Early family experiences and helping behaviors of physical therapists.

[Curtis KA, Davis CM, Trimble TK, Papoulids DK. Early family experiences and helping behavior of physical therapists. Phys Ther. 1995; 75:1089-1100.]

Key Words: Compassion, Family of origin, Physical therapist, Role perceptions.

Helping Behavior

The health professions, including physical therapy, are helping professions. Goals in providing help may include reducing or eliminating pain or disease, restoring function, or reducing disability by providing appropriate treatment. The type of help offered may vary with the professional role of the helper, the underlying philosophies influencing treatment approach, and the nature of patients' problems.

Physical therapists share educational backgrounds and philosophies that prepare them for helping roles centered on the restoration of function. Despite a common educational background and similar philosophies, individual therapists may choose to help patients in different ways and to varying extents. What accounts for these differences in helping behavior? Some authors[1,2] contend that the type of help that health professionals provide for patients depends on the attitudes and values that the health professionals developed during childhood and adolescence.

The role that a person takes in the family may influence the development of effective or ineffective helping behaviors. A child who has frequently felt responsible for fixing a parent's problems, for example, may develop similar behaviors as an adult, such as taking on responsibility for patient problems that are well beyond the helper's control. Thus, an individual who has had family experiences in which such personal boundaries are not clear may develop helping behaviors that encourage helpee dependence and overextend the helper's personal resources.[1] In addition to the influence of helping roles in the family, it is also important to recognize additional factors that motivate helping behavior.

Brickman's Models of Helping and Coping

In trying to explain helping behavior, attribution theorists have explored the relationship of helper perceptions to their willingness to give help.[3,4] Blickman et al[5] identified the influence of helper perceptions of helpee responsibility for the problem and solution on helping behavior and conceptualized four models to represent these motivational factor. These models vary levels of responsibility (blame) for a problem and responsibility for the solution to the problem (Fig. 1). Depending on how the helper perceives the helpee's situation, the helper may be mote or less willing to give help. These models can be applied to enhance our understanding of helping behavior in typical medical and rehabilitation situations.

In Brickman's moral model, the patient has high responsibility for both the problem and the solution. In this situation, a patient may have caused a problem by an unhealthy lifestyle, high-risk activity, or neglect and is responsible for taking appropriate action to generate and follow through with a solution (treatment) to the problem.

In contrast, in the enlightenment model, the patient has high responsibility for the problem but low responsibility for the solution. This could be a patient who caused the problem as in the moral model, but who is unwilling or unable to generate solutions to the problem.

In the compensatory model, the patient has low responsibility for the problem but high responsibility for the solution. This model is exemplified by an innocent victim of an accident or a person with a severe illness who is still able to seek and fully participate in a solution (treatment) to the problem.

Finally, in the medical model, the patient has low responsibility for both the problem and the solution. This might be a victim of an accident or a person with a severe fullness who is unwilling or unable to participate in solutions to the problem. Multiple factors such as the nature of the solution, cognitive involvement, or immaturity might prevent the patient's participation in seeking or implementing a solution to the problem.

Factors Motivating Helping

Kahan studied helping responses in physician's assistants and nurses who responded to scenarios representing Brickman's four models (Kahan M, unpublished research, 1988). She found that these subjects were most willing to help when a hypothetical patient met the description of the medical model and were less willing to help when the patient met the description of the moral model. Similarly, subjects felt the most anger for hypothetical patients representing the moral model and the most pity for patients representing the medical model. These findings illustrate that helpers are most willing to help when the helpee is determined not to be responsible for either the problem or the solution to the problem. In contrast, they are least willing to help when the helpee is determined to be responsible for both causing the problem and implementing the solution to the problem.

Studies have shown that people are most willing to help when they are faced with problems they perceive to be beyond the helpee's control.[6] Such research on helping behavior has focused on examining perceptions of the cause of the helpee's problem. Our perceptions of the locus of the cause, the stability of the cause, and the controllability of the cause of the helpee's problem, as well as our emotional responses, determine our willingness to help.

Research in the attribution field has shown that if the helper perceives the locus of the cause of the problem to be internal, that is, the problem has been caused by an action or trait of the helpee, the helper will be less inclined to help. Conversely, helpers are more inclined to help if the cause is perceived to be externally caused.[6-8] Stability refers to the helper's perception of the potential changeability of the current problem. If the problem is seen as changeable or unstable, a person will be more motivated to offer help to effect change for the better than if the problem is perceived as unchangeable or stable.[6-8] Controllability refers to the helpee's control of factors that caused the problem. If the cause of the problem is felt to be within the helpee's control, helpers are less willing to help than if the cause of the problem is perceived to be out of the helpee's control.[4,6]

Numerous studies[4,9-11] have shown high correlations between emotional responses and helping behavior. if a person reacts with sympathy to another person's problem, then there will be a greater tendency to help. If a person reacts with anger to another person's problem, this reduces the tendency to help.[4,9] Thus, our perceptions of the factors causing a given situation and our feelings about the situation all influence our willingness to provide help.

Additionally, an empathic set or perspective may augment the helping response. Betancourt[11] reported that subjects who were instructed to consider the helpee's perspective felt that the helpee's problem was less controllable, felt greater empathic emotion, and were more likely to help than subjects who had been instructed to take a more objective perspective.

The Consequences of Providing Excessive Help

Excessive helping behavior in adult health care professionals can have negative outcomes for both the patient and the provider.[2] When the provider fosters patient dependence by offering excessive help, the patient may have difficulty in taking appropriate responsibility for his or her own health care needs.[12] Additionally, providers may create further dependence and create a sense of overload by failing to delegate appropriate responsibility to supportive personnel or family members.[12] these types of ineffective helping relationships, patients fail to function independently and look to health care providers to solve their problems. From a different perspective, providers who habitually encourage patient dependence and overextend themselves in patient relationships may eventually feel overwhelmed and frustrated by their inability solve patient problems.[12] These types of feelings are reported to contribute to health care provider burnout, job dissatisfaction, and attrition.[12-15]

Identifying the factors that motivate excessive or ineffective helping behavior in physical therapists may be valuable in designing interventions to foster patient independence and prevent the negative consequences of provider overextension. Additionally, in the current climate of health care reform, physical therapist time will be increasingly allocated toward patient evaluation, case management Consultation, and education, rather than toward extended hands-on treatment. Recognition of factors that influence a person's willingness to delegate to and educate others will become increasingly important as physical therapist roles shift to that of consultant and case manager on the health care team.

The purpose of this study was to determine the influences of the health care provider's early family experiences and perceptions of the patient's situation in motivating provider helping responses. We hypothesized that therapists who reported early family experiences that were characterized by unclear personal boundaries (dysfunctional family) would respond to patients with greater tendencies to help all patients and would see themselves as less replaceable in offering this help. Consistent with the findings of previous studies, we expected that subjects would offer the most help, feel less replaceable, and feel most compassionate about the patient who was perceived to have low responsibility for both the problem and the solution (medical model).

Method

Subjects

The subjects for this study were recruited from 500 randomly selected active members of the American Physical Therapy Association. From this sample, 226 subjects returned a survey packet, out of which 221 survey packets were usable, yielding a 44% response rate. The average age for this group was 34.3 years (SD=7.0), with a range from 22 to 66 years. The sample was 81.9% female and 99.5% Caucasian. Most (82.3%) of the subjects held an entry-level physical therapy degree at the baccalaureate level, 9.1% had received a postbaccalaureate certificate, and 8.6% held a master's degree.

The average time since graduation was 10.7 years (SD=6.6), with a range 1 to 41 years. The subjects averaged 9.6 years of full-time clinical experience since graduation, with a range of 1 to 30 years. They were employed in a variety of work settings (ie, 31% in acute care hospitals, 26% in private offices, 10% in home health care, 80% in rehabilitation centers, 5% in skilled nursing facilities, 4% in schools [pediatrics], 3% in academic institutions, and 12% in all other settings).

The subjects reported working an average of 41.2 (SD=10.2) hours per week, with a range of 15 to 70 hours per week. This total included an average of 31.9 (SD=12.3) hours per week in clinical practice, ranging from 0 to 60 hours per week. Other work activities accounted for an average of 10.3 (SD = 10.6) hours, with a range from 0 to 44 hours.

Instruments

We designed a 12-page questionnaire booklet consisting of three measurement instruments: a demographics data sheet, the Family of Origin Scale (FOS), and the Helping Questionnaire.

Demographics data sheet This questionnaire elicited information about subject age, gender, ethnic origin, educational background, time since graduation, and current employment site. Subjects also estimated the hours spent per week in current job responsibilities, including time spent specifically in patient care responsibilities.

Family of origin scale. The FOS is a validated self-report of perceived levels of psychological and emotional health in the family in which an individual has his or her beginnings.[16-18] This instrument consists of 40 items that measure the quality of intrafamily communication, such as responsibility, power structure, autonomy, intimacy, acceptance of separation and loss, perception of reality, family mood, tone, range of feelings, and respect for others. Responses to this instrument were scored on a five-point Likert scale, anchored at "strongly agree" and "strongly disagree." Subjects responded to items such as "In my family, I felt free to express my own opinions." Half of these items were worded negatively and were scored by reversing the Likert scale.

Scores may range from 40 to 200, with the highest scores indicating a high level of perceived family health. Previous researchers,[6-18] have reported scores ranging between 62 and 198 and a test-retest rehability of .97.

Helping questionnaire. The third instrument was a questionnaire developed especially for this study, which consisted. of four patient care stimulus scenarios, representing Brickman's four models varying patient responsibility for the problem and patient responsibility for the solution. These four scenarios represented typical patients referred for physical therapy in each of the medical, compensatory, enlightenment, and moral models (Appendix). For example, the patient for the medical model, who has low responsibility for both the problem and the solution, was portrayed as a deconditioned patient receiving renal dialysis and awaiting a kidney transplant. In contrast, die patient for the moral model, who has high responsibility for both the problem and the solution, was portrayed as a reckless motorcyclist who caused a multiplecar collision and must undergo treatment for multiple lower-extremity fractures. The four scenarios served as the experimental manipulation for this study and allowed for within-subjects comparison of helping responses in a 2X2 matrix identical to the Brickman models (Fig. 1).

Subjects responded to a series of questions following each patient scenario. We included six questions to explore three constructs serving as dependent variables relating to helping behavior (1) willingness to help, (2) perceived replaceability, and (3) feelings of compassion. Each of the four scenarios was followed by the same six questions, which enabled us to compare subject responses for each cell of the model.

Definitions, Item Content, and Scoring

The first dependent variable, willingness to help, included three items that measured how much the therapist was willing to exert extra effort within the normal boundaries of the physical therapist role to meet the patient's needs, such as making extra telephone calls or researching information to help a patient. Subjects responded on a seven-point scale, anchored at 1 for "not at all willing" and 7 for "very willing." Thus, the possible scores ranged from 3 to 21 on this variable. A high score on this variable would indicate a high degree of willingness to help.

The second variable, perceived replaceability, was measured by two items that centered on delegating responsibility to or replacing the therapist with a colleague or family member. one of these items, for example, read "How easy would it be for another physical therapist to replace you in your role with this patient?" Subjects responded on a seven-point scale, anchored at 1 for "not at all easy" and 7 for "very easy." Thus, the minimum possible score was 2 and the maximum score was 14 on this variable. A high score on this variable would represent a greater tendency to delegate responsibility and to discourage patient dependence on the provider.

The last dependent variable, compassion, was measured by having the subjects rate the degree to which they felt compassion for the patient in each scenario. Subjects responded to this single item on a seven-point scale, anchored at 1 for "not at all" and 7 for "very strongly."

Instrument Development

Scenarios were first developed and evaluated for content and accurate representation of Brickman's models by expert review for content validity. An initial 16 scenarios were then pilottested among 28 physical therapy student subjects for subject perceptions of stability, severity, patient responsibility for the problem, and patient responsibility for the solution for the patient in each scenario. The most representative scenario of each of the four models was selected for inclusion in the final instrument. The Helping Questionnaire items were also reviewed by experts, revised, and pilottested among 21 physical thempy student subjects. An intraclass correlation (ICC[1,1]) was performed on the final questionnaire items to estimate test-retest reliability for the three dependent variables. The average values (r) for test-retest reliability across all four scenarios were .862 for the construct willingness to help, .734 for perceived replaceability, and .817 for compassion.

Procedure

The three instruments were presented in a 12-page questionnaire booklet, which was sent to a random sample of 500 physical therapists with a cover letter explaining the project and a postage-paid return envelope. Two weeks later, a postcard was sent to all 500 subjects, encouraging their participation. Four weeks after the initial mailing, a follow-up letter and another copy of the questionnaire booklet were sent to nonrespondents.

Data Analysis

The Likert scale scores for the 20 negative items in the FOS were reversed, and then the scores were summed. We then categorized the subjects into three subject pools divided by lower, middle, and upper thirds of the FOS score frequency distribution. The lowest third was classified as dysfunctional, the middle third as neutral, and the upper third as functional, in accordance with the methods reported for past studies using the FOS.[16]

The three dependent variables (willingness to help, perceived replaceability, and compassion) were analyzed using a series of 2X2x3-factor (patient responsibility for the problem, patient responsibility for the solution, and FOS grouping) analyses of variance (ANOVAs). These ANOVAs measured the intrasubject differences in helping responses for the four scenarios representing each of the Brickman models and the intersubject differences for the three FOS subgroups.

Results

Family of Origin Scale

In our sample, the FOS results yielded a mean score of 147 (SD=31.1, range=53-200). The group was divided into thirds based on the subjects' FOS scores. Seventy-one subjects with scores between 53 and 138 were classified as dysfunctional, 73 subjects with scores between 140 and 160 were classified as neutral, and 74 subjects with scores between 161 and 200 were classified as functional. The partitioning of the sample within these ranges of scores is similar to previously reported scores.[16]

Helping Questionnaire

As expected, Helping Questionnaire responses varied with patient responsibility for each of the four scenarios. Subject responses to the repeated six questionnaire items were analyzed by factor analysis for each of the four scenarios to assess the validity of the questionnaire items in representing the three separate constructs willingness to help, perceived replaceability, and compassion. Factor loading following orthogonal transformation showed a three-factor solution, representing these three constructs, consistently across all four patient scenarios (Tab. 1).

[TABULAR DATA 1 OMITTED]

Each of the dependent variables was then analyzed separately, and responses were compared among scenarios. For the dependent variable willingness to help, responses ranged from a low score of 5 to a high score of 21. The lowest mean score ([chi bar] = 15.9, SD=3.7) was found for the patient representing the enlightenment model, and the highest mean score ([chi bar] = 18.2, SD=2.3) was found for the patient representing the compensatory model. Scores for the variable perceived replaceability ranged from 3 to 14, with the highest mean score ([chi bar] = 11.1, SD=2.1) for the patient representing the moral model and the lowest mean score ([chi bar] = 10.4, SD=2.1) for the patient representing the compensatory model (Tab. 2).

Overall, subjects expressed high levels of compassion for the patients described in the Helping Questionnaire scenarios. Scores ranged from a minimum of 1 to a maximum of 7 for the dependent variable compassion. The highest mean score for this variable was found for the hypothetical patient in the compensatory model ([chi bar] = 6.1, SD=0.8), and the lowest mean score was found for the patient in the moral model ([chi bar] = 4.4, SD = 1.4) (Tab. 2). There were positive correlations between the variables willingness to help and compassion in the medical model (r=.119, P=.08), in the compensatory model (r=.223, P<.001), in the enlightenment model (r=.489, P<.0001), and in the moral model (r=.264, P<.0001).

[TABULAR DATA 2 OMITTED]

Analysis-of-Variance Results

A series of 2x2x3 ANOVAs (patient responsibility for the problem, patient responsibility for the solution, and FOS grouping) were performed to analyze the influence of early family experiences and variations in patient responsibility on subject responses to the three dependent variables (willingness to help, perceived replaceability, and compassion). Preliminary analyses of FOS scores and Helping Questionnaire responses by gender and by years of experience yielded no significant differences; therefore, all data were pooled for all subsequent analyses.

Family of origin. There were significant main effects of the independent variable FOS group on the dependent variables willingness to help and compassion (Figs. 2, 3). The functional group demonstrated significantly higher levels of willingness to help and scored significantly higher on feelings of compassion compared with the dysfunctional and neutral groups (Figs. 2, 3). Post hoc testing by a series of Scheffe's F tests showed that there were significant differences in each of the dependent variables (willingness to help, perceived replace ability, and compassion) between the average scores of the functional and dysfunctional groups and between the average scores of the functional and neutral groups, but not between the scores of the neutral and dysfunctional groups (P<.05).

There were no significant interactions of the independent variable FOS group with either of the other two independent variables, responsibility for the problem and responsibility for the solution. Thus, the differences in FOS group responses were maintained consistently across all scenarios (Tab. 3).

[TABULAR DATA 3 OMITTED]

Responsibility for the problem and solution. The 2X2X3-factor ANOVA showed significant main effects for the independent variable responsibility for the problem on afl three dependent variables (willingness to help, perceived replaceability, and compassion) and significant main effects for the independent variable responsibility for the solution on the dependent variable willingness to help (Figs. 2-4). Compared with their responses for patients with high responsibility for the problem (enlightenment and moral models), subjects were more willing to help, felt less able to be replaced, and felt more compassionate for patients with low responsibility for causing the problem (medical and compensatory models) (Tabs. 2, 3).

The ANOVA results showed significant interaction effects of the independent variables of responsibility for the problem and responsibility for the solution on all three of the dependent variables (willingness to help, perceived replaceability, and compassion). Subjects were most willing to help, felt least replaceable, and were most compassionate toward the patient who represented Brickman's compensatory model, followed by the medical model, followed by the enlightenment model, and were least willing to help the patient who represented the moral model. Using a series of Scheffe's F tests, we found post hoc differences (P<.05) in subject responses for the dependent variables willingness to help and compassion between each of the four patient scenarios, with one exception. Only the post hoc differences in subject responses between the enlightenment and moral models failed to reach statistical significance. However, for the dependent variable, perceived replaceability, there were only post hoc differences between the medical and moral models, between the compensatory and enlightenment models, and between the compensatory and moral models (Figs. 2-4).

Discussion

Family of Origin and Helping Behavior

We hypothesized that therapists from dysfunctional families of origin would be uncomfortable with delegation and offer excessive help, tending to create patient dependency and the potential for overextension of provider resources. Contrary to our hypothesis, when compared with the neutral and dysfunctional FOS groups, the functional FOS group had a higher score for the dependent variables willingness to help and compassion. These results were inconsistent with our hypotheses; we had expected that the dysfunctional FOS group would show the highest scores on these two variables.

Further, our results showed little influence of the FOS score for the dependent variable perceived replaceability, indicating that subjects in all FOS groups were equally as likely to delegate treatment responsibility in any patient case. To understand these unexpected findings, we generated several potential explanations.

Affect-Mediated Helping Behavior

Attributional studies on helping behavior support an affect-mediated model of helping (Kohan M, unpublished research, 1988).[3,4,6-8] This model hypothesizes that perceptions of helpee controllability lead to affective reactions such as pity, which then motivate helping behavior. Subject helping responses in our study also seemed to be mediated by the affective reaction of compassion. Feelings of compassion were positively related to a greater willingness to help. These findings are consistent with the attribution literature and seem to apply in all FOS groups. The relationship of affect to helping response may partially explain some of the differences seen among the FOS groups, as the functional FOS group tended to react with stronger feelings of compassion than did the dysfunctional and neutral FOS groups.

Why did the subjects in the functional FOS group react with stronger feelings of compassion than did the subjects in the neutral and dysfunctional FOS groups? It may be helpful to think of the subjects in the functional FOS group as holding a common view or value different than the dysfunctional and neutral FOS groups.

Emphatic Set

Betancourt[11] reported that our perceptions of controllability, affective reactions (feelings), and helping tendencies are influenced by prior instructions to think about the needs or feelings of a needy person. This empathic set sensitizes the helper to perceive that the situation is less controllable by the helpee and to feel more sympathy, which results in a greater helping response. Perhaps the subjects in the functional FOS group hold a cognitive perspective or value in common, which causes strong feelings of compassion, thus motivating their willingness to help. This shared value or perspective might be related to their helping roles in their families, or it may reflect an interaction of their family experiences with the strong influences of professional socialization.

Role Socialization 

We can look to role socialization as another possible influence on helping behavior. Role socialization in the workplace can be a potent force in shaping professional behavior, both in the precareer entry period (professional training) and during new graduate practice.[19] We can expect that the subjects in our study have been influenced by professional role socialization, as they were experienced therapists with an average of 10.7 years since graduation. The effect of early family experience on helping behavior may be mitigated by professional role models in shaping helping attitudes and behaviors. Although evidence indicates that excessive willingness to help and perceptions of irreplaceability will most likely be detrimental to the practicing health care professional, these behaviors may be a strongly socialized and highly valued characteristic of the physical therapy profession. Despite family background, after many months of exposure to these professional values, these values and not the family experiences may become the driving force in explaining helping behavior. Perhaps individuals in the functional FOS group were more vulnerable to the influence of professional role models who encouraged or modeled excessive helping behavior. Additionally, these individuals who comprised the functional FOS group may have other personality characteristics or propensities in common that serve as intervening variables in motivating excessive helping behavior.

Limitations of the Family of Origin Scale

We must consider that the FOS score may not accurately reflect the actual dynamics of early family experiences that specifically influence the helping relationship and may represent what the subject wants to believe. It might be reasonable to expect such protective responses with such sensitive issues such as familial interactions. The FOS would allow these protective responses to occur, because this self-report instrument can only measure subjects' perceptions of their early family experiences.[17]

Helping Behavior and the Brickman Models

Regardless of their family experiences, therapists showed the highest willingness to help and the least willingness to be replaced following the Brickman model scenarios in which the patients had low responsibility for the problem (medical and compensatory models). This finding is consistent with those of previous studies in the helping domain.[9]

We found it interesting that the patient who represented the compensatory model, not the medical model, stimulated the strongest helping tendencies. Because the patient in the compensatory model represents the two conditions that seemed to optimally motivate helping (low responsibility for the problem, high responsibility, for the solution) and that evoked high levels of the affect compassion, it would follow that this patient elicited the highest scores on the dependent variable willingness to help and the lowest scores on perceptions of replaceability.

The compensatory model, represented by a scenario describing a patient who is recovering from the removal of a benign spinal cord tumor and has extensive lower-extremity weakness, is a familiar and comfortable one for most physical therapists. This patient scenario follows the typical model of rehabilitation, in which the patient enters the rehabilitation system from an acute care episode (surgery); has limited responsibility for the problem; and is then expected to participate actively in the rehabilitation, education, and coping process. Although k was encouraging that physical therapists were most willing to help this patient, who probably, evoked familiar feelings, it is also distressing that these same therapists may be most likely to overextend themselves and fail to delegate treatment responsibilities when indicated. These types of actions may be related to the high levels of burnout recently reported in a study of rehabilitation hospital therapists.[14]

Implications for Clinical Practice, Education, and Policy

The results of our study have implications for clinical practice, physical therapy education programs, and health care policy. In clinical practice, it is important that health care professionals recognize their biases toward different patients. we believe accepting feelings as valid and then actively working to restrict the influence of these feelings on the delivery of services is required for professional care. Equally important is the acceptance of conditions that cannot be changed. Overextension of provider resources may develop if therapists believe that all problems can be helped only by the provider or if therapists offer help to the same extent to an patients. Discernment is critical; therapists must be able to distinguish accurately the extent to which their help can make a difference.

Because therapists are more likely to offer the most help to patients with low responsibility for their medical problem, the patient mix that therapists treat may influence their tendencies to chronically overextend their personal resources. Realistic institutional expectations, supervisory support for appropriate delegation and patient/family education, a patient load that includes a variety of diagnoses, and a mix of a patient care and nonpatient care responsibilities may provide the therapists with opportunities to achieve a sense of balance, in the workplace. Such influences in the work environment may counter therapists' tendencies to offer excessive help or overextend their personal resources and be useful in preventing feelings of burnout and job dissatisfaction.

In physical therapy curricula, academic and clinical educators need to emphasize the role of the, physical therapist as that of an effective agent for change. Ideally, the motivating foundation from which the future professional provides service or effective help should be the projected potential for change of the patient's functional problems, rather than the feelings evoked or attributions about patient responsibility for the medical problem.

Third-party payers often require that therapists predict potential for change in patients. As part of standard Medicare documentation requirements, therapists must address the rehabilitation potential of the patient to justify the provision of services. With this value in mind, if a patient's functional problem is likely to change via physical therapy intervention, the therapist should recommend delivery of the appropriate service. If the patient's functional problem is not likely to change, however, the therapist should make the decision not to pursue unrealistic goals. Patient responsibility for the original problem should not influence the type, quality, or quantity of treatment to that patient.

Finally, these findings have implications for health care policy. The results of this study showed that physical therapists are less willing to help patients who they feel are responsible for their problems. Both providers and recipients of care must be careful that health care policies, such as reimbursement for services and allocation of research funds, are, not driven by perceptions of patients' responsibility for their problems. For example, it does not serve the needs of society to have funding denied because policy-makers hold die recipients of care responsible for acquiring or contributing to their disease or problems by controllable lifestyle choices. These biases in our thinking may lead to unconscious decisions with major implications for large segments of the population.

Limitations of the Study

In survey research, sampling error leads us to question whether the respondent sample was truly representative of the population of physical therapists. Although three mailings were done, a response rate of less than 50% is sufficiently low to exercise caution in generalizing these findings to the physical therapist population at large.

Another limitation of this study could be our method of partitioning the subjects into three groups of equal size by the frequency distribution of their scores on the FOS. Although this method was similar to those of other reported studies, it may have led to the nonsignificant differences observed between the FOS groups. The distribution of scores in the dysfunctional group spanned an 85-point range, with over half of the scores falling with a 25-point range of the lowest neutral group scores. The range of scores in the neutral and functional groups covered only 21 and 39 points, respectively. Thus, it is likely that many members of the dysfunctional and functional groups had experiences similar to those of the members of the neutral group, possibly accounting for the lack of significant differences among the FOS groups. Therefore, early family experiences may have greater influence on a therapist's helping behavior than the results of this study suggest.

Finally, this study was based on subject responses to hypothetical patient descriptions. Our operationalization of the constructs willingness to help, perceived replaceability, and compassion in the Helping Questionnaire may have oversimplified these phenomena. Measuring these variables as projected responses for hypothetical patients may n account for the interaction of expressed helping tendencies with real-life variables such as coping strategies, support systems, self-restraint, and various personality characteristics, which may markedly influence actual behavior.

Future Research

Future exploration in this area is warranted because many questions remain unanswered and the results of this study generate new questions. The relative contributions of early family experiences and the professional socialization process could be compared. Certainly, both factors influence personal feelings as well as helping behaviors, and it would be pertinent to discern how both family experiences and professional socialization influence professional behavior. The influence on helping behavior of age and cultural and gender differences of patients and therapists also remains unexplored. This information would be valuable for designing interventions to influence helping behavior.

Conclusion

The purpose of this study was to explore how early family experiences and variations in perceived patient responsibility for the patient's problem and treatment motivate helping behaviors in physical therapists. Results showed that dysfunctional early family experiences, as measured by the FOS, do not seem to promote excessive willingness to help or feelings of irreplaceability, as was previously hypothesized. Instead, the results of this study showed that those who reported functional early family experiences actually indicated a greater propensity to provide help, especially for patients with low responsibility for their medical problems.

The physical therapists who participated in this study were most inclined to help and least likely to delegate treatment responsibility for those patients who have low responsibility for their medical problems. Such feelings and helping tendencies may cause a disparate allocation of treatment and staff resources. This phenomenon may contribute to chronic overextension of provider resources. The findings of this study help to isolate the determinants of helping behavior and may provide insight into how our educational and institutional systems can address these issues for the optimal benefit of both staff and patients.

Acknowledgments

We wish to express our appreciation to Sandra Graham, PhD, University of California, Los Angeles, and Miriam Kahan, PhD, Charles Drew School of Medicine, for their assistance with study design and development of instruments. We also thank E Brooks Applegate, PhD, for his support with interpretation of the results of data analysis and Eleftheria Sidiropoulou, PhD, for her assistance with manuscript review.

[Figures 1-4 ILLUSTRATION OMITTED]

Appendix. Brickman Model Patient Scenarios

Medical Model

This patient is a 43-year-old man with renal failure due to a side effect of an antibiotic he was taking for septicemia resulting from an abscess on his leg. The leg infection has cleared and is healing nicely. He currently has weakness, lethargy, and confusion, which improves with dialysis. He is not showing signs of returning renal function and is expected to need a kidney transplant in the future.

Compensatory Model

This patient is a 40-year-old man a spinal cord tumor at the L-2 level. The tumor has been resected completely, leaving him wth lower-extremity weakness. Most patients at his age and with this degree of involvement do not achieve full ambulatory independence with assistive devices or orthoses.

Enlightenment Model

This is a 43-year-old man who is experiencing memory loss due to chronic alcohol abuse. He sustained a fractured femur after falling downstairs. His blood alcohol level was .34 when he was admitted to the hospital. He underwent internal fixation of the femoral fracture with an intramedullary rod. He is currently having delirium tremens, is verbally abusive to the staff, and is being medicated with Valium[R] to control the "dt's."

Moral Model

This patient is a 38-year-old man who was in an accident while driving his motorcycle without a helmet and wearing shorts. He was speeding on a city street, and he reportedly ran a red light and crashed while swerving to miss an oncoming car. The occupants of the other car were killed trying to avoid him. He sustained multiple fractures of the fight lower extremity and peripheral nerve damage, which has left his tibialis anterior muscle paralyzed. Due to the nature of his injuries, he is not expected to regain this muscle function and will have permanent shortening of the right limb by 2.54 to 5.08 cm (1-2 in).

References

[1] Cauthorne-Lindstrom C, Hrabe D. Codependent behaviors in managers: a script for failure. Nursing Management. 1990; 21:34-39. [2] Herrick CA. Codependency: characteristics, risks, progression and strategies for healing. Nursing Forum. 1992; 27(3):12-19. [3] Weiner B. An Attributional Theory of Motivation and Emotion. New York, NY: Springer-Verlag New York Inc; 1986. [4] Weiner B. A cognitive attribution-emotional-action model of helping behavior: an analysis of judgments of help giving. J. Pers Soc Psychol. 1980; 39:186-200. [5] Brickman P, Rabinowitz VC, Karuza J, et al. Models of helping and coping. A Psychol. 1982;37:368-384. [6] Meyer JP, Mulherin A. From attribution to helping: an analysis of mediating effect of affect and expectancy. J Pers Soc Psychol. 1980; 39:201-210. [7] Sharrock R, Day A, Qazi F, Brewin CR. Explanations by professional care staff, optimism and helping behavior: an application of attribution theory. Psychol Med. 1990; 20:849-855. [8] Reisenzein R. A structural equation analysis of Weiner's attribution-affect model of helping behavior. J Pers Soc Psychol. 1986; 50:1123-1133. [9] Batson CD, Duncan BD, Ackerman P, et al. Is emphatic emotion a source of altruistic motivation? J Pers Soc Psychol. 1981; 40:290-302. [10] Schmidt G, Weiner B. An attribution-affectaction theory of behavior: replications of judgments of help-giving. Personality and Social Psychology Bulletin. 1988; 14:610-621. [11] Betancourt H. An attribution-empathy model of helping behavior: behavioral intentions and judgments of help-giving. Personality and Social Psychology Bulletin. 1990; 16: 573-591. [12] Williams CA. Empathy and burnout in male and female helping professionals. Res Nurs Health. 1989; 12:169-178. [13] Maslach C. Burnout. The Cost of Caring. New York, NY: Prentice-Hall Press; 1982. [14] Donahoe E. Nawawi A, Wilker L, et al. Factors associated with burnout of physical therapists in Massachusetts rehabilitation hospitals. Phys Ther. 1993; 73:750-761. [15] Deckard GJ, Present R. Impact of role stress on physical therapists emotional and physical well-being. Phys Ther. 1989; 69:713-718. [16] Hovestadt AJ, Ander son WT, Piercy FP, et al. The family of origin scale. Journal of Marital and Family Therapy. 1985; 11:287-297. [17] Lee RE, Gordon NG, O'Dell JW. The validity and use of the family of origin scale. journal of Marital and Family Therapy. 1989; 15:19-27. [18] Mazer GE, Mangrum OL, Hovestadt AJ, Brashear RL. Further validation of the family of origin scale: a factor analysis, Journal of Marital and Family Therapy, 1990; 16:423-426. [19] Jacobsen B. Role model concepts before and after the formal professional socialization period. Phys Ther. 1980; 60:188-193.

KA Curtis, PhD, PT, is Associate Professor, Department of Physical Therapy, California State University, Fresno, 2345 E San Ramon Ave, Fresno, CA 93740-0029 (USA) (kathleen [e-mail]). She was Assistant Professor, Division of Physical Therapy, Department of Orthopaedics and Rehabilitation, University of Miami School of Medicine, Coral Gables, FL 33146, at the time this study was conducted. Address all correspondence to Dr Curtis.

CM Davis, EdD, PT, is Associate Professor, Division of Physical Therapy, Department of Orthopaedics and Rehabilitation, University of Miami School of Medicine.

TK Trimble, PT, is Staff Physical Therapist, Florida Sportsmedicine Center, South Miami, FL 33243.

DK Papoulidis, PT, is Staff Physical Therapist, Department of Physical Therapy, Jackson Memorial Hospital, Miami, FL 33136.

Ms Trimble and Ms Papoulidis were graduate students, Division of Physical Therapy, Department of Orthopaedics and Rehabilitation, University of Miami School of Medicine, at the time the study was conducted.

This study was approved by the Behavioral Sciences Subcommittee for the Protection of Human Subjects, University of Miami.

This article was submitted January 17, 1995, and was accepted August 4, 1995.