Lost in Translation (LiT)

Translational medicine is a roller coaster with occasional brilliant successes and a large majority of failures. Lost in Translation 1 (‘LiT1’), beginning in the 1950s, was a golden era built upon earlier advances in experimental physiology, biochemistry and pharmacology, with a dash of serendipity, that led to the discovery of many new drugs for serious illnesses. LiT2 saw the large-scale industrialization of drug discovery using high-throughput screens and assays based on affinity for the target molecule. The links between drug development and university sciences and medicine weakened, but there were still some brilliant successes. In LiT3, the coverage of translational medicine expanded from molecular biology to drug budgets, with much greater emphasis on safety and official regulation. Compared with R&D expenditure, the number of breakthrough discoveries in LiT3 was disappointing, but monoclonal antibodies for immunity and inflammation brought in a new golden era and kinase inhibitors such as imatinib were breakthroughs in cancer. The pharmaceutical industry is trying to revive the LiT1 approach by using phenotypic assays and closer links with academia. LiT4 faces a data explosion generated by the genome project, GWAS, ENCODE and the ‘omics’ that is in danger of leaving LiT4 in a computerized cloud. Industrial laboratories are filled with masses of automated machinery while the scientists sit in a separate room viewing the results on their computers. Big Data will need Big Thinking in LiT4 but with so many unmet medical needs and so many new opportunities being revealed there are high hopes that the roller coaster will ride high again.     

Extracting general practice data for timely vaccine coverage estimates: The England experience

In England, primary care providers use standardised coding systems to record health events such as vaccination as well as patient characteristics. This information can be automatically extracted to estimate coverage for vaccine programmes delivered through primary care, in the general population as well as in specific geographical, ethnic, age or clinical groups. This system provides timely vaccine coverage estimates as well as the flexibility to extract tailored data in order to directly inform a continuously evolving national vaccine programme. It is however limited by the quality and completeness of clinical coding in primary care. A centralised, individual-level register would however improve data quality, completeness and reliability and remains the gold standard.     

Utility-preserving privacy protection of textual healthcare documents

The adoption of ITs by medical organisations makes possible the compilation of large amounts of healthcare data, which are quite often needed to be released to third parties for research or business purposes. Many of this data are of sensitive nature, because they may include patient-related documents such as electronic healthcare records. In order to protect the privacy of individuals, several legislations on healthcare data management, which state the kind of information that should be protected, have been defined. Traditionally, to meet with current legislations, a manual redaction process is applied to patient-related documents in order to remove or black-out sensitive terms. This process is costly and time-consuming and has the undesired side effect of severely reducing the utility of the released content. Automatic methods available in the literature usually propose ad-hoc solutions that are limited to protect specific types of structured information (e.g. e-mail addresses, social security numbers, etc.); as a result, they are hardly applicable to the sensitive entities stated in current regulations that do not present those structural regularities (e.g. diseases, symptoms, treatments, etc.). To tackle these limitations, in this paper we propose an automatic sanitisation method for textual medical documents (e.g. electronic healthcare records) that is able to protect, regardless of their structure, sensitive entities (e.g. diseases) and also those semantically related terms (e.g. symptoms) that may disclose the former ones. Contrary to redaction schemes based on term removal, our approach improves the utility of the protected output by replacing sensitive terms with appropriate generalisations retrieved from several medical and general-purpose knowledge bases. Experiments conducted on highly sensitive documents and in coherency with current regulations on healthcare data privacy show promising results in terms of the practical privacy and utility of the protected output.     

B-Mode Ultrasound Combined with Color Doppler and Strain Elastography in the Diagnosis of Non-mass Breast Lesions: A Prospective Study

Non-mass breast lesions on ultrasound (US) are areas without an associated mass. The purpose of this study was to evaluate whether combining B-mode US with color Doppler US and strain elastography (SE) improves US differentiation between benign and malignant non-mass breast lesions and the decision for biopsy. In this prospective study, three different radiologists analyzed the US images of 77 non-mass lesions independently and recorded Breast Imaging Reporting and Data System (BI-RADS) categories for four data sets. The image characteristics and BI-RADS categories of the four data sets were analyzed by another radiologist. The final diagnosis was made on the basis of pathologic findings. Values for area under the receiver operating curve (AUC), sensitivity, specificity and accuracy were compared among the data sets. The AUC of B-mode US combined with both color Doppler US and SE was greater than that of B-mode US alone (0.666 vs. 0.828) (p = 0.011). The specificity of making the decision for biopsy increased from 6.5% to 38.7% when B-mode US was combined with color Doppler and SE, without a statistically significant change in sensitivity (p < 0.001). Combined use of color Doppler and SE could improve the diagnostic value of B-mode US in distinguishing benign from malignant non-mass breast lesions and the specificity of making the decision for biopsy of non-mass breast lesions.     

Fetal death certificate data quality: a tale of two U.S. counties

PurposeDescribe the relative frequency and joint effect of missing and misreported fetal death certificate (FDC) data and identify variations by key characteristics.MethodsStillbirths were prospectively identified during 2006–2008 for a multisite population-based case-control study. For this study, eligible mothers of stillbirths were not incarcerated residents of DeKalb County, Georgia, or Salt Lake County, Utah, aged ≥13 years, with an identifiable FDC. We identified the frequency of missing and misreported (any departure from the study value) FDC data by county, race/ethnicity, gestational age, and whether the stillbirth was antepartum or intrapartum.ResultsData quality varied by item and was highest in Salt Lake County. Reporting was generally not associated with maternal or delivery characteristics. Reasons for poor data quality varied by item in DeKalb County: some items were frequently missing and misreported; however, others were of poor quality due to either missing or misreported data.ConclusionsFDC data suffer from missing and inaccurate data, with variations by item and county. Salt Lake County data illustrate that high quality reporting is attainable. The overall quality of reporting must be improved to support consequential epidemiologic analyses for stillbirth, and improvement efforts should be tailored to the needs of each jurisdiction.     

Brief communication: The Mid-South Clinical Data Research Network

The Mid-South Clinical Data Research Network (CDRN) encompasses three large health systems: (1) Vanderbilt Health System (VU) with electronic medical records for over 2 million patients, (2) the Vanderbilt Healthcare Affiliated Network (VHAN) which currently includes over 40 hospitals, hundreds of ambulatory practices, and over 3 million patients in the Mid-South, and (3) Greenway Medical Technologies, with access to 24 million patients nationally. Initial goals of the Mid-South CDRN include: (1) expansion of our VU data network to include the VHAN and Greenway systems, (2) developing data integration/interoperability across the three systems, (3) improving our current tools for extracting clinical data, (4) optimization of tools for collection of patient-reported data, and (5) expansion of clinical decision support. By 18 months, we anticipate our CDRN will robustly support projects in comparative effectiveness research, pragmatic clinical trials, and other key research areas and have the capacity to share data and health information technology tools nationally.     

Implementing partnership-driven clinical federated electronic health record data sharing networks

ObjectiveBuilding federated data sharing architectures requires supporting a range of data owners, effective and validated semantic alignment between data resources, and consistent focus on end-users. Establishing these resources requires development methodologies that support internal validation of data extraction and translation processes, sustaining meaningful partnerships, and delivering clear and measurable system utility. We describe findings from two federated data sharing case examples that detail critical factors, shared outcomes, and production environment results.MethodsTwo federated data sharing pilot architectures developed to support network-based research associated with the University of Washington’s Institute of Translational Health Sciences provided the basis for the findings. A spiral model for implementation and evaluation was used to structure iterations of development and support knowledge share between the two network development teams, which cross collaborated to support and manage common stages.ResultsWe found that using a spiral model of software development and multiple cycles of iteration was effective in achieving early network design goals. Both networks required time and resource intensive efforts to establish a trusted environment to create the data sharing architectures. Both networks were challenged by the need for adaptive use cases to define and test utility.ConclusionAn iterative cyclical model of development provided a process for developing trust with data partners and refining the design, and supported measureable success in the development of new federated data sharing architectures.     

Patterns of aged care use among older Australian women: A prospective cohort study using linked data

BackgroundWomen live longer than men and have an increased need for long-term care. The objective of this study was to identify patterns of aged care use among older Australian women and to examine how these patterns were associated with their demographic and health-related characteristics.MethodsThe sample consisted of 8768 women from the 1921–1926 birth cohort of the Australian Longitudinal Study on Women’s Health (ALSWH), who had survived to age 75-80 years. ALSWH survey and linked administrative aged care and death datasets from 2001 to 2011 were utilized. Patterns of aged care use were identified using a repeated measure latent class analysis.ResultsWe identified four patterns of aged care use over time, differentiated by timing of service onset, types of service use and time of death. Approximately 41% of the sample were non-users or using basic home and community care (HACC), while 24% were at high risk of using moderate to high-level HACC/community aged care package (CACP). Only 11% had a greater risk of using residential aged care (RAC) over time. Being widowed, residing in remote/regional areas, having difficulty in managing income, having a chronic condition, reporting poor/fair self-rated health, and lower SF-36 quality of life scores were associated with an increased odds of being a member of the following classes: 1) moderate to high-level HACC/CACP, 2) increasing RAC, and 3) early mortality, compared with the non-user class.ConclusionsDistinct patterns of aged care use were identified. These results will facilitate future capacity planning for aged care systems in Australia.