Jennifer K. Carroll, MD, MPH

Fewer than half of all Americans get sufficient physical activity, defined as 30 minutes or more per day, at least 5 times per week.¹ The need to increase physical activity applies particularly to underserved populations: they are even less likely to get enough physical activity, and are thus even more likely to suffer greater burden of disease.^2,3

The purpose of this systematic review was to assess clinical trials of clinician-initiated counseling interventions for promoting physical activity in under-served populations. We define under-served populations as individuals from minority ethnic backgrounds (such as African Americans, Hispanics, and Asian Americans), or vulnerable populations such as people with low educational attainment, low income, lack of insurance, or those residing in rural communities.

Primary care interventions are linked to a change in habits

Primary care physicians can have a significant impact on their patients’ physical activity. Individuals with a regular primary care physician are more likely to report attempts to change their physical activity habits.⁴ However, underserved populations are more likely to have inconsistent access to medical care, which may contribute to their greater risk of conditions linked to inadequate physical activity, such as diabetes, hypertension, and obesity.

Only about 25% of patients in primary care settings report receiving any counseling on physical activity.⁵ Those who are middle-aged or have a baccalaureate degree or higher are more likely to report such advice; African Americans and foreign-born immigrants are less likely to report it.

A study by Taira et al⁶ examined the relationship between patient income and discussion of health risk behaviors. Low-income patients were more likely to be obese and smoke than high-income patients; however, physicians were less likely to discuss diet and exercise with low-income patients. Among all the patients with whom some discussion occurred in this study, low-income patients were much more likely to attempt to change behavior based on physician advice than were high-income patients.

Clinical trials within^7,8 and outside the US^9-11 support the potential value of physical activity counseling in primary care. In these studies, as little as 3 to 5 minutes of patient-clinician communication about physical activity was linked to short-term improvement in patients’ exercise habits. As few as 2 or 3 office visits over 6 months were associated with increases in patients’ physical activity levels up to 1 year later. Other features that contributed to their success included having a brief (<3 minutes) counseling component for clinicians, supplementing the counseling with a written exercise prescription, having follow-up contact, and tailoring the counseling to patients’ needs and concerns.

These results are promising for primary care clinicians, whose longitudinal relationships with their patients afford them repeated opportunities to intervene to promote physical activity.

Few studies have focused on the underserved

A review by Taylor et al² of physical activity interventions in low-income, ethnic minority, or disabled populations identified 14 community-based studies, mostly with quasi-experimental “pre/post” study designs. Ten studies included ethnic minorities, but physical activity was documented in just 2 studies at baseline, and these 2 studies did not include any postintervention follow-up. None of the 10 interventions was conducted in a primary care setting.

Another recent review¹² found that studies that were ethnically inclusive placed greater emphasis on involving communities and building coalitions right from study inception, and they tailored messages (and messengers) that were culturally specific. Several of these studies showed better outcomes among ethnic minority participants than the white participants they sampled.

Taken together, previous reviews have examined the effectiveness of primary care interventions for the general population,^13,14 as well as community-based programs for underserved populations.² However, little information exists about effective physical activity counseling strategies for underserved groups in primary care.

Methods

Looking for studies in underserved populations

We conducted a systematic review of the literature involving clinical trials in the US, looking for trials where counseling interventions are initiated by primary care clinicians, and that assessed behavioral change related to physical activity.

Inclusion criteria

TABLE 1 shows the inclusion criteria and search terms for the literature review. We searched Ovid, Medline, CINAHL, PsycINFO, PubMed, Cochrane, and HealthSTAR for studies published between 1966 and 2005. We also searched bibliographies of retrieved articles, and contacted experts in the field in an effort to obtain other relevant data.

The principal investigator (JKC) reviewed titles and abstracts of all potentially relevant articles to determine whether they met eligibility criteria. Studies that met the criteria were retrieved and abstracted.

Using these predefined criteria, data were extracted from each eligible article. Studies were also rated according to the Strength of Recommendation Taxonomy (SORT), because of its emphasis on patient-oriented outcomes and the quality, quantity, and consistency of evidence.¹⁵

TABLE 1
Inclusion criteria and search terms

Results

6 of 8 studies report increases in physical activity

We reviewed a total of 253 titles and abstracts. Eight studies^16-23 met our inclusion criteria. We were not able to locate any clinical trials that both 1) examined the effect of primary care clinician counseling on physical activity outcomes, and 2) had a study population focused on an underserved group. TABLE 2 (available at www.jfponline.com) shows the characteristics of these 8 studies.

Although we sought trials that defined “primary care clinician” as a professional—such as MD, nurse practitioner (NP), or physician assistant (PA)—who provides longitudinal primary health care, several of these studies considered dieticians, exercise physiologists, or health care workers as primary care clinicians.

Only 1 study²⁰ examined physical activity counseling with an intervention that incorporated a follow-up visit by the primary care clinician, and looked at the long-term effect on physical activity as an outcome. Thus, the degree to which the clinician’s counseling influenced the physical activity outcome in these studies is unclear.

Identifying underserved groups

Information on race or ethnicity (which tended to be reported as a single variable), level of education, and income of participants was reported in the demographic data of all studies’ results, but relationships between these variables and physical activity outcomes were not consistently reported. One study²³ stratified participants by race/ethnicity and health center; 2 studies^16,21 reported analyses and findings for participants according to ethnicity, income, and educational level, as that was their focus.

Overall, however, it is not clear to what extent the interventions succeeded for various underserved groups, even if they were included as participants.

Study designs and the nature of exercise interventions

Seven^16,18-23 of these studies (88%) were randomized controlled trials; the unit of randomization and control group varied. Trials were conducted at 1 or multiple (up to 11) primary care sites. Use of more than 1 method to recruit participants—such as mailings, use of office staff to promote/recruit, advertising, and community announcements—tended to be most effective.

Intervention types included phone and mail interventions,^17-23 computer-based interventions,^18,19,21 visits from a community health worker,^22,23 group classes,^16,22,23 directly supervised physical activity sessions,^16,22,23 clinician counseling,^16-23 and prescription protocols (eg, written, guided action plans).^17-23 Those delivering the intervention varied, and included primary care physicians,^17-23 nurse practitioners or physician assistants,^17-19,23,23 nutritionists,¹⁶ exercise physiologists,¹⁶ community health educators,^20,22,23 and other study personnel.^19,21 Specific elements of interventions that were likely to contribute to patients’ success included addressing financial or environmental/safety issues for exercise,¹⁶ use of trained office staff to provide exercise counseling,^18-20,23 and offering flexibility in choice by tailoring the goals and plans to the patients’ needs and interests.^17-23

The “dose” of clinician counseling varied from very brief (1 to 3 minutes of direct contact on 1 occasion) to more extended (>5 minutes of direct counseling over repeated intervals). Duration of follow-up for the 8 studies ranged from 4 months to 2 years.

Several studies designed their interventions to make the clinician counseling brief,^17-20,23 in order to enhance feasibility for busy primary care settings. Three studies^16,21,22 described strategies they used for tailoring the intervention to a specific culture, or for addressing issues of literacy for the written materials. Two studies^16,22 reported that their study staffs were ethnically or culturally representative of the targeted population.

The difficulty of maintaining adherence to physical activity

Three studies^18,19,21 reported having difficulty with attrition among their minority participants; they did not, however, include information specific to minorities in their physical activity outcomes. Studies with highest retention rates (>80%) tended to specifically address barriers to participation, including cultural issues, or they used a “lead-in” period.^16,20,21,23

The studies with the best adherence and retention among black and Hispanic participants, and those participants with low educational attainment,^16,21 used baseline qualitative data regarding management of health behaviors when they designed their interventions. For example, 1 study¹⁶ mentioned cultural adaptations derived from prior qualitative work—such as using program materials that extensively depicted African American individuals, families, and community settings—and using language in the intervention reflecting social values and situations relevant to African Americans.

How exercise data were reported

Six of the 8 (75%) studies^{16,17,19,20,22,23} reported some improvement in short-term physical activity outcomes (TABLE 2, available at www.jfponline.com); however, there was considerable heterogeneity in how these studies measured physical activity outcomes. All 8 incorporated a self-report measure of physical activity, such as the Patient-centered Assessment and Counseling for Exercise (PACE),^17-19 Paffenbarger Physical Activity Questionnaire (PPAQ),¹⁷ 7-day Physical Activity Recall (PAR),^17,20,21,23 and other self-report recall measures to assess physical activity. (A RESOURCE LIST of these instruments is available at www.jfponline.com.) Two studies also measured “states of change,”^17,20 but these states were not consistently defined.

Three studies^17,20,23 included objective measures of physical activity, such as accelerometers; in these studies, there was not substantial variance in physical activity outcomes between the objective and subjective measures.

Discussion

More study needed in the underserved

This review reflects in part the difficult task of designing and implementing realistic interventions for the underserved in primary care. However, interventions must be replicated in these populations before we can necessarily assume that findings from other trials are generalizable, due to issues of access, financial resources, health literacy, beliefs, cultural differences, self-efficacy, and other logistic barriers to traditional care that disproportionately affect underserved groups.

Integrate known personal, social, and environmental factors

Several studies^24-26 have explored the social, demographic, and environmental factors associated with physical activity in minority populations. These studies shed light on the reasons why clinical trials that focus on white, affluent, educated populations might not be generalizable to underserved groups.

To be maximally effective, any interventions for promoting physical activity in the underserved need to find ways to address any cultural or financial barriers, and incorporate factors associated with success. For example, among African American and Hispanic women, having lower “social role strain,” higher attendance at religious services, and a greater feeling that one’s neighborhood was safe were all associated with increased likelihood of exercise.^24-26 Such studies suggest that differences in beliefs, resources, self-efficacy, prior experience, and competing life demands can all contribute to promoting physical activity in some underserved groups. Practically, such findings encourage clinicians to work with patients to help them identify sources of social support and positive influences on their health, and help them articulate internal strengths and personal attributes to succeed in behavioral change.

Despite the variations in training or means of communication in the studies we identified, 2 studies used interventions that were successful at explicitly anticipated and addressed barriers to physical activity.^16,21 These 2 studies also had interventionists who represented the communities of interest, and they used cultural adaptations to promote exercise where appropriate. Thus, limited data suggest that some primary care–based programs improve physical activity in underserved patients, but the effects of communication from the primary care clinician on physical activity is lacking, consistent with other work in the field.^12,27

Promising strategies include office prompts, brief counseling

Primary care clinicians face many time pressures, fiscal constraints, administrative burdens, and competing priorities; these make addressing health promotion behaviors such as physical activity quite difficult. These issues are magnified for clinicians practicing in medically underserved areas. Despite these many challenges, promising opportunities do exist.

On a systems level, practice-based systems to manage chronic diseases have been successfully developed and implemented in the primary care setting; such systems can be tested to promote physical activity, as well. These practice-based approaches include patient registry data, office prompts, and other electronic systems to promote clinician counseling. For example, studies in this review using computer-based programs in primary care offices were feasible and effective.^18,19,21

Bodenheimer²⁸ has argued for a redesign of primary care systems to more effectively address chronic conditions rather than acute care needs. Several health care systems have successfully implemented the pillars of such a redesign imperative, and they have shown convincingly the promise of addressing competing priorities, physician competence and confidence, motivation, and durability in improving patient self-management.²⁸

At the level of the clinician-patient relationship, data suggest that patient physical activity can be increased (at least in the short term) by counseling that:

• is brief (5 minutes or less)^17-20,23
• is focused/goal-oriented^17-23
• is molded to the patient’s specific health needs^17-23
• is delivered over multiple contacts (whether it be office visits, telephone, or group sessions)^17-23
• contains a written plan to achieve goals.^17-23

We do not know what “dose-response” relationship exists for primary care clinician communication with patients over the long term, and what effect repeated counseling would have on long-term sustainability of physical activity levels. This is even less clear for underserved groups. It is also unknown to what extent collaborative links with community programs might increase physical activity when added to primary care–based counseling. Future research should evaluate the optimal “dose-response” to the interventions, the effect of repeated visits and continuity of care, and the effect of community-based referrals for physical activity programs for underserved populations in primary care.

Limitations of this review

Because our inclusion criteria were strict, we omitted potentially meaningful studies that were less directly relevant to our aims. For example, there has been substantial creative community-based work with underserved populations in the US to promote physical activity, and many innovations have been designed by researchers outside the US. Results from these programs and trials should be incorporated into primary care settings working with underserved populations.

Another limitation is that our definition of “underserved” is not the only possible definition. The most marginalized underserved groups with the least access to the health care system (such as the uninsured or homeless) were more likely to be omitted from our results, because we wanted to examine physical activity programs among patients in primary care settings.

Finally, this review did not address the need to understand the connection between sustained improvements in physical activity and patient-oriented health outcomes for underserved populations.

Conclusion

Information on exercise counseling interventions in primary care for the underserved is limited: these groups have not been included in the majority of clinical trials of physical activity thus far. Physical activity interventions need to be replicated in underserved populations before we can assume their results are generalizable. Though characteristics of existing studies show promise, future research on physical activity in underserved populations should assess the effect of practice-based systems on reducing barriers and promoting physical activity, the dose-response effect of clinician counseling on physical activity outcomes, and the effect of the physician-patient relationship and continuity of care on physical activity outcomes.

Funding

This study was supported by grant 1R25CA102618 from the National Cancer Institute.

Correspondence
Jennifer K. Carroll, MD, MPH, University of Rochester School of Medicine, Family Medicine Research Programs, 1381 South Avenue, Rochester, NY 14620; jennifer_carroll@urmc.rochester.edu

Fewer than half of all Americans get sufficient physical activity, defined as 30 minutes or more per day, at least 5 times per week.¹ The need to increase physical activity applies particularly to underserved populations: they are even less likely to get enough physical activity, and are thus even more likely to suffer greater burden of disease.^2,3

The purpose of this systematic review was to assess clinical trials of clinician-initiated counseling interventions for promoting physical activity in under-served populations. We define under-served populations as individuals from minority ethnic backgrounds (such as African Americans, Hispanics, and Asian Americans), or vulnerable populations such as people with low educational attainment, low income, lack of insurance, or those residing in rural communities.

Primary care interventions are linked to a change in habits

Primary care physicians can have a significant impact on their patients’ physical activity. Individuals with a regular primary care physician are more likely to report attempts to change their physical activity habits.⁴ However, underserved populations are more likely to have inconsistent access to medical care, which may contribute to their greater risk of conditions linked to inadequate physical activity, such as diabetes, hypertension, and obesity.

Only about 25% of patients in primary care settings report receiving any counseling on physical activity.⁵ Those who are middle-aged or have a baccalaureate degree or higher are more likely to report such advice; African Americans and foreign-born immigrants are less likely to report it.

A study by Taira et al⁶ examined the relationship between patient income and discussion of health risk behaviors. Low-income patients were more likely to be obese and smoke than high-income patients; however, physicians were less likely to discuss diet and exercise with low-income patients. Among all the patients with whom some discussion occurred in this study, low-income patients were much more likely to attempt to change behavior based on physician advice than were high-income patients.

Clinical trials within^7,8 and outside the US^9-11 support the potential value of physical activity counseling in primary care. In these studies, as little as 3 to 5 minutes of patient-clinician communication about physical activity was linked to short-term improvement in patients’ exercise habits. As few as 2 or 3 office visits over 6 months were associated with increases in patients’ physical activity levels up to 1 year later. Other features that contributed to their success included having a brief (<3 minutes) counseling component for clinicians, supplementing the counseling with a written exercise prescription, having follow-up contact, and tailoring the counseling to patients’ needs and concerns.

These results are promising for primary care clinicians, whose longitudinal relationships with their patients afford them repeated opportunities to intervene to promote physical activity.

Few studies have focused on the underserved

A review by Taylor et al² of physical activity interventions in low-income, ethnic minority, or disabled populations identified 14 community-based studies, mostly with quasi-experimental “pre/post” study designs. Ten studies included ethnic minorities, but physical activity was documented in just 2 studies at baseline, and these 2 studies did not include any postintervention follow-up. None of the 10 interventions was conducted in a primary care setting.

Another recent review¹² found that studies that were ethnically inclusive placed greater emphasis on involving communities and building coalitions right from study inception, and they tailored messages (and messengers) that were culturally specific. Several of these studies showed better outcomes among ethnic minority participants than the white participants they sampled.

Taken together, previous reviews have examined the effectiveness of primary care interventions for the general population,^13,14 as well as community-based programs for underserved populations.² However, little information exists about effective physical activity counseling strategies for underserved groups in primary care.

Methods

Looking for studies in underserved populations

We conducted a systematic review of the literature involving clinical trials in the US, looking for trials where counseling interventions are initiated by primary care clinicians, and that assessed behavioral change related to physical activity.

Inclusion criteria

TABLE 1 shows the inclusion criteria and search terms for the literature review. We searched Ovid, Medline, CINAHL, PsycINFO, PubMed, Cochrane, and HealthSTAR for studies published between 1966 and 2005. We also searched bibliographies of retrieved articles, and contacted experts in the field in an effort to obtain other relevant data.

The principal investigator (JKC) reviewed titles and abstracts of all potentially relevant articles to determine whether they met eligibility criteria. Studies that met the criteria were retrieved and abstracted.

Using these predefined criteria, data were extracted from each eligible article. Studies were also rated according to the Strength of Recommendation Taxonomy (SORT), because of its emphasis on patient-oriented outcomes and the quality, quantity, and consistency of evidence.¹⁵

TABLE 1
Inclusion criteria and search terms

Results

6 of 8 studies report increases in physical activity

We reviewed a total of 253 titles and abstracts. Eight studies^16-23 met our inclusion criteria. We were not able to locate any clinical trials that both 1) examined the effect of primary care clinician counseling on physical activity outcomes, and 2) had a study population focused on an underserved group. TABLE 2 (available at www.jfponline.com) shows the characteristics of these 8 studies.

Although we sought trials that defined “primary care clinician” as a professional—such as MD, nurse practitioner (NP), or physician assistant (PA)—who provides longitudinal primary health care, several of these studies considered dieticians, exercise physiologists, or health care workers as primary care clinicians.

Only 1 study²⁰ examined physical activity counseling with an intervention that incorporated a follow-up visit by the primary care clinician, and looked at the long-term effect on physical activity as an outcome. Thus, the degree to which the clinician’s counseling influenced the physical activity outcome in these studies is unclear.

Identifying underserved groups

Information on race or ethnicity (which tended to be reported as a single variable), level of education, and income of participants was reported in the demographic data of all studies’ results, but relationships between these variables and physical activity outcomes were not consistently reported. One study²³ stratified participants by race/ethnicity and health center; 2 studies^16,21 reported analyses and findings for participants according to ethnicity, income, and educational level, as that was their focus.

Overall, however, it is not clear to what extent the interventions succeeded for various underserved groups, even if they were included as participants.

Study designs and the nature of exercise interventions

Seven^16,18-23 of these studies (88%) were randomized controlled trials; the unit of randomization and control group varied. Trials were conducted at 1 or multiple (up to 11) primary care sites. Use of more than 1 method to recruit participants—such as mailings, use of office staff to promote/recruit, advertising, and community announcements—tended to be most effective.

Intervention types included phone and mail interventions,^17-23 computer-based interventions,^18,19,21 visits from a community health worker,^22,23 group classes,^16,22,23 directly supervised physical activity sessions,^16,22,23 clinician counseling,^16-23 and prescription protocols (eg, written, guided action plans).^17-23 Those delivering the intervention varied, and included primary care physicians,^17-23 nurse practitioners or physician assistants,^17-19,23,23 nutritionists,¹⁶ exercise physiologists,¹⁶ community health educators,^20,22,23 and other study personnel.^19,21 Specific elements of interventions that were likely to contribute to patients’ success included addressing financial or environmental/safety issues for exercise,¹⁶ use of trained office staff to provide exercise counseling,^18-20,23 and offering flexibility in choice by tailoring the goals and plans to the patients’ needs and interests.^17-23

The “dose” of clinician counseling varied from very brief (1 to 3 minutes of direct contact on 1 occasion) to more extended (>5 minutes of direct counseling over repeated intervals). Duration of follow-up for the 8 studies ranged from 4 months to 2 years.

Several studies designed their interventions to make the clinician counseling brief,^17-20,23 in order to enhance feasibility for busy primary care settings. Three studies^16,21,22 described strategies they used for tailoring the intervention to a specific culture, or for addressing issues of literacy for the written materials. Two studies^16,22 reported that their study staffs were ethnically or culturally representative of the targeted population.

The difficulty of maintaining adherence to physical activity

Three studies^18,19,21 reported having difficulty with attrition among their minority participants; they did not, however, include information specific to minorities in their physical activity outcomes. Studies with highest retention rates (>80%) tended to specifically address barriers to participation, including cultural issues, or they used a “lead-in” period.^16,20,21,23

The studies with the best adherence and retention among black and Hispanic participants, and those participants with low educational attainment,^16,21 used baseline qualitative data regarding management of health behaviors when they designed their interventions. For example, 1 study¹⁶ mentioned cultural adaptations derived from prior qualitative work—such as using program materials that extensively depicted African American individuals, families, and community settings—and using language in the intervention reflecting social values and situations relevant to African Americans.

How exercise data were reported

Six of the 8 (75%) studies^{16,17,19,20,22,23} reported some improvement in short-term physical activity outcomes (TABLE 2, available at www.jfponline.com); however, there was considerable heterogeneity in how these studies measured physical activity outcomes. All 8 incorporated a self-report measure of physical activity, such as the Patient-centered Assessment and Counseling for Exercise (PACE),^17-19 Paffenbarger Physical Activity Questionnaire (PPAQ),¹⁷ 7-day Physical Activity Recall (PAR),^17,20,21,23 and other self-report recall measures to assess physical activity. (A RESOURCE LIST of these instruments is available at www.jfponline.com.) Two studies also measured “states of change,”^17,20 but these states were not consistently defined.

Three studies^17,20,23 included objective measures of physical activity, such as accelerometers; in these studies, there was not substantial variance in physical activity outcomes between the objective and subjective measures.

Discussion

More study needed in the underserved

This review reflects in part the difficult task of designing and implementing realistic interventions for the underserved in primary care. However, interventions must be replicated in these populations before we can necessarily assume that findings from other trials are generalizable, due to issues of access, financial resources, health literacy, beliefs, cultural differences, self-efficacy, and other logistic barriers to traditional care that disproportionately affect underserved groups.

Integrate known personal, social, and environmental factors

Several studies^24-26 have explored the social, demographic, and environmental factors associated with physical activity in minority populations. These studies shed light on the reasons why clinical trials that focus on white, affluent, educated populations might not be generalizable to underserved groups.

To be maximally effective, any interventions for promoting physical activity in the underserved need to find ways to address any cultural or financial barriers, and incorporate factors associated with success. For example, among African American and Hispanic women, having lower “social role strain,” higher attendance at religious services, and a greater feeling that one’s neighborhood was safe were all associated with increased likelihood of exercise.^24-26 Such studies suggest that differences in beliefs, resources, self-efficacy, prior experience, and competing life demands can all contribute to promoting physical activity in some underserved groups. Practically, such findings encourage clinicians to work with patients to help them identify sources of social support and positive influences on their health, and help them articulate internal strengths and personal attributes to succeed in behavioral change.

Despite the variations in training or means of communication in the studies we identified, 2 studies used interventions that were successful at explicitly anticipated and addressed barriers to physical activity.^16,21 These 2 studies also had interventionists who represented the communities of interest, and they used cultural adaptations to promote exercise where appropriate. Thus, limited data suggest that some primary care–based programs improve physical activity in underserved patients, but the effects of communication from the primary care clinician on physical activity is lacking, consistent with other work in the field.^12,27

Promising strategies include office prompts, brief counseling

Primary care clinicians face many time pressures, fiscal constraints, administrative burdens, and competing priorities; these make addressing health promotion behaviors such as physical activity quite difficult. These issues are magnified for clinicians practicing in medically underserved areas. Despite these many challenges, promising opportunities do exist.

On a systems level, practice-based systems to manage chronic diseases have been successfully developed and implemented in the primary care setting; such systems can be tested to promote physical activity, as well. These practice-based approaches include patient registry data, office prompts, and other electronic systems to promote clinician counseling. For example, studies in this review using computer-based programs in primary care offices were feasible and effective.^18,19,21

Bodenheimer²⁸ has argued for a redesign of primary care systems to more effectively address chronic conditions rather than acute care needs. Several health care systems have successfully implemented the pillars of such a redesign imperative, and they have shown convincingly the promise of addressing competing priorities, physician competence and confidence, motivation, and durability in improving patient self-management.²⁸

At the level of the clinician-patient relationship, data suggest that patient physical activity can be increased (at least in the short term) by counseling that:

• is brief (5 minutes or less)^17-20,23
• is focused/goal-oriented^17-23
• is molded to the patient’s specific health needs^17-23
• is delivered over multiple contacts (whether it be office visits, telephone, or group sessions)^17-23
• contains a written plan to achieve goals.^17-23

We do not know what “dose-response” relationship exists for primary care clinician communication with patients over the long term, and what effect repeated counseling would have on long-term sustainability of physical activity levels. This is even less clear for underserved groups. It is also unknown to what extent collaborative links with community programs might increase physical activity when added to primary care–based counseling. Future research should evaluate the optimal “dose-response” to the interventions, the effect of repeated visits and continuity of care, and the effect of community-based referrals for physical activity programs for underserved populations in primary care.

Limitations of this review

Because our inclusion criteria were strict, we omitted potentially meaningful studies that were less directly relevant to our aims. For example, there has been substantial creative community-based work with underserved populations in the US to promote physical activity, and many innovations have been designed by researchers outside the US. Results from these programs and trials should be incorporated into primary care settings working with underserved populations.

Another limitation is that our definition of “underserved” is not the only possible definition. The most marginalized underserved groups with the least access to the health care system (such as the uninsured or homeless) were more likely to be omitted from our results, because we wanted to examine physical activity programs among patients in primary care settings.

Finally, this review did not address the need to understand the connection between sustained improvements in physical activity and patient-oriented health outcomes for underserved populations.

Conclusion

Information on exercise counseling interventions in primary care for the underserved is limited: these groups have not been included in the majority of clinical trials of physical activity thus far. Physical activity interventions need to be replicated in underserved populations before we can assume their results are generalizable. Though characteristics of existing studies show promise, future research on physical activity in underserved populations should assess the effect of practice-based systems on reducing barriers and promoting physical activity, the dose-response effect of clinician counseling on physical activity outcomes, and the effect of the physician-patient relationship and continuity of care on physical activity outcomes.

Funding

This study was supported by grant 1R25CA102618 from the National Cancer Institute.

Correspondence
Jennifer K. Carroll, MD, MPH, University of Rochester School of Medicine, Family Medicine Research Programs, 1381 South Avenue, Rochester, NY 14620; jennifer_carroll@urmc.rochester.edu

Fewer than half of all Americans get sufficient physical activity, defined as 30 minutes or more per day, at least 5 times per week.¹ The need to increase physical activity applies particularly to underserved populations: they are even less likely to get enough physical activity, and are thus even more likely to suffer greater burden of disease.^2,3

The purpose of this systematic review was to assess clinical trials of clinician-initiated counseling interventions for promoting physical activity in under-served populations. We define under-served populations as individuals from minority ethnic backgrounds (such as African Americans, Hispanics, and Asian Americans), or vulnerable populations such as people with low educational attainment, low income, lack of insurance, or those residing in rural communities.

Primary care interventions are linked to a change in habits

Primary care physicians can have a significant impact on their patients’ physical activity. Individuals with a regular primary care physician are more likely to report attempts to change their physical activity habits.⁴ However, underserved populations are more likely to have inconsistent access to medical care, which may contribute to their greater risk of conditions linked to inadequate physical activity, such as diabetes, hypertension, and obesity.

Only about 25% of patients in primary care settings report receiving any counseling on physical activity.⁵ Those who are middle-aged or have a baccalaureate degree or higher are more likely to report such advice; African Americans and foreign-born immigrants are less likely to report it.

A study by Taira et al⁶ examined the relationship between patient income and discussion of health risk behaviors. Low-income patients were more likely to be obese and smoke than high-income patients; however, physicians were less likely to discuss diet and exercise with low-income patients. Among all the patients with whom some discussion occurred in this study, low-income patients were much more likely to attempt to change behavior based on physician advice than were high-income patients.

Clinical trials within^7,8 and outside the US^9-11 support the potential value of physical activity counseling in primary care. In these studies, as little as 3 to 5 minutes of patient-clinician communication about physical activity was linked to short-term improvement in patients’ exercise habits. As few as 2 or 3 office visits over 6 months were associated with increases in patients’ physical activity levels up to 1 year later. Other features that contributed to their success included having a brief (<3 minutes) counseling component for clinicians, supplementing the counseling with a written exercise prescription, having follow-up contact, and tailoring the counseling to patients’ needs and concerns.

These results are promising for primary care clinicians, whose longitudinal relationships with their patients afford them repeated opportunities to intervene to promote physical activity.

Few studies have focused on the underserved

A review by Taylor et al² of physical activity interventions in low-income, ethnic minority, or disabled populations identified 14 community-based studies, mostly with quasi-experimental “pre/post” study designs. Ten studies included ethnic minorities, but physical activity was documented in just 2 studies at baseline, and these 2 studies did not include any postintervention follow-up. None of the 10 interventions was conducted in a primary care setting.

Another recent review¹² found that studies that were ethnically inclusive placed greater emphasis on involving communities and building coalitions right from study inception, and they tailored messages (and messengers) that were culturally specific. Several of these studies showed better outcomes among ethnic minority participants than the white participants they sampled.

Taken together, previous reviews have examined the effectiveness of primary care interventions for the general population,^13,14 as well as community-based programs for underserved populations.² However, little information exists about effective physical activity counseling strategies for underserved groups in primary care.

Methods

Looking for studies in underserved populations

We conducted a systematic review of the literature involving clinical trials in the US, looking for trials where counseling interventions are initiated by primary care clinicians, and that assessed behavioral change related to physical activity.

Inclusion criteria

TABLE 1 shows the inclusion criteria and search terms for the literature review. We searched Ovid, Medline, CINAHL, PsycINFO, PubMed, Cochrane, and HealthSTAR for studies published between 1966 and 2005. We also searched bibliographies of retrieved articles, and contacted experts in the field in an effort to obtain other relevant data.

The principal investigator (JKC) reviewed titles and abstracts of all potentially relevant articles to determine whether they met eligibility criteria. Studies that met the criteria were retrieved and abstracted.

Using these predefined criteria, data were extracted from each eligible article. Studies were also rated according to the Strength of Recommendation Taxonomy (SORT), because of its emphasis on patient-oriented outcomes and the quality, quantity, and consistency of evidence.¹⁵

TABLE 1
Inclusion criteria and search terms

Results

6 of 8 studies report increases in physical activity

We reviewed a total of 253 titles and abstracts. Eight studies^16-23 met our inclusion criteria. We were not able to locate any clinical trials that both 1) examined the effect of primary care clinician counseling on physical activity outcomes, and 2) had a study population focused on an underserved group. TABLE 2 (available at www.jfponline.com) shows the characteristics of these 8 studies.

Although we sought trials that defined “primary care clinician” as a professional—such as MD, nurse practitioner (NP), or physician assistant (PA)—who provides longitudinal primary health care, several of these studies considered dieticians, exercise physiologists, or health care workers as primary care clinicians.

Only 1 study²⁰ examined physical activity counseling with an intervention that incorporated a follow-up visit by the primary care clinician, and looked at the long-term effect on physical activity as an outcome. Thus, the degree to which the clinician’s counseling influenced the physical activity outcome in these studies is unclear.

Identifying underserved groups

Information on race or ethnicity (which tended to be reported as a single variable), level of education, and income of participants was reported in the demographic data of all studies’ results, but relationships between these variables and physical activity outcomes were not consistently reported. One study²³ stratified participants by race/ethnicity and health center; 2 studies^16,21 reported analyses and findings for participants according to ethnicity, income, and educational level, as that was their focus.

Overall, however, it is not clear to what extent the interventions succeeded for various underserved groups, even if they were included as participants.

Study designs and the nature of exercise interventions

Seven^16,18-23 of these studies (88%) were randomized controlled trials; the unit of randomization and control group varied. Trials were conducted at 1 or multiple (up to 11) primary care sites. Use of more than 1 method to recruit participants—such as mailings, use of office staff to promote/recruit, advertising, and community announcements—tended to be most effective.

Intervention types included phone and mail interventions,^17-23 computer-based interventions,^18,19,21 visits from a community health worker,^22,23 group classes,^16,22,23 directly supervised physical activity sessions,^16,22,23 clinician counseling,^16-23 and prescription protocols (eg, written, guided action plans).^17-23 Those delivering the intervention varied, and included primary care physicians,^17-23 nurse practitioners or physician assistants,^17-19,23,23 nutritionists,¹⁶ exercise physiologists,¹⁶ community health educators,^20,22,23 and other study personnel.^19,21 Specific elements of interventions that were likely to contribute to patients’ success included addressing financial or environmental/safety issues for exercise,¹⁶ use of trained office staff to provide exercise counseling,^18-20,23 and offering flexibility in choice by tailoring the goals and plans to the patients’ needs and interests.^17-23

The “dose” of clinician counseling varied from very brief (1 to 3 minutes of direct contact on 1 occasion) to more extended (>5 minutes of direct counseling over repeated intervals). Duration of follow-up for the 8 studies ranged from 4 months to 2 years.

Several studies designed their interventions to make the clinician counseling brief,^17-20,23 in order to enhance feasibility for busy primary care settings. Three studies^16,21,22 described strategies they used for tailoring the intervention to a specific culture, or for addressing issues of literacy for the written materials. Two studies^16,22 reported that their study staffs were ethnically or culturally representative of the targeted population.

The difficulty of maintaining adherence to physical activity

Three studies^18,19,21 reported having difficulty with attrition among their minority participants; they did not, however, include information specific to minorities in their physical activity outcomes. Studies with highest retention rates (>80%) tended to specifically address barriers to participation, including cultural issues, or they used a “lead-in” period.^16,20,21,23

The studies with the best adherence and retention among black and Hispanic participants, and those participants with low educational attainment,^16,21 used baseline qualitative data regarding management of health behaviors when they designed their interventions. For example, 1 study¹⁶ mentioned cultural adaptations derived from prior qualitative work—such as using program materials that extensively depicted African American individuals, families, and community settings—and using language in the intervention reflecting social values and situations relevant to African Americans.

How exercise data were reported

Six of the 8 (75%) studies^{16,17,19,20,22,23} reported some improvement in short-term physical activity outcomes (TABLE 2, available at www.jfponline.com); however, there was considerable heterogeneity in how these studies measured physical activity outcomes. All 8 incorporated a self-report measure of physical activity, such as the Patient-centered Assessment and Counseling for Exercise (PACE),^17-19 Paffenbarger Physical Activity Questionnaire (PPAQ),¹⁷ 7-day Physical Activity Recall (PAR),^17,20,21,23 and other self-report recall measures to assess physical activity. (A RESOURCE LIST of these instruments is available at www.jfponline.com.) Two studies also measured “states of change,”^17,20 but these states were not consistently defined.

Three studies^17,20,23 included objective measures of physical activity, such as accelerometers; in these studies, there was not substantial variance in physical activity outcomes between the objective and subjective measures.

Discussion

More study needed in the underserved

This review reflects in part the difficult task of designing and implementing realistic interventions for the underserved in primary care. However, interventions must be replicated in these populations before we can necessarily assume that findings from other trials are generalizable, due to issues of access, financial resources, health literacy, beliefs, cultural differences, self-efficacy, and other logistic barriers to traditional care that disproportionately affect underserved groups.

Integrate known personal, social, and environmental factors

Several studies^24-26 have explored the social, demographic, and environmental factors associated with physical activity in minority populations. These studies shed light on the reasons why clinical trials that focus on white, affluent, educated populations might not be generalizable to underserved groups.

To be maximally effective, any interventions for promoting physical activity in the underserved need to find ways to address any cultural or financial barriers, and incorporate factors associated with success. For example, among African American and Hispanic women, having lower “social role strain,” higher attendance at religious services, and a greater feeling that one’s neighborhood was safe were all associated with increased likelihood of exercise.^24-26 Such studies suggest that differences in beliefs, resources, self-efficacy, prior experience, and competing life demands can all contribute to promoting physical activity in some underserved groups. Practically, such findings encourage clinicians to work with patients to help them identify sources of social support and positive influences on their health, and help them articulate internal strengths and personal attributes to succeed in behavioral change.

Despite the variations in training or means of communication in the studies we identified, 2 studies used interventions that were successful at explicitly anticipated and addressed barriers to physical activity.^16,21 These 2 studies also had interventionists who represented the communities of interest, and they used cultural adaptations to promote exercise where appropriate. Thus, limited data suggest that some primary care–based programs improve physical activity in underserved patients, but the effects of communication from the primary care clinician on physical activity is lacking, consistent with other work in the field.^12,27

Promising strategies include office prompts, brief counseling

Primary care clinicians face many time pressures, fiscal constraints, administrative burdens, and competing priorities; these make addressing health promotion behaviors such as physical activity quite difficult. These issues are magnified for clinicians practicing in medically underserved areas. Despite these many challenges, promising opportunities do exist.

On a systems level, practice-based systems to manage chronic diseases have been successfully developed and implemented in the primary care setting; such systems can be tested to promote physical activity, as well. These practice-based approaches include patient registry data, office prompts, and other electronic systems to promote clinician counseling. For example, studies in this review using computer-based programs in primary care offices were feasible and effective.^18,19,21

Bodenheimer²⁸ has argued for a redesign of primary care systems to more effectively address chronic conditions rather than acute care needs. Several health care systems have successfully implemented the pillars of such a redesign imperative, and they have shown convincingly the promise of addressing competing priorities, physician competence and confidence, motivation, and durability in improving patient self-management.²⁸

At the level of the clinician-patient relationship, data suggest that patient physical activity can be increased (at least in the short term) by counseling that:

• is brief (5 minutes or less)^17-20,23
• is focused/goal-oriented^17-23
• is molded to the patient’s specific health needs^17-23
• is delivered over multiple contacts (whether it be office visits, telephone, or group sessions)^17-23
• contains a written plan to achieve goals.^17-23

We do not know what “dose-response” relationship exists for primary care clinician communication with patients over the long term, and what effect repeated counseling would have on long-term sustainability of physical activity levels. This is even less clear for underserved groups. It is also unknown to what extent collaborative links with community programs might increase physical activity when added to primary care–based counseling. Future research should evaluate the optimal “dose-response” to the interventions, the effect of repeated visits and continuity of care, and the effect of community-based referrals for physical activity programs for underserved populations in primary care.

Limitations of this review

Because our inclusion criteria were strict, we omitted potentially meaningful studies that were less directly relevant to our aims. For example, there has been substantial creative community-based work with underserved populations in the US to promote physical activity, and many innovations have been designed by researchers outside the US. Results from these programs and trials should be incorporated into primary care settings working with underserved populations.

Another limitation is that our definition of “underserved” is not the only possible definition. The most marginalized underserved groups with the least access to the health care system (such as the uninsured or homeless) were more likely to be omitted from our results, because we wanted to examine physical activity programs among patients in primary care settings.

Finally, this review did not address the need to understand the connection between sustained improvements in physical activity and patient-oriented health outcomes for underserved populations.

Conclusion

Information on exercise counseling interventions in primary care for the underserved is limited: these groups have not been included in the majority of clinical trials of physical activity thus far. Physical activity interventions need to be replicated in underserved populations before we can assume their results are generalizable. Though characteristics of existing studies show promise, future research on physical activity in underserved populations should assess the effect of practice-based systems on reducing barriers and promoting physical activity, the dose-response effect of clinician counseling on physical activity outcomes, and the effect of the physician-patient relationship and continuity of care on physical activity outcomes.

Funding

This study was supported by grant 1R25CA102618 from the National Cancer Institute.

Correspondence
Jennifer K. Carroll, MD, MPH, University of Rochester School of Medicine, Family Medicine Research Programs, 1381 South Avenue, Rochester, NY 14620; jennifer_carroll@urmc.rochester.edu