Development of reusable logic for determination of statin exposure-time from electronic health records.

OBJECTIVE: We aim to quantify HMG-CoA reductase inhibitor (statin) prescriber-intended exposure-time using a generalizable algorithm that interrogates data stored in the electronic health record (EHR). MATERIALS AND METHODS: This study was conducted using the Marshfield Clinic (MC) Personalized Medicine Research Project (PMRP) a central Wisconsin-based population and biobank with, on average, 30 years of electronic health data available in the independently-developed MC Cattails MD EHR. Individuals with evidence of statin exposure were identified from the electronic records, and manual chart abstraction of all mentions of prescribed statins was completed. We then performed electronic chart abstraction of prescriber-intended exposure time for statins, using previously identified logic to capture pill-splitting events, normalizing dosages to atorvastatin-equivalent dose. Four models using iterative training sets were tested to capture statin end-dates. Calculated cumulative provider-intended exposures were compared to manually abstracted gold-standard measures of ordered statin prescriptions, and aggregate model results (totals) for training and validation populations were compared. The most successful model was the one with the smallest discordance between modeled and manually abstracted Atorvastatin 10mg/year Equivalents (AEs). RESULTS: Of the approximately 20,000 patients enrolled in the PMRP, 6243 were identified with statin exposure during the study period (1997-2011), 59.8% of whom had been prescribed multiple statins over an average of approximately 11 years. When the best-fit algorithm was implemented and validated by manual chart review for the statin-ordered population, it was found to capture 95.9% of the correlation between calculated and expected statin provider-intended exposure time for a random validation set, and the best-fit model was able to predict intended statin exposure to within a standard deviation of 2.6 AEs, with a standard error of +0.23 AEs. CONCLUSION: We demonstrate that normalized provider-intended statin exposure time can be estimated using a combination of structured clinical data sources, including a medications ordering system and a clinical appointment coordination system, supplemented with text data from clinical notes.

Underutilization of Lynch syndrome screening in a multisite study of patients with colorectal cancer.

PURPOSE: The aim of this study was to examine Lynch syndrome screening of patients with metastatic colorectal cancer in integrated health-care-delivery organizations. METHODS: We determined the availability of Lynch syndrome screening criteria and actual Lynch syndrome screening in the medical records of 1,188 patients diagnosed with metastatic colorectal cancer between 2004 and 2009 at seven institutions in the Cancer Research Network. RESULTS: We found infrequent use of Lynch syndrome screening (41/1,188). Family history was available for 937 of the 1,188 patients (79%). There was sufficient information to assess Lynch syndrome risk using family history-based criteria in 719 of the 937 patients (77%) with family history documentation. In 391 individuals with a family history of a Lynch syndrome-associated cancer, 107 (27%) could not be evaluated due to missing information such as age of cancer onset. Eleven percent of patients who met the Bethesda criteria and 25% of individuals who met the Amsterdam II criteria were screened for Lynch syndrome. Recommended guidelines were adhered to during screening, but no testing method was preferred. CONCLUSION: The information required for Lynch syndrome screening decisions is routinely collected but seldom used. There is a critical gap between collection of family history and its use to guide Lynch syndrome screening, which may support a case for implementation of universal screening guidelines.

Analysis of urological procedures in men who died from prostate cancer using a population-based approach.

UNLABELLED: What's known on the subject? and What does the study add? Very few studies have examined end-of-life urological studies in men with prostate cancer. These studies reported fewer procedures in men who received primary therapy for prostate cancer. However, these studies were typically single institution or had a short follow-up period. The present study is the first population-based study examining end-of-life urological procedures and uses a geographic region encompassing 385 000 patients. Furthermore, this study incorporates both hospital- and office-based procedures. This approach has not been previously undertaken. OBJECTIVE: To determine using a population-based approach whether men with end-stage prostate cancer who had definitive primary therapy might require fewer urological interventions. Repeated urological procedures can impact health-related quality of life in patients dying from prostate cancer. PATIENTS AND METHODS: Using the Marshfield Epidemiological Study Area (MESA) database and tumour registry, we compared end-of-life interventions in men who died from prostate cancer between 1991 and 2009. Patient charts were queried for urological procedures using International Classification of Disease Modification, 9th edition (ICD9) codes for 3 years before death. Clinicopathological information was examined including whether the patient had a history of primary therapy (radiation or radical prostatectomy). RESULTS: Among 280 patients dying from prostate cancer, 52 (19%) required 153 urological procedures during the last 3 years of life. The frequency of procedures increased closer to death. The most common procedures involved nephrostomy tube (56%), Foley catheter (24%) and transurethral resection of the prostate (10%). Clinicopathological features did not predict the need for an end-of-life urological procedure. There was no difference in the frequency of upper or lower tract procedures in surgery or radiation patients compared with patients without primary therapy (P = 0.556 and P = 0.508). Using a Kaplan-Meier analysis, there were no differences between groups in the proportion of patients not requiring a procedure (n = 280; P = 0.179). CONCLUSIONS: This is the first population-based study to examine the frequency of urological procedures in patients with end-stage prostate cancer. A minority of patients (19%) required urological procedures during the final 3 years of life. A history of surgery or radiation did not influence the overall risk for urological intervention.

Tamoxifen-induced venothromboembolic events: exploring validation of putative genetic association.

OBJECTIVE: A pilot study to examine accrual rates, efficiency of data capture approaches, study design and genotyping capacity for a future genetic validation study was undertaken. DESIGN: The process pilot evaluated feasibility of applying a matched case-control design to validate association of two candidate estrogen receptor (ER) single nucleotide polymorphisms (SNPs) with incidence of venothromboembolic events (VTE) in breast cancer patients treated with tamoxifen where criteria included frequency matching by age, number of years diagnosed with breast cancer within 4-year intervals, and geographic residency. SETTING: The study was conducted at Marshfield Clinic, in central Wisconsin. PARTICIPANTS: Study-eligible cases with a breast cancer diagnosis between 1994 and 2006 who experienced a VTE within 5 years of last tamoxifen exposure were matched at a ratio of 1:4 to controls with a breast cancer diagnosed between 1994 and 2006 with no VTE history following tamoxifen exposure for ≥2 years. METHODS: Feasibility of enrolling, phenotyping, and genotyping 20% of the total number of validated eligible cases and controls was tested in order to project enrollment rates and assess probability of enrolling the projected sample size for the prospective validation study and adequacy of planned data capture. Conditional logistic regression analysis was conducted for the matched case-control study design. RESULTS: Enrollment accruals included 19 of 24 targeted cases (79%), and 74 of 96 (77%) targeted controls. Electronic data capture for most variables was nearly 100%. No unexpected statistically significant differences were observed between cases and controls. Capacity to conduct in-house screening for rs2234689 (ER1 PvuII), rs9340799 (ER1 XbaI), rs13146272 (CYP4V2), rs2227589 (SERPINC 1) and rs1613662 (GP6) was successfully established. Association of GP6 with VTE was further validated (P=0.0403; OR, 0.19). CONCLUSION: Accrual rates to the larger prospective study will require a multi-center design to ensure enrollment of adequate numbers of cases and controls for achieving the projected sample size required to validate association of the ER SNPs. To prevent study failure due to poor accrual, the importance of conducting feasibility studies before launching large scale validation studies of genetic association and adverse drug events, is discussed.

Historical prostate cancer screening and treatment outcomes from a single institution.

OBJECTIVE: To quantify outcomes of individuals diagnosed and treated for prostate cancer in a single institution. DESIGN: Retrospective electronic chart abstraction. SETTING: Marshfield Clinic, the largest private multispecialty group practice in Wisconsin, and one of the largest in the United States, provides health care services annually to approximately 385,000 unique patients through 1.8 million annual patient encounters. PARTICIPANTS: Individuals within the Marshfield Clinic cancer registry who had been diagnosed with prostate cancer between 1960 and 2009. METHODS: Electronic chart abstraction from the cancer registry and the electronic medical record was conducted (N=6,181). Data abstracted included age at diagnosis; stage and grade of tumor; prostate specific antigen (PSA) values before, at, and after diagnosis; initial cancer treatment; follow-up time; subsequent cancer treatments; evidence of metastasis; age of death; and cause of death, if known. RESULTS: The average age of prostate cancer diagnosis has decreased from 70-71 years in the 1960's and 1970's to an average age at diagnosis of 67 years in the 2000's (P<0.001). This decrease in age occurred within the decades of implementation of PSA screening. Approximately 74% of men diagnosed with prostate cancer within the PSA screening era had at least one PSA test, and the presence of a PSA test did not appear to change treatment outcome. Age, grade, and stage were the biggest predictors of prostate cancer outcome. There was no difference in event-free survival between current treatment types (radical prostatectomy, brachytherapy, photon treatment, or intensity-modulated radiation therapy) (2003 or later) when stratified by age (greater than 85%, 5-year event-free survival P=0.85); however, more events occurred with older external beam radiation treatment regimens (1993-2003) (70% to 75%, 5-year event-free survival P=0.001). CONCLUSION: Individuals diagnosed and treated for prostate cancer within the Marshfield Clinic comprehensive care setting follow national trends with a decreased age of diagnosis since the advent of PSA screening. Outcomes for individuals treated within the Clinic system are also comparable to national trends.

Population based allele frequencies of disease associated polymorphisms in the Personalized Medicine Research Project.

BACKGROUND: There is a lack of knowledge regarding the frequency of disease associated polymorphisms in populations and population attributable risk for many populations remains unknown. Factors that could affect the association of the allele with disease, either positively or negatively, such as race, ethnicity, and gender, may not be possible to determine without population based allele frequencies.Here we used a panel of 51 polymorphisms previously associated with at least one disease and determined the allele frequencies within the entire Personalized Medicine Research Project population based cohort. We compared these allele frequencies to those in dbSNP and other data sources stratified by race. Differences in allele frequencies between self reported race, region of origin, and sex were determined. RESULTS: There were 19544 individuals who self reported a single racial category, 19027 or (97.4%) self reported white Caucasian, and 11205 (57.3%) individuals were female. Of the 11,208 (57%) individuals with an identifiable region of origin 8337 or (74.4%) were German.41 polymorphisms were significantly different between self reported race at the 0.05 level. Stratification of our Caucasian population by self reported region of origin revealed 19 polymorphisms that were significantly different (p = 0.05) between individuals of different origins. Further stratification of the population by gender revealed few significant differences in allele frequencies between the genders. CONCLUSIONS: This represents one of the largest population based allele frequency studies to date. Stratification by self reported race and region of origin revealed wide differences in allele frequencies not only by race but also by region of origin within a single racial group. We report allele frequencies for our Asian/Hmong and American Indian populations; these two minority groups are not typically selected for population allele frequency detection. Population wide allele frequencies are important for the design and implementation of studies and for determining the relevance of a disease associated polymorphism for a given population.

Differentiation of Hispanic biogeographic ancestry with 80 ancestry informative markers.

Ancestry informative single nucleotide polymorphisms (SNPs) can identify biogeographic ancestry (BGA); however, population substructure and relatively recent admixture can make differentiation difficult in heterogeneous Hispanic populations. Utilizing unrelated individuals from the Genomic Origins and Admixture in Latinos dataset (GOAL, n = 160), we designed an 80 SNP panel (Setser80) that accurately depicts BGA through STRUCTURE and PCA. We compared our Setser80 to the Seldin and Kidd panels via resampling simulations, which models data based on allele frequencies. We incorporated Admixed American 1000 Genomes populations (1000 G, n = 347), into a combined populations dataset to determine robustness. Using multinomial logistic regression (MLR), we compared the 3 panels on the combined dataset and found overall MLR classification accuracies: 93.2% Setser80, 87.9% Seldin panel, 71.4% Kidd panel. Naïve Bayesian classification had similar results on the combined dataset: 91.5% Setser80, 84.7% Seldin panel, 71.1% Kidd panel. Although Peru and Mexico were absent from panel design, we achieved high classification accuracy on the combined populations for Peru (MLR = 100%, naïve Bayes = 98%), and Mexico (MLR = 90%, naïve Bayes = 83.4%) as evidence of the portability of the Setser80. Our results indicate the Setser80 SNP panel can reliably classify BGA for individuals of presumed Hispanic origin.

Functional characterization of the HOXB13 promoter region.

Homeobox (HOX) genes are crucial regulators of cell growth and differentiation. These genes initiate and control gene expression cascades that drive development. More recently, the absent or aberrant expression of HOX genes has been implicated in cancer development. Despite the observance of these expression changes, the regulation of the HOX genes in adult tissues and how these genes become deregulated in cancerous tissues still needs much investigation. We characterized the promoter region of the HOXB13 gene. A 3 kb region upstream of the HOXB13 gene, which included the 5'UTR, increased reporter gene expression in LNCaP cells by approximately 99 fold over the promoterless control construct. A highly conserved 179 base pair fragment containing only the 5'UTR of the HOXB13 gene constituted a minimal promoter in the LNCaP cell line. Strong promoter activity was seen in the presence or absence of testosterone, although testosterone exposure did decrease expression in LNCaP cells by 50%. In an androgen insensitive cell line Du145, no sensitivity to testosterone was detected and a consistent low basal level of expression was observed. Since HOXB13 expression is highly tissue specific, we investigated the ability of the promoter to drive expression in tissues other than prostate. We observed highest expression in LNCaP cells with low levels of expression in lung, retinoblastoma, and colon cancer cells and higher expression in MCF7 breast cancer cells.

Development of a multi-institutional cohort to facilitate cardiovascular disease biomarker validation using existing biorepository samples linked to electronic health records.

BACKGROUND: Emerging biomarkers for acute myocardial infarction (AMI) may enhance conventional risk-prediction algorithms if they are informative and associated with risk independently of established predictors. In this study, we constructed a cohort for testing emerging biomarkers for AMI in managed-care populations using existing biospecimen repositories linked to electronic health records (EHR). HYPOTHESIS: Electronic health record-based biorepositories collected by healthcare systems can be federated to provide large, methodologically sound testing sets for biomarker validation. METHODS: Subjects ages 40 to 80 years were selected from 2 existing population-based biospecimen repositories. Incident AMI status and covariates were ascertained from the EHR. An ad hoc model for AMI risk was parameterized and validated. Simulation was used to test incremental gains in performance due to the inclusion of biomarkers in this model. Gains in performance were assessed in terms of area under the receiver operating characteristic curve (ROC-AUC) and case reclassification. RESULTS: A total of 18 329 individuals (57% female) contributed 108 400 person-years of EHR follow-up. The crude AMI incidence was 10.8 and 5.0 per 1000 person-years among males and females, respectively. Compared with the model with risk factors alone, inclusion of a simulated biomarker yielded substantial gains in sensitivity without loss of specificity. Furthermore, a net ROC-AUC gain of 13.3% was observed, as well as correct reclassification of 9.8% of incident cases (79 of 806) that were otherwise not considered statin-indicated at baseline under the National Cholesterol Education Program Adult Treatment Panel III criteria. CONCLUSIONS: More research is needed to assess incremental contribution of emerging biomarkers for AMI prediction in managed-care populations.

Development of a fingerprinting panel using medically relevant polymorphisms.

BACKGROUND: For population based biorepositories to be of use, rigorous quality control and assurance must be maintained. We have designed and validated a panel of polymorphisms for individual sample identification consisting of 36 common polymorphisms that have been implicated in a wide range of diseases and an additional sex marker. This panel uniquely identifies our biorepository of approximately 20,000 samples and would continue to uniquely identify samples in biorepositories of over 100 million samples. METHODS: A panel of polymorphisms associated with at least one disease state in multiple populations was constructed using a cut-off of 0.20 or greater confirmed minor allele frequency in a European Caucasian population. The fingerprinting assay was tested using the MALDI-TOF mass spectrometry method of allele determination on a Sequenom platform with a panel of 28 Caucasian HapMap samples; the results were compared with known genotypes to ensure accuracy. The frequencies of the alleles were compared to the expected frequencies from dbSNP and any genotype that did not achieve Hardy Weinberg equilibrium was excluded from the final assay. RESULTS: The final assay consisted of the AMG sex marker and 36 medically relevant polymorphisms with representation on each chromosome, encompassing polymorphisms on both the Illumina 550K bead array and the Affymetrix 6.0 chip (with over a million polymorphisms) platform. The validated assay has a P(ID) of 6.132 x 10(-15) and a Psib(ID) of 3.077 x 10(-8). This assay allows unique identification of our biorepository of 20,000 individuals as well and ensures that as we continue to recruit individuals they can be uniquely fingerprinted. In addition, diseases such as cancer, heart disease diabetes, obesity, and respiratory disease are well represented in the fingerprinting assay. CONCLUSION: The polymorphisms in this panel are currently represented on a number of common genotyping platforms making QA/QC flexible enough to accommodate a large number of studies. In addition, this panel can serve as a resource for investigators who are interested in the effects of disease in a population, particularly for common diseases.