John G. Scott, MD, PhD

Acute respiratory tract (ART) infections, such as common cold, bronchitis, pharyngitis, sinusitis, and otitis media, are among the most common problems seen in primary care practice.¹ Unnecessary use of antibiotics for these infections is a major worldwide problem both in terms of cost² and as a contributor to the development of antibiotic-resistant bacteria.³

Although there is some evidence that physicians misdiagnose many viral infections as bacterial,^4,5 recent studies suggest that the reasons for unnecessary antibiotic prescribing are more complex, having as much or more to do with patient and physician expectations as with physicians’ diagnostic skills.^6-8 These studies are limited to describing perceptions of behavior rather than actual behavior, because of their use of interview and focus group data. Consequently, we do not know what actually happens during outpatient visits for ART infections that leads to antibiotic prescribing.

Two studies by Stivers^9,10 underscore the importance of directly observing what transpires during encounters with pediatric ART infection patients. Stivers’ examination of videotaped visits found that, in some cases, parental pressure for antibiotics influenced the physician’s decision to prescribe. This finding has not been replicated, however, in family practice settings, where both adults and children are seen. We used direct observation of outpatient visits to family physicians for ART infections to analyze the effects of physician-patient communication on unnecessary antibiotic prescribing. By understanding the ways these communication patterns influence prescribing behavior, practicing family physicians can develop strategies to deliver more appropriate care for ART infections.

Methods

These data were collected as part of the Prevention & Competing Demands in Primary Care Study, which was an in-depth observational study begun in October 1996 and completed in August 1999 that examined the organizational and clinical structures and process of community-based family practices. Each of 18 purposefully selected practices was studied using a multimethod comparative case study design that involved extensive direct observation of clinical encounters and office systems by field researchers who spent 4 weeks or more in each practice. Field researchers directly observed and dictated descriptions of approximately 30 patient encounters with each of the more than 50 clinicians. Details of the sampling and data collection are available elsewhere in this issue.¹¹

Data Analysis and Interpretation

Encounters related to ART infection were identified in the database using search terms for symptoms and diagnoses including: sore throat, runny nose, congestion, cough, drainage, postnasal drainage, earache, cold, upper respiratory infection, pharyngitis, sinusitis, bronchitis, and otitis. ART infection was identified as the principal or associated diagnosis in 316 outpatient visits of a total of 1637 observed encounters; 298 had sufficiently rich data for analysis. The encounters were first coded for antibiotic use or nonuse.

Before any qualitative analysis began, visits during which antibiotics were prescribed were further characterized as appropriate or unnecessary according to guidelines by the Centers for Disease Control and Prevention (CDC) for judicious use of antibiotics for children¹² and adults.Table W1¹³* Two family physicians assigned appropriate/unnecessary codes independently. Inter-rater reliability was good (k=0.71). All disagreements were resolved by discussion.

Subsequently, the text for each outpatient visit was read independently by 2 family physicians, a medical anthropologist, a nurse, and a communication specialist. This research team discussed individual encounters as a group to identify emerging patterns of physician-patient interaction.

Results

Women made up 59% of the study sample; 64% were 16 years or older (and classified as adults). Antibiotics were prescribed in 204 of the 298 ART infection encounters (68%). Antibiotic use was unnecessary according to the CDC guidelines in 164 of these (80%). Adults were more likely than children to receive unnecessary antibiotics Table 1.

Our analysis identified 6 different types of patient behaviors that advocated for medication, particularly antibioticsTable 2. These behaviors fell into 3 broad categories: explicit requests, presentation of chief complaint, and appeals to lifeworld circumstances. Multiple pressures were noted in many encounters.

While patients occasionally made direct requests for antibiotics, they much more frequently positioned themselves indirectly for receiving antibiotic treatment by the way they presented the chief complaint. Four distinct approaches were identified: symptoms only,⁹ candidate diagnosis,⁹ implied candidate diagnosis,⁹ and portraying the severity and inability to shake the illness.

A second category of indirect approach used life-world circumstances¹⁰ (eg, an upcoming family vacation) or a past history with successful antibiotic treatment to formulate appeals for antibiotics in the current encounter. In those cases in which antibiotics were clearly unnecessary, physicians often rationalized their prescribing practices by finding symptoms or assigning diagnoses to justify antibiotic use. Each of these patient pressures, as well as the physician-rationalizing behavior, is illustrated with sample visits. The samples are taken directly from transcribed field notes, but the names have been altered to protect the identity of patients and clinicians.

Explicit Request

Explicit requests for antibiotics were observed in only 6% of cases (n=15). For example:

Claire asked the patient, “How are you doing?” and she said, “Well, I’m coughing up phlegm, I ache and I have chills and a sore throat.” Claire said, “You have bronchoconstriction, and 3 times a day, if you need to, you should use proventil.” The patient asked if she could have an antibiotic for her cold; cephalexin has worked in the past. Claire said that she would get her cephalexin and also some samples of an inhaler.

Presentation of the Chief Complaint

Patients frequently put pressure on the physician for treatment during the presentation of the chief complaint, the exception being the symptoms-only presentation. This is different from the other indirect pressures, which usually occurred during different parts of the medical encounter.

Symptoms-only presentation (eg, “I have a cough and a sore throat.”) In the symptoms-only approach (n=15), the patient reports his or her symptoms with little embellishment. This approach does not pressure physicians for antibiotic treatment.

Candidate diagnosis (eg, “I think I’ve got strep throat.”) In contrast, patients also presented their chief complaint to the physician by offering a candidate diagnosis (n=18). As shown in the following example, the patient responds by offering a diagnosis. This is a way of indirectly advocating for antibiotic treatment.

A 21-year-old white woman went to see Dr. Maxwell with an acute problem of congestion. Dr Maxwell said, “Well, how are you doing?” The patient said, “It sounds like bronchitis. It started about 4 days ago.”

Implied candidate diagnosis (eg, “My throat hurts; it’s red; and it has white spots.”) The implied candidate diagnosis is a hybrid of the symptoms-only and the candidate diagnosis approaches (n=48). When presenting their chief complaint, patients reported very specific symptoms that indexed a particular diagnosis. For example:

A 29-year-old woman went to see Dr Redmond with swollen glands, congestion, and white spots on her throat. When Dr Redmond and I went into the examination room, the patient had a pink paper top on, and Dr Redmond told her that her throat culture was negative.

The patient reports that she has swollen glands, congestion, and white spots on her throat. The symptoms specifically index a particular condition (strep throat). The patient’s presentation of symptoms clearly implies a diagnosis of strep throat, and the physician ordered a strep culture before seeing the patient.

Candidate diagnoses and implied candidate diagnoses delicately assert that the nature of the patient’s problem is already known. The reason for the medical visit is to seek treatment for the patient’s already known condition. When candidate and implied candidate diagnoses point to a condition the patient believes to be treatable (eg, bronchitis, strep throat, ear infection), this way of presenting the chief complaint looks directly ahead to a treatment involving a prescription for an antibiotic and thus indirectly pressures the physician to prescribe one.

Portraying the severity of one’s illness (eg, “I can’t shake this, Doc.”) The most common strategy was for patients to subtly pressure physicians for medication by portraying the severity of their condition and their inability to shake the illness on their own (n=99). For example:

The patient was sitting up on the table, and right away he told Dr Lamont, “I just can’t shake it. I feel like the back of my throat has raw hamburger hanging in it.” Dr Lamont checked the patient’s throat well, and the patient said, “This has lasted 4 days and it has been getting worse today.” Dr Lamont checked the patient’s ears, glands, and lungs. “I’m going to give you a shot of penicillin, slow release. It’s some kind of an infection. It may be a virus.”

Portraying the severity of one’s illness may not in and of itself advocate for medication; however, portrayals of the severity of one’s condition were usually accompanied by other actions implicating the need for medication. By opening the encounter with the announcement “I just can’t shake it,” the patient implies that he needs help in getting well. This subtly suggests the need for a prescription medication to alleviate his sore throat. At the end of this visit, the patient receives an antibiotic shot.

Appeals to Nonmedical Circumstances

Patients also used nonmedical circumstances to advocate for medication. These behaviors tended to occur after the problem presentation in the encounter and either centered on some important event, such as a big examination or a trip out of town (n=16), or focused on a previous positive experience with antibiotics for themselves or a family member (n=39).

Appealing to life-world circumstance (eg, “But I’m going to Disney World.”) This patient uses an upcoming trip to make an appeal to the clinician to prescribe medication:

The patient is a 33-year-old man coming in with an acute problem of a sore throat. The patient stated that he had been trying to manage this on his own, but he was taking his wife and 2 children to Disney World at the end of the week and was becoming worried that he was still going be sick and not able to enjoy a trip that they had saved so long for. He also told Dr Liam: “I know we’ll just get to Florida, and the kids will get sick, and then we’ll all be sick again. Dr. Liam said, “Well, we can have you bring them in, but then we’d be treating them for something that they haven’t gotten. Let me think about this a bit.” He does the rapid strep test, and it’s negative. Dr Liam reported the news of a negative strep test and said, “Many times we get a 50% false-negative, so I’m gonna go ahead and put you on an antibiotic and see if we can’t get you feeling better.” With this the patient said, “Well, what do you think I should do about my kids?” Dr. Liam asked if the kids were seen in this clinic, and the patient responded that they had never been seen there before. Dr Liam said, “Well, I’ll go ahead and give you a script for erythromycin in case these kids get sick down in Florida. If they do, go ahead and give them the medicine; if they don’t, throw away the prescription.”

This case is interesting because once he is treated with an antibiotic, the patient uses the same argument to make an appeal for antibiotics for his children (both of whom have never been seen by this physician).

Previous positive experience with antibiotics (eg, “I got an antibiotic for this before.”). Patients also appealed to other nonmedical contingencies to advocate for antibiotic treatment. For example:

Our next patient was a 51-year-old woman complaining of a cold and laryngitis. The doctor asked the patient about her symptoms. The patient responded, saying that she had been taking medication during the end of December for the same symptoms; they had cleared after taking antibiotics, and now they were back again.

The patient indirectly makes an appeal for antibiotic treatment by stating that she received antibiotics in the past for the same symptoms that she has now.

Patients used several variations of this approach. These included stating that another physician prescribed an antibiotic for this illness in the past; that others in the family are sick with an illness for which they received antibiotics; that they have a history of illness for which antibiotics are regularly prescribed; and that they were recently taking an antibiotic for an illness that has not improved (with the idea that an antibiotic is needed again).

Effectiveness of Patient Pressures

Physicians prescribed an antibiotic unnecessarily in 80% of the encounters in which some patient pressure was observed. They seemed able to resist certain types of pressures better than others. Unnecessary antibiotics were prescribed for a smaller percentage of implied candidate diagnoses and candidate diagnoses and for a larger percentage of direct patient requests and previous positive experiences with antibiotics Table 3.

Physicians’ Response to Prescribing an Unnecessary Antibiotic

When physicians prescribed an antibiotic unnecessarily, they often rationalized this practice by finding symptoms or assigning diagnoses that, to them, justified prescribing antibiotics. Physicians used various rationales, such as red throat or enlarged tonsils; severe, prolonged, or productive cough; yellow or green mucus; sinus tenderness on palpation; associated chronic disease; history of previous infection; and the desire to “cover” the patient “just in case.” None of these rationales are supported by evidence as correlating with bacterial infection. An example of this kind of rationalization follows:

This is a 20-year-old woman coming in with a complaint of a worsening cough. She said that her chest had a prickly, burning sensation, and it hurt to breathe. Dr Hart asked if she was able to bring anything up. She said that she really couldn’t. It was just a really terrible barky cough. Following the physician examination, the physician told the patient that her lungs basically sounded clear, but she could certainly hear some rough bronchial sounds. With this, she said. “What I think is happening here with your cold is that it is probably ending up in a bronchitis-type situation, and probably what we should do is put you on an antibiotic and order a decongestant.”

Discussion

This investigation, in agreement with the pediatric studies of Stivers,^9,10 suggests that the connection between patient diagnosis and physician prescribing is highly complex, involving patient presentation and physician-patient communication as much as, if not more than, physician diagnostic skills. Also, these data suggest that physicians are better able to resist patient pressures that are framed in medical terms such as candidate diagnoses or implied candidate diagnoses but are much less able to resist pressures that are not medicalized, such as portraying severity of illness and use of life-world circumstances. Thus, it is not surprising that past interventions designed to increase physician knowledge regarding when to prescribe antibiotics have had limited success.^14,15 Physicians appear to be trying to maximize patient satisfaction by giving antibiotic-seeking patients what they want. Our findings show the need to modify current thinking about the diagnostic and treatment process to reduce the use of antibiotics. Rather than thinking of these processes as physician controlled, the powerful role patients play in this interaction must be considered.

Our study has important implications for future research. From a methodologic standpoint, our findings illustrate the importance of qualitative evaluation of directly observed medical encounters. The patterns of patient behavior observed could not have been discerned using survey, interview, or focus group data.

Limitations

Because these data were collected by field researchers who were unaware that ART infection would be a focus of our study, it is possible that there were other patient symptoms and behavior related to ART infection, as well as physician behaviors related to antibiotic prescribing, that were not recorded. The data were sufficiently rich, however, to easily and reliably apply the CDC guidelines for appropriate use of antibiotics. Any unrecorded behaviors might add to, but not substantially change, our conclusions that patients indirectly pressure their physicians for treatment, and physicians respond by giving antibiotics. Studies using videotaped encounters might uncover such additional important patient and physician behaviors. Since the patient population studied was limited to a single midwestern state, it is possible that other populations with a different ethnic or racial mix might behave differently. Future research in this area should attempt to include such populations. Finally, too few encounters per physician were observed in this study to evaluate whether particular physicians were high or low prescribers (such a pattern has been reported by De Sutter and colleagues¹⁶).

Conclusions

Physicians should be educated about the subtle approaches patients use to pressure them for antibiotic treatment and should be shown techniques for responding to these pressures without prescribing antibiotics unnecessarily. Our findings also suggest the need to increase patients’ awareness both of the dangers and lack of effectiveness of using antibiotics for ART infections and of the amount of influence that patients have on antibiotic prescribing. Macfarlane and coworkers17 have shown that use of patient education materials reduces visits for ART infection. Additional approaches to decreasing patient pressure for antibiotic prescriptions are needed to diminish antibiotic overuse and its public health consequences.

Acknowledgments

Our study was funded by the Agency for Healthcare Research and Quality Grant R01 HS08776. Dr Scott is a postdoctoral fellow supported by the Health Resources and Services Administration (HRSA) PE1011 and the Agency for Healthcare Research and Quality (AHRQ) HS09788. Analysis of these data was supported by a Research Center grant from the American Academy of Family Physicians (Center for Research in Family Practice and Primary Care). Drs Jaen and Crabtree are associated with the Center for Research in Family Practicer and Primary Care, Cleveland, New Brunswick, Allentown. and San Antonio. The authors wish to thank the family physicians of Nebraska who were willing to open their practices to us. We also thank Kurt C. Stange, MD, PhD, for his thoughtful comments on drafts of this manuscript.

Related Resources

U.S. Centers for Disease Control and Prevention—Promoting Appropriate Antibiotic Use in the Community http://www.cdc.gov/antibioticresistance/tools.htm

A vast resource of of patient education resources.

Acute respiratory tract (ART) infections, such as common cold, bronchitis, pharyngitis, sinusitis, and otitis media, are among the most common problems seen in primary care practice.¹ Unnecessary use of antibiotics for these infections is a major worldwide problem both in terms of cost² and as a contributor to the development of antibiotic-resistant bacteria.³

Although there is some evidence that physicians misdiagnose many viral infections as bacterial,^4,5 recent studies suggest that the reasons for unnecessary antibiotic prescribing are more complex, having as much or more to do with patient and physician expectations as with physicians’ diagnostic skills.^6-8 These studies are limited to describing perceptions of behavior rather than actual behavior, because of their use of interview and focus group data. Consequently, we do not know what actually happens during outpatient visits for ART infections that leads to antibiotic prescribing.

Two studies by Stivers^9,10 underscore the importance of directly observing what transpires during encounters with pediatric ART infection patients. Stivers’ examination of videotaped visits found that, in some cases, parental pressure for antibiotics influenced the physician’s decision to prescribe. This finding has not been replicated, however, in family practice settings, where both adults and children are seen. We used direct observation of outpatient visits to family physicians for ART infections to analyze the effects of physician-patient communication on unnecessary antibiotic prescribing. By understanding the ways these communication patterns influence prescribing behavior, practicing family physicians can develop strategies to deliver more appropriate care for ART infections.

Methods

These data were collected as part of the Prevention & Competing Demands in Primary Care Study, which was an in-depth observational study begun in October 1996 and completed in August 1999 that examined the organizational and clinical structures and process of community-based family practices. Each of 18 purposefully selected practices was studied using a multimethod comparative case study design that involved extensive direct observation of clinical encounters and office systems by field researchers who spent 4 weeks or more in each practice. Field researchers directly observed and dictated descriptions of approximately 30 patient encounters with each of the more than 50 clinicians. Details of the sampling and data collection are available elsewhere in this issue.¹¹

Data Analysis and Interpretation

Encounters related to ART infection were identified in the database using search terms for symptoms and diagnoses including: sore throat, runny nose, congestion, cough, drainage, postnasal drainage, earache, cold, upper respiratory infection, pharyngitis, sinusitis, bronchitis, and otitis. ART infection was identified as the principal or associated diagnosis in 316 outpatient visits of a total of 1637 observed encounters; 298 had sufficiently rich data for analysis. The encounters were first coded for antibiotic use or nonuse.

Before any qualitative analysis began, visits during which antibiotics were prescribed were further characterized as appropriate or unnecessary according to guidelines by the Centers for Disease Control and Prevention (CDC) for judicious use of antibiotics for children¹² and adults.Table W1¹³* Two family physicians assigned appropriate/unnecessary codes independently. Inter-rater reliability was good (k=0.71). All disagreements were resolved by discussion.

Subsequently, the text for each outpatient visit was read independently by 2 family physicians, a medical anthropologist, a nurse, and a communication specialist. This research team discussed individual encounters as a group to identify emerging patterns of physician-patient interaction.

Results

Women made up 59% of the study sample; 64% were 16 years or older (and classified as adults). Antibiotics were prescribed in 204 of the 298 ART infection encounters (68%). Antibiotic use was unnecessary according to the CDC guidelines in 164 of these (80%). Adults were more likely than children to receive unnecessary antibiotics Table 1.

Our analysis identified 6 different types of patient behaviors that advocated for medication, particularly antibioticsTable 2. These behaviors fell into 3 broad categories: explicit requests, presentation of chief complaint, and appeals to lifeworld circumstances. Multiple pressures were noted in many encounters.

While patients occasionally made direct requests for antibiotics, they much more frequently positioned themselves indirectly for receiving antibiotic treatment by the way they presented the chief complaint. Four distinct approaches were identified: symptoms only,⁹ candidate diagnosis,⁹ implied candidate diagnosis,⁹ and portraying the severity and inability to shake the illness.

A second category of indirect approach used life-world circumstances¹⁰ (eg, an upcoming family vacation) or a past history with successful antibiotic treatment to formulate appeals for antibiotics in the current encounter. In those cases in which antibiotics were clearly unnecessary, physicians often rationalized their prescribing practices by finding symptoms or assigning diagnoses to justify antibiotic use. Each of these patient pressures, as well as the physician-rationalizing behavior, is illustrated with sample visits. The samples are taken directly from transcribed field notes, but the names have been altered to protect the identity of patients and clinicians.

Explicit Request

Explicit requests for antibiotics were observed in only 6% of cases (n=15). For example:

Claire asked the patient, “How are you doing?” and she said, “Well, I’m coughing up phlegm, I ache and I have chills and a sore throat.” Claire said, “You have bronchoconstriction, and 3 times a day, if you need to, you should use proventil.” The patient asked if she could have an antibiotic for her cold; cephalexin has worked in the past. Claire said that she would get her cephalexin and also some samples of an inhaler.

Presentation of the Chief Complaint

Patients frequently put pressure on the physician for treatment during the presentation of the chief complaint, the exception being the symptoms-only presentation. This is different from the other indirect pressures, which usually occurred during different parts of the medical encounter.

Symptoms-only presentation (eg, “I have a cough and a sore throat.”) In the symptoms-only approach (n=15), the patient reports his or her symptoms with little embellishment. This approach does not pressure physicians for antibiotic treatment.

Candidate diagnosis (eg, “I think I’ve got strep throat.”) In contrast, patients also presented their chief complaint to the physician by offering a candidate diagnosis (n=18). As shown in the following example, the patient responds by offering a diagnosis. This is a way of indirectly advocating for antibiotic treatment.

A 21-year-old white woman went to see Dr. Maxwell with an acute problem of congestion. Dr Maxwell said, “Well, how are you doing?” The patient said, “It sounds like bronchitis. It started about 4 days ago.”

Implied candidate diagnosis (eg, “My throat hurts; it’s red; and it has white spots.”) The implied candidate diagnosis is a hybrid of the symptoms-only and the candidate diagnosis approaches (n=48). When presenting their chief complaint, patients reported very specific symptoms that indexed a particular diagnosis. For example:

A 29-year-old woman went to see Dr Redmond with swollen glands, congestion, and white spots on her throat. When Dr Redmond and I went into the examination room, the patient had a pink paper top on, and Dr Redmond told her that her throat culture was negative.

The patient reports that she has swollen glands, congestion, and white spots on her throat. The symptoms specifically index a particular condition (strep throat). The patient’s presentation of symptoms clearly implies a diagnosis of strep throat, and the physician ordered a strep culture before seeing the patient.

Candidate diagnoses and implied candidate diagnoses delicately assert that the nature of the patient’s problem is already known. The reason for the medical visit is to seek treatment for the patient’s already known condition. When candidate and implied candidate diagnoses point to a condition the patient believes to be treatable (eg, bronchitis, strep throat, ear infection), this way of presenting the chief complaint looks directly ahead to a treatment involving a prescription for an antibiotic and thus indirectly pressures the physician to prescribe one.

Portraying the severity of one’s illness (eg, “I can’t shake this, Doc.”) The most common strategy was for patients to subtly pressure physicians for medication by portraying the severity of their condition and their inability to shake the illness on their own (n=99). For example:

The patient was sitting up on the table, and right away he told Dr Lamont, “I just can’t shake it. I feel like the back of my throat has raw hamburger hanging in it.” Dr Lamont checked the patient’s throat well, and the patient said, “This has lasted 4 days and it has been getting worse today.” Dr Lamont checked the patient’s ears, glands, and lungs. “I’m going to give you a shot of penicillin, slow release. It’s some kind of an infection. It may be a virus.”

Portraying the severity of one’s illness may not in and of itself advocate for medication; however, portrayals of the severity of one’s condition were usually accompanied by other actions implicating the need for medication. By opening the encounter with the announcement “I just can’t shake it,” the patient implies that he needs help in getting well. This subtly suggests the need for a prescription medication to alleviate his sore throat. At the end of this visit, the patient receives an antibiotic shot.

Appeals to Nonmedical Circumstances

Patients also used nonmedical circumstances to advocate for medication. These behaviors tended to occur after the problem presentation in the encounter and either centered on some important event, such as a big examination or a trip out of town (n=16), or focused on a previous positive experience with antibiotics for themselves or a family member (n=39).

Appealing to life-world circumstance (eg, “But I’m going to Disney World.”) This patient uses an upcoming trip to make an appeal to the clinician to prescribe medication:

The patient is a 33-year-old man coming in with an acute problem of a sore throat. The patient stated that he had been trying to manage this on his own, but he was taking his wife and 2 children to Disney World at the end of the week and was becoming worried that he was still going be sick and not able to enjoy a trip that they had saved so long for. He also told Dr Liam: “I know we’ll just get to Florida, and the kids will get sick, and then we’ll all be sick again. Dr. Liam said, “Well, we can have you bring them in, but then we’d be treating them for something that they haven’t gotten. Let me think about this a bit.” He does the rapid strep test, and it’s negative. Dr Liam reported the news of a negative strep test and said, “Many times we get a 50% false-negative, so I’m gonna go ahead and put you on an antibiotic and see if we can’t get you feeling better.” With this the patient said, “Well, what do you think I should do about my kids?” Dr. Liam asked if the kids were seen in this clinic, and the patient responded that they had never been seen there before. Dr Liam said, “Well, I’ll go ahead and give you a script for erythromycin in case these kids get sick down in Florida. If they do, go ahead and give them the medicine; if they don’t, throw away the prescription.”

This case is interesting because once he is treated with an antibiotic, the patient uses the same argument to make an appeal for antibiotics for his children (both of whom have never been seen by this physician).

Previous positive experience with antibiotics (eg, “I got an antibiotic for this before.”). Patients also appealed to other nonmedical contingencies to advocate for antibiotic treatment. For example:

Our next patient was a 51-year-old woman complaining of a cold and laryngitis. The doctor asked the patient about her symptoms. The patient responded, saying that she had been taking medication during the end of December for the same symptoms; they had cleared after taking antibiotics, and now they were back again.

The patient indirectly makes an appeal for antibiotic treatment by stating that she received antibiotics in the past for the same symptoms that she has now.

Patients used several variations of this approach. These included stating that another physician prescribed an antibiotic for this illness in the past; that others in the family are sick with an illness for which they received antibiotics; that they have a history of illness for which antibiotics are regularly prescribed; and that they were recently taking an antibiotic for an illness that has not improved (with the idea that an antibiotic is needed again).

Effectiveness of Patient Pressures

Physicians prescribed an antibiotic unnecessarily in 80% of the encounters in which some patient pressure was observed. They seemed able to resist certain types of pressures better than others. Unnecessary antibiotics were prescribed for a smaller percentage of implied candidate diagnoses and candidate diagnoses and for a larger percentage of direct patient requests and previous positive experiences with antibiotics Table 3.

Physicians’ Response to Prescribing an Unnecessary Antibiotic

When physicians prescribed an antibiotic unnecessarily, they often rationalized this practice by finding symptoms or assigning diagnoses that, to them, justified prescribing antibiotics. Physicians used various rationales, such as red throat or enlarged tonsils; severe, prolonged, or productive cough; yellow or green mucus; sinus tenderness on palpation; associated chronic disease; history of previous infection; and the desire to “cover” the patient “just in case.” None of these rationales are supported by evidence as correlating with bacterial infection. An example of this kind of rationalization follows:

This is a 20-year-old woman coming in with a complaint of a worsening cough. She said that her chest had a prickly, burning sensation, and it hurt to breathe. Dr Hart asked if she was able to bring anything up. She said that she really couldn’t. It was just a really terrible barky cough. Following the physician examination, the physician told the patient that her lungs basically sounded clear, but she could certainly hear some rough bronchial sounds. With this, she said. “What I think is happening here with your cold is that it is probably ending up in a bronchitis-type situation, and probably what we should do is put you on an antibiotic and order a decongestant.”

Discussion

This investigation, in agreement with the pediatric studies of Stivers,^9,10 suggests that the connection between patient diagnosis and physician prescribing is highly complex, involving patient presentation and physician-patient communication as much as, if not more than, physician diagnostic skills. Also, these data suggest that physicians are better able to resist patient pressures that are framed in medical terms such as candidate diagnoses or implied candidate diagnoses but are much less able to resist pressures that are not medicalized, such as portraying severity of illness and use of life-world circumstances. Thus, it is not surprising that past interventions designed to increase physician knowledge regarding when to prescribe antibiotics have had limited success.^14,15 Physicians appear to be trying to maximize patient satisfaction by giving antibiotic-seeking patients what they want. Our findings show the need to modify current thinking about the diagnostic and treatment process to reduce the use of antibiotics. Rather than thinking of these processes as physician controlled, the powerful role patients play in this interaction must be considered.

Our study has important implications for future research. From a methodologic standpoint, our findings illustrate the importance of qualitative evaluation of directly observed medical encounters. The patterns of patient behavior observed could not have been discerned using survey, interview, or focus group data.

Limitations

Because these data were collected by field researchers who were unaware that ART infection would be a focus of our study, it is possible that there were other patient symptoms and behavior related to ART infection, as well as physician behaviors related to antibiotic prescribing, that were not recorded. The data were sufficiently rich, however, to easily and reliably apply the CDC guidelines for appropriate use of antibiotics. Any unrecorded behaviors might add to, but not substantially change, our conclusions that patients indirectly pressure their physicians for treatment, and physicians respond by giving antibiotics. Studies using videotaped encounters might uncover such additional important patient and physician behaviors. Since the patient population studied was limited to a single midwestern state, it is possible that other populations with a different ethnic or racial mix might behave differently. Future research in this area should attempt to include such populations. Finally, too few encounters per physician were observed in this study to evaluate whether particular physicians were high or low prescribers (such a pattern has been reported by De Sutter and colleagues¹⁶).

Conclusions

Physicians should be educated about the subtle approaches patients use to pressure them for antibiotic treatment and should be shown techniques for responding to these pressures without prescribing antibiotics unnecessarily. Our findings also suggest the need to increase patients’ awareness both of the dangers and lack of effectiveness of using antibiotics for ART infections and of the amount of influence that patients have on antibiotic prescribing. Macfarlane and coworkers17 have shown that use of patient education materials reduces visits for ART infection. Additional approaches to decreasing patient pressure for antibiotic prescriptions are needed to diminish antibiotic overuse and its public health consequences.

Acknowledgments

Our study was funded by the Agency for Healthcare Research and Quality Grant R01 HS08776. Dr Scott is a postdoctoral fellow supported by the Health Resources and Services Administration (HRSA) PE1011 and the Agency for Healthcare Research and Quality (AHRQ) HS09788. Analysis of these data was supported by a Research Center grant from the American Academy of Family Physicians (Center for Research in Family Practice and Primary Care). Drs Jaen and Crabtree are associated with the Center for Research in Family Practicer and Primary Care, Cleveland, New Brunswick, Allentown. and San Antonio. The authors wish to thank the family physicians of Nebraska who were willing to open their practices to us. We also thank Kurt C. Stange, MD, PhD, for his thoughtful comments on drafts of this manuscript.

Related Resources

U.S. Centers for Disease Control and Prevention—Promoting Appropriate Antibiotic Use in the Community http://www.cdc.gov/antibioticresistance/tools.htm

A vast resource of of patient education resources.

Acute respiratory tract (ART) infections, such as common cold, bronchitis, pharyngitis, sinusitis, and otitis media, are among the most common problems seen in primary care practice.¹ Unnecessary use of antibiotics for these infections is a major worldwide problem both in terms of cost² and as a contributor to the development of antibiotic-resistant bacteria.³

Although there is some evidence that physicians misdiagnose many viral infections as bacterial,^4,5 recent studies suggest that the reasons for unnecessary antibiotic prescribing are more complex, having as much or more to do with patient and physician expectations as with physicians’ diagnostic skills.^6-8 These studies are limited to describing perceptions of behavior rather than actual behavior, because of their use of interview and focus group data. Consequently, we do not know what actually happens during outpatient visits for ART infections that leads to antibiotic prescribing.

Two studies by Stivers^9,10 underscore the importance of directly observing what transpires during encounters with pediatric ART infection patients. Stivers’ examination of videotaped visits found that, in some cases, parental pressure for antibiotics influenced the physician’s decision to prescribe. This finding has not been replicated, however, in family practice settings, where both adults and children are seen. We used direct observation of outpatient visits to family physicians for ART infections to analyze the effects of physician-patient communication on unnecessary antibiotic prescribing. By understanding the ways these communication patterns influence prescribing behavior, practicing family physicians can develop strategies to deliver more appropriate care for ART infections.

Methods

These data were collected as part of the Prevention & Competing Demands in Primary Care Study, which was an in-depth observational study begun in October 1996 and completed in August 1999 that examined the organizational and clinical structures and process of community-based family practices. Each of 18 purposefully selected practices was studied using a multimethod comparative case study design that involved extensive direct observation of clinical encounters and office systems by field researchers who spent 4 weeks or more in each practice. Field researchers directly observed and dictated descriptions of approximately 30 patient encounters with each of the more than 50 clinicians. Details of the sampling and data collection are available elsewhere in this issue.¹¹

Data Analysis and Interpretation

Encounters related to ART infection were identified in the database using search terms for symptoms and diagnoses including: sore throat, runny nose, congestion, cough, drainage, postnasal drainage, earache, cold, upper respiratory infection, pharyngitis, sinusitis, bronchitis, and otitis. ART infection was identified as the principal or associated diagnosis in 316 outpatient visits of a total of 1637 observed encounters; 298 had sufficiently rich data for analysis. The encounters were first coded for antibiotic use or nonuse.

Before any qualitative analysis began, visits during which antibiotics were prescribed were further characterized as appropriate or unnecessary according to guidelines by the Centers for Disease Control and Prevention (CDC) for judicious use of antibiotics for children¹² and adults.Table W1¹³* Two family physicians assigned appropriate/unnecessary codes independently. Inter-rater reliability was good (k=0.71). All disagreements were resolved by discussion.

Subsequently, the text for each outpatient visit was read independently by 2 family physicians, a medical anthropologist, a nurse, and a communication specialist. This research team discussed individual encounters as a group to identify emerging patterns of physician-patient interaction.

Results

Women made up 59% of the study sample; 64% were 16 years or older (and classified as adults). Antibiotics were prescribed in 204 of the 298 ART infection encounters (68%). Antibiotic use was unnecessary according to the CDC guidelines in 164 of these (80%). Adults were more likely than children to receive unnecessary antibiotics Table 1.

Our analysis identified 6 different types of patient behaviors that advocated for medication, particularly antibioticsTable 2. These behaviors fell into 3 broad categories: explicit requests, presentation of chief complaint, and appeals to lifeworld circumstances. Multiple pressures were noted in many encounters.

While patients occasionally made direct requests for antibiotics, they much more frequently positioned themselves indirectly for receiving antibiotic treatment by the way they presented the chief complaint. Four distinct approaches were identified: symptoms only,⁹ candidate diagnosis,⁹ implied candidate diagnosis,⁹ and portraying the severity and inability to shake the illness.

A second category of indirect approach used life-world circumstances¹⁰ (eg, an upcoming family vacation) or a past history with successful antibiotic treatment to formulate appeals for antibiotics in the current encounter. In those cases in which antibiotics were clearly unnecessary, physicians often rationalized their prescribing practices by finding symptoms or assigning diagnoses to justify antibiotic use. Each of these patient pressures, as well as the physician-rationalizing behavior, is illustrated with sample visits. The samples are taken directly from transcribed field notes, but the names have been altered to protect the identity of patients and clinicians.

Explicit Request

Explicit requests for antibiotics were observed in only 6% of cases (n=15). For example:

Claire asked the patient, “How are you doing?” and she said, “Well, I’m coughing up phlegm, I ache and I have chills and a sore throat.” Claire said, “You have bronchoconstriction, and 3 times a day, if you need to, you should use proventil.” The patient asked if she could have an antibiotic for her cold; cephalexin has worked in the past. Claire said that she would get her cephalexin and also some samples of an inhaler.

Presentation of the Chief Complaint

Patients frequently put pressure on the physician for treatment during the presentation of the chief complaint, the exception being the symptoms-only presentation. This is different from the other indirect pressures, which usually occurred during different parts of the medical encounter.

Symptoms-only presentation (eg, “I have a cough and a sore throat.”) In the symptoms-only approach (n=15), the patient reports his or her symptoms with little embellishment. This approach does not pressure physicians for antibiotic treatment.

Candidate diagnosis (eg, “I think I’ve got strep throat.”) In contrast, patients also presented their chief complaint to the physician by offering a candidate diagnosis (n=18). As shown in the following example, the patient responds by offering a diagnosis. This is a way of indirectly advocating for antibiotic treatment.

A 21-year-old white woman went to see Dr. Maxwell with an acute problem of congestion. Dr Maxwell said, “Well, how are you doing?” The patient said, “It sounds like bronchitis. It started about 4 days ago.”

Implied candidate diagnosis (eg, “My throat hurts; it’s red; and it has white spots.”) The implied candidate diagnosis is a hybrid of the symptoms-only and the candidate diagnosis approaches (n=48). When presenting their chief complaint, patients reported very specific symptoms that indexed a particular diagnosis. For example:

A 29-year-old woman went to see Dr Redmond with swollen glands, congestion, and white spots on her throat. When Dr Redmond and I went into the examination room, the patient had a pink paper top on, and Dr Redmond told her that her throat culture was negative.

The patient reports that she has swollen glands, congestion, and white spots on her throat. The symptoms specifically index a particular condition (strep throat). The patient’s presentation of symptoms clearly implies a diagnosis of strep throat, and the physician ordered a strep culture before seeing the patient.

Candidate diagnoses and implied candidate diagnoses delicately assert that the nature of the patient’s problem is already known. The reason for the medical visit is to seek treatment for the patient’s already known condition. When candidate and implied candidate diagnoses point to a condition the patient believes to be treatable (eg, bronchitis, strep throat, ear infection), this way of presenting the chief complaint looks directly ahead to a treatment involving a prescription for an antibiotic and thus indirectly pressures the physician to prescribe one.

Portraying the severity of one’s illness (eg, “I can’t shake this, Doc.”) The most common strategy was for patients to subtly pressure physicians for medication by portraying the severity of their condition and their inability to shake the illness on their own (n=99). For example:

The patient was sitting up on the table, and right away he told Dr Lamont, “I just can’t shake it. I feel like the back of my throat has raw hamburger hanging in it.” Dr Lamont checked the patient’s throat well, and the patient said, “This has lasted 4 days and it has been getting worse today.” Dr Lamont checked the patient’s ears, glands, and lungs. “I’m going to give you a shot of penicillin, slow release. It’s some kind of an infection. It may be a virus.”

Portraying the severity of one’s illness may not in and of itself advocate for medication; however, portrayals of the severity of one’s condition were usually accompanied by other actions implicating the need for medication. By opening the encounter with the announcement “I just can’t shake it,” the patient implies that he needs help in getting well. This subtly suggests the need for a prescription medication to alleviate his sore throat. At the end of this visit, the patient receives an antibiotic shot.

Appeals to Nonmedical Circumstances

Patients also used nonmedical circumstances to advocate for medication. These behaviors tended to occur after the problem presentation in the encounter and either centered on some important event, such as a big examination or a trip out of town (n=16), or focused on a previous positive experience with antibiotics for themselves or a family member (n=39).

Appealing to life-world circumstance (eg, “But I’m going to Disney World.”) This patient uses an upcoming trip to make an appeal to the clinician to prescribe medication:

The patient is a 33-year-old man coming in with an acute problem of a sore throat. The patient stated that he had been trying to manage this on his own, but he was taking his wife and 2 children to Disney World at the end of the week and was becoming worried that he was still going be sick and not able to enjoy a trip that they had saved so long for. He also told Dr Liam: “I know we’ll just get to Florida, and the kids will get sick, and then we’ll all be sick again. Dr. Liam said, “Well, we can have you bring them in, but then we’d be treating them for something that they haven’t gotten. Let me think about this a bit.” He does the rapid strep test, and it’s negative. Dr Liam reported the news of a negative strep test and said, “Many times we get a 50% false-negative, so I’m gonna go ahead and put you on an antibiotic and see if we can’t get you feeling better.” With this the patient said, “Well, what do you think I should do about my kids?” Dr. Liam asked if the kids were seen in this clinic, and the patient responded that they had never been seen there before. Dr Liam said, “Well, I’ll go ahead and give you a script for erythromycin in case these kids get sick down in Florida. If they do, go ahead and give them the medicine; if they don’t, throw away the prescription.”

This case is interesting because once he is treated with an antibiotic, the patient uses the same argument to make an appeal for antibiotics for his children (both of whom have never been seen by this physician).

Previous positive experience with antibiotics (eg, “I got an antibiotic for this before.”). Patients also appealed to other nonmedical contingencies to advocate for antibiotic treatment. For example:

Our next patient was a 51-year-old woman complaining of a cold and laryngitis. The doctor asked the patient about her symptoms. The patient responded, saying that she had been taking medication during the end of December for the same symptoms; they had cleared after taking antibiotics, and now they were back again.

The patient indirectly makes an appeal for antibiotic treatment by stating that she received antibiotics in the past for the same symptoms that she has now.

Patients used several variations of this approach. These included stating that another physician prescribed an antibiotic for this illness in the past; that others in the family are sick with an illness for which they received antibiotics; that they have a history of illness for which antibiotics are regularly prescribed; and that they were recently taking an antibiotic for an illness that has not improved (with the idea that an antibiotic is needed again).

Effectiveness of Patient Pressures

Physicians prescribed an antibiotic unnecessarily in 80% of the encounters in which some patient pressure was observed. They seemed able to resist certain types of pressures better than others. Unnecessary antibiotics were prescribed for a smaller percentage of implied candidate diagnoses and candidate diagnoses and for a larger percentage of direct patient requests and previous positive experiences with antibiotics Table 3.

Physicians’ Response to Prescribing an Unnecessary Antibiotic

When physicians prescribed an antibiotic unnecessarily, they often rationalized this practice by finding symptoms or assigning diagnoses that, to them, justified prescribing antibiotics. Physicians used various rationales, such as red throat or enlarged tonsils; severe, prolonged, or productive cough; yellow or green mucus; sinus tenderness on palpation; associated chronic disease; history of previous infection; and the desire to “cover” the patient “just in case.” None of these rationales are supported by evidence as correlating with bacterial infection. An example of this kind of rationalization follows:

This is a 20-year-old woman coming in with a complaint of a worsening cough. She said that her chest had a prickly, burning sensation, and it hurt to breathe. Dr Hart asked if she was able to bring anything up. She said that she really couldn’t. It was just a really terrible barky cough. Following the physician examination, the physician told the patient that her lungs basically sounded clear, but she could certainly hear some rough bronchial sounds. With this, she said. “What I think is happening here with your cold is that it is probably ending up in a bronchitis-type situation, and probably what we should do is put you on an antibiotic and order a decongestant.”

Discussion

This investigation, in agreement with the pediatric studies of Stivers,^9,10 suggests that the connection between patient diagnosis and physician prescribing is highly complex, involving patient presentation and physician-patient communication as much as, if not more than, physician diagnostic skills. Also, these data suggest that physicians are better able to resist patient pressures that are framed in medical terms such as candidate diagnoses or implied candidate diagnoses but are much less able to resist pressures that are not medicalized, such as portraying severity of illness and use of life-world circumstances. Thus, it is not surprising that past interventions designed to increase physician knowledge regarding when to prescribe antibiotics have had limited success.^14,15 Physicians appear to be trying to maximize patient satisfaction by giving antibiotic-seeking patients what they want. Our findings show the need to modify current thinking about the diagnostic and treatment process to reduce the use of antibiotics. Rather than thinking of these processes as physician controlled, the powerful role patients play in this interaction must be considered.

Our study has important implications for future research. From a methodologic standpoint, our findings illustrate the importance of qualitative evaluation of directly observed medical encounters. The patterns of patient behavior observed could not have been discerned using survey, interview, or focus group data.

Limitations

Because these data were collected by field researchers who were unaware that ART infection would be a focus of our study, it is possible that there were other patient symptoms and behavior related to ART infection, as well as physician behaviors related to antibiotic prescribing, that were not recorded. The data were sufficiently rich, however, to easily and reliably apply the CDC guidelines for appropriate use of antibiotics. Any unrecorded behaviors might add to, but not substantially change, our conclusions that patients indirectly pressure their physicians for treatment, and physicians respond by giving antibiotics. Studies using videotaped encounters might uncover such additional important patient and physician behaviors. Since the patient population studied was limited to a single midwestern state, it is possible that other populations with a different ethnic or racial mix might behave differently. Future research in this area should attempt to include such populations. Finally, too few encounters per physician were observed in this study to evaluate whether particular physicians were high or low prescribers (such a pattern has been reported by De Sutter and colleagues¹⁶).

Conclusions

Physicians should be educated about the subtle approaches patients use to pressure them for antibiotic treatment and should be shown techniques for responding to these pressures without prescribing antibiotics unnecessarily. Our findings also suggest the need to increase patients’ awareness both of the dangers and lack of effectiveness of using antibiotics for ART infections and of the amount of influence that patients have on antibiotic prescribing. Macfarlane and coworkers17 have shown that use of patient education materials reduces visits for ART infection. Additional approaches to decreasing patient pressure for antibiotic prescriptions are needed to diminish antibiotic overuse and its public health consequences.

Acknowledgments

Our study was funded by the Agency for Healthcare Research and Quality Grant R01 HS08776. Dr Scott is a postdoctoral fellow supported by the Health Resources and Services Administration (HRSA) PE1011 and the Agency for Healthcare Research and Quality (AHRQ) HS09788. Analysis of these data was supported by a Research Center grant from the American Academy of Family Physicians (Center for Research in Family Practice and Primary Care). Drs Jaen and Crabtree are associated with the Center for Research in Family Practicer and Primary Care, Cleveland, New Brunswick, Allentown. and San Antonio. The authors wish to thank the family physicians of Nebraska who were willing to open their practices to us. We also thank Kurt C. Stange, MD, PhD, for his thoughtful comments on drafts of this manuscript.

Related Resources

U.S. Centers for Disease Control and Prevention—Promoting Appropriate Antibiotic Use in the Community http://www.cdc.gov/antibioticresistance/tools.htm

A vast resource of of patient education resources.