Dark blue bars show prescribing of antibiotic courses of extended duration, defined as more than 28 days. Light blue bars show prescribing of antibiotic courses of short duration, defined as 28 days or less. To account for varying frequency of dermatology encounters between years, prescription counts were divided by the number of dermatology visits to calculate rates per 100 dermatology encounters.
A and B, Extended courses are more than 28 days, and short courses are 28 days or less. To account for varying frequency of dermatology encounters between years, prescription counts were divided by the number of dermatology visits for the associated diagnosis to calculate rates per 100 dermatology encounters with the associated diagnosis. SSTI indicates skin and soft-tissue infection.
Green, yellow, and red boxes show more frequent, intermediate, and less frequent antibiotic classes, respectively. ER indicates extended release; LD, low dose.
eTable. Antibiotic Tiers
Customize your JAMA Network experience by selecting one or more topics from the list below.
Barbieri JS, Bhate K, Hartnett KP, Fleming-Dutra KE, Margolis DJ. Trends in Oral Antibiotic Prescription in Dermatology, 2008 to 2016. JAMA Dermatol. 2019;155(3):290–297. doi:10.1001/jamadermatol.2018.4944
In what settings do dermatologists most frequently prescribe antibiotics, and how is this use changing over time?
Between 2008 and 2016 among 985 866 courses of oral antibiotics prescribed by 11 986 unique dermatologists in this repeated cross-sectional analysis, there was a decrease in overall antibiotic prescribing from 3.36 to 2.13 courses per 100 visits. However, there was an increase in prescribing associated with surgical visits from 3.92 to 6.65 courses per 100 visits.
Overall antibiotic use by dermatologists is declining; however, the use of oral antibiotics after surgical visits is increasing, and the value of this practice should be carefully evaluated.
Dermatologists prescribe more oral antibiotic courses per clinician than any other specialty, and this use puts patients at risk of antibiotic-resistant infections and antibiotic-associated adverse events.
To characterize the temporal trends in the diagnoses most commonly associated with oral antibiotic prescription by dermatologists, as well as the duration of this use.
Design, Setting, and Participants
Repeated cross-sectional analysis of antibiotic prescribing by dermatologists from January 1, 2008, to December 31, 2016. The setting was Optum Clinformatics Data Mart (Eden Prairie, Minnesota) deidentified commercial claims data. Participants were dermatology clinicians identified by their National Uniform Claim Committee taxonomy codes, and courses of oral antibiotics prescribed by these clinicians were identified by their National Drug Codes.
Claims for oral antibiotic prescriptions were consolidated into courses of therapy and associated with the primary diagnosis from the most recent visit. Courses were stratified into those of extended duration (>28 days) and those of short duration (≤28 days).
Main Outcomes and Measures
Frequency of antibiotic prescribing and associated diagnoses. Poisson regression models were used to assess for changes in the frequency of antibiotic prescribing over time.
Between 2008 and 2016 among 985 866 courses of oral antibiotics prescribed by 11 986 unique dermatologists, overall antibiotic prescribing among dermatologists decreased 36.6% (1.23 courses per 100 visits) from 3.36 (95% CI, 3.34-3.38) to 2.13 (95% CI, 2.12-2.14) courses per 100 visits with a dermatologist (prevalence rate ratio for annual change, 0.931; 95% CI, 0.930-0.932), with much of this decrease occurring among extended courses for acne and rosacea. Oral antibiotic use associated with surgical visits increased 69.6% (2.73 courses per 100 visits) from 3.92 (95% CI, 3.83-4.01) to 6.65 (95% CI, 6.57-6.74) courses per 100 visits associated with a surgical visit (prevalence rate ratio, 1.061; 95% CI, 1.059-1.063).
Conclusions and Relevance
Continuing to develop alternatives to oral antibiotics for noninfectious conditions, such as acne, can improve antibiotic stewardship and decrease complications from antibiotic use. In addition, the rising use of postoperative antibiotics after surgical visits is concerning and may put patients at unnecessary risk of adverse events. Future studies are needed to identify the value of this practice and the risk of adverse events.
Antibiotic resistance is a growing concern both for the effectiveness of therapies for dermatologic disease and for the treatment of infectious diseases.1-4 Dermatologists prescribe more oral antibiotic courses per clinician than any other specialty, and many of these courses of antibiotics are prescribed for several months in duration.5-8 Oral antibiotics are frequently used for acne, rosacea, and other inflammatory conditions due to their potential anti-inflammatory properties.9-15 In addition, dermatologists also prescribe perioperative and postoperative oral antibiotics to prevent surgical complications.
This antibiotic use can have clinical consequences, including the development of antimicrobial resistance.16 Oral antibiotic therapy in the treatment of acne is associated with disruption of the normal oropharyngeal flora and resultant pharyngitis.17-21 The use of tetracycline-class antibiotics may also be associated with the development of inflammatory bowel disease and collagen vascular diseases.22-24 Furthermore, chronic antibiotic use has been linked in some studies25,26 to an increased risk of colon and breast cancer, which is thought to be mediated through disruption of the microbiome. As a result, there have been calls to reduce antibiotic use throughout medicine. Multiple clinical guidelines for acne recommend reducing antibiotic use through nonantimicrobial therapies and by limiting the duration of antibiotic therapy.4,27-32
While antibiotics are prescribed for a variety of conditions in dermatology, the frequency and duration of the use of oral antibiotics by dermatologists for conditions other than acne have not been well characterized. In addition, because there are few randomized clinical trials demonstrating efficacy of antibiotic therapy and even fewer comparing the effectiveness of antibiotic and nonantibiotic therapies in dermatology, identifying the most frequent areas of use can guide future studies to evaluate optimal prescribing practices in these settings. The objective of this study was to characterize the temporal trends in the diagnoses most commonly associated with oral antibiotic use by dermatologists, as well as the duration of this use.
This study was a repeated cross-sectional analysis of antibiotic prescribing by dermatologists from January 1, 2008, to December 31, 2016, using Optum Clinformatics Data Mart (Eden Prairie, Minnesota) commercial claims data. This source includes deidentified administrative commercial claims data for approximately 12 to 14 million privately insured patients annually in the United States. The patient population available in the data source is similar to the demographics of the US population with respect to sex, age, and geographic distribution.33 These data include both medical and pharmacy claims, as well as patient demographic information, such as age and sex. This study was deemed exempt from review and approval by the Institutional Review Board of the University of Pennsylvania. Informed consent of participants was not applicable because this was a claims database study.
Dermatology clinicians were identified by their National Uniform Claim Committee taxonomy codes. The analysis was limited to courses of oral antibiotics, with prescriptions identified by their National Drug Codes. Claims for oral antibiotic prescriptions were consolidated into courses of therapy, with the start date defined as the date of the first prescription of the series and the end date defined as the date of the last prescription in the series, plus the number of days of medication supplied. To account for potential delays between prescriptions due to nonadherence, prior authorizations, or other factors, prescriptions separated by less than 30 days were considered to be part of the same course of therapy.6-8,34,35
Courses of therapy were stratified into those of extended duration (>28 days) and those of short duration (≤28 days). This cutoff was chosen to attempt to separate prescriptions that may be given for acute infections (eg, Lyme disease) from those for more chronic dermatologic diseases (eg, acne and hidradenitis suppurativa). These courses of therapy were then associated with the primary diagnosis from the most recent clinic visit to a dermatologist before the date when the prescription was filled by the patient. For visits with multiple diagnoses, a tier system was used to associate the antibiotic prescription with the most likely diagnosis for which it was prescribed (eTable in the Supplement). The diagnosis in the highest tier was considered the primary diagnosis. Surgical visits were defined by Current Procedural Terminology codes for a destruction, excision, repair, or Mohs surgery. When multiple diagnoses from the same tier were present, the first diagnosis coded was chosen as the primary diagnosis. Prescriptions filled more than 28 days after the most recent visit with a dermatologist were excluded.
To account for varying frequency of dermatology encounters between years, prescription counts were divided by the number of dermatology visits with the associated diagnosis to calculate annual rates per 100 visits. Poisson regression models were used to assess for changes in the frequency of antibiotic prescribing over time. When appropriate, data are summarized using descriptive statistics. Statistical analyses were performed using a software program (Stata, version 14; StataCorp LP).
Between 2008 and 2016, there were 985 866 courses of oral antibiotics prescribed by 11 986 unique dermatologists. The median time between the date of the associated visit and the date on which the prescription was filled was 0 days (interquartile range, 0-2 days). The most commonly prescribed antibiotics were doxycycline hyclate (26.3%), minocycline (25.8%), and cephalexin (19.9%).
Between 2008 and 2016, overall antibiotic prescribing among dermatologists decreased 36.6% (1.23 courses per 100 visits) from 3.36 (95% CI, 3.34-3.38) to 2.13 (95% CI, 2.12-2.14) courses per 100 visits with a dermatologist (Figure 1), with a prevalence rate ratio (PRR) estimating an annual change in prescribing rate of 0.931 (95% CI, 0.930-0.932). For courses of extended duration, prescribing decreased 53.2% (1.30 courses per 100 visits) from 2.45 (95% CI, 2.43-2.47) to 1.15 (95% CI, 1.14-1.16) courses per 100 visits with a dermatologist (PRR, 0.900; 95% CI, 0.899-0.901). For courses of short duration, prescribing increased 8.4% (0.07 courses per 100 visits) from 0.91 (95% CI, 0.90-0.92) to 0.98 (95% CI, 0.98-0.99) courses per 100 visits with a dermatologist (PRR, 1.018; 95% CI, 1.014-1.023 for 2008-2011 and PRR, 0.990; 95% CI, 0.987-0.993 for 2012-2016).
In 2016, a total of 57.5% of prescriptions for an extended course of antibiotic therapy were associated with acne, and 13.5% of prescriptions for an extended course of antibiotic therapy were associated with rosacea. Adjusted for the frequency of the diagnosis, acne, rosacea, and hidradenitis suppurativa were the diagnoses most commonly associated with extended courses of antibiotic therapy (Figure 2A). Between 2008 and 2016, prescribing associated with acne decreased 28.1% (3.31 courses per 100 visits) from 11.76 (95% CI, 11.65-11.86) to 8.45 (95% CI, 8.36-8.54) courses per 100 visits with a diagnosis of acne (PRR, 0.960; 95% CI, 0.959-0.961). Prescribing associated with rosacea decreased 18.1% (1.97 courses per 100 visits) from 10.89 (95% CI, 10.67-11.11) to 8.92 (95% CI, 8.73-9.11) courses per 100 visits with a diagnosis of rosacea (PRR, 1.086; 95% CI, 1.073-1.100 for 2008-2011 and PRR, 0.927; 95% CI, 0.924-0.931 for 2012-2016). Prescribing associated with hidradenitis suppurativa increased by 3.2% (0.27 courses per 100 visits) from 8.75 (95% CI, 7.75-9.74) to 9.02 (95% CI, 8.41-9.63) courses per 100 visits with a diagnosis of hidradenitis (PRR, 1.137; 95% CI, 1.080-1.197 for 2008-2011 and PRR, 0.958; 95% CI, 0.947-0.970 for 2012-2016).
Skin and soft-tissue infections and surgical visits were the most common diagnoses associated with short courses of antibiotic therapy. Adjusted for the frequency of the diagnosis, skin and soft-tissue infections, surgical visits, hidradenitis suppurativa, and cysts were the most common diagnoses associated with short courses of antibiotic therapy (Figure 2B). Between 2008 and 2016, oral antibiotic use associated with surgical visits increased 69.6% (2.73 courses per 100 visits) from 3.92 (95% CI, 3.83-4.01) to 6.65 (95% CI, 6.57-6.74) courses per 100 visits associated with a surgical visit (PRR, 1.061; 95% CI, 1.059-1.063). Prescribing associated with cysts increased 35.3% (0.44 courses per 100 visits) from 1.24 (95% CI, 1.17-1.31) to 1.68 (95% CI, 1.62-1.74) courses per 100 visits associated with a diagnosis of a cyst (PRR, 1.004; 95% CI, 0.999-1.008).
Throughout the study period, course duration remained similar across major diagnostic categories, including acne, rosacea, surgical visits, and cysts (Table). The most commonly prescribed antibiotics associated with a diagnosis of acne were doxycycline hyclate, minocycline, and extended-release minocycline. The most commonly prescribed antibiotics associated with a diagnosis of rosacea were doxycycline hyclate, extended-release doxycycline, and minocycline. The most commonly prescribed antibiotics associated with surgical visits were cephalexin and doxycycline hyclate, and the median duration was 7 days (interquartile range, 5-10 days) for these courses (Figure 3).
During the study period, prescribing of brand-name, extended-release preparations of minocycline for acne decreased from 21.3% to 17.1%. Similarly, prescribing of brand-name, extended-release preparations of doxycycline for rosacea decreased from 24.6% to 17.0%. In both cases, prescribing for doxycycline monohydrate increased. For hidradenitis suppurativa, the use of clindamycin increased during the study period from 3.0% to 8.8% and for rifampin from 3.8% to 5.9% (Figure 3).
This repeated cross-sectional analysis demonstrated that antibiotic prescribing among dermatologists has substantially decreased over the past decade. In the context of the 39 million annual office visits to dermatologists,36 this absolute change of 1.23 courses per 100 visits is estimated to result in almost 480 000 fewer antibiotic courses per year being prescribed by dermatologists in 2016 than in 2008. Given that dermatologists were identified in 2013 as the most frequent prescribers of oral antibiotics per clinician,5 this decreased overall use is encouraging. While the prescribing frequency for antibiotic courses of extended duration (>28 days) decreased substantially during the study period, there has been growth in the use of antibiotic courses of short duration (≤28 days), particularly among those associated with surgical visits.
Much of the decrease in extended courses of antibiotic therapy is associated with visits for acne and rosacea. Although recent guidelines27 suggest limiting the duration of therapy in this patient population, course duration has remained stable over time, suggesting that this decrease may be due to fewer patients being treated with antibiotics rather than patients being treated for a shorter duration.29,34 Another possible cause of decreased antibiotic use may be patient preference. Given increasing concerns about complications associated with antibiotic use, patients may be more reluctant to opt for extended courses of antibiotics. It is also possible that improved topical treatments and increased use of alternative systemic treatments may have decreased reliance on oral antibiotics for the treatment of these conditions. For instance, there is increasing evidence to support the use of spironolactone as an alternative to oral antibiotics for women with acne, and the use of spironolactone has grown in recent years.8,37-40
The increasing use of postoperative antibiotics associated with visits for surgical procedures is concerning. The risk of surgical site infections resulting from dermatologic procedures, including Mohs surgery, is low. Procedures in the groin, skin grafts, wedge excisions of the lip or ear, and procedures below the knee may be associated with higher surgical site infection risk, and single-dose perioperative antibiotics may help decrease the risk of surgical site infection for these higher-risk cases.41-43 However, multiple prospective studies of clean-contaminated head and neck procedures, including those that breach the mucosa, have not identified increased efficacy between regimens of 24 hours and longer regimens of 3 to 7 days, although these studies were not conducted in an office-based surgery setting.44-47 In addition, a 2008 advisory statement on antibiotic prophylaxis recommends single-dose perioperative antibiotics for patients at increased risk of surgical site infection.48 Finally, guidelines from the American Heart Association49 and the American Academy of Orthopedic Surgeons50 recommend limited use of single-dose perioperative antibiotics for the prevention of infective endocarditis and joint infections, respectively, and neither guideline supports prolonged courses of postoperative antibiotics.
While guideline recommendations generally do not support extended postoperative courses of oral antibiotics, a 2012 survey51 sent to American College of Mohs Surgery members identified that many antibiotic prescribing practices reported by surgeons were not aligned with guideline recommendations and concluded that dermatologic surgeons prescribe more antibiotics than needed for infection prevention. In addition, only about 70% of surgeons reported that they were familiar with the 2007 guidelines of the American Heart Association48 and the 2008 advisory statement on antibiotic prophylaxis in dermatologic surgery.51
It is estimated that approximately 1 in 1000 oral antibiotic prescriptions results in an emergency department visit for associated complications.52 Overuse of oral antibiotics is also associated with significant changes to the microbiome, and prior exposure to antibiotics is also associated with increased risk for the development of Clostridium difficile and antibiotic-resistant infections.53,54 Given the low rate of infectious complications, even for Mohs surgery, and the lack of evidence to support the use of prolonged rather than single-dose perioperative regimens, the postoperative courses of antibiotics identified in this study may increase risks to patients without substantial benefits. Just as the shift from topical antibiotics to plain white petrolatum has improved outcomes at reduced cost for postoperative wound care, there may be an opportunity to optimize oral antibiotic prescribing in dermatologic surgery.55 Additional evidence, including data from well-controlled prospective studies, is needed to determine the appropriate role for perioperative and postoperative oral antibiotics for dermatologic procedures, particularly for Mohs surgery, in which the risk of postoperative complications may be higher and the morbidity of these complications is more significant.
For many of the conditions evaluated herein, the most common antibiotics prescribed remained constant throughout the study period. There was a decrease in prescribing of brand-name, extended-release preparations of doxycycline and minocycline for acne and rosacea, as well as a shift toward increased use of doxycycline monohydrate, which may be related to the high cost of these medications and the dramatic price increases for generic doxycycline hyclate56 that occurred around 2012. There has also been an increase in the use of clindamycin and rifampin among courses of antibiotics associated with visits for hidradenitis suppurativa. The results of 2 retrospective case series57,58 suggested that such treatment can be effective for patients with hidradenitis suppurativa. Given the growth in the use of these antibiotics, additional prospective controlled studies or comparative effectiveness studies may be warranted to better characterize the efficacy of this treatment regimen for patients with hidradenitis suppurativa. There has also been an increase in the use of tetracycline-class antibiotics associated with surgical visits. While the underlying factors related to this shift are unclear, it is possible that the increased prescribing may be related to efforts to improve scar cosmesis through inhibition of matrix metalloproteases by tetracycline antibiotics.59 However, evidence to support this practice is limited. A 2007 study60 reported smaller mean scar size in 4 rabbits randomized to receive a maximum subtoxic dose of minocycline than in 4 control rabbits, but clinical studies supporting the effectiveness of oral tetracycline-class antibiotics to improve scar cosmesis are lacking.59
The results of this study should be interpreted in the context of the study design. Because dermatology visits may have multiple diagnosis codes and antibiotic courses may be prescribed for reasons other than the diagnoses coded at the visit, there is the possibility for misclassification herein with respect to the associated diagnoses for the antibiotic course. However, the antibiotics were prescribed by dermatologists, and by using a tier system to identify the most relevant diagnoses (eg, for a visit with diagnoses coded for cellulitis, actinic keratosis, and a surgical visit, the prescription would be associated with cellulitis), we have attempted to reduce the influence of this potential source of bias. Since the International Classification of Diseases, Ninth Revision was replaced with the International Classification of Diseases and Related Health Problems, Tenth Revision, Clinical Modification in 2015, it is important to consider coding differences when examining trends across these periods. These coding changes may help explain the decrease in antibiotic prescribing rates for skin and soft-tissue infections in 2015 and 2016; for more commonly associated diagnoses, such as acne, rosacea, hidradenitis suppurativa, and surgical visits, there were no significant changes in coding between revisions. Finally, because perioperative antibiotics administered in the office may not generate an associated pharmacy claim, we were unable to examine trends in the in-office use of perioperative antibiotics.
While dermatologists were once the most frequent prescribers of antibiotics per clinician, the prescribing of antibiotics by dermatologists is declining, particularly for extended courses of antibiotics given to patients with chronic dermatologic conditions, such as acne and rosacea. Opportunities may exist to improve antibiotic stewardship further, and the Centers for Disease Control and Prevention has developed a framework for improved antibiotic stewardship in the outpatient setting.61 There is rising use of prolonged postoperative courses of antibiotics associated with visits for surgical procedures, which may put patients at unnecessary risk of adverse events given the available evidence and guideline recommendations. Future studies are needed to identify the value of this practice with respect to patient outcomes and antibiotic stewardship.
Accepted for Publication: November 6, 2018.
Corresponding Author: John S. Barbieri, MD, MBA, Department of Dermatology, University of Pennsylvania Perelman School of Medicine, Perelman Center for Advanced Medicine South Tower 7, 3400 Civic Center Blvd, Philadelphia, PA 19104 (firstname.lastname@example.org).
Published Online: January 16, 2019. doi:10.1001/jamadermatol.2018.4944
Author Contributions: Dr Barbieri had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Barbieri, Bhate, Hartnett, Margolis.
Acquisition, analysis, or interpretation of data: Barbieri, Fleming-Dutra, Margolis.
Drafting of the manuscript: Barbieri, Bhate, Margolis.
Critical revision of the manuscript for important intellectual content: Barbieri, Hartnett, Fleming-Dutra, Margolis.
Statistical analysis: Barbieri, Margolis.
Obtained funding: Margolis.
Administrative, technical, or material support: Barbieri, Bhate, Fleming-Dutra.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was funded in part by award 1P30AR069589-01 from the National Institute of Arthritis and Musculoskeletal Skin Diseases (NIAMS). Dr Barbieri is supported by award T32-AR-007465 from the NIAMS and receives partial salary support through a Pfizer Fellowship Grant to the Trustees of the University of Pennsylvania.
Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Meeting Presentation: This study was presented as a poster at the International Investigative Dermatology 2018 Meeting; May 18, 2018; Orlando, Florida.
Create a personal account or sign in to: