DURING THE past decade research has shown that most adolescents engage in at least some risk-taking behaviors,1-4 that they begin to do so at younger ages than in previous generations,3, 5 and that the prevalence of risk taking increases with age.4-6 The health-risk behaviors of adolescents include unprotected sex, use of alcohol, tobacco, and drugs, and intentional and unintentional injury.3-7 During the 1980s, primary care providers began to develop in-depth screening and assessment tools that would give primary care providers a thorough understanding of a specific problem such as alcohol use.8-18 The drawback of this approach was that the time required for in-depth screening and assessment of a single, categorical problem could preclude screening for other problems that the adolescent might have.19
In 1992, with funding from the Division of Adolescent and School Health of the Centers for Disease Control and Prevention, the American Medical Association (AMA) published the Guidelines for Adolescent Preventive Services (GAPS).20 The GAPS recommendations described the types of services that should be included in comprehensive clinical preventive services visits for adolescent patients between 11 and 21 years of age. Similar recommendations for adolescents appeared in Bright Futures: Guidelines for Health Supervision of Infants, Children, and Adolescents, sponsored by the Maternal and Child Health Bureau and published in 1994.21 Both publications recommended comprehensive clinical preventive services, which included screening, anticipatory guidance, and targeted counseling associated with primary and secondary prevention, as well as immunizations. Both sets of recommendations also addressed confidentiality, the need for annual visits, the need for individualized care, and health guidance for parents. Taken as a whole, the recommendations provided a framework for the organization and content of clinical preventive health services.
After publishing the GAPS recommendations, the AMA developed materials to help providers (ie, nurse practitioners or physicians) incorporate comprehensive health services for adolescent patients into clinical practice. The materials included a screening questionnaire for adolescent patients consisting of a brief medical history, a list of biomedical health concerns that the adolescent patient might have, and a list of health-risk behaviors that, taken together, would form a health-risk behavior profile. A screening questionnaire for parents also was developed. It comprised a family history of select medical conditions as well as a list of parents' concerns. Manuals for providers were developed that included suggestions for how to incorporate clinical preventive services for adolescent patients into routine office procedures.22-23
In 1994, the Robert Wood Johnson Foundation provided the AMA with funding to assess the feasibility of implementing comprehensive clinical preventive services using the GAPS model and to identify factors that could facilitate or impede the implementation of these services. In this article one component of the feasibility study is described: providers' responsiveness to health-risk behaviors reported by adolescent patients on the GAPS screening questionnaire. Providers' responsiveness in the present analysis is defined as a response by primary care providers to reported adolescent health-risk behaviors, as indicated on the debriefing interview with patients or by provider notes in the medical record.
Ideally, in a single clinical encounter, providers would respond to each health-risk behavior reported by a patient. However, this is not always possible. The reported condition may be severe or complex (eg, pregnancy or suicide ideation and/or attempt), the adolescent patient may present with too many health-risk behaviors to be addressed in a reasonable amount of time, or the provider may be ill equipped to deal with the reported condition. For these reasons, it is likely that many adolescent patients will have to schedule a subsequent visit for providers to address health concerns and health-risk behaviors, and others will be referred to another provider for treatment. The emphasis on prevention during the visit may be dropped altogether among patients who present with urgent problems (eg, suspected pregnancy or human immunodeficiency virus).
One goal of this study was to learn how often adolescent patients have most, some, or few of their reported health-risk behaviors addressed by their providers in a single visit. A second goal was to gauge the degree to which other factors could affect a provider's responsiveness. For example, it was expected that a provider's responsiveness would be lower for an adolescent patient who reported a relatively severe or a large number of health-risk behaviors.
A greater number of reported biomedical health concerns was also expected to lower a provider's responsiveness. This decrease is because biomedical health concerns, while legitimate, could prevent providers from addressing the psychosocial health-risk behaviors that were intended to be the focus of the preventive services visit.
A positive relationship was expected between a provider's responsiveness and visit duration. In general, providers who spent more time with patients were expected to have higher responsiveness scores, with all else equal.
Whether the adolescent patient was new to the practice (ie, had never seen the provider before) or an established patient was also taken into consideration. At issue was whether providers would find it more or less difficult to discuss highly sensitive topics with patients with whom they had a long-standing relationship compared with patients they were meeting for the first time.
The effects of the sex and professional role (physician vs nurse practitioner) of the provider were also included in the analysis. Some studies24-25 have shown that female providers are more responsive than male providers to health-risk behaviors reported by patients. Because nurse practitioners may spend more time with patients, they might also be more responsive than physicians. The specialty of the physician was not included in the analysis because there were too few providers per specialty to make reliable estimates.
The age of the patient was examined to determine whether older adolescents (aged 15-18 years) would be more likely than younger adolescents (aged 11-14 years) to report a greater number of health-risk behaviors and more severe risk behaviors. The sex of the adolescent patient was also included to see whether there was a significant difference in the number and severity of reported risk behaviors, a pattern that could affect a provider's responsiveness.
PARTICIPANTS AND METHODS
PROTOCOL
Between January and August 1994, AMA staff recruited 7 clinical sites in the Chicago, Ill, area to assess the usability of forms and manuals developed to help primary care providers deliver a comprehensive clinical preventive services visit to adolescent patients and parents in their practice, using the GAPS model. The sites included 3 staff-model health maintenance organizations, 2 private practices, 1 community clinic, and 1 university-affiliated clinic. The sites served diverse groups that included whites, African Americans, Latinos, and Asian Americans, as well as poor through upper-middle-class families. Approval to conduct the study was obtained from a local university as well as the institutional review board at each participating site (except the 2 private practices). To defray expenses associated with needed immunizations and a potentially extended clinical visit, each site was given $100 per patient seen in the study. It is assumed that this payment helped sites overcome their reticence to participate in the study; there is no indication that providers increased visit duration as a result of the payment, particularly since it was given to the site and not the individual provider.
Prior to the study, AMA staff visited each site, described the goals of GAPS as a model of clinical preventive services, and explained what would be asked of participating sites, providers, adolescent patients, and parents. Sites were asked to incorporate visits based on GAPS into their schedule 2 days a week until each provider had seen 10 adolescent patients and their parents. (On other days, GAPS forms and protocols were not used to deliver preventive services.) Interested providers were invited to participate, but none were forced to participate. Providers were asked to give parents and adolescent patients health education, anticipatory guidance, and counseling, as needed, and to administer immunizations to adolescent patients, as appropriate. They were also asked to use GAPS forms and materials, including screening questionnaires.
Provider orientation for the study included a review of the GAPS recommendations, the screening questionnaire and other forms associated with the visit, and patient recruitment procedures. The orientation was not intended to teach providers how to incorporate GAPS into their clinical practice. In fact, an important goal of the study was to identify strategies for implementing preventive services for adolescent patients in clinical settings and incorporate this information into future training programs. American Medical Association staff contacted providers periodically during the study and once afterward to discuss problems encountered during implementation and solutions to address those problems. Providers were told that parents and adolescent patients would be debriefed about their experience with GAPS, but responsiveness was not mentioned as a topic of interest. During the orientation, each provider received a notebook containing health education brochures for parents and adolescent patients developed by various organizations, which addressed topics covered in the GAPS recommendations.
Adolescent patients between the ages of 11 and 21 years with nonacute conditions were eligible to participate in the study. During several weeks, patients and their parents were recruited for the study in 1 of 3 ways: (1) patients with existing health maintenance appointments were called by the receptionist and invited to participate in the study; (2) patients who called the provider to schedule a health maintenance appointment were invited to participate; and (3) patients who were due for an annual health maintenance visit were invited to participate when the receptionist called them to schedule their annual appointment. When contacting parents, the receptionist explained that the primary care provider was working with the AMA on a study of preventive care for adolescent patients. Parents and adolescent patients would be asked to complete a screening questionnaire, sign a consent form at the clinical site prior to the visit, and participate in a debriefing interview at the conclusion of the visit. The visit would include a physical examination, a clinical interview, and related laboratory tests, as needed, and each adolescent patient would receive 2 coupons for movie tickets to thank them for their participation. Parents and adolescent patients were each assured of confidentiality. After the clinical visit and before leaving the office, parents and adolescent patients met separately with a researcher from the AMA who used a semistructured interview for a confidential debriefing interview about topics discussed during the visit and how satisfied they were with the visit and the provider.
QUESTIONNAIRES
When adolescent patients and parents or guardians arrived at the clinical site, they were asked to sit in separate areas of the waiting room to complete their respective screening questionnaires and consent forms. The self-administered patient screening questionnaire included 5 sections: (1) a brief medical history; (2) a checklist of 24 biological health concerns (eg, acne or headaches); (3) a list of major changes in family life during the past year; (4) 3 open-ended self-perception questions; and (5) a health-risk behavior profile. The health-risk behavior profile included 11 topics: diet and weight, physical activity, puberty and sexual behavior, school performance, relationships with friends and family, injury prevention, tobacco, alcohol, and drug use, weapons and violence, and emotional health. Questions under each topic heading were designed to elicit reporting of specific health-risk behaviors and 1 question about whether the adolescent patient was concerned about or wanted more information about the topic. There were 52 questions with forced yes or no answers in the health-risk profile; 22 screening questions also included a "not sure" column.
The debriefing interview with the adolescent patient after the visit included questions about whether the adolescent was a new patient; specifics about the visit (eg, a review of confidentiality, physical examination, or immunizations); topics discussed with the provider; comfort with the topics discussed; satisfaction with the visit and visit duration; and candid reporting of alcohol, other drug, and/or tobacco use and sexual activity. Questions about physical, sexual, or emotional abuse or suicide were excluded because these topics were considered too sensitive, and "positive" responses to these questions required mandatory reporting.
KEY VARIABLES
Variables included in the analysis were the provider's responsiveness, the number and severity of reported health-risk behaviors, the number of competing health concerns, visit duration, sex of the provider and the patient, professional role of the provider, and whether the adolescent patient was new to the practice.
Responsiveness was calculated in 2 ways, at the topic level (to determine whether providers tended to respond to certain topics and avoid others) and at the item level (to determine overall responsiveness). Responsiveness at the topic level was calculated as follows: If an adolescent patient reported at least 1 health-risk behavior within a topic on the screening questionnaire (eg, safety) and reported during the debriefing interview after the visit that the topic was discussed, the provider was defined as responsive.
Responsiveness was also calculated at the item level to gauge the provider's responsiveness to specific health-risk behaviors reported by the adolescent patient and not just to the topic more generally. For example, using a bicycle helmet or seat belt were separate questions under the topic heading "safety." If the patient indicated risk on both questions, but the provider addressed only 1 of them, the provider was considered responsive at the topic level but only partially responsive at the item level.
Item-level responsiveness was determined by matching patient responses during the debriefing interview with the adolescent patient's responses to the health-risk behavior questions on the screening questionnaire. The provider was considered responsive to an item if 2 conditions were met. First, the adolescent patient indicated risk (or responded not sure) to the question on the screening questionnaire. Second, the adolescent patient reported during the debriefing interview that the provider addressed that issue during the visit. If the health-risk behavior was reported on the screening questionnaire but not discussed during the visit, the provider was considered unresponsive for that item. Health-risk behaviors that were not reported on the screening questionnaire but were identified as problems during the visit were excluded from consideration unless the provider changed the adolescent patient's response on the screening questionnaire.
Matching reported and discussed items for the highly sensitive issues of suicide, abuse, and some sexual behaviors was handled differently. The adolescent patient's responses on the screening questionnaire were matched using only provider notes in the patient record. Although highly sensitive behavior reported during a clinical interview may not be recorded in the patient's medical record and, therefore, would underestimate provider's responsiveness, this was deemed the most appropriate compromise.
Once matching was completed, the total number of health-risk items discussed during the visit per adolescent patient were summed and divided by the number of health-risk items reported by the adolescent patient on the screening questionnaire, creating a percentage score for responsiveness (0-100).
Health-risk behaviors were defined in terms of high, moderate, and low severity. Decisions about how to categorize behaviors were made by asking participating providers to identify which health-risk behaviors they considered to be an immediate threat, which were life-threatening rather than unhealthy, and which posed primarily long-term consequences.
High-severity behaviors included suicide ideation or attempt, eating disorder, and physical, sexual, or emotional abuse. Moderate-severity behaviors included drinking and driving, self or friends ever having had sexual intercourse, and drug use by self or friends. Low-severity behaviors included all other conditions, such as insufficient exercise, imbalanced daily diet, declining school grades, tension at home, no seat belt use, a physical fight in the last 3 months, smoking cigarettes or living with smokers, and often feeling sad or down.
The number of competing health concerns was calculated by summing the number of items checked on the list of 24 biomedical health concerns on the screening questionnaire (eg, headaches, frequent colds, and problems with various parts of the body).
Visit duration was defined as the number of minutes that the adolescent patient and the provider spent together in the examination room. An AMA staff member on site who recorded visit duration on a time sheet measured this through observation.
RESULTS
 |
Fifteen primary care providers in 7 sites participated in the study. They included 3 nurse practitioners, 7 pediatricians, 4 family practice physicians, and 1 internist.
All adolescent patients and parents who were invited to participate agreed to be a part of the study. However, of the initial 122 adolescent patients and parents who were invited to participate, 12 either canceled their appointments or failed to come to their scheduled appointment. A total of 102 patients received a comprehensive clinical preventive services visit, but complete screening questionnaires and debriefing interviews were obtained for 96 adolescent patients and parents. Six adolescent patients had to leave before the final interview could be completed, and AMA staff were unable to successfully contact the adolescent patients after at least 4 follow-up attempts. One adolescent patient was excluded from the sample because he reported 41 health-risk behaviors; including this patient would have distorted the analysis. Of the 95 adolescent patients included in the final sample, 61 (64%) were between 11 and 14 years of age and 34 (36%) were 15 to 18 years of age. Fifty-three adolescent patients (56%) were female and 42 (44%) were male.
The average visit, which included only the physical examination and clinical interview with the provider, was 29 minutes. Visits ranged from 10 to 72 minutes, plus an average of 12 minutes for the adolescent patient to complete the screening questionnaire prior to the clinical interview.
Of the 95 patients, 94 (99%) reported at least 1 health-risk behavior on the screening questionnaire. Of the 11 topics covered in the health-risk behavior profile section of the screening questionnaire, the most frequently reported were diet and/or weight (84 patients [88%]) and safety (81 patients [85%]). As shown in Table 1, more than half indicated a problem or concern about relationships with family and/or friends (57 patients [60%]), physical development and sexual behavior (53 patients [56%]), tobacco use (49 patients [52%]), and weapons and violence (49 patients [52%]). The less frequently reported (but not uncommon) health-risk behavior topics were school performance (41 patients [43%]), physical activity (39 patients [41%]), alcohol use (36 patients [38%]), drug-related issues (33 patients [35%]), and emotional issues, such as depression, abuse, and suicide (33 patients [35%]).
|
|
|
|
Table 1. Reported Health-Risk Topics and Providers' Responsiveness
|
|
|
The mean (SD) number of specific health-risk behaviors reported by patients was 10 (6.2) of a possible 52. The number of health-risk behaviors reported by patients ranged from 0 to 27. Females reported a mean (SD) of 11 (6.3) health-risk behaviors, and males reported 9 (6.0), but the difference was not statistically significant by a 2-tailed test (t93 = 1.35; P>.10). The mean (SD) number of health-risk behaviors reported by younger adolescent patients was 10 (6.5), the same as older adolescent patients (10 [5.8]). The mean (SD) number of health-risk behaviors reported by new patients was 10 (7.1) and for established patients was 9 (5.7), but the difference was not statistically significant by a 2-tailed test (t92 = - .75; P>.10).
Of the 95 patients, 18 (19%) reported 1 or more high-severity health-risk behaviors. An additional 30 adolescent patients (32%) reported 1 or more moderate-severity health-risk behaviors, and 47 (49%) reported no or low-severity health-risk behaviors.
Adolescent patients indicated a mean (SD) of 2.5 (2.7) competing health concerns on the screening questionnaire, although the number ranged from 0 to 13 of 24 possible health concerns. The most commonly reported health concerns were height and weight (32 patients [34%]), muscle or joint pain (20 patients [21%]), headaches (15 patients [16%]), and sleep (14 patients [15%]).
Overall, providers' responsiveness was relatively high. A mean (SD) of 7 (4.6) health-risk behaviors were discussed during the clinical interview. For 22 patients (23%), the provider discussed every health-risk behavior reported on the screening questionnaire. For an additional 58 patients (62%), half or more (but not all) of reported health-risk behaviors were discussed. For the remaining 14 patients (15%), fewer than half of reported health-risk behaviors were discussed.
The relationship between the reported health-risk behavior topics and the provider's responsiveness appears in Table 1. Although providers were highly responsive to the 3 health-risk behaviors most often reported by adolescent patients (diet and/or weight [79%], safety [86%], and family and/or friends [77%]), they were also highly responsive to reported problems with physical activity (87%) and school (80%). Providers' responsiveness was lowest, below 60%, for health-risk behaviors involving weapons or violence (59%) and depression or abuse (55%).
The correlations between providers' responsiveness and the predictor variables appear in Table 2. Providers' responsiveness was negatively correlated with the number of health-risk behaviors (r = - 0.24), severity of reported health-risk behaviors (r = - 0.29), and number of competing health concerns (r = - 0.24), and the relationships were statistically significant. Specifically, providers' responsiveness dropped as the number and severity of health-risk behaviors increased. A similar pattern existed for competing health concerns; providers' responsiveness declined as the number of competing health concerns reported by adolescent patients increased. Contrary to expectations, providers' responsiveness was not significantly correlated with visit duration (r = - 0.02). Univariate correlations also were not statistically significant for the professional role of the provider (r = - 0.17), the sex of the provider (r = - 0.13) and the adolescent patient (r = - 0.17), the age of the patient (r = 0.09), and whether the adolescent patient was new to the practice (r = -0.004).
|
|
|
|
Table 2. Intercorrelations Between Providers' Responsiveness and Predictor Variables
|
|
|
Other noteworthy relationships observed among the predictor variables in the correlation matrix included a statistically positive relationship between the number and severity of health-risk behaviors (r = 0.66) such that adolescent patients with a greater number of problems also tended to have relatively severe problems. Second, the number of reported health-risk behaviors was positively correlated with visit duration (r = 0.29). Providers spent significantly more time with higher-risk adolescent patients than those who presented with few problems. Third, females in the sample were significantly more likely than males to report competing health concerns. Because responsiveness drops as the number of reported competing health concerns increases, and females are more likely than males to report competing health concerns, it was important to test whether the effect of competing health concerns was suppressing a real relationship between sex of the adolescent patient and responsiveness. When differences in competing health concerns were controlled for, a significant correlation between sex of the adolescent patient and the provider's responsiveness was identified (partial r = - 0.26; P<.05).
Using multiple regression, provider responsiveness was regressed on the linear combination of the number and severity of the reported health-risk behavior, the number of competing health concerns, and sex of the adolescent patient. As shown in Table 3, the severity of the reported health-risk behavior, the number of competing health concerns, and the adolescent patient's sex were statistically significant. The equation containing these 3 variables accounted for 19% of the variance in the provider's responsiveness. The number of reported health-risk behaviors was not significant, an indication that severity is a more critical determinant of responsiveness than the number of health-risk behaviors reported by the adolescent patient.
|
|
|
|
Table 3. Multiple Regression Equation Predicting Providers' Responsiveness*
|
|
|
The finding that providers appear to be less responsive to adolescent males than females was unexpected. In an effort to clarify the relationship between the sex of the adolescent patient and responsiveness, a second regression analysis was conducted. For this analysis, a dichotomous variable of "troubledness" was created. Adolescent patients who reported a moderate or severe health-risk behavior and 2 or more competing health concerns were coded 1 and considered highly troubled. All other adolescent patients were coded 0 and considered less troubled. Responsiveness was then regressed on the sex of the adolescent patient, troubledness, and the interaction of the patient's sex and troubledness. As shown in Table 4, being highly troubled was a significant predictor of responsiveness, as was the interaction between troubledness and the adolescent patient's sex, but sex was no longer a significant predictor. This interaction effect was such that there was a more precipitous drop in responsiveness for highly troubled males than females (Figure 1).
|
|
|
|
Table 4. Multiple Regression Equation for Sex and "Troubledness"*
|
|
|
|
|
|
|
Differences in physicians' or nurse practitioners' responsiveness to troubled males and females aged 11 to 18 years.
|
|
|
COMMENT
This study has shown that almost all the adolescent patients reported health-risk behaviors on a screening questionnaire, including behaviors that are sensitive in nature. While the degree of underreporting of actual health-risk behaviors is unknown, the fact that 19 adolescent patients (20%) reported 1 or more high-severity health-risk behaviors and an additional 30 adolescent patients (32%) reported 1 or more moderate-severity health-risk behaviors demonstrates that a substantial proportion of the adolescent patients were candid with their health care providers.
This study also demonstrated that comprehensive clinical preventive services for adolescent patients could be delivered within 30 minutes, on average. It is worth noting that none of the providers had used the GAPS questionnaire or forms prior to the study. Thus, visit duration for all providers in this study may reflect a learning curve. It is likely that visit duration would decrease as providers became more familiar and comfortable with GAPS questionnaires and protocol, while maintaining or improving responsiveness.
Perhaps most important, this study has shown that providers try hard to discuss in a single encounter all health-risk behaviors reported by an adolescent patient. For 22 adolescent patients (23%), the provider addressed every reported health-risk behavior, and for an additional 59 adolescent patients (62%), more than half (but not all) health-risk behaviors were addressed. The health-risk behaviors least likely to be addressed by providers were those involving emotional issues (depression, suicide, and abuse) and weapons and/or violence. The reasons why providers were less likely to respond to these health-risk behaviors are not known. It could be that depression and/or suicide and abuse are difficult for some providers to address, especially those who have relatively few patients in this age group in their practice. It may also be that some providers do not record in the patient's file a report of abuse or contemplated or attempted suicide. This omission would also underestimate responsiveness. With respect to violence, it could be that providers were simply unsure about how to help adolescent patients with this problem and may not have had an alternative source for referral.
The findings suggest that the most critical determinants of responsiveness are severity of health-risk behavior, competing health concerns, and the sex of the adolescent patient. When confronted with severe problems or a relatively large number of competing health concerns reported by the adolescent patient, responsiveness declined. One of the more curious but least understood findings is the more dramatic decline in responsiveness among troubled adolescent males compared with troubled adolescent females. This differential decline merits further study to determine if it is replicated and, if so, why this is the case.
The results of this study, while promising in terms of demonstrating providers' responsiveness to adolescent patients during comprehensive clinical preventive services visits, has its limitations. First, the relatively small sample size of both adolescent patients and primary care providers limits the generalizability of the findings to the population as a whole. Second, the absence of a control group precludes demonstrating whether adolescent patient reporting of health-risk behaviors increased as a result of the GAPS screening questionnaire and whether providers who used GAPS materials were more responsive than providers who did not use GAPS materials. Finally, there is no way to gauge the degree to which adolescent patients underreported or overreported during the debriefing interview topics discussed during the visit. Underreporting would understate a provider's responsiveness and overreporting would exaggerate responsiveness. Similarly, the degree to which providers failed to document in the patient's record topics discussed during the visit would also understate responsiveness, and some (especially highly sensitive) topics may be more likely than others to have been left out of the patient's record. It would be useful for future research to address whether the screening questionnaires and other materials helped patients and providers increase the number and sensitivity of topics discussed during the clinical visit and demonstrate this through the use of experimental and control groups of adolescents.
The widespread adoption of preventive care for adolescents into routine practice hinges on several factors. Individual primary care providers and health system administrators need to be assured that these services can be delivered in a reasonable period of time, that adolescents will report health-risk behaviors, and that reported problems can be treated effectively by the individual provider or through referral. Without such assurance it is less likely that adolescents will receive the screening and health guidance they need.
There are several strategies that could be used to improve providers' responsiveness. For example, developing and disseminating algorithms that would help providers quickly identify and assess reported health-risk behaviors might enable some providers to respond to a wider range of problems. Training designed to improve skills in assessment and treatment might also enable some providers to tackle problems they currently avoid or simply fail to address routinely. Finally, access to appropriate referrals for specific problems might also improve providers' responsiveness. Some providers might be more willing to delve more deeply into a potential problem if they knew they had a way to meet the patient's needs. Alternative technologies, written health education material, or ancillary personnel might also be used during a comprehensive preventive services visit to enhance overall responsiveness to health-risk behaviors reported by adolescents.
