Relation of Time Spent in an Encounter With the Use of Antibiotics in Pediatric Office Visits for Viral Respiratory Infections FREE

Children with presumably viral respiratory infections are seen commonly in pediatric ambulatory practice and often given antibiotics unnecessarily.^1- 2 Unnecessary antibiotic usage leads to the increasingly serious problem of resistant bacteria.^3- 4 Although antibiotic prescribing practices for pediatric respiratory infections have become more judicious and in line with recommendations, antibiotics are still overused.^5- 6

Many factors for inappropriate antibiotic use have been identified.⁷ These include physician factors such as productivity concerns, perceived patient pressure, diagnostic uncertainty, as well as patient factors like anxiety, loss of work time, and misconceptions about what antibiotics do.² One of the purported physician factors for why physicians prescribe antibiotics for upper respiratory infections (URIs), based on focus groups, is to save time at office visits.⁸ Information from these focus groups indicates that physicians feel that it is faster to write a prescription for antibiotics than to explain why they are not needed. Physicians in many settings are under capitated arrangements with incentives to increase patient volume to meet productivity goals, thus getting patients in and out in a timely manner may play a role in antibiotic prescribing decisions.⁹ There is little information, however, that validates the perceptions of these physician by demonstrating that prescribing antibiotics for these viral infections actually leads to shorter visits that save time. In 1 study involving adults with URIs, antibiotic use was associated with a shorter visit duration by 1 minute.¹⁰ According to some authors, “Of all the factors leading to antibiotic overprescription, economic pressures may be particularly important but are the least well studied.”⁸

The purpose of this study was to examine the relationship between time spent in pediatric office encounters and the prescribing of antibiotics for upper respiratory infections.

The data used in this study come from the 2000 National Ambulatory Medical Care Survey (NAMCS). The NAMCS is a national survey conducted annually by the National Center for Health Statistics (NCHS) for the Centers for Disease Control and Prevention designed to meet the need for objective, reliable information about the provision and use of ambulatory medical care services in the United States.¹¹ The survey collects data on patient visits, as the primary unit of analysis, and office-based physician practices in the United States. Hospital-based clinics and federal sites are not included. The survey has a 3-tiered, multistage probability design based on geographic location, physician specialty, and individual visits within the practice. The NCHS weighs each visit by taking into account practice location and physician specialty. The weighing calculations also account for practices that were invited to participate but declined to do so. This weighing of survey clusters allows for data extrapolation to national estimates for all survey items. National estimates are considered reliable with a standard error of 30% or less, which generally corresponds to a sample of at least 30 patient visits. It is the only survey of office-based physicians in the United States that collects prescribing information and allows for unbiased national estimates.

Physicians are randomly selected from the master files of the American Medical Association and the American Osteopathic Association. Each physician is randomly assigned to a 1-week reporting period. During this period, data for a systematic random sample of visits are recorded by the physician or office staff on a standardized encounter form provided for that purpose and checked for completeness by NCHS field staff. The physician response rate in 2000 was 68% after accounting for those who declined to participate or did not see patients during the study period. The 2000 NAMCS was conducted on encounters from 27 369 patients from 1388 physicians.

Clinical and demographic data are collected for each visit. All patient demographic information is de-identified to prevent linkage with individual patients. Physician variables include geographic location, urban or rural site, and self-selected specialty. Clinical variables include up to 6 medication entries coded according to the National Drug Code Directory¹² and up to 3 diagnoses (1 primary and 2 secondary) coded according to the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM).¹³

To examine the use of antibiotics for primarily viral upper respiratory tract infections, we examined all office visits with the primary diagnoses of acute nasopharyngitis (common cold), (ICD-9-CM code 460), acute URIs of multiple or unspecified sites (ICD-9-CM code 465), acute bronchitis (ICD-9-CM code 466), and bronchitis not otherwise specified (ICD-9-CM code 490). In our assessment of overuse of antibiotics, we are aware that even though the primary diagnosis may not warrant prescribing antibiotics, a secondary or tertiary diagnosis may suggest the need for antibiotics. Because other conditions that warrant antiobiotic use may be present in some cases, we excluded from consideration visits with secondary or tertiary diagnoses of nonsuppurative otitis media (ICD-9-CM 381-381.4), suppurative otitis media (ICD-9-CM 382.xx), acute sinusitis (ICD-9-CM 461.xx), chronic sinusitis (ICD-9-CM 473.xx), acute pharyngitis (ICD-9-CM 462), acute tonsillitis (ICD-9-CM 463), streptococcal sore throat (ICD-9-CM 034.0), and pneumonia (ICD-9-CM 481.xx-486.xx). We also excluded bacterial infections (ICD-9-CM 041.xx) and urinary tract infections (ICD-9-CM 590.xx, 595.xx, 597.xx, 599.0). Additionally, we excluded secondary or tertiary diagnoses of other diseases due to the following: viruses and Chlamydiae (ICD-9-CM 078.xx), viral and chlamydial infection in conditions classified elsewhere and of unspecified site (ICD-9-CM 079.xx), syphilis and other venereal diseases (ICD-9-CM 090.x × -099.xx), unspecified urogenital trichomoniasis (ICD-9-CM 131.00), cystic fibrosis (ICD-9-CM 277.0), sickle cell anemia (ICD-9-CM 282.6x), mastoiditis and related conditions (ICD-9-CM 383.xx), acute laryngitis and tracheitis (ICD-9-CM 464.1x-464.3x), chronic nasopharyngitis (ICD-9-CM 472.2), chronic disease of tonsils and adenoids (ICD-9-CM 474.xx), chronic laryngitis and laryngotracheitis (ICD-9-CM 476.x), other diseases of upper respiratory tract (ICD-9-CM 478.xx), inflammatory disease of female pelvic organs (ICD-9-CM 614.x × -616.xx), cellulites and abscess of finger and toe (ICD-9-CM 681.xx), other cellulites and abscess (ICD-9-CM 682.x), impetigo (ICD-9-CM 684), unspecified local infection of skin and subcutaneous tissue (ICD-9-CM 686.9), osteomyelitis, periostitis, and other infections involving bone (ICD-9-CM 730.xx), venereal diseases (ICD-9-CM V01.6), gonorrhea (ICD-9-CM V02.7), and other venereal diseases (ICD-9-CM V02.8). An ICD-9-CM code with a designation of “x” indicates all codes (0-9) for that digit. Additionally, because the NAMCS does not allow for a differentiation of dosage length for treatment, we excluded visits with the secondary or tertiary diagnoses of acne (ICD-9-CM 706.1). We excluded these diagnoses as well as emphysema (ICD-9-CM 492.xx) and chronic bronchitis (ICD-9-CM 491.xx) because of evidence of effectiveness of antibiotic use in acute exacerbations.¹⁴

Antibiotic prescription for each visit was identified by using both trade and generic names of antimicrobial agents assigned to each medication prescribed by the NAMCS. The NAMCS survey format does not allow for direct linkage between diagnosis and medication but up to 6 medications are recorded for each visit. A visit was categorized as receiving antibiotics if any of the following medications were recorded as 1 of the 6 listed: penicillin, amoxicillin, ampicillin, erythromycin, tetracycline, doxycycline, trimethoprim, sulfamethoxazole and trimethoprim-sulfamethoxazole (narrow-spectrum agents), and cephalosporins, other macrolides, fluoroquinolones, and amoxillicin–clavulanic acid (broad-spectrum agents).

For each visit selected, the physician or a member of the physician’s staff provided information about the characteristics of the patient including the visit duration. The primary outcome variable in the analysis was face-to-face time a physician spent with a patient. The NAMCS includes an item labeled “time spent with a physician.” The intent of this item is to get the total time spent in face-to-face contact between the patient and the physician. The NCHS instructions for this item are to include the length of time the sample physician spent with the patient. Time spent waiting to see the doctor or receiving care from someone other than the doctor, ie, a nurse administering an injection, are not to be included. Also time spent on the patient, but not with the patient, such as reviewing records, are not to be included. The NCHS instructions also direct physicians to complete the patient record form immediately after each visit. To minimize data collection workload and facilitate timely data entry, all physicians in the sample complete only 30 records per their sampling week.

In the 2000 NAMCS, 16.2% of the visits had missing values for this variable. In these cases, values were imputed by randomly assigning a value from a patient record form with similar characteristics based on physician specialty, geographic region, and 3-digit ICD-9-CM codes for the primary diagnosis. Imputation for this variable was performed regardless of whether antibiotics were prescribed for the visit.

Because the form of payment may have an influence on the desire for the physician to see more patients per day, an assessment of whether the encounter was funded through a prepaid plan was included in the multivariate analysis. Prepaid status was determined by whether the visit was capitated or through a health maintenance organization plan as defined in a previous analysis using NAMCS for national visit length estimates.¹⁵ NAMCS has separate entries for both variables. Results were combined and dichotomized as participation or nonparticipation in a prepaid plan in the analysis.

It has been speculated that the patient-physician relationship may influence the therapeutic interaction. Consequently, as a potential control variable we included whether the physician of record was the patient’s primary care physician. The survey question, “Are you the patient’s primary care physician?” was recorded as yes, no, or unknown. In the analysis, this variable was dichotomized as primary or nonprimary care physician.

Only children and adolescents aged 18 years or younger were included in the study. Sex and race were also included in the multivariate analysis as control variables. Owing to a low percentage of other ethnic groups in the sample, race was dichotomized as white or black.

We used the statistical software STATA Intercooled version 8.2 (Stata Corp, College Station, Tex) to analyze all data. The masked survey design variables were programmed into STATA as recommended by the Ambulatory Statistics Branch of the National Center for Health Statistics.¹⁶ These design variables take into account the complex multistage sample design and weighing in the survey that allow for making population estimates and generation of variance estimates that result in conservative tests of significance.¹⁶

The assumption that t tests make, which is that every observation has the same variance, could not be made in the present sampling design. Consequently, we used simple linear regression as a substitute method for a t test to examine the statistical significance of the difference between the mean physician visit time for patients who received antibiotics and those who did not. Physician time spent with the patient was modeled as the dependent continuous variable with receipt of antibiotics modeled as the sole, dichotomized, categorical, independent variable. No control variables were included in this aspect of the analysis.

To control for potential confounding variables, a multivariate logistic regression model was developed to determine the relationship between antibiotic prescription and the time a physician spent with the patient while controlling for the patient’s age, sex, race, prepaid plan participation, and whether they were seen by their primary care physician. In this model, the outcome variable—physician time spent with a patient—was dichotomized as above or below the median physician visit time of 15 minutes.

The study was determined to be exempt from review by the Institutional Review Committee at Lancaster General Hospital, Lancaster, Pa.

The process for the selection of visits for the analysis is outlined in the Figure. The 2000 NAMCS is based on a total of 27 671 visit records representing more than 335 million estimated annual visits to office-based physicians in the United States. There were 4727 records for children and adolescents aged 18 years or younger, representing more than 170 million annual visits. Within this age group, 335 records had a primary diagnosis of URI or bronchitis. The final total was 269 records after excluding 55 for secondary or tertiary diagnoses of bacterial infections that would justify an antibiotic prescription, 3 for visit time recorded as 0, and 8 for having only 1 physician per geographic cluster. The remaining 269 visit records represent an estimated 12 366 162 national annual visits for these 2 diagnoses in this age group.

The characteristics of the sample are represented in Table 1. Demographically, the sample records are predominantly of white children aged 5 years or younger. In terms of economic factors, most patients had private insurance and almost half were in prepaid health plans, ie, either health maintenance organization or another capitated arrangement. About three quarters of the patients were seen by pediatricians and most by their primary care physician. Almost 90% of the visits were for an acute problem, with the common cold being the primary diagnosis for over 80% of the visits. The overall rate of antibiotic prescriptions (not included in the table) was 30%; 27% of those with a URI were given antibiotics and 46% of those with acute bronchitis were also given antiobiotics.

All results represent combined visits for URIs and bronchitis. In univariate analysis, there was no difference in the mean time a physician spent with a patient when comparing those visits with receipt or nonreceipt of an antibiotic prescription. Mean physician time (SE) spent with children with either colds or bronchitis was 14.24 (0.85) minutes when antibiotics were prescribed and 14.18 (1.03) minutes when they were not prescribed (P = .95).

The multivariate model showed that the prescription of antibiotics was not associated with whether the time a physician spent with a patient was above or below the median encounter time of 15 minutes, while controlling for demographic variables, such as patient participation in a prepaid plan, and whether the patient’s primary care physician attended the visit (Table 2).

Our analysis of ambulatory care visits to United States physicians in the year 2000, comparing physician-patient face-to-face time for presumed viral respiratory infections, treated with or without antibiotics, reveals that the time a physician spends with a patient in an encounter is not associated with the prescription of antibiotics. This lack of association remained when 2 factors that might be expected to influence the amount of time a physician spends in the examination room with a patient were taken into account. The analysis included variables for patient participation in a prepaid health plan and whether the patient’s primary care physician was the one attending the encounter. Neither of these factors, when analyzed independently or when modeled with the receipt of an antibiotic prescription, was associated with whether the amount of time a physician spent with a patient was above or below the median encounter time. The implications of these results are that when physicians see children with these diagnoses, it is unlikely that an antibiotic prescription will significantly reduce the amount of time the physician spends in the examination room. Although it is not possible to make definitive conclusions from this analysis, the implication may be that taking the time to explain that antibiotics are not indicated, or suggest alternatives such as bronchodilators, does not significantly add to the amount of time a physician spends in the examination room with a patient. As mentioned previously, information from physician focus groups has shown that saving time is a factor in a physician’s decision to prescribe an antibiotic.⁸ Our findings, demonstrating a lack of association between antibiotic prescribing and time in the room with the patient, are important for physicians who believe that antibiotic prescribing for these diagnoses saves time; further, it may decrease the incentives to prescribe antibiotics inappropriately. The results of this study, by addressing one of the key physician productivity reasons for inappropriate antibiotic use, provide more information about 1 of the factors responsible for antibiotic resistance.

Several other studies have addressed the question of physician time expenditure in relationship to antibiotic prescribing. In a prospective observational study of 306 pediatric patients, encounters were audiotaped to determine if the words used by physicians in describing examination findings were associated with antibiotic prescription.¹⁷ The time the physician spent in the examination room was also recorded. The median visit length in this study was 4 minutes and visits in which antibiotics were prescribed were a nonsignificant 31 seconds shorter (P = .11) compared with visits without antibiotics. The setting was limited to 2 pediatric practices that served educated families with high incomes (75% had annual incomes >$50 000). Therefore it is difficult to compare the median visit time with the results from a nationally representative sample as used in our analysis. A recent report of NAMCS data on visit duration compared with direct observation data concluded that NAMCS values are overestimated by 29%.¹⁸ The authors postulated that the discrepancy could be because of physician inclusion of time spent outside the examination room reviewing records in their visit duration estimates when completing NAMCS record forms. Nonetheless, the lack of association between prescribing antibiotics and physician time in the examination room mirrors the results in this analysis.

Another study addressed the relationship between antibiotic prescription and encounter time in adults with URIs.¹⁰ The mean visit time was 14.2 minutes when antibiotics were prescribed and visits associated with the receipt of antibiotics were 1 minute shorter. However, the analysis included upper respiratory diagnoses such as streptococcal sore throat, acute tonsillitis, and otitis media that are typically treated with antibiotics, and multivariate analysis did not control for patient participation in prepaid plans and whether the visit was attended by the patient’s primary care physician, as was done in our study. Consequently, we feel that our analysis provides data that fills a gap in the existing knowledge regarding overuse of antibiotics in children, including when care is rendered in prepaid plan settings.

There are several limitations to our study. First, the primary outcome variable “time spent with a physician” was not necessarily directly measured. The NAMCS instructions state to include the length of time the sample physician spent with the patient and not time spent waiting to see the doctor or receiving care from someone other than the doctor. So although the NAMCS goal was clearly to obtain data on the time the physician was in the room with the patient, it is possible that some of these measurements were estimated from the physician’s appointment calendar, rather than actually measured by support staff. Although the actual visit lengths may not be accurate, there is no reason to believe that the percentage of estimated measurements would vary for children receiving antibiotics because the data was collected without knowledge of this study’s aims.

Second, even though we attempted to use a rigorous method for linking the medication prescribed during a visit to the primary diagnosis, it is possible that the antibiotics were actually prescribed for a different condition than was addressed in the visit. Conditions other than viral respiratory infections may have required longer visits, thus increasing the visit time in the antibiotic group. However, we feel that this is not likely because 89% of the sample were seen for an acute problem consistent with a diagnosis of URI or bronchitis.

Third, NAMCS provides the ability to investigate the prescribing practices of office-based physicians, but does not provide information on emergency department or hospital ambulatory care center visits. It is possible that in these potentially more impersonal environments, prescribing antibiotics for viral respiratory infections could save time. In these settings, explanations of why antibiotics are not indicated or suggestions of over-the-counter alternatives by an unknown provider may meet with more parental resistance and longer face-to-face time with the physician. However, although our study was limited to data from office settings, our analysis found no relationship with visit time and whether patients were seen by their primary care physicians.

In conclusion, our analysis shows that prescribing antibiotics for children with presumably viral respiratory infections is not associated with the amount of time a physician spends in the room with a patient. This information is important in addressing one of the physician productivity factors responsible for inappropriate antibiotic prescribing, and in ultimately slowing the development of antibiotic resistance.

Correspondence: Andrew Coco, MD, MS, Medical Director, Louise Von Hess Health Research Institute, Lancaster General Hospital, 555 N Duke St, Lancaster, PA 17604 (ascoco@lancastergeneral.org).

Accepted for Publication: July 8, 2005.

Funding/Support: This study was supported by the American Academy of Family Physicians (Leawood, Kan) through their research stimulation grant program.