Abstract
Objective
To develop and validate a scale that assesses the readiness of gynecologic oncology patients to engage in advance care planning.
Methods
The Advance Care Planning Readiness Scale (ACPRS) was validated across three independent samples of gynecologic oncology patients. In Step I, patients underwent cognitive interviewing to determine if the scale items were comprehensible and applicable to patients. Based on this, modifications to the scale (addition, removal, and merger of items) were completed. In Step II, the revised scale was administered to a new sample of patients to assess scale reliability and validity. An exploratory factor analysis determined if the scale loaded onto unique factors. In Step III, the revised scale was administered to a third sample of patients and a confirmatory factor analysis was conducted to test the factor structure proposed in Step II. Associations between ACPRS score and completion of advance directives were evaluated.
Results
Based on patients’ responses, the original ACPRS used in Step I was modified to the ACPRS used in Step II. The final 8-item ACPRS is a valid, reliable (Cronbach’s α = 0.81) scale and has two primary factors. Women with medical power of attorney documents and living wills had higher ACPRS total scores than those who did not have these advance directives (p = 0.0030). Women with do-not-resuscitate orders (DNRs) had higher ACPRS total scores than women without DNRs (p = 0.0176).
Conclusions
The ACPRS is a valid and reliable 8-item scale that assesses the readiness of gynecologic oncology patients to discuss advance care planning issues.
Introduction
Communication regarding advance care planning is an important, but often neglected, component of caring for patients with cancer (1). Early advance care planning allows the patient to transition from active treatment of cancer to the management of the physical, spiritual, and emotional needs of the terminally ill patient. Advance care planning is defined as the activities that the patient participates in with her healthcare team and her family to determine her end-of-life care (EOLC) goals (2, 3). This planning includes the creation of advance directives such as a Medical Power of Attorney, Living Will, and Do Not Resuscitate (DNR) order and discussions between the patient, the healthcare team, and the patient’s family regarding the patient’s overall EOLC goals relating to the management of her disease and the transition to the end of life (2, 3).
Research demonstrates that early EOLC discussions can improve the likelihood that patients’ final wishes are honored at the end of life and may result in decreased distress for the patient (4). Early EOLC discussions are associated with a decreased number of invasive procedures and treatments at the end of life, resulting in increased quality of life (QOL) and decreased health-care costs (5–7). Both the American Society of Clinical Oncology (ASCO) and the National Comprehensive Cancer Network (NCCN) recommend that advance care planning discussions occur early in a cancer patient’s disease course (8, 9).
Patients experience differing levels of comfort discussing EOLC issues with their providers, making this an area that could benefit from interventions targeted at improving patient-provider communication. Specifically, the ability to objectively assess a patient’s interest in discussing EOLC issues could allow for better patient/provider communication by identifying those patients who are interested in and open to talking about their EOLC preferences. Unfortunately, no standardized scale currently exists to assist providers in assessing oncology patients’ willingness and readiness to discuss these topics. The goal of this study was to develop and validate a scale that assesses the readiness of gynecologic oncology patients to engage in EOLC discussions, which are a critical part of advance care planning.
Materials and Methods
The Advance Care Planning Readiness Scale (ACPRS) was validated across three independent samples (Steps I-III) of gynecologic oncology patients at The University of Texas MD Anderson Cancer Center. Patients were eligible to participate in the IRB approved study if they: 1) had a gynecologic malignancy, 2) were currently receiving treatment or in active surveillance at our institution, 3) had an ECOG (Eastern Cooperative Oncology Group) performance score of 2 or less, 4) were able to speak, read, and understand English, 5) were able to provide written informed consent. Patients meeting the eligibility criteria were identified using clinic templates and the electronic medical record. Eligible patients were then recruited from the Gynecology Clinic at their follow-up appointment. Patients were intentionally recruited to evenly represent women receiving treatment and those not in order to allow equal representation of patients undergoing treatment and those in surveillance. Participation was voluntary and without compensation. The anonymity of the participants was maintained throughout the study and patients’ attending physician (treating provider) had no knowledge of patient survey responses. All results were anonymized before statistical analyses. The three steps necessary for validation of a scale were conducted in the present study. Step I is the formation and development of the scale. Step II tests the reliability and validity of the scale and examines how many factors (or subcategories) the scale has. Step III confirms that the number of factors identified in Step II is correct. The methodology of each step in the scale validation is described below.
Step I Methodology
The initial ACPRS was a 13-item scale developed by generating a larger pool of items based on issues covered in the EOLC literature, physicians’ experiences with patients, and items pilot-tested by the investigators. Previous unpublished work by one of the authors of this protocol (AJB) pre-tested an end-of-life care readiness scale composed of two factors: Willingness to Discuss EOLC and the Acceptance of EOLC among 140 patients seen in a Gynecologic Oncology clinic in Baltimore, MD. This evaluation demonstrated good internal reliability within those two factors with a Cronbach’s α = 0.77 for the Willingness to Discuss EOLC factor and Cronbach’s α = 0.71 for the Acceptance of EOLC factor. Scale validation was not done during this study but it demonstrated the potential internal validity of this type of construct. Additional items were added to the pilot scale after discussion of important topics relating to end-of-life care with an expert panel of gynecologic oncologists, communication specialists, and survey development experts. The final pool of items included in the current study were reviewed and modified by this expert panel. The 13-item ACPRS had a seven-point Likert-type response scale (1 = strongly disagree, 7 = strongly agree). Higher scores on the ACPRS indicate increased overall readiness to discuss and make EOLC decisions.
Twenty-five patients were enrolled in Step I. Patients completed a background questionnaire regarding age, race/ethnicity, marital status, education, religion, and occupation. Details regarding the patient’s cancer, current treatment, and presence of DNR and Advance Directives were obtained from the medical record. Participants completed the self-administered 13-item ACPRS. Following scale completion, trained research personnel conducted a cognitive interview regarding the 13-item ACPRS using methods outlined by Rubin and Rubin (10). All interviews were recorded via a digital recorder and audited by the study staff who completed the cognitive interviews in order to collect data for the thematic analysis.
Step II Methodology
The modified 11-item scale created from Step I (see results) was tested for reliability and validity in Step II with an independent sample of 110 patients. Patients were asked to complete the Step I background questionnaire, the modified 11-item ACPRS, and three additional measures to assess convergent and divergent validity (detailed below). Details regarding the patient’s cancer were obtained from the medical record as in Step I. Patients received an addressed, stamped envelope and were asked to take the ACPRS home and complete it again one week after their initial participation in the study in order to assess test-retest reliability.
Questionnaires Used in Step II
PEACE Questionnaire
A validated 12-item measure composed of two subscales: a 7-item Struggle with Illness subscale and a 5-item Peaceful Acceptance subscale (11). All items are measured using a 4-point Likert type scale (1 = not at all, 4 = to a large extent). Higher scores indicate, respectively, increased struggle with illness or peaceful acceptance. The Struggle with Illness subscale has a significant positive association with advance care planning activities such as living will formation or healthcare proxy designation (11). The Peaceful Acceptance subscale has a significant inverse relationship with feeding tube usage at the end of life. (11) This scale was used to assess the convergent validity of the ACPRS. We hypothesized that the ACPRS would be positively associated with the Struggle with Illness and the Peaceful Acceptance subscales.
Templers’ Death Anxiety Scale (DAS)
A validated scale composed of 15 true/false questions that assess an individual’s level of anxiety surrounding death (12). A higher score indicates increased death anxiety. This scale was used to assess the convergent validity of the ACPRS. We hypothesized that the ACPRS would be negatively associated with the DAS.
Shortened Marlowe-Crowne Social Desirability Scale (MCSDS)
A validated scale composed of 10 true/false questions that assess an individual’s endorsement of socially desirable characteristics (13, 14). A higher score indicates increased concern for being viewed as having socially desirable characteristics. This scale was utilized to assess the divergent validity of the ACPRS because the two were not expected to be associated.
Step III: Methodology
A confirmatory factor analysis (CFA) of the ACPRS was conducted with an independent sample of 110 patients in order to verify the results obtained in Step II. Patients completed the Step I background questionnaire and the 11-item ACPRS. Details regarding the patient’s cancer were obtained from the medical record as in Step I.
Step I: Statistical Analyses
Thematic analysis was utilized to code the qualitative data from the cognitive interviews to determine which items should be modified (to enhance understanding), cut (due to irrelevance to patients), or added (due to importance to patients) to the 13-item ACPRS prior to beginning Step II.
Step II: Statistical Analyses
Cronbach’s alpha was calculated to assess internal reliability. An exploratory factor analysis (EFA) was conducted to determine the factors/constructs that the ACPRS measured. Convergent validity was assessed by correlating ACPRS scores with PEACE Struggle, PEACE Acceptance, and DAS scores. A Bonferroni correction was applied to control for Type I errors when assessing convergent validity; therefore a cutoff of p<0.0167 for correlations was used to assess convergence in order to achieve a familywise Type I error of 0.05.
Divergent validity was assessed by correlating the ACPRS with the MCSDS and testing HA: ρ = 0. If the 90% confidence interval for this correlation was wholly contained within this interval [−0.30, 0.30], we concluded that the scales diverged. Finally, test-retest reliability was assessed by correlating ACPRS scores with those taken a week later by a subset of participants (those who completed the scale within 7 days post-assessment).
Step III: Statistical Analyses
A confirmatory factor analysis (CFA) was conducted to verify that the results obtained from Step II and the EFA performed were reproducible (Figure 1). Model fit was evaluated by examining: 1) the root mean square error of approximation (RMSEA)(15), 2) the Tucker Lewis Index (TLI), also known as the Non-Normed Fit Index (NNFI), 3) the Comparative Fit Index (CFI)(16) and the standardized root mean square residual (SRMR) (18).
Figure 1. Model depicting standardized results from Confirmatory Factor Analysis of modified 8-item Advance Care Planning Readiness Scale.
Figure 1. Model depicting standardized results from Confirmatory Factor Analysis (CFA) of modified 8-item Advance Care Planning Readiness Scale (ACPRS). Item 1 in diagram corresponds with Item 2 of original scale; Item 2 in diagram corresponds with Item 4 of original scale. Item 3 above corresponds with previous Item 5; Item 4 above corresponds with previous Item 6, Item 5 above corresponds with previous Item 7; Item 6 above corresponds with previous Item 8; Item 7 above corresponds with previous Item 9; and Item 8 above corresponds with previous Item 11.
Results
Step I: Results
Patients from this sample were predominately white (84%) and had endometrial (44%) or ovarian cancer (40%) (Table 1). Participation rate was 93% (25/27). Most found the ACPRS clear and straightforward. A recurring theme suggested by patients was to add qualifiers to the questions, such as specifying what “medical treatment” meant (e.g., “CPR”) as well as using “if my cancer were terminal/untreatable” to specify when EOLC discussions or planning would occur. Using the information collected during the cognitive interviewing in Step I, the original 13-item ACPRS used in Step I was modified to form the 11-item ACPRS used in Step II (Supplementary Table 1).
Table 1.
Demographic and Clinical Summary of Steps I, II and III
| Step 1 N (%) |
Step 2 N (%) |
Step 3 N (%) |
p-value | ||
|---|---|---|---|---|---|
| Age | N | 25 | 110 | 110 | 0.8008 |
| Mean (SD) | 60.01 (10.47) | 58.82 (12.06) | 58.18 (14.19) | ||
| Median | 63 | 61 | 58 | ||
| Min - Max | 41 – 81 | 24 – 82 | 28 – 85 | ||
|
| |||||
| Race | White | 21 (84.00%) | 84 (76.36%) | 82 (74.55%) | 0.2474 |
| African American | 1 (4.00%) | 4 (3.64%) | 9 (8.18%) | ||
| Hispanic | 1 (4.00%) | 18 (16.36%) | 16 (14.55%) | ||
| Asian | 0 (0.00%) | 3 (2.73%) | 2 (1.82%) | ||
| Other | 2 (8.00%) | 1 (0.91%) | 1 (0.91%) | ||
|
| |||||
| Education | Elementary, HS, GED | 7 (28.00%) | 31 (28.18%) | 48 (43.64%) | 0.0441 |
| Associate, Undergraduate, Graduate | 18 (72.00%) | 79 (71.82%) | 62 (56.36%) | ||
|
| |||||
| Cancer Site | Ovarian | 10 (40.00%) | 51 (46.36%) | 59 (53.64%) | 0.2981 |
| Uterine | 11 (44.00%) | 38 (34.55%) | 26 (23.64%) | ||
| Cervical | 4 (16.00%) | 17 (15.45%) | 17 (15.45%) | ||
| Vulvar/Vaginal | 0 (0.00%) | 4 (3.64%) | 8 (7.27%) | ||
|
| |||||
| Stage | Not Staged | 0 (0.00%) | 12 (11.01%) | 13 (11.82%) | 0.7273 |
| Stage I | 7 (29.17%) | 31 (28.44%) | 27 (24.55%) | ||
| Stage II | 2 (8.33%) | 9 (8.26%) | 13 (11.82%) | ||
| Stage III | 11 (45.83%) | 43 (39.45%) | 39 (35.45%) | ||
| Stage IV | 4 (16.67%) | 14 (12.84%) | 18 (16.36%) | ||
| Unknown/Missing | 1 | 1 | 0 | ||
|
| |||||
| Cancer Management | Surveillance | 13 (52.00%) | 55 (50.00%) | 46 (41.82%) | 0.4125 |
| Primary/Recurrence | 12 (48.00%) | 55 (50.00%) | 64 (58.18%) | ||
|
| |||||
| Current Treatment | Current Treatment | 13 (52.00%) | 55 (50.00%) | 55 (50.00%) | > 0.9999 |
| No Treatment or Maintenance | 12 (48.00%) | 55 (50.00%) | 55 (50.00%) | ||
Step II Results
Patients from this sample were predominately white (76.36%) and had ovarian (46%) or endometrial cancer (35%) (Table 1). Participation rate was 74% (110/149).
Reliability
The initial assessment of reliability indicated that Items 1, 3 and 10 (from the modified 11-item scale administered in Step II) had low correlations with the total score (< 0.3). Accordingly, these items were removed from the ACPRS, resulting in the final 8-item scale (Table 2). Reliability of the modified 8-item scale was good with Cronbach’s α = 0.81 (Table 3). Reliability was assessed separately for patients with active disease and those in surveillance. Reliability in both cases was similarly good (Supplementary Tables 2 and 3). Test-retest reliability for the modified ACPRS was adequate (r = 0.73; 95% CI: 0.60 – 0.83) (Supplementary Table 4). The modified 8-item ACPRS was used for all subsequent analyses (Table 4).
Table 2.
8-item Advance Care Planning Readiness Scale (ACPRS)
| Item | Correlation with Total | Action Taken | On final, 8 item ACPRS? |
|---|---|---|---|
| Item 1. I have thought about whether or not I would want medical help to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover.* | 0.19 | Removed item | No |
| Item 2. It is important to tell my family whether or not I want medical help to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover. | 0.40 | Kept item | Yes |
| Item 3. I am uncomfortable thinking about whether or not I would want medical help to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover.* | 0.24 | Removed item | No |
| Item 4. I feel ok talking with my doctor about what medical help I would want to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover. | 0.48 | Kept item | Yes |
| Item 5. If I have tried many treatments and my cancer cannot be cured, I feel ok with my doctor talking to me about medications that will reduce the symptoms of my cancer but will not treat my cancer. | 0.68 | Kept item | Yes |
| Item 6. I feel ok with my doctor talking to me about hospice care. | 0.72 | Kept item | Yes |
| Item 7. I feel ok talking with my family about what medical help I would want to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover. | 0.57 | Kept item | Yes |
| Item 8. If my cancer is not responding to treatment and I am prevented from doing most of the things I enjoy, there is no point in continuing treatment. | 0.54 | Kept item | Yes |
| Item 9. I would prefer to live a shorter life with mild symptoms of my cancer than live a longer life with severe symptoms of my cancer. | 0.33 | Kept item | Yes |
| Item 10. I will want to try all cancer treatments possible even if my doctor tells me there is no cure for my cancer. * | 0.16 | Removed item | No |
| Item 11. I am peaceful when talking about the end of my life. | 0.37 | Kept item | Yes |
Item was removed from the Advance Care Planning Readiness Scale because it had a low correlation with the total score (<0.3).
Table 3.
Reliability for 8-item Advance Care Planning Readiness Scale: Step II Study Population
| Item | Mean (SD) | Correlation with Total | Cronbach’s α |
|---|---|---|---|
| Total Score | 44.86 (7.77) N: 99 Median: 46 Min - Max: 11 – 56 |
0.81 | |
| Item 2. It is important to tell my family whether or not I want medical help to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover. | 6.05 (1.66) | 0.41 | 0.80 |
| Item 4. I feel ok talking with my doctor about what medical help I would want to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover. | 5.86 (1.48) | 0.49 | 0.79 |
| Item 5. If I have tried many treatments and my cancer cannot be cured, I feel ok with my doctor talking to me about medications that will reduce the symptoms of my cancer but will not treat my cancer. | 6.01 (1.42) | 0.69 | 0.76 |
| Item 6. I feel ok with my doctor talking to me about hospice care. | 5.94 (1.32) | 0.76 | 0.75 |
| Item 7. I feel ok talking with my family about what medical help I would want to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover. | 6.12 (1.16) | 0.58 | 0.77 |
| Item 8. If my cancer is not responding to treatment and I am prevented from doing most of the things I enjoy, there is no point in continuing treatment. | 4.58 (1.92) | 0.52 | 0.78 |
| Item 9. I would prefer to live a shorter life with mild symptoms of my cancer than live a longer life with severe symptoms of my cancer. | 5.05 (1.59) | 0.30 | 0.81 |
| Item 11. I am peaceful when talking about the end of my life. | 5.25 (1.51) | 0.41 | 0.80 |
Table 4.
Advance Care Planning Readiness Scale’s Association with Demographics, Disease, and Advance Care Planning Documents (Step II Study Population, n=110*).
| Characteristic | N | Total Score | N | Willingness | N | Acceptance | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| Mean (SD) | Median (Min-Max) | Mean (SD) | Median (Min-Max) | Mean (SD) | Median (Min-Max) | |||||
| Age | p = 0.1339 r = 0.15 |
p = 0.3764 r = 0.09 |
p = 0.0755 r = 0.18 |
|||||||
|
| ||||||||||
| Race | p = 0.0582 | p = 0.0664 | p = 0.2154 | |||||||
| White | 76 | 45.67 (7.32) | 46.5 (26 – 56) | 76 | 45.67 (7.32) | 46.5 (26 – 56) | 76 | 45.67 (7.32) | 46.5 (26 – 56) | |
| Non-White | 23 | 42.17 (8.74) | 44.0 (11 – 54) | 23 | 42.17 (8.74) | 44.0 (11 – 54) | 23 | 42.17 (8.74) | 44.0 (11 – 54) | |
|
| ||||||||||
| Education | p = 0.3911 | p = 0.5256 | p = 0.4022 | |||||||
| Elementary, HS, GED | 28 | 43.79 (6.72) | 44.0 (27 – 56) | 28 | 43.79 (6.72) | 44.0 (27 – 56) | 28 | 43.79 (6.72) | 44.0 (27 – 56) | |
| Associate, Undergraduate, Graduate | 71 | 45.28 (8.15) | 46.0 (11 – 56) | 71 | 45.28 (8.15) | 46.0 (11 – 56) | 71 | 45.28 (8.15) | 46.0 (11 – 56) | |
|
| ||||||||||
| Cancer Site | p = 0.6967 | p = 0.9773 | p = 0.2900 | |||||||
| Ovarian | 50 | 45.62 (7.20) | 46.5 (27 – 56) | 50 | 45.62 (7.20) | 46.5 (27 – 56) | 50 | 45.62 (7.20) | 46.5 (27 – 56) | |
| Uterine | 30 | 44.07 (9.12) | 46.0 (11 – 56) | 30 | 44.07 (9.12) | 46.0 (11 – 56) | 30 | 44.07 (9.12) | 46.0 (11 – 56) | |
| Cervical | 16 | 43.56 (7.42) | 43.0 (28 – 56) | 16 | 43.56 (7.42) | 43.0 (28 – 56) | 16 | 43.56 (7.42) | 43.0 (28 – 56) | |
| Vulvar/Vaginal | 3 | 47.00 (5.20) | 44.0 (44 – 53) | 3 | 47.00 (5.20) | 44.0 (44 – 53) | 3 | 47.00 (5.20) | 44.0 (44 – 53) | |
|
| ||||||||||
| Stage | p = 0.5929 | p = 0.6418 | p = 0.7464 | |||||||
| Not Staged | 12 | 45.75 (7.36) | 47.5 (30 – 56) | 12 | 45.75 (7.36) | 47.5 (30 – 56) | 12 | 45.75 (7.36) | 47.5 (30 – 56) | |
| Stage I | 27 | 46.19 (6.95) | 46.0 (28 – 56) | 27 | 46.19 (6.95) | 46.0 (28 – 56) | 27 | 46.19 (6.95) | 46.0 (28 – 56) | |
| Stage II | 7 | 47.14 (4.95) | 45.0 (42 – 55) | 7 | 47.14 (4.95) | 45.0 (42 – 55) | 7 | 47.14 (4.95) | 45.0 (42 – 55) | |
| Stage III | 41 | 43.49 (9.16) | 45.0 (11 – 56) | 41 | 43.49 (9.16) | 45.0 (11 – 56) | 41 | 43.49 (9.16) | 45.0 (11 – 56) | |
| Stage IV | 11 | 44.36 (6.23) | 45.0 (34 – 56) | 11 | 44.36 (6.23) | 45.0 (34 – 56) | 11 | 44.36 (6.23) | 45.0 (34 – 56) | |
|
| ||||||||||
| Current Treatment | p = 0.9781 | p = 0.4817 | p = 0.2996 | |||||||
| Surveillance | 49 | 44.84 (7.66) | 46.0 (11 – 56) | 49 | 30.37 (5.41) | 31.0 (5 – 35) | 49 | 14.47 (4.07) | 15.0 (5 – 21) | |
| Current Treatment | 50 | 44.88 (7.95) | 46.0 (26 – 56) | 50 | 29.60 (5.40) | 30.0 (13 – 35) | 50 | 15.28 (3.65) | 16.0 (6 – 21) | |
|
| ||||||||||
| Have DNR | p = 0.0176 | p = 0.0826 | p = 0.0322 | |||||||
| Yes | 35 | 47.09 (5.60) | 47.0 (30 – 56) | 35 | 31.09 (3.60) | 32.0 (20 – 35) | 35 | 16.00 (3.16) | 16.0 (9 – 21) | |
| No | 64 | 43.64 (8.53) | 44.0 (11 – 56) | 64 | 29.38 (6.10) | 30.0 (5 – 35) | 64 | 14.27 (4.10) | 14.0 (5 – 21) | |
|
| ||||||||||
| Have Living Will or Medical Power of Attorney | p = 0.0030 | p = 0.0328 | p = 0.0031 | |||||||
| Yes | 50 | 47.12 (6.79) | 47.5 (27 – 56) | 50 | 31.12 (4.81) | 32.0 (13 – 35) | 50 | 16.00 (3.27) | 16.0 (9 – 21) | |
| No | 49 | 42.55 (8.09) | 44.0 (11 – 56) | 49 | 28.82 (5.74) | 30.0 (5 – 35) | 49 | 13.73 (4.12) | 14.0 (5 – 21) | |
Exploratory Factor Analysis (EFA)
Principal components factoring was used to extract factors from the data. An examination of the scree plot indicated that two factors would be sufficient to explain 70% of the variability in the data, suggesting that the 8-item ACPRS represents a two-factor scale (Table 5). An examination of the rotated factors indicated that Items 2, 4, 5, 6 and 7 of the 8-item ACPRS loaded heavily onto Factor 1 (Willingness to Discuss EOLC) and Items 8, 9 and 11 loaded heavily onto Factor 2 (Acceptance of EOLC). With the exception of Item 11, uniqueness ranged from 0.19 to 0.32. Accordingly, each item other than Item 11 was assumed to measure a unique component of the two factors.
Table 5.
Factor loadings for final version of Advance Care Planning Readiness Scale after rotation (Step II Study Population, n=110).
| Item | Factor1 (Willingness to Discuss EOLC*) | Factor2 (Acceptance of EOLC*) | Uniqueness |
|---|---|---|---|
| Item 2. It is important to tell my family whether or not I want medical help to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover. | 0.91 | −0.21 | 0.27 |
| Item 4. I feel ok talking with my doctor about what medical help I would want to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover. | 0.83 | −0.02 | 0.32 |
| Item 5. If I have tried many treatments and my cancer cannot be cured, I feel ok with my doctor talking to me about medications that will reduce the symptoms of my cancer but will not treat my cancer. | 0.75 | 0.25 | 0.24 |
| Item 6. I feel ok with my doctor talking to me about hospice care. | 0.73 | 0.33 | 0.19 |
| Item 7. I feel ok talking with my family about what medical help I would want to keep my heart and lungs working (cardiopulmonary resuscitation (CPR) or letting a machine breathe for me) if I were unconscious and unlikely to recover. | 0.83 | 0.03 | 0.29 |
| Item 8. If my cancer is not responding to treatment and I am prevented from doing most of the things I enjoy, there is no point in continuing treatment. | 0.13 | 0.77 | 0.32 |
| Item 9. I would prefer to live a shorter life with mild symptoms of my cancer than live a longer life with severe symptoms of my cancer. | −0.16 | 0.91 | 0.24 |
| Item 11. I am peaceful when talking about the end of my life. | 0.14 | 0.59 | 0.57 |
EOLC= End of life care
Validity
The 8-item ACPRS was positively correlated with the PEACE Acceptance subscale (r = 0.35, p = 0.0005) and negatively correlated with the PEACE Struggle subscale (r = −0.28, p = 0.0055), providing evidence for convergent validity. It was not significantly correlated with the DAS. It was adequately divergent from the MCSDS (r = −0.05; 90% CI: −0.23–0.12). Validation of the scale held when examining only those with active disease, but failed to find convergent validity when examining only those in surveillance (Supplementary Table 4).
ACPRS Associations with Demographic, Medical, and EOLC Outcomes
Table 4 provides the association between the ACPRS score with demographic, medical, and EOLC outcomes. Women with a living will or medical power of attorney had higher ACPRS total, willingness, and acceptance scores than those who did not have advance directives (p = 0.0030, p = 0.0328, and p = 0.0031, respectively). Additionally, women with DNRs had higher ACPRS acceptance and total scores than women without DNRs (p = 0.0322 and p = 0.0176, respectively). Cancer site or stage did not impact patients’ ACPRS score.
Step III Results
Patients from this sample were predominately white (75%) and had ovarian cancer (54%) (Table 1). Participation rate was 84% (110/131).
Confirmatory Factor Analysis (CFA)
An initial CFA indicated a modestly fitting model (RMSEA = 0.18, Pr[RMSEA ≤ 0.05] < 0.001; CFI = 0.89; TLI = 0.84; SRMR = 0.07).(15–17). However, modification indices indicated that the error term for Item 3 (Item 5 on the original scale) was correlated with the error term for Item 4 (MI = 36.12) and subsequently Item 5 (MI = 14.87) (Items 6 and 7 on the original scale). Because correlation between e3 with both e4 and e5 seemed plausible, a final CFA was constructed accounting for these correlations between error terms. The fit of this final model was much improved (RMSEA = 0.096; Pr[RMSEA ≤ 0.05] = 0.06; CFI = 0.97; TLI = 0.96; SRMR = 0.07), giving evidence that the ACPRS does measure two factors: 1) Willingness to Discuss EOLC and 2) Acceptance of EOLC (Figure 1).
Conclusions
The results of our study indicate that the ACPRS is a valid and reliable scale composed of two factors: 1) Willingness to Discuss EOLC and 2) Acceptance of EOLC. The scale was associated with patients’ advance care planning documentation, with those with a higher total ACPRS score being more likely to have advance care documentation such as a Medical Power of Attorney, a Living Will, or a DNR order. This suggests that the ACPRS may be useful in assessing a patient’s interest in engaging in advance care planning conversations with her medical provider and may indicate to providers when patients are ready to initiate advance care planning documentation.
Convergent validity of the ACPRS was assessed by examining the association between the ACPRS and the PEACE scale and the DAS. We found a positive correlation with the PEACE Acceptance subscale but a negative correlation with the PEACE Struggle subscale. The negative correlation observed between the ACPRS and the PEACE Struggle subscale inherently makes sense, but the directionality is not as originally hypothesized. We had originally thought the two scales would be positively correlated based on findings described in the methods section. The negative correlation noted between the ACPRS and the PEACE Struggle subscale may be due to attributes of the PEACE scale that were not taken into consideration prior to data collection for our study. For example, the original PEACE study consisted only of patients with terminal disease, but 47% of our subjects were free of their cancer and in surveillance (11). Additionally, of those being treated for cancer, terminal disease was not an eligibility requirement for our study. It is possible that because patients in the PEACE study were terminally ill, they were more likely to have addressed EOLC issues when compared to our sample. Additionally, 50% of the patients in the PEACE study were men, but our study only consisted of women. Men may struggle with their illness differently than women and may be more likely to implement advance care planning documentation, thus leading to a positive correlation between Struggle with Illness and advance care planning documents.
The ACPRS had a trend toward a negative correlation with the DAS, but the correlation was not significant as originally hypothesized. This could be because the analysis was not properly powered to assess whether or not the relationship was statistically significant since many patients choose not to complete the DAS. Specifically, many patients chose not to complete this scale because they felt it was too “morbid” since it asked pointed questions regarding how the patient felt about death.
Divergent validity of the ACPRS was assessed by examining the relationship between the ACPRS and the MCSDS. The ACPRS was not significantly correlated with the MCSDS, confirming the divergent validity of the scale. Reliability testing of the ACPRS indicate that the total ACPRS as well as its two subscales (Willingness to Discuss EOLC and Acceptance of EOLC) are internally consistent. Moreover, the test-retest reliability indicates that this measure is reliable over time.
The strengths of our study include that it adhered to stringent psychometric scale development methodology and was developed in conjunction with input from physicians, psychometricians, and patients. The patients in our study were at varying points in their cancer management, which allows for the scale to be generalized to a broader population that includes patients in all stages of cancer management.
Our study was limited by the fact that it was restricted to female participants with a gynecologic malignancy. Additionally, our patients were American and the majority were white. The fact that our study population consisted of American, white women limits the ability to generalize the ACPRS to other populations. Accordingly, the scale in its current state is not reliable to use among patients who are not women, do not speak English, and are not American (i.e. other cultures outside of the one that the ACPRS was validated in). This is a limitation present in all scale validation studies. Future developments of the scale should include male participants, other cultures, and other cancer types to assess whether the scale is valid to administer to other populations outside of American women with a gynecologic cancer. Additionally, future work should examine what potential “cut off” points using this scale would indicate that a patient is ready and willing to begin discussing advance care planning and end-of-life care. This could help guide clinicians regarding when to have these difficult conversations.
Early discussion regarding advance care planning goals for cancer patients is a quality care indicator that is likely to be more scrutinized in the future. Providers must develop ways to improve communication with patients regarding these topics. Tools that assess a patient’s level of comfort and interest in these discussions may encourage the conversations to occur in a timely and efficient manner. Although EOLC conversations are challenging, it is imperative that providers encourage patients to engage in these difficult conversations prior to progressing to an advanced state of illness that renders them unable to make decisions for themselves (18). The ACPRS is an 8-item scale that is valid and reliable. The scale provides a foundation for a tool that may be utilized to assess a gynecologic oncology patient’s readiness to discuss EOLC topics and their acceptance of the need of EOLC support. We plan to conduct further evaluation of the ACPRS scale in other gynecologic oncology patient populations in order to identify the ACPRS score that constitutes patient readiness to discuss EOLC, to evaluate how often the scale should be administered, and to evaluate the best time to administer the scale to patients.
Supplementary Material
Acknowledgments
This research was supported in part by the National Institutes of Health through MD Anderson Cancer Center’s Support Grant P30 CA016672 and T32 CA101642. Dr. Shen’s work on this project was supported by a cancer prevention fellowship from the National Cancer Institute (T32-CA009461). Dr. Parker’s work on this project was supported by a National Cancer Institute Cancer Center Core Grant No. P30-CA008748.
Footnotes
There are no conflicts of interest in the included manuscript.
References
- 1.Díaz-Montes TP, Johnson MK, Giuntoli RL, Brown AJ. Importance and timing of end-of-life care discussions among gynecologic oncology patients. Am J Hosp Palliat Care. 2013;30(1):59–67. doi: 10.1177/1049909112444156. [DOI] [PubMed] [Google Scholar]
- 2.Heyland DK, Dodek P, Rocker G, Groll D, Gafni A, Pichora D, et al. What matters most in end-of-life care: perceptions of seriously ill patients and their family members. CMAJ. 2006;174(5):627–33. doi: 10.1503/cmaj.050626. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sahm S, Will R, Hommel G. What are cancer patients’ preferences about treatment at the end of life, and who should start talking about it? A comparison with healthy people and medical staff. Support Care Cancer. 2005;13(4):206–14. doi: 10.1007/s00520-004-0725-z. [DOI] [PubMed] [Google Scholar]
- 4.Mack JW, Weeks JC, Wright AA, Block SD, Prigerson HG. End-of-life discussions, goal attainment, and distress at the end of life: predictors and outcomes of receipt of care consistent with preferences. J Clin Oncol. 2010;28(7):1203–8. doi: 10.1200/JCO.2009.25.4672. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Lopez-Acevedo M, Havrilesky LJ, Broadwater G, Kamal AH, Abernethy AP, Berchuck A, et al. Timing of end-of-life care discussion with performance on end-of-life quality indicators in ovarian cancer. Gynecol Oncol. 2013;130(1):156–61. doi: 10.1016/j.ygyno.2013.04.010. [DOI] [PubMed] [Google Scholar]
- 6.Mack JW, Cronin A, Keating NL, Taback N, Huskamp HA, Malin JL, et al. Associations between end-of-life discussion characteristics and care received near death: a prospective cohort study. J Clin Oncol. 2012;30(35):4387–95. doi: 10.1200/JCO.2012.43.6055. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Wright AA, Mack JW, Kritek PA, Balboni TA, Massaro AF, Matulonis UA, et al. Influence of patients’ preferences and treatment site on cancer patients’ end-of-life care. Cancer. 2010;116(19):4656–63. doi: 10.1002/cncr.25217. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.National Comprehensive Cancer Network. [Accession date: 1/9/17];Practice Guidelines in Oncology. [Available from: http://www.nccn.org/professionals/physician_gls/f_guidelines.asp - supportive.
- 9.Smith TJ, Temin S, Alesi ER, Abernethy AP, Balboni TA, Basch EM, et al. American Society of Clinical Oncology provisional clinical opinion: the integration of palliative care into standard oncology care. J Clin Oncol. 2012;30(8):880–7. doi: 10.1200/JCO.2011.38.5161. [DOI] [PubMed] [Google Scholar]
- 10.Rubin HJ, Rubin IS. Qualitative interviewing: The art of hearing data. 3. Washington DC: SAGE Publications; 2012. [Google Scholar]
- 11.Mack JW, Nilsson M, Balboni T, Friedlander RJ, Block SD, Trice E, et al. Peace, Equanimity, and Acceptance in the Cancer Experience (PEACE): validation of a scale to assess acceptance and struggle with terminal illness. Cancer. 2008;112(11):2509–17. doi: 10.1002/cncr.23476. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Templer DI. The construction and validation of a Death Anxiety Scale. J Gen Psychol. 1970;82(2d Half):165–77. doi: 10.1080/00221309.1970.9920634. [DOI] [PubMed] [Google Scholar]
- 13.Strahan R, Gerbasi KC. Short, homogeneous versions of the Marlowe-Crowne Social Desirability Scale. Journal of clinical psychology. 1972;28:191–3. [Google Scholar]
- 14.Crowne D, Marlowe D. Journal of Consulting Psychology. 24. 1960. A new scale of social desirability independent of psychopathology; pp. 349–54. [DOI] [PubMed] [Google Scholar]
- 15.Browne MW, Cudeck R, Bollen KA. Alternative ways of assessing model fit. In: Bollen KA, Long J, editors. Testing structural equarion models. Newbury Park, CA: SAGE; 1993. pp. 136–62. [Google Scholar]
- 16.Bentler PM. Comparative fit indexes in structural models. Psychological bulletin. 1990;107(2):238–46. doi: 10.1037/0033-2909.107.2.238. [DOI] [PubMed] [Google Scholar]
- 17.Hu Lt, Bentler PM. Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal. 1999;6(1):1–55. [Google Scholar]
- 18.Zaros MC, Curtis JR, Silveira MJ, Elmore JG. Opportunity lost: end-of-life discussions in cancer patients who die in the hospital. J Hosp Med. 2013;8(6):334–40. doi: 10.1002/jhm.1989. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.