. We find that the detection algorithm can identify such designated abnormalities in the data with high sensitivity with specificity held at 97%, but more importantly, early during an outbreak. The proposed methodology can be applied to a broad range of influenza-like infectious diseases, whether naturally occurring or a result of bioterrorism, and thus can be an integral component of a real-time surveillance system.
Cassa, Iancu, Olson, Mandl. A software tool for creating simulated outbreaks to benchmark surveillance systems. BMC Med Inform Decis Mak. 2005;5:22.
BACKGROUND: Evaluating surveillance systems for the early detection of bioterrorism is particularly challenging when systems are designed to detect events for which there are few or no historical examples. One approach to benchmarking outbreak detection performance is to create semi-synthetic datasets containing authentic baseline patient data (noise) and injected artificial patient clusters, as signal. METHODS: We describe a software tool, the AEGIS Cluster Creation Tool (AEGIS-CCT), that enables users to create simulated clusters with controlled feature sets, varying the desired cluster radius, density, distance, relative location from a reference point, and temporal epidemiological growth pattern. AEGIS-CCT does not require the use of an external geographical information system program for cluster creation. The cluster creation tool is an open source program, implemented in Java and is freely available under the Lesser GNU Public License at its Sourceforge website. Cluster data are written to files or can be appended to existing files so that the resulting file will include both existing baseline and artificially added cases. Multiple cluster file creation is an automated process in which multiple cluster files are created by varying a single parameter within a user-specified range. To evaluate the output of this software tool, sets of test clusters were created and graphically rendered. RESULTS: Based on user-specified parameters describing the location, properties, and temporal pattern of simulated clusters, AEGIS-CCT created clusters accurately and uniformly. CONCLUSION: AEGIS-CCT enables the ready creation of datasets for benchmarking outbreak detection systems. It may be useful for automating the testing and validation of spatial and temporal cluster detection algorithms.
McMurry, Gilbert, Reis, Chueh, Kohane, Mandl. A self-scaling, distributed information architecture for public health, research, and clinical care. J Am Med Inform Assoc. 2007;14:527–33.
OBJECTIVE: This study sought to define a scalable architecture to support the National Health Information Network (NHIN). This architecture must concurrently support a wide range of public health, research, and clinical care activities. STUDY DESIGN: The architecture fulfils five desiderata: (1) adopt a distributed approach to data storage to protect privacy, (2) enable strong institutional autonomy to engender participation, (3) provide oversight and transparency to ensure patient trust, (4) allow variable levels of access according to investigator needs and institutional policies, (5) define a self-scaling architecture that encourages voluntary regional collaborations that coalesce to form a nationwide network. RESULTS: Our model has been validated by a large-scale, multi-institution study involving seven medical centers for cancer research. It is the basis of one of four open architectures developed under funding from the Office of the National Coordinator of Health Information Technology, fulfilling the biosurveillance use case defined by the American Health Information Community. The model supports broad applicability for regional and national clinical information exchanges. CONCLUSIONS: This model shows the feasibility of an architecture wherein the requirements of care providers, investigators, and public health authorities are served by a distributed model that grants autonomy, protects privacy, and promotes participation.

R

Dunn, Bourgeois, Murthy, Mandl, Day, Coiera. The role and impact of research agendas on the comparative-effectiveness research among antihyperlipidemics. Clin Pharmacol Ther. 2012;91:685–91.
Although it is well established that funding source influences the publication of clinical trials, relatively little is known about how funding influences trial design. We examined a public trial registry to determine how funding source shapes trial design among trials involving antihyperlipidemics. We used an automated process to identify and analyze 809 trials from a set of 72,564. Three networks representing industry-, collaboratively, and non-industry-funded trials were constructed. Each network comprised 18 drugs as nodes connected according to the number of comparisons made between them. The results indicated that industry-funded trials were more likely to compare across drugs and examine dyslipidemia as a condition, and less likely to register safety outcomes. The source of funding for clinical trials had a measurable effect on trial design, which helps quantify differences in research agendas. Improved monitoring of current clinical trials may be used to more closely align research agendas to clinical needs.