Machine Learning to Predict Tamoxifen Nonadherence Among US Commercially Insured Patients With Metastatic Breast Cancer.

PURPOSE: Adherence to tamoxifen citrate among women diagnosed with metastatic breast cancer can improve survival and minimize recurrence. This study aimed to use real-world data and machine learning (ML) methods to classify tamoxifen nonadherence. METHODS: A cohort of women diagnosed with metastatic breast cancer from 2012 to 2017 were identified from IBM MarketScan Commercial Claims and Encounters and Medicare claims databases. Patients with < 80% proportion of days coverage in the year following treatment initiation were classified as nonadherent. Training and internal validation cohorts were randomly generated (4:1 ratio). Clinical procedures, comorbidity, treatment, and health care encounter features in the year before tamoxifen initiation were used to train logistic regression, boosted logistic regression, random forest, and feedforward neural network models and were internally validated on the basis of area under receiver operating characteristic curve. The most predictive ML approach was evaluated to assess feature importance. RESULTS: A total of 3,022 patients were included with 40% classified as nonadherent. All models had moderate predictive accuracy. Logistic regression (area under receiver operating characteristic 0.64) was interpreted with 94% sensitivity (95% CI, 89 to 92) and 0.31 specificity (95% CI, 29 to 33). The model accurately classified adherence (negative predictive value 89%) but was nondiscriminate for nonadherence (positive predictive value 48%). Variable importance identified top predictive factors, including age ≥ 55 years and pretreatment procedures (lymphatic nuclear medicine, radiation oncology, and arterial surgery). CONCLUSION: ML using baseline administrative data predicts tamoxifen nonadherence. Screening at treatment initiation may support personalized care, improve health outcomes, and minimize cost. Baseline claims may not be sufficient to discriminate adherence. Further validation with enriched longitudinal data may improve model performance.

Burden and risk factors for gastrointestinal symptom distress in HIV patients in the modern antiretroviral era.

In the modern antiretroviral (ARV) era, there is limited knowledge about the prevalence and risk factors for HIV patient-reported gastrointestinal (GI) symptoms (diarrhoea/soft stool, nausea/vomiting, bloating/painful abdomen, loss of appetite, and weight loss/wasting) and distress. We prospectively analysed data (2007-2014) on distressing GI symptoms from the Ontario HIV Treatment Network Cohort Study, which follows people attending HIV clinics. Using generalized estimating equations with a logit link, we estimated the associations of psychosocial, demographic, behavioural, and clinical factors with each GI symptoms compared to asymptomatic and non-bothersome symptoms. Among 1532 included participants, 80.4% were male, mean age was 45 years, and 64.6% reported being men who have sex with men. Most were Caucasian (56.3%), a median time since HIV diagnosis of 9.8 years (interquartile range (IQR): 4.1-16.9), and 83.1% were on ARV. More than two-thirds (68.7% (95% confidence intervals (CI): 63.1% to 69.2%)) reported one or more symptoms with a median of 1.2 (IQR: 0-1.7). The proportion remained stable over time since HIV diagnosis and ARV initiation. Risk factors varied for multivariable models. A strong association with Centre for Epidemiologic Studies Depression scale scores of ≥23 was found for all symptoms. Adjusted odds ratios (95% CI) were 1.72 (1.39-2.12), 2.95 (2.33-3.72), 2.20 (1.81-2.68), 4.97 (3.99-6.19), and 2.98 (2.52-3.82) for diarrhoea, nausea/vomiting, bloating, loss of appetite, and weight loss, respectively. With the exception of bloating, odds were significantly lower for those on ARV containing integrase inhibitors and greater for patients reporting current cannabis use. GI symptoms in the modern ARV era are highly prevalent and may arise as a common pathway of distress in response to psychosocial vulnerabilities, regardless of the stage of diagnosis. These findings support the need for integrated approaches to address psychological and physical distress in HIV disease.

Surveillance in Patients With Barrett's Esophagus for Early Detection of Esophageal Adenocarcinoma: A Systematic Review and Meta-Analysis.

OBJECTIVES: Although endoscopic surveillance of patients with Barrett's esophagus (BE) has been widely implemented for early detection of esophageal adenocarcinoma (EAC), its justification has been debated. This systematic review aimed to evaluate benefits, safety, and cost effectiveness of surveillance for patients with BE. METHODS: MEDLINE, EMBASE, EconLit, Scopus, Cochrane, and CINAHL were searched for published human studies that examined screening practices, benefits, safety, and cost effectiveness of surveillance among patients with BE. Reviewers independently reviewed eligible full-text study articles and conducted data extraction and quality assessment, with disagreements resolved by consensus. Random effects meta-analyses were performed to assess the incidence of EAC, EAC/high-grade dysplasia (HGD), and annual stage-specific transition probabilities detected among BE patients under surveillance, and relative risk of mortality among EAC patients detected during surveillance compared with those not under surveillance. RESULTS: A total of 51 studies with 11,028 subjects were eligible; the majority were of high quality based on the Newcastle-Ottawa quality scale. Among BE patients undergoing endoscopic surveillance, pooled EAC incidence per 1,000 person-years of surveillance follow-up was 5.5 (95% confidence interval (CI): 4.2-6.8) and pooled EAC/HGD incidence was 7.7 (95% CI: 5.7-9.7). Pooled relative mortality risk among surveillance-detected EAC patients compared with nonsurveillance-detected EAC patients was 0.386 (95% CI: 0.242-0.617). Pooled annual stage-specific transition probabilities from nondysplastic BE to low-grade dysplasia, high-grade dysplasia, and EAC were 0.019, 0.003, and 0.004, respectively. There was, however, insufficient scientific evidence on safety and cost effectiveness of surveillance for BE patients. CONCLUSIONS: Our findings confirmed a low incidence rate of EAC among BE patients undergoing surveillance and a reduction in mortality by 61% among those who received regular surveillance and developed EAC. Because of knowledge gaps, it is important to assess safety of surveillance and health-care resource use and costs to supplement existing evidence and inform a future policy decision for surveillance programs.

Colorectal cancer among persons with HIV: protocol for a systematic review and meta-analysis.

BACKGROUND: As persons with HIV live longer, data regarding the epidemiology of colorectal cancer are required to optimize the long-term management of these patients. The purpose of this systematic review and meta-analysis is to synthesize evidence regarding the incidence of colorectal cancer in persons with HIV. METHODS/DESIGN: Our primary outcome is the standardized incidence ratio of colorectal cancer among persons with HIV relative to rates in persons not living with HIV. Our secondary objectives are to summarize the evidence for differences with respect to stage at diagnosis, site of disease, and mortality due to colorectal cancer. We will search electronic bibliographic databases from their inception date, as well as conference proceedings and reference lists of included articles. Two investigators will independently screen citations and full-text articles, conduct data abstraction, and appraise study quality. We will examine clinical, methodological, and statistical heterogeneity among studies prior to conducting meta-analysis. Random effects meta-analysis methods will be employed to estimate standardized incidence ratios. These data will inform the development of guidelines for colorectal cancer screening in persons with HIV. SYSTEMATIC REVIEW REGISTRATION: PROSPERO CRD42014013449.

The effect of HIV-hepatitis C co-infection on bone mineral density and fracture: a meta-analysis.

OBJECTIVE: There is a variable body of evidence on adverse bone outcomes in HIV patients co-infected with hepatitis C virus (HCV). We examined the association of HIV/HCV co-infection on osteoporosis or osteopenia (reduced bone mineral density; BMD) and fracture. DESIGN: Systematic review and random effects meta-analyses. METHODS: A systematic literature search was conducted for articles published in English up to 1 April 2013. All studies reporting either BMD (g/cm2, or as a T-score) or incident fractures in HIV/HCV co-infected patients compared to either HIV mono-infected or HIV/HCV uninfected/seronegative controls were included. Random effects meta-analyses estimated the pooled odds ratio (OR) and the relative risk (RR) and associated 95% confidence intervals (CI). RESULTS: Thirteen eligible publications (BMD N = 6; Fracture = 7) of 2,064 identified were included with a total of 427,352 subjects. No publications reported data on HCV mono-infected controls. Meta-analysis of cross-sectional studies confirmed that low bone mineral density was increasingly prevalent among co-infected patients compared to HIV mono-infected controls (pooled OR 1.98, 95% CI 1.18, 3.31) but not those uninfected (pooled OR 1.47, 95% CI 0.78, 2.78). Significant association between co-infection and fracture was found compared to HIV mono-infected from cohort and case-control studies (pooled RR 1.57, 95% CI 1.33, 1.86) and compared to HIV/HCV uninfected from cohort (pooled RR 2.46, 95% CI 1.03, 3.88) and cross-sectional studies (pooled OR 2.30, 95% CI 2.09, 2.23). CONCLUSIONS: The associations of co-infection with prevalent low BMD and risk of fracture are confirmed in this meta-analysis. Although the mechanisms of HIV/HCV co-infection's effect on BMD and fracture are not well understood, there is evidence to suggest that adverse outcomes among HIV/HCV co-infected patients are substantial.