John M. Hickner, MD, MS

I pride myself on judicious use of antibiotics for acute upper respiratory infections (URIs). The easy patients to treat are those with classic viral symptoms of clear rhinorrhea, nonproductive cough, and a mild to moderate illness of short duration. Most of these patients readily accept my reassurance and advice for symptomatic treatment.

Patients with acute bronchitis are a bit tougher to treat. They often have productive cough and are moderately ill for a longer duration. Even though antibiotics are of marginal, if any, benefit for acute bronchitis,¹ approximately 30% of my patients receive an antibiotic prescription despite my best efforts to the contrary. I usually resort to the popular last-ditch tactic of the backup prescription. If 50% of my patients fill those backup prescriptions as Couchman and colleagues² found, my actual prescribing rate for acute bronchitis is only 15%—not bad for this largely viral syndrome.

The good news is that the anti-antibiotic scuffle with patients may be happening less often. The public campaign by the Centers for Disease Control and Prevention to reduce inappropriate use of antibiotics appears to be reaching some of my patients. They are more likely to accept my advice for symptomatic treatment than they were 5 years ago. However, there is still a great need for educating patients and physicians about appropriate use of antibiotics for respiratory tract infections, as illustrated by the survey of college students by Zoorob and colleagues³ in this issue of JFP. When confronted with scenarios typical of viral URIs, 50% of those bright young adults would seek medical care and an antibiotic prescription.

Antibiotics for Acute Sinusitis

Acute sinusitis is a horse (or discharge, maybe?) of a different color. I used to think that sinusitis was the easy one to handle. Cheek pain plus green discharge equals antibiotic. The patient goes away satisfied, and I go on to the next coughing patient. Not so fast. This last port of refuge for antibiotic advocates is crumbling, too. Consider the following sobering facts:

Sinusitis rarely occurs in isolation and is most often accompanied by nasal cavity inflammation, resulting in the new designation rhinosinusitis.⁴

Most cases of rhinosinusitis are caused by viruses. Maxillary sinus radiographs of young adults with typical viral URIs showed mucosal abnormalities in 39% of cases on the seventh day of illness,⁵ and computed tomography (CT) scans were abnormal in 87% of similar cases.⁶

When based on signs and symptoms, the diagnosis of acute sinusitis is correct in approximately 50% or less of cases.^7-11 We probably are not this accurate in routine practice.

Randomized clinical trials of antibiotic treatment of rhinosinusitis have shown no effect when the diagnosis was based on clinical findings alone¹² or on clinical findings confirmed by plain radiographs.¹³

Despite the negative results of these randomized clinical trials, more than 90% of patients with a diagnosis of sinusitis by primary care physicians receive an antibiotic prescription.^13,14

Antibiotics have little effect on the course of rhinosinusitis, because the clinical diagnosis of bacterial sinusitis is so difficult. The signs and symptoms of viral infection of the paranasal sinuses mimic those of bacterial infection. Several investigators have attempted to identify clinical findings specific to bacterial sinusitis using a high-quality reference standard (sinus puncture and aspiration of purulent secretions, positive bacterial culture of aspirated secretions, or positive CT scan of the sinuses).^7,8,11 Maxillary facial pain, tooth pain, and purulent nasal discharge (ie, white, not green) are most discriminating, but even with these seemingly specific findings the ability to diagnose bacterial sinusitis accurately is poor. Because viral rhinosinusitis and rhinosinusitis with minimal bacterial suprainfection are so much more common than significant bacterial sinusitis, the few truly antibiotic-responsive bacterial infections are diluted out in clinical trials. We therefore see no effect of antibiotic treatment.

When Are Antibiotics Effective?

Antibiotics may be effective in some cases of acute sinusitis, but which ones? The dilemma is that we do not have a practical clinical test for ferreting out the few patients with sinusitis-like illness who would truly benefit from antibiotic treatment. (Most patients would prefer a trial of an antibiotic to sinus puncture for definitive diagnosis.) A single randomized trial of antibiotic treatment of rhinosinusitis has shown a modest benefit of such treatment when the diagnosis was made on the basis of a positive CT scan of the sinuses¹⁶ (56% of patients treated with placebo, 82% of those treated with penicillin, and 89% of those treated with amoxicillin were substantially better on day 10 of treatment). But CT sinus scans are expensive and not readily available in outpatient practice.

Until a better test is discovered and because most cases of sinusitis resolve without antibiotic treatment, providing reassurance, analgesics, and perhaps decongestants for symptomatic relief is the preferred treatment for mild to moderate cases of less than 7 days’ duration. Patients with typical sinusitis symptoms and more severe facial pain probably do benefit from early antibiotic treatment.¹⁷

However, I find that watchful waiting is ineffective for patients who complain of recurring sinus infections. These patients present within the first few days of illness, insist that they have a sinus infection just like the last one, and want to “catch it before it gets too bad.” This argument sounds sensible enough, and I find it hard to refuse the request. Are these sinusitis-prone patients more likely to have a bacterial infection than patients with similar complaints and no past history of sinusitis?

Antibiotics for Sinusitis-Prone Patients

Alho and coworkers¹⁸ present results of a rare attempt to explore this question in this issue of JFP. They recruited 23 adults who claimed to have suffered from recurrent sinus infections (sinusitis-prone group) and 25 who did not, all of whom had self-diagnosed colds of 48 to 96 hours’ duration. They compared the clinical courses, CT findings, and viral and bacterial cultures of nasal secretions obtained by nasoendoscopy. The sinusitis-prone group (defined as at least 2 episodes in the past year) had significantly more facial pain and sinusitis-like changes on CT scan (65% vs 36%). However, the sinusitis-prone group was just as likely to have a positive nasal viral culture as the control group (70% vs 64%). Paradoxically, the sinusitis-prone group was less likely to have a positive culture for pathogenic bacteria on nasoendoscopy (9% vs 40%), a finding the authors claim correlates with bacterial growth in the sinuses.

Alho and colleagues suggest that these sinusitis-prone patients usually have viral respiratory tract infections, just like the non-sinusitis–prone patients. They believe that because the sinusitis-prone patients tend to have more severe facial pain from their colds, they are more likely to seek care, and they are therefore more likely to be prescribed antibiotics inappropriately for this viral infection. My equally plausible explanation is that sinusitis-prone patients are truly more likely to develop complicating bacterial sinus infection early in the course of their illnesses, and they should be treated more aggressively with early antibiotic treatment. We cannot know the correct explanation because Alho and coworkers did not randomly assign subjects with respect to antibiotic treatment and because no diagnostic sinus punctures were done to determine who truly had a bacterial sinus infection. In a study of clinical predictors Hansen and colleagues11 found no difference in bacterial sinusitis in patients with and without a history of sinusitis, a finding that supports the interpretation of Alho and coworkers.

Time to Change Prescription Habits?

In light of the findings of Alho and colleagues, should I change my practice of using antibiotics for sinusitis-prone patients? Their findings and those of Hansen and coworkers are intriguing enough for me to think twice before reflexively prescribing an antibiotic. I will delve more carefully into the patient’s history to convince myself that the past episodes sound more like significant sinus infections than allergy or simple URIs. In doubtful cases, a plain Waters sinus radiograph, if normal, effectively rules out bacterial infection (negative predictive value of approximately 90%, meaning that 90% of symptomatic patients with a normal radiograph do not have sinusitis). A positive radiograph, however, does not rule in bacterial infection, and in difficult cases a sinus CT may be helpful. Otherwise, it is necessary to go back to finding common ground with the patient and negotiating treatment. Clearly, a next step on the rhinosinusitis research agenda is a randomized clinical trial of antibiotic treatment for sinusitis-prone patients. Ultimately, an accurate, inexpensive, and convenient diagnostic test is needed before we can base antibiotic treatment of sinusitis-like illness on firm scientific grounds.

I pride myself on judicious use of antibiotics for acute upper respiratory infections (URIs). The easy patients to treat are those with classic viral symptoms of clear rhinorrhea, nonproductive cough, and a mild to moderate illness of short duration. Most of these patients readily accept my reassurance and advice for symptomatic treatment.

Patients with acute bronchitis are a bit tougher to treat. They often have productive cough and are moderately ill for a longer duration. Even though antibiotics are of marginal, if any, benefit for acute bronchitis,¹ approximately 30% of my patients receive an antibiotic prescription despite my best efforts to the contrary. I usually resort to the popular last-ditch tactic of the backup prescription. If 50% of my patients fill those backup prescriptions as Couchman and colleagues² found, my actual prescribing rate for acute bronchitis is only 15%—not bad for this largely viral syndrome.

The good news is that the anti-antibiotic scuffle with patients may be happening less often. The public campaign by the Centers for Disease Control and Prevention to reduce inappropriate use of antibiotics appears to be reaching some of my patients. They are more likely to accept my advice for symptomatic treatment than they were 5 years ago. However, there is still a great need for educating patients and physicians about appropriate use of antibiotics for respiratory tract infections, as illustrated by the survey of college students by Zoorob and colleagues³ in this issue of JFP. When confronted with scenarios typical of viral URIs, 50% of those bright young adults would seek medical care and an antibiotic prescription.

Antibiotics for Acute Sinusitis

Acute sinusitis is a horse (or discharge, maybe?) of a different color. I used to think that sinusitis was the easy one to handle. Cheek pain plus green discharge equals antibiotic. The patient goes away satisfied, and I go on to the next coughing patient. Not so fast. This last port of refuge for antibiotic advocates is crumbling, too. Consider the following sobering facts:

Sinusitis rarely occurs in isolation and is most often accompanied by nasal cavity inflammation, resulting in the new designation rhinosinusitis.⁴

Most cases of rhinosinusitis are caused by viruses. Maxillary sinus radiographs of young adults with typical viral URIs showed mucosal abnormalities in 39% of cases on the seventh day of illness,⁵ and computed tomography (CT) scans were abnormal in 87% of similar cases.⁶

When based on signs and symptoms, the diagnosis of acute sinusitis is correct in approximately 50% or less of cases.^7-11 We probably are not this accurate in routine practice.

Randomized clinical trials of antibiotic treatment of rhinosinusitis have shown no effect when the diagnosis was based on clinical findings alone¹² or on clinical findings confirmed by plain radiographs.¹³

Despite the negative results of these randomized clinical trials, more than 90% of patients with a diagnosis of sinusitis by primary care physicians receive an antibiotic prescription.^13,14

Antibiotics have little effect on the course of rhinosinusitis, because the clinical diagnosis of bacterial sinusitis is so difficult. The signs and symptoms of viral infection of the paranasal sinuses mimic those of bacterial infection. Several investigators have attempted to identify clinical findings specific to bacterial sinusitis using a high-quality reference standard (sinus puncture and aspiration of purulent secretions, positive bacterial culture of aspirated secretions, or positive CT scan of the sinuses).^7,8,11 Maxillary facial pain, tooth pain, and purulent nasal discharge (ie, white, not green) are most discriminating, but even with these seemingly specific findings the ability to diagnose bacterial sinusitis accurately is poor. Because viral rhinosinusitis and rhinosinusitis with minimal bacterial suprainfection are so much more common than significant bacterial sinusitis, the few truly antibiotic-responsive bacterial infections are diluted out in clinical trials. We therefore see no effect of antibiotic treatment.

When Are Antibiotics Effective?

Antibiotics may be effective in some cases of acute sinusitis, but which ones? The dilemma is that we do not have a practical clinical test for ferreting out the few patients with sinusitis-like illness who would truly benefit from antibiotic treatment. (Most patients would prefer a trial of an antibiotic to sinus puncture for definitive diagnosis.) A single randomized trial of antibiotic treatment of rhinosinusitis has shown a modest benefit of such treatment when the diagnosis was made on the basis of a positive CT scan of the sinuses¹⁶ (56% of patients treated with placebo, 82% of those treated with penicillin, and 89% of those treated with amoxicillin were substantially better on day 10 of treatment). But CT sinus scans are expensive and not readily available in outpatient practice.

Until a better test is discovered and because most cases of sinusitis resolve without antibiotic treatment, providing reassurance, analgesics, and perhaps decongestants for symptomatic relief is the preferred treatment for mild to moderate cases of less than 7 days’ duration. Patients with typical sinusitis symptoms and more severe facial pain probably do benefit from early antibiotic treatment.¹⁷

However, I find that watchful waiting is ineffective for patients who complain of recurring sinus infections. These patients present within the first few days of illness, insist that they have a sinus infection just like the last one, and want to “catch it before it gets too bad.” This argument sounds sensible enough, and I find it hard to refuse the request. Are these sinusitis-prone patients more likely to have a bacterial infection than patients with similar complaints and no past history of sinusitis?

Antibiotics for Sinusitis-Prone Patients

Alho and coworkers¹⁸ present results of a rare attempt to explore this question in this issue of JFP. They recruited 23 adults who claimed to have suffered from recurrent sinus infections (sinusitis-prone group) and 25 who did not, all of whom had self-diagnosed colds of 48 to 96 hours’ duration. They compared the clinical courses, CT findings, and viral and bacterial cultures of nasal secretions obtained by nasoendoscopy. The sinusitis-prone group (defined as at least 2 episodes in the past year) had significantly more facial pain and sinusitis-like changes on CT scan (65% vs 36%). However, the sinusitis-prone group was just as likely to have a positive nasal viral culture as the control group (70% vs 64%). Paradoxically, the sinusitis-prone group was less likely to have a positive culture for pathogenic bacteria on nasoendoscopy (9% vs 40%), a finding the authors claim correlates with bacterial growth in the sinuses.

Alho and colleagues suggest that these sinusitis-prone patients usually have viral respiratory tract infections, just like the non-sinusitis–prone patients. They believe that because the sinusitis-prone patients tend to have more severe facial pain from their colds, they are more likely to seek care, and they are therefore more likely to be prescribed antibiotics inappropriately for this viral infection. My equally plausible explanation is that sinusitis-prone patients are truly more likely to develop complicating bacterial sinus infection early in the course of their illnesses, and they should be treated more aggressively with early antibiotic treatment. We cannot know the correct explanation because Alho and coworkers did not randomly assign subjects with respect to antibiotic treatment and because no diagnostic sinus punctures were done to determine who truly had a bacterial sinus infection. In a study of clinical predictors Hansen and colleagues11 found no difference in bacterial sinusitis in patients with and without a history of sinusitis, a finding that supports the interpretation of Alho and coworkers.

Time to Change Prescription Habits?

In light of the findings of Alho and colleagues, should I change my practice of using antibiotics for sinusitis-prone patients? Their findings and those of Hansen and coworkers are intriguing enough for me to think twice before reflexively prescribing an antibiotic. I will delve more carefully into the patient’s history to convince myself that the past episodes sound more like significant sinus infections than allergy or simple URIs. In doubtful cases, a plain Waters sinus radiograph, if normal, effectively rules out bacterial infection (negative predictive value of approximately 90%, meaning that 90% of symptomatic patients with a normal radiograph do not have sinusitis). A positive radiograph, however, does not rule in bacterial infection, and in difficult cases a sinus CT may be helpful. Otherwise, it is necessary to go back to finding common ground with the patient and negotiating treatment. Clearly, a next step on the rhinosinusitis research agenda is a randomized clinical trial of antibiotic treatment for sinusitis-prone patients. Ultimately, an accurate, inexpensive, and convenient diagnostic test is needed before we can base antibiotic treatment of sinusitis-like illness on firm scientific grounds.

I pride myself on judicious use of antibiotics for acute upper respiratory infections (URIs). The easy patients to treat are those with classic viral symptoms of clear rhinorrhea, nonproductive cough, and a mild to moderate illness of short duration. Most of these patients readily accept my reassurance and advice for symptomatic treatment.

Patients with acute bronchitis are a bit tougher to treat. They often have productive cough and are moderately ill for a longer duration. Even though antibiotics are of marginal, if any, benefit for acute bronchitis,¹ approximately 30% of my patients receive an antibiotic prescription despite my best efforts to the contrary. I usually resort to the popular last-ditch tactic of the backup prescription. If 50% of my patients fill those backup prescriptions as Couchman and colleagues² found, my actual prescribing rate for acute bronchitis is only 15%—not bad for this largely viral syndrome.

The good news is that the anti-antibiotic scuffle with patients may be happening less often. The public campaign by the Centers for Disease Control and Prevention to reduce inappropriate use of antibiotics appears to be reaching some of my patients. They are more likely to accept my advice for symptomatic treatment than they were 5 years ago. However, there is still a great need for educating patients and physicians about appropriate use of antibiotics for respiratory tract infections, as illustrated by the survey of college students by Zoorob and colleagues³ in this issue of JFP. When confronted with scenarios typical of viral URIs, 50% of those bright young adults would seek medical care and an antibiotic prescription.

Antibiotics for Acute Sinusitis

Acute sinusitis is a horse (or discharge, maybe?) of a different color. I used to think that sinusitis was the easy one to handle. Cheek pain plus green discharge equals antibiotic. The patient goes away satisfied, and I go on to the next coughing patient. Not so fast. This last port of refuge for antibiotic advocates is crumbling, too. Consider the following sobering facts:

Sinusitis rarely occurs in isolation and is most often accompanied by nasal cavity inflammation, resulting in the new designation rhinosinusitis.⁴

Most cases of rhinosinusitis are caused by viruses. Maxillary sinus radiographs of young adults with typical viral URIs showed mucosal abnormalities in 39% of cases on the seventh day of illness,⁵ and computed tomography (CT) scans were abnormal in 87% of similar cases.⁶

When based on signs and symptoms, the diagnosis of acute sinusitis is correct in approximately 50% or less of cases.^7-11 We probably are not this accurate in routine practice.

Randomized clinical trials of antibiotic treatment of rhinosinusitis have shown no effect when the diagnosis was based on clinical findings alone¹² or on clinical findings confirmed by plain radiographs.¹³

Despite the negative results of these randomized clinical trials, more than 90% of patients with a diagnosis of sinusitis by primary care physicians receive an antibiotic prescription.^13,14

Antibiotics have little effect on the course of rhinosinusitis, because the clinical diagnosis of bacterial sinusitis is so difficult. The signs and symptoms of viral infection of the paranasal sinuses mimic those of bacterial infection. Several investigators have attempted to identify clinical findings specific to bacterial sinusitis using a high-quality reference standard (sinus puncture and aspiration of purulent secretions, positive bacterial culture of aspirated secretions, or positive CT scan of the sinuses).^7,8,11 Maxillary facial pain, tooth pain, and purulent nasal discharge (ie, white, not green) are most discriminating, but even with these seemingly specific findings the ability to diagnose bacterial sinusitis accurately is poor. Because viral rhinosinusitis and rhinosinusitis with minimal bacterial suprainfection are so much more common than significant bacterial sinusitis, the few truly antibiotic-responsive bacterial infections are diluted out in clinical trials. We therefore see no effect of antibiotic treatment.

When Are Antibiotics Effective?

Antibiotics may be effective in some cases of acute sinusitis, but which ones? The dilemma is that we do not have a practical clinical test for ferreting out the few patients with sinusitis-like illness who would truly benefit from antibiotic treatment. (Most patients would prefer a trial of an antibiotic to sinus puncture for definitive diagnosis.) A single randomized trial of antibiotic treatment of rhinosinusitis has shown a modest benefit of such treatment when the diagnosis was made on the basis of a positive CT scan of the sinuses¹⁶ (56% of patients treated with placebo, 82% of those treated with penicillin, and 89% of those treated with amoxicillin were substantially better on day 10 of treatment). But CT sinus scans are expensive and not readily available in outpatient practice.

Until a better test is discovered and because most cases of sinusitis resolve without antibiotic treatment, providing reassurance, analgesics, and perhaps decongestants for symptomatic relief is the preferred treatment for mild to moderate cases of less than 7 days’ duration. Patients with typical sinusitis symptoms and more severe facial pain probably do benefit from early antibiotic treatment.¹⁷

However, I find that watchful waiting is ineffective for patients who complain of recurring sinus infections. These patients present within the first few days of illness, insist that they have a sinus infection just like the last one, and want to “catch it before it gets too bad.” This argument sounds sensible enough, and I find it hard to refuse the request. Are these sinusitis-prone patients more likely to have a bacterial infection than patients with similar complaints and no past history of sinusitis?

Antibiotics for Sinusitis-Prone Patients

Alho and coworkers¹⁸ present results of a rare attempt to explore this question in this issue of JFP. They recruited 23 adults who claimed to have suffered from recurrent sinus infections (sinusitis-prone group) and 25 who did not, all of whom had self-diagnosed colds of 48 to 96 hours’ duration. They compared the clinical courses, CT findings, and viral and bacterial cultures of nasal secretions obtained by nasoendoscopy. The sinusitis-prone group (defined as at least 2 episodes in the past year) had significantly more facial pain and sinusitis-like changes on CT scan (65% vs 36%). However, the sinusitis-prone group was just as likely to have a positive nasal viral culture as the control group (70% vs 64%). Paradoxically, the sinusitis-prone group was less likely to have a positive culture for pathogenic bacteria on nasoendoscopy (9% vs 40%), a finding the authors claim correlates with bacterial growth in the sinuses.

Alho and colleagues suggest that these sinusitis-prone patients usually have viral respiratory tract infections, just like the non-sinusitis–prone patients. They believe that because the sinusitis-prone patients tend to have more severe facial pain from their colds, they are more likely to seek care, and they are therefore more likely to be prescribed antibiotics inappropriately for this viral infection. My equally plausible explanation is that sinusitis-prone patients are truly more likely to develop complicating bacterial sinus infection early in the course of their illnesses, and they should be treated more aggressively with early antibiotic treatment. We cannot know the correct explanation because Alho and coworkers did not randomly assign subjects with respect to antibiotic treatment and because no diagnostic sinus punctures were done to determine who truly had a bacterial sinus infection. In a study of clinical predictors Hansen and colleagues11 found no difference in bacterial sinusitis in patients with and without a history of sinusitis, a finding that supports the interpretation of Alho and coworkers.

Time to Change Prescription Habits?

In light of the findings of Alho and colleagues, should I change my practice of using antibiotics for sinusitis-prone patients? Their findings and those of Hansen and coworkers are intriguing enough for me to think twice before reflexively prescribing an antibiotic. I will delve more carefully into the patient’s history to convince myself that the past episodes sound more like significant sinus infections than allergy or simple URIs. In doubtful cases, a plain Waters sinus radiograph, if normal, effectively rules out bacterial infection (negative predictive value of approximately 90%, meaning that 90% of symptomatic patients with a normal radiograph do not have sinusitis). A positive radiograph, however, does not rule in bacterial infection, and in difficult cases a sinus CT may be helpful. Otherwise, it is necessary to go back to finding common ground with the patient and negotiating treatment. Clearly, a next step on the rhinosinusitis research agenda is a randomized clinical trial of antibiotic treatment for sinusitis-prone patients. Ultimately, an accurate, inexpensive, and convenient diagnostic test is needed before we can base antibiotic treatment of sinusitis-like illness on firm scientific grounds.

An increase in the prevalences of childhood and adolescent obesity in the United States has been documented in a series of national surveys conducted in the last 30 years. These include the National Health Evaluation surveys, NHES II (1963-1965) and NHES III (1966-1970), and the National Health and Nutrition Examination surveys, NHANES I (1971-1974), NHANES II (1976-1980), and NHANES III (1988-1994).^1-4 Obesity in childhood and adolescence may lead to disorders in adulthood, including hypertension, diabetes, coronary heart disease, and some cancers.⁵ The strongest connection between chronic disease and death is linked to adolescent rather than adult obesity.⁶ Between 1980 and 1994 there was a 6% increase in the absolute prevalence of obesity in US children.⁷ Data from NHANES III, the most recent national survey of weight and height in children and adults, showed that 14% of children aged 6 to 11 years and 12% of those aged 12 to 17 years were overweight.⁷

We noticed that the prevalence of obesity in our Michigan primary care practices appeared higher than that reported from national or state surveys. We confirmed our impression by measuring a consecutive sample of 6912 patients in 19 practices in the summer of 1996. Analysis of the adult data revealed that 54.5% of the men and 52.6% of the women were overweight, nearly double the prevalences cited in the 1993 Michigan Behavioral Risk Factor Survey (MBRFS): 29.5% for the men and 28.7% for the women.⁸ MBRFS obesity prevalences are slightly higher than contemporary national prevalences. Comparison with county-specific data from the MBRFS gave the same results. Adjusting the analysis for age, race, socioeconomic status, and chronic diseases associated with obesity such as diabetes and heart disease, did not significantly affect the results. This high prevalence of obesity in primary care patients as compared with population-based surveys has not been previously described, and the reason for it is unknown.

The purpose of our report is to describe the prevalences of obesity in of children and adolescents in our Michigan primary care practices and to compare those prevalences to national and state prevalences. We hypothesized that the children enrolled in our practices would have higher prevalences of obesity than national and state prevalences derived from surveys of community-dwelling populations.

Methods

Practices, Enrollment, and Measurement

We included 19 family practice offices of the Upper Peninsula Research Network (UPRNet) in our study. UPRNet is a family practice research network in rural northern Michigan. These practices are in rural locations, and community populations range from 300 to 15,000. Eleven of the offices are publicly funded community health centers. The populations served are typical of rural northern Michigan, with more than the state average of elderly and poor patients. To include urban and suburban children, we also asked 2 family practice clinics serving middle and lower socioeconomic groups in the Lansing area to participate.

In the spring and early summer of 1996, we invited consecutive patients aged 4 to 17 years visiting any of the participating practices for care to take part in our study, and we obtained informed consent. We chose spring and early summer to avoid those seasons when people wear heavy clothing and boots and to avoid enrolling vacationers who did not reside in the study areas. Patients with appointments specifically to see a dietitian, those known to be pregnant, and those unable to stand on a scale were excluded from the study.

Heights were measured using 1 of 2 methods, depending on the practice ’s preference; either with a measuring arm attached to a wall or with a measuring arm attached to a scale. All measuring arms were checked to be certain they were parallel to the floor. Weights were taken on balance beam scales calibrated less than a month before the study began. Patients removed their coats and shoes; no weight adjustments were made for clothing. Each office nursing staff reviewed the measuring procedures before participating in the study. Trained on-site study coordinators (office managers or nursing managers) supervised the data gathering process to ensure adherence.

The University Committee on Research Involving Human Subjects of Michigan State University approved our study protocol.

Definitions of Obesity in Children and Adolescents

There is no internationally accepted measure to define obesity in children. Methods for classifying childhood and adolescent obesity have been developed on the basis of comparison with growth charts (percentile weight for height by age and sex), percentile of triceps skin fold thickness, and percentile cutoffs for body mass index (BMI). BMI is emerging as the prefer red standard. The Expert Committee on Clinical Guidelines for Overweight in Adolescent Preventive Services⁸ recommends that children and adolescents with a BMI greater than the 85th percentile but less than the 95th be classified as “at risk of overweight” and those with a BMI greater than the 95th percentile be classified as “overweight.” We used that classification system.

For the comparisons, we followed the method used by Troiano.¹ For the baseline standard, he used the BMI distributions derived from the National Health Examination surveys (NHES II for children aged 6 to 11 years and NHES III for children aged 12 to 17 years). He calculated the percentage of children above the 85th and 95th percentile cutoffs of NHES for NHANES I, NHANES II, and NHANES III data. Similarly, we calculated the percentage of children and adolescents in the Michigan primary care practices with BMIs greater than the 85th and 95th percentiles as defined by NHES. For a contemporary regional comparison group, we obtained data from the Kuntzelman Fitness for Youth Program (KFYP), a 1997 statewide survey of Michigan schoolchildren. In that survey, a nonrandom sample of schoolchildren from 41 Michigan school districts were weighed and measured. Because 90% of our primary care patients were white, we used only the data on white children from the KFYP database. We summarized the KFYP data in the same way as the other surveys.

Results

A total of 993 children aged 4 to 17 years were measured and weighed in the Michigan primary care practices. There were 502 boys and 491 girls. Of these patients, 90% were white; 3% were Native American; 3% were African American; 1% was Hispanic; 1% was of another race; and 2% did not state their race. The Table 1 shows the percentage of children in the Michigan primary care practices above the 85th and 95th percentiles based on the NHES data and compared with NHANES data and Michigan schoolchildren (KFYP). The prevalences of obesity in the United States have increased over time, and the patients in our primary care practices have the highest prevalences of obesity in all age and sex groups when compared with the national surveys. For example, for boys aged 6 to 11, the percentage of those overweight in NHANES was 11%, whereas 20% of the Michigan primary care patients in this age group were overweight. In all groups, except for girls aged 6 to 11 years, prevalences of obesity of the primary care patients were greater than those of the Michigan schoolchildren in the KFYP survey.

Discussion

A larger than expected proportion of the children and adolescents in our primary care practices in Michigan are at risk for being overweight or are overweight. The prevalences of obesity in children and adolescents presenting for care in these practices are nearly 2 times greater than the national prevalences measured in NHANES II and NHANES III an d somewhat greater than a contemporary sample of Michigan schoolchildren. The absolute prevalences of obesity in the children and adolescents seen in our primary care practices were 3% to 9% higher than in the Michigan schoolchildren in all age-sex categories, except for girls aged 6 to 11 years.

It is not surprising that obesity is more prevalent in our sample than in the NHANES II and NHANES III surveys; NHANES data was gathered from 1976 to 1980 and 1988 to 1994, respectively, and obesity prevalences are steadily increasing. However, the reason that our patients are more obese than a contemporary sample of Michigan schoolchildren is unknown. The difference may be caused by systematic bias, as neither of the surveys was based on random sampling. However, in the analysis of our adult patients, the prevalence of obesity of patients was much greater than local community prevalences, even after adjusting for socioeconomic status and medical conditions known to be associated with obesity.⁸

The association between overweight and socio-economic status in children and adolescents is weak and not statistically significant.⁹ We speculate that obese children and adolescents have more physical and psychological complaints than healthy-weight individuals, causing them to visit their physicians more frequently. If they visit the practice more frequently, obese patients are more likely to be included in a survey of consecutive patients, thereby explaining the higher prevalences of obesity in primary care patients. Perhaps obese children or their parents are more worried about their health and therefore more likely to visit the physician for minor complaints. However, no association has been found between obesity and mental illness, although research regarding such an association is considered inconclusive at this time.¹⁰

Conclusions

The prevalences of obesity for children and adolescents presenting for care in Michigan primary care practices are higher than the state and national prevalences. Further studies using larger sample sizes, random sampling, and more settings are necessary to confirm or refute our findings. If this prevalence of obesity in primary care patients is confirmed, explanations for the differences should be explored.

Acknowledgments

We would like to thank the following people for their contributions to this study: Charles Kuntzelman, EdD, for providing us with the Michigan schoolchildren data; Nan Kreher, MEd, administration; Teresa Ettenhofer, administration and data analysis; and Kimber Gauthier, Jenny Folcik, and Celeste St. John, data entry. We would also like to thank the UPRNet and Michigan State University clinics that collected data. They are: Alpena Medical Arts, Alpena; Doctors Park Family Physicians, Escanaba; Dr Victoria Macki’s office, Newberry; OSF Medical Group, Gladstone and Escanaba; Gwinn Medical Center, Gwinn; Marquette Medical Clinic, Iron River; Burns Clinic, St Ignace; Alcona Health Center, Lincoln and Ossineke; East Jordan Family Health Center, East Jord an; Ewen Medical Clinic, Ewen; Northern Menominee Health Center, Spalding; Northern Michigan Health Services, Houghton Lake; Michigan State University-Family Practice at Clinical Center site, East Lansing; and Michigan State University-Family Practice at the Family Health Center, St. Lawrence Hospital, Lansing.

An increase in the prevalences of childhood and adolescent obesity in the United States has been documented in a series of national surveys conducted in the last 30 years. These include the National Health Evaluation surveys, NHES II (1963-1965) and NHES III (1966-1970), and the National Health and Nutrition Examination surveys, NHANES I (1971-1974), NHANES II (1976-1980), and NHANES III (1988-1994).^1-4 Obesity in childhood and adolescence may lead to disorders in adulthood, including hypertension, diabetes, coronary heart disease, and some cancers.⁵ The strongest connection between chronic disease and death is linked to adolescent rather than adult obesity.⁶ Between 1980 and 1994 there was a 6% increase in the absolute prevalence of obesity in US children.⁷ Data from NHANES III, the most recent national survey of weight and height in children and adults, showed that 14% of children aged 6 to 11 years and 12% of those aged 12 to 17 years were overweight.⁷

We noticed that the prevalence of obesity in our Michigan primary care practices appeared higher than that reported from national or state surveys. We confirmed our impression by measuring a consecutive sample of 6912 patients in 19 practices in the summer of 1996. Analysis of the adult data revealed that 54.5% of the men and 52.6% of the women were overweight, nearly double the prevalences cited in the 1993 Michigan Behavioral Risk Factor Survey (MBRFS): 29.5% for the men and 28.7% for the women.⁸ MBRFS obesity prevalences are slightly higher than contemporary national prevalences. Comparison with county-specific data from the MBRFS gave the same results. Adjusting the analysis for age, race, socioeconomic status, and chronic diseases associated with obesity such as diabetes and heart disease, did not significantly affect the results. This high prevalence of obesity in primary care patients as compared with population-based surveys has not been previously described, and the reason for it is unknown.

The purpose of our report is to describe the prevalences of obesity in of children and adolescents in our Michigan primary care practices and to compare those prevalences to national and state prevalences. We hypothesized that the children enrolled in our practices would have higher prevalences of obesity than national and state prevalences derived from surveys of community-dwelling populations.

Methods

Practices, Enrollment, and Measurement

We included 19 family practice offices of the Upper Peninsula Research Network (UPRNet) in our study. UPRNet is a family practice research network in rural northern Michigan. These practices are in rural locations, and community populations range from 300 to 15,000. Eleven of the offices are publicly funded community health centers. The populations served are typical of rural northern Michigan, with more than the state average of elderly and poor patients. To include urban and suburban children, we also asked 2 family practice clinics serving middle and lower socioeconomic groups in the Lansing area to participate.

In the spring and early summer of 1996, we invited consecutive patients aged 4 to 17 years visiting any of the participating practices for care to take part in our study, and we obtained informed consent. We chose spring and early summer to avoid those seasons when people wear heavy clothing and boots and to avoid enrolling vacationers who did not reside in the study areas. Patients with appointments specifically to see a dietitian, those known to be pregnant, and those unable to stand on a scale were excluded from the study.

Heights were measured using 1 of 2 methods, depending on the practice ’s preference; either with a measuring arm attached to a wall or with a measuring arm attached to a scale. All measuring arms were checked to be certain they were parallel to the floor. Weights were taken on balance beam scales calibrated less than a month before the study began. Patients removed their coats and shoes; no weight adjustments were made for clothing. Each office nursing staff reviewed the measuring procedures before participating in the study. Trained on-site study coordinators (office managers or nursing managers) supervised the data gathering process to ensure adherence.

The University Committee on Research Involving Human Subjects of Michigan State University approved our study protocol.

Definitions of Obesity in Children and Adolescents

There is no internationally accepted measure to define obesity in children. Methods for classifying childhood and adolescent obesity have been developed on the basis of comparison with growth charts (percentile weight for height by age and sex), percentile of triceps skin fold thickness, and percentile cutoffs for body mass index (BMI). BMI is emerging as the prefer red standard. The Expert Committee on Clinical Guidelines for Overweight in Adolescent Preventive Services⁸ recommends that children and adolescents with a BMI greater than the 85th percentile but less than the 95th be classified as “at risk of overweight” and those with a BMI greater than the 95th percentile be classified as “overweight.” We used that classification system.

For the comparisons, we followed the method used by Troiano.¹ For the baseline standard, he used the BMI distributions derived from the National Health Examination surveys (NHES II for children aged 6 to 11 years and NHES III for children aged 12 to 17 years). He calculated the percentage of children above the 85th and 95th percentile cutoffs of NHES for NHANES I, NHANES II, and NHANES III data. Similarly, we calculated the percentage of children and adolescents in the Michigan primary care practices with BMIs greater than the 85th and 95th percentiles as defined by NHES. For a contemporary regional comparison group, we obtained data from the Kuntzelman Fitness for Youth Program (KFYP), a 1997 statewide survey of Michigan schoolchildren. In that survey, a nonrandom sample of schoolchildren from 41 Michigan school districts were weighed and measured. Because 90% of our primary care patients were white, we used only the data on white children from the KFYP database. We summarized the KFYP data in the same way as the other surveys.

Results

A total of 993 children aged 4 to 17 years were measured and weighed in the Michigan primary care practices. There were 502 boys and 491 girls. Of these patients, 90% were white; 3% were Native American; 3% were African American; 1% was Hispanic; 1% was of another race; and 2% did not state their race. The Table 1 shows the percentage of children in the Michigan primary care practices above the 85th and 95th percentiles based on the NHES data and compared with NHANES data and Michigan schoolchildren (KFYP). The prevalences of obesity in the United States have increased over time, and the patients in our primary care practices have the highest prevalences of obesity in all age and sex groups when compared with the national surveys. For example, for boys aged 6 to 11, the percentage of those overweight in NHANES was 11%, whereas 20% of the Michigan primary care patients in this age group were overweight. In all groups, except for girls aged 6 to 11 years, prevalences of obesity of the primary care patients were greater than those of the Michigan schoolchildren in the KFYP survey.

Discussion

A larger than expected proportion of the children and adolescents in our primary care practices in Michigan are at risk for being overweight or are overweight. The prevalences of obesity in children and adolescents presenting for care in these practices are nearly 2 times greater than the national prevalences measured in NHANES II and NHANES III an d somewhat greater than a contemporary sample of Michigan schoolchildren. The absolute prevalences of obesity in the children and adolescents seen in our primary care practices were 3% to 9% higher than in the Michigan schoolchildren in all age-sex categories, except for girls aged 6 to 11 years.

It is not surprising that obesity is more prevalent in our sample than in the NHANES II and NHANES III surveys; NHANES data was gathered from 1976 to 1980 and 1988 to 1994, respectively, and obesity prevalences are steadily increasing. However, the reason that our patients are more obese than a contemporary sample of Michigan schoolchildren is unknown. The difference may be caused by systematic bias, as neither of the surveys was based on random sampling. However, in the analysis of our adult patients, the prevalence of obesity of patients was much greater than local community prevalences, even after adjusting for socioeconomic status and medical conditions known to be associated with obesity.⁸

The association between overweight and socio-economic status in children and adolescents is weak and not statistically significant.⁹ We speculate that obese children and adolescents have more physical and psychological complaints than healthy-weight individuals, causing them to visit their physicians more frequently. If they visit the practice more frequently, obese patients are more likely to be included in a survey of consecutive patients, thereby explaining the higher prevalences of obesity in primary care patients. Perhaps obese children or their parents are more worried about their health and therefore more likely to visit the physician for minor complaints. However, no association has been found between obesity and mental illness, although research regarding such an association is considered inconclusive at this time.¹⁰

Conclusions

The prevalences of obesity for children and adolescents presenting for care in Michigan primary care practices are higher than the state and national prevalences. Further studies using larger sample sizes, random sampling, and more settings are necessary to confirm or refute our findings. If this prevalence of obesity in primary care patients is confirmed, explanations for the differences should be explored.

Acknowledgments

We would like to thank the following people for their contributions to this study: Charles Kuntzelman, EdD, for providing us with the Michigan schoolchildren data; Nan Kreher, MEd, administration; Teresa Ettenhofer, administration and data analysis; and Kimber Gauthier, Jenny Folcik, and Celeste St. John, data entry. We would also like to thank the UPRNet and Michigan State University clinics that collected data. They are: Alpena Medical Arts, Alpena; Doctors Park Family Physicians, Escanaba; Dr Victoria Macki’s office, Newberry; OSF Medical Group, Gladstone and Escanaba; Gwinn Medical Center, Gwinn; Marquette Medical Clinic, Iron River; Burns Clinic, St Ignace; Alcona Health Center, Lincoln and Ossineke; East Jordan Family Health Center, East Jord an; Ewen Medical Clinic, Ewen; Northern Menominee Health Center, Spalding; Northern Michigan Health Services, Houghton Lake; Michigan State University-Family Practice at Clinical Center site, East Lansing; and Michigan State University-Family Practice at the Family Health Center, St. Lawrence Hospital, Lansing.

An increase in the prevalences of childhood and adolescent obesity in the United States has been documented in a series of national surveys conducted in the last 30 years. These include the National Health Evaluation surveys, NHES II (1963-1965) and NHES III (1966-1970), and the National Health and Nutrition Examination surveys, NHANES I (1971-1974), NHANES II (1976-1980), and NHANES III (1988-1994).^1-4 Obesity in childhood and adolescence may lead to disorders in adulthood, including hypertension, diabetes, coronary heart disease, and some cancers.⁵ The strongest connection between chronic disease and death is linked to adolescent rather than adult obesity.⁶ Between 1980 and 1994 there was a 6% increase in the absolute prevalence of obesity in US children.⁷ Data from NHANES III, the most recent national survey of weight and height in children and adults, showed that 14% of children aged 6 to 11 years and 12% of those aged 12 to 17 years were overweight.⁷

We noticed that the prevalence of obesity in our Michigan primary care practices appeared higher than that reported from national or state surveys. We confirmed our impression by measuring a consecutive sample of 6912 patients in 19 practices in the summer of 1996. Analysis of the adult data revealed that 54.5% of the men and 52.6% of the women were overweight, nearly double the prevalences cited in the 1993 Michigan Behavioral Risk Factor Survey (MBRFS): 29.5% for the men and 28.7% for the women.⁸ MBRFS obesity prevalences are slightly higher than contemporary national prevalences. Comparison with county-specific data from the MBRFS gave the same results. Adjusting the analysis for age, race, socioeconomic status, and chronic diseases associated with obesity such as diabetes and heart disease, did not significantly affect the results. This high prevalence of obesity in primary care patients as compared with population-based surveys has not been previously described, and the reason for it is unknown.

The purpose of our report is to describe the prevalences of obesity in of children and adolescents in our Michigan primary care practices and to compare those prevalences to national and state prevalences. We hypothesized that the children enrolled in our practices would have higher prevalences of obesity than national and state prevalences derived from surveys of community-dwelling populations.

Methods

Practices, Enrollment, and Measurement

We included 19 family practice offices of the Upper Peninsula Research Network (UPRNet) in our study. UPRNet is a family practice research network in rural northern Michigan. These practices are in rural locations, and community populations range from 300 to 15,000. Eleven of the offices are publicly funded community health centers. The populations served are typical of rural northern Michigan, with more than the state average of elderly and poor patients. To include urban and suburban children, we also asked 2 family practice clinics serving middle and lower socioeconomic groups in the Lansing area to participate.

In the spring and early summer of 1996, we invited consecutive patients aged 4 to 17 years visiting any of the participating practices for care to take part in our study, and we obtained informed consent. We chose spring and early summer to avoid those seasons when people wear heavy clothing and boots and to avoid enrolling vacationers who did not reside in the study areas. Patients with appointments specifically to see a dietitian, those known to be pregnant, and those unable to stand on a scale were excluded from the study.

Heights were measured using 1 of 2 methods, depending on the practice ’s preference; either with a measuring arm attached to a wall or with a measuring arm attached to a scale. All measuring arms were checked to be certain they were parallel to the floor. Weights were taken on balance beam scales calibrated less than a month before the study began. Patients removed their coats and shoes; no weight adjustments were made for clothing. Each office nursing staff reviewed the measuring procedures before participating in the study. Trained on-site study coordinators (office managers or nursing managers) supervised the data gathering process to ensure adherence.

The University Committee on Research Involving Human Subjects of Michigan State University approved our study protocol.

Definitions of Obesity in Children and Adolescents

There is no internationally accepted measure to define obesity in children. Methods for classifying childhood and adolescent obesity have been developed on the basis of comparison with growth charts (percentile weight for height by age and sex), percentile of triceps skin fold thickness, and percentile cutoffs for body mass index (BMI). BMI is emerging as the prefer red standard. The Expert Committee on Clinical Guidelines for Overweight in Adolescent Preventive Services⁸ recommends that children and adolescents with a BMI greater than the 85th percentile but less than the 95th be classified as “at risk of overweight” and those with a BMI greater than the 95th percentile be classified as “overweight.” We used that classification system.

For the comparisons, we followed the method used by Troiano.¹ For the baseline standard, he used the BMI distributions derived from the National Health Examination surveys (NHES II for children aged 6 to 11 years and NHES III for children aged 12 to 17 years). He calculated the percentage of children above the 85th and 95th percentile cutoffs of NHES for NHANES I, NHANES II, and NHANES III data. Similarly, we calculated the percentage of children and adolescents in the Michigan primary care practices with BMIs greater than the 85th and 95th percentiles as defined by NHES. For a contemporary regional comparison group, we obtained data from the Kuntzelman Fitness for Youth Program (KFYP), a 1997 statewide survey of Michigan schoolchildren. In that survey, a nonrandom sample of schoolchildren from 41 Michigan school districts were weighed and measured. Because 90% of our primary care patients were white, we used only the data on white children from the KFYP database. We summarized the KFYP data in the same way as the other surveys.

Results

A total of 993 children aged 4 to 17 years were measured and weighed in the Michigan primary care practices. There were 502 boys and 491 girls. Of these patients, 90% were white; 3% were Native American; 3% were African American; 1% was Hispanic; 1% was of another race; and 2% did not state their race. The Table 1 shows the percentage of children in the Michigan primary care practices above the 85th and 95th percentiles based on the NHES data and compared with NHANES data and Michigan schoolchildren (KFYP). The prevalences of obesity in the United States have increased over time, and the patients in our primary care practices have the highest prevalences of obesity in all age and sex groups when compared with the national surveys. For example, for boys aged 6 to 11, the percentage of those overweight in NHANES was 11%, whereas 20% of the Michigan primary care patients in this age group were overweight. In all groups, except for girls aged 6 to 11 years, prevalences of obesity of the primary care patients were greater than those of the Michigan schoolchildren in the KFYP survey.

Discussion

A larger than expected proportion of the children and adolescents in our primary care practices in Michigan are at risk for being overweight or are overweight. The prevalences of obesity in children and adolescents presenting for care in these practices are nearly 2 times greater than the national prevalences measured in NHANES II and NHANES III an d somewhat greater than a contemporary sample of Michigan schoolchildren. The absolute prevalences of obesity in the children and adolescents seen in our primary care practices were 3% to 9% higher than in the Michigan schoolchildren in all age-sex categories, except for girls aged 6 to 11 years.

It is not surprising that obesity is more prevalent in our sample than in the NHANES II and NHANES III surveys; NHANES data was gathered from 1976 to 1980 and 1988 to 1994, respectively, and obesity prevalences are steadily increasing. However, the reason that our patients are more obese than a contemporary sample of Michigan schoolchildren is unknown. The difference may be caused by systematic bias, as neither of the surveys was based on random sampling. However, in the analysis of our adult patients, the prevalence of obesity of patients was much greater than local community prevalences, even after adjusting for socioeconomic status and medical conditions known to be associated with obesity.⁸

The association between overweight and socio-economic status in children and adolescents is weak and not statistically significant.⁹ We speculate that obese children and adolescents have more physical and psychological complaints than healthy-weight individuals, causing them to visit their physicians more frequently. If they visit the practice more frequently, obese patients are more likely to be included in a survey of consecutive patients, thereby explaining the higher prevalences of obesity in primary care patients. Perhaps obese children or their parents are more worried about their health and therefore more likely to visit the physician for minor complaints. However, no association has been found between obesity and mental illness, although research regarding such an association is considered inconclusive at this time.¹⁰

Conclusions

The prevalences of obesity for children and adolescents presenting for care in Michigan primary care practices are higher than the state and national prevalences. Further studies using larger sample sizes, random sampling, and more settings are necessary to confirm or refute our findings. If this prevalence of obesity in primary care patients is confirmed, explanations for the differences should be explored.

Acknowledgments

We would like to thank the following people for their contributions to this study: Charles Kuntzelman, EdD, for providing us with the Michigan schoolchildren data; Nan Kreher, MEd, administration; Teresa Ettenhofer, administration and data analysis; and Kimber Gauthier, Jenny Folcik, and Celeste St. John, data entry. We would also like to thank the UPRNet and Michigan State University clinics that collected data. They are: Alpena Medical Arts, Alpena; Doctors Park Family Physicians, Escanaba; Dr Victoria Macki’s office, Newberry; OSF Medical Group, Gladstone and Escanaba; Gwinn Medical Center, Gwinn; Marquette Medical Clinic, Iron River; Burns Clinic, St Ignace; Alcona Health Center, Lincoln and Ossineke; East Jordan Family Health Center, East Jord an; Ewen Medical Clinic, Ewen; Northern Menominee Health Center, Spalding; Northern Michigan Health Services, Houghton Lake; Michigan State University-Family Practice at Clinical Center site, East Lansing; and Michigan State University-Family Practice at the Family Health Center, St. Lawrence Hospital, Lansing.