Identification of patients at risk of death from cancer surgery should aid in preoperative preparation. The purpose of this study is to assess and adjust the age-adjusted Charlson comorbidity index (ACCI) to identify cancer patients with increased risk of perioperative mortality. We identified 156,151 patients undergoing surgery for one of the ten common cancers between 2007 and 2011 in the Taiwan National Health Insurance Research Database. Half of the patients were randomly selected, and a multivariate logistic regression analysis was used to develop an adjusted-ACCI score for estimating the risk of 90-day mortality by variables from the original ACCI. The score was validated. The association between the score and perioperative mortality was analyzed. The adjusted-ACCI score yield a better discrimination on mortality after cancer surgery than the original ACCI score, with c-statics of 0.75 versus 0.71. Over 80 years of age, 70–80 years, and renal disease had the strongest impact on mortality, hazard ratios 8.40, 3.63, and 3.09 (P < 0.001), respectively. The overall 90-day mortality rates in the entire cohort varied from 0.9%, 2.9%, 7.0%, and 13.2% in four risk groups stratifying by the adjusted-ACCI score; the adjusted hazard ratio for score 4–7, 8–11, and ≥ 12 was 2.84, 6.07, and 11.17 (P < 0.001), respectively, in 90-day mortality compared to score 0–3. The adjusted-ACCI score helps to identify patients with a higher risk of 90-day mortality after cancer surgery. It might be particularly helpful for preoperative evaluation of patients over 80 years of age.

Cancer survival correlates not only with the features of primary malignancy but also with the degree of underlying comorbidities. Of the multiple methods used for evaluating the impact of comorbidities on survival, the Charlson and Elixhauser methods are most common. This study compared these 2 comorbidity measures for predicting survival in oral cancer patients. Using the Taiwan National Health Insurance claims data (2008-2011), we acquired data regarding patients' characteristics, comorbidities, and survival from 3583 oral cancer patients. Comorbidity was classified according to both the Charlson comorbidity and Elixhauser comorbidity based on the International Classification of Diseases, 9th Revision. The Elixhauser comorbidity score and Charlson comorbidity score were also calculated. The prediction of survival was determined using measures of discrimination, including the Akaike information criterion and Harrell C (C-statistic). The mean age of the study cohort was 52 ± 10 years, and 94.9% of the patients were male. The median follow-up time was 30.1 months, and the 3-year overall survival was 61.6%. Elixhauser comorbidity method added higher discrimination, compared with the Charlson comorbidity method (Harrell C, 0.677 vs 0.651). Furthermore, the Elixhauser comorbidity score outperformed the Charlson comorbidity score in continuous variable (Harrell C, 0.654 vs 0.646) and category (Harrell C, 0.658 vs 0.645). The Elixhauser method is a superior comorbidity risk-adjustment model for oral cancer survival prediction. Utilization of the Elixhauser comorbidity method may be encouraged for risk adjustment in oral cancer study.

Background.: The relationship between low socioeconomic status (SES) and aggressiveness of end-of-life (EOL) care in cancer patients of working age (older than 18 years and younger than 65 years) is not clear. We assessed the association between aggressiveness of EOL care and differences in SES among working-age terminal cancer patients from Taiwan between 2009 and 2011. Methods.: A total of 32,800 cancer deaths were identified from the Taiwan National Health Insurance Research Database. The indicators of aggressive EOL care (chemotherapy, more than one emergency room [ER] visit or hospital admission, more than 14 days of hospitalization, intensive care unit [ICU] admission, and death in an acute care hospital) in the last month of life were examined. The associations between SES and the indicators were explored. Results.: Up to 81% of the cancer deaths presented at least one indicator of aggressive EOL care. Those who were aged 35–44 years and male, had low SES, had metastatic malignant disease, lived in urban areas, or were in hospitals with more abundant health care resources were more likely to receive aggressive EOL care. In multilevel logistic regression analyses, high-SES cancer deaths had less chemotherapy (p < .001), fewer ER visits (p < .001), fewer ICU admissions (p < .001), and lower rates of dying in acute hospitals (p < .001) compared with low-SES cancer deaths. Conclusion.: Working-age terminal cancer patients in Taiwan received aggressive EOL care. EOL cancer care was even more aggressive in those with low SES. Public health strategies should continue to focus on low-SES patients to provide them with better EOL cancer care.

To understand the impact of socioeconomic status (SES) on the diagnosis of primary open-angle glaucoma (POAG) and primary angle closure glaucoma (PACG) in Taiwan. Subjects with glaucoma were identified from the National Health Insurance Research Database of year 2006, which included one million randomly selected insurants. Individuals who had ≥4 ambulatory visits within one year which had the diagnosis code of POAG (ICD-9-CM 365.11 or 365.12) or PACG (365.23) and concurrent prescription of anti-glaucoma medication or surgery were selected. Individual SES was represented by monthly income calculated from the insurance premium. Neighborhood SES was defined based on neighborhood household income averages. Urbanization level of habitation was categorized into 3 levels. The odds ratio of having POAG or PACG in subjects with different SES was evaluated by using multiple logistic regression analysis. In total, 752 and 561 subjects with POAG and PACG, respectively, who were treated on a regular basis, were identified. The diagnosis of glaucoma was affected by age, gender, frequency of healthcare utilization, individual SES, and urbanization level of habitation. With the adjustment of age, gender, healthcare utilization, neighborhood SES and level of urbanization, subjects with lower income were more likely to be diagnosed as PACG, but less likely as POAG. Subjects with more frequent healthcare utilization were more likely to be diagnosed with glaucoma. Subjects with low SES were more susceptible to PACG, but subjects with high SES were more likely to be diagnosed as POAG. This information is useful for the design and target participant setting in glaucoma education and screening campaign to maximize the efficacy of limited resources in preventing glaucoma blindness.

Introduction.: Although palliative chemotherapy during end-of-life care is used for relief of symptoms in patients with metastatic cancer, chemotherapy may lead to more aggressive end-of-life care and less use of hospice service. This is a population-based study of the association between palliative chemotherapy and aggressiveness of end-of-life care. Patients and Methods.: Using the National Health Insurance Research Database of Taiwan, we identified 49,920 patients with metastatic cancer who underwent palliative chemotherapy from January 1, 2009, to December 31, 2011. Patients who received chemotherapy 2–6 months before death were included. Aggressiveness of end-of-life care was examined by previously reported indicators. Cardiopulmonary resuscitation and endotracheal tube intubation were included as indicators of aggressive end-of-life care. The association between palliative chemotherapy and hospice care was studied. Results.: Palliative chemotherapy was associated with more aggressive treatment. After adjustment for patient age, sex, Charlson Comorbidity Index score, cancer group, primary physician’s specialty, postdiagnosis survival, hospital characteristics, hospital caseload, urbanization, and geographic regions, more than one emergency room visit (p < .001), more than one intensive care unit admission (p < .001), and endotracheal intubation (p = .02) during end-of-life care were significantly more common in patients receiving palliative chemotherapy. Patients who did not receive palliative chemotherapy received more hospice care in the last 6 months of life (p < .001). Conclusion.: Although the decision to initiate palliative chemotherapy was made several months before death, this study showed that palliative chemotherapy was associated with more aggressive end-of-life care, including more emergency room visits and intensive care unit admissions, and endotracheal intubation. The patients who received palliative chemotherapy received less hospice service toward the end of life.

The impact of important preexisting comorbidities, such as liver and renal disease, on the outcome of liver resection remains unclear. Identification of patients at risk of mortality will aid in improving preoperative preparations. The purpose of this study is to develop and validate a population-based score based on available preoperative and predictable parameters predicting 90-day mortality after liver resection using data from a hepatitis endemic country. We identified 13,159 patients who underwent liver resection between 2002 and 2006 in the Taiwan National Health Insurance Research Database. In a randomly selected half of the total patients, multivariate logistic regression analysis was used to develop a prediction score for estimating the risk of 90-day mortality by patient demographics, preoperative liver disease and comorbidities, indication for surgery, and procedure type. The score was validated with the remaining half of the patients. Overall 90-day mortality was 3.9%. Predictive characteristics included in the model were age, preexisting cirrhosis-related complications, ischemic heart disease, heart failure, cerebrovascular disease, renal disease, malignancy, and procedure type. Four risk groups were stratified by mortality scores of 1.1%, 2.2%, 7.7%, and 15%. Preexisting renal disease and cirrhosis-related complications were the strongest predictors. The score discriminated well in both the derivation and validation sets with c-statistics of 0.75 and 0.75, respectively. This population-based score could identify patients at risk of 90-day mortality before liver resection. Preexisting renal disease and cirrhosis-related complications had the strongest influence on mortality. This score enables preoperative risk stratification, decision-making, quality assessment, and counseling for individual patients.