Identifying Nonpatient Authors of Patient Portal Secure Messages in Oncology: A Proof-of-Concept Demonstration of Natural Language Processing Methods.

PURPOSE: Patient portal secure messages are not always authored by the patient account holder. Understanding who authored the message is particularly important in an oncology setting where symptom reporting is crucial to patient treatment. Natural language processing has the potential to detect messages not authored by the patient automatically. METHODS: Patient portal secure messages from the Memorial Sloan Kettering Cancer Center were retrieved and manually annotated as a predicted unregistered proxy (ie, not written by the patient) or a presumed patient. After randomly splitting the annotated messages into training and test sets in a 70:30 ratio, a bag-of-words approach was used to extract features and then a Least Absolute Shrinkage and Selection Operator (LASSO) model was trained and used for classification. RESULTS: Portal secure messages (n = 2,000) were randomly selected from unique patient accounts and manually annotated. We excluded 335 messages from the data set as the annotators could not determine if they were written by a patient or proxy. Using the remaining 1,665 messages, a LASSO model was developed that achieved an area under the curve of 0.932 and an area under the precision recall curve of 0.748. The sensitivity and specificity related to classifying true-positive cases (predicted unregistered proxy-authored messages) and true negatives (presumed patient-authored messages) were 0.681 and 0.960, respectively. CONCLUSION: Our work demonstrates the feasibility of using unstructured, heterogenous patient portal secure messages to determine portal secure message authorship. Identifying patient authorship in real time can improve patient portal account security and can be used to improve the quality of the information extracted from the patient portal, such as patient-reported outcomes.

Autonomous detection, grading, and reporting of postoperative complications using natural language processing.

INTRODUCTION: Natural language processing, a computer science technique that allows interpretation of narrative text, is infrequently used to identify surgical complications. We designed a natural language processing algorithm to identify and grade the severity of deep venous thrombosis and pulmonary embolism (together: venous thromboembolism). METHODS: Patients from our 2011-2014 American College of Surgeons National Surgical Quality Improvement Project cohorts with a duplex ultrasound or a computerized tomography angiography of the chest performed within 30 days of surgery were divided into training and validation datasets. A "bag of words" approach classified the reports; other electronic health record data classified the venous thromboembolism's severity. RESULTS: Of the 10,295 American College of Surgeons National Surgical Quality Improvement Project patients, 251 were used in our deep venous thromboses validation cohort (273 total ultrasounds) and 506 in our pulmonary embolisms cohort (552 total computerized tomography angiographies). For deep venous thromboses the sensitivity and specificity were 85.1% and 94.6%, while for pulmonary embolisms they were 90% and 98.7%. Most discordances were due to lack of imaging documentation of a deep venous thrombosis (28/41, 68.3%) or pulmonary embolism (6/6, 100%). Most deep venous thromboses (28 patients, 54.6%) and pulmonary embolisms (25 patients, 75.8%) required administration of therapeutic intravenous or subcutaneous anticoagulation. CONCLUSION: Natural language processing can reliably detect the presence of postoperative venous thromboembolisms, and its use should be expanded for the detection of other conditions from narrative documentation.

Assessing Estimates of Patient Safety Derived From Coded Data.

The Agency for Healthcare Research and Quality (AHRQ) developed Patient Safety Indicators (PSI) with the goal of detecting and preventing adverse events using administrative data. Use of PSIs to measure safety and quality of care raises questions since they rely heavily on coded data. The goal of this paper is to review literature on AHRQ PSIs, specifically their accuracy in detecting adverse events, and how recent coding changes affect the accuracy of these estimates. The Healthcare Utilization Project (HCUP) National Inpatient Sample (NIS) contains data on more than seven million hospital stays yearly. Trending these estimates over time indicate there has been a decrease in rates for the majority of PSIs, with Accidental Puncture or Laceration, Central Venous Catheter Blood Stream Infections, Iatrogenic Pneumothorax, and Deaths from complications decreasing significantly each year from 2008-2012. Birth Trauma, Retained Foreign Bodies, Post-op Sepsis and Post-Op Respiratory Failure PSIs show no improvement from 2008-2012. A literature review on accuracy of PSIs and the affect of the Present on Admission (POA) indicator on PSI estimates reveals mixed results. PSIs serve as a useful tool in identifying problem areas in quality of care but should be used cautiously in determining hospital performance.

Impact of Robotic Platforms on Surgical Approach and Costs in the Management of Morbidly Obese Patients with Newly Diagnosed Uterine Cancer.

BACKGROUND: Minimally invasive surgery (MIS) is associated with decreased complication rates, length of hospital stay, and cost compared with laparotomy. Robotic-assisted surgery-a method of laparoscopy-addresses many of the limitations of standard laparoscopic instrumentation, thus leading to increased rates of MIS. We sought to assess the impact of robotics on the rates and costs of surgical approaches in morbidly obese patients with uterine cancer. METHODS: Patients who underwent primary surgery at our institution for uterine cancer from 1993 to 2012 with a BMI ≥40 mg/m(2) were identified. Surgical approaches were categorized as laparotomy (planned or converted), laparoscopic, robotic, or vaginal. We identified two time periods based on the evolving use of MIS at our institution: laparoscopic (1993-2007) and robotic (2008-2012). Direct costs were analyzed for cases performed from 2009 to 2012. RESULTS: We identified 426 eligible cases; 299 performed via laparotomy, 125 via MIS, and 2 via a vaginal approach. The rates of MIS for the laparoscopic and robotic time periods were 6 % and 57 %, respectively. The rate of MIS was 78 % in this morbidly obese cohort in 2012; 69 % were completed robotically. The median length of hospital stay was 5 days (range 2-37) for laparotomy cases and 1 day (range 0-7) for MIS cases (P < 0.001). The complication rate was 36 and 15 %, respectively (P < 0.001). The rate of wound-related complications was 27 and 6 %, respectively (P < 0.001). Laparotomy was associated with the highest cost. CONCLUSIONS: The robotic platform provides significant health and cost benefits by increasing MIS rates in this patient population.

High burden of palliative needs among older intensive care unit survivors transferred to post-acute care facilities. a single-center study.

RATIONALE: Adults with chronic critical illness (tracheostomy after ≥ 10 d of mechanical ventilation) have a high burden of palliative needs, but little is known about the actual use and potential need of palliative care services for the larger population of older intensive care unit (ICU) survivors discharged to post-acute care facilities. OBJECTIVES: To determine whether older ICU survivors discharged to post-acute care facilities have potentially unmet palliative care needs. METHODS: We examined electronic records from a 1-year cohort of 228 consecutive adults ≥ 65 years of age who had their first medical-ICU admission in 2009 at a single tertiary-care medical center and survived to discharge to a post-acute care facility (excluding hospice). Use of palliative care services was defined as having received a palliative care consultation. Potential palliative care needs were defined as patient characteristics suggestive of physical or psychological symptom distress or anticipated poor prognosis. We examined the prevalence of potential palliative needs and 6-month mortality. MEASUREMENTS AND MAIN RESULTS: The median age was 78 years (interquartile range, 71-84 yr), and 54% received mechanical ventilation for a median of 7 days (interquartile range, 3-16 d). Six subjects (2.6%) received a palliative care consultation during the hospitalization. However, 88% had at least one potential palliative care need; 22% had chronic wounds, 37% were discharged on supplemental oxygen, 17% received chaplaincy services, 23% preferred to not be resuscitated, and 8% were designated "comfort care." The 6-month mortality was 40%. CONCLUSIONS: Older ICU survivors from a single center who required postacute facility care had a high burden of palliative care needs and a high 6-month mortality. The in-hospital postcritical acute care period should be targeted for palliative care assessment and intervention.

A prognostic model for 6-month mortality in elderly survivors of critical illness.

BACKGROUND: Although 1.4 million elderly Americans survive hospitalization involving intensive care annually, many are at risk for early mortality following discharge. No models that predict the likelihood of death after discharge exist explicitly for this population. Therefore, we derived and externally validated a 6-month postdischarge mortality prediction model for elderly ICU survivors. METHODS: We derived the model from medical record and claims data for 1,526 consecutive patients aged ≥ 65 years who had their first medical ICU admission in 2006 to 2009 at a tertiary-care hospital and survived to discharge (excluding those patients discharged to hospice). We then validated the model in 1,010 patients from a different tertiary-care hospital. RESULTS: Six-month mortality was 27.3% and 30.2% in the derivation and validation cohorts, respectively. Independent predictors of mortality (in descending order of contribution to the model's predictive power) were a do-not-resuscitate order, older age, burden of comorbidity, admission from or discharge to a skilled-care facility, hospital length of stay, principal diagnoses of sepsis and hematologic malignancy, and male sex. For the derivation and external validation cohorts, the area under the receiver operating characteristic curve was 0.80 (SE, 0.01) and 0.71 (SE, 0.02), respectively, with good calibration for both (P = 0.31 and 0.43). CONCLUSIONS: Clinical variables available at hospital discharge can help predict 6-month mortality for elderly ICU survivors. Variables that capture elements of frailty, disability, the burden of comorbidity, and patient preferences regarding resuscitation during the hospitalization contribute most to this model's predictive power. The model could aid providers in counseling elderly ICU survivors at high risk of death and their families.