Structured Approach for Evaluating Strategies for Cancer Ascertainment Using Large-Scale Electronic Health Record Data

Impact of Helicobacter pylori Infection and Treatment on Colorectal Cancer in a Large, Nationwide Cohort

PURPOSE

Helicobacter pylori is the most common cause of infection-associated cancer worldwide. We aimed to evaluate the impact of H. pylori infection and treatment on colorectal cancer (CRC) incidence and mortality.

PATIENTS

US Veterans who completed H. pylori testing between 1999 and 2018.

METHODS

We conducted a retrospective cohort analysis among adults within the Veterans Health Administration who completed testing for H. pylori. The primary exposures were (1) H. pylori test result (positive/negative) and (2) H. pylori treatment (untreated/treated) among H. pylori-positive individuals. The primary outcomes were CRC incidence and mortality. Follow-up started at the first H. pylori testing and continued until the earliest of incident or fatal CRC, non-CRC death, or December 31, 2019.

RESULTS

Among 812,736 individuals tested for H. pylori, 205,178 (25.2%) tested positive. Being H. pylori-positive versus H. pylori-negative was associated with higher CRC incidence and mortality. H. pylori treatment versus no treatment was associated with lower CRC incidence and mortality (absolute risk reduction 0.23%-0.35%) through 15-year follow-up. Being H. pylori-positive versus H. pylori-negative was associated with an 18% (adjusted hazard ratio [adjusted HR], 1.18 [95% CI, 1.12 to 1.24]) and 12% (adjusted HR, 1.12 [95% CI, 1.03 to 1.21]) higher incident and fatal CRC risk, respectively. Individuals with untreated versus treated H. pylori infection had 23% (adjusted HR, 1.23 [95% CI, 1.13 to 1.34]) and 40% (adjusted HR, 1.40 [95% CI, 1.24 to 1.58]) higher incident and fatal CRC risk, respectively. The results were more pronounced in the analysis restricted to individuals with nonserologic testing.

CONCLUSION

H. pylori positivity may be associated with small but statistically significant higher CRC incidence and mortality; untreated individuals, especially those with confirmed active infection, appear to be most at risk.

Risk of Esophageal Cancer in Achalasia: A Matched Cohort Study Using the Nationwide Veterans Affairs Achalasia Cohort

INTRODUCTION

Achalasia is a postulated risk factor of esophageal cancer (EC); however, EC-associated risk in achalasia is understudied. We aimed to evaluate EC risk among individuals within the nationwide Veterans Affairs Achalasia Cohort.

METHODS

We conducted a matched cohort study among US veterans aged 18 years or older from 1999 to 2019. Individuals with achalasia were age matched and sex matched 1:4 to individuals without achalasia. Follow-up continued from study entry until diagnosis with incident/fatal EC (primary outcome), death from non-EC-related causes, or end of the study follow-up (December 31, 2019). Association between achalasia and EC risk was examined using Cox regression models.

RESULTS

We included 9,315 individuals in the analytic cohort (median age 55 years; 92% male): 1,863 with achalasia matched to 7,452 without achalasia. During a median 5.5 years of follow-up, 17 EC occurred (3 esophageal adenocarcinoma, 12 squamous cell carcinoma, and 2 unknown type) among individuals with achalasia, compared with 15 EC (11 esophageal adenocarcinoma, 1 squamous cell carcinoma, and 3 unknown type) among those without achalasia. EC incidence for those with achalasia was 1.4 per 1,000 person-years, and the median time from achalasia diagnosis to EC development was 3.0 years (Q1-Q3: 1.3-9.1). Individuals with achalasia had higher cumulative EC incidence at 5, 10, and 15 years of follow-up compared with individuals without achalasia, and EC risk was 5-fold higher (hazard ratio 4.6, 95% confidence interval: 2.3-9.2).

DISCUSSION

Based on substantial EC risk, individuals with achalasia may benefit from a high index of suspicion and endoscopic surveillance for EC.

A deep learning algorithm to predict risk of pancreatic cancer from disease trajectories

Pancreatic cancer is an aggressive disease that typically presents late with poor outcomes, indicating a pronounced need for early detection. In this study, we applied artificial intelligence methods to clinical data from 6 million patients (24,000 pancreatic cancer cases) in Denmark (Danish National Patient Registry (DNPR)) and from 3 million patients (3,900 cases) in the United States (US Veterans Affairs (US-VA)). We trained machine learning models on the sequence of disease codes in clinical histories and tested prediction of cancer occurrence within incremental time windows (CancerRiskNet). For cancer occurrence within 36 months, the performance of the best DNPR model has area under the receiver operating characteristic (AUROC) curve = 0.88 and decreases to AUROC (3m) = 0.83 when disease events within 3 months before cancer diagnosis are excluded from training, with an estimated relative risk of 59 for 1,000 highest-risk patients older than age 50 years. Cross-application of the Danish model to US-VA data had lower performance (AUROC = 0.71), and retraining was needed to improve performance (AUROC = 0.78, AUROC (3m) = 0.76). These results improve the ability to design realistic surveillance programs for patients at elevated risk, potentially benefiting lifespan and quality of life by early detection of this aggressive cancer.

Adenoma Detection Rate and Clinical Characteristics Influence Advanced Neoplasia Risk After Colorectal Polypectomy

BACKGROUND AND AIMS

Postpolypectomy risk stratification for subsequent metachronous advanced neoplasia (MAN) is imprecise and does not account for colonoscopist adenoma detection rate (ADR). Our aim was to assess association of ADR with MAN and create a prediction model for postpolypectomy risk stratification incorporating ADR and other factors.

METHODS

We conducted a retrospective cohort study of individuals with baseline polypectomy and subsequent surveillance colonoscopy from 2004 to 2016 within the U.S. Department of Veterans Affairs (VA). Clinical factors, polyp findings, and baseline colonoscopist ADR were considered for the model. Model performance (sensitivity, specificity, and area under the curve) for identifying individuals with MAN was compared with 2020 U.S. Multi-Society Task Force on Colorectal Cancer (USMSTF) surveillance recommendations.

RESULTS

A total of 30,897 individuals were randomly assigned 2:1 into independent model training and validation sets. Increasing age, male sex, diabetes, current smoking, adenoma number, polyp location, adenoma ≥10 mm or with tubulovillous/villous features, and decreasing colonoscopist ADR were independently associated with MAN. A range of 1.48- to 1.66-fold increased risk for MAN was observed for ADR in the lowest 3 quintiles (ADR <19.7%-39.3%) vs the highest quintile (ADR >47.0%). When the final model selected based on the training set was applied to the validation set, improved sensitivity and specificity over 2020 USMSTF risk stratification were achieved (P = .001), with an area under the curve of 0.62 (95% confidence interval, 0.60-0.64).

CONCLUSIONS

Colonoscopist ADR is associated with MAN. Combining clinical factors and ADR for risk stratification has potential to improve postpolypectomy risk stratification. Improving ADR is likely to improve postpolypectomy outcomes.

Performance of a rule-based semi-automated method to optimize chart abstraction for surveillance imaging among patients treated for non-small cell lung cancer

Background

We aim to develop and test performance of a semi-automated method (computerized query combined with manual review) for chart abstraction in the identification and characterization of surveillance radiology imaging for post-treatment non-small cell lung cancer patients.

Methods

A gold standard dataset consisting of 3011 radiology reports from 361 lung cancer patients treated at the Veterans Health Administration from 2008 to 2016 was manually created by an abstractor coding image type, image indication, and image findings. Computerized queries using a text search tool were performed to code reports. The primary endpoint of query performance was evaluated by sensitivity, positive predictive value (PPV), and F1 score. The secondary endpoint of efficiency compared semi-automated abstraction time to manual abstraction time using a separate dataset and the Wilcoxon rank-sum test.

Results

Query for image type demonstrated the highest sensitivity of 85%, PPV 95%, and F1 score 0.90. Query for image indication demonstrated sensitivity 72%, PPV 70%, and F1 score 0.71. The image findings queries ranged from sensitivity 75–85%, PPV 23–25%, and F1 score 0.36–0.37. Semi-automated abstraction with our best performing query (image type) improved abstraction times by 68% per patient compared to manual abstraction alone (from median 21.5 min (interquartile range 16.0) to 6.9 min (interquartile range 9.5), p < 0.005).

Conclusions

Semi-automated abstraction using the best performing query of image type improved abstraction efficiency while preserving data accuracy. The computerized query acts as a pre-processing tool for manual abstraction by restricting effort to relevant images. Determining image indication and findings requires the addition of manual review for a semi-automatic abstraction approach in order to ensure data accuracy.

Ten or More Cumulative Lifetime Adenomas Are Associated with Increased Risk for Advanced Neoplasia and Colorectal Cancer

BACKGROUND

Colorectal cancer (CRC) screening guidelines recommend frequent colonoscopies and consideration of genetic testing in individuals with ≥10 cumulative adenomas. However, it is unclear how these guidelines apply to routine practice.

AIMS

We estimated the proportion of participants found to have ≥10 cumulative adenomas in a screening population and described their outcomes of advanced neoplasia (AN), CRC, and extra-colonic malignancy.

METHODS

We performed a secondary analysis of VA CSP#380, which includes 3121 veterans aged 50-75 who were followed up to 10 years after screening colonoscopy. We calculated the cumulative risk of ≥10 cumulative adenomas by Kaplan-Meier method. We compared baseline risk factors in those with and without ≥10 cumulative adenomas as well as the risk for AN (adenoma ≥1 cm, villous adenoma or high-grade dysplasia, or CRC) and extra-colonic malignancy by multivariate logistic regression.

RESULTS

The cumulative risk of ≥10 cumulative adenomas over 10.5 years was 6.51% (95% CI 4.38%-9.62%). Age 60-69 or 70-75 at baseline colonoscopy was the only factors associated with the finding of ≥10 cumulative adenomas. Compared to those with 0-9 cumulative adenomas, participants with ≥10 cumulative adenomas were more likely to have had AN (OR 17.03; 95% CI 9.41-30.84), including CRC (OR 7.00; 95% CI 2.84-17.28), but not extra-colonic malignancies.

CONCLUSIONS

Approximately 6.5% of participants in this screening population were found to have ≥10 cumulative adenomas over 10.5 years, which was uncommon before age 60. These participants were found to have AN and CRC significantly more often compared to those with lower cumulative adenomas.

Socioeconomic Disparities in Pancreas Cancer Resection and Survival in the Veterans Health Administration

BACKGROUND

Disparities based on socioeconomic factors such as race, ethnicity, marital status, and insurance status are associated with pancreatic cancer resection, but these disparities are usually not observed for survival after resection. It is unknown if there are disparities when patients undergo their treatment in a non-fee-for-service, equal-access healthcare system such as the Veterans Health Administration (VHA).

METHODS

Patients having T1-T3 M0 pancreatic adenocarcinoma diagnosed between 2006 and 2017 were identified from the VHA Corporate Data Warehouse. Socioeconomic, demographic, and tumor variables associated with resection and survival were assessed.

RESULTS

In total, 2580 patients with early-stage pancreatic cancer were identified. The resection rate was 36.5%. Surgical resection was independently associated with younger age [odds ratio (OR) 0.94, p < 0.001], White race (OR 1.35, p = 0.028), married status (OR 1.85, p = 0.001), and employment status (retired vs. unemployed, OR 1.41, p = 0.008). There were no independent associations with Hispanic ethnicity, geographic region, or Social Deprivation Index. Resection was associated with significantly improved survival (median 21 vs. 8 months, p = 0.001). Among resected patients, survival was independently associated with younger age (HR 1.019, p = 0.002), geographic region (South vs. Pacific West, HR 0.721, p = 0.005), and employment (employed vs. unemployed, HR 0.752, p = 0.029). Race, Hispanic ethnicity, marital status, and Social Deprivation Index were not independently associated with survival after resection.

CONCLUSIONS

Race, marital status, and employment status are independently associated with resection of pancreatic cancer in the VHA, whereas geographic region and employment status are independently associated with survival after resection. Further studies are warranted to determine the basis for these inequities.

Ascertainment of Veterans With Metastatic Prostate Cancer in Electronic Health Records: Demonstrating the Case for Natural Language Processing

PURPOSE

Prostate cancer (PCa) is among the leading causes of cancer deaths. While localized PCa has a 5-year survival rate approaching 100%, this rate drops to 31% for metastatic prostate cancer (mPCa). Thus, timely identification of mPCa is a crucial step toward measuring and improving access to innovations that reduce PCa mortality. Yet, methods to identify patients diagnosed with mPCa remain elusive. Cancer registries provide detailed data at diagnosis but are not updated throughout treatment. This study reports on the development and validation of a natural language processing (NLP) algorithm deployed on oncology, urology, and radiology clinical notes to identify patients with a diagnosis or history of mPCa in the Department of Veterans Affairs.

PATIENTS AND METHODS

Using a broad set of diagnosis and histology codes, the Veterans Affairs Corporate Data Warehouse was queried to identify all Veterans with PCa. An NLP algorithm was developed to identify patients with any history or progression of mPCa. The NLP algorithm was prototyped and developed iteratively using patient notes, grouped into development, training, and validation subsets.

RESULTS

A total of 1,144,610 Veterans were diagnosed with PCa between January 2000 and October 2020, among which 76,082 (6.6%) were identified by NLP as having mPCa at some point during their care. The NLP system performed with a specificity of 0.979 and sensitivity of 0.919.

CONCLUSION

Clinical documentation of mPCa is highly reliable. NLP can be leveraged to improve PCa data. When compared to other methods, NLP identified a significantly greater number of patients. NLP can be used to augment cancer registry data, facilitate research inquiries, and identify patients who may benefit from innovations in mPCa treatment.

Data Matching to Support Analysis of Cancer Epidemiology Among Veterans Compared With Non-Veteran Populations-An Exemplar in Brain Tumors

PURPOSE

State and national cancer registries do not systematically include Veteran data, which hinders analysis of the diagnosis patterns, treatment trajectories, and clinical outcomes of Veterans compared with non-Veteran populations. This study used data matching approaches to compare cases included in the Oncology Domain of the Veterans Affairs (VA) Corporate Data Warehouse and the Ohio Cancer Incidence Surveillance System, using brain tumors as an exemplar.

METHODS

We used direct data matching, on the basis of protected health information (PHI) common to both databases, to compare primary brain tumors from Veterans and non-Veterans diagnosed from 2000 to 2016. Working with this matched data set, we used six data elements that did not contain PHI, to assess the feasibility of using deterministic data matching to compare Veterans and non-Veterans.

RESULTS

Between 2000 and 2016, 223 Veterans from Ohio had a primary brain tumor; of those, 55 (25%) were not included in Ohio Cancer Incidence Surveillance System. Direct data matching showed that Veterans experienced a greater proportion of glioblastomas (41%) compared with non-Veterans (21%). Sex did not account for this difference. Deterministic data matching within the matched data set found that 75% (126 of 168) of Veterans had exact matches for at least five of six non-PHI variables common to both databases.

CONCLUSION

This study indicated that direct and deterministic data matching approaches to compare brain tumors in Veterans and in non-Veterans is feasible. This approach has the potential to promote comparisons of the distribution of tumors, the impact of chemical and environmental exposures, treatment trajectories, and clinical outcomes among Veteran and non-Veteran populations with brain tumors as well as other cancers and rare diseases.

Use and Outcomes of SBRT for Early Stage NSCLC Without Pathologic Confirmation in the Veterans Health Care Administration

Purpose

Stereotactic body radiation therapy (SBRT) use has increased among patients without pathologic confirmation (PC) of lung cancer. Empirical SBRT without PC raises concerns about variation in workup and patient selection, but national trends have not been well described. In this study, we assessed patterns of empirical SBRT use, workup, and causes of death among a large national non-small cell lung cancer (NSCLC) cohort.

Methods and Materials

We identified 2221 patients treated with SBRT for cT1-T2aN0M0 NSCLC in the Veterans Affairs health care system from 2008 to 2015. We reviewed their pretreatment workup and assessed associations between absence of PC and clinical and demographic factors. We compared causes of death between PC and non-PC groups and used Cox proportional hazards modeling to compare overall survival and lung cancer specific survival (LCSS) between these groups.

Results

Treatment without PC varied from 0% to 61% among Veterans Affairs medical centers, with at least 5 cases of stage I NSCLC. Overall, 14.9% of patients were treated without PC and 8.8% did not have a biopsy attempt. Ten percent of facilities were responsible for almost two-thirds (62%) of cases of treatment without PC. Of non-PC patients, 95.5% had positron emission tomography scans, 40.6% had biopsy procedures attempted, and 12.7% underwent endobronchial ultrasound. Non-PC patients were more likely to have cT1 tumors and live outside the histoplasmosis belt. Age, sex, smoking status, and Charlson comorbidity index were similar between groups. Lung cancer was the most common cause of death in both groups. Overall survival was similar between groups, whereas non-PC patients had better LCSS (hazard ratio = 0.77, P = .031).

Conclusions

Empirical SBRT use varied widely among institutions and appropriate radiographic workup was consistently used in this national cohort. Future studies should investigate determinants of variation and reasons for higher LCSS among non-PC patients.

A strategy for validation of variables derived from large-scale electronic health record data

Purpose:

Standardized approaches for rigorous validation of phenotyping from large-scale electronic health record (EHR) data have not been widely reported. We proposed a methodologically rigorous and efficient approach to guide such validation, including strategies for sampling cases and controls, determining sample sizes, estimating algorithm performance, and terminating the validation process, hereafter referred to as the San Diego Approach to Variable Validation (SDAVV).

Methods:

We propose sample size formulae which should be used prior to chart review, based on pre-specified critical lower bounds for positive predictive value (PPV) and negative predictive value (NPV). We also propose a stepwise strategy for iterative algorithm development/validation cycles, updating sample sizes for data abstraction until both PPV and NPV achieve target performance.

Results:

We applied the SDAVV to a Department of Veterans Affairs study in which we created two phenotyping algorithms, one for distinguishing normal colonoscopy cases from abnormal colonoscopy controls and one for identifying aspirin exposure. Estimated PPV and NPV both reached 0.970 with a 95% confidence lower bound of 0.915, estimated sensitivity was 0.963 and specificity was 0.975 for identifying normal colonoscopy cases. The phenotyping algorithm for identifying aspirin exposure reached a PPV of 0.990 (a 95% lower bound of 0.950), an NPV of 0.980 (a 95% lower bound of 0.930), and sensitivity and specificity were 0.960 and 1.000.

Conclusions:

A structured approach for prospectively developing and validating phenotyping algorithms from large-scale EHR data can be successfully implemented, and should be considered to improve the quality of “big data” research.

Time to Colonoscopy After Abnormal Stool-Based Screening and Risk for Colorectal Cancer Incidence and Mortality

BACKGROUND AND AIMS

The optimal time interval for diagnostic colonoscopy completion after an abnormal stool-based colorectal cancer (CRC) screening test is uncertain. We examined the association between time to colonoscopy and CRC outcomes among individuals who underwent diagnostic colonoscopy after abnormal stool-based screening.

METHODS

We performed a retrospective cohort study of veterans age 50 to 75 years with an abnormal fecal occult blood test (FOBT) or fecal immunochemical test (FIT) between 1999 and 2010. We used multivariable Cox proportional hazards to generate CRC-specific incidence and mortality hazard ratios (HRs) and 95% confidence intervals (CI) for 3-month colonoscopy intervals, with 1 to 3 months as the reference group. Association of time to colonoscopy with late-stage CRC diagnosis was also examined.

RESULTS

Our cohort included 204,733 patients. Mean age was 61 years (SD 6.9). Compared with patients who received a colonoscopy at 1 to 3 months, there was an increased CRC risk for patients who received a colonoscopy at 13 to 15 months (HR 1.13; 95% CI 1.00-1.27), 16 to 18 months (HR 1.25; 95% CI 1.10-1.43), 19 to 21 months (HR 1.28; 95% CI: 1.11-1.48), and 22 to 24 months (HR 1.26; 95% CI 1.07-1.47). Compared with patients who received a colonoscopy at 1 to 3 months, mortality risk was higher in groups who received a colonoscopy at 19 to 21 months (HR 1.52; 95% CI 1.51-1.99) and 22 to 24 months (HR 1.39; 95% CI 1.03-1.88). Odds for late-stage CRC increased at 16 months.

CONCLUSIONS

Increased time to colonoscopy is associated with higher risk of CRC incidence, death, and late-stage CRC after abnormal FIT/FOBT. Interventions to improve CRC outcomes should emphasize diagnostic follow-up within 1 year of an abnormal FIT/FOBT result.

Trends of lymphoma incidence in US veterans with rheumatoid arthritis, 2002-2017

OBJECTIVE

Past epidemiological studies have consistently demonstrated a link between rheumatoid arthritis (RA) and the incidence of lymphoma and it has been posited that high systemic inflammatory activity is a major risk determinant of lymphomagenesis. Given advances in the therapeutic armamentarium for RA management in recent years, the resulting lower level of disease activity could have led to a decline in lymphoma incidence in patients with RA. This study examined recent trends in lymphoma incidence in US veterans with RA.

METHODS

Patients with RA were identified in the Veterans Affairs (VA) Corporate Data Warehouse. Lymphoma incidence was identified through the end of 2018 from the VA Central Cancer Registry and compared among patients diagnosed during 2003-2005, 2006-2008, 2009-2011 and 2012-2014.

RESULTS

Among persons diagnosed with RA during 2003-2005, the incidence of lymphoma in the next 6 years was 2.0 per 1000 person-years. There was a steady decline in lymphoma incidence during the corresponding 6 years following diagnosis in the subsequent three cohorts, with a rate of 1.5 per 1000 person-years in the 2012-2014 cohort (incidence relative to that in the 2003-2005 cohort=0.79 (95% CI 0.58 to 1.1)). There was no similar decline in lymphoma incidence in VA patients diagnosed with osteoarthritis.

CONCLUSION

We observed a decline in lymphoma incidence in recent years among American veterans with RA. Further studies are needed to evaluate the specific factors driving this decline.

Risk Factors for Early-Onset Colorectal Cancer

BACKGROUND & AIMS

Colorectal cancer (CRC) incidence and mortality are increasing among persons younger than 50 years old in the United States, but risk factors associated with early-onset CRC (EOCRC) have not been widely studied.

METHODS

We conducted a case-control study of US veterans 18 to 49 years old who underwent colonoscopy examinations from 1999 through 2014. EOCRC cases were identified from a national cancer registry; veterans who were free of CRC at their baseline colonoscopy through 3 years of follow-up were identified as controls. We collected data on age, sex, race/ethnicity, body weight, body mass index (BMI), diabetes, smoking status, and aspirin use. Multivariate-adjusted EOCRC odds were estimated for each factor, with corresponding 95% confidence interval (CI) values.

RESULTS

Our final analysis included 651 EOCRC cases and 67,416 controls. Median age was 45.3 years, and 82.3% were male. Higher proportions of cases were older, male, current smokers, nonaspirin users, and had lower BMIs, compared with controls (P < .05). In adjusted analyses, increasing age and male sex were significantly associated with increased risk of EOCRC, whereas aspirin use and being overweight or obese (relative to normal BMI) were significantly associated with decreased odds of EOCRC. In post hoc analyses, weight loss of 5 kg or more within the 5-year period preceding colonoscopy was associated with higher odds of EOCRC (odds ratio 2.23; 95% CI 1.76-2.83).

CONCLUSIONS

In a case-control study of veterans, we found increasing age and male sex to be significantly associated with increased risk of EOCRC, and aspirin use to be significantly associated with decreased risk; these factors also affect risk for CRC onset after age 50. Weight loss may be an early clinical sign of EOCRC. More intense efforts are required to identify the factors that cause EOCRC and signs that can be used to identify individuals at highest risk.

Metformin Is Associated With Reduced Odds for Colorectal Cancer Among Persons With Diabetes

INTRODUCTION

Metformin may be associated with reduced colorectal cancer (CRC) risk, but findings from previous studies have been inconsistent and had insufficient sample sizes to examine whether the association differs by anatomic site. This study examined whether metformin was associated with reduced CRC risk, both overall and stratified by anatomic site, in a large sample of persons with diabetes who underwent colonoscopy.

METHODS

We performed a case-control study of US Veterans with prevalent diabetes who underwent colonoscopy between 1999 and 2014 using Department of Veterans Affairs electronic health record data. Cases were defined by presence of CRC at colonoscopy, while controls had normal colonoscopy. The primary exposure was metformin use at time of colonoscopy (yes/no). Association of metformin exposure with CRC (further stratified by proximal, distal, or rectal subsite) was examined using multivariable and multinomial logistic regression and summarized by odds ratios (ORs) with 95% confidence intervals (CIs).

RESULTS

We included 6,650 CRC patients and 454,507 normal colonoscopy patients. CRC cases were older and had lower metformin exposure. Metformin was associated with 8% relative reduction in CRC odds (OR: 0.92, 95% CI: 0.87-0.96). By subsite, metformin was associated with a 14% statistically significant reduced rectal cancer odds (OR: 0.86, 95% CI: 0.78-0.94) but no reduced distal or proximal cancer odds.

DISCUSSION

Metformin was associated with reduced CRC odds-particularly rectal cancer-in a large sample of persons with diabetes undergoing colonoscopy.

Development and evaluation of a computable phenotype to identify pediatric patients with leukemia and lymphoma treated with chemotherapy using electronic health record data

BACKGROUND

Widespread implementation of electronic health records (EHR) has created new opportunities for pediatric oncology observational research. Little attention has been given to using EHR data to identify patients with pediatric hematologic malignancies.

METHODS

This study used EHR-derived data in a pediatric clinical data research network, PEDSnet, to develop and evaluate a computable phenotype algorithm to identify pediatric patients with leukemia and lymphoma who received treatment with chemotherapy. To guide early development, multiple computable phenotype-defined cohorts were compared to one institution's tumor registry. The most promising algorithm was chosen for formal evaluation and consisted of at least two leukemia/lymphoma diagnoses (Systematized Nomenclature of Medicine codes) within a 90-day period, two chemotherapy exposures, and three hematology-oncology provider encounters. During evaluation, the computable phenotype was executed against EHR data from 2011 to 2016 at three large institutions. Classification accuracy was assessed by masked medical record review with phenotype-identified patients compared to a control group with at least three hematology-oncology encounters.

RESULTS

The computable phenotype had sensitivity of 100% (confidence interval [CI] 99%, 100%), specificity of 99% (CI 99%, 100%), positive predictive value (PPV) and negative predictive value (NPV) of 100%, and C-statistic of 1 at the development institution. The computable phenotype performance was similar at the two test institutions with sensitivity of 100% (CI 99%, 100%), specificity of 99% (CI 99%, 100%), PPV of 96%, NPV of 100%, and C-statistic of 0.99.

CONCLUSION

The EHR-based computable phenotype is an accurate cohort identification tool for pediatric patients with leukemia and lymphoma who have been treated with chemotherapy and is ready for use in clinical studies.

Risk factors for colorectal cancer significantly vary by anatomic site

Objective

To conduct an anatomic site-specific case–control study of candidate colorectal cancer (CRC) risk factors.

Design

Case–control study of US veterans with >1 colonoscopy during 1999–2011. Cases had cancer registry-identified CRC at colonoscopy, while controls were CRC free at colonoscopy and within 3 years of colonoscopy. Primary outcome was CRC, stratified by anatomic site: proximal, distal, or rectal. Candidate risk factors included age, sex, race/ethnicity, body mass index, height, diabetes, smoking status, and aspirin exposure summarised by adjusted ORs and 95% CIs.

Results

21 744 CRC cases (n=7017 rectal; n=7039 distal; n=7688 proximal) and 612 646 controls were included. Males had significantly higher odds relative to females for rectal cancer (OR=2.84, 95% CI 2.25 to 3.58) than distal cancer (OR=1.84, 95% CI 1.50 to 2.24). Relative to whites, blacks had significantly lower rectal cancer odds (OR=0.88, 95% CI 0.82 to 0.95), but increased distal (OR=1.27, 95% CI 1.19 to 1.37) and proximal odds (OR=1.62, 95% CI 1.52 to 1.72). Diabetes prevalence was more strongly associated with proximal (OR=1.29, 95% CI 1.22 to 1.36) than distal (OR=1.15, 95% CI 1.08 to 1.22) or rectal cancer (OR=1.12, 95% CI 1.06 to 1.19). Current smoking was more strongly associated with rectal cancer (OR=1.81, 95% CI 1.68 to 1.95) than proximal cancer (OR=1.53, 95% CI 1.43 to 1.65) or distal cancer (OR=1.46, 95% CI 1.35 to 1.57) compared with never smoking. Aspirin use was significantly more strongly associated with reduced rectal cancer odds (OR=0.71, 95% CI 0.67 to 0.76) than distal (OR=0.85, 95% CI 0.81 to 0.90) or proximal (OR=0.91, 95% CI 0.86 to 0.95).

Conclusion

Candidate CRC risk factor associations vary significantly by anatomic site. Accounting for site may enable better insights into CRC pathogenesis and cancer control strategies.

A Framework for Leveraging "Big Data" to Advance Epidemiology and Improve Quality: Design of the VA Colonoscopy Collaborative

OBJECTIVE

To describe a framework for leveraging big data for research and quality improvement purposes and demonstrate implementation of the framework for design of the Department of Veterans Affairs (VA) Colonoscopy Collaborative.

METHODS

We propose that research utilizing large-scale electronic health records (EHRs) can be approached in a 4 step framework: 1) Identify data sources required to answer research question; 2) Determine whether variables are available as structured or free-text data; 3) Utilize a rigorous approach to refine variables and assess data quality; 4) Create the analytic dataset and perform analyses. We describe implementation of the framework as part of the VA Colonoscopy Collaborative, which aims to leverage big data to 1) prospectively measure and report colonoscopy quality and 2) develop and validate a risk prediction model for colorectal cancer (CRC) and high-risk polyps.

RESULTS

Examples of implementation of the 4 step framework are provided. To date, we have identified 2,337,171 Veterans who have undergone colonoscopy between 1999 and 2014. Median age was 62 years, and 4.6 percent (n = 106,860) were female. We estimated that 2.6 percent (n = 60,517) had CRC diagnosed at baseline. An additional 1 percent (n = 24,483) had a new ICD-9 code-based diagnosis of CRC on follow up.

CONCLUSION

We hope our framework may contribute to the dialogue on best practices to ensure high quality epidemiologic and quality improvement work. As a result of implementation of the framework, the VA Colonoscopy Collaborative holds great promise for 1) quantifying and providing novel understandings of colonoscopy outcomes, and 2) building a robust approach for nationwide VA colonoscopy quality reporting.