Changes In Functional Status Related To health Maintenance Visits To Family Physicians

Preventive care is not sought by patients or provided by physicians at the level recommended by national organizations.¹ This may be because of inadequate attention resulting from a physician’s lack of training,²⁴ forgetting to provide preventive care,⁵ negative attitude toward such care,⁶ or low confidence in its effectiveness.⁷ The low level of preventive services delivery may also be caused by inadequate reimbursement to physicians,⁸ out-of-pocket costs to patients, patient fears of finding disease, patients’ health beliefs,⁹ lack of agreement between the physician and the patient regarding the need for behavioral change,¹⁰ and lack of information given to the patient.¹¹ It is likely that the interaction among multiple factors results in the lack of provision of preventive care. Our previous work¹¹ suggested an additional explanation: We found statistically significant declines in emotional and social functioning of patients who had been advised to change health-related behaviors, even though no organic illness was diagnosed. Similar results were reported by Stoate,¹² who found patients with no acute complaints felt worse after receiving routine preventive care. Assuming the declines in functioning are representative of a widespread phenomenon, this may explain patients’ resistance to purely preventive care and provide insight into why physicians cite overall lack of gratification and satisfaction with providing it.^2,3

These findings may be an extension of other known negative effects of preventive medicine. It is known that the diagnosis (or labeling) of asymptomatic patients with diseases is associated with negative outcomes. For example, the diagnosis of asymptomatic hypertension has been associated with a greater number of sick days,¹³ as well as lower income.^14,15 The investigators did not find a decrease in psychological well-being, however. It is likely that similar functional changes could occur with the diagnoses of other conditions. For example, the cessation of smoking can cause physical symptoms (withdrawal) and be associated with the onset of depression.¹⁶ Starting a health promotion habit, such as physical exercise, may cause temporary symptoms as well. Patients may feel guilty if not engaging in the healthy behavior that was recommended or disappointed if results from behavioral change are not immediate. Changes in the way that a family functions may result from the knowledge of a new diagnosis or new behaviors, such as dietary changes.

We hypothesized that the pressure exerted by the physician’s advice challenges patients with limited confidence in their ability to manage change and causes a decrease in social and emotional functioning. To investigate this phenomenon, we conducted an observational study involving patients presenting for health maintenance in an academic family practice center.

Methods

Our study was completed in 3 phases: recruitment and baseline data collection, postvisit data collection, and telephone follow-up of patients at 1 month and 3 months. Patients from all socioeconomic strata aged 18 years and older presenting for health maintenance visits at the Family Practice Center of Bowman Gray School of Medicine were eligible. The Family Practice Center is an academic office where family-physician faculty, residents, and physician assistants care for patients. Appointment lists were screened to identify likely candidates, excluding those who were younger than 18 years and those presenting for acute care. Patients who met the inclusion criteria were contacted by telephone before their clinic visits, and after providing informed consent were given the Duke Health Profile (DUKE)¹⁷ and a brief intake questionnaire. We used the intake questionnaire to gather information on the reasons for visiting the clinician, previous experience with the clinician, and visit expectations. The DUKE profile is a 17-item questionnaire with 6 health measures (physical, mental, social, general, perceived health, and self-esteem) and 4 dysfunction measures (anxiety, depression, pain, and disability). The DUKE takes a broad view of health, has been validated in family practice populations, and is easy to administer.¹⁸

Enrolled patients were randomized into 2 groups in a 1:2 ratio: an observation-only group and a group that received a debriefing interview after their visit to the family practice clinician. As they left the examination room, the interview group was asked to complete a brief questionnaire that included a patient satisfaction instrument, the Patient/Doctor Interaction Scale (PDIS),¹⁹ and a debriefing instrument. The debriefing instrument addressed the patient’s views about the visit, specific behavioral changes recommended by the provider (with no preset response set), patient’s perception of the need for behavioral change, methods suggested to accomplish the change, and the patient’s perception of the likelihood of success in accomplishing the change. The PDIS is a 17-item patient satisfaction scale that assesses the portion of patient satisfaction involving interactions with the physician; we modified it slightly by adding 3 more general satisfaction questions. It was developed and validated in a family practice office and has been shown to be related to higher recall rates.²⁰ The scale has balanced positive and negative questions, uses a 5-point scale, and has an adequate completion rate. The clinicians were also asked to complete a brief questionnaire characterizing their perceptions of encounters with patients enrolled in both groups.

There was telephone follow-up of all patients at 1 month and 3 months after the visit to the clinician. A maximum of 6 attempts was made to contact participants. The telephone calls included repeat administration of the DUKE to assess functional status, questions about additional visits to the clinician or other healthcare providers, and about progress toward achieving recommended behavioral changes. We included all data in the analysis, in concordance with the intention-to-treat principle.

Data from completed forms were entered into a database by a trained, experienced research assistant. Before entry, each form was inspected for completeness, ambiguity of responses, or other irregularities. All unclear responses were referred to the investigators. Range checks were conducted periodically as data were entered to detect errors and were repeated as part of the data cleaning procedures before analysis. Descriptive statistics were calculated on all variables, including the DUKE subscale scores and the PDIS scores. Initial statistical analysis was carried out to test for differences in DUKE means between patients randomized to the observation-only group and the debriefing group. Differences in the PDIS and DUKE subscale means were tested at baseline, 1 month, and 3 months using a repeated measures approach (SAS subroutine PROC MIXED, SAS Institute, Cary, NC).

Results

We recruited participants during a 9-month period beginning in September 1995. Including the 3-month follow-up period, data collection was completed in 12 months. Of the 208 patients recruited, 68 (34%) were randomized to the observation-only group, and 132 (68%) were assigned to receive the debriefing interview. In the observation-only group, 64 (94%) patients were successfully contacted for the 1-month assessment, and 62 (91%) for the 3-month assessment. Of the 132 patients assigned to the debriefing group, 2 refused to complete the debriefing interview. Of those completing the debriefing interview, 123 (93%) were successfully contacted for the 1-month assessment, and 122 (92%) for the 3-month assessment.

The average age of the enrolled patients was 47.4 years (standard deviation [SD] =11.9, range=19-76 years) and 68.0% were women; 32.5% were African American, 65.0% were white, and the rest represented a variety of ethnic groups Table 1. The average educational level of the patients was 14.2 years (SD=3.0). A percent of 12.8 reported annual family incomes less than $15,000, 13.4% between $15,000 and $25,000, and more than half had incomes of $35,000 or more. The interview group had higher income levels than the observation group; otherwise there were no significant differences.

The reasons for patient visits were Papanicolaou tests, pelvic examinations, and routine health maintenance, although some of these visits incorporated a follow-up of a chronic condition. No statistically significant differences were found between the study groups by reason for visit. After the visits, 63 (48%) of the 132 patients in the interview group reported that their clinician recommended a specific type of behavioral change. Of the patients reporting having been given a recommendation, 11 were asked to quit smoking; 15 to change medications or the way that medications were taken; 33 to alter their diet, exercise level, or lose weight; and 4 received recommendations related to stress reduction. Additional miscellaneous changes were also recommended. It is interesting that patients sometimes reported some form of mental health behavioral change, but alcohol was only mentioned rarely by the patients. We expected behavioral changes related to alcohol use or abuse to be mentioned frequently.

PDIS assessment only occurred at baseline. The mean PDIS score for all observations was 55.8 (SD=5.5, range=35-66). No significant differences in satisfaction were found between the observation and interview groups, and there was no difference in patient satisfaction found between patients who reported receiving a recommendation for behavioral change and those who did not.

We performed the analysis of the outcomes (ie, changes in functional status by recommendation to change behavior), in 2 steps. In the first step, we examined the effect of the debriefing interview by comparing the DUKE subscale scores for the interview and observation groups at 1 month and 3 months. No significant differences were found for any of the DUKE subscales. Based on results indicating that the debriefing interviews had no effect on outcomes, we created 2 groups for subsequent analyses. One group consisted of patients who reported that they received behavioral change recommendations. The second group included all other patients enrolled in the study. This is a conservative method that would tend to underestimate differences because a portion of the observation group may have received behavioral change recommendations. Analysis of differences between these 2 groups during the 3 observations was carried out using a repeated measures approach. A repeated measures model (using SAS PROC MIXED) was fit for each of the DUKE subscale means. The model included time of assessment (baseline, 1 month, 3 months), race, age, sex, and educational level as independent variables. Table 1, Table 2 shows the means for each of the DUKE subscales at each assessment for patients who did and did not receive recommendations to change behavior. As Table 2 indicates, no significant differences (P <.05) were found between patients who did and did not receive a recommendation to change behavior at baseline. At the 1-month follow-up assessment, the mean scores for mental health, social health, and self-esteem were lower for patients who received a behavioral change recommendation. At 3 months, the differences in mental, social, and self-esteem found at 1 month persisted, and the means coresforanxiety, anxiety/depression, and depression were worse for patients receiving recommendations. Race and sex were not significantly associated with differences for any of the DUKE subscale scores shown in Table 2. Education, however, was significantly associated with every subscale. Age was associated with self-esteem. The functional status scores for patients who received recommendations to change behavior declined as educational level increased. Older, better-educated patients who received behavioral change recommendations were the most likely to report reduced functional status and self-esteem during the 3-month follow-up period. Finally, as the results in Table 2 show, the effects observed in the patients who received recommendations were consistent at the 1-month and 3-month assessments. The differences were all in the negative direction except for the pain subscale score. Reports of pain decreased from the baseline to the 1-month assessment and then were higher at the 3-month assessment.

The Effect of Specific Recommendations

To further investigate influences on DUKE subscale scores, we evaluated the effect of the type of change recommended by the clinician for patients given a specific type of recommendation.

Four categories of recommended changes were found: medication compliance, diet and exercise, smoking cessation, and stress control. Table 3 shows the mean DUKE subscale scores by the type of change recommended. As Table 3 indicates, statistically significant differences among the mean scores by type of change recommended were found at the 3-month assessment for disability, mental health, and self-esteem. Post hoc examination of the mean scores in these cases suggested that disability scores for patients who were asked to stop smoking were worse than those of patients asked to improve medication compliance or change their diet. Mental health scores for patients who were asked to stop smoking were significantly poorer than for those asked to try to control stress. No group differences were found for self-esteem scores.

Discussion

The results from our study provide further support that certain elements of patient functioning decline 3 months after behavioral change is recommended by a clinician. These results confirm our previous findings¹¹ and those reported by others.¹² We found that social and emotional functioning varied according to whether the patient reported that their clinician recommended that they change a behavior. The observation group had functional declines similar to the entire interview group; the declines fell intermediately between those of patients who reported being asked to make behavioral changes (who had functional declines) and those who reported no behavioral changes (who had no functional declines). This suggests that the debriefing after the office visit had no impact on the outcomes, and reinforces that it was the behavioral change recommendation that led to the declines in self-reported functional status.

Negative reactions by patients to recommendations for behavioral change may be one reason that such changes are not recommended more often by clinicians. Patients may also resist getting recommendations for behavioral change by not raising the subject with the clinician or by avoiding visits where such recommendations would be likely to occur (such as physical examinations). Physicians may avoid recommending changes to reduce potential conflict with the patient.

Declines in functioning were greatest when the physician recommended that the patient stop smoking, and potential declines were seen for patients with diet or exercise recommendations. No detectable functional decline occurred for those patients who had been given recommendations to make medication-related or dietary changes. This suggests that behavioral changes are perceived as being more difficult for patients than simple medication changes and that being faced with a recommendation to make a behavioral change is associated with lower levels of functioning. The patients who reported being told to quit smoking had marked increases in depression and anxiety and overall disability at 3 months after the visit. These results are consistent with the literature that smoking and depression are interlinked.¹⁶ Our study also supports the use of antidepressants such as buproprion²¹ for helping patients to quit smoking, but our study was done before the drug was approved and commonly used for this indication.

Limitations

The results from our study are subject to limitations and should be interpreted cautiously. First, we did not include sufficient numbers to allow testing of the association between success in the specific behavioral changes and functional decline. It is possible that the decline in functional status is limited to patients who were not successful in changing behavior. It is also possible that the declines are focused on a select group of behavioral changes, such as smoking. Further study is needed to test such associations. Second, data were collected at only 1 family practice center. The patient population is diverse but not necessarily representative of the community at large. Our study was also completed in North Carolina, a state known for tobacco consumption, which may have affected the results pertaining to smoking-related behavioral changes and functional decline. Further studies should emphasize whether functional declines reverse with a longer time frame and whether there is a relationship with successful behavioral change. Research should also consider whether physician behaviors can have a positive impact on a patient’s functional status.

Acknowledgments

Our research was supported by a grant from the American Academy of Family Physicians Foundation (#G9609). We would like to thank Dottie Greek for excellence in project management.

Preventive care is not sought by patients or provided by physicians at the level recommended by national organizations.¹ This may be because of inadequate attention resulting from a physician’s lack of training,²⁴ forgetting to provide preventive care,⁵ negative attitude toward such care,⁶ or low confidence in its effectiveness.⁷ The low level of preventive services delivery may also be caused by inadequate reimbursement to physicians,⁸ out-of-pocket costs to patients, patient fears of finding disease, patients’ health beliefs,⁹ lack of agreement between the physician and the patient regarding the need for behavioral change,¹⁰ and lack of information given to the patient.¹¹ It is likely that the interaction among multiple factors results in the lack of provision of preventive care. Our previous work¹¹ suggested an additional explanation: We found statistically significant declines in emotional and social functioning of patients who had been advised to change health-related behaviors, even though no organic illness was diagnosed. Similar results were reported by Stoate,¹² who found patients with no acute complaints felt worse after receiving routine preventive care. Assuming the declines in functioning are representative of a widespread phenomenon, this may explain patients’ resistance to purely preventive care and provide insight into why physicians cite overall lack of gratification and satisfaction with providing it.^2,3

These findings may be an extension of other known negative effects of preventive medicine. It is known that the diagnosis (or labeling) of asymptomatic patients with diseases is associated with negative outcomes. For example, the diagnosis of asymptomatic hypertension has been associated with a greater number of sick days,¹³ as well as lower income.^14,15 The investigators did not find a decrease in psychological well-being, however. It is likely that similar functional changes could occur with the diagnoses of other conditions. For example, the cessation of smoking can cause physical symptoms (withdrawal) and be associated with the onset of depression.¹⁶ Starting a health promotion habit, such as physical exercise, may cause temporary symptoms as well. Patients may feel guilty if not engaging in the healthy behavior that was recommended or disappointed if results from behavioral change are not immediate. Changes in the way that a family functions may result from the knowledge of a new diagnosis or new behaviors, such as dietary changes.

We hypothesized that the pressure exerted by the physician’s advice challenges patients with limited confidence in their ability to manage change and causes a decrease in social and emotional functioning. To investigate this phenomenon, we conducted an observational study involving patients presenting for health maintenance in an academic family practice center.

Methods

Our study was completed in 3 phases: recruitment and baseline data collection, postvisit data collection, and telephone follow-up of patients at 1 month and 3 months. Patients from all socioeconomic strata aged 18 years and older presenting for health maintenance visits at the Family Practice Center of Bowman Gray School of Medicine were eligible. The Family Practice Center is an academic office where family-physician faculty, residents, and physician assistants care for patients. Appointment lists were screened to identify likely candidates, excluding those who were younger than 18 years and those presenting for acute care. Patients who met the inclusion criteria were contacted by telephone before their clinic visits, and after providing informed consent were given the Duke Health Profile (DUKE)¹⁷ and a brief intake questionnaire. We used the intake questionnaire to gather information on the reasons for visiting the clinician, previous experience with the clinician, and visit expectations. The DUKE profile is a 17-item questionnaire with 6 health measures (physical, mental, social, general, perceived health, and self-esteem) and 4 dysfunction measures (anxiety, depression, pain, and disability). The DUKE takes a broad view of health, has been validated in family practice populations, and is easy to administer.¹⁸

Enrolled patients were randomized into 2 groups in a 1:2 ratio: an observation-only group and a group that received a debriefing interview after their visit to the family practice clinician. As they left the examination room, the interview group was asked to complete a brief questionnaire that included a patient satisfaction instrument, the Patient/Doctor Interaction Scale (PDIS),¹⁹ and a debriefing instrument. The debriefing instrument addressed the patient’s views about the visit, specific behavioral changes recommended by the provider (with no preset response set), patient’s perception of the need for behavioral change, methods suggested to accomplish the change, and the patient’s perception of the likelihood of success in accomplishing the change. The PDIS is a 17-item patient satisfaction scale that assesses the portion of patient satisfaction involving interactions with the physician; we modified it slightly by adding 3 more general satisfaction questions. It was developed and validated in a family practice office and has been shown to be related to higher recall rates.²⁰ The scale has balanced positive and negative questions, uses a 5-point scale, and has an adequate completion rate. The clinicians were also asked to complete a brief questionnaire characterizing their perceptions of encounters with patients enrolled in both groups.

There was telephone follow-up of all patients at 1 month and 3 months after the visit to the clinician. A maximum of 6 attempts was made to contact participants. The telephone calls included repeat administration of the DUKE to assess functional status, questions about additional visits to the clinician or other healthcare providers, and about progress toward achieving recommended behavioral changes. We included all data in the analysis, in concordance with the intention-to-treat principle.

Data from completed forms were entered into a database by a trained, experienced research assistant. Before entry, each form was inspected for completeness, ambiguity of responses, or other irregularities. All unclear responses were referred to the investigators. Range checks were conducted periodically as data were entered to detect errors and were repeated as part of the data cleaning procedures before analysis. Descriptive statistics were calculated on all variables, including the DUKE subscale scores and the PDIS scores. Initial statistical analysis was carried out to test for differences in DUKE means between patients randomized to the observation-only group and the debriefing group. Differences in the PDIS and DUKE subscale means were tested at baseline, 1 month, and 3 months using a repeated measures approach (SAS subroutine PROC MIXED, SAS Institute, Cary, NC).

Results

We recruited participants during a 9-month period beginning in September 1995. Including the 3-month follow-up period, data collection was completed in 12 months. Of the 208 patients recruited, 68 (34%) were randomized to the observation-only group, and 132 (68%) were assigned to receive the debriefing interview. In the observation-only group, 64 (94%) patients were successfully contacted for the 1-month assessment, and 62 (91%) for the 3-month assessment. Of the 132 patients assigned to the debriefing group, 2 refused to complete the debriefing interview. Of those completing the debriefing interview, 123 (93%) were successfully contacted for the 1-month assessment, and 122 (92%) for the 3-month assessment.

The average age of the enrolled patients was 47.4 years (standard deviation [SD] =11.9, range=19-76 years) and 68.0% were women; 32.5% were African American, 65.0% were white, and the rest represented a variety of ethnic groups Table 1. The average educational level of the patients was 14.2 years (SD=3.0). A percent of 12.8 reported annual family incomes less than $15,000, 13.4% between $15,000 and $25,000, and more than half had incomes of $35,000 or more. The interview group had higher income levels than the observation group; otherwise there were no significant differences.

The reasons for patient visits were Papanicolaou tests, pelvic examinations, and routine health maintenance, although some of these visits incorporated a follow-up of a chronic condition. No statistically significant differences were found between the study groups by reason for visit. After the visits, 63 (48%) of the 132 patients in the interview group reported that their clinician recommended a specific type of behavioral change. Of the patients reporting having been given a recommendation, 11 were asked to quit smoking; 15 to change medications or the way that medications were taken; 33 to alter their diet, exercise level, or lose weight; and 4 received recommendations related to stress reduction. Additional miscellaneous changes were also recommended. It is interesting that patients sometimes reported some form of mental health behavioral change, but alcohol was only mentioned rarely by the patients. We expected behavioral changes related to alcohol use or abuse to be mentioned frequently.

PDIS assessment only occurred at baseline. The mean PDIS score for all observations was 55.8 (SD=5.5, range=35-66). No significant differences in satisfaction were found between the observation and interview groups, and there was no difference in patient satisfaction found between patients who reported receiving a recommendation for behavioral change and those who did not.

We performed the analysis of the outcomes (ie, changes in functional status by recommendation to change behavior), in 2 steps. In the first step, we examined the effect of the debriefing interview by comparing the DUKE subscale scores for the interview and observation groups at 1 month and 3 months. No significant differences were found for any of the DUKE subscales. Based on results indicating that the debriefing interviews had no effect on outcomes, we created 2 groups for subsequent analyses. One group consisted of patients who reported that they received behavioral change recommendations. The second group included all other patients enrolled in the study. This is a conservative method that would tend to underestimate differences because a portion of the observation group may have received behavioral change recommendations. Analysis of differences between these 2 groups during the 3 observations was carried out using a repeated measures approach. A repeated measures model (using SAS PROC MIXED) was fit for each of the DUKE subscale means. The model included time of assessment (baseline, 1 month, 3 months), race, age, sex, and educational level as independent variables. Table 1, Table 2 shows the means for each of the DUKE subscales at each assessment for patients who did and did not receive recommendations to change behavior. As Table 2 indicates, no significant differences (P <.05) were found between patients who did and did not receive a recommendation to change behavior at baseline. At the 1-month follow-up assessment, the mean scores for mental health, social health, and self-esteem were lower for patients who received a behavioral change recommendation. At 3 months, the differences in mental, social, and self-esteem found at 1 month persisted, and the means coresforanxiety, anxiety/depression, and depression were worse for patients receiving recommendations. Race and sex were not significantly associated with differences for any of the DUKE subscale scores shown in Table 2. Education, however, was significantly associated with every subscale. Age was associated with self-esteem. The functional status scores for patients who received recommendations to change behavior declined as educational level increased. Older, better-educated patients who received behavioral change recommendations were the most likely to report reduced functional status and self-esteem during the 3-month follow-up period. Finally, as the results in Table 2 show, the effects observed in the patients who received recommendations were consistent at the 1-month and 3-month assessments. The differences were all in the negative direction except for the pain subscale score. Reports of pain decreased from the baseline to the 1-month assessment and then were higher at the 3-month assessment.

The Effect of Specific Recommendations

To further investigate influences on DUKE subscale scores, we evaluated the effect of the type of change recommended by the clinician for patients given a specific type of recommendation.

Four categories of recommended changes were found: medication compliance, diet and exercise, smoking cessation, and stress control. Table 3 shows the mean DUKE subscale scores by the type of change recommended. As Table 3 indicates, statistically significant differences among the mean scores by type of change recommended were found at the 3-month assessment for disability, mental health, and self-esteem. Post hoc examination of the mean scores in these cases suggested that disability scores for patients who were asked to stop smoking were worse than those of patients asked to improve medication compliance or change their diet. Mental health scores for patients who were asked to stop smoking were significantly poorer than for those asked to try to control stress. No group differences were found for self-esteem scores.

Discussion

The results from our study provide further support that certain elements of patient functioning decline 3 months after behavioral change is recommended by a clinician. These results confirm our previous findings¹¹ and those reported by others.¹² We found that social and emotional functioning varied according to whether the patient reported that their clinician recommended that they change a behavior. The observation group had functional declines similar to the entire interview group; the declines fell intermediately between those of patients who reported being asked to make behavioral changes (who had functional declines) and those who reported no behavioral changes (who had no functional declines). This suggests that the debriefing after the office visit had no impact on the outcomes, and reinforces that it was the behavioral change recommendation that led to the declines in self-reported functional status.

Negative reactions by patients to recommendations for behavioral change may be one reason that such changes are not recommended more often by clinicians. Patients may also resist getting recommendations for behavioral change by not raising the subject with the clinician or by avoiding visits where such recommendations would be likely to occur (such as physical examinations). Physicians may avoid recommending changes to reduce potential conflict with the patient.

Declines in functioning were greatest when the physician recommended that the patient stop smoking, and potential declines were seen for patients with diet or exercise recommendations. No detectable functional decline occurred for those patients who had been given recommendations to make medication-related or dietary changes. This suggests that behavioral changes are perceived as being more difficult for patients than simple medication changes and that being faced with a recommendation to make a behavioral change is associated with lower levels of functioning. The patients who reported being told to quit smoking had marked increases in depression and anxiety and overall disability at 3 months after the visit. These results are consistent with the literature that smoking and depression are interlinked.¹⁶ Our study also supports the use of antidepressants such as buproprion²¹ for helping patients to quit smoking, but our study was done before the drug was approved and commonly used for this indication.

Limitations

The results from our study are subject to limitations and should be interpreted cautiously. First, we did not include sufficient numbers to allow testing of the association between success in the specific behavioral changes and functional decline. It is possible that the decline in functional status is limited to patients who were not successful in changing behavior. It is also possible that the declines are focused on a select group of behavioral changes, such as smoking. Further study is needed to test such associations. Second, data were collected at only 1 family practice center. The patient population is diverse but not necessarily representative of the community at large. Our study was also completed in North Carolina, a state known for tobacco consumption, which may have affected the results pertaining to smoking-related behavioral changes and functional decline. Further studies should emphasize whether functional declines reverse with a longer time frame and whether there is a relationship with successful behavioral change. Research should also consider whether physician behaviors can have a positive impact on a patient’s functional status.

Acknowledgments

Our research was supported by a grant from the American Academy of Family Physicians Foundation (#G9609). We would like to thank Dottie Greek for excellence in project management.

Preventive care is not sought by patients or provided by physicians at the level recommended by national organizations.¹ This may be because of inadequate attention resulting from a physician’s lack of training,²⁴ forgetting to provide preventive care,⁵ negative attitude toward such care,⁶ or low confidence in its effectiveness.⁷ The low level of preventive services delivery may also be caused by inadequate reimbursement to physicians,⁸ out-of-pocket costs to patients, patient fears of finding disease, patients’ health beliefs,⁹ lack of agreement between the physician and the patient regarding the need for behavioral change,¹⁰ and lack of information given to the patient.¹¹ It is likely that the interaction among multiple factors results in the lack of provision of preventive care. Our previous work¹¹ suggested an additional explanation: We found statistically significant declines in emotional and social functioning of patients who had been advised to change health-related behaviors, even though no organic illness was diagnosed. Similar results were reported by Stoate,¹² who found patients with no acute complaints felt worse after receiving routine preventive care. Assuming the declines in functioning are representative of a widespread phenomenon, this may explain patients’ resistance to purely preventive care and provide insight into why physicians cite overall lack of gratification and satisfaction with providing it.^2,3

These findings may be an extension of other known negative effects of preventive medicine. It is known that the diagnosis (or labeling) of asymptomatic patients with diseases is associated with negative outcomes. For example, the diagnosis of asymptomatic hypertension has been associated with a greater number of sick days,¹³ as well as lower income.^14,15 The investigators did not find a decrease in psychological well-being, however. It is likely that similar functional changes could occur with the diagnoses of other conditions. For example, the cessation of smoking can cause physical symptoms (withdrawal) and be associated with the onset of depression.¹⁶ Starting a health promotion habit, such as physical exercise, may cause temporary symptoms as well. Patients may feel guilty if not engaging in the healthy behavior that was recommended or disappointed if results from behavioral change are not immediate. Changes in the way that a family functions may result from the knowledge of a new diagnosis or new behaviors, such as dietary changes.

We hypothesized that the pressure exerted by the physician’s advice challenges patients with limited confidence in their ability to manage change and causes a decrease in social and emotional functioning. To investigate this phenomenon, we conducted an observational study involving patients presenting for health maintenance in an academic family practice center.

Methods

Our study was completed in 3 phases: recruitment and baseline data collection, postvisit data collection, and telephone follow-up of patients at 1 month and 3 months. Patients from all socioeconomic strata aged 18 years and older presenting for health maintenance visits at the Family Practice Center of Bowman Gray School of Medicine were eligible. The Family Practice Center is an academic office where family-physician faculty, residents, and physician assistants care for patients. Appointment lists were screened to identify likely candidates, excluding those who were younger than 18 years and those presenting for acute care. Patients who met the inclusion criteria were contacted by telephone before their clinic visits, and after providing informed consent were given the Duke Health Profile (DUKE)¹⁷ and a brief intake questionnaire. We used the intake questionnaire to gather information on the reasons for visiting the clinician, previous experience with the clinician, and visit expectations. The DUKE profile is a 17-item questionnaire with 6 health measures (physical, mental, social, general, perceived health, and self-esteem) and 4 dysfunction measures (anxiety, depression, pain, and disability). The DUKE takes a broad view of health, has been validated in family practice populations, and is easy to administer.¹⁸

Enrolled patients were randomized into 2 groups in a 1:2 ratio: an observation-only group and a group that received a debriefing interview after their visit to the family practice clinician. As they left the examination room, the interview group was asked to complete a brief questionnaire that included a patient satisfaction instrument, the Patient/Doctor Interaction Scale (PDIS),¹⁹ and a debriefing instrument. The debriefing instrument addressed the patient’s views about the visit, specific behavioral changes recommended by the provider (with no preset response set), patient’s perception of the need for behavioral change, methods suggested to accomplish the change, and the patient’s perception of the likelihood of success in accomplishing the change. The PDIS is a 17-item patient satisfaction scale that assesses the portion of patient satisfaction involving interactions with the physician; we modified it slightly by adding 3 more general satisfaction questions. It was developed and validated in a family practice office and has been shown to be related to higher recall rates.²⁰ The scale has balanced positive and negative questions, uses a 5-point scale, and has an adequate completion rate. The clinicians were also asked to complete a brief questionnaire characterizing their perceptions of encounters with patients enrolled in both groups.

There was telephone follow-up of all patients at 1 month and 3 months after the visit to the clinician. A maximum of 6 attempts was made to contact participants. The telephone calls included repeat administration of the DUKE to assess functional status, questions about additional visits to the clinician or other healthcare providers, and about progress toward achieving recommended behavioral changes. We included all data in the analysis, in concordance with the intention-to-treat principle.

Data from completed forms were entered into a database by a trained, experienced research assistant. Before entry, each form was inspected for completeness, ambiguity of responses, or other irregularities. All unclear responses were referred to the investigators. Range checks were conducted periodically as data were entered to detect errors and were repeated as part of the data cleaning procedures before analysis. Descriptive statistics were calculated on all variables, including the DUKE subscale scores and the PDIS scores. Initial statistical analysis was carried out to test for differences in DUKE means between patients randomized to the observation-only group and the debriefing group. Differences in the PDIS and DUKE subscale means were tested at baseline, 1 month, and 3 months using a repeated measures approach (SAS subroutine PROC MIXED, SAS Institute, Cary, NC).

Results

We recruited participants during a 9-month period beginning in September 1995. Including the 3-month follow-up period, data collection was completed in 12 months. Of the 208 patients recruited, 68 (34%) were randomized to the observation-only group, and 132 (68%) were assigned to receive the debriefing interview. In the observation-only group, 64 (94%) patients were successfully contacted for the 1-month assessment, and 62 (91%) for the 3-month assessment. Of the 132 patients assigned to the debriefing group, 2 refused to complete the debriefing interview. Of those completing the debriefing interview, 123 (93%) were successfully contacted for the 1-month assessment, and 122 (92%) for the 3-month assessment.

The average age of the enrolled patients was 47.4 years (standard deviation [SD] =11.9, range=19-76 years) and 68.0% were women; 32.5% were African American, 65.0% were white, and the rest represented a variety of ethnic groups Table 1. The average educational level of the patients was 14.2 years (SD=3.0). A percent of 12.8 reported annual family incomes less than $15,000, 13.4% between $15,000 and $25,000, and more than half had incomes of $35,000 or more. The interview group had higher income levels than the observation group; otherwise there were no significant differences.

The reasons for patient visits were Papanicolaou tests, pelvic examinations, and routine health maintenance, although some of these visits incorporated a follow-up of a chronic condition. No statistically significant differences were found between the study groups by reason for visit. After the visits, 63 (48%) of the 132 patients in the interview group reported that their clinician recommended a specific type of behavioral change. Of the patients reporting having been given a recommendation, 11 were asked to quit smoking; 15 to change medications or the way that medications were taken; 33 to alter their diet, exercise level, or lose weight; and 4 received recommendations related to stress reduction. Additional miscellaneous changes were also recommended. It is interesting that patients sometimes reported some form of mental health behavioral change, but alcohol was only mentioned rarely by the patients. We expected behavioral changes related to alcohol use or abuse to be mentioned frequently.

PDIS assessment only occurred at baseline. The mean PDIS score for all observations was 55.8 (SD=5.5, range=35-66). No significant differences in satisfaction were found between the observation and interview groups, and there was no difference in patient satisfaction found between patients who reported receiving a recommendation for behavioral change and those who did not.

We performed the analysis of the outcomes (ie, changes in functional status by recommendation to change behavior), in 2 steps. In the first step, we examined the effect of the debriefing interview by comparing the DUKE subscale scores for the interview and observation groups at 1 month and 3 months. No significant differences were found for any of the DUKE subscales. Based on results indicating that the debriefing interviews had no effect on outcomes, we created 2 groups for subsequent analyses. One group consisted of patients who reported that they received behavioral change recommendations. The second group included all other patients enrolled in the study. This is a conservative method that would tend to underestimate differences because a portion of the observation group may have received behavioral change recommendations. Analysis of differences between these 2 groups during the 3 observations was carried out using a repeated measures approach. A repeated measures model (using SAS PROC MIXED) was fit for each of the DUKE subscale means. The model included time of assessment (baseline, 1 month, 3 months), race, age, sex, and educational level as independent variables. Table 1, Table 2 shows the means for each of the DUKE subscales at each assessment for patients who did and did not receive recommendations to change behavior. As Table 2 indicates, no significant differences (P <.05) were found between patients who did and did not receive a recommendation to change behavior at baseline. At the 1-month follow-up assessment, the mean scores for mental health, social health, and self-esteem were lower for patients who received a behavioral change recommendation. At 3 months, the differences in mental, social, and self-esteem found at 1 month persisted, and the means coresforanxiety, anxiety/depression, and depression were worse for patients receiving recommendations. Race and sex were not significantly associated with differences for any of the DUKE subscale scores shown in Table 2. Education, however, was significantly associated with every subscale. Age was associated with self-esteem. The functional status scores for patients who received recommendations to change behavior declined as educational level increased. Older, better-educated patients who received behavioral change recommendations were the most likely to report reduced functional status and self-esteem during the 3-month follow-up period. Finally, as the results in Table 2 show, the effects observed in the patients who received recommendations were consistent at the 1-month and 3-month assessments. The differences were all in the negative direction except for the pain subscale score. Reports of pain decreased from the baseline to the 1-month assessment and then were higher at the 3-month assessment.

The Effect of Specific Recommendations

To further investigate influences on DUKE subscale scores, we evaluated the effect of the type of change recommended by the clinician for patients given a specific type of recommendation.

Four categories of recommended changes were found: medication compliance, diet and exercise, smoking cessation, and stress control. Table 3 shows the mean DUKE subscale scores by the type of change recommended. As Table 3 indicates, statistically significant differences among the mean scores by type of change recommended were found at the 3-month assessment for disability, mental health, and self-esteem. Post hoc examination of the mean scores in these cases suggested that disability scores for patients who were asked to stop smoking were worse than those of patients asked to improve medication compliance or change their diet. Mental health scores for patients who were asked to stop smoking were significantly poorer than for those asked to try to control stress. No group differences were found for self-esteem scores.

Discussion

The results from our study provide further support that certain elements of patient functioning decline 3 months after behavioral change is recommended by a clinician. These results confirm our previous findings¹¹ and those reported by others.¹² We found that social and emotional functioning varied according to whether the patient reported that their clinician recommended that they change a behavior. The observation group had functional declines similar to the entire interview group; the declines fell intermediately between those of patients who reported being asked to make behavioral changes (who had functional declines) and those who reported no behavioral changes (who had no functional declines). This suggests that the debriefing after the office visit had no impact on the outcomes, and reinforces that it was the behavioral change recommendation that led to the declines in self-reported functional status.

Negative reactions by patients to recommendations for behavioral change may be one reason that such changes are not recommended more often by clinicians. Patients may also resist getting recommendations for behavioral change by not raising the subject with the clinician or by avoiding visits where such recommendations would be likely to occur (such as physical examinations). Physicians may avoid recommending changes to reduce potential conflict with the patient.

Declines in functioning were greatest when the physician recommended that the patient stop smoking, and potential declines were seen for patients with diet or exercise recommendations. No detectable functional decline occurred for those patients who had been given recommendations to make medication-related or dietary changes. This suggests that behavioral changes are perceived as being more difficult for patients than simple medication changes and that being faced with a recommendation to make a behavioral change is associated with lower levels of functioning. The patients who reported being told to quit smoking had marked increases in depression and anxiety and overall disability at 3 months after the visit. These results are consistent with the literature that smoking and depression are interlinked.¹⁶ Our study also supports the use of antidepressants such as buproprion²¹ for helping patients to quit smoking, but our study was done before the drug was approved and commonly used for this indication.

Limitations

The results from our study are subject to limitations and should be interpreted cautiously. First, we did not include sufficient numbers to allow testing of the association between success in the specific behavioral changes and functional decline. It is possible that the decline in functional status is limited to patients who were not successful in changing behavior. It is also possible that the declines are focused on a select group of behavioral changes, such as smoking. Further study is needed to test such associations. Second, data were collected at only 1 family practice center. The patient population is diverse but not necessarily representative of the community at large. Our study was also completed in North Carolina, a state known for tobacco consumption, which may have affected the results pertaining to smoking-related behavioral changes and functional decline. Further studies should emphasize whether functional declines reverse with a longer time frame and whether there is a relationship with successful behavioral change. Research should also consider whether physician behaviors can have a positive impact on a patient’s functional status.

Acknowledgments

Our research was supported by a grant from the American Academy of Family Physicians Foundation (#G9609). We would like to thank Dottie Greek for excellence in project management.