Sunday, January 10, 2016: 9:00-10:30
Kai Chong Tong Auditorium, G/F (Jockey Club School of Public Health and Primary Care Building at Prince of Wales Hospital)

Ya-Chen Tina Shih, PhD1, Chia-Chin Li2, Brian Chiu, PhD3, Ashleigh Guadagnolo, MD4 and Chun-Ru Chien, MD, PhD2, (1)Department of Health Services Research, University of Texas MD Anderson Cancer Center, Houston, TX, (2)Department of Radiation Oncology, China Medical University Hospital, Taichung, Taiwan, (3)Department of Public Health Sciences, University of Chicago, Chicago, IL, (4)Department of Radiation Oncology, University of Texas MD Anderson Cancer Center, Houston, TX
Purpose: Overly aggressive cancer treatment toward the end of life (EOL) is considered an indicator of poor-quality care. While radiotherapy can be used to alleviate symptoms arising from cancer, prolonged courses of radiotherapy may subject patients to additional harms, higher financial burden, and reduced quality of life. We hypothesized that overly aggressive EOL care is more likely among younger cancer patients.   

Method(s): We defined the duration of EOL as the last 30 days of life and used data from the population-based Taiwan cancer registry linked to Taiwan death registry to construct an EOL study cohort from patients diagnosed with lung, esophageal, and colorectal cancer between 2008 and 2010 and died before December 31, 2012. We identified radiotherapy use from the National Health Insurance claims data. We conducted logistic regression to compare the EOL radiotherapy use between elderly (≥ 65) and non-elderly cancer patients while controlling for cancer type, gender, region, socioeconomic status, comorbidity, and year of death. We compared EOL medical costs between patients with and without radiotherapy; all costs were normalized to 2014 NT dollars and converted to USD via purchasing power parity index. We calculated the proportion of radiotherapy-related costs in total medical costs in EOL and quantified patients whose proportion was at the top 10% percentile as those with overly aggressive EOL radiotherapy and examine the associated factors with logistic regression.

Result(s): Our study cohort consisted of 26954 patients (15226 lung, 3032 esophageal, and 8696 colorectal cancer). EOL radiotherapy use was reported in 7.65% patients, with a higher percentage observed among non-elderly patients (11% vs. 6.15%, P<.0001). Non-elderly patients were more likely to receive EOL radiotherapy after controlling for confounders [adjusted odds ratio (AOR): 1.83, 95% CI 1.66 – 2.02]. Total EOL medical cost was higher among patients who received radiotherapy (Mean: $10410 vs. $6309, P<.0001). Among patients with EOL radiotherapy, overly aggressive care was found to be more common among nonelderly patients in unadjusted (11.74% vs. 8.57%, P=0.017) and adjusted analyses.

Conclusion(s): Our study found younger cancer patients are more susceptible to overly aggressive EOL care, suggesting that communications to initiate early palliative care are especially important among these patients for whom the true prognosis of disease may be masked by the desire to defeat cancer.


Sze-chuan Suen, MS1, Margaret L. Brandeau, PhD1 and Jeremy D. Goldhaber-Fiebert, PhD2, (1)Department of Management Science and Engineering, Stanford University, Stanford, CA, (2)Stanford Health Policy, Centers for Health Policy and Primary Care and Outcomes Research, Department of Medicine, Stanford University, Stanford, CA

Purpose: Patients on first-line tuberculosis (TB) treatment may not be cured if their TB strains are drug-resistant (DR).  However, the sputum-smear (SS) test used for patient monitoring cannot identify bacterial strain, so drug-sensitivity testing (DST) is required to identify such patients for alternative treatment.  Currently, India performs DST in the fourth month of treatment.  We determine the optimal time to administer DST and to identify the patterns of SS results that should prompt DST. If DST is administered too soon, many patients without DR TB will be unnecessarily tested; if administered too late, patients with DR TB may continue to transmit disease and decline in health.  It is critical to determine the optimal timing for DST because India is planning to adopt the fast but expensive Xpert technology system for DST, increasing the cost of unnecessary testing.

Method(s): We formulate a partially observed Markov decision process (POMDP) to determine the optimal timing of SS test information collection and DST. We calculate parameters such as patient response to treatment, dynamics while on treatment (the possibility of default or death), and discounted lifetime costs and health benefits using clinical studies and our previously published TB microsimulation model. We solve the POMDP using value iteration on a constrained feasible belief set.

Result(s): India's current policy appears suboptimal given relatively high national estimates of TB transmission. For these estimated values, DST should be administered to all patients upon initial TB diagnosis. After accounting for averted downstream transmission, we project that this testing sequence could save thousands of dollars per TB patient in discounted net monetary benefits. However, in settings where the risk of transmission is much lower, a patient's SS test result sequence can change the optimal DST timing.  Figure 1 shows the optimal patient testing path in an environment without TB transmission.

Conclusion(s):  India should revise its drug sensitivity testing protocol for the first-line national TB treatment program to provide DST at initial TB diagnosis in areas of average or high drug-resistant TB transmission, and may wish to consider individually tailored DST regimens in low transmission areas to reduce financial costs.

Figure 1: Optimal testing path for patients in no-transmission environments



Ming Tai-Seale, PhD1, Atul Gupta2 and Ellis Dillon, PhD1, (1)Palo Alto Medical Foundation Research Institute, Palo Alto, CA, (2)Stanford University and Palo Alto Medical Foundation, Palo Alto, CA

Elderly patients often receive care from multiple health care providers. Because they face many preference sensitive care decisions, it is critical that their medical records document their preferences in an accessible manner. Poor documentation of patient preferences in the electronic health record (EHR) may jeopardize the ability of health care providers to honor these preferences. Since 2010, the Palo Alto Medical Foundation (PAMF) established a palliative care program across its 4 divisions. We aim to characterize advance care planning (ACP) documentation practices and determine if the palliative care program changed ACP practices, using EHR data from 2005-2014. Both advance health care directives (AHCD) and physician orders for life-sustaining treatment (POLST) are considered ACP. 


This analysis included PAMF patients ≥65 between 2013-2014 with no EHR record of ACP before 2013. If their EHR’s Problem List has an ICD-9-code for ACP, it is considered accessible. Logistic regression analysis examined the relationship between having accessible ACP documentation and explanatory variables, including (1) type of serious illness (defined by the presence of those conditions listed in the National Committee for Quality Assurance Palliative Care/hospice measurement set); (2) travel distance to the nearest palliative care specialists; and (3) patient and physician characteristics.


Among 65,253 patients ≥65 who did not have any ACP prior to 2013, 10.24% had accessible AHCD, 0.82% had accessible POLST, 0.54% had accessible AHCD and POLST, leaving 88.4% without accessible ACP at the end of 2014. Among those with at least one serious illness, 12.09% had accessible AHCD, 5.42% had accessible POLST, 3.62% had both, leaving 78.87% without accessible ACP. Patients with chronic obstructive pulmonary disease are more likely to have AHCD (OR=1.094, p<0.01) and POLST (OR=1.215, p<0.01). Patients with brain cancer (OR=4.165, p<0.01), Esophageal cancer (OR=4.697, p<0.05) and debility (OR=1.923, p<0.01) are more likely to have accessible POLST. Male, Asian, Chinese, and Black patients are less likely to have accessible AHCD. The travel distance to palliative care specialists lowers the likelihood of having accessible AHCD (OR=0.904, p<0.01) and POLST (OR=0.866, p<0.01). 


Overwhelming majority of older patients’ EHR does not have accessible ACP, even those seriously ill. Systematic efforts are needed to eliminate gender and racial disparities and integrate palliative care into other areas of medical practice to enhance care for vulnerable elders.


Ariane Plaisance, Bsc, Université Laval, Québec, QC, Canada and Patrick Michel Archambault, CISSS CA, Secteur Alphonse-Desjardins (CHAU Hôtel-Dieu de Lévis), Lévis, QC, Canada

To assess critical care specialits and intensive care unit (ICU) patients' needs for better evidence and value based decision making prior to the development of a cardiopulmonary resuscitation (CPR) decision aid adjustable to patients' characteristics.


This study took place in the ICU at the Hôtel-Dieu de Lévis (Canada), a closed medical and surgical ICU with 18 beds and staffed by 5 critical care specialists. We conducted three weeks of observation of patients, family members, intensivists and other allied health professionals discussions about advance care planning. We specifically observed 5 dyads of attending intensivists and patients discussing the difficult choice between conducting cardiopulmonary resuscitation (CPR) or no CPR in the case of cardiac arrest. We also interviewed 5 intensivists about their needs for easier and better decision making between CPR and no CPR.

We then employed user centered design and rapid prototyping to explore different ways to explain the risks and benefits of CPR to patients. We also explored different ways to elicit patients' values and preferences. We interviewed 9 patients about the acceptability and relevance of the information presented. Discussions between intensivists and patients were recorded and a standardised observation grid was used to collect patients’ comments and sociodemographic data. Field notes, verbatim and content extracted from the observation grids were content-analyzed.

Our observations and rapid prototyping will inform the adaption of different existing decision aids in various formats (paper, video, web). We will house the different versions of our decision aid in a wiki that will enable future adjustments of our tool to various contexts and patients characteristics.


Our qualitative content analysis revealed that patients and their family members are most concerned about the risks of losing functional autonomy following successful CPR. However, they lack knowledge about the purpose of CPR, the survival rate and functional outcomes after CPR. We also observed a lag between the level of care documented in the patient’s chart and  their values, preferences and medical condition.


Basic understanding of what is a cardiac arrest, what is CPR and the risks and benefits of CPR is needed in order to reach a free and informed concern. Use of different formats of decision aid could improve advance care planning communication between intensivists and patients.


Dawn Fairlie, APRN, PhD, College of Staten Island, City University of New York, Staten Island, NY

The purpose of this study was to investigate the relationship between end of life terminologies and decisional conflict in surrogate decision makers using a convenience sample of 234 adults age 50 and older.  


After randomizing participants into two groups; and each received a vignette and personalized it with the name of a loved one. The vignettes varied only in the use of the words “Do Not Resuscitate (DNR)” and “Allow Natural Death (AND)”.  The Decisional Conflict Scale (DCS) was administered and demographic data were collected.


There was no difference in total DCS score based on AND and DNR versions.  Significant findings include the following:                                                                                                                            

  1. AND respondents perceived their decision as a good decision, and were eight times more likely to sign the document than DNR participants. This indicated that framing influences surrogate decision making at the end of life.                                                                                                      
  2. Experienced decision makers (EDMs) emerged as a discreet group and had lower mean total DCS scores and lower mean subscores. This indicated that prior experience is an important aspect of end of life decision making.                                                                                                                
  3. AND and EDM participants were more likely to perceive their decision as good and were more likely to be to be sure of their decision. This indicated that experienced decision makers respond more favorably to the words Allow Natural Death.
  4. Respondents to the DNR version were likely to not sign or postpone signing.  
  5. Respondents to the AND version were more likely to make a decision.                                          
  6. Participants were more likely to withdraw from the study when the term DNR was used.   
  7. Participants who reported frequent religious attendance were more satisfied with their decisions.


The results of this preliminary study of primarily Caucasian, fairly well educated, predominantly affluent respondents indicate that the completion of ADs in this demographic population can be influenced by information framing, prior experience with EOL decision-making. Communication modalities, information framing, and decision aids can be critical aspects of EOL communication.  A larger, multi-site study with a more diverse sample would help to verify the results of this study.


Huiqin Yang, PhD, MSc.,, University of York, York, United Kingdom and Carl Thompson, PhD, University of Leeds, Leeds, United Kingdom
Purpose: ,

Time constraint is a significant factor that influences clinical judgement and decision making. This study aims to evaluate nurses’ diagnostic reasoning policies and performance under time constraint in a simulated acute care setting.

Method(s): ,

Ninety-seven nurses were exposed to 25 clinical scenarios under time constraint and no time constraint conditions in a simulated acute care setting.  In 12 of the scenarios only 20 seconds per judgement was allowed, in the other 13 scenarios no time constraint existed. The proportion of correct judgments in both situations was calculated. The logistic regression modelling was used to assess the relationship between their judgments and information cues utilised.  Relative weights were calculated to capture the degree of attention paid to particular information cues.


There was no significant difference in the proportion of correct judgments between the time constraint and no time constraint conditions. However, time constraint significantly impacted on the reasoning policies used. Results showed that nurses used fewer cues to reach their clinical judgements under time constraint, with the relative weight of heart rate being much smaller in the time constraint model. 


Time constraint had a significant impact on nurses’ diagnostic reasoning policies but not outcomes. Nurses tended to use less information cues to reach their judgements under time pressure without the expense of reasoning performance. Nurses showed adaptive reasoning when making clinical judgements under time constraints.