Validation of the health care surrogate preferences scale.
Abstract: Recent advances in health care technology have increased the number of health care decisions made by acute care patients and those who act on their behalf, known as health care surrogates. This study reports on the validation of a new measure, the Health Care Surrogate Preferences Scale. Designed to assess the willingness of adults to perform and convey the duties required to communicate patient preferences, the scale offers a promising tool for use by social workers in health care settings. Development, evaluation, application of the new measure, and future research needs are discussed.

Key words: decision-making scale; end-of-life decisions; health care decision making; health care proxy; health care surrogate
Subject: Medical care (Methods)
Medical care (Technology application)
Medical care (United States)
Health services administration (Analysis)
Health services administration (Forecasts and trends)
Authors: Buckey, Julia W.
Abell, Neil
Pub Date: 07/01/2004
Publication: Name: Social Work Publisher: National Association of Social Workers Audience: Academic Format: Magazine/Journal Subject: Sociology and social work Copyright: COPYRIGHT 2004 National Association of Social Workers ISSN: 0037-8046
Issue: Date: July, 2004 Source Volume: 49 Source Issue: 3
Topic: Event Code: 010 Forecasts, trends, outlooks Canadian Subject Form: Medical care (Private); Medical care (Private); Medical care (Private) Computer Subject: Technology application; Market trend/market analysis
Product: Product Code: 9105210 Health Care Services; 8000001 Medical & Health Services NAICS Code: 92312 Administration of Public Health Programs; 62 Health Care and Social Assistance SIC Code: 8000 HEALTH SERVICES
Geographic: Geographic Scope: United States Geographic Code: 1USA United States
Accession Number: 119780300
Full Text: Advancing medical technology and increasing numbers of incapacitated patients in hospitals and nursing homes have augmented cultural demands to use technology to extend the lives of dying patients (Fletcher, Lombardo, Marshall, & Miller, 1997). In the United States, end-of-life treatment decisions have been examined through the lens of individual patient autonomy (American College of Obstetricians and Gynecologists, 1996; Hardwig, 1993; Keigher, 1994). In today's health care environment, a dual emphasis on cost containment and patient's rights fuels speculation on the types of medical care and treatment available to patients and their surrogates (Kapp, 2001).

Although patients prefer family members as surrogates (Hines et al., 2001), they are reluctant to discuss treatment preferences in advance, as evidenced by relatively low rates of advance directive completion (from 10 percent to 25 percent; Dubler, 1995). Although elderly people without family may choose professionals as surrogates (High, 1990), when physician-family member surrogacy was compared, physicians made decisions consistent with their formal medical roles, not as surrogates ethically bound to follow ex pressed patient wishes; family members typically felt greater burdens when making decisions, yet identified more with patient preferences (Silberfeld, Grundstein, Stephens, & Deber, 1996). To date, research interest has focused on decisions of medical patients rather than on those of health care surrogates responsible for making from 60 percent to 90 percent of intensive care treatment decisions (Hines et al.; Swigart, Lidz, Butterworth, & Arnold, 1996).

No single decision theory completely addresses the issues facing health care surrogates. Social problem-solving theory characterizes decision making as "a complex, cognitive-affective-behavioral process ... of different components ... and skills" (D'Zurilla & Nezu, 1990, p. 156). In developing the Social Problem-Solving Inventory (SPSI), D'Zurilla and Nezu assessed individual strengths and deficiencies across varying problem-solving contexts. The SPSI's decision-making subscale measures social decision making in daily living but does not capture the essential moral and ethical nuances of decisions made in health care environments.

Focused on individual uniqueness and a need for clear communication among clinicians, patients, and families, the clinical pragmatism model adapts clinical bioethics theory to the clinician-patient context, viewing end-of-life health care decisions as moral problems with moral considerations (Fins, Bacchetta, & Miller, 1997). Criticized for overemphasizing medical facts (Jansen, 1998; Tong, 1997), it prompted inclusion of conveyance As a current scale component.

In general, health care surrogates should exercise choices reflecting those of patients based on intimate knowledge of another's preferences. Statutes further imply that health care surrogates honor ethical and moral obligations to patients based on previously established relationships and personal, reciprocal bonds of trust developed over time (Fletcher et al., 1997). Yet, state statutes vary in their attempts to define and clarify responsibilities of patients and surrogates, so that resulting ambiguity or conflict can frequently place decisions back in the hands of physicians (Keigher, 1994). Nonetheless, state statutes governing health care surrogate duties and responsibilities (for example, Florida Health Care Surrogate Act, 2002) offer some direction in specifying tasks for scale inclusion and identifying the "interrelationships suitable for the development of a new instrument" (Nunnally & Bernstein, 1994, p. 107).

Based on these observations and concerns, this study reports the development of the Health Care Surrogate Preferences Scale (HSPS), a rapid assessment instrument designed to measure the willingness of surrogates to perform their required duties and to convey the wishes and needs of an incapacitated friend or family member to others. No standardized instrument is available to similarly evaluate the feelings, attitudes, or preferences of adults assuming these roles. Our hope is that this tool will be useful to practitioners and researchers working with those entrusted with ensuring that their loved ones' wishes and preferences prevail.

Method

Scale Construction

The HSPS is designed to address two health care surrogate dimensions: duties and conveyance (see sample items in the Appendix). It is based on the assumption that health care surrogates must be willing to make a decision and then act on it responsibly, with self-assurance and an unwavering commitment to the preferences of another.

Therefore, willingness, as related to the HSPS, was defined as a readiness to do what is necessary to act on the health care needs and wishes of another who is no longer capable of independent action. The two dimensions--duties and conveyance--emerge from this global construct definition. Duties represent a readiness to perform tasks or actions associated with the roles of a health care surrogate, including planning for and consenting to physical or mental health treatments, assessing and sharing information with medical providers, securing supportive care and services, and otherwise acting on a principal's prior instructions. Willingness to complete surrogate duties also requires conveyance, or sharing of a task through communication with others regarding the principal's wishes or needs. Conveyance becomes the method of accomplishing or finalizing health care surrogate duties.

Using Nunnally and Bernstein's (1994) domain sampling model, an initial 35-item pool was generated. An expert panel of practitioners and professors in medical, clinical, and geriatric social work specialties examined initial items for content validity. Experts were provided with clear definitions of the constructs, then asked to rate each of the 35 HSPS items on two aspects: (1) the degree to which the item reflected the construct definition and (2) the conceptual clarity of each item. Judges used a five-point scale, with 5 = excellent (recommending item retention) and 1 = poor (recommending rejection). Also, each member suggested rewording any item that appeared unclear.

The HSPS, a simple pencil-and-paper instrument, was designed to measure the readiness of potential health care surrogates to fulfill the multiple roles assigned to them when acting on behalf of incapacitated people. Participants were asked to register a level of agreement on a seven-point Likert scale, where responses ranged from 1 = never to 7 = always. High scores reflected greater willingness to perform and convey surrogate duties; lower scores reflected greater hesitancy. In the original item pool, four reverse score items (representing either low or absent levels of the respondent's willingness) were included to discourage acquiescence bias (DeVellis, 2003).

>From the original global HSPS totaling 35 items, the two intended subscales of duty (HSPS Duty) and conveyance (HSPS Conveyance) were proposed (see sample items in Appendix). The HSPS scores were obtained by computing mean scores for the Duty subscale (items 1 to 10), the Conveyance subscale (items 11 to 20), and the total HSPS (global) scale. As the psychometric properties of the HSPS depended on summative item contributions, individual item interpretation was not recommended.

Additional Instruments

We hypothesized that discriminant construct validity would be indicated if neither Duty nor Conveyance subscale scores were strongly related to factors hypothesized not to be associated (that is, age, gender, education, ethnicity, or religious preferences of the participants). In performing health care surrogate duties, commitment to the patient was hypothesized to be of primary importance and thus would be unrelated to the age, gender, education, ethnicity, or religious preference of those in the surrogate role. Convergent construct validity would be indicated through higher positive correlations of the HSPS Duty subscale with instruments reflecting the sound decision-making capability expected of one performing as a surrogate. We also hypothesized that willing surrogates should have high self-esteem, given the magnitude of the decisions and actions required of them, so high correlations between the HSPS Duty subscale and measures of self-esteem were expected. Finally, surrogate action to fulfill patient wishes conveyed the ethical and moral convictions inherent in a strong, close relationship with another and would be expected to result in high correlations between like measures and the HSPS Conveyance subscale.

To help test these construct validation hypotheses, other scales were included with the HSPS. Respondents were asked to complete the decision-making subscale (DMS) of the SPSI (D'Zurilla & Nezu, 1990). The DMS reports a Cronbach's alpha coefficient of .75 and measures an individual's capacity to make general decisions in everyday life. The second measure, the 40-item Self-Esteem Rating Scale (SERS) by Nugent and Thomas (1993), with an alpha coefficient of .97, measures respondents' regard for self. Finally, the Miller Social Intimacy Scale (MSIS) (Miller & Lefcourt, 1982), with alpha coefficients ranging from .86 to .91, measures the level of intimacy between adults, with alpha coefficients ranging from .86 to .91. We hypothesized that evidence of convergent construct validity would be established if respondents scoring high on the HSPS also scored high on the DMS, SERS, and/or MSIS.

Sample

Following approval by the institutional review board, nonprobability sample members--faculty and students at a major southern public university and professionals and paraprofessionals in three major southern cities, were requested to complete the survey. Course instructors, and hospital and hospice agency administrators, personally known to the first author, were approached to request instrument completion by students, professional, paraprofessional, and volunteer staffs. This purposive sample was selected from multiple communities in the state to reflect a more heterogeneous population, blending rural with urban subjects diverse in age, education, and religious preference. This type of sample provides a wide variety of respondents to test the broad applicability of questions composing a new instrument (Rubin & Babbie, 1997).

Data Collection

The packet administered to respondents contained the HSPS, the three convergent construct validity measures, and an explanatory cover letter with a declaratory statement of implied consent. Potential emotional reactions or anxiety experienced by the participants were addressed through provision of a services resource list. Participation was completely voluntary; no other activity was required; and participants could cease participation at any time without penalty.

A total of 350 instruments were delivered to the sites and administered by the first author or research associates; instruments were completed individually and in group settings. To standardize methods across settings, a verbal script was provided for research associates to read before administering the test in locations where the researcher could not be present.

Results

Respondents

A total of 195 responses were received. Seven instruments contained incomplete data and were excluded from the study. Of the remaining 188 respondents, 35 were male and 152 were female; one respondent did not provide gender. The average age was 45 years (SD = 16.4 years). The average number of years of formal education was 16.2 (SD = 2.4 years). Of the total sample, 155 (82.4 percent) were white American, 20 (10.6 percent) were African American, three (1.6 percent) were Native American, four (2.1 percent) were Hispanic American, three (1.6 percent) were Asian American, and two (1.1 percent) were of unspecified race or ethnicity; one person did not respond. Religious preference was as follows: 34 (18.1 percent) Catholic, 77 (41.0 percent) Protestant, 16 (8.5 percent) Jewish, 48 (25.5 percent) other, and 13 (6.9 percent) unspecified.

Respondents reported a relatively high decision-making capacity (DMS: M = 2.88, SD = .57, where 4 = extremely true); a positive self-concept (SERS: M = 1.67, SD = .74, where 7 = almost always or -7 = never [one-half these items are negatively scored]); and relatively strong relationships with intimate others (MSIS: M = 4.20, SD = .54, where 5 = almost always).

Content Validity

Judges' ratings of the fit between item content and HSPS construct definitions, as described earlier, were examined. Items that failed to achieve a mean rating of [less than or equal to] 1.5 from the ideal (5) would be considered for exclusion. No items fell below this threshold; therefore the original 35 items were retained for data collection and psychometric analysis.

Reliability

Cronbach's alpha coefficient was computed as an index of reliability for the individual subscales and factors. Springer and colleagues (2002) proposed standards for assessing alpha on tools used with individual clients, labeling coefficients < .70 as "unacceptable" and those > .85 as "respectable." In this study Cronbach's coefficient alpha was initially computed on the original subscale item pools, and items that made the weakest contributions to alpha were subsequently removed.

Within the original 35-item scale, reliability analysis of the initial 20-item HSPS Duty subscale produced an alpha of .81. Three items were identified for exclusion to increase alpha; two items, with alphas slightly lower than the initial subscale value, were excluded for pragmatic considerations, reducing the subscale to 15 items yielding an alpha of .89. These items were retained as the hypothesized Duty subscale for confirmatory factor analysis (CFA), detailed later in this article. Based on those reliability and CFA results and to maximize ease of scale completion, scoring, and interpretation, five additional Duty items were eliminated. This process resulted in an alpha coefficient of .86 for the final 10-item Duty subscale.

An alpha of .71 resulted from the initial reliability analysis of the 15-item HSPS Conveyance subscale. Following removal of five items to enhance reliability and factor structure and to maximize scale utility, the resulting 10-item Conveyance subscale demonstrated an alpha coefficient of .78.

The 20-item global HSPS alpha coefficient was .89. This coefficient and that for the HSPS Duty subscale exceed Springer and colleagues' (2002) recommended criterion for use with individuals. Although the HSPS Conveyance subscale falls marginally below their recommended minimum standard for individual use (alpha [greater than or equal to] .80), it exceeds the Nunnally and Bernstein (1994) level suggested for nomothetic (that is, group or research) applications (alpha [greater than or equal to] .70).

In addition to reported reliabilities, standard error of measurement (SEM) coefficients were computed as indicators of measurement error for the current sample (Nunnally & Bernstein, 1994). For the Duty subscale, SEM = .303; for the Conveyance subscale, SEM = .342; and for the global HSPS, SEM = .235. Adhering to Hudson's (1999) recommendation that SEMs

should be 5 percent or less of the total scale score range (for HSPS, 5 percent of 6.0 range = .3), the global HSPS SEM is well below the threshold value; both subscale SEMs slightly exceeded the target value.

Factorial Validity

Factorial validity, or factorial composition, is determined in part when scale items have higher positive correlations with the total scale score than with the score of any other scales used in the study (Nunnally & Bernstein, 1994). Corrected item total correlations for the HSPS Duty and HSPS Conveyance subscales were calculated to remove the item self-correlations in each, because without such handling the), are subject to producing exaggerated validity estimates (Nunnally & Bernstein). As reported in Table 1, correlations between the HSPS Duty, HSPS Conveyance, and global HSPS scores and scores for the DMS, SERS, and MSIS revealed stronger loadings for all items on the HSPS than on the other included validity instruments.

Because scale items were developed to match specific constructs, the multiple groups method of CFA (Nunnally & Bernstein, 1994) was used to further examine the factor structure of the HSPS. Following generation of a matrix correlating individual item responses with their intended and unintended subscales, column data were examined to determine whether any items not intended for a specific subscale correlated more strongly with that total subscale score than items that were intended. Second, row data were examined to determine whether individual item responses correlated more strongly with their intended subscale score than with the other HSPS subscale score. The HSPS Duty and HSPS Conveyance columns in Table 1 display the final version of this analysis. As a general criterion, item correlations with factors about .60 are considered moderately high (Nunnally & Bernstein).

Initial CFA included the 15-item Duty and 10-item Conveyance subscales resulting from initial reliability analyses. Inspection of initial column data revealed two items intended for HSPS Duty that loaded weaker on their subscale score than did several HSPS Conveyance items, and three additional items intended for HSPS Duty that loaded higher on HSPS Conveyance than did the lowest item intended for the Conveyance subscale. Based on these data, five initial Duty items were eliminated, creating a pragmatically desirable HSPS with 20 items.

The final HSPS consisted of two 10-item Duty and Conveyance subscales clearly revealing two hypothesized factors (see Table 1). Confirmation of factors previously identified in the HSPS development offered further evidence of the HSPS factorial validity (DeVellis, 2003). Examination of the columns revealed that all items of the Duty and Conveyance subscales correlated more strongly with their own subscale score than its counterpart. Examination of the rows showed that each item correlated more strongly with its own subscale than with its counterpart.

Construct Validity

Two types of construct validity, discriminant and convergent, were examined through the correlations found in Table 2. Discriminant validity demographic variables of age, gender, education, and ethnicity performed as expected, with only religious preference producing a significant negative correlation with HSPS Conveyance. Mean [r.sup.2] statistics for discriminant correlations were trivial. Mean [r.sup.2] statistics are also reported for each set of convergent measures in Table 2. Although all correlations were significant and positive, as hypothesized, the small effect sizes provide only weak preliminary evidence of convergent construct validity (Cohen, 1988).

Discussion

Results of this initial validation are promising for use of the HSPS in social work practice, particularly in health care settings and work with elderly populations. Based on this analysis, the HSPS has good reliability, content, and factorial validity, and preliminary evidence for construct validity. The short 20-item format is easy to administer, score, and interpret, making it suitable (within constraints described) for use as a rapid assessment instrument in practice and research. These characteristics are particularly valuable in emergency medical settings when choosing a health care surrogate or proxy is imminent. The HSPS adds to the literature by filling gaps in psychometric instrument development regarding bioethical and medical decision making.

Regarding discriminant construct validity, demographic variables performed as expected, with the minor exception of religious preference: Catholic respondents, followed by Protestants, appeared slightly more willing to convey the wishes of another when acting as surrogates. These results should be interpreted with caution, as the "other" religious category included a range of affiliations, from Pentecostal to agnostic.

Participants reported strong decision-making skills, as indicated by scores on the DMS, positively correlated with HSPS scores in convergent construct validity analyses. Also, positive correlations between SERS and HSPS scores suggest that individual self-esteem levels may factor into surrogate role performance, especially in conveyance. Finally, positive correlations between the MSIS and the Conveyance subscale suggest that the stronger the degree of intimacy shared between people, the stronger the surrogate's willingness to communicate patient desires to others. Respondents appeared more willing to convey another's wishes when they placed a higher value on themselves and on another with whom they shared a close bond of intimacy.

Still, the relatively weak supportive evidence for convergent construct validity constitutes a study limitation. Although a number of hypotheses were confirmed, effect sizes were small. The fact that there are no known instruments to evaluate the willingness of people to act as health care surrogates limited options for convergent construct validity measures, yet underscored the importance of the HSPS as an initial instrument to fill this measurement gap.

Future research should also obtain a larger sample. For factor analyses, our sample (N = 188) did not meet the 10 subjects per item recommendation (N = 350) of Nunnally and Bernstein (1994) or the suggested "fair" size (N = 200) of Tabachnick and Fidell (1996). Findings from a larger sample would strengthen information on HSPS psychometric properties.

However, potential difficulties exist in accessing desired research participants. People currently performing as health care surrogates may be less willing to participate in research because of inherent role stress. Researchers must acknowledge the privacy needs of these individuals and respect them accordingly. These factors add to the difficulty in obtaining improved clinical samples.

The relative lack of substantial theory addressing surrogate decision making presents a dilemma for researchers, clinicians, and clients. In the absence of a well-developed model, we have constructed beginning propositions that may shed light on these issues. Preliminary evidence for convergent construct validity illustrates that future attention to empirical and conceptual efforts is needed for service providers to close the gap between patient and family needs and our knowledge base.

Applications for Social Work Practice

In the ongoing concerns over the rights of incapacitated medical patients, the HSPS offers a preliminary measure for use by social workers in hospitals, nursing homes, and adult congregate living facilities. Social workers may find the HSPS useful as a planning tool when preparing advance directives, identifying potential surrogates, or targeting structured interventions. Responses to the HSPS may help clients identify with surrogate responsibilities, their own capacity for assertive involvement, or their fears in facing the death of a loved one.

The HSPS may also help expand the role of social workers in proactive decision making.

Use of a paper-and-pencil tool in the acute care setting may increase objectivity and produce chartable scale scores, thereby strengthening interdisciplinary communication (Christ & Sormanti, 1999). The HSPS may be especially useful in assisting with surrogate selection during judicial or acute care emergencies, when no advance directive exists, or when the courts are committed to naming supportive proxies or guardians most likely to act on behalf of patients.

In critical care settings, social work clients, family members, and friends face end-of-life decision crises requiring the ability to make sound decisions for loved ones. The HSPS offers a viable, beginning tool to help social workers perform the essential and historical tasks of the profession: supporting surrogate understanding of their feelings, perceptions, and options and helping surrogates follow through on honoring commitments to those unable to act on their own.

Appendix

Instructions and Sample Items: Health Care Surrogate Preferences Scale (HSPS)

At some time, we may be asked to make medical decisions for others. The following statements are designed to measure your feelings about making those decisions for the person whom you consider to be your close friend (a friend, spouse, or your closest relative). For each statement, fill in the blank with a number from the key below that most clearly represents the way you feel. This is not a test, so there are no right of wrong answers.

1 = Never 2 = Almost Never 3 = Rarely 4 = Sometimes 5 = Frequently 6 = Almost Always 7 = Always

Sample Duty Items

-- I would try to make the best decisions I could for my friend's medical care. -- If necessary, I could put making health care decisions for my friend ahead of my needs. -- If I disagreed with a doctor about my friend's medical care, I'd seek a second opinion.

Sample Conveyance Items

-- I believe I could persuade others to do what my friend would want them to do. -- If I believe I am right, I will stand up to a doctor in order to do what my friend would have wanted. -- I would "go to bat" for my friends anytime, but especially when they are sick.

References

American College of Obstetricians and Gynecologists, Committee on Ethics. (1996). End-of-life decision making: Understanding the goals of care. Women's Health Issues, 6, 287-294.

Christ, G. H., & Sormanti, M. (1999). Advancing social work practice in end-of-life care. Social Work in Health Care, 30, 81-99.

Cohen, J. (1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.

DeVellis, R. F. (2003). Scale development: Theory and applications (2nd ed.). Thousand Oaks, CA: Sage Publications.

Dubler, N. N. (1995). The doctor-proxy relationship: The neglected condition. Kennedy Institute of Ethics Journal, 5, 289-306.

D'Zurilla, T. J., & Nezu, A. M. (1990). Development and preliminary evaluation of the social problem-solving inventory. Journal of Consulting and Clinical Psychology, 2, 156-163.

Fins, J. J., Bacchetta, M. D., & Miner, F. G. (1997). Clinical pragmatism: A method of moral problem solving. Kennedy Institute of Ethics Journal, 7, 129-145.

Fletcher, J. E., Lombardo, P. A., Marshall, M. F., & Miller, F. G. (1997). Introduction to clinical ethics (2nd ed.). Frederick, MD: University Publishing Group.

Florida Health Care Surrogate Act, Title XLIV FL Statutes. [subsections] 765.205 (2002). Retrieved November 6, 2003, from http://www.flsenate.gov/Statutes/index.cfm?App_mode=Display_Statute&Search_ String=&URL=h0765/SEC201.HTM&Title=->2003->Ch0765->Section%20201

Hardwig, L (1993). The problem of proxies with interests of their own: Toward a better theory of proxy decisions. Journal of Clinical Ethics, 4, 20-27.

High, D. M. (1990). Old and alone: Surrogate health care decision making for the elderly without families. Journal of Aging Studies, 4, 277-288.

Hines, S. C., Glover, J. J., Babrow, A. S., Holley, J. L., Badzek, L. A., & Moss, A. H. (2001). Improving advance care planning by accommodating family preferences. Journal of Palliative Medicine, 4, 481-489.

Hudson, W. W. (1999). Measuring personal and social problems: Methods for scale development. Unpublished manuscript, Florida State University, Tallahassee.

Jansen, L. A. (1998). Assessing clinical pragmatism. Kennedy Institute of Ethics Journal, 8, 23-36.

Kapp, M. B. (2001). Economic influences on end-of-life care: Empirical evidence and ethical speculation. Death Studies, 25, 251-263.

Keigher, S. M. (1994). Patient rights and dying: Policy restraint and the states [National Health Line]. Health d, Social Work, 19, 298-303.

Miller, R. S., & Lefcourt, H. M. (1982). The assessment of social intimacy. Journal of Personality Assessment, 46, 514-518.

Nugent, W. R., & Thomas, J. W. (1993). Validation of the Self-Esteem Rating Scale. Research on Social Work Practice, 3, 191-207.

Nunnally, J. C., & Bernstein, I. H. (1994). Psychometric theory (3rd ed.). New York: McGraw-Hill.

Rubin, A., & Babbie, E. (1997). Research methods for social work (3rd ed.). Pacific Grove, CA: Brooks/ Cole.

Silberfeld, M., Grundstein, A. R., Stephens, D., & Deber, R. (1996). Family and physicians' views of surrogate decision-making: The roles and how to choose. International Psychogeriatrics, 8, 589-596.

Springer, D., Abell, N., & Nugent, W. (2002). Creating and validating rapid assessment instruments for practice and research: Part two. Research on Social Work Practice, 12, 805-832.

Swigart, V., Lidz, C., Butterworth, V., & Arnold, R. (1996). Letting go: Family willingness to forgo life support. Heart & Lung, 25, 483-494.

Tabachnick, B. G., & Fidell, L. S. (1996). Using multivariate statistics (3rd ed.). Northridge, CA: HarperCollins.

Tong, R. (1997). The promises and perils of pragmatism: Commentary on Fins, Bacchetta, and Miller. Kennedy Institute of Ethics Journal, 7, 147-152.

Julia W. Buckey, ACSW, is a doctoral candidate, School of Social Work, Florida State University, UCC 2500, Tallahassee, FL 32306-2570; e-mail: jcb5405@gamet.acns.fsu.edu. Neil Abell, PhD, LCSW, is associate professor, School of Social Work, Florida State University, Tallahassee. Earlier versions of this article were presented at Social Work 2000, November, Baltimore; the 53rd Annual Meeting of the Gerontological Society of America, November 2000, Washington, DC; and the 23rd Annual Meeting of the Southern Gerontological Society, May 2002, Orlando, FL.

Original manuscript received September 13,2001 Final revision received November 2, 2002 Accepted December 16, 2002
Table 1

Confirmatory Factorial Validity of the HSPS: Correlations of Item
Responses with Subscale and Related Scale Scores

HSPS Items              Duty     Conveyance     HSPS

Duty subscale
  1                    .57 **      .46 **      .58 **
  2                    .56 **      .33 **      .50 **
  3                    .62 **      .45 **      .61 **
  4                    .74 **      .37 **      .63 **
  5                    .80 **      .34 **      .65 **
  6                    .66 **      .43 **      .61 **
  7                    .66 **      .41 **      .60 **
  8                    .73 **      .46 **      .67 **
  9                    .72 **      .46 **      .67 **
  10                   .68 **      .40 **      .61 **
Conveyance subscale
  11                   .38 **      .49 **      .48 **
  12                   .21 **      .60 **      .44 **
  13                   .35 **      .61 **      .52 **
  14                   .31 **      .70 **      .55 **
  15                   .50 **      .69 **      .66 **
  16                   .54 **      .67 **      .67 **
  17                   .24 **      .48 **      .39 **
  18                   .50 **      .57 **      .59 **
  19                   .38 **      .65 **      .56 **
  20                   .41 **      .52 **      .52 **

HSPS Items               DMS        SERS      MSIS

Duty subscale
  1                     .17 *      .08       .12
  2                     .12        .17 *     .02
  3                     .11        .13       .23 **
  4                     .18 *      .12       .17 *
  5                     .10 *      .06       .12
  6                     .11        .20 **    .12
  7                     .07        .11       .11
  8                     .01        .11       .07
  9                     .10        .19 **    .15 *
  10                    .13        .04       .02
Conveyance subscale
  11                   -.02        .10       .06
  12                    .07        .10       .18 *
  13                    .07        .30 **    .10
  14                    .16 *      .20 **    .26 **
  15                    .18 *      .30 **    .16 *
  16                    .20 **     .14       .18 *
  17                    .14        .42 **    .05
  18                    .08        .25 **    .20 **
  19                    .02       -.09       .28 **
  20                    .22 **     .11       .20 **

NOTE: HSPS = Health Care Surrogate Preferences Scale; DMS =
decision-making subscale of the SPSI (D'Zurilla, T. J., & Nezu, A. M.,
1990); SERS = Self-Esteem Rating Scale (Nugent, W. R., & Thomas, J. W.,
1993); MSIS = Miller Social Intimacy Scale (Miller, R. S., & Lefcourt,
H. M., 1982).

* p<.05. ** p<.01.

Table 2

Convergent and Discriminant Construct Validity of the HSPS

                      HSPS Duty          HSPS Conveyance
Variable               Subscale              Subscale

                    r      [r.sup.2]      r       [r.sup.2]

Discriminant

Age                -.14        .02      -.02         .00
Gender             -.02        .00       .02         .00
Education           .07        .00       .13         .02
Ethnicity           .04        .00      -.08         .01
Religious
  preference       -.08        .01      -.15 *       .02
M                              .01                   .01

Convergent

DMS                .16 *       .03       .18 *       .03
SERS               .17 *       .03       .28 **      .08
MSIS               .16 *       .03       .29 **      .08
M                              .03                   .06

Variable              Global HSPS

                    r       [r.sup.2]

Discriminant
Age               -.09         .01
Gender             .00         .00
Education          .11         .01
Ethnicity         -.02         .00
Religious
  preference      -.12         .01
M                              .01

Convergent

DMS                .19 **      .04
SERS               .24 **      .06
MSIS               .25 **      .06
M                              .05

NOTE: HSPS = Health Care Surrogate Preferences Scale;
DMS = decision-making subscale of the SPSI (D'Zurilla,
T. J., & Nezu, A. M., 1990); SERS = Self-Esteem Rating Scale
(Nugent, W. R., & Thomas, J. W., 1993); MSIS = Miller Social
Intimacy Scale (Miller, R. S., & Lefcourt, H. M., 1982).

* p< .05. ** p < .01.
Gale Copyright: Copyright 2004 Gale, Cengage Learning. All rights reserved.