Antibiotic Prescribing Patterns for Sinusitis Within a Direct-to-Consumer Virtual Urgent Care

Introduction

Telemedicine services are experiencing dramatic growth throughout the healthcare sector. There are numerous drivers for this growth, including the economic necessity for cost containment, patient demand for more convenient care, healthcare systems' need for demand smoothing, and a merger-driven centralization of healthcare resources. Although aggregate telemedicine utilization data are lacking in the academic environment, industry experts estimate that 7 million patients will use telemedicine in 2018. ¹ Historically, the majority of telemedicine visits have been hosted through private for-profit companies such as Doctor on Demand™ or American Well™. However, telemedicine capacity is also on the rise within academic healthcare, as evidenced by exponential growth in telemedicine publications. ² Such dramatic growth can be disruptive and destabilizing, and concerns have arisen that telemedicine may not meet expected quality standards, especially in a direct-to-consumer model. ³

A virtual urgent care direct-to-consumer platform (termed VUC) offers quick, affordable and accessible healthcare for minor conditions such as sore throat, urinary tract infections, conjunctivitis, colds, and sinusitis—many of which can easily be treated with a course of antibiotics. Yet in an age of worsening antibiotic resistance, antibiotic stewardship and nonuse practice guidelines are a critical pillar for any quality-oriented care program. ⁴ Previously published data suggest that virtual visits, either through video or questionnaire, may be associated with increased antibiotic utilization. ^3,5 If telemedicine is to reach its potential to transform healthcare, attention must be given to quality initiatives such as antibiotic stewardship.

This was a descriptive study with the goal of comparing antibiotic utilization for acute sinusitis among patients treated in the VUC with those treated in traditional “brick and mortar” urgent care (TUC) within the same health system. In addition, we sought to explore factors that drove antibiotic utilization for sinusitis when care was delivered virtually.

Methods

Results

Four hundred twenty-two (n = 422) patients were treated in the UCHealth VUC between September 1, 2016 and September 11, 2017. Of these, 57 patients were given a diagnosis of acute sinusitis. A sample of 100 sequential patients, treated for acute sinusitis within the same time window in a TUC, was used as a comparison. This cohort of patients treated for acute sinusitis formed the primary sample for analysis. Characteristics of the primary sample are given in Table 1.

Of the 57 patients treated for acute sinusitis within the virtual urgent care, 39 patients were prescribed an oral antibiotic (67%). This included amoxicillin (13%), amoxicillin–clavulanate (28%), azithromycin (51%), doxycycline (4%), and levofloxacin (4%). Of the 100 patients treated for acute sinusitis at TUC, 92 patients were given an oral antibiotic (92%): amoxicillin (26%), amoxicillin–clavulanate (16%), azithromycin (25%), cefdinir (1%), clarithromycin (2%), doxycycline (27%), levofloxacin (2%), or minocycline (1%). Figure 1 shows a lower rate of antibiotic utilization for acute sinusitis within the VUC than within TUC (p < 0.001).

OPEN IN VIEWER

Fig. 1.

Antibiotic use for acute sinusitis.

Age and sex were not associated with the use of an oral antibiotic (p = 0.781, p = 0.555) when care was delivered virtually for acute sinusitis. In addition, whether a patient was evaluated by telephone or by video was not associated with oral antibiotic utilization (p = 0.978).

Discussion

In this study, we demonstrate that antibiotic utilization for acute sinusitis was lower when care was delivered through a direct-to-consumer virtual urgent care model than when delivered through traditional urgent care. Antibiotic stewardship is a critical factor in the delivery of quality medical care. Telehealth offerings are growing at a rapid pace in both the for-profit and nonprofit healthcare sectors. This growth promises to be disruptive to traditional care delivery models. However, medical care delivered virtually will not succeed if it is substandard, especially so if it is concomitantly cast as innovative. Previous studies have suggested that antibiotic utilization is higher when care is delivered virtually. Uscher-Pines et al. found that antibiotic utilization rates for pharyngitis and bronchitis were higher when care was delivered through a for-profit direct-to-consumer virtual visit than when care was delivered through in-person office visits. ⁶ In a follow-up study, these same authors found higher rates of antibiotic utilization for bronchitis when care was delivered virtually. ³ In a study of e-visits, defined by the authors as secure online questionnaires submitted asynchronously through a medical portal, antibiotic utilization was higher for sinusitis than during office visits. ⁵

Lowering antibiotic utilization for acute sinusitis has been identified as an important priority given that most acute sinusitis is viral in etiology. ⁷ Previous studies demonstrate antibiotic utilization rates ranging from 57% to 94%. ^5,8,9 High rates of antibiotic utilization for acute sinusitis are likely driven by several factors. First, there is no gold standard test to differentiate between viral and bacterial sinusitis. Second, past high antibiotic usage may have deleteriously shifted current patient expectations regarding the appropriateness of antibiotics for any acute sinus infection. However, these same challenges frame a virtual care environment ideally suited for study. Most VUC visits occur without the benefit of diagnostic testing; in the case of acute sinusitis, this does not prohibit the clinician from making an appropriate diagnosis. Also, management of patient expectations is entirely reasonable through a live interactive video platform.

In this article, we demonstrate that antibiotic utilization was lower when care was delivered virtually than when care was delivered by traditional urgent care. In the most general terms, this suggests that quality medical care, at least measured through antibiotic utilization, is possible when care is delivered virtually. Why was antibiotic utilization lower in this study? Antibiotic utilization is undoubtedly driven by a multitude of factors. Provider training level, practice culture, specialty, patient expectations, and care delivery modality have been identified as likely contributors. ¹⁰ The relative strength of each of these contributors requires further investigation. In this study, virtual urgent care providers were board-certified or board-eligible emergency physicians practicing in an academic setting. These providers may have been more aware of specific institutional antibiotic prescribing guidelines. In contrast, the traditional urgent care providers were a more heterogeneous mix of providers dominated by advanced practice providers. It is certainly possible that differences in the level of provider training dwarfed any effect driven by care modality. However, it is a heartening observation that care delivered virtually for sinusitis does not implicitly come at the price of increased antibiotic utilization.

In previous studies, age has been associated with higher antibiotic utilization for management of acute sinusitis. ⁸ However, neither age nor gender was associated with increased antibiotic utilization within the virtual care cohort. However, the small sample size of this cohort does limit the generalizability of this observation. In addition, whether the virtual visit was conducted by phone or live interactive video appeared to have no relationship on antibiotic utilization. Historically, the bulk of virtual visits have been conducted by phone instead of video. Ubiquitous smart phones will likely change this trend. In theory, a video visit allows the provider to more easily determine “sick vs. not sick” when compared with phone visits. This is especially important given that most direct-to-consumer virtual visits occur without the benefit of vital signs. Anecdotally, older emergency physicians within this cohort were more comfortable delivering care by video than by phone. Future studies should explore provider preference and determine where objective value is added when care is delivered by video rather than by telephone.

Five different antibiotics were used to treat acute sinusitis in the virtual care cohort. In the TUC cohort, eight antibiotics were used, suggesting increased variation in care management. It is possible that much of this practice variability is driven by the fact that there were more visits (and more providers) within the TUC arm (100 vs. 57), which allowed more opportunity for variation. Nonetheless, it is worth exploring in future studies whether care variation can be limited through provider training or clinical decision support. It is also worth noting that the use of azithromycin is inappropriately prevalent in both cohorts; azithromycin is not a first-line antibiotic for treatment of acute sinusitis as recommended by the Infectious Disease Society of America. ⁴ There is likely an opportunity for clinical decision support to lower azithromycin utilization for acute sinusitis, although this requires further study.

Limitations

The sample size of this study was relatively small, which may have led to an increased risk of type 2 error, especially in the observations that age, sex, and care delivery modality were not associated with antibiotic utilization. In addition, finding an appropriate control group to compare against the VUC cohort was a challenge. For the purposes of this study, the authors had to make a determination on whether to optimize homogeneity in the patient population or in the provider population. The virtual care cohort in this study used academic emergency physicians. However, comparing antibiotic utilization with brick and mortar emergency departments (EDs) visits for acute sinusitis would have likely introduced selection bias, as the patients presenting to an ED would likely have been sicker. In this study, provider heterogeneity between the virtual and traditional care groups may have acted as a confounder.

Conclusions

Antibiotic utilization for acute sinusitis was lower when care was delivered through a direct-to-consumer virtual urgent care model than when delivered through to traditional urgent care. When care was delivered virtually, age, sex, and care modality (telephone vs. video) were not associated with antibiotic utilization for acute sinusitis. This suggests that it is possible to deliver virtual care without increasing antibiotic utilization. Further study is required to explore determinants of antibiotic utilization in virtual care environments.