Background Prior text analysis of R01 critiques suggested that female applicants may be disadvantaged in NIH peer review, particularly for R01 renewals. NIH altered its review format in 2009. The authors examined R01 critiques and scoring in the new format for differences due to principal investigator (PI) sex. Method The authors analyzed 739 critiques—268 from 88 unfunded and 471 from 153 funded applications for grants awarded to 125 PIs (M = 76, 61% F = 49, 39%) at the University of Wisconsin-Madison between 2010 and 2014. The authors used 7 word categories for text analysis: ability, achievement, agentic, negative evaluation, positive evaluation, research, and standout adjectives. The authors used regression models to compare priority and criteria scores, and results from text analysis for differences due to PI sex and whether the application was for a new (Type 1) or renewal (Type 2) R01. Results Approach scores predicted priority scores for all PIs’ applications (P<.001); but scores and critiques differed significantly for male and female PIs’ Type 2 applications. Reviewers assigned significantly worse priority, approach, and significance scores to female than male PIs’ Type 2 applications, despite using standout adjectives (e.g., “outstanding,” “excellent”) and making references to ability in more of their critiques (P<.05 for all comparisons). Conclusions The authors’ analyses suggest that subtle gender bias may continue to operate in the post-2009 NIH review format in ways that could lead reviewers to implicitly hold male and female applicants to different standards of evaluation, particularly for R01 renewals.

Background: Women are less successful than men in renewing R01 grants from the National Institutes of Health. Continuing to probe text mining as a tool to identify gender bias in peer review, we used algorithmic text mining and qualitative analysis to examine a sample of critiques from men's and women's R01 renewal applications previously analyzed by counting and comparing word categories. Methods: We analyzed 241 critiques from 79 Summary Statements for 51 R01 renewals awarded to 45 investigators (64% male, 89% white, 80% PhD) at the University of Wisconsin-Madison between 2010 and 2014. We used latent Dirichlet allocation to discover evaluative ''topics'' (i.e., words that co-occur with high probability). We then qualitatively examined the context in which evaluative words occurred for male and female investigators. We also examined sex differences in assigned scores controlling for investigator productivity. Results: Text analysis results showed that male investigators were described as ''leaders'' and ''pioneers'' in their ''fields,'' with ''highly innovative'' and ''highly significant research.'' By comparison, female investigators were characterized as having ''expertise'' and working in ''excellent'' environments. Applications from men received significantly better priority, approach, and significance scores, which could not be accounted for by differences in productivity. Conclusions: Results confirm our previous analyses suggesting that gender stereotypes operate in R01 grant peer review. Reviewers may more easily view male than female investigators as scientific leaders with significant and innovative research, and score their applications more competitively. Such implicit bias may contribute to sex differences in award rates for R01 renewals.

Purpose To compare overall colorectal cancer (CRC) screening rates for patients who were eligible and due for CRC screening and who were with and without insurance coverage for computed tomographic (CT) colonography for CRC screening. Materials and Methods The institutional review board approved this retrospective cohort study, with a waiver of consent. This study used longitudinal electronic health record data from 2005 through 2010 for patients managed by one of the largest multispecialty physician groups in the United States. It included 33 177 patients under age 65 who were eligible and due for CRC screening and managed by the participating health system. Stratified Cox regression models provided propensity-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for the relationship between CT colonography coverage and CRC screening. Results After adjustment, patients who had insurance coverage for CT colonography and were due for CRC screening had a 48% greater likelihood of being screened for CRC by any method compared with those without coverage who were due for CRC screening (HR, 1.48; 95% CI: 1.41, 1.55). Similarly, patients with CT colonography coverage had a greater likelihood of being screened with CT colonography (HR, 8.35; 95% CI: 7.11, 9.82) and with colonoscopy (HR, 1.38; 95% CI: 1.31, 1.45) but not with fecal occult blood test (HR, 1.00; 95% CI: 0.91, 1.10) than those without such insurance coverage. Conclusion Insurance coverage of CT colonography for CRC screening was associated with a greater likelihood of a patient being screened and a greater likelihood of being screened with a test that helps both to detect cancer and prevent cancer from developing (CT colonography or colonoscopy). RSNA, 2017.

Aims. Colorectal cancer (CRC) screening is underutilized. Increasing CRC screening rates requires interventions targeting multiple barriers at each level of the healthcare organization (patient, provider, and system). We examined groups of primary care providers (PCPs) based on perceptions of screening barriers and the relationship to CRC screening rates to inform approaches for conducting barrier assessments prior to designing and implementing quality improvement interventions. Methods. We conducted a retrospective cohort study linking EHR and survey data. PCPs with complete survey responses for questions addressing CRC screening barriers were included (N = 166 PCPs; 39,430 patients eligible for CRC screening). Cluster analysis identified groups of PCPs. Multivariate logistic regression estimated odds ratios and 95% confidence intervals for predictors of membership in one of the PCP groups. Results. We found two distinct groups: (1) PCPs identifying multiple barriers to CRC screening at patient, provider, and system levels (N = 75) and (2) PCPs identifying no major barriers to screening (N = 91). PCPs in the top half of CRC screening performance were more likely to identify multiple barriers than the bottom performers (OR, 4.14; 95% CI, 2.43–7.08). Conclusions. High-performing PCPs can more effectively identify CRC screening barriers. Targeting high-performers when conducting a barrier assessment is a novel approach to assist in designing quality improvement interventions for CRC screening.