Association of Physician Adenoma Detection Rates With Postcolonoscopy Colorectal Cancer.

Importance: Although colonoscopy is frequently performed in the United States, there is limited evidence to support threshold values for physician adenoma detection rate as a quality metric. Objective: To evaluate the association between physician adenoma detection rate values and risks of postcolonoscopy colorectal cancer and related deaths. Design, Setting, and Participants: Retrospective cohort study in 3 large integrated health care systems (Kaiser Permanente Northern California, Kaiser Permanente Southern California, and Kaiser Permanente Washington) with 43 endoscopy centers, 383 eligible physicians, and 735 396 patients aged 50 to 75 years who received a colonoscopy that did not detect cancer (negative colonoscopy) between January 2011 and June 2017, with patient follow-up through December 2017. Exposures: The adenoma detection rate of each patient's physician based on screening examinations in the calendar year prior to the patient's negative colonoscopy. Adenoma detection rate was defined as a continuous variable in statistical analyses and was also dichotomized as at or above vs below the median for descriptive analyses. Main Outcomes and Measures: The primary outcome (postcolonoscopy colorectal cancer) was tumor registry-verified colorectal adenocarcinoma diagnosed at least 6 months after any negative colonoscopy (all indications). The secondary outcomes included death from postcolonoscopy colorectal cancer. Results: Among 735 396 patients who had 852 624 negative colonoscopies, 440 352 (51.6%) were performed on female patients, median patient age was 61.4 years (IQR, 55.5-67.2 years), median follow-up per patient was 3.25 years (IQR, 1.56-5.01 years), and there were 619 postcolonoscopy colorectal cancers and 36 related deaths during more than 2.4 million person-years of follow-up. The patients of physicians with higher adenoma detection rates had significantly lower risks for postcolonoscopy colorectal cancer (hazard ratio [HR], 0.97 per 1% absolute adenoma detection rate increase [95% CI, 0.96-0.98]) and death from postcolonoscopy colorectal cancer (HR, 0.95 per 1% absolute adenoma detection rate increase [95% CI, 0.92-0.99]) across a broad range of adenoma detection rate values, with no interaction by sex (P value for interaction = .18). Compared with adenoma detection rates below the median of 28.3%, detection rates at or above the median were significantly associated with a lower risk of postcolonoscopy colorectal cancer (1.79 vs 3.10 cases per 10 000 person-years; absolute difference in 7-year risk, -12.2 per 10 000 negative colonoscopies [95% CI, -10.3 to -13.4]; HR, 0.61 [95% CI, 0.52-0.73]) and related deaths (0.05 vs 0.22 cases per 10 000 person-years; absolute difference in 7-year risk, -1.2 per 10 000 negative colonoscopies [95%, CI, -0.80 to -1.69]; HR, 0.26 [95% CI, 0.11-0.65]). Conclusions and Relevance: Within 3 large community-based settings, colonoscopies by physicians with higher adenoma detection rates were significantly associated with lower risks of postcolonoscopy colorectal cancer across a broad range of adenoma detection rate values. These findings may help inform recommended targets for colonoscopy quality measures.

Associations between molecular characteristics of colorectal serrated polyps and subsequent advanced colorectal neoplasia.

PURPOSE: BRAF mutation and DNA hypermethylation have linked sessile serrated adenomas/polyps (SSA/Ps) to serrated colorectal cancer (CRC) in cross-sectional studies, but they have not been evaluated in a longitudinal study. We aimed to evaluate the associations between molecular markers of serrated polyps and subsequent advanced colorectal neoplasia. METHODS: Study subjects included Kaiser Permanente Washington members aged 20-75 years who received an index colonoscopy between 1/1/1998 and 12/31/2007 and had hyperplastic polyps (HPs) or SSA/Ps according to study pathology review. Polyps from index colonoscopies were removed and assayed for BRAF mutation, CpG island methylator phenotype (CIMP), and MLH1 methylation. Pathology reports and biopsies from the subsequent lower gastrointestinal endoscopy through 1/1/2013 were reviewed for advanced colorectal neoplasia. We identified additional incident CRC cases through linkage to the Seattle-Puget Sound Surveillance Epidemiology and End Results registry. We used generalized estimating equations to calculate adjusted odds ratios (OR) and 95% confidence intervals (CI) for subsequent advanced colorectal neoplasia, comparing index serrated polyps with different molecular markers. RESULTS: We included 553 individuals with index serrated polyps (420 HPs and 133 SSA/Ps) and 795 subsequent endoscopies. The prevalence of BRAF-mutant, CIMP-high, and MLH1-methylated serrated polyps were 51%, 4%, and 2%, respectively. BRAF and CIMP were not associated with subsequent advanced colorectal neoplasia. MLH1-methylated SSP/As were significantly more likely to have subsequent advanced neoplasia (OR = 4.66, 95% CI 1.06-20.51). CONCLUSION: Our results suggest that BRAF-mutant and CIMP-high serrated polyps are not associated with subsequent advanced colorectal neoplasia. Among SSA/Ps, MLH1 methylation may be an important marker to identify high-risk CRC precursors.

The association between colorectal sessile serrated adenomas/polyps and subsequent advanced colorectal neoplasia.

PURPOSE: Colorectal cancer (CRC) screening guidelines recommend increased surveillance of individuals with sessile serrated adenomas/polyps (SSA/Ps), but there is uncertainty about the risk associated with SSA/Ps. We aimed to determine the association between SSA/Ps and subsequent advanced colorectal neoplasia. METHODS: This case-control study included Kaiser Permanente Washington (KPWA) members who received an index colonoscopy between 1/1/1998 and 12/31/2007, and had hyperplastic polyps (HPs) or SSA/Ps but no conventional adenomas according to study pathologist histologic review. Subsequent pathology reports and biopsies through 1/1/2013 were reviewed for advanced colorectal neoplasia. We linked to the Seattle-Puget Sound Surveillance Epidemiology and End Results (SEER) registry to identify additional CRC cases. We used generalized estimating equations with a logit link to estimate adjusted odds ratios (ORs) and 95% confidence intervals (CIs) for advanced colorectal neoplasia, comparing those with SSA/Ps to those with HPs. RESULTS: There were 161 individuals with index SSA/Ps, 548 with HPs, and 918 subsequent endoscopies included in analyses. Of those with index SSA/Ps, 19 had subsequent advanced colorectal neoplasia; 39 with HPs had subsequent advanced colorectal neoplasia. Compared to those with HPs, those with SSA/Ps were not statistically significantly more likely to have subsequent advanced colorectal neoplasia (adjusted OR 1.79; CI 0.98-3.28). Polyp size ≥ 10 mm, right colon location, and the presence of multiple serrated polyps were also not associated with advanced colorectal neoplasia. CONCLUSIONS: Our results suggest that there is not a strong association between SSA/Ps and subsequent advanced colorectal neoplasia during the 5 years following SSA/P removal.

Changes in use of opioid therapy after colon cancer diagnosis: a population-based study.

PURPOSE: To describe patterns of opioid use in cancer survivors. METHODS: In a cohort study of colon cancer patients diagnosed during 1995-2014 and enrolled at two Kaiser Permanente regions, we constructed quarterly measures of opioid use from 1 year before cancer diagnosis through 5 years after diagnosis to examine changes in use. Measures included any use, incident use, regular use (use ≥ 45 days in a 91-day quarter), and average daily dose (converted to morphine milligram equivalent, MME). We also assessed temporal trends of opioid use. RESULTS: Of 2,039 colon cancer patients, 11-15% received opioids in the four pre-diagnosis quarters, 68% in the first quarter after diagnosis, and 15-17% in each subsequent 19 quarters. Regular opioid use increased from 3 to 5% pre-diagnosis to 5-7% post diagnosis. Average dose increased from 15 to 17 MME/day pre-diagnosis to 14-22 MME/day post diagnosis (excluding the quarter in which cancer was diagnosed). Among post-diagnosis opioid users, 73-95% were on a low dose (< 20 MME/day). Over years, regular use of opioids increased in survivorship with no change in dosage. CONCLUSION: Opioid use slightly increased following a colon cancer diagnosis, but high-dose use was rare. Research is needed to differentiate under- versus over-treatment of cancer pain.

Cardiovascular medication use and risks of colon cancer recurrences and additional cancer events: a cohort study.

BACKGROUND: Cardiovascular medications may be associated with cancer development, but little is known about their association with cancer recurrence. Medications such as statins and antihypertensives may be commonly used among colon cancer survivors, who are, on average, diagnosed in their mid-60s. We described the associations between statins and antihypertensive medications and colon cancer recurrence in a large, population-based study. METHODS: We conducted a cohort study among adults with stage I-IIIA colon cancer diagnosed in 1995-2014 in two Kaiser Permanente regions, Colorado and Washington. Statin and antihypertensive use were obtained from electronic pharmacy dispensing data. People were classified as medication users on the date of their first dispensing after cohort entry, which started 90 days after completing cancer treatment, continuing through the earliest of death, health plan disenrollment, or chart abstraction. We collected outcome information from medical record abstraction and tumor registries on colon cancer recurrences and second primary cancers. Using Cox proportional hazards multivariable models, we estimated hazard ratios (HRs) with 95% confidence intervals (CIs) for colon cancer recurrences and any cancer event (recurrences and new primaries at any anatomic site) comparing medication users to non-users. RESULTS: Among 2039 people, 937 (46%) used statins and 1425 (70%) used antihypertensives at any point during a median of 4.9 years of follow-up; 460 people had any additional cancer event, including 152 with a colon cancer recurrence. Statin use was not associated with colon cancer recurrence (HR = 1.09, 95%CI = 0.65-1.85) or any cancer event (HR = 1.12, 95%CI = 0.85-1.47), nor was antihypertensive use associated with recurrence (HR = 0.73, 95%CI = 0.44-1.21) or any cancer event (HR = 0.93, 95%CI = 0.70-1.24). CONCLUSIONS: Our results suggest no association between cardiovascular medication use and the risk of recurrence or any additional cancer, and may provide reassurance to colon cancer survivors.

Use of antidepressants after colon cancer diagnosis and risk of recurrence.

OBJECTIVE: Prior research examining the association between use of antidepressants after colon cancer diagnosis and risk of recurrence is scant. We evaluated this association among colon cancer patients diagnosed at two integrated health care delivery systems in the United States. METHODS: We conducted a cohort study of stage I to IIIA colon cancer patients diagnosed at greater than or equal to 18 years of age at Kaiser Permanente Colorado and Kaiser Permanente Washington during 1995 to 2014. We used pharmacy records to identify dispensings for antidepressants and tumor registry records and patients' medical charts to identify cancer recurrences. Using Cox proportional hazards models, we estimated the adjusted hazard ratio (HR) of colon cancer recurrence comparing patients who used antidepressants after diagnosis to those who did not. We also evaluated the risk associated with use of selective serotonin reuptake inhibitors (SSRIs) and tricyclic antidepressants (TCAs) separately. RESULTS: Among the 1923 eligible colon cancer patients, 807 (42%) used an antidepressant after diagnosis and 139 had a colon cancer recurrence during an average 5.6 years of follow-up. Use of antidepressants after colon cancer diagnosis was not associated with risk of recurrence (HR: 1.14; 95% confidence interval [CI], 0.69-1.87). The HR for use of SSRIs was 1.22 (95% CI, 0.64-2.30), and for TCAs, it was 1.18 (95% CI, 0.68-2.07). CONCLUSIONS: Our findings suggest that use of antidepressants after colon cancer diagnosis was common and not associated with risk of recurrence. Future larger studies with greater power to examine risk associated with individual antidepressants would be valuable additions to the evidence base.

Risk of colon cancer recurrence in relation to diabetes.

PURPOSE: To describe the association between diabetes and colon cancer recurrence. METHODS: We conducted a cohort study at two integrated health care delivery systems in the United States. Using tumor registry data, we identified patients aged ≥ 18 years when diagnosed with stage I-IIIA adenocarcinomas of the colon during 1995-2014. Pre-existing diabetes was ascertained via diagnosis codes. Medical records were reviewed for eligibility and to abstract recurrence and covariate information. Recurrence was ascertained beginning 90 days after the end of colon cancer treatment (i.e., cohort entry). Recurrence of any cancer or a new primary cancer at any site was a secondary outcome. We used multivariable Cox proportional hazards models to estimate hazard ratios (HR) and 95% confidence intervals (CIs) for the associations between diabetes at cohort entry and study outcomes. RESULTS: Among the 1,923 eligible patients, 393 (16.7%) had diabetes at cohort entry. Diabetes was not associated with recurrence (HR 0.87; 95% CI 0.56-1.33) or with any subsequent cancer (HR 1.09; 95% CI 0.85-1.40). When the definition of recurrence included second primary colorectal cancer, risk was non-significantly higher in patients with diabetes than without diabetes. CONCLUSIONS: The risk of colon cancer recurrence appears to be similar in patients with and without diabetes at diagnosis. IMPACT: Future studies should evaluate the association between diabetes and colorectal cancer outcomes, especially second primary colon cancers, in larger populations.

Cancer Survivors' Reported Discussions with Health Care Providers About Follow-Up Care and Receipt of Written Care Plans.

Prior studies reveal gaps in cancer survivors' discussions with health care providers about follow-up care and receipt of care plans; however, whether survivorship care planning may vary by cancer type is not known. We surveyed 615 survivors of breast, colorectal, prostate, lung cancer, and melanoma enrolled in three health plans to examine cancer survivors' self-reported discussions of follow-up care, including the need for surveillance, late and long-term effects, emotional needs, and health behaviors. We assessed whether cancer survivors received a written treatment summary and post-treatment care instructions. Most (92%) survivors reported having a discussion about the need for surveillance; 75%, late and long-term effects; 69%, lifestyle and health behaviors; and 53%, emotional and social needs. Most (88%) reported receiving post-treatment care instructions and 47%, a treatment summary. While there was little difference among survivors' receipt of surveillance or health behavior recommendations by cancer type (p = 0.85 and p = 0.66, respectively), discussions of late and long-term effects occurred among 82% of prostate, 78% of breast, 73% of melanoma, 72% of colorectal, and 67% of lung survivors (p = 0.06). Approximately half of survivors reported discussions of emotional needs, with modest differences by cancer type (p = 0.08). Our findings indicate that most patient-provider discussions cover information on surveillance, with less emphasis on late and long-term effects, lifestyle and health behaviors, and substantially less focusing on emotional and social needs. No or modest differences in discussions occurred by cancer type. Whether tailoring information to individual cancer survivor needs is beneficial should be examined.

Association of worry about cancer to benefit finding and functioning in long-term cancer survivors.

PURPOSE: Worry about cancer recurrence or progression is associated with negative effects of cancer, such as worse physical functioning, but associations with positive changes post-cancer, such as benefit finding, are unknown. We measured the proportion of patients reporting frequent worry about cancer recurrence or progression and examined the association between worry about cancer recurrence or progression to benefit finding and functioning in cancer. METHODS: We surveyed 594 long-term (5-10 years post-diagnosis) survivors of cancer (breast, prostate, colorectal, lung, melanoma) in this cross-sectional study. The survey asked about worry about cancer recurrence/progression, negative effects of cancer on mental and physical function, and benefit finding as a result of the cancer (positive effects). Multivariate regressions estimated associations of worry about cancer with negative and positive effects of cancer. RESULTS: Worrying about cancer often or all the time was reported by 19.6% of survivors. Worry about cancer was related to worse functioning (odds ratio (OR) range 1.40 to 1.46, all p's < .01). Worry about recurrence/progression was unrelated to benefit finding (all p's > .10). CONCLUSIONS: Worry about cancer was associated with negative, but not positive, effects of cancer. Treating worry about cancer is unlikely to reduce benefit finding after cancer. Given the high prevalence of worry about cancer and relationship to negative effects of cancer, clinical care should attempt to address this worry for long-term survivors.

The Cancer Research Network: a platform for epidemiologic and health services research on cancer prevention, care, and outcomes in large, stable populations.

PURPOSE: The ability to collect data on patients for long periods prior to, during, and after a cancer diagnosis is critical for studies of cancer etiology, prevention, treatment, outcomes, and costs. We describe such data capacities within the Cancer Research Network (CRN), a cooperative agreement between the National Cancer Institute (NCI) and organized health care systems across the United States. METHODS: Data were extracted from each CRN site's virtual data warehouse using a centrally written and locally executed program. We computed the percent of patients continuously enrolled ≥1, ≥5, and ≥10 years before cancer diagnosis in 2012-2015 (year varied by CRN site). To describe retention after diagnosis, we computed the cumulative percentages enrolled, deceased, and disenrolled each year after the diagnosis for patients diagnosed in 2000. RESULTS: Approximately 8 million people were enrolled in ten CRN health plans on December 31, 2014 or 2015 (year varied by CRN site). Among more than 30,000 recent cancer diagnoses, 70 % were enrolled for ≥5 years and 56 % for ≥10 years before diagnosis. Among 25,274 cancers diagnosed in 2000, 28 % were still enrolled in 2010, 45 % had died, and 27 % had disenrolled from CRN health systems. CONCLUSIONS: Health plan enrollment before cancer diagnosis was generally long in the CRN, and the proportion of patients lost to follow-up after diagnosis was low. With long enrollment histories among cancer patients pre-diagnosis and low post-diagnosis disenrollment, the CRN provides an excellent platform for epidemiologic and health services research on cancer incidence, outcomes, and costs.

Predicting suicide death after emergency department visits with mental health or self-harm diagnoses.

OBJECTIVE: Use health records data to predict suicide death following emergency department visits. METHODS: Electronic health records and insurance claims from seven health systems were used to: identify emergency department visits with mental health or self-harm diagnoses by members aged 11 or older; extract approximately 2500 potential predictors including demographic, historical, and baseline clinical characteristics; and ascertain subsequent deaths by self-harm. Logistic regression with lasso and random forest models predicted self-harm death over 90 days after each visit. RESULTS: Records identified 2,069,170 eligible visits, 899 followed by suicide death within 90 days. The best-fitting logistic regression with lasso model yielded an area under the receiver operating curve of 0.823 (95% CI 0.810-0.836). Visits above the 95th percentile of predicted risk included 34.8% (95% CI 31.1-38.7) of subsequent suicide deaths and had a 0.303% (95% CI 0.261-0.346) suicide death rate over the following 90 days. Model performance was similar across subgroups defined by age, sex, race, and ethnicity. CONCLUSIONS: Machine learning models using coded data from health records have moderate performance in predicting suicide death following emergency department visits for mental health or self-harm diagnosis and could be used to identify patients needing more systematic follow-up.

Effects of the COVID-19 Pandemic on Animal-Assisted Activities in Pediatric Hospitals.

INTRODUCTION: The goal of this study was to document current hospital-based animal-assisted activities (AAA) practices. METHOD: We contacted 20 hospitals and asked about their AAA programs, including COVID-19 precautions. RESULTS: Eighteen of 20 hospitals responded. Before 2020, all offered either in-person only (n = 17) or both in-person and virtual AAA visits (n = 1). In early 2022, 13 provided in-person visits; the five hospitals that had not resumed in-person visits planned to restart. Most hospitals stopped group visits. Most required that patients and handlers be free of COVID-19 symptoms and that handlers be vaccinated and wear masks and eye protection. Most did not require COVID-19 vaccination for patients. None required handlers to test negative for COVID-19. DISCUSSION: The COVID-19 pandemic impacted hospital-based pediatric AAA. Future studies should assess the effectiveness of virtual AAA and of precautions to prevent COVID-19 transmission between patients and AAA volunteers.

A Randomized Controlled Trial of Animal-assisted Activities for Pediatric Oncology Patients: Psychosocial and Microbial Outcomes.

INTRODUCTION: Evidence about the effectiveness and safety of dog visits in pediatric oncology is limited. METHOD: We conducted a randomized controlled trial (n=26) of dog visits versus usual care among pediatric oncology inpatients. Psychological functioning and microbial load from hand wash samples were evaluated. Parental anxiety was a secondary outcome. RESULTS: We did not observe a difference in the adjusted mean present functioning score (-3.0; 95% confidence interval [CI], -12.4 to 6.4). The difference in microbial load on intervention versus control hands was -0.04 (95% CI, -0.60 to 0.52) log10 CFU/mL, with an upper 95% CI limit below the prespecified noninferiority margin. Anxiety was lower in parents of intervention versus control patients. DISCUSSION: We did not detect an effect of dog visits on functioning; however, our study was underpowered by low recruitment. Visits improved parental anxiety. With hand sanitization, visits did not increase hand microbial levels. CLINICAL TRIAL REGISTRATION: Clinicaltrials.gov NCT03471221.

Risk of Colorectal Cancer and Colorectal Cancer Mortality Beginning Ten Years after a Negative Colonoscopy, among Screen-Eligible Adults 76 to 85 Years Old.

BACKGROUND: Few empirical data are available to inform older adults' decisions about whether to screen or continue screening for colorectal cancer based on their prior history of screening, particularly among individuals with a prior negative exam. METHODS: Using a retrospective cohort of older adults receiving healthcare at three Kaiser Permanente integrated healthcare systems in Northern California (KPNC), Southern California (KPSC), and Washington (KPWA), we estimated the cumulative risk of colorectal cancer incidence and mortality among older adults who had a negative colonoscopy 10 years earlier, accounting for death from other causes. RESULTS: Screen-eligible adults ages 76 to 85 years who had a negative colonoscopy 10 years earlier were found to be at a low risk of colorectal cancer diagnosis, with a cumulative incidence of 0.39% [95% CI, 0.31%-0.48%) at 2 years that increased to 1.29% (95% CI, 1.02%-1.61%) at 8 years. Cumulative mortality from colorectal cancer was 0.04% (95% CI, 0.02%-0.08%) at 2 years and 0.46% (95% CI, 0.30%-0.70%) at 8 years. CONCLUSIONS: These low estimates of cumulative colorectal cancer incidence and mortality occurred in the context of much higher risk of death from other causes. IMPACT: Knowledge of these results could bear on older adults' decision to undergo or not undergo further colorectal cancer screening, including choice of modality, should they decide to continue screening. See related commentary by Lieberman, p. 6.

Risk of Colorectal Cancer and Colorectal Cancer Mortality Beginning One Year after a Negative Fecal Occult Blood Test, among Screen-Eligible 76- to 85-Year-Olds.

BACKGROUND: Colorectal cancer screening is universally recommended for adults ages 45 to 75 years. Noninvasive fecal occult blood tests are effective screening tests recommended by guidelines. However, empirical evidence to inform older adults' decisions about whether to continue screening is sparse, especially for individuals with prior screening. METHODS: This study used a retrospective cohort of older adults at three Kaiser Permanente integrated healthcare systems (Northern California, Southern California, Washington) and Parkland Health. Beginning 1 year following a negative stool-based screening test, cumulative risks of colorectal cancer incidence, colorectal cancer mortality (accounting for deaths from other causes), and non-colorectal cancer mortality were estimated. RESULTS: Cumulative incidence of colorectal cancer in screen-eligible adults ages 76 to 85 with a negative fecal occult blood test 1 year ago (N = 118,269) was 0.23% [95% confidence interval (CI), 0.20%-0.26%] after 2 years and 1.21% (95% CI, 1.13%-1.30%) after 8 years. Cumulative colorectal cancer mortality was 0.03% (95% CI, 0.02%-0.04%) after 2 years and 0.33% (95% CI, 0.28%-0.39%) after 8 years. Cumulative risk of death from non-colorectal cancer causes was 4.81% (95% CI, 4.68%-4.96%) after 2 years and 28.40% (95% CI, 27.95%-28.85%) after 8 years. CONCLUSIONS: Among 76- to 85-year-olds with a recent negative stool-based test, cumulative colorectal cancer incidence and mortality estimates were low, especially within 2 years; death from other causes was over 100 times more likely than death from colorectal cancer. IMPACT: These findings of low absolute colorectal cancer risk, and comparatively higher risk of death from other causes, can inform decision-making regarding whether and when to continue colorectal cancer screening beyond age 75 among screen-eligible adults.

Complex modeling with detailed temporal predictors does not improve health records-based suicide risk prediction.

Suicide risk prediction models can identify individuals for targeted intervention. Discussions of transparency, explainability, and transportability in machine learning presume complex prediction models with many variables outperform simpler models. We compared random forest, artificial neural network, and ensemble models with 1500 temporally defined predictors to logistic regression models. Data from 25,800,888 mental health visits made by 3,081,420 individuals in 7 health systems were used to train and evaluate suicidal behavior prediction models. Model performance was compared across several measures. All models performed well (area under the receiver operating curve [AUC]: 0.794-0.858). Ensemble models performed best, but improvements over a regression model with 100 predictors were minimal (AUC improvements: 0.006-0.020). Results are consistent across performance metrics and subgroups defined by race, ethnicity, and sex. Our results suggest simpler parametric models, which are easier to implement as part of routine clinical practice, perform comparably to more complex machine learning methods.

Validating a predictive algorithm for suicide risk with Alaska Native populations.

INTRODUCTION: The American Indian/Alaska Native (AI/AN) suicide rate in Alaska is twice the state rate and four times the U.S. rate. Healthcare systems need innovative methods of suicide risk detection. The Mental Health Research Network (MHRN) developed suicide risk prediction algorithms in a general U.S. METHODS: We applied MHRN predictors and regression coefficients to electronic health records of AI/AN patients aged ≥13 years with behavioral health diagnoses and primary care visits between October 1, 2016, and March 30, 2018. Logistic regression assessed model accuracy for predicting and stratifying risk for suicide attempt within 90 days after a visit. We compared expected to observed risk and assessed model performance characteristics. RESULTS: 10,864 patients made 47,413 primary care visits. Suicide attempt occurred after 589 (1.2%) visits. Visits in the top 5% of predicted risk accounted for 40% of actual attempts. Among visits in the top 0.5% of predicted risk, 25.1% were followed by suicide attempt. The best fitting model had an AUC of 0.826 (95% CI: 0.809-0.843). CONCLUSIONS: The MHRN model accurately predicted suicide attempts among AI/AN patients. Future work should develop clinical and operational guidance for effective implementation of the model with this population.

Evaluating and Improving Cancer Screening Process Quality in a Multilevel Context: The PROSPR II Consortium Design and Research Agenda.

BACKGROUND: Cancer screening is a complex process involving multiple steps and levels of influence (e.g., patient, provider, facility, health care system, community, or neighborhood). We describe the design, methods, and research agenda of the Population-based Research to Optimize the Screening Process (PROSPR II) consortium. PROSPR II Research Centers (PRC), and the Coordinating Center aim to identify opportunities to improve screening processes and reduce disparities through investigation of factors affecting cervical, colorectal, and lung cancer screening in U.S. community health care settings. METHODS: We collected multilevel, longitudinal cervical, colorectal, and lung cancer screening process data from clinical and administrative sources on >9 million racially and ethnically diverse individuals across 10 heterogeneous health care systems with cohorts beginning January 1, 2010. To facilitate comparisons across organ types and highlight data breadth, we calculated frequencies of multilevel characteristics and volumes of screening and diagnostic tests/procedures and abnormalities. RESULTS: Variations in patient, provider, and facility characteristics reflected the PROSPR II health care systems and differing target populations. PRCs identified incident diagnoses of invasive cancers, in situ cancers, and precancers (invasive: 372 cervical, 24,131 colorectal, 11,205 lung; in situ: 911 colorectal, 32 lung; precancers: 13,838 cervical, 554,499 colorectal). CONCLUSIONS: PROSPR II's research agenda aims to advance: (i) conceptualization and measurement of the cancer screening process, its multilevel factors, and quality; (ii) knowledge of cancer disparities; and (iii) evaluation of the COVID-19 pandemic's initial impacts on cancer screening. We invite researchers to collaborate with PROSPR II investigators. IMPACT: PROSPR II is a valuable data resource for cancer screening researchers.

Evaluation of Electronic Health Record-Based Suicide Risk Prediction Models on Contemporary Data.

BACKGROUND: Suicide risk prediction models have been developed by using information from patients' electronic health records (EHR), but the time elapsed between model development and health system implementation is often substantial. Temporal changes in health systems and EHR coding practices necessitate the evaluation of such models in more contemporary data. OBJECTIVES: A set of published suicide risk prediction models developed by using EHR data from 2009 to 2015 across seven health systems reported c-statistics of 0.85 for suicide attempt and 0.83 to 0.86 for suicide death. Our objective was to evaluate these models' performance with contemporary data (2014-2017) from these systems. METHODS: We evaluated performance using mental health visits (6,832,439 to mental health specialty providers and 3,987,078 to general medical providers) from 2014 to 2017 made by 1,799,765 patients aged 13+ across the health systems. No visits in our evaluation were used in the previous model development. Outcomes were suicide attempt (health system records) and suicide death (state death certificates) within 90 days following a visit. We assessed calibration and computed c-statistics with 95% confidence intervals (CI) and cut-point specific estimates of sensitivity, specificity, and positive/negative predictive value. RESULTS: Models were well calibrated; 46% of suicide attempts and 35% of suicide deaths in the mental health specialty sample were preceded by a visit (within 90 days) with a risk score in the top 5%. In the general medical sample, 53% of attempts and 35% of deaths were preceded by such a visit. Among these two samples, respectively, c-statistics were 0.862 (95% CI: 0.860-0.864) and 0.864 (95% CI: 0.860-0.869) for suicide attempt, and 0.806 (95% CI: 0.790-0.822) and 0.804 (95% CI: 0.782-0.829) for suicide death. CONCLUSION: Performance of the risk prediction models in this contemporary sample was similar to historical estimates for suicide attempt but modestly lower for suicide death. These published models can inform clinical practice and patient care today.

Predicting suicide attempts and suicide deaths among adolescents following outpatient visits.

BACKGROUND: Few studies report on machine learning models for suicide risk prediction in adolescents and their utility in identifying those in need of further evaluation. This study examined whether a model trained and validated using data from all age groups works as well for adolescents or whether it could be improved. METHODS: We used healthcare data for 1.4 million specialty mental health and primary care outpatient visits among 256,823 adolescents across 7 health systems. The prediction target was 90-day risk of suicide attempt following a visit. We used logistic regression with least absolute shrinkage and selection operator (LASSO) and generalized estimating equations (GEE) to predict risk. We compared performance of three models: an existing model, a recalibrated version of that model, and a newly-learned model. Models were compared using area under the receiver operating curve (AUC), sensitivity, specificity, positive predictive value and negative predictive value. RESULTS: The AUC produced by the existing model for specialty mental health visits estimated in adolescents alone (0.796; [0.789, 0.802]) was not significantly different than the AUC of the recalibrated existing model (0.794; [0.787, 0.80]) or the newly-learned model (0.795; [0.789, 0.801]). Predicted risk following primary care visits was also similar: existing (0.855; [0.844, 0.866]), recalibrated (0.85 [0.839, 0.862]), newly-learned (0.842, [0.829, 0.854]). LIMITATIONS: The models did not incorporate non-healthcare risk factors. The models relied on ICD9-CM codes for diagnoses and outcome measurement. CONCLUSIONS: Prediction models already in operational use by health systems can be reliably employed for identifying adolescents in need of further evaluation.