Data were obtained from the National Ambulatory Medical Care Survey (NAMCS) for a 5-year period and retrospectively analyzed. A detailed description of the survey is described elsewhere.²¹ The NAMCS is based on random samples of visits to non–federally employed office-based physicians who are primarily engaged in direct patient care. Trained interviewers provide the physicians with survey materials and instructions, and each physician records data on his or her patients' symptoms, diagnoses, and medications ordered or provided for a 1-week period. The survey also provides statistics on the demographic characteristics of patients and services provided, including information on diagnostic procedures, patient management, and planned future treatment. The NAMCS has a three-stage probability sampling design that randomly collects data from different geographic areas, medical specialties, and patient visits. 

The patients included in our study were broadly identified through the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM)²² diagnostic codes for acute URTIs of unspecific or multiple sites (code 465) as the primary diagnosis by osteopathic physicians. Patient characteristics analyzed included age, ethnicity, sex, concurrent condition, and new patient status. Patients with a concurrent condition were specifically identified by a second or third diagnosis of acute bronchitis, acute otitis media, acute pharyngitis, acute sinusitis, or asthma during the visit using ICD-9-CM codes 461, 462, 466, 382, or 493, respectively. 

Physicians were categorized according to medication prescribing, diagnostic testing practices (complete blood cell count, Streptococcus test), physiotherapy, specialty, whether he or she was a patient's PCP, practice location (metropolitan or nonmetropolitan [as defined by the US Office of Management and Budget]), practice type (single or group), and employment type (owner or non-owner). Although physiotherapy is broadly defined by NAMCS as “any form of physical therapy, including heat, light, sound or physical pressure, movement, or manipulation,” we assumed that osteopathic manipulative treatment was the form of “physiotherapy” most likely to have been used by osteopathic physicians in this situation. 

Statistical analyses were performed using SAS software (Version 8.1; SAS Institute, Cary, NC). Bivariate statistics were used to describe the frequency of acute URTIs among patient visits based on the characteristics of patients and physicians. Multiple logistic regression analysis assessed the independent contribution of patient and physician characteristics to the probability of prescribing antibiotics for URTIs. Prescribing antibiotics was the binary dependent variable. Profiles of patients and physicians were the independent variables. Adjusted odds ratios and 95% confidence intervals for the probability of prescribing antibiotics were calculated for each predictor to explain the strength of the association after controlling for patient age, race, sex, and practice location. The stepwise selection procedure was used for entering independent variables into the model, and only those variables that met the .05 significance entry level were reported. We used the “patient visit weight” provided by the NAMCS in every statistical procedure to reflect the national estimates. 

From July 1, 1997, through June 30, 2001, there were 9.6 million patient visits to osteopathic physicians for acute URTIs (Table). 

Adults represented 55.6% of the visits. Most patients were white (87.8% [8.43 million]), and slightly more than half were female (53.4% [5.13 million]). Patients who had a concurrent condition such as acute bronchitis, acute otitis media, acute pharyngitis, acute sinusitis, or asthma, accounted for 7.2% (0.69 million) of the visits. New patients accounted for 9.3% (0.89 million) of the visits. Patient characteristics that were significantly associated with higher rates of antibiotic prescribing were age (>18 y), ethnicity (nonwhite), sex (female), comorbidity, and patient status (new) (Table). Adult, nonwhite, female, and new patients were each about 1.5 times more likely to receive antibiotics compared with their counterparts. Patients with concurrent diagnoses were greater than 3 times more likely to receive antibiotics compared with those who did not have concurrent conditions. 

Antibiotics were prescribed in 56.4% (5.41 million) of all visits for URTIs. The prescribing of antibiotics for URTIs declined slightly over the 5-year period, with averages of 57.7% (5.54 million), 61% (5.86 million), 49.1% (4.91 million), 52.7% (5.06 million), and 52.9% (5.08 million) from 1997 through 2001, respectively (Figure). Two or more medications in addition to the antibiotics were prescribed in about 49% (4.70 million) of the visits. Diagnostic tests, such as complete blood cell count and Streptococcus test, were ordered in 64.8% (6.22 million) of the visits. No antibiotics were prescribed for patients who received physiotherapy (0.2% [0.02 million]). However, we were unable to conduct further analysis for this group because of the small sample size. 

The physician was the patient's PCP in 88.4% (8.49 million) of the visits. Ninety percent of the patients with URTIs were seen by family physicians, and the other 10% (0.96 million) were seen by general internists and pediatricians. More than 70% (6.72 million) of the physicians' offices were within metropolitan areas. More than 65% (6.34 million) of the physicians were owners of, and 28% (2.68 million) were not owners of their practices. 

Antibiotics were more likely to be prescribed when multiple medications such as nonsteroidal anti-inflammatory drugs and combination decongestant-antihistamine medications were prescribed (odds ratio [OR], 1.49; 95% confidence interval [CI], 1.42–1.56) and when diagnostic tests were performed during the visit (OR, 1.15; 95% CI, 1.05–1.27). Family physicians were greater than 3 times more likely to prescribe antibiotics compared with general internists or pediatricians (OR, 3.18; 95% CI, 2.72–3.72). When the physician was the patient's PCP, he or she was also more likely to prescribe antibiotics (OR, 2.98; 95% CI, 2.58–3.45). Physicians who were not owners of their practice (OR, 1.33; 95% CI, 1.18–1.49), those in a group practice (OR, 1.47; 95% CI, 1.22–1.52), and those practicing in nonmetropolitan areas (OR, 1.83; 95% CI, 1.61–2.08) were more likely to prescribe antibiotics. 

A number of factors were found to influence physicians' prescribing of antibiotics. Among patient factors, being an adult, female, and nonwhite were identified as influential factors for prescribing antibiotics. Because this study did not assess how those factors influence antibiotic prescribing patterns for URTIs, further studies may be needed to clarify their importance. 

Patients with acute nonspecific URTIs and a concurrent condition were treated with antibiotics more than 3 times as often as patients with acute URTIs alone, despite published practice guidelines that do not recommend antibiotics for most patients with these conditions.^3,4,15,21 The origin of such acute infections is usually viral, and very little benefit has been demonstrated from treatment with antibiotics.^{3,13,15,18,19} Nyquist et al,²³ who also used the NAMCS database for their 1998 study, suggested that physicians might specify a concurrent diagnosis to validate their prescribing of antibiotics. We contend that documenting a second diagnosis would tend rather to result in under-rather than overreporting of actual antibiotic prescribing for simple URTIs. 

Physician-related factors also determined whether patients were given antibiotics. Antibiotic prescribing may also reflect diagnostic uncertainty, as antibiotics were prescribed much more often when a diagnosis was made on clinical grounds alone. In pharyngitis, for example, treatment with antibiotics is recommended only in the presence of group A β-hemolytic Streptococcus,^3,4,14,15 which is the causal agent in only 5% to 15% of all cases of adult pharyngitis.^15,24 Diagnosis based on clinical presentation alone is unreliable; even experienced physicians have only moderate success in predicting positive culture results.²⁴ In the present study, it seems that the more severe the patient presentation was (ie, those for whom additional medications were given and diagnostic tests were performed), the more likely it was that he or she was given antibiotics. 

Interestingly, “physiotherapy” was performed by osteopathic physicians in 0.2% (0.02 million) patient visits, but no antibiotics were prescribed in this group. Again, we assumed that the NAMCS's broadly defined physiotherapy, when used by osteopathic physicians, was probably osteopathic manipulative treatment. Although we were unable to assess the significance of this finding due to the small sample size, the physiotherapy used by these osteopathic physicians for URTIs may have been an important means not only for symptomatic relief and enhancement of patient satisfaction, but also for reduction of unnecessary antibiotic prescribing. Further studies are needed to validate this finding. 

Our findings that family physicians prescribed antibiotics more than 3 times as often as general internists or pediatricians and that physicians who were not owners of their practice group were 1.5 times more likely to prescribe antibiotics are consistent with the findings of other studies.^16,23 Some researchers have attributed high prescribing rates to peer pressure,²⁵ patient volume,^16,25 legal concerns,²⁵ and the physician's desire to validate the office visit.^26,27 In our analysis, however, comparison of specialty prescribing rates did not include a direct comparison of age-stratified patient populations. Given the much greater likelihood of most patients being seen by family physicians or their PCPs, our results may be a function of patient population demographics rather than of specialty or philosophy. An ongoing patient-PCP relationship resulted in antibiotics being prescribed 3 times as often. 

Physicians' perceptions of patients' wishes may also influence their prescribing behavior. Patient expectation is cited by physicians as affecting their decision to prescribe antibiotics for nonspecific URTIs more often than any other factor.^4,14,16,23 However, there is considerable evidence that patient satisfaction is not dependent on whether their initial expectations are met, but rather on the nature and quality of the communication and interaction between patient and physician.^{3,4,14–16,20} While many patients expect to receive antibiotics, most patients report the primary reason for the visit to be a desire for symptom relief.²⁸ Most patients stated that they would be satisfied without a prescription for antibiotics if the physician explained the reasons for his or her decision.²⁸ In many cases, physicians have cited reducing the number of return visits as a reason for prescribing antibiotics.^{4,14,16,23,29} The perception by physicians that giving antibiotics will make a return visit less likely may also explain why antibiotics were prescribed more in nonmetropolitan areas. In these areas, return visits may also be more time consuming and more burdensome for the patient. However, according to many studies, prescribing antibiotics for viral conditions results in more return visits rather than fewer.^3,14–16 When prescribed medications do not bring about an improvement or resolution of URTIs, patients are likely to return. Moreover, the prescribing of antibiotics tends to validate the idea that minor illnesses require a visit to the doctor, and to reinforce patients' beliefs that antibiotics are always necessary.^3,4,16 

Prescribing antibiotics for acute nonspecific URTIs appears to be influenced by a complex interaction between patient and physician factors. Osteopathic physicians are uniquely qualified to lead the effort to reduce unnecessary antibiotic use because judicious prescribing—not only of antibiotics, but of all medications—is a fundamental tenet of the holistic osteopathic philosophy. Physician education efforts, perhaps in the continuing medical education setting, that include the integration of osteopathic manipulative procedures with a focus on the relief of symptoms should be emphasized and may provide greater success in reducing unnecessary antibiotic prescribing among osteopathic physicians.^30,31