Machine learning can accurately predict pre-admission baseline hemoglobin and creatinine in intensive care patients.

Patients admitted to the intensive care unit frequently have anemia and impaired renal function, but often lack historical blood results to contextualize the acuteness of these findings. Using data available within two hours of ICU admission, we developed machine learning models that accurately (AUC 0.86-0.89) classify an individual patient's baseline hemoglobin and creatinine levels. Compared to assuming the baseline to be the same as the admission lab value, machine learning performed significantly better at classifying acute kidney injury regardless of initial creatinine value, and significantly better at predicting baseline hemoglobin value in patients with admission hemoglobin of <10 g/dl.

Patterns of care for anal cancer in the United States - a comparison between academic and community cancer centers.

BACKGROUND: Management of squamous cell carcinoma of the anus (SCCA) is becoming more relevant, as its incidence increases. The purpose of this study was to investigate possible differences in patient population and care delivery for SCCA between academic and community cancer programs in the United States. METHODS: A review of available data from the American College of Surgeons Committee on Cancer National Cancer DataBase focused on gender, age, race, type of health insurance, comorbidity score, distance traveled for care, stage at diagnosis, and therapy utilization (surgery, chemotherapy, and radiation therapy) as first course of treatment (FCT). The analysis included 38,766 patients treated for SCCA. Of them, 14,422 patients received treatment at Academic Cancer Programs (ACPs), while 24,344 were treated at Community Cancer Programs (CCPs) between the years 2003 and 2013. RESULTS: Over the 11-year study period, ACPs had significantly more male patients, of younger age, a greater non-white race population, with more Medicaid or no insurance coverage, who traveled farther for cancer center care (p < 0.001). There was no difference between ACPs and CCPs with respect to Charlson co-morbidity score and stage of SCCA at diagnosis. For stage 0 patients, use of chemotherapy was 8% for ACPs, 9% for CCPs, and use of radiotherapy was 10% for ACPs and 14% for CCPs. The incidence of stage unknown was identical at both ACPs and CCPs (11.5%). CCPs had a greater overall utilization of radiation therapy as FCT for stage 0, I, II and IV patients (p < 0.001). CONCLUSIONS: Our study indicates that gender, demographic and socio-economic differences exist in the patient population with SCCA accessing different cancer programs in the US. The high incidence of stage unknown patients reflects ongoing challenges in the pre-treatment phase. A significant percentage of stage 0 patients received systemic chemotherapy and/or radiotherapy, rather than surgery alone. Despite comparable stage at diagnosis and comorbidity scores between ACPs and CCPs, there appear to be variations in treatment choices, especially with the use of radiotherapy, with associated cost and toxicity risks. Further analysis and monitoring of SCCA management in the US may lead to improved compliance with NCCN guidelines.

Challenges in Management of Squamous Cell Carcinoma of the Anus in New England and Across the United States: A Review of the National Cancer Data Base.

BACKGROUND: Management of squamous cell carcinoma of the anus (SCCA) is becoming more relevant, as its incidence continues to increase. The purpose of this study was to evaluate regional and national data to assess trends in epidemiology, access to cancer center care, and overall management strategies in SCCA. STUDY DESIGN: A review of available data from the American College of Surgeons Committee on Cancer National Cancer Data Base focused on incidence, sex, age, stage at diagnosis, distance traveled for care, and utilization of therapy as first course of treatment (FCT). The analysis included 40,817 patients treated for SCCA at 1513 cancer centers in the United States, of which 2347 patients were treated at 109 cancer centers in New England, between the years 2003 and 2013. RESULTS: Over the 11-year period, incidence of SCCA increased by 76% in the United States and by 83.8% in New England. Stage was unknown in 11.7% of all US cases, significantly higher than more common cancers, for example, breast (4.3%), prostate (6%), or colon (7.8%) (P<0.001). Patients in southern New England, compared with northern New England, traveled <10 miles more often (53.4% vs. 38.1%) (P<0.001), and>25 miles less often (14.3% vs. 28.7%) (P<0.001). Cases of early stage SCCA (0, I) were more frequent in southern New England (29.2%) than northern New England (21.7%) (P=0.0025), whereas more advanced stage (II to IV) cases occurred less frequently in southern New England (60.1%) than northern New England (72%) (P<0.001). Overall, the most common FCT was chemoradiotherapy, utilized in 49.3% of cases, followed by chemoradiotherapy plus surgery in 19.4% of cases. Stage unknown patients were treated with chemoradiotherapy in 34.6% of cases, with surgery alone in 20.2%, and with chemoradiotherapy plus surgery in 15.4% of cases. CONCLUSIONS: The incidence of SCCA is steadily increasing. Its frequency of stage unknown is significantly higher than other common cancer sites. Travel distance and stage at diagnosis data may reflect regional differences in cancer center care access. Although chemoradiotherapy remains the most commonly utilized FCT, challenges in accurate staging and inconsistent use of additional prognostic variables may affect optimal treatment.