Measuring Families’ Perceptions of Care Across a Health Care System: Preliminary Experience with the Family Assessment of Treatment at End of Life Short Form (FATE-S)

Article Outline

• Abstract
• Introduction
• Methods
  • Survey Development
  • Sample and Setting
  • Data Collection
  • Psychometric Analysis
  • Nonresponse Bias
  • Validity
• Results
  • Psychometric Analysis
  • Nonresponse Bias
  • Validity
• Discussion
• Disclosures and Acknowledgments
• References
• Copyright

Abstract 

Context

Because the Family Evaluation of Treatment at End of Life (FATE) survey was too long for routine use in the Veterans Administration (VA) health care system to measure quality of care, a shorter instrument was developed.

Objectives

To evaluate the short version of the FATE survey for use as a nationwide quality measure in the VA health care system.

Methods

Fifty-one VA medical centers, including acute and long-term care, participated in this nationwide telephone survey. Family members of the patients were eligible if the patients died in a participating facility. One family member per patient was selected from medical records using predefined eligibility criteria and invited to participate. The survey consists of 14 items describing key aspects of the patient’s care in his or her last month of life, one global rating, and two open-ended questions for additional comments.

Results

Interviews were completed with 2827 family members. Overall, the survey showed excellent psychometric characteristics, with good homogeneity (e.g., Cronbach’s α=0.84) and strong evidence of discriminant validity. Two survey items have been targeted for quality improvement efforts in multisite collaboratives.

Conclusion

Surveys of surrogates offer an important source of quality data that can be used to improve the quality of end-of-life care and promote accountability.

Key Words: End-of-life care, quality improvement, measurement, veterans

Back to Article Outline

Introduction 

A growing body of research has identified numerous problems with end-of-life care in the United States,¹, ², ³ and the National Institutes of Health State-of-the-Science Conference on Improving End-of-Life Care highlighted many elements of care that are both effective and underused.⁴ For instance, symptom management, advance care planning, caregiver concerns, and continuity of care are often inadequate.⁵ In addition, the management of pain and dyspnea and provision of emotional support could be improved.³, ⁶ Finally, health care providers are often unaware of patients’ preferences for life-sustaining treatment,⁷, ⁸ and patients and families may feel that they do not receive enough information about the patient’s illness and treatment options.³, ⁹, ¹⁰ These problems, in turn, have led to calls for coordinated efforts to measure and improve the quality of end-of-life care.⁴, ⁵, ¹¹, ¹², ¹³

Unfortunately, large-scale quality measurement efforts have been slow to develop. Although the National Mortality Followback Study provided a valuable glimpse of families’ experiences with end-of-life care across the country, its funding was not renewed.¹⁴, ¹⁵ In addition, although efforts like the Family Evaluation of Hospice Care include large numbers of patients, it is limited to hospice organizations, and its use is currently voluntary.¹⁶

The lack of a widely used tool prompted the development of a research tool, the Family Assessment of Treatment at End of Life (FATE), to be used in the Veterans Administration (VA) health care system. The FATE was designed to measure quality of care for quality improvement and research purposes and has undergone previous testing and validation.¹⁷, ¹⁸, ¹⁹, ²⁰ However, it was felt to be too long for routine use as a large-scale quality measure, and a shorter instrument was developed for nationwide use.

In this article, we report the initial experience with the short version of the FATE (FATE-S) as a quality measure, including its response characteristics, susceptibility to nonresponse bias, and validity. We also describe the VA’s Center for Performance Reporting and Outcome Measurement to Improve the Standard of care at End-of-life (PROMISE) and its implementation of the FATE-S. We will conclude by suggesting ways in which the VA’s experience with this measurement effort offers lessons for measurement and accountability to non-VA health care systems.

Back to Article Outline

Methods 

Survey Development 

The initial survey version was developed as the 34-item FATE, whose development, validity, and psychometric properties have been described elsewhere.¹⁷, ¹⁸ To create a survey that could be used on a national scale without undue costs or burdens on family members, we created a shorter 16-item version (the FATE-S) for use as a national quality measure. Thirteen specific items and one global assessment were selected from the FATE with input from a national panel based on their content, psychometric characteristics, and potential usefulness in driving quality improvement. Specifically, panelists were asked to agree on those items that provide a meaningful reflection of quality and that health care providers could reasonably be expected to improve. We also added two open-ended questions that ask family members to describe positive and negative impressions of the care that the patient received.

Stata statistical software (Stata for Windows, Version 10.0; StataCorp LP, College Station, TX) was used for all analyses described below. Because this project was conducted as part of the VA’s operations, institutional review board (IRB) approval was not obtained for data collection. An application for secondary use of data for publication was granted an exemption by the IRB of the Philadelphia VA Medical Center, where the PROMISE Center is located.

Sample and Setting 

Patients who died in a participating VA medical center were identified using a data extraction algorithm in the VA’s electronic medical record (EMR) system. We excluded those patients who died within 24 hours of admission and/or in the emergency department, unless they had been admitted to a participating hospital or nursing home in the preceding month. We also excluded patients who died as a result of suicide or accident. During infrequent periods when the number of deaths exceeded the interviewers’ capacity to conduct interviews, patients were selected at random for omission from the sample. For each patient, the potential respondent (generally a family member) was identified in the following order of priority: 1) patient’s next of kin, 2) primary contact named in the EMR, and 3) individual named as durable power of attorney for health care.

Data Collection 

Approximately four weeks after the patient’s death, family members were sent a letter that described the survey and provided a toll-free telephone number that they could use to opt out. Beginning approximately two weeks later, interviewers made up to three attempts to contact family members, including at least one attempt after 5pm local time or on the weekend. We excluded family members who did not have a working telephone number, those who did not speak English or Spanish, and those with hearing impairment or other health conditions that precluded a telephone interview. We also excluded family members who said that they could not evaluate the care that the patient received in the last month of life, but we offered them the opportunity to identify a more knowledgeable informant, who was then contacted instead. Families were informed that participation in the survey was voluntary and no compensation would be provided.

Because one participating facility had a high prevalence of family members who preferred to complete surveys in Spanish, we created a Spanish version of the FATE-S. This version was translated and back translated by two individuals. The original and back-translated English versions were compared, and discrepancies were resolved by consensus among the two translators and the PROMISE Center staff.

Family members who were reached by telephone were given the option of completing the survey immediately or scheduling a time that was more convenient. At the conclusion of the survey, families were asked if they would like to be referred to a counselor for bereavement support, a hospital representative for questions about veterans’ benefits, or for other questions or concerns. Medical record reviews were used to determine eligibility and define key veteran characteristics (e.g., age, ethnicity, and site of death) and processes of care (e.g., palliative care consult and chaplain visits).

If the interviewer was still unable to reach the family member after three attempts, we sent a mail survey that included the same items as the phone survey. Surveys were sent with a cover letter and a self-addressed prepaid return envelope. Each survey was labeled with a unique number to allow matching of the survey to the patient. In addition, those family members who were excluded from phone interviews because of hearing impairment or other health conditions that precluded a telephone interview also were sent a mail survey.

Psychometric Analysis 

Each of the 14 specific items (Table 1) is coded as either 1 or 0, reflecting the best possible answer vs. all others. For instance, the item that assesses whether providers spoke in an understandable way is scored as 1 if providers “always” spoke in an understandable way and as “0” if the family said that providers “usually,” “sometimes,” or “never” spoke in an understandable way. In addition, an extra screening item is used for the pain item, which contributes to the FATE-S score if a patient had any pain in the last month of life and if the family was able to assess the patient’s pain management during that period. However, if a family member is unsure whether a patient had pain, or is uncertain how often the pain made the patient uncomfortable, the item is treated as missing.

Table 1. Survey Items

To derive the FATE-S score, each patient’s “best possible” responses are summed and divided by the number of nonmissing responses. Because patients have different numbers of nonmissing responses, the denominators for these proportions vary among patients. However, scores are only calculated for surveys in which at least 10 items are completed. The FATE-S score is expressed as a proportion, standardized on a 0 to 100 scale, in which a higher number indicates better perceptions of care (i.e., the proportion of FATE-S items for which the patient received the best possible care).

Several techniques were used to evaluate the usefulness of the FATE-S as a quality measure. First, we examined the survey’s floor and ceiling effects and distribution of responses for each item. We also evaluated the survey’s homogeneity using Cronbach’s alpha (α). As a test of criterion validity, we examined the association between scores for each of the 14 specific items and the global assessment using the rank sum test.

Nonresponse Bias 

Next, we looked for patient and family characteristics associated with survey completion that might create nonresponse bias. To do this, we evaluated the characteristics in Table 2 as potential predictors of survey completion. Next, those variables whose association with response had a P-value <0.25 were considered for inclusion in a multivariable logistic regression model in which the dependent variable was survey completion. That model was used to calculate the likelihood of a completed survey for each patient in the sample, after considering all variables in the model. We used these probabilities to calculate a weight for each patient that corresponded to the inverse of the probability that a survey would be completed for that patient. Thus, patients who were underrepresented in the sample (e.g., African Americans) were assigned a higher weight to correct for their lower response rate. We checked the appropriateness of the model by comparing the weighted sample characteristics with those of the entire population.

Table 2. Patient and Family Characteristics (n=2827)

Characteristics	n (%)
Patient’s age
Mean (range)	74 (21–101)
Patient’s gender
Female	60 (2)
Patient’s ethnicity
White, non-Hispanic	2150 (76)
Nonwhite	601 (21)
Unknown	76 (3)
Respondent’s relationship
Spouse/partner	1158 (41)
Child	857 (30)
Sibling	410 (15)
Parent	86 (3)
Other	316 (11)
Site of death
Inpatient ward	655 (23)
Nursing home	341 (12)
Intensive care unit	643 (23)
Inpatient hospice unit	1169 (41)
Other	19 (1)
Palliative consult	1747 (62)
Selected diagnoses
Cancer	1204 (43)
Heart disease	737 (26)
Lung disease	852 (30)
Dementia	695 (25)
Region
New England	549 (20)
Northeast	720 (26)
Mid-Atlantic	297 (11)
Southeast	652 (23)
Midwest	334 (12)
Southwest	275 (10)

Validity 

Third, we evaluated the survey’s discriminant validity. That is, we assessed its ability to detect differences among groups that should have different scores. Specifically, we examined groups for which large differences in scores had been identified in previous analysis of the FATE (e.g., patients with/without a palliative care consult; patients with/without inpatient hospice care; patients who were/were not seen by a chaplain; and patients for whom a goals discussion was/was not documented). We used either linear or logistic regression models in which the outcome variables were either the FATE-S score or items scores, respectively. These models were weighted to account for nonresponse bias as described above. Models were clustered by facility and used robust jackknife estimates of standard errors.

Back to Article Outline

Results 

Medical record review identified a total of 5513 patients who died in one of 51 participating VA facilities between July 2008 and March 2009. Of these, 274 were selected at random for exclusion to manage interviewer workload. An additional 684 were determined to be ineligible. (Two patients committed suicide, 565 had missing or inaccurate contact information, and 94 died within 24 hours of admission. Two respondents did not speak English or Spanish, and 21 respondents had health conditions that made the interview impossible.) Of the remaining patients (n=4555), 2359 family members completed a telephone survey (52% of those who were eligible) and 468 completed a mail survey. Telephone surveys required approximately five to 10 minutes to complete.

Mail surveys and telephone interviews had similar means (62 vs. 63, respectively) and medians (70 vs. 69, respectively). Mail and telephone results also had identical ranges (0–100 for both) and very similar interquartile ranges (41–85; 46–85). Therefore, the results of both mail surveys and telephone interviews were included in the analysis, producing a final sample of 2827 (62% of those patients who were eligible).

Psychometric Analysis 

FATE-S scores were slightly skewed, with a predominance of higher scores. The mean and median were 63 and 69 of a possible 100, respectively, with the following percentiles: (10th percentile: 23; 25th: 45; 75th: 85; and 90th: 92). Items showed a wide distribution of responses and had means between 29% and 86% (corresponding to the proportions of respondents who indicated the patient received the best possible care for each item).

The average missing and “unknown” rate for FATE-S items was 5% (n=147) (Table 1). The item with the lowest missing rate asked families whether they were kept informed (1%; 26). The highest was observed with the pain item 22% (n=599), typically because families were unable to assess the severity of the patient’s pain. This item was more likely to be missing among patients with noncancer diagnoses (402 of 1640 vs. 197 of 1211, 24% vs. 16%; χ² P<0.001), but its completion rate was unrelated to other patient characteristics (e.g., ethnicity, age, and site of death) or the relationship of the family member to the patient. The FATE-S had good homogeneity (Cronbach’s α=0.84) well above the accepted cutoff of 0.70 for scales used for between-group comparisons.²¹

Nonresponse Bias 

Patient and family characteristics for completed interviews are described in Table 2. There were several differences between respondents (n=2827) and nonrespondents (n=2254). In a multivariable logistic regression model, nine independent predictors of response were identified: family relationship (spouse vs. other) (odds ratio [OR]=1.41; 1.26–1.58; P<0.001), patient age, expressed in 10-year increments (OR 1.14; 1.07–1.22; P<0.001), race (non-Hispanic white vs. other) (OR=1.45; 1.27–1.65; P<0.001), dementia (OR=1.34; 1.12–1.60; P=0.001), death in a hospice unit (OR=1.24; 1.10–1.40; P<0.001), death in a nursing home (OR=1.39; 1.03–1.86; P=0.029), presence of a Do Not Resuscitate (DNR) order (OR=1.26; 1.07–1.48; P=0.006), a documented chaplain visit (OR=1.22; 1.07–1.39; P=0.004), and documented discussion of the patient’s goals with a family member (OR=1.49; 1.21–1.81; P<0.001).

This model was used to calculate each patient’s probability of response and its inverse, the sample weight (range 1.23–5.45; interquartile range 1.58–2.10). These weights were checked by applying them to the sample and then by comparing the weighted sample with the entire population of eligible patients. Most weighted sample characteristics were identical to those of the entire sample (e.g., respondent relationship, patient ethnicity, presence of dementia, documented discussion with a family member, and patient age). Other weighted sample characteristics differed only slightly from the entire sample (e.g., death in a hospice unit: 38% vs. 35%; death in long-term care: 11% vs. 10%; presence of a DNR order: 87% vs. 86%; and documentation of a chaplain contact: 32% vs. 30%). These weights were used in tests of the FATE-S’s validity described below, except as noted.

Validity 

We examined validity in three ways. First, we used a linear regression model, clustered by facility (e.g., 51 clusters) and weighted to account for nonresponse bias, to examine the association between the FATE-S score (based on all items except the global item) and the single-item global rating. This five-point global rating has been widely used in other instruments³, ²², ²³ and provides a test of criterion validity. The FATE-S score was strongly associated with the global rating (β=0.17, 95% confidence interval [CI]:0.16, 0.17; P<0.001). Significant associations were also present in logistic regression models between the global rating and each of the FATE-S items (P<0.001 for all) (Table 1).

Next, we looked for differences in scores across facilities. The FATE-S score showed considerable variation among facilities (unweighted range 32–88; interquartile range 58–67). The overall and interquartile ranges (unweighted) among long-term care facilities (32–88; 62–73, respectively) were slightly larger than those among acute care hospitals (46–77; 57–66).

Finally, we used bivariate regression to evaluate the discriminant validity of the FATE-S by comparing scores among groups for which the quality of palliative care would be expected to differ. For instance, as expected, patients who were seen by a palliative care consult service in the last month of life had significantly better FATE-S scores compared with those who did not (mean 64 vs. 60; β=0.04; 95% CI:0.01, 0.06; P=0.005). Similarly, patients who died in a hospice unit had higher scores compared with those who died in other settings (66 vs. 61; β=0.06; 95% CI:0.02, 0.09; P=0.001), and those who died in an acute care ward had lower scores compared with those who died in other settings (58 vs. 65; β=−0.06; 95% CI: −0.09, −0.03; P<0.001). Patients who died with a DNR order had higher scores than other patients (64 vs. 54; β=0.10; 95% CI: 0.06, 0.14; P<0.001). Patients also had better scores when a discussion about the patient’s goals with a family member was documented (64 vs. 55; β=0.10; 95% CI: 0.06, 0.13; P<0.001).

Because the FATE-S was designed to assess the impact of quality improvement activities that often target a single aspect of care, we also examined the discriminant validity of selected items. In this analysis, we used logistic regression models and focused on items for which we could easily collect data from the medical record about processes of care that should be associated with an item’s score. For instance, families for whom at least one chaplain visit was documented were more likely to report that they received enough spiritual support (62% vs. 50%; β=0.42; 95% CI: 0.24, 0.60; P<0.001). Families for whom a chaplain visit was documented were also more likely to report that they received enough emotional support before the patient’s death (64% vs. 57%; β=0.26; 95% CI: 0.08, 0.44; P=0.005) and after the patient’s death (66% vs. 60%; β=0.28; 95% CI: 0.08, 0.48; P=0.007).

When there was a documented discussion with a family member about the patient’s goals, families were more likely to report that they received enough information about the patient’s condition and treatment (69 vs. 50; β=0.78; 95% CI: 0.48, 1.11; P<0.001). They were also more likely to say that they knew what to expect as the patient neared the end of life (71 vs. 48; β=1.00; 95% CI: 0.69, 1.31; P<0.001). They were also more likely to say that the patient’s health care providers took enough time to listen (74 vs. 64; β=0.47; 95% CI: 0.16, 0.80; P=004).

Back to Article Outline

Discussion 

Increasing recognition of numerous problems with end-of-life care has led to calls for a broad approach to quality measurement.¹¹, ¹², ¹³ The FATE-S and its associated quality measurement tools represent one novel way to measure families’ perspectives of the quality of end-of-life care that is delivered across a national health care system. Two results in particular support the use of the FATE-S as a quality measure.

First, the survey offers strong face validity. Beginning with the results of a literature review and extensive qualitative family interviews, its content was refined through iterative reviews by an expert panel to select those items that would be most useful in driving quality improvement. Specifically, the panel was asked to select those items that reflected high-quality care and which provided actionable information to guide improvement. The same procedures of review, together with extensive cognitive testing, were used to optimize item wording.

Second, the results reported here support the FATE-S’s discriminant validity. For instance, the palliative consults and hospice units in this sample are associated with higher FATE-S scores. Similarly, there is clear evidence of associations between processes of care (e.g., chaplain visits) and FATE-S items (e.g., families’ perceptions of spiritual support). Together, these results suggest that the FATE-S may be useful in defining variations in quality and guiding improvement efforts.

Despite preliminary evidence of the usefulness of the FATE-S, this study has three limitations that should be noted. First, the FATE-S has been evaluated only among veterans, who comprise an older population that is almost entirely men. Therefore, further testing would be needed if the FATE-S were to be used in other settings.

Second, as is true of many surveys of bereaved family members,¹, ⁹, ²⁴, ²⁵, ²⁶, ²⁷ this study had a modest response rate, which creates the possibility of selection bias. However, this is not necessarily a concern, particularly when, as in this case, a great deal is known about the potential sample. This allows surveys to be weighted to account for many sources of nonresponse bias including patient and family characteristics and site of care, reducing the risk of significant sources of nonresponse bias. However, as with any survey, it is still possible that patients for whom interviews were not conducted may differ with respect to other unmeasured characteristics.

Third, techniques that are often used in survey development, such as factor analysis and item response theory, were not used in the present study. Although factor analysis was used to develop the original FATE instrument, the FATE-S items were selected based on expert opinion. This clinimetric approach (vs. a purely psychometric approach) was felt to be the best way to develop an instrument that would be accepted by clinicians. Nevertheless, further research is needed to define the FATE-S’s factor structure and better understand relationships among its items.

The VA has made a commitment to use the FATE-S to improve end-of-life care throughout the health care system in several ways. First, it will be used by the VA’s Comprehensive End of Life Care Implementation Center to drive quality improvements at the patient care level based on proven practices as identified by bereaved family survey results. The FATE-S is being used as the driver and main outcome measure for a series of multisite quality improvement collaboratives, directed by another author (C. L.). These collaboratives have focused on improving scores for a single FATE-S item at a time and typically involve 20–30 facilities. Two collaboratives have been initiated in 2009, and two more are scheduled for 2010. Second, there is the potential for a set of quality indicators and data extraction procedures created by the Quality Improvement Resource Center, directed by one of the authors (K. L.), to be used to evaluate needs and priorities for improving the quality of end-of-life care throughout the VA. Third, the FATE-S is the key measure of the impact of the VA’s nationwide effort to build palliative care infrastructure, including palliative care teams and inpatient units in the VA system.

The FATE-S offers an important source of quality data that can be used to improve the end-of-life care of a large population of patients in an integrated health care system. It has, therefore, a key element of the VA’s national efforts in this area. However, these results can and should be complemented by other data sources. Indeed, after-death surveys are only one element, albeit an important one, of a comprehensive quality measurement program.¹³ Other potential data sources include administrative data, prospective data from patients and families, and data that define structures of care. By collecting these types of data in an organized and rigorous way, the VA has the potential to become a model of the kind of comprehensive measurement efforts that have been proposed for national and international use.¹³

Back to Article Outline

Disclosures and Acknowledgments 

This material is based on the work supported by the Department of Veterans Affairs, Veterans Health Administration, Office of Research and Development, HSR&D. The authors declare no conflicts of interest.

The contents of this work do not represent the views of the Department of Veterans Affairs or the U.S. government.

Back to Article Outline

References 

1. Hamel MB, Phillips RS, Davis RB, et al. Outcomes and cost-effectiveness of initiating dialysis and continuing aggressive care in seriously ill hospitalized adults. SUPPORT Investigators. Study to Understand Prognoses and Preferences for Outcomes and Risks of Treatments. Ann Intern Med. 1997;127:195–202
   • View In Article
   • MEDLINE
2. In:  Field MJ,  Cassell CK editor. Approaching death. Improving care at the end of life. Washington, DC: National Academy Press; 1997;Financial and Economic Issues in End-Of-Life Care (pp.154–87)
   • View In Article
3. Teno J, Clarridge B, Casey V, et al. Family perspectives on end-of-life care at the last place of care. JAMA. 2004;291:88–93
   • View In Article
   • CrossRef
4. National Institutes of Health . State-of-the-Science conference statement on improving end-of-life care. In: State-of-the-science conference on improving end-of-life care. Bethesda, MD: NIH; 2004;Available from: http://consensus.nih.gov/2004/2004EndOfLifeCareSOS024html.htm
   • View In Article
5. Lorenz KA, Lynn J, Dy SM, et al. Evidence for improving palliative care at the end of life: a systematic review. Ann Intern Med. 2008;148:147–159
   • View In Article
6. Tarzian AJ, Hoffman DE. Barriers to managing pain in the nursing home: findings from a statewide survey. J Am Med Dir Assoc. 2004;5:82–88
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (112 KB)
   • CrossRef
7. Teno JM, Hakim RB, Knaus WA, et al. Preferences for cardiopulmonary resuscitation: physician-patient agreement and hospital resource use. The SUPPORT Investigators. J Gen Intern Med. 1995;10:179–186
   • View In Article
   • MEDLINE
   • CrossRef
8. Campbell ML, Guzman JA. Impact of a proactive approach to improve end-of-life care in a medical ICU. Chest. 2003;123:266–271
   • View In Article
   • MEDLINE
   • CrossRef
9. Baker R, Wu AW, Teno JM, et al. Family satisfaction with end-of-life care in seriously ill hospitalized adults. J Am Geriatr Soc. 2000;48:S61–S69
   • View In Article
   • MEDLINE
10. Lynn J, Teno J, Phillips R, et al. Perceptions by family members of the dying experience of older and seriously ill patients. Ann Intern Med. 1997;126:97–106
    • View In Article
    • MEDLINE
11. Singer PA, Wolfson M. “The best places to die”. BMJ. 2003;327:173–174
    • View In Article
    • CrossRef
12. Singer PA, Bowman KW. Quality care at the end of life. BMJ. 2002;324:1291–1292
    • View In Article
    • CrossRef
13. Casarett DJ, Teno J, Higginson I. How should nations measure the quality of end-of-life care for older adults? Recommendations for an international minimum data set. J Am Geriatr Soc. 2006;54:1765–1771
    • View In Article
    • MEDLINE
    • CrossRef
14. Weitzen S, Teno JM, Fennell M, Mor V. Factors associated with site of death: a national study of where people die. Med Care. 2003;41:323–335
    • View In Article
    • MEDLINE
    • CrossRef
15. Teno J, Clarridge B, Casey V, et al. Family perspectives on end-of-life care at the last place of care. JAMA. 2004;291:88–93
    • View In Article
    • CrossRef
16. Connor SR, Teno J, Spence C, Smith N. Family evaluation of hospice care: results from voluntary submission of data via website. J Pain Symptom Manage. 2005;30:9–17
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (106 KB)
    • CrossRef
17. Casarett D, Pickard A, Bailey FA, et al. A nationwide VA palliative care quality measure: the family assessment of treatment at the end of life. J Palliat Med. 2008;11:68–75
    • View In Article
    • CrossRef
18. Casarett DJ, Pickard AP, Bailey FA, et al. Important aspects of end-of-life care for veterans: implications for measurement and quality improvement. J Pain Symptom Manage. 2008;35:115–125
    • View In Article
19. Casarett DJ, Pickard AP, Bailey FA, et al. Do palliative care consultations improve patient outcomes?. J Am Geriatr Soc. 2008;56:593–599
    • View In Article
    • CrossRef
20. Finlay E, Shreve S, Casarett D. Nationwide veterans affairs quality measure for cancer: the family assessment of treatment at end of life. J Clin Oncol. 2008;26:3838–3844
    • View In Article
    • CrossRef
21. Nunnally JC. Psychometric theory. 2nd ed.. New York: McGraw-Hill; 1978;
    • View In Article
22. Teno JM, Claridge B, Casey V, Edgman-Levitan S, Fowler J. Validation of toolkit after-death bereaved family member interview. J Pain Symptom Manage. 2001;22:752–758
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (62 KB)
    • CrossRef
23. Connor SR, Tecca M, LundPerson J, Teno J. Measuring hospice care: the National Hospice and Palliative Care Organization National Hospice Data Set. J Pain Symptom Manage. 2004;28:316–328
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (122 KB)
    • CrossRef
24. Patrick DL, Engelberg RA, Curtis JR. Evaluating the quality of dying and death. J Pain Symptom Manage. 2001;22:717–726
    • View In Article
    • Abstract
    • Full Text
    • Full-Text PDF (210 KB)
    • CrossRef
25. Tolle SW, Tilden VP, Rosenfeld AG, Hickman SE. Family reports of barriers to optimal care of the dying. Nurs Res. 2000;49:310–317
    • View In Article
    • MEDLINE
    • CrossRef
26. Somogyi-Zalud E, Zhong Z, Lynn J, Hamel M. Elderly persons' last six months of life: findings from the Hospitalized Elderly Longitudinal Project. J Am Geriatr Soc. 2000;48:S131–S139
    • View In Article
    • MEDLINE
27. Hanson L, Danis M, Garrett J. What is wrong with end-of-life care? Opinions of bereaved family members. J Am Geriatr Soc. 1997;45:1339–1344
    • View In Article
    • MEDLINE