Background. In response to the dramatic emergence of resistant pneumococci, more judicious use of antibiotics has been advocated. Physician beliefs, their prescribing practices, and the attitudes of patients have been evaluated previously in separate studies.
Methods. This 3-part study included a statewide mailed survey, office chart reviews, and parent telephone interviews. We compared survey responses of 366 licensed pediatricians and family physicians in Georgia to recently published recommendations on diagnosis and treatment of upper respiratory infections (URIs). We further evaluated 25 randomly selected pediatricians from 119 surveyed in the Atlanta metropolitan area. For each, charts from the first 30 patients between the ages of 12 and 72 months seen on a randomly selected date were reviewed for encounters during the preceding year. A sample of parents from each practice were interviewed by telephone.
Results. In the survey, physicians agreed that overuse of antibiotics is a major factor contributing to the development of antibiotic resistance (97%), and that they should consider selective pressure for resistance in their decisions on providing antibiotic treatment for URIs in children in their practices (83%). However, many reported practices do not conform to the recently published principles for judicious antibiotic use. For example, 69% of physicians considered purulent rhinitis a diagnostic finding for sinusitis; 86% prescribed antibiotics for bronchitis regardless of the duration of cough; and 42% prescribed antibiotics for the common cold. Reported practices by family physicians were more often at odds with the published principles: they were significantly more likely than pediatricians to omit pneumatic otoscopy (46% vs 25%); to omit the requirement for prolonged symptoms to diagnose sinusitis (median 4 vs 10 days); and to omit laboratory testing for pharyngitis (27% vs 14%). Of the 7531 encounters analyzed in the chart review, 43% resulted in an antibiotic prescription, including 11% of checkups, 18% of telephone calls, and 72% of visits for URIs. There was wide variability in the overall antibiotic use rates among the 25 physicians (1–10 courses per child per year). There was an even wider variability in some diagnosis-specific rates; bronchitis and sinusitis in particular. Those with the highest antibiotic prescribing rates had up to 30% more return office visits. Physicians who prescribed antibiotics for purulent rhinitis were more likely to see parents who believed that their children should be evaluated for cold symptoms.
Conclusions. Physicians recognize the problem of antibiotic resistance but their reported practices are not in line with recently published recommendations for most pediatric URIs. The actual prescribing practices of pediatricians are often considerably different from their close colleagues. Patient beliefs are correlated with their own physician's practices.
The spread of bacteria resistant to common antimicrobial agents, particularly multidrug-resistant pneumococci, has led to renewed interest in antimicrobial prescribing patterns and the promotion of judicious antimicrobial use. The 5 major indications for outpatient antimicrobial prescribing are all upper respiratory infections (URIs), and children are given a disproportionate number of these prescriptions.1 Many of the prescriptions are unnecessary, being given for viral infections such as the common cold.2 Reducing inappropriate prescribing has been advocated as one way to protect patients against resistant infections,3 and a national campaign has been developed to improve understanding of resistance by the general public and prescribing by physicians.4 A recently published set of recommendations was developed collaboratively by members of the Centers for Disease Control and Prevention, the American Academy of Pediatrics, and the American Academy of Family Physicians to identify specific situations in which unnecessary antibiotic use could be curtailed without compromising patient care (herein referred to as “the principles”).5–9
The success of efforts to improve antimicrobial prescribing by physicians depends on a detailed understanding of current prescribing patterns. Some physicians report a reluctance to reduce inappropriate prescribing because of the concern that patients will simply obtain antibiotics from colleagues whose routine practice may differ substantially from their own.10 Recent publications have reviewed mailed surveys,11 existing databases,12,,13 and selected practices.14 The factors leading to antimicrobial overuse are complex, involving physician beliefs, the constraints of daily practice, and patient expectations. The interplay of these factors has not been examined in a single study.
We report the results of a survey of pediatricians and family physicians in Georgia, a chart review of the practices of a population-based sample of pediatricians in Atlanta, and parent interviews with the patients of these physicians. We compare survey responses to the subsequently published principles, and correlate the physician practices with the beliefs of their patients.
From lists of all 1230 family physicians and 823 pediatricians licensed in Georgia, we randomly selected 75% of the members of each specialty. This sample size was based on an estimated 30% to 40% response rate and an estimated final sample of ∼500. Selected physicians were mailed a survey in June 1996 regarding outpatient antimicrobial use practices for URIs for children ages 5 and under. Physicians were eligible to complete the survey if they provided primary care for at least 10 children <5 years of age per week. Two reminder letters and surveys were sent to nonrespondents with postage-paid return envelopes. A random sample of nonrespondents was contacted with at least 3 follow-up calls encouraging a written response or completion of an abbreviated version of the survey by telephone. These responses were analyzed separately and were not included for comparison with original respondents.
We used the subsequently published principles,5–9 as a guide to appropriate diagnosis and treatment of upper respiratory conditions. Responses to the questionnaire were tabulated and compared with these principles. Because responses from pediatricians and nonpediatricians consistently differed, these groups were analyzed separately. Comparisons were made by using χ2or Fisher's exact tests, as appropriate, for dichotomous variables, and t tests or Wilcoxon rank-sum tests, as appropriate, for continuous variables.
Of 366 surveys returned, 119 were completed by pediatricians practicing in metropolitan Atlanta. The list of 119 was randomized, and 28 pediatricians were approached for consent to review charts in their offices. Each physician who agreed to participate in the study was assigned an index date randomly selected from 1996. The first 30 children who visited the physician on the index date for any purpose and who met age criteria were included in the study. In group practices in which >1 physician treated the child but 1 physician was assigned as the primary care provider, any child visiting the office on the index date and for whom the selected physician provided primary care was eligible for the study. In group practices in which no single physician was assigned to any child, children who specifically visited the selected doctor on the index date were chosen.
From each chart, demographic information about each child was obtained, and a visit log for the 12 months before the index date was completed. For any visit or telephone call in which the child either was diagnosed as having a URI or was prescribed an antibiotic, both the diagnosis(es) and antibiotic(s) prescribed were noted. The diagnosis noted was that actually written in the chart by the physician (as opposed to their billing code). On <10% of occasions, the physician's diagnosis was not listed but was deduced by abstractors using a predetermined algorithm.
Limited resources precluded interviewing all parents of children whose charts were reviewed. To select an unbiased sample, the list of 25 participating physicians was randomly ordered, as was each list of the 30 charts reviewed. Practices were then interviewed sequentially. After the first 15 patients on each list were selected, an introductory letter was mailed to the parent(s) by their physician, and the parent(s) were contacted by a trained interviewer who used a standardized form. Attempts to reach each patient's parent(s) were limited to 3 separate calls at different times of the day.
Office characteristics, patient populations, and antibiotic prescribing practices among the 25 pediatricians were described, and predictors of high and low antibiotic prescribers were sought using a χ2 or t test, as appropriate. We adjusted the tests for the effects of clustering introduced by the 3-stage sampling scheme (physicians, patients, parents), by using the robust variance estimators provided in Sudaan software (Research Triangle Institute, Research Triangle Park, NC). Associations between physician practices and patient beliefs were evaluated using the logistic regression module in Sudaan to adjust for clustering.
Of 1541 surveys mailed, 695 were ineligible (incorrect address, retired, saw <10 pediatric patients per week) and 366 (43%) of 846 eligible returned analyzable surveys. In addition, 28 of 56 eligible nonrespondents later agreed to answer questions by phone.
Fifty-eight percent of respondents were pediatricians and 42% were family practitioners. The characteristics of pediatricians and family physicians were similar, except that pediatricians tended to be slightly older, treated more children per week, and were more likely to be female (Table 1). The characteristics and responses of 28 nonrespondents who agreed to answer questions by phone were not significantly different from those of the respondents.
In several of the survey questions, physicians were asked whether they agreed or disagreed with specific statements (noted here in quotation marks). Most physicians (97%) agreed that “overuse of antibiotics is a major factor contributing to the development of antibiotic resistance.” Physicians agreed that antibiotic resistance is important on an individual level: that “prior antibiotic use increases the risk that a child will develop a resistant infection” (78%) and that “resistance has resulted in treatment failure for children with URIs in my practice” (69%). Physicians also agreed that they “should consider selective pressure for resistance in decisions on providing antibiotic treatment for URIs in children in my practice” (83%).
Fifty-eight percent of all physicians responded that their decision to provide an antibiotic for nonspecific URI was affected by parental pressure, and 49% were influenced by parents specifically asking for therapy for a nonspecific URI (ie, common cold). Physician perception of pressures to provide antibiotics varied by specialty. For example, 18% of pediatricians versus 34% of family physicians (P = .001) agreed that “avoiding the cost of a return visit, particularly in managed care patients, is an incentive to provide antibiotics to children with nonspecific URIs.”
The diagnostic and treatment practices of Georgia physicians differed in some important ways from those recommended in the principles subsequently published for each of the 5 major upper respiratory conditions (Table 2). Family physicians were consistently less likely to report practices supported by the principles.
Physicians responded with 68 different combinations of findings that they believed were essential to diagnosing sinusitis. Family physicians and pediatricians also differed regarding the duration of symptoms they required to diagnose sinusitis. Family physicians who reported duration of nasal discharge was an important factor required a median of 4 days of nasal discharge while pediatricians required a median of 10 days (Wilcoxon P < .001) with a minority (15%) requiring >14 days.
For bronchitis, durations required by pediatricians and those required by family physicians were similar, with an overall median of 4 days. However, there was no consistent set of diagnostic findings for bronchitis. Physicians reported 102 different combinations of necessary findings to diagnose bronchitis.
Many physicians (55%) acknowledged that they “occasionally provide antibiotics to a child with a nonspecific URI to prevent otitis media.” Forty-two percent of family physicians versus 16% of pediatricians (P = .001) agreed that “a child with purulent nasal discharge for 2 to 3 days will likely have a protracted illness if not treated with antibiotics.”
Overall, an antibiotic was prescribed in 43% of the 7531 encounters. Antibiotics were prescribed for 19% of 1446 telephone encounters and 49% of 6084 office visit encounters. Visits for any upper respiratory complaint resulted in an antibiotic 65% of the time, but well-child checkups also resulted in prescriptions 11% of the time.
Antibiotics were consistently prescribed when the final diagnosis was any of 5 URIs; otitis media (96% of 1723 encounters), pharyngitis (82% of 729 encounters), sinusitis (95% of 341 encounters), bronchitis (85% of 293 encounters), or purulent rhinitis (97% of 157 encounters). In 649 encounters in which the common cold was diagnosed, an antibiotic was prescribed 31% of the time, although 9 physicians rarely or never prescribed an antibiotic if the common cold was diagnosed (≤0.1 prescription per child-year).
There was a wide variation in the rate of antibiotic prescribing among the 25 practitioners, ranging from 1 to 10 prescriptions per child per year. This variation was not fully explained by any practice characteristic, physician profile, or patient demographics, although the power to explore these potential explanations was limited because only 25 practitioners were in the sample. Notably, however, physicians with no Medicaid patients prescribed more antibiotics, as did those with <50% of patients who were African-American (Table 3).
Diagnosis and prescribing rates for specific upper respiratory diagnoses were even more variable than overall rates. Of particular note was the variability of prescribing for bronchitis, for which 16% of physicians never prescribed an antibiotic. Some physicians who rarely prescribed antibiotics for bronchitis commonly prescribed for sinusitis, and vice-versa (Fig 1).
Certain diagnosis-specific prescribing was associated with lower overall antibiotic prescribing rates. Physicians who rarely treated sinusitis (≤1 prescription per 30 child-years) also prescribed fewer antibiotics overall (3.1 vs 4.8 prescriptions per child-year,P = .01). Those who rarely treated bronchitis or purulent rhinitis also were low overall prescribers (2.1 vs 4.6 prescriptions per child-year, P = .005).
Those physicians who prescribed ≤1 antibiotic per child-year for probable viral conditions had lower rates of office visits per child per year. This was true for bronchitis or purulent rhinitis (7.3 visits per child-year for low prescribers versus 10.4 for others,P = .09) as well as for sinusitis (7.7 visits per child-year for low prescribers versus 11.0 for others,P = .002).
Interviews were conducted with 211 parents from 16 of the physician practices reviewed (mean of 13.2 parents per physician, range 8–15). There was a direct correlation between the prescribing patterns of physicians and the beliefs of the parents of their patients. For example, physicians who prescribed >5 antibiotic courses per child per year were more likely to see parents who believed that a child should be brought to the doctor for a runny nose (P = .001). The association was even more apparent for diagnosis-specific prescribing. There was a direct correlation between physicians who prescribed antibiotics for purulent rhinitis and parents who believed they should be evaluated in the office if their child had a runny nose (P = .002; Fig 2).
In this 3-part study, we were able to compare physician beliefs with their reported practice and their actual practice and evaluate how these physician practices correlate with the beliefs of their patients. In the chart review, we found evidence of a wide variation in physicians' prescribing practices for pediatric upper respiratory infections even though all study participants were office-based pediatricians in the same metropolitan area. The best predictor of whether a physician was a high or low prescriber was whether that physician frequently used certain diagnostic labels. In fact, making a diagnosis of an upper respiratory condition was tantamount to giving a prescription for that condition. Two approaches may help to reduce unnecessary antibiotic prescribing: improving diagnostic practices for all upper respiratory conditions, and improving antibiotic prescribing practices for certain of these conditions. How, then, could physicians make more accurate diagnoses and what is influencing their prescribing practices for these diagnoses?
Physicians most clearly diverged from the published principles in their diagnosis and treatment of purulent rhinitis, bronchitis, and the common cold. For example, many physicians reported that the purulence of sputum or nasal discharge is an important diagnostic feature. Many physicians, family physicians in particular, were concerned that purulent nasal discharge would lead to a protracted illness or otitis media. This is consistent with a survey of physicians in Virginia, that found that 71% of family practitioners and 53% of pediatricians would prescribe antimicrobial agents immediately for a 10-month-old infant with scant, green, mucopurulent nasal secretions of 1 day's duration.15 The belief that purulent nasal discharge is an indication for antibiotics seems to be common despite evidence that purulence of nasal discharge does not indicate bacterial infection.16–19 Physicians also diverged from the principles by prescribing β-lactam antimicrobials for bronchitis. Although controlled trials have failed to demonstrate the benefit of antibiotic treatment for acute bronchitis,20 some reports have implicated Mycoplasma pneumoniae21,,22 orChlamydia pneumoniae23,,24 in prolonged cough illness. β-lactam antimicrobials are ineffective against these pathogens, and yet these are the agents that the majority of physicians in our study prescribed for bronchitis. For the common cold, for which standards are unambiguous and nearly universally accepted, physicians still report that they prescribe antibiotics, although not quite to the extent found in a recent review of Kentucky Medicaid claims.2
Some conditions may be more amenable to targeted intervention because physicians agree with the principles even though they do not practice them. For example, physicians report that middle ear effusion is important to document for diagnosing acute otitis media. However, 46% of family physicians do not routinely perform pneumatic otoscopy, and 33% of those who stated that pneumatic otoscopy is essential do not routinely perform it. Although most physicians have been taught pneumatic otoscopy in medical school, many have not continued to practice the technique and may have lost those skills; or they may avoid pneumatic otoscopy because it is more time-consuming and is technically difficult.
Many physicians are overprescribing antibiotics despite their belief that antimicrobial resistance is a problem and that it is linked to overuse of antibiotics. What is influencing high prescribing rates for upper respiratory diagnoses? Several studies have attributed high prescribing rates to parental pressure, patient volume, legal concerns, and the physician's desire to validate the office visit.25–27 A recent editorial emphasized the role of patient expectations in driving inappropriate antimicrobial prescribing and the responsibility of physicians to educate parents about antimicrobial use.28 Our survey supports the hypothesis that parental pressure affects prescribing rates. In our chart review we also found a strong correlation between the practices of physicians and the beliefs of their patients. This correlation likely reflects the central role of physicians in educating their own patients and those of their close colleagues about when to seek care for upper respiratory conditions. It is also possible that patients who uniformly expect antimicrobial treatment seek out like-minded physicians, but a recent study found that only a small proportion of patients would seek a different physician because of the prescribing pattern of their own practitioner.10 Physicians in that study cited unrealistic patient expectations as the single most important reason for their own inappropriate prescribing. The current findings indicate that these unrealistic patient expectations may be generated, at least in part, by previous inappropriate prescribing by the physician.
The wide variability in prescribing practices documented among the physicians in this study and among physicians elsewhere29,,30 may also be an impediment to reducing inappropriate use. A concern expressed by physicians is that if they reduce their prescribing, their patients will have little difficulty obtaining a prescription from a colleague with less stringent diagnostic standards.10 This problem is compounded by the fact that the correct practice is not known or agreed on for some conditions such as bronchitis. Bronchitis has not received much attention in pediatric literature; from 1995 through July 1998 there were only 3 English language pediatric articles on Medline, compared with 33 for sinusitis. The large number of combinations of symptoms reported (1.6 times as many combinations as were cited for sinusitis) and the lack of consensus on duration of symptoms needed for diagnosis may result from this lack of published data. Because of the lack of standards and the inconsistency of physicians in both diagnosis and treatment of bronchitis, it is a syndrome for which published principles may be useful in laying out a standard for discussion and debate. A uniform and agreed-on set of standards for diagnosis and treatment of upper respiratory infections has been reported to be important in accomplishing reduced antimicrobial use among colleagues.31 The Centers for Disease Control and Prevention, the American Academy of Pediatrics, and others have worked cooperatively toward developing such standards.32
Why are family physicians consistently more likely than pediatricians to prescribe antibiotics for URIs? Differences in rates of antimicrobial prescribing between pediatricians and family physicians have been reported by others for colds, URIs, and bronchitis.12 Family physicians may be exposed to different literature or have different priorities for providing care. For example, the greater duration of signs required for diagnosis of sinusitis by pediatricians may reflect the emphasis on documenting 10 days of symptoms in a series of studies published in the pediatric literature by Wald and others.33–37 Alternatively, it may be that family physicians are aware of and agree with the recommendations in the pediatric literature but choose not to incorporate them in their practice. This may reflect a differing focus on the interest of the child versus those of the family. For example, requiring a child to return after 10 days of nasal discharge rather than prescribing an antibiotic after only 4 days may not be in the best interest of a family with limited funds or an inability to make a follow-up visit. In fact, a significantly greater number of family physicians than pediatricians in our survey were concerned with avoiding the cost of a return visit.
According to the chart review, however, physicians who prescribed antibiotics more judiciously had significantly (up to 30%) less total office visits per child per year. In a fee-for-service system, revenues might be increased if patients are taught to seek care for probable viral conditions. In a capitated system, however, these unnecessary office visits are costly. Physicians and patients in capitated systems are unlikely to spontaneously modify ingrained care-seeking and prescribing behavior.14 Special efforts to reeducate patients and reorient physicians may prove to be effective and cost-beneficial, in addition to decreasing the pressure driving antibiotic resistance.
Our survey had several potentially important limitations. The response rate was relatively low, raising concern about potential differences between those who chose to or not to respond. A random sample of nonrespondents suggested that they were a similar population and had similar beliefs about antimicrobial use and resistance. However, the nonrespondent sample was small and may not have given us the power to detect smaller differences between respondents and nonrespondents. The survey was only conducted in Georgia and therefore may not be generalizable to the practices and beliefs of physicians in other parts of the United States. It is also possible that the findings in the chart review are not representative of pediatric upper respiratory prescribing in general. The focus on pediatricians is likely to lead to underestimates of inappropriate prescribing because pediatricians are consistently found to have lower prescribing rates for upper respiratory infections than nonpediatricians.12,,13 On the other hand, studying physicians in Atlanta may tend to provide overestimates of inappropriate prescribing, because there is some evidence that physicians in the southern United States prescribe antibiotics more liberally than in other regions.12 Older physicians may be more likely to be “high prescribers,”13 but we had a limited number of older physicians with whom to evaluate this hypothesis. Actually, individual physician prescribing may be even more variable than reported here, because some patients were seen by >1 physician in the practice, and averaging several physicians in the same practice would tend to blunt any unusual prescribing patterns.
As was recently reported from Scandinavia,30 we found prescribing patterns for URIs to vary widely from physician to physician. No patient or physician characteristic adequately explained this variation other than individual physician diagnostic style. Although parent beliefs varied as well, these beliefs were closely correlated with the practices of their children's physicians. Georgia physicians recognize the problem of antibiotic resistance and its link to the overuse of antibiotics, yet many do not practice judicious antibiotic use. This survey highlights specific areas where current practice differs from published principles. Improved education of medical students and residents and the establishment of more uniform standards for pediatric antimicrobial prescribing may help reduce the variation in physician practice and unnecessary antibiotic prescribing. Education of their own patient population about the association between antimicrobial misuse and antimicrobial resistance is an important responsibility of pediatricians and family physicians, that should be actively supported by national and local public education efforts. The Centers for Disease Control and Prevention has developed materials to provide such education that are available on the Internet (www.cdc.gov) or can be provided in response to a fax request (fax 404-639-4702).
This research was supported in part by an appointment to the Research Participation Program at the Centers for Disease Control and Prevention, National Center for Infectious Diseases, Division of Bacterial and Mycotic Diseases, administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the United States Department of Energy and the Centers for Disease Control and Prevention.
We thank the pediatricians in Atlanta who invited us into their practices to conduct the study, and we acknowledge the valuable contributions of Cynthia-Lou Coleman, Doug McLaughlin, Joan Stebel, and Anika Foster, who conducted telephone interviews. The authors thank Heidi Bauer and Doug McLaughlin for their assistance with data management. We also thank Kathleen Toomey for assistance with study design and conduct.
- URI =
- upper respiratory infection
- ↵Dowell SF, Marcy SM, Phillips WR, Gerber MA, Schwartz B. Otitis media—principles of judicious use of antimicrobial agents. Pediatrics. 1998;101:165–171. Supplement
- ↵Rosenstein N, Phillips WR, Gerber MA, Marcy SM, Schwartz B, Dowell SF. The common cold—principles of judicious use of antimicrobial agents. Pediatrics. 1998;101:181–184. Supplement
- ↵O'Brien KL, Dowell SF, Schwartz B, Marcy SM, Phillips WR, Gerber MA. Acute sinusitis—principles of judicious use of antimicrobial agents. Pediatrics. 1998;101:174–177. Supplement
- ↵O'Brien KL, Dowell SF, Schwartz B, Marcy SM, Phillips WR, Gerber MA. Cough illness/bronchitis—principles of judicious use of antimicrobial agents. Pediatrics. 1998;101:178–181. Supplement
- ↵Schwartz B, Marcy SM, Phillips WR, Gerber MA, Dowell SF. Pharyngitis—principles of judicious use of antimicrobial agents. Pediatrics. 1998;101:171–174. Supplement
- Barden LS,
- Dowell SF,
- Schwartz B,
- Lackey C
- McIsaac WJ,
- Goel V
- Hays GC,
- Mullard JE
- Foy HM,
- Cooney MK,
- Maletzky AJ,
- Grayston JT
- Grayston JT,
- Campbell LA,
- Kuo CC
- Grob PR
- Bauchner H,
- Philipp B
- ↵CDC: The CAUSE (careful antibiotic use to prevent resistance). Newsletter. October, 1997; www.cdc.gov/ncidod/dbmd/cause/oct97.htm
- ↵Dowell SF, Marcy SM, Phillips WR, Gerber MA, Schwartz B. Principles of judicious use of antimicrobial agents for pediatric upper respiratory tract infections. Pediatrics. 1998;101:163–165. Supplement
- Wald ER
- Wald ER
- Copyright © 1999 American Academy of Pediatrics