Use of electronic medical records (EMR) for oncology outcomes research: assessing the comparability of EMR information to patient registry and health claims data

Renal response in real-world carfilzomib- vs bortezomib-treated patients with relapsed or refractory multiple myeloma

In the phase 3 ENDEAVOR study, carfilzomib-dexamethasone (Kd) improved survival over bortezomib-dexamethasone (Vd) in patients with relapsed or refractory multiple myeloma (RRMM), regardless of baseline renal function. This real-world study compared renal response in patients with RRMM (1-3 prior lines) and renal impairment (estimated glomerular filtration rate ≤50 mL/min) treated with Kd vs Vd. Electronic medical records data from the Oncology Services Comprehensive Electronic Records database were assessed (from January 2012 through February 2018). Time to renal response (defined according to International Myeloma Working Group criteria) was evaluated using the Kaplan-Meier method and log-rank test. Incidence rate ratios (IRRs) and 95% confidence intervals (CIs) were calculated for renal overall response (ROR) and renal complete response (RCR) using Cox proportional hazard models adjusted for baseline covariates. Included were 543 Kd-treated and 1005 Vd-treated patients. In line 2 (2L), compared with Vd, Kd achieved significantly higher ROR (51.4% vs 39.6%; P < .0001) and RCR (26.6% vs 22.2%; P = .0229). After baseline covariate adjustment, 2L patients receiving Kd vs Vd were 45% more likely to achieve ROR (IRR, 1.45; 95% CI, 1.18-1.78), and 68% were more likely to achieve RCR (IRR, 1.68; 95% CI, 1.24-2.28). The renal response benefit with Kd remained consistent in 2L to line 4 (4L). In a combined analysis of patients receiving Kd and Vd (2L and 2L-4L), renal responders had longer overall survival and time to next treatment than renal nonresponders. These results demonstrate improved real-world effectiveness of Kd over Vd in RRMM renal rescue, and the positive association between renal response and improved survival.

A Conceptual Model of an Oncology Information System

Introduction

Oncologists are usually faced with a huge amount of diagnostic and therapeutic data in the process of cancer care. However, they do not have access to the integrated data. This research aimed to present a conceptual model of an oncology information system based on the users' requirements.

Methods

This study was conducted in 2019 and composed of two phases. Initially, a questionnaire was designed, and clinical experts (n=34) were asked to identify the most important data elements and functional requirements in an oncology information system. In the second phase, conceptual, structural and behavioral diagrams of the system were drawn based on the results of the first phase. These diagrams were also reviewed and validated by five experts.

Results

Most of the data elements and all functional requirements were found important by the experts. The data elements were related to different phases of cancer care including screening, prevention, diagnosis, treatment, mental care and pain relief, and end-of-life care. Then, conceptual, structural and behavioral diagrams of the system were designed and approved by the experts or revised based on their comments.

Conclusion

The conceptual model and the diagrams presented in the current study can be used for developing an oncology information system. This system will be able to manage patients' cancer data from screening to the end-of-life care. However, the system needs to be designed and implemented in a real healthcare setting to see how it can meet users' requirements.

Appropriateness of granulocyte colony-stimulating factor use in patients receiving chemotherapy by febrile neutropenia risk level

OBJECTIVE

Inappropriate granulocyte colony-stimulating factor use with myelosuppressive chemotherapy has been reported. Using the Oncology Services Comprehensive Electronic Records electronic medical record database, prophylactic granulocyte colony-stimulating factor (pegfilgrastim/filgrastim) use in cancer patients was assessed by febrile neutropenia risk level.

METHODS

Patients with nonmetastatic or metastatic breast, head/neck, colorectal, ovarian/gynecologic, lung cancer, or non-Hodgkin's lymphoma who received myelosuppressive chemotherapy from June 2013 to May 2014 were included. Prophylactic granulocyte colony-stimulating factor use with high-risk, intermediate-risk, and low-risk chemotherapy and distribution of National Comprehensive Cancer Network risk factors with intermediate-risk regimens were assessed.

RESULTS

Overall, 86,189 patients received ∼4.2 million chemotherapy cycles (high risk, 9%; intermediate risk, 48%; low risk, 43%). Prophylactic granulocyte colony-stimulating factor was given in 24% of cycles (high risk, 59%; intermediate risk, 29%; low risk, 11%). For nonmetastatic solid tumors, granulocyte colony-stimulating factor was given in 78% (high risk), 31% (intermediate risk), and 6% (low risk) of cycles. For metastatic solid tumors or non-Hodgkin's lymphoma, granulocyte colony-stimulating factor was given in 50% (high risk), 27% (intermediate risk), and 11% (low risk) of cycles. Among patients receiving intermediate-risk regimens with granulocyte colony-stimulating factor, febrile neutropenia risk factors were identified in 56% (95% confidence interval, 51.1-60.9%) of patients with nonmetastatic solid tumors (n = 400) and in 70% (64.5-73.5%) of patients with metastatic solid tumors or non-Hodgkin's lymphoma (n = 400).

CONCLUSION

Prophylactic granulocyte colony-stimulating factor use was appropriately highest for high-risk regimens and lowest for low-risk regimens yet still potentially underused in high risk regimens, overused in low-risk regimens, and not appropriately targeted in intermediate-risk regimens, indicating a need for further education on febrile neutropenia risk evaluation and appropriate granulocyte colony-stimulating factor use.

Relevant Word Order Vectorization for Improved Natural Language Processing in Electronic Health Records

Electronic health records (EHR) represent a rich resource for conducting observational studies, supporting clinical trials, and more. However, much of the data contains unstructured text, presenting an obstacle to automated extraction. Natural language processing (NLP) can structure and learn from text, but NLP algorithms were not designed for the unique characteristics of EHR. Here, we propose Relevant Word Order Vectorization (RWOV) to aid with structuring. RWOV is based on finding the positional relationship between the most relevant words to predicting the class of a text. This facilitates machine learning algorithms to use the interaction of not just keywords but positional dependencies (e.g. a relevant word occurs 5 relevant words before some term of interest). As a proof-of-concept, we attempted to classify the hormone receptor status of breast cancer patients treated at the University of Kansas Medical Center, comparing RWOV to other methods using the F1 score and AUC. RWOV performed as well as, or better than other methods in all but one case. For F1 score, RWOV had a clear edge on most tasks. AUC tended to be closer, but for HER2, RWOV was significantly better for most comparisons. These results suggest RWOV should be further developed for EHR-related NLP.

Linkage between Utah All Payers Claims Database and Central Cancer Registry

OBJECTIVE

To evaluate the linkage of claims from the Utah All Payers Claims Database (APCD) and Utah Cancer Registry (UCR).

DATA SOURCES

Secondary data from 2013 and 2014 Utah APCD and 2013 UCR cases.

STUDY DESIGN

This is a descriptive analysis of the quality of linkage between APCD claims data and cancer registry cases.

DATA COLLECTION/EXTRACTION METHODS

We used the LinkPlus software to link Utah APCD and UCR data.

PRINCIPAL FINDINGS

We were able to link 82.4 percent (9441/11 453) of the UCR reportable cancer cases with APCD claims. Of those linked, 66 percent were perfect matches.

CONCLUSIONS

The quality of identifiers is high, evidence that claims data can potentially supplement cancer registry data for use in research.

Linkage between Utah All Payers Claims Database and Central Cancer Registry

OBJECTIVE

To evaluate the linkage of claims from the Utah All Payers Claims Database (APCD) and Utah Cancer Registry (UCR).

DATA SOURCES

Secondary data from 2013 and 2014 Utah APCD and 2013 UCR cases.

STUDY DESIGN

This is a descriptive analysis of the quality of linkage between APCD claims data and cancer registry cases.

DATA COLLECTION/EXTRACTION METHODS

We used the LinkPlus software to link Utah APCD and UCR data.

PRINCIPAL FINDINGS

We were able to link 82.4 percent (9441/11 453) of the UCR reportable cancer cases with APCD claims. Of those linked, 66 percent were perfect matches.

CONCLUSIONS

The quality of identifiers is high, evidence that claims data can potentially supplement cancer registry data for use in research.

Electronic Health Record-Based Registries: Clinical Research Using Registries in Colon and Rectal Surgery

Electronic health records (EHRs) or electronic medical records (EMRs) contain a vast amount of clinical data that can be useful for multiple purposes including research. Disease registries are collections of data in predefined formats for population management, research, and other purposes. There are differences between EHRs and registries in the data structure, data standards, and protocols. Proprietary EHR systems use different coding systems and data standards, which are usually kept secret. For EHR data to flow seamlessly into registries, there is the need for interoperability between EHR systems and between EHRs and registries. The levels of interoperability required include functional, structural, and semantic interoperability. EHR data can be manually mapped to registry data, but that is a tedious, resource-intensive endeavor. The development of data standards that can be used as building blocks for both EHRs and registries will help overcome the problem of interoperability.

Change in Survival in Metastatic Breast Cancer with Treatment Advances: Meta-Analysis and Systematic Review

Background

Metastatic breast cancer (MBC) treatment has changed substantially over time, but we do not know whether survival post-metastasis has improved at the population level.

Methods

We searched for studies of MBC patients that reported survival after metastasis in at least two time periods between 1970 and the present. We used meta-regression models to test for survival improvement over time in four disease groups: recurrent, recurrent estrogen (ER)-positive, recurrent ER-negative, and de novo stage IV. We performed sensitivity analyses based on bias in some studies that could lead earlier cohorts to include more aggressive cancers.

Results

There were 15 studies of recurrent MBC (N = 18 678 patients; 3073 ER-positive and 1239 ER-negative); meta-regression showed no survival improvement among patients recurring between 1980 and 1990, but median survival increased from 21 (95% confidence interval [CI] = 18 to 25) months to 38 (95% CI = 31 to 47) months from 1990 to 2010. For ER-positive MBC patients, median survival increased during 1990–2010 from 32 (95% CI = 23 to 43) to 57 (95% CI = 37 to 87) months, and for ER-negative MBC patients from 14 (95% CI = 11 to 19) to 33 (95% CI = 21 to 51) months. Among eight studies (N = 35 831) of de novo stage IV MBC, median survival increased during 1990–2010 from 20 (95% CI = 16 to 24) to 31 (95% CI = 24 to 39) months. Results did not change in sensitivity analyses.

Conclusion

By bridging studies over time, we demonstrated improvements in survival for recurrent and de novo stage IV MBC overall and across ER-defined subtypes since 1990. These results can inform patient-doctor discussions about MBC prognosis and therapy.

Biases introduced by filtering electronic health records for patients with "complete data"

Objective

One promise of nationwide adoption of electronic health records (EHRs) is the availability of data for large-scale clinical research studies. However, because the same patient could be treated at multiple health care institutions, data from only a single site might not contain the complete medical history for that patient, meaning that critical events could be missing. In this study, we evaluate how simple heuristic checks for data "completeness" affect the number of patients in the resulting cohort and introduce potential biases.

Materials and Methods

We began with a set of 16 filters that check for the presence of demographics, laboratory tests, and other types of data, and then systematically applied all 216 possible combinations of these filters to the EHR data for 12 million patients at 7 health care systems and a separate payor claims database of 7 million members.

Results

EHR data showed considerable variability in data completeness across sites and high correlation between data types. For example, the fraction of patients with diagnoses increased from 35.0% in all patients to 90.9% in those with at least 1 medication. An unrelated claims dataset independently showed that most filters select members who are older and more likely female and can eliminate large portions of the population whose data are actually complete.

Discussion and Conclusion

As investigators design studies, they need to balance their confidence in the completeness of the data with the effects of placing requirements on the data on the resulting patient cohort.

How Confounder Strength Can Affect Allocation of Resources in Electronic Health Records

When electronic health record (EHR) data are used, multiple approaches may be available for measuring the same variable, introducing potentially confounding factors. While additional information may be gleaned and residual confounding reduced through resource-intensive assessment methods such as natural language processing (NLP), whether the added benefits offset the added cost of the additional resources is not straightforward. We evaluated the implications of misclassification of a confounder when using EHRs. Using a combination of simulations and real data surrounding hospital readmission, we considered smoking as a potential confounder. We compared ICD-9 diagnostic code assignment, which is an easily available measure but has the possibility of substantial misclassification of smoking status, with NLP, a method of determining smoking status that more expensive and time-consuming than ICD-9 code assignment but has less potential for misclassification. Classification of smoking status with NLP consistently produced less residual confounding than the use of ICD-9 codes; however, when minimal confounding was present, differences between the approaches were small. When considerable confounding is present, investing in a superior measurement tool becomes advantageous.

Estimation of State-Level Prevalence of Hepatitis C Virus Infection, US States and District of Columbia, 2010

BACKGROUND.

Hepatitis C virus (HCV) infection is the most common chronic blood-borne infection in the United States and a leading cause of morbidity and mortality. Previous analyses of the US National Health and Nutrition Examination Survey (NHANES) indicated approximately 3.6 million noninstitutionalized persons with antibody to HCV (anti-HCV). However, state-level prevalence remains less understood and cannot be estimated reliably from NHANES alone.

METHODS.

We used 3 publicly available government data sources to estimate anti-HCV prevalence in each US state among noninstitutionalized persons aged ≥18 years. A small-area estimation model combined indirect standardization of NHANES-based prevalence with logistic regression modeling of mortality data, listing acute or chronic HCV infection as a cause of death, from the National Vital Statistics System during 1999-2012. Model results were combined with US Census population sizes to estimate total number and prevalence of persons with antibody to HCV in 2010.

RESULTS.

National anti-HCV prevalence was 1.67% (95% confidence interval [CI], 1.53-1.90), or 3 911 800 (95% CI, 3 589 400- 4 447 500) adults in 2010. State-specific prevalence ranged from 0.71% (Illinois) to 3.34% (Oklahoma). The West census region had the highest region-specific prevalence (2.14% [95% CI, 1.96-2.48]); 10 of 13 states had rates above the national average. The South had the highest number of persons with anti-HCV (n = 1561600 [95% CI, 1 427 700-1 768 900]). The Midwest had the lowest region-specific prevalence (1.14% [95% CI, 1.04%-1.30%]).

CONCLUSIONS.

States in the US West and South have been most impacted by hepatitis C. Estimates of HCV infection burden are essential to guide policy and programs to optimally prevent, detect, and cure infection.

Complex temporal topic evolution modelling using the Kullback-Leibler divergence and the Bhattacharyya distance

The rapidly expanding corpus of medical research literature presents major challenges in the understanding of previous work, the extraction of maximum information from collected data, and the identification of promising research directions. We present a case for the use of advanced machine learning techniques as an aide in this task and introduce a novel methodology that is shown to be capable of extracting meaningful information from large longitudinal corpora and of tracking complex temporal changes within it. Our framework is based on (i) the discretization of time into epochs, (ii) epoch-wise topic discovery using a hierarchical Dirichlet process-based model, and (iii) a temporal similarity graph which allows for the modelling of complex topic changes. More specifically, this is the first work that discusses and distinguishes between two groups of particularly challenging topic evolution phenomena: topic splitting and speciation and topic convergence and merging, in addition to the more widely recognized emergence and disappearance and gradual evolution. The proposed framework is evaluated on a public medical literature corpus.

-Omic and Electronic Health Record Big Data Analytics for Precision Medicine

OBJECTIVE

Rapid advances of high-throughput technologies and wide adoption of electronic health records (EHRs) have led to fast accumulation of -omic and EHR data. These voluminous complex data contain abundant information for precision medicine, and big data analytics can extract such knowledge to improve the quality of healthcare.

METHODS

In this paper, we present -omic and EHR data characteristics, associated challenges, and data analytics including data preprocessing, mining, and modeling.

RESULTS

To demonstrate how big data analytics enables precision medicine, we provide two case studies, including identifying disease biomarkers from multi-omic data and incorporating -omic information into EHR.

CONCLUSION

Big data analytics is able to address -omic and EHR data challenges for paradigm shift toward precision medicine.

SIGNIFICANCE

Big data analytics makes sense of -omic and EHR data to improve healthcare outcome. It has long lasting societal impact.

Burden of symptoms associated with development of metastatic bone disease in patients with breast cancer

Purpose

Women with breast cancer frequently develop painful bone metastases. This retrospective study was designed to longitudinally characterize patterns of patient-reported symptoms among patients with breast cancer relative to the diagnosis of bone metastases.

Methods

Patient records were identified from the Oncology Services Comprehensive Electronic Records (OSCER) database which includes outpatient oncology practices across the USA. Symptom burden was assessed by Patient Care Monitor (PCM) assessments, which are administered as part of routine care in a subset of these practices. Eligible patients were women diagnosed with breast cancer (ICD-9-CM 174.xx) who developed bone metastases (ICD-9-CM 198.5) and had ≥1 PCM assessment between January 2007 and December 2012. The pre-specified endpoint was the occurrence of moderate to severe symptom burden, defined as PCM score ≥4 (0–10 scale).

Results

One thousand one hundred five women (median age, 61) met the eligibility criteria. Worsening of symptoms, particularly fatigue and pain, occurred in the months leading up to the diagnosis of bone metastases. After bone metastases diagnosis, the rate of increase in the proportion of patients experiencing moderate/severe symptoms slowed, but continued to climb during follow-up. Median time to moderate/severe symptoms was 0.9 month for fatigue, 1 month for pain, 2.9 months for trouble sleeping, and 7.7 months for numbness/tingling. Half of the patients received bone-targeted agents after diagnosis of bone metastases.

Conclusions

Symptom burden, especially pain and fatigue, increased both before and after the diagnosis of bone metastases, highlighting the need for proactive monitoring and management of symptoms in breast cancer patients.

Electronic supplementary material

The online version of this article (doi:10.1007/s00520-016-3154-x) contains supplementary material, which is available to authorized users.

Development of a large urban longitudinal HIV clinical cohort using a web-based platform to merge electronically and manually abstracted data from disparate medical record systems: technical challenges and innovative solutions

OBJECTIVE

Electronic medical records (EMRs) are being increasingly utilized to conduct clinical and epidemiologic research in numerous fields. To monitor and improve care of HIV-infected patients in Washington, DC, one of the most severely affected urban areas in the United States, we developed a city-wide database across 13 clinical sites using electronic data abstraction and manual data entry from EMRs.

MATERIALS AND METHODS

To develop this unique longitudinal cohort, a web-based electronic data capture system (Discovere®) was used. An Agile software development methodology was implemented across multiple EMR platforms. Clinical informatics staff worked with information technology specialists from each site to abstract data electronically from each respective site's EMR through an extract, transform, and load process.

RESULTS

Since enrollment began in 2011, more than 7000 patients have been enrolled, with longitudinal clinical data available on all patients. Data sets are produced for scientific analyses on a quarterly basis, and benchmarking reports are generated semi-annually enabling each site to compare their participants' clinical status, treatments, and outcomes to the aggregated summaries from all other sites.

DISCUSSION

Numerous technical challenges were identified and innovative solutions developed to ensure the successful implementation of the DC Cohort. Central to the success of this project was the broad collaboration established between government, academia, clinics, community, information technology staff, and the patients themselves.

CONCLUSIONS

Our experiences may have practical implications for researchers who seek to merge data from diverse clinical databases, and are applicable to the study of health-related issues beyond HIV.

Conducting record review studies in clinical practice

Clinical record review or chart review is a previously recorded data to answer clinical queries. Such a study can be used to answer specific clinical questions in a relatively easy and less resource intensive manner. But these studies may be constrained by the limited information retrievable and inadequacy of records. Various types of data sources may be available for conducting such reviews (like case charts, computerized registries, etc), each with specific strengths and weaknesses. The procedure usually consists of drawing up the research question, identifying the appropriate data source, devising a data extraction plan, extracting the data, checking for errors, data analysis, and appropriate archiving and dissemination of the findings. The ethical aspects in such studies primarily pertain to issues of informed consent and confidentiality. This paper provides a broad overview of how to go about a clinical record review, and serves as a ready reference for those who would like to undertake such record reviews.

Patients with bone metastases from solid tumors initiating treatment with a bone-targeted agent in 2011: a descriptive analysis using oncology clinic data in the US

PURPOSE

Three bone-targeted agents (BTAs) are approved in the USA for prevention of bone complications among solid tumor patients with bone metastases: two intravenous bisphosphonates (IV BP) (pamidronate and zoledronic acid), and one subcutaneous receptor activator of nuclear factor-kappaB (RANK) ligand inhibitor (denosumab). Using electronic medical record data from outpatient community and hospital-affiliated oncology clinics, we examined the characteristics of patients who initiated treatment with a BTA in 2011 and followed them for a maximum of 12 months.

METHODS

Adult patients with bone metastasis secondary to solid tumors newly treated with a BTA during 2011 were identified from the Oncology Services Comprehensive Electronic Records (OSCER) database. We examined patient characteristics at BTA initiation, treatment patterns, and compliance during a 12-month period. Sensitivity analyses were performed in a subgroup of patients who had confirmed 12 months of follow-up data.

RESULTS

Denosumab patients (N = 1,594) were older (65 % ≥65 years vs. 60 % ≥65 years), further along in their disease progression (time since bone metastasis diagnosis: 16 % ≥2 years vs. 10 % ≥2 years), less likely to switch BTA (overall: 6 vs. 14 %; subgroup: 8 vs. 19 %), and more compliant with treatment (overall: median doses of 7 vs. 4; subgroup: 11 vs. 8) compared to IV BP patients (N = 1,975). Findings were consistent across gender, age, tumor type, naïve, and transition strata.

CONCLUSIONS

Patients receiving denosumab and IV BPs may differ. Despite higher age and more advanced disease, patients treated with denosumab are more likely to stay on treatment and have better compliance.

Integrating patient reported outcomes with clinical cancer registry data: a feasibility study of the electronic Patient-Reported Outcomes From Cancer Survivors (ePOCS) system

Background

Routine measurement of Patient Reported Outcomes (PROs) linked with clinical data across the patient pathway is increasingly important for informing future care planning. The innovative electronic Patient-reported Outcomes from Cancer Survivors (ePOCS) system was developed to integrate PROs, collected online at specified post-diagnostic time-points, with clinical and treatment data in cancer registries.

Objective

This study tested the technical and clinical feasibility of ePOCS by running the system with a sample of potentially curable breast, colorectal, and prostate cancer patients in their first 15 months post diagnosis.

Methods

Patients completed questionnaires comprising multiple Patient Reported Outcome Measures (PROMs) via ePOCS within 6 months (T1), and at 9 (T2) and 15 (T3) months, post diagnosis. Feasibility outcomes included system informatics performance, patient recruitment, retention, representativeness and questionnaire completion (response rate), patient feedback, and administration burden involved in running the system.

Results

ePOCS ran efficiently with few technical problems. Patient participation was 55.21% (636/1152) overall, although varied by approach mode, and was considerably higher among patients approached face-to-face (61.4%, 490/798) than by telephone (48.8%, 21/43) or letter (41.0%, 125/305). Older and less affluent patients were less likely to join (both P<.001). Most non-consenters (71.1%, 234/329) cited information technology reasons (ie, difficulty using a computer). Questionnaires were fully or partially completed by 85.1% (541/636) of invited participants at T1 (80 questions total), 70.0% (442/631) at T2 (102-108 questions), and 66.3% (414/624) at T3 (148-154 questions), and fully completed at all three time-points by 57.6% (344/597) of participants. Reminders (mainly via email) effectively prompted responses. The PROs were successfully linked with cancer registry data for 100% of patients (N=636). Participant feedback was encouraging and positive, with most patients reporting that they found ePOCS easy to use and that, if asked, they would continue using the system long-term (86.2%, 361/419). ePOCS was not administratively burdensome to run day-to-day, and patient-initiated inquiries averaged just 11 inquiries per month.

Conclusions

The informatics underlying the ePOCS system demonstrated successful proof-of-concept – the system successfully linked PROs with registry data for 100% of the patients. The majority of patients were keen to engage. Participation rates are likely to improve as the Internet becomes more universally adopted. ePOCS can help overcome the challenges of routinely collecting PROs and linking with clinical data, which is integral for treatment and supportive care planning and for targeting service provision.

The PSA tracker: a computerised health care system initiative in Northern Ireland

INTRODUCTION

[corrected] The follow-up of men with prostate cancer forms a large part of many urologists workload. However, a rising PSA usually announces disease progression long before any clinically apparent symptom. Thus, many men can be safely monitored with PSA measurement alone. To facilitate this process, PSA tracking software was introduced to remotely monitor PSA results, minimising the work required for follow-up.

METHODS

Stable prostate cancer patients were into the PSA tracker. When each PSA test was performed, the result was reviewed. The program automatically generated patient reminder letters, summary reports for clinic use and all correspondence to patients and primary care physicians.

RESULTS

Since 2006, 65 patients have been entered into the PSA tracker. Median age was 81 (57-94) years. 274 outpatient appointments have been saved, indicating a potential saving of £32,000. More importantly it increased the capacity of the department to assess new patients. For the individual patient, the system has saved them, a median of 3 appointments each.

CONCLUSION

Remote follow-up of prostate cancer is associated with significant savings for both healthcare organisations and individual patients. This example, further demonstrates the benefits of implanting healthcare software for patients and hospitals.

Caveats for the use of operational electronic health record data in comparative effectiveness research

The growing amount of data in operational electronic health record systems provides unprecedented opportunity for its reuse for many tasks, including comparative effectiveness research. However, there are many caveats to the use of such data. Electronic health record data from clinical settings may be inaccurate, incomplete, transformed in ways that undermine their meaning, unrecoverable for research, of unknown provenance, of insufficient granularity, and incompatible with research protocols. However, the quantity and real-world nature of these data provide impetus for their use, and we develop a list of caveats to inform would-be users of such data as well as provide an informatics roadmap that aims to insure this opportunity to augment comparative effectiveness research can be best leveraged.

Extracting data from electronic medical records: validation of a natural language processing program to assess prostate biopsy results

OBJECTIVE

The extraction of specific data from electronic medical records (EMR) remains tedious and is often performed manually. Natural language processing (NLP) programs have been developed to identify and extract information within clinical narrative text. We performed a study to assess the validity of an NLP program to accurately identify patients with prostate cancer and to retrieve pertinent pathologic information from their EMR.

MATERIALS AND METHODS

A retrospective review was performed of a prospectively collected database including patients from the Southern California Kaiser Permanente Medical Region that underwent prostate biopsies during a 2-week period. A NLP program was used to identify patients with prostate biopsies that were positive for prostatic adenocarcinoma from all pathology reports within this period. The application then processed 100 consecutive patients with prostate adenocarcinoma to extract 10 variables from their pathology reports. The extraction and retrieval of information by NLP was then compared to a blinded manual review.

RESULTS

A consecutive series of 18,453 pathology reports were evaluated. NLP correctly detected 117 out of 118 patients (99.1%) with prostatic adenocarcinoma after TRUS-guided prostate biopsy. NLP had a positive predictive value of 99.1% with a 99.1% sensitivity and a 99.9% specificity to correctly identify patients with prostatic adenocarcinoma after biopsy. The overall ability of the NLP application to accurately extract variables from the pathology reports was 97.6%.

CONCLUSIONS

Natural language processing is a reliable and accurate method to identify select patients and to extract relevant data from an existing EMR in order to establish a prospective clinical database.

Completeness of bladder cancer staging in the Danish Cancer Registry, 2004-2009

Objective

To investigate the completeness of tumor, node, and metastasis (TNM) staging for invasive bladder cancer in the Danish Cancer Registry (DCR).

Methods

From the DCR, we retrieved data on TNM stage, year of diagnosis, sex, and age of all-incident invasive bladder cancer patients between 2004 and 2009. Data on comorbidity was obtained from the Danish National Patient Register. We estimated the completeness of TNM registration in the DCR overall and stratified the analysis by sex, age, year of cancer diagnosis, and Charlson comorbidity score. Through knowledge of pathophysiology and clinical coding practice, we designed a clinically based algorithm that allowed tumors with certain missing TNM-stage components to be placed in localized, regional, distant, and unknown categories.

Results

The overall completeness of TNM staging for bladder cancer was 44.1% (95% confidence interval [CI]: 42.7–45.5). Completeness decreased from 60.9% (95% CI: 40.6–78.6) in patients aged 0–39 years to 25.5% (95% CI: 23.2–27.9) in patients aged 80 years or older. Among patients with a low level of comorbidity, completeness was 48.4% (95% CI: 46.6–50.3), decreasing to 34.0% (95% CI: 30.4–37.8) among those with a high level of comorbidity. The highest proportion of missing TNM data was found for registration of lymph node metastases. Defining T1 cancer as completely registered, regardless of missing N and M stage, increased TNM-registration completeness to 61.8%. When we applied a clinically based algorithm, only 29.6% of tumors had an unknown stage.

Conclusion

The overall completeness of TNM staging for bladder cancer in the DCR was low, especially with increasing age and high level of comorbidity. Thus, restricting analyses to bladder cancer patients with complete data on stage may produce substantially selected study populations. Careful considerations should thus be made on handling missing data.

The use of regional platforms for managing electronic health records for the production of regional public health indicators in France

Background

In France, recent developments in healthcare system organization have aimed at strengthening decision-making and action in public health at the regional level. Firstly, the 2004 Public Health Act, by setting 100 national and regional public health targets, introduced an evaluative approach to public health programs at the national and regional levels. Meanwhile, the implementation of regional platforms for managing electronic health records (EHRs) has also been under assessment to coordinate the deployment of this important instrument of care within each geographic area. In this context, the development and implementation of a regional approach to epidemiological data extracted from EHRs are an opportunity that must be seized as soon as possible. Our article addresses certain design and organizational aspects so that the technical requirements for such use are integrated into regional platforms in France. The article will base itself on organization of the Rhône-Alpes regional health platform.

Discussion

Different tools being deployed in France allow us to consider the potential of these regional platforms for epidemiology and public health (implementation of a national health identification number and a national information system interoperability framework). The deployment of the Rhône-Alpes regional health platform began in the 2000s in France. By August 2011, 2.6 million patients were identified in this platform. A new development step is emerging because regional decision-makers need to measure healthcare efficiency. To pool heterogeneous information contained in various independent databases, the format, norm and content of the metadata have been defined. Two types of databases will be created according to the nature of the data processed, one for extracting structured data, and the second for extracting non-structured and de-identified free-text documents.

Summary

Regional platforms for managing EHRs could constitute an important data source for epidemiological surveillance in the context of epidemic alerts, but also in monitoring a number of indicators of infectious and chronic diseases for which no data are yet available in France.