Early Stage Bladder Cancer: Do Pathology Reports Tell Us What We Need to Know?

Abstract

Objective To assess a large national sample of bladder cancer pathology reports to determine if they contained the components necessary for clinical decision making. Methods We examined a random sample of 507 bladder cancer pathology reports from the national Veterans Administration (VA) Corporate Data Warehouse to assess whether each included information on the four report components explicitly recommended by the College of American Pathologists’ protocol for the examination of such specimens: histology, grade, presence vs. absence of muscularis propria in the specimen, and microscopic extent. We then assessed variation in the proportion of reports lacking at least one component across VA facilities. Results 108 of 507 reports (21%) lacked at least one of the four components, with microscopic extent and presence vs. absence of muscularis propria in the specimen most commonly missing (each in 11% of reports). There was wide variation across facilities in the proportion of reports lacking at least one component, ranging from 0% to 80%. Conclusion One fifth of bladder cancer pathology reports lack information needed for clinical decision making. The wide variation in incomplete report rates across facilities implies that some facilities already have implemented best practices assuring complete reporting while others have room for improvement. Future work to better understand barriers and facilitators of complete reporting may lead to interventions that improve bladder cancer care.

Development of a Natural Language Processing Engine to Generate Bladder Cancer Pathology Data for Health Services Research

“…In our set of bladder pathology reports, only 20% included an explicitly stated tumor T-stage. 12 To derive a T-stage, we had to evaluate statements used by pathologists to describe depth of invasion. The language used in these statements was highly variable, which led to higher levels of text misinterpretation by the NLP engine.…”

“…NLP would likely be more accurate if pathologists were using more standardized language to describe depth of invasion. The College of American Pathologists suggests use of a synoptic report which includes standardized language, 20 but this has only rarely been adopted for bladder pathology reports within VA. 12 Wider use of standardized language could facilitate both NLP and clinical care in the future. Nevertheless and in spite of the variability in language, we were able to identify tumors with lamina propria invasion (T1) with a high PPV of 0.91 in the final cleaned data set, and our accuracy is comparable to that of a prior study that was focused on abstracting staging information from full text lung cancer pathology reports (0.68 in the current study versus 0.72 in the prior study).…”

“…Our goal was to sample a nationally representative set of bladder cancer pathology reports from this Corporate Data Warehouse. As described in a previous report focused on the content of these reports, 12 we accomplished this via a three step process. First, we identified patients who were diagnosed with bladder cancer between 2005 and 2011.…”

“…This gold standard was created via a rigorous annotation process, during which two human reviewers highlighted full text statements and categorized them according to their meaning as previously described. 12 In brief, we first developed an annotation schema, outlining categories of information and content to be abstracted from the reports. Next, two annotators (FRS, a urologic oncologist, and EAP, a urology chief resident) used a specialized annotation tool (ChartReview 15 ) to independently highlight and categorize statements in each report containing information on the following variables: histology, invasion (presence versus absence and depth), grade, and statements regarding presence of carcinoma in situ and of muscularis propria in the specimen.…”

Assembling and validating data from multiple sources to study care for Veterans with bladder cancer

“…Thus, our a priori expectation was that receipt of bladder cancer care in VA should be significantly more common among patients who have full text pathology reports than among those who have not (Table 1 ). As previously described, we identified full text pathology reports based on the report title indicating a pathology report and on presence of at least one of the three keywords “bladder”, “urethra”, or “ureter” within the full text [ 12 ]. We then used the chi-squared test to compare the proportion of patients receiving bladder cancer care among those who did versus who did not have pathology reports available.…”