- Email: [email protected]
Physician attitudes toward confidential care for adolescents
ADOLESCENT MEDICINE
Physician attitudes toward confidential care for adolescents We examined factors that determine whether pediatricians will grant confidential care to adolescents. Through four vignettes, in which adolescents of different ages and maturity requested confidential care. we assessed the willingness o f physicians to give such care for four problems: request for contraceptive, diagnosis o f gonorrhea, intrauterine device found incidentally on x-ray study, and illicit drug use. A random sample o f 932 physicians responded to the vignettes. Overall, physicians agreed to give confidential care to adolescents in 75% o f the vignettes. They were more likely to give confidentiality in cases involving sexual activity. They supported confidential care for 87 % o f patients requesting contraceptives, but for only 54 % reporting illicit drug use. The proportion o f physicians supporting confidentiality increased with age and maturity o f the minors. Seventeen-year-old mature adolescents seeking contraceptives were given confidentiality by 97% o f physicians. Thus physician responses to vignettes indicated that they strongly support confidential care for adolescents as represented in the clinical vignettes. (J PEDIATR 106:517, 1985)
Joan Lovett, M.D., and Michael S. Wald, J.D. Stanford,
California
WHETHER ADOLESCENTS should be able to receive confidential health care is an issue with social, legal, and medical implications. Opponents believe that adolescents need adult guidance in health care decisions, especially those related to sexual activity or drug use. They argue that such guidance should come through family involvement, because the family is most familiar with the child's needs. In addition, some commentators oppose confidential care because they believe that parents have a right to know when their children, for whom they are responsible, receive health care.
In contrast, proponents of confidential health care regard it as essential for a trusting relationship between adolescent and physician. Furthermore, they assert that minors will not utilize certain medical services unless guaranteed confidential care. Although most proponents of confidential care encourage parental involvement when such involvement would benefit the adolescent, they would not make notification a prerequisite for treatment. The degree to which physicians may give confidential care is now being debated in courts and in the federal and IUD
From the Stanford Center Jor the Study o f Youth Development. At the time o f this research Dr. Lovett was a Robert Wood Johnson Clinical Scholar at Stanford Medical School. Mr. Wald is a Professor o f Law at Stanford University and a Member o f the Stanford Center for the Study o f Youth Development. The opinions expressed in this paper are not necessarily those of the Robert Wood Johnson Foundation or o f the Center for the Study o f Youth Development. Submitted for publication Oct. 12, 1983; accepted Aug. 10, 1984. Reprint requests: Joan Lovett, M,D., 25 Northampton, Berkeley, CA 94707.
Intrauterine device
state governments. Thirty states and the District of Columbia have laws requiring that physicians give confidential care when treating adolescents for certain conditions, usually those related to contraception, abortion, venereal disease, and drug abuse. However, other states require parental notificatibn before a minor can receive abortion or contraceptive services. ~ Recently, the U.S. Department of Health and Human Services initiated efforts to link federal aid to clinics providing contraceptive care with a requirement that parents be notified when minors seek contracep-
The Journal o f P E D I A T R I C S
517
5 18
Lovett and Wald
fives. At the time of writing, these regulations were being litigated. The physician whose adolescent patient asks that her or his parents not be involved in decisions about sexual activity or drug use is faced with an ethical dilemma, in addition to potential legal conflicts. The American Medical Association, in its Principles o f Medical Ethics', states: "A physician may not reveal the confidence entrusted to him in the course of medical attendance, or the deficiencies he may observe in the character of patients, unless he is required to do so by law or unless it becomes necessary in order to protect the welfare of the individual or of the community." This standard provides little guidance as to which actions would best "protect the welfare of the individual.''2 The issue is whether granting confidential care is detrimental for adolescents because of the absence of parental involvement or whether it is beneficial because it facilitates care for those who would not seek medical attention if parental involvement were required. Very little is known about physicians' attitudes toward granting confidentiality and the factors they consider important in making decisions regarding confidentiality. We present data on how practicing physicians decide whether to give confidential care to adolescents seeking treatment for a sex- or drug-related problem. METHODS Study design. A sample of i427 pediatricians was presented with a series of vignettes describing case histories of adolescents seeking confidential medical care. They were asked whether they would honor the patient's request for confidentiality. By systematically varying the age, maturity, and other characteristics of the patients in the vignettes, we were able to assess some of the variables that these physicians used in making their decisions. The study was designed to determine whether the physician's tendency to give confidential care varied with selected attributes of the female adolescent, the type of medical problem, or the physicians' characteristics. We chose to present standardized case vignettes to elicit physicians' responses, so that physician and patient variables could be disentangled. Case vignettes were used rather than direct observation or clinical simulations using actors so that a broader sampling of physicians and cases was possible2 W e acknowledge that there are limitations to the vignette approach: it assumes that the relevant variables are known and only the weights to be assigned to them are not, and that enough data needed for the decision are present. Because the judgments being studied are made on hypothetical cases, critics can raise the question of whether findings can be generalized to actual clinical
The Journal of Pediatrics March 1985
practice, the same objection raised with almost any experimental study. 3 Our decision to embed variables within the vignettes was based on a review of the literature as well as on a pilot study we conducted. Direct questioning about attitudes has been shown to be an unreliable predictor of actual behavior. Nisbett and Bellows 4 argue persuasively that when subjects are asked to introspect about their reasoning processes, their reports are based on prior implicit causal theories. 4 In a pilot study, we presented physicians with vignettes and asked them to list factors that would influence whether they would give confidential care to the adolescents described. Age, maturity, and duration of patient-physician relationship were listed most frequently, but we noted that physicians changed their responses as they proceeded through the four vignettes, because they were eager to base their decisions on consistent variables. Embedding the variables in the vignettes eliminated the physicians' need to appear consistent in their decisionmaking criteria. Based on the considerations described, we used vignettes to simulate clinical situations in which an adolescent requests that a physician not tell her parents that she is sexually active or that she uses illicit drugsl The vignettes included (I) a request for contraceptives, (2) a diagnosis of and request for treatment of gonorrhea, (3) an IUD found incidentally on x-ray examination, and (4) disclosure of illicit drug use. These types of medical problems were chosen because sexual behaviors and drug use are common reasons adolescents initiate confidential consultation with a physician and because legal controversy centers around these problems. The patients described in the vignettes were all females. Although adolescent males are involved in drug use and are treated for venereal disease, only females are affected by confidentiality rulings regarding use of the birth control pill or IUD. We chose to describe only female patients in the vignettes because inclusion of cases involving males would have introduced an additional variable in only two of the cases, making comparison across vignette types impossible. Differential treatment or attitudes toward adolescent males and females could be explored in a future study, holding constant problem type, age and maturity of the adolescent, and duration of physician-patient relationship. All physicians were sent a packet including all four problems. However, the information about the patients was varied in each vignette. The patient was described by age (14 or 17 years), maturity (mature or somewhat immature), and duration of patient-physician relationship (first patient contact or well known to physician). The variables of age and maturity were used as indicators of the
Volume 106 Number 3
Confidential care for adolescents
5 19
Table I. Percent of physicians who would give confidential care according to characteristics of adolescent and duration of physician-family relationship Adolescent characteristic
Age 17 years Age 14 years Mature Immature Well known Unknown Total
Vignette Drugs
Contraceptives
60 48 61 46 57 52 54
94 79 89 84 88 86 87
1UD
Gonorrhea
Total
91 70 88 73 81 80 80
87 68 84 72 77 79 78
83* 66 8 t* 69 76 74 75
*P < 0.0001, 17 yearsvs 14 yearS,mature vs immature. adolescent's capacity for responsible decision making and the likelihood that she would need advice on decisions. We looked at duration of relationship between the physician and the patient's family to see whether an established relationship with the parents would induce the physician to involve the parents. Every packet included two vignettes in which the adolescent described was 14 years of age, two vignettes in which she was described as 17 years, two vignettes in which she was mature, and so on. Anticipating that the order of the vignette within the packet might influence physicians' decisions, we varied the order. The vignettes were sent to all 589 physician members of the Society for AdolesCent Medicine (SAM) and to a random sample of 838 (of 16,436) Board Certified (nonSAM) members of the American Academy of Pediatricians (AAP) throughout the United States. The physicians were asked to respond to each vignette by either "Yes, I would agree to give confidential care" or "No, I would not agree to give confidential care." We also asked each physician to provide a brief description of the most 9 important factor influencing the decision, regardless of whether that factor was present in the vignette, in addition, we collected background information about the physician (age, gender, practice setting, parenthood status, experience with adolescents, membership in SAM or AAP, and state where practice is located). Seventy percent of the 1427 questionnaires were returned. Questionnaire packets were categorized as usable if all four responses were either yes or no. Nine hundred thirty-two respondents returned usable packets, and the remainder were not included in the analysis. The response rate differed between SAM members (81%) and non-SAM members (64%). Analysis of data. The study design allowed calculation of the overall percentages of confidential responses as well as comparisons of the variables embedded in the vignettes
(medical problem, mature vs immature adolescent, 14 vs 17 years of age, well known vs unknown) and comparisons of physician variables (e.g., old vs young, male vs female). We calculated the proportion of physicians replying yes or no in each of the 32 vignettes. We then calculated the effect of each of the variables on the physicians' responses. For example, we determined the effect of maturity by finding the average percent of yes responses to the mature a d o l e s c e n t (Pmat) minus the average p e r c e n t of yes responses to the immature adolescent (P~. . . . ) (i.e., Pmat - P~. . . . . . . ). w e defined the maturity effect as i/2(Pmat- Pimm~t). Thus, if Pm~t= 0.80 and Pi. . . . ~--" 0.70, t h e n emit - P~...... = 0.10 and l/2(Pm~t- P~..... ) = 0.05, the maturity effect. This effect impiies that maturity elicits 5% more yes responses than the mean. Analysis of variance was performed to determine the significance of the adolescent's age, duration of patient-physician relationship, request for contraceptive, IUD, gonorrhea, and drug problem on a proportion of yes responses. Analysis of variance also allowed comparisons among physician variables. RESULTS Overall, the physicians strongly supported confidentiality. The mean percentage of yes responses for all the vignettes was 75%. Only 4% responded that they would not give confidential care in any of the four vignettes, whereas 40% (371 respondents) would give confidential care in all four situations. However, the proportions of yes responses varied widely, with 97% of mature 17-year-old patients seeking contraceptives receiving responses in favor of confidentiality, whereas only 40% of immature 14-year-old drug users received confidentiality. The proportion of physicians supporting Confidentiality increased in cases involving older and more mature minors and in situations involving contraception rather than drug use. The 17-year-old
520
Lovett and Wald
The Journal of Pediatrics March 1985
Table II. Percent of physicians giving confidentialitj for various problems bY age and maturity
Medical )roblem Age of
Drugs
adolescent (yr)
Mature
17 14
69* 56
Contraceptives
I Immature
Mature
52 40
I Immature
97* 81
1UD MaCure
9l 77
96* 80
Gonorrhea
I Immature 86 60
Mature
Immature
90* 77
84 60
*P < 0.0001. Older, more mature adolescentswere more likelyto receiveconfidentialcare, regardless of type of medical problem.
Table IlL Percent of physicians giving confidential care by characteristics of physiciaus
Physician characteristic
I Drugs
Age (yr) (mean = 44 yr) >44 47 <44 62 Sex Male 54 Female 54 SAM member Yes 67 No 43 Children Yes 54 No 54 Clinical time with adolescents (%) (mean = 20%) >20 63 <20 47
Contraceptives
IUD
Gonorrhea
Total
83 89
77 84
70 85
69 80*
86 88
79 84
77 79
74 76
95 79
91 74
89 70
85* 67
85 89
80 80
77 79
74 72
90 83
85 77
83 74
80* 70
*P < 0.0001: Physicians<44 years, SAM member, >20% of clinicaltime with adolescents.
patient elicited substantially more yes responses than the 14-year-old, regardless of the medical problem (Table I). The age effect was 8.4% (P < 0.0001). Greater maturity also significantly increased the relative frequency of the confidential response, regardless of the medical problem~ Overall, 81% of mature adolescents and 69% of immature adolescents were granted confidentiality, regardless of the medical problem. The maturity effect was 5.7% (P < 0.0001). An unexpected finding was that physicians were equally likely to give confidential care whether they had known the adolescent and her family for many years or were meeting the adolescent for the first time. The familiarity effect was 0.2% and was not statistically significant. The age and maturity effects held essentially constant across vignette type and did not depend on one another or on familiarity; that is, there were no statistically significant interactions among the variables (vignette type, age, maturity, familiarity) (Table II). Physicians were most likely to agree to maintain confidentiality when the adolescent requested contraceptives
(86%), had an I U D as an incidental x-ray finding (80%), oi" had a diagnosis of gonorrhea (78%). In coritrast, only 54% were willing to grant confidentiality regarding drug Use. The order of the vignettes Within the PaCket also made a difference in the proportion of yes responses accorded a particular vignette. This effect was most pronounced with the vignette about drug use. When placed first in the packet it received 62% yes responses, whereas placed last in the packet it received 45% yes responses. The physicians' background was correlated with their responses (Table 1II). Younger physicians, those who spend more than 20% of their clinical time with adolescent patients, and those who are S A M members were most likely to agree to confidential care. Physicians who were most likely t o agree to confidentiality practice in the Pacific Coast states, New England, Alaska, Hawaii, and the Rocky Mountain states. Willingness to give confidential care did not vary with gender or parenthood status of the respondent. Many of the physicians combined private practice with hospital or academic positions, so it was difficult to determine a
Volume 106 Number 3
relationship between practice setting and confidentiality choices. DISCUSSION This study indicates that physicians, especially those who are younger than 44 years, those who are SAM members, and those who spend more than 20% of their clinical time with adolescents, are strongly disposed to give confidential care to adolescents represented in clinical vignettes. Although the physicians did take into account the patient's age and maturity (many specified these as important factors in response to our question "Please name the single most important factor influencing your decision"), they tended to give confidential care regardless of age or maturity. Physicians' markedly different responses to the drug cases as compared with cases related to sexuality strengthen our premise that decisions were based on subject area considered, not just on the characteristics of age or maturity of the adolescent or the duration of the patientphysician relationship. Perhaps reluctance to give confidential care to a drug user reflects the physicians' experiences or the context in which they practice clinically. Even among patients under care of specialists in adolescent medicine, drug users are often referred to drug treatment centers for group therapy, whereas physicians with experience in adolescent medicine routinely handle contraceptive needs or treat venereal infections. Perhaps some physicians believe that successful management of drug use requires parental supervision, particularly for younger teens. Sexual relationships during the teen years may be regarded as premature, whereas drug use at any age is illegal and is usually perceived as dysfunctional. The order of presentation of vignettes to the physicians influenced their willingness to give confidential care, especially with the vignette involving illicit drug use. Placed first in the packet it received 62% yes responses, whereas placed last in the packet it received only 45% yes responses. Perhaps this change reflects the fact that this situation appeared to require more parental involvement after the physicians contrasted it with the cases involving sexual behavior. Perhaps in such clinical situations, in which decision-making guidelines are unclear, physicians tend to refuse confidentiality because they are influenced by weariness or fatigue. Nevertheless, the responses indicate that, overall, physicians believe giving confidential care is the best way to protect the well-being of adolescent patients, egpecially
Confidential care f o r adolescents
521
those who are sexually active. However, a number of the physicians indicated that, although they would give confidential care, they often would try to persuade the adolescent to include her parents in the discussion or might agree to confidentiality contingent on the adolescent's response to their counseling about the problem. Some also indicated that they would be less likely to try to involve parents who might not act in a helpful manner. During the next few years, as debate continues over federal regulations governing confidentiality for adolescents, physicians' attitudes and behavior will be important to policy makers. Our study provides baseline data about physicians' attitudes regarding confidentiality. It raises questions for future research about differential treatment of male and female patients, delineation of special concerns about confidentiality for drug users, and the effect of varying the order of case presentation on physicians' decision making. Despite the limitations of the vignette approach, it seems useful in examining attitudes toward confidentiality and in eliciting some of the criteria used in decision making by a large sample of physicians. Although the physicians' replies are meaningful only insofar as their reactions to hypothetical situations approximate their responses to real clinical situations, it may be instructive that a representative national sample of physicians agreed to confidential care for adolescents in three fourths of the clinical vignettes. We thank Lincoln Moses, David Wright, and Mary DoMing for statistical design and analysis; Halsted Holman, Diana Dutton, and Fellows of the Clinical Scholar Program; J. Merrill Carlsmith, Sanford M. Dornbusch, and Ruth T. Gross, M.D., for their help in planning the study and reviewingthe manuscript; Robert Haggerty, M.D., and Iris Litt, M.D., for their endorsement of the study; Douglas Solomonfor precodingthe questionnaire;members of the American Academy of Pediatrics and the Society for Adolescent Medicine who participated in the study; and Sally Bridges, Bonnie Obrig, and Betsy Seroggs for secretarial assistance. REFERENCES
1. The legal status of adolescents. San Francisco, 1980, Scientific Analysis Corp. 2. Wilson JP: The rights of adolescents in the mental health system. Boston, 1978, DC Heath. 3. Elstein AS: Psychological approaches to decision making in health care. Am Behav Sci 25:18, 1982. 4. Nisbett R, Bellows N: Verbal reports about causal influences on social judgments: Private access versus public theories. J Pers Soc Psychol 35:613, 1977.
Physician attitudes toward confidential care for adolescents We examined factors that determine whether pediatricians will grant confidential care to adolescents. Through four vignettes, in which adolescents of different ages and maturity requested confidential care. we assessed the willingness o f physicians to give such care for four problems: request for contraceptive, diagnosis o f gonorrhea, intrauterine device found incidentally on x-ray study, and illicit drug use. A random sample o f 932 physicians responded to the vignettes. Overall, physicians agreed to give confidential care to adolescents in 75% o f the vignettes. They were more likely to give confidentiality in cases involving sexual activity. They supported confidential care for 87 % o f patients requesting contraceptives, but for only 54 % reporting illicit drug use. The proportion o f physicians supporting confidentiality increased with age and maturity o f the minors. Seventeen-year-old mature adolescents seeking contraceptives were given confidentiality by 97% o f physicians. Thus physician responses to vignettes indicated that they strongly support confidential care for adolescents as represented in the clinical vignettes. (J PEDIATR 106:517, 1985)
Joan Lovett, M.D., and Michael S. Wald, J.D. Stanford,
California
WHETHER ADOLESCENTS should be able to receive confidential health care is an issue with social, legal, and medical implications. Opponents believe that adolescents need adult guidance in health care decisions, especially those related to sexual activity or drug use. They argue that such guidance should come through family involvement, because the family is most familiar with the child's needs. In addition, some commentators oppose confidential care because they believe that parents have a right to know when their children, for whom they are responsible, receive health care.
In contrast, proponents of confidential health care regard it as essential for a trusting relationship between adolescent and physician. Furthermore, they assert that minors will not utilize certain medical services unless guaranteed confidential care. Although most proponents of confidential care encourage parental involvement when such involvement would benefit the adolescent, they would not make notification a prerequisite for treatment. The degree to which physicians may give confidential care is now being debated in courts and in the federal and IUD
From the Stanford Center Jor the Study o f Youth Development. At the time o f this research Dr. Lovett was a Robert Wood Johnson Clinical Scholar at Stanford Medical School. Mr. Wald is a Professor o f Law at Stanford University and a Member o f the Stanford Center for the Study o f Youth Development. The opinions expressed in this paper are not necessarily those of the Robert Wood Johnson Foundation or o f the Center for the Study o f Youth Development. Submitted for publication Oct. 12, 1983; accepted Aug. 10, 1984. Reprint requests: Joan Lovett, M,D., 25 Northampton, Berkeley, CA 94707.
Intrauterine device
state governments. Thirty states and the District of Columbia have laws requiring that physicians give confidential care when treating adolescents for certain conditions, usually those related to contraception, abortion, venereal disease, and drug abuse. However, other states require parental notificatibn before a minor can receive abortion or contraceptive services. ~ Recently, the U.S. Department of Health and Human Services initiated efforts to link federal aid to clinics providing contraceptive care with a requirement that parents be notified when minors seek contracep-
The Journal o f P E D I A T R I C S
517
5 18
Lovett and Wald
fives. At the time of writing, these regulations were being litigated. The physician whose adolescent patient asks that her or his parents not be involved in decisions about sexual activity or drug use is faced with an ethical dilemma, in addition to potential legal conflicts. The American Medical Association, in its Principles o f Medical Ethics', states: "A physician may not reveal the confidence entrusted to him in the course of medical attendance, or the deficiencies he may observe in the character of patients, unless he is required to do so by law or unless it becomes necessary in order to protect the welfare of the individual or of the community." This standard provides little guidance as to which actions would best "protect the welfare of the individual.''2 The issue is whether granting confidential care is detrimental for adolescents because of the absence of parental involvement or whether it is beneficial because it facilitates care for those who would not seek medical attention if parental involvement were required. Very little is known about physicians' attitudes toward granting confidentiality and the factors they consider important in making decisions regarding confidentiality. We present data on how practicing physicians decide whether to give confidential care to adolescents seeking treatment for a sex- or drug-related problem. METHODS Study design. A sample of i427 pediatricians was presented with a series of vignettes describing case histories of adolescents seeking confidential medical care. They were asked whether they would honor the patient's request for confidentiality. By systematically varying the age, maturity, and other characteristics of the patients in the vignettes, we were able to assess some of the variables that these physicians used in making their decisions. The study was designed to determine whether the physician's tendency to give confidential care varied with selected attributes of the female adolescent, the type of medical problem, or the physicians' characteristics. We chose to present standardized case vignettes to elicit physicians' responses, so that physician and patient variables could be disentangled. Case vignettes were used rather than direct observation or clinical simulations using actors so that a broader sampling of physicians and cases was possible2 W e acknowledge that there are limitations to the vignette approach: it assumes that the relevant variables are known and only the weights to be assigned to them are not, and that enough data needed for the decision are present. Because the judgments being studied are made on hypothetical cases, critics can raise the question of whether findings can be generalized to actual clinical
The Journal of Pediatrics March 1985
practice, the same objection raised with almost any experimental study. 3 Our decision to embed variables within the vignettes was based on a review of the literature as well as on a pilot study we conducted. Direct questioning about attitudes has been shown to be an unreliable predictor of actual behavior. Nisbett and Bellows 4 argue persuasively that when subjects are asked to introspect about their reasoning processes, their reports are based on prior implicit causal theories. 4 In a pilot study, we presented physicians with vignettes and asked them to list factors that would influence whether they would give confidential care to the adolescents described. Age, maturity, and duration of patient-physician relationship were listed most frequently, but we noted that physicians changed their responses as they proceeded through the four vignettes, because they were eager to base their decisions on consistent variables. Embedding the variables in the vignettes eliminated the physicians' need to appear consistent in their decisionmaking criteria. Based on the considerations described, we used vignettes to simulate clinical situations in which an adolescent requests that a physician not tell her parents that she is sexually active or that she uses illicit drugsl The vignettes included (I) a request for contraceptives, (2) a diagnosis of and request for treatment of gonorrhea, (3) an IUD found incidentally on x-ray examination, and (4) disclosure of illicit drug use. These types of medical problems were chosen because sexual behaviors and drug use are common reasons adolescents initiate confidential consultation with a physician and because legal controversy centers around these problems. The patients described in the vignettes were all females. Although adolescent males are involved in drug use and are treated for venereal disease, only females are affected by confidentiality rulings regarding use of the birth control pill or IUD. We chose to describe only female patients in the vignettes because inclusion of cases involving males would have introduced an additional variable in only two of the cases, making comparison across vignette types impossible. Differential treatment or attitudes toward adolescent males and females could be explored in a future study, holding constant problem type, age and maturity of the adolescent, and duration of physician-patient relationship. All physicians were sent a packet including all four problems. However, the information about the patients was varied in each vignette. The patient was described by age (14 or 17 years), maturity (mature or somewhat immature), and duration of patient-physician relationship (first patient contact or well known to physician). The variables of age and maturity were used as indicators of the
Volume 106 Number 3
Confidential care for adolescents
5 19
Table I. Percent of physicians who would give confidential care according to characteristics of adolescent and duration of physician-family relationship Adolescent characteristic
Age 17 years Age 14 years Mature Immature Well known Unknown Total
Vignette Drugs
Contraceptives
60 48 61 46 57 52 54
94 79 89 84 88 86 87
1UD
Gonorrhea
Total
91 70 88 73 81 80 80
87 68 84 72 77 79 78
83* 66 8 t* 69 76 74 75
*P < 0.0001, 17 yearsvs 14 yearS,mature vs immature. adolescent's capacity for responsible decision making and the likelihood that she would need advice on decisions. We looked at duration of relationship between the physician and the patient's family to see whether an established relationship with the parents would induce the physician to involve the parents. Every packet included two vignettes in which the adolescent described was 14 years of age, two vignettes in which she was described as 17 years, two vignettes in which she was mature, and so on. Anticipating that the order of the vignette within the packet might influence physicians' decisions, we varied the order. The vignettes were sent to all 589 physician members of the Society for AdolesCent Medicine (SAM) and to a random sample of 838 (of 16,436) Board Certified (nonSAM) members of the American Academy of Pediatricians (AAP) throughout the United States. The physicians were asked to respond to each vignette by either "Yes, I would agree to give confidential care" or "No, I would not agree to give confidential care." We also asked each physician to provide a brief description of the most 9 important factor influencing the decision, regardless of whether that factor was present in the vignette, in addition, we collected background information about the physician (age, gender, practice setting, parenthood status, experience with adolescents, membership in SAM or AAP, and state where practice is located). Seventy percent of the 1427 questionnaires were returned. Questionnaire packets were categorized as usable if all four responses were either yes or no. Nine hundred thirty-two respondents returned usable packets, and the remainder were not included in the analysis. The response rate differed between SAM members (81%) and non-SAM members (64%). Analysis of data. The study design allowed calculation of the overall percentages of confidential responses as well as comparisons of the variables embedded in the vignettes
(medical problem, mature vs immature adolescent, 14 vs 17 years of age, well known vs unknown) and comparisons of physician variables (e.g., old vs young, male vs female). We calculated the proportion of physicians replying yes or no in each of the 32 vignettes. We then calculated the effect of each of the variables on the physicians' responses. For example, we determined the effect of maturity by finding the average percent of yes responses to the mature a d o l e s c e n t (Pmat) minus the average p e r c e n t of yes responses to the immature adolescent (P~. . . . ) (i.e., Pmat - P~. . . . . . . ). w e defined the maturity effect as i/2(Pmat- Pimm~t). Thus, if Pm~t= 0.80 and Pi. . . . ~--" 0.70, t h e n emit - P~...... = 0.10 and l/2(Pm~t- P~..... ) = 0.05, the maturity effect. This effect impiies that maturity elicits 5% more yes responses than the mean. Analysis of variance was performed to determine the significance of the adolescent's age, duration of patient-physician relationship, request for contraceptive, IUD, gonorrhea, and drug problem on a proportion of yes responses. Analysis of variance also allowed comparisons among physician variables. RESULTS Overall, the physicians strongly supported confidentiality. The mean percentage of yes responses for all the vignettes was 75%. Only 4% responded that they would not give confidential care in any of the four vignettes, whereas 40% (371 respondents) would give confidential care in all four situations. However, the proportions of yes responses varied widely, with 97% of mature 17-year-old patients seeking contraceptives receiving responses in favor of confidentiality, whereas only 40% of immature 14-year-old drug users received confidentiality. The proportion of physicians supporting Confidentiality increased in cases involving older and more mature minors and in situations involving contraception rather than drug use. The 17-year-old
520
Lovett and Wald
The Journal of Pediatrics March 1985
Table II. Percent of physicians giving confidentialitj for various problems bY age and maturity
Medical )roblem Age of
Drugs
adolescent (yr)
Mature
17 14
69* 56
Contraceptives
I Immature
Mature
52 40
I Immature
97* 81
1UD MaCure
9l 77
96* 80
Gonorrhea
I Immature 86 60
Mature
Immature
90* 77
84 60
*P < 0.0001. Older, more mature adolescentswere more likelyto receiveconfidentialcare, regardless of type of medical problem.
Table IlL Percent of physicians giving confidential care by characteristics of physiciaus
Physician characteristic
I Drugs
Age (yr) (mean = 44 yr) >44 47 <44 62 Sex Male 54 Female 54 SAM member Yes 67 No 43 Children Yes 54 No 54 Clinical time with adolescents (%) (mean = 20%) >20 63 <20 47
Contraceptives
IUD
Gonorrhea
Total
83 89
77 84
70 85
69 80*
86 88
79 84
77 79
74 76
95 79
91 74
89 70
85* 67
85 89
80 80
77 79
74 72
90 83
85 77
83 74
80* 70
*P < 0.0001: Physicians<44 years, SAM member, >20% of clinicaltime with adolescents.
patient elicited substantially more yes responses than the 14-year-old, regardless of the medical problem (Table I). The age effect was 8.4% (P < 0.0001). Greater maturity also significantly increased the relative frequency of the confidential response, regardless of the medical problem~ Overall, 81% of mature adolescents and 69% of immature adolescents were granted confidentiality, regardless of the medical problem. The maturity effect was 5.7% (P < 0.0001). An unexpected finding was that physicians were equally likely to give confidential care whether they had known the adolescent and her family for many years or were meeting the adolescent for the first time. The familiarity effect was 0.2% and was not statistically significant. The age and maturity effects held essentially constant across vignette type and did not depend on one another or on familiarity; that is, there were no statistically significant interactions among the variables (vignette type, age, maturity, familiarity) (Table II). Physicians were most likely to agree to maintain confidentiality when the adolescent requested contraceptives
(86%), had an I U D as an incidental x-ray finding (80%), oi" had a diagnosis of gonorrhea (78%). In coritrast, only 54% were willing to grant confidentiality regarding drug Use. The order of the vignettes Within the PaCket also made a difference in the proportion of yes responses accorded a particular vignette. This effect was most pronounced with the vignette about drug use. When placed first in the packet it received 62% yes responses, whereas placed last in the packet it received 45% yes responses. The physicians' background was correlated with their responses (Table 1II). Younger physicians, those who spend more than 20% of their clinical time with adolescent patients, and those who are S A M members were most likely to agree to confidential care. Physicians who were most likely t o agree to confidentiality practice in the Pacific Coast states, New England, Alaska, Hawaii, and the Rocky Mountain states. Willingness to give confidential care did not vary with gender or parenthood status of the respondent. Many of the physicians combined private practice with hospital or academic positions, so it was difficult to determine a
Volume 106 Number 3
relationship between practice setting and confidentiality choices. DISCUSSION This study indicates that physicians, especially those who are younger than 44 years, those who are SAM members, and those who spend more than 20% of their clinical time with adolescents, are strongly disposed to give confidential care to adolescents represented in clinical vignettes. Although the physicians did take into account the patient's age and maturity (many specified these as important factors in response to our question "Please name the single most important factor influencing your decision"), they tended to give confidential care regardless of age or maturity. Physicians' markedly different responses to the drug cases as compared with cases related to sexuality strengthen our premise that decisions were based on subject area considered, not just on the characteristics of age or maturity of the adolescent or the duration of the patientphysician relationship. Perhaps reluctance to give confidential care to a drug user reflects the physicians' experiences or the context in which they practice clinically. Even among patients under care of specialists in adolescent medicine, drug users are often referred to drug treatment centers for group therapy, whereas physicians with experience in adolescent medicine routinely handle contraceptive needs or treat venereal infections. Perhaps some physicians believe that successful management of drug use requires parental supervision, particularly for younger teens. Sexual relationships during the teen years may be regarded as premature, whereas drug use at any age is illegal and is usually perceived as dysfunctional. The order of presentation of vignettes to the physicians influenced their willingness to give confidential care, especially with the vignette involving illicit drug use. Placed first in the packet it received 62% yes responses, whereas placed last in the packet it received only 45% yes responses. Perhaps this change reflects the fact that this situation appeared to require more parental involvement after the physicians contrasted it with the cases involving sexual behavior. Perhaps in such clinical situations, in which decision-making guidelines are unclear, physicians tend to refuse confidentiality because they are influenced by weariness or fatigue. Nevertheless, the responses indicate that, overall, physicians believe giving confidential care is the best way to protect the well-being of adolescent patients, egpecially
Confidential care f o r adolescents
521
those who are sexually active. However, a number of the physicians indicated that, although they would give confidential care, they often would try to persuade the adolescent to include her parents in the discussion or might agree to confidentiality contingent on the adolescent's response to their counseling about the problem. Some also indicated that they would be less likely to try to involve parents who might not act in a helpful manner. During the next few years, as debate continues over federal regulations governing confidentiality for adolescents, physicians' attitudes and behavior will be important to policy makers. Our study provides baseline data about physicians' attitudes regarding confidentiality. It raises questions for future research about differential treatment of male and female patients, delineation of special concerns about confidentiality for drug users, and the effect of varying the order of case presentation on physicians' decision making. Despite the limitations of the vignette approach, it seems useful in examining attitudes toward confidentiality and in eliciting some of the criteria used in decision making by a large sample of physicians. Although the physicians' replies are meaningful only insofar as their reactions to hypothetical situations approximate their responses to real clinical situations, it may be instructive that a representative national sample of physicians agreed to confidential care for adolescents in three fourths of the clinical vignettes. We thank Lincoln Moses, David Wright, and Mary DoMing for statistical design and analysis; Halsted Holman, Diana Dutton, and Fellows of the Clinical Scholar Program; J. Merrill Carlsmith, Sanford M. Dornbusch, and Ruth T. Gross, M.D., for their help in planning the study and reviewingthe manuscript; Robert Haggerty, M.D., and Iris Litt, M.D., for their endorsement of the study; Douglas Solomonfor precodingthe questionnaire;members of the American Academy of Pediatrics and the Society for Adolescent Medicine who participated in the study; and Sally Bridges, Bonnie Obrig, and Betsy Seroggs for secretarial assistance. REFERENCES
1. The legal status of adolescents. San Francisco, 1980, Scientific Analysis Corp. 2. Wilson JP: The rights of adolescents in the mental health system. Boston, 1978, DC Heath. 3. Elstein AS: Psychological approaches to decision making in health care. Am Behav Sci 25:18, 1982. 4. Nisbett R, Bellows N: Verbal reports about causal influences on social judgments: Private access versus public theories. J Pers Soc Psychol 35:613, 1977.