Trends in US Outpatient Antibiotic Prescriptions During the Coronavirus Disease 2019 Pandemic

Abstract

Background

The objective of our study was to describe trends in US outpatient antibiotic prescriptions from January through May 2020 and compare with trends in previous years (2017–2019).

Methods

We used data from the IQVIA Total Patient Tracker to estimate the monthly number of patients dispensed antibiotic prescriptions from retail pharmacies from January 2017 through May 2020. We averaged estimates from 2017 through 2019 and defined expected seasonal change as the average percent change from January to May 2017–2019. We calculated percentage point and volume changes in the number of patients dispensed antibiotics from January to May 2020 exceeding expected seasonal changes. We also calculated average percent change in number of patients dispensed antibiotics per month in 2017–2019 versus 2020. Data were analyzed overall and by agent, class, patient age, state, and prescriber specialty.

Results

From January to May 2020, the number of patients dispensed antibiotic prescriptions decreased from 20.3 to 9.9 million, exceeding seasonally expected decreases by 33 percentage points and 6.6 million patients. The largest changes in 2017–2019 versus 2020 were observed in April (–39%) and May (–42%). The number of patients dispensed azithromycin increased from February to March 2020 then decreased. Overall, beyond-expected decreases were greatest among children (≤19 years) and agents used for respiratory infections, dentistry, and surgical prophylaxis.

Conclusions

From January 2020 to May 2020, the number of outpatients with antibiotic prescriptions decreased substantially more than would be expected because of seasonal trends alone, possibly related to the coronavirus disease 2019 pandemic and associated mitigation measures.

antibiotic, antibiotic stewardship, outpatient, COVID-19

The coronavirus disease 2019 (COVID-19) pandemic and related mitigation measures, such as school closures and stay-at-home orders, substantially affected healthcare- and medication-seeking patterns throughout the United States [1–4]. These measures also potentially affected transmission of infections commonly managed with outpatient antibiotics. Antibiotics are among the most commonly prescribed medications in US outpatient healthcare, and antibiotic stewardship is a focus of national- and state-level efforts, underscoring the importance of evaluating impacts of the COVID-19 pandemic on outpatient antibiotic prescribing. Prescribing of azithromycin, which was under exploration for treatment of COVID-19, initially increased during the pandemic [2, 3]; however, examination of prescribing for other antibiotics during the COVID-19 pandemic has been limited. Outpatient antibiotic use varies by season and is typically highest in winter, likely related to the incidence of respiratory diseases [5], with decreases during the spring. However, effects of the COVID-19 pandemic, including associated mitigation measures and changes in healthcare delivery, on outpatient antibiotic use during spring 2020 is unknown. The objective of our analysis was to describe and compare trends in US outpatient antibiotic prescriptions between January 2020 and May 2020 (the first several months of the COVID-19 pandemic) with historic trends to assess how antibiotic prescribing changed during the pandemic.

METHODS

Data Source and Study Population

We used 2017–2020 data from IQVIA Total Patient Tracker (TPT), a proprietary IQVIA dataset, to estimate the number of unique patients dispensed antibiotic prescriptions from retail pharmacies in the United States. IQVIA collects data from approximately 48 900 US retail pharmacies, representing more than 3.5 billion transactions annually and covering 92% of all retail prescriptions, and generates national projections of numbers of unique, individual patients receiving dispensed prescriptions. IQVIA TPT data are updated monthly and are available approximately 1 month after collection, with data available through May 2020 at the time of this analysis. IQVIA TPT contains data on drug active ingredient, drug class, patient age, patient sex, prescriber specialty, and state. These data have previously been used for regulatory oversight [6], public health surveillance [7, 8] and studies of medication dispensing trends [3]. We included only systemic (oral or parenteral) antibiotics and excluded antibiotics prescribed by veterinarians. IQVIA TPT data were accessed through an online portal.

We categorized antibiotic prescriptions by active ingredient and therapeutic class. We categorized patient age as: ≤19 years, 20 to 64 years, and ≥65 years. We further categorized the ≤19 years group into ≤4 years and 5 to 19 years to examine differences in trends in younger and older children.

Data Analysis

We estimated the number of patients dispensed antibiotic prescriptions each month by agent, therapeutic class, and patient age group. Although estimates for all antibiotics were not available at the state and specialty level, select antibiotic information was available, thus we estimated the number of patients dispensed prescriptions for commonly used antibiotics by prescriber specialty (for amoxicillin, amoxicillin-clavulanate, azithromycin, cefdinir, and cephalexin) and by state (for amoxicillin and azithromycin). We chose 1 month from each season, January (winter) and May (spring), to evaluate seasonal changes. We averaged the number of patients dispensed antibiotic prescriptions in 2017–2019 to create a historic baseline before the pandemic for comparison with estimates from 2020. We calculated the following endpoints for each stratum (see Supplemental Table 1): (1) average percent change in the number of patients dispensed antibiotic prescriptions in 2017–2019 versus 2020 for each month; (2) change in the number of patients dispensed antibiotic prescriptions from January to May for each time period (2017–2019 average and 2020); and (3) percent change in the number of patients dispensed antibiotic prescriptions from January 2020 to May 2020 for each time period.

Using these values, we calculated additional endpoints to evaluate changes in 2020 compared with what would be seasonally expected. We considered the percent change in average numbers of patients dispensed antibiotic prescriptions from January to May in 2017–2019 to be the expected seasonal change. To estimate changes in the number of patients dispensed antibiotic prescriptions beyond expected seasonal changes, we calculated the difference between the 2017–2019 average percent change and the 2020 observed percent change and reported this difference as the additional percentage change. For example, a 15% average decrease from January 2017–2019 to May 2017–2019 and a 50% decrease from January 2020 to May 2020, would result in an additional percentage change of –35 percentage points. We estimated seasonally expected changes in numbers of patients dispensed antibiotic prescriptions by multiplying the observed January 2020 numbers of patients dispensed prescriptions by the seasonally expected (2017–2019 average) January to May percent change. We estimated the number of patients dispensed antibiotic prescriptions above or below seasonal expectation by subtracting the seasonally expected change from the observed change in the number of patients dispensed antibiotic prescriptions from January to May 2020.

For comparisons among age groups and states, we used US Census population projections [9] for 2017–2020 to calculate rates of antibiotic dispensing per 1000 persons.

The IQVIA TPT online portal was used to generate reports that were further analyzed using SAS 9.4 (SAS Institute, Cary, NC). Analyses were conducted by the US Centers for Disease Control and Prevention (CDC) as part of public health surveillance activities. This activity was reviewed by CDC and was conducted consistent with applicable federal law and CDC policy (See eg, 45 C.F.R. part 46.102(l)(2), 21 C.F.R. part 56; 42 U.S.C. §241(d); 5 U.S.C. §552a; 44 U.S.C. §3501 et seq.)

RESULTS

From January 2020 to May 2020, the number of patients dispensed antibiotic prescriptions decreased from 20.3 million to 9.9 million, which is 6.6 million fewer than seasonally expected, representing an additional decrease of 33 percentage points beyond the expected seasonal decline (Table 1). Monthly numbers of patients with antibiotic prescriptions in 2020 were only slightly lower than 2017–2019 averages in January through March, with decreases of 4% in January and February and 9% in March (Figure 1). Much larger differences were seen in April and May; the number of patients with prescriptions were 39% and 42% lower, respectively, than 2017–2019 averages for these months.

Table 1.

Estimated Number of Patients with Antibiotic Prescriptions Dispensed from Retail Pharmacies and Beyond Seasonally Expected Changes by Antibiotic Class and Agent, United States, January 2017 Through May 2020^a

	2017–2019 Average				2020 Observed				2020 January to May 2020 Change Beyond Seasonally Expected
	No. Patients Dispensed Antibiotic Prescriptions, in Millions		January to May Change		No. Patients Dispensed Antibiotic Prescriptions, in Millions		January to May Change
Antibiotic Class and Selected Agentsb	January	May	No., in Millions	%c	January	May	No., in Millions	%c	Additional Percentaged	No., in Millionse
Total	21.1	17.2	−3.9	−19	20.3	9.9	−10.4	−51	−33	−6.6
Penicillins	5.7	4.7	−1.0	−17	5.6	2.3	−3.3	−59	−42	−2.3
Amoxicillin	5.3	4.3	−0.9	−18	5.2	2.0	−3.2	−61	−43	−2.2
Penicillin V	0.4	0.4	0.0f	−6	0.3	0.2	−0.1	−33	−27	−0.1
Macrolides	4.9	3.0	−1.9	−39	4.3	1.2	−3.1	−73	−34	−1.5
Azithromycin	4.7	2.9	−1.9	−39	4.1	1.1	−3.0	−74	−34	−1.4
Cephalosporins	3.2	2.9	−0.3	−9	3.3	1.9	−1.4	−42	−33	−1.1
Cephalexin	1.6	1.8	0.1	8	1.7	1.4	−0.3	−18	−26	−0.4
Cefdinir	1.1	0.8	−0.3	−26	1.1	0.3	−0.8	−72	−46	−0.5
Extended spectrum beta−lactams	2.9	2.1	−0.8	−27	3.0	1.1	−1.9	−62	−35	−1.1
Amoxicillin-clavulanate	2.9	2.1	−0.8	−27	3.0	1.1	−1.9	−62	−35	−1.1
Fluoroquinolones	2.1	1.7	−0.4	−17	1.6	1.0	−0.5	−35	−18	−0.3
Ciprofloxacin	1.2	1.1	0.0f	−3	0.9	0.8	−0.2	−20	−17	−0.2
Levofloxacin	0.9	0.6	−0.3	−35	0.6	0.3	−0.4	−56	−21	−0.1
Tetracyclines	2.0	1.9	−0.1	−5	2.4	1.5	−0.8	−35	−30	−0.7
Doxycycline	1.7	1.6	−0.1	−6	2.1	1.3	−0.8	−38	−31	−0.6
Sulfonamides and combinations	1.5	1.5	0.01	1	1.3	1.1	−0.2	−17	−18	−0.2
Sulfamethoxazole-trimethoprim	1.4	1.5	0.0f	1	1.3	1.1	−0.2	−18	−19	−0.2
Lincosamides	0.8	0.8	0.0f	3	0.7	0.6	−0.1	−20	−23	−0.2
Clindamycin	0.8	0.8	0.0f	3	0.7	0.6	−0.1	−20	−23	−0.2
Other	0.0f	0.1	0.0f	10	0.1	0.0f	0.0f	−23	−33	0.0f

^aData are from IQVIA Total Patient Tracker (January 2017–May 2020) and were accessed July 16, 2020.

^bClasses and agents do not sum to total as patients with ≥1 dispensed antibiotic prescription are only counted once in the total and not all agents are shown.

^cPercent change from January to May calculated as: (No. patients with antibiotic prescriptions in May – No. patients with antibiotic prescriptions in January)/No. patients with antibiotic prescriptions in January.

^dAdditional percentage change calculated as: Observed percent change from January to May in 2020 – Average percent change from January to May between 2017 and 2019.

^eChange beyond seasonally expected in the number of patients with antibiotic prescriptions calculated as: observed change in the number of patients with dispensed antibiotics prescriptions from January to May 2020 – (2017–2019 average percent change in the number of patients with dispensed antibiotic prescriptions from January to May × Number of patients with dispensed antibiotic prescriptions in January 2020).

^fNumber has an absolute value of <50 000 and is rounded to 0.0 million.

Figure 1.

Estimated percent change in the number of patients with antibiotic prescriptions dispensed from retail pharmacies by month, 2017–2019 versus 2020, United States. Only systemic antibiotics were included. Data are from IQVIA Total Patient Tracker (January 2017–May 2020) and were accessed July 16, 2020.

Larger than seasonally expected decreases from January 2020 to May 2020 in the number of patients dispensed antibiotic prescriptions occurred across antibiotic classes and agents (Table 1). The largest beyond-expected decreases were observed among the penicillin (2.3 million fewer patients), macrolide (1.5 million fewer patients), and cephalosporin (1.1 million fewer patients) classes. By agent, amoxicillin accounted for 2.2 million fewer patients than expected and 34% of the total beyond seasonally expected decrease in patients dispensed antibiotic prescriptions, whereas azithromycin accounted for 1.4 million fewer patients and 21% of the total beyond expected decrease. The largest additional percentage changes in patients dispensed antibiotic prescriptions (January to May 2020 compared with January to May 2017–2019 average) were observed for cefdinir (–46 percentage points) and amoxicillin (–43 percentage points). Although, overall, the number of patients dispensed azithromycin prescriptions decreased from January 2020 to May 2020, the number of patients dispensed azithromycin prescriptions increased by 5% from February 2020 to March 2020, followed by a 71% decrease from March 2020 to May 2020 (Figure 2). In March 2020, the number of patients dispensed azithromycin prescriptions was similar to that in 2017–2019; however, in May 2020 it was 62% lower than the historic average.

Figure 2.

Estimated number of unique patients with prescriptions dispensed from retail pharmacies, by month for (A) all antibiotics and (B) azithromycin, 2017–2019 versus 2020, United States. Data are from IQVIA Total Patient Tracker and were accessed July 16, 2020.

The observed 2020 January to May decrease in the number of patients dispensed amoxicillin and azithromycin prescriptions exceeded the expected seasonal decline across states. However, in March 2020, the number of patients dispensed azithromycin prescriptions exceeded the historical average in 11 states (Supplemental Table 2): New York (37% higher in 2020 than 2017–2019 average), New Jersey (32%), Florida (16%), Oklahoma (7%), Louisiana (6%), Georgia (5%), Alabama (5%), Texas (3%), Mississippi (3%), Arkansas (2%), and Idaho (1%). In April 2020, the number of patients dispensed azithromycin prescriptions was lower than the historical average in all states except New York and New Jersey, where 2020 numbers exceeded historical averages by 10%. In May 2020, the number of patients dispensed azithromycin prescriptions was lower than the historical average in all states. For amoxicillin, the number of patients dispensed prescriptions was lower in 2020 than the historical average in all states in March through May.

Decreases in the number of patients dispensed antibiotic prescriptions from January to May 2020 exceeded expected values across age groups (Figure 3). Standardized per 1000 persons, the beyond seasonally expected decrease among children was double the decrease in adults 20–64 years: –31.4 versus –14.4 patients with antibiotic prescriptions per 1000 persons. Decreases beyond expected were greater among children ≤4 years (–41.3 patients with prescriptions per 1000 persons) compared with children 5–19 years (–28.3 per 1000 persons). The decrease in the number of patients dispensed antibiotic prescriptions beyond expected was also higher among older adults ≥65 years (–21.7 patients with antibiotic prescriptions per 1000 persons) compared with adults 20–64 years (–14.4 per 1000 persons).

Figure 3.

Observed versus expected number of patients dispensed antibiotic prescriptions from retail pharmacies per 1000 persons by age group, January 2020 and May 2020, United States. Estimated numbers of unique patients dispensed antibiotic prescriptions from January 2020 and May 2020 standardized per 1000 persons. Population estimates are based on US Census population projections [9]. Only systemic antibiotics were included. Data are from IQVIA Total Patient Tracker and were accessed July 16, 2020.

Among patients dispensed amoxicillin prescriptions, the greatest decreases beyond seasonally expected declines occurred in patients treated by primary care prescribers (pediatrics, emergency medicine), gastroenterologists, nurse practitioners, and physician assistants (Table 2). A substantial decrease, –45 percentage points, was also observed in patients treated by orthopedic surgeons. Among patients dispensed cephalexin prescriptions, otolaryngologists and additional surgical specialties (plastic surgery: -31 percentage points and other surgery: –29 percentage points) were also among the specialties with the greatest declines beyond expected. Notably, percent changes in monthly numbers of patients with antibiotic prescriptions between 2017–2019 and 2020 decreased in March and April and then increased in May among patients treated by dentists, medical subspecialists, and surgeons whereas declines continued in May among primary care specialties (Supplemental Table 3).

Table 2.

Estimated Number of Patients With Antibiotic Prescriptions Dispensed From Retail Pharmacies and Beyond Seasonally Expected Changes by Select Antibiotic Agent and Prescriber Specialty, United States, January 2017 through May 2020^a

	2017–2019 Average				2020 Observed				2020 January to May Change Beyond Seasonally Expected
	No. Patients Dispensed Antibiotic Prescriptions		January to May Change		No. Patients Dispensed Antibiotic Prescriptions		January to May Change
Prescriber Specialtyb	January	May	No.	%c	January	May	No.	%c	Additional Percentaged	No.e
Amoxicillin
Pediatrics	886 936	657 138	−229 798	−26	804 580	106 388	−698 191	−87	−61	−489 731
Gastroenterology	17 587	18 573	986	6	17 442	7900	−9542	−55	−60	−10 520
Nurse practitioner	871 027	666 221	−204 807	−24	1 008 406	220 770	−787 636	−78	−55	−550 527
Physician assistants	506 634	389 337	−117 297	−23	532 135	117 931	−414 204	−78	−55	−291 003
Emergency medicine	148 624	115 199	−33 425	−22	133 292	35 893	−97 399	−73	−51	−67 422
Amoxicillin clavulanate
Pediatrics	266 003	205 517	−60 485	−23	236 923	49 449	−187 474	−79	−56	−133 601
Allergy	14 926	11 916	−3009	−20	13 750	4819	−8932	−65	−45	−6159
Nurse practitioner	669 411	470 824	−198 587	−30	828 517	236 059	−592 459	−72	−42	−346 671
Physician assistants	388 765	280 720	−108 045	−28	437 653	136 263	−301 391	−69	−41	−179 758
Otolaryngology	67 109	56 001	−11 108	−17	62 521	27 805	−34 716	−56	−39	−24 367
Azithromycin
Pediatrics	313 826	220 262	−93 564	−30	261 476	36 417	−225 060	−86	−56	−147 104
Infectious diseases	16 819	14 945	−1874	−11	16 613	6304	−10 308	−62	−51	−8457
Otolaryngology	29 703	22 319	−7383	−25	25 340	9143	−16 197	−64	−39	−9898
Nurse practitioner	909 422	568 311	−341 111	−38	968 808	238 143	−730 665	−75	−38	−367 279
Physician assistants	598 614	353 974	−244 640	−41	556 381	117 035	−439 347	−79	−38	−211 966
Cefdinir
Pediatrics	259 819	202 681	−57 138	−22	236 715	42 021	−194 694	−82	−60	−142 637
Nurse practitioner	252 457	187 013	−65 444	−26	324 093	74 670	−249 423	−77	−51	−165 408
Otolaryngology	29 238	25 146	−4091	−14	29 110	10 824	−18 286	−63	−49	−14 213
Allergy	5919	5013	−906	−15	5298	1910	−3388	−64	−49	−2577
Physician assistants	132 228	95 915	−36 313	−27	152 504	37 430	−115 074	−75	−48	−73 192
Cephalexin
Physician assistants	228 303	264 606	36 303	16	259 959	200 874	−59 086	−23	−39	−100 422
Otolaryngology	18 731	19 111	380	2	16 958	10 671	−6287	−37	−39	−6631
Nurse practitioner	236 914	276 944	40 030	17	303 127	240 432	−62 695	−21	−38	−113 913
Pediatrics	96 966	101 657	4691	5	90 005	62 297	−27 707	−31	−36	−32 062
Emergency medicine	122 163	140 905	18 742	15	129 980	108 445	−21 535	−17	−32	−41 477

^aData are from IQVIA Total Patient Tracker (January 2017–May 2020) and were accessed July 16, 2020.

^bSpecialties by agent do not sum to agent total as patients with ≥1 dispensed antibiotic prescription are only counted once in the total and only the top 5 specialties are shown for each antibiotic.

^cPercent change from January to May calculated as: (No. patients with antibiotic prescriptions in May – No. patients with antibiotic prescriptions in January)/No. patients with antibiotic prescriptions in January.

^dAdditional percentage change calculated as: Observed percent change from January to May in 2020 – Average percent change from January to May between 2017 and 2019.

^eChange beyond seasonally expected in the number of patients with antibiotic prescriptions calculated as: Observed change in the no. of patients with dispensed antibiotics prescriptions from January to May 2020 – (2017–2019 average percent change in the number of patients with dispensed antibiotic prescriptions from January to May × No. of patients with dispensed antibiotic prescriptions in January 2020).

DISCUSSION

From January to May 2020, more than 6 million fewer outpatients were dispensed antibiotic prescriptions from retail pharmacies than would be expected based on the same timeframe in prior years. Decreases were seen across all antibiotic classes and agents, with the greatest beyond seasonally expected declines among agents commonly prescribed for respiratory diseases, dentistry, and surgical prophylaxis. Temporal and geographic azithromycin trends varied from trends in other antibiotic classes with increases in the number of patients dispensed azithromycin prescriptions. Declines beyond seasonal expectations were greatest in children, especially young children (≤4 years), followed by older adults. The greatest beyond-seasonally expected decreases during the COVID-19 pandemic period examined here were observed in prescribing by primary care, advanced practice providers, and select subspecialty and surgical providers, although monthly trends varied between these groups.

We observed marked beyond seasonally expected declines in outpatient antibiotic prescribing during the COVID-19 pandemic. These decreases may be driven by declines in non-COVID-19 disease transmission and by changes in healthcare access and care-seeking. Studies have found declines in US outpatient visits during the pandemic. Several studies of emergency departments found visits for non-COVID-19–related conditions decreased >40% during the pandemic [1, 4, 10]. Data published by the Commonwealth Fund from a convenience sample of US outpatient visits showed a decline of almost 60% in March [11], mirroring the decrease in outpatients dispensed antibiotic prescriptions from March to April. Similarly, a study of a sample of patients with commercial and Medicare Advantage insurance found that outpatient visit rates decreased by around 50% in March [12]. However, these studies demonstrated rebounds in number of outpatient visits in April and May [11, 12], differing from trends observed in antibiotic prescribing. Pandemic-related declines in routine vaccination [13, 14] and chronic disease medication dispensing [2] have also been observed. In addition, a study using a nationally representative survey found that an estimated 41% of adults in the United States have delayed medical care during the pandemic because of concerns about COVID-19 [15]. Decreases in antibiotic prescriptions from dentists, surgical, and subspecialty prescribers and in agents commonly used in dentistry and surgical prophylaxis suggest that decreases in elective and subspecialty medical care may have contributed to declines in antibiotic prescriptions. Greater decreases observed in antibiotic prescriptions for children compared with adults may be indicative of declines in respiratory disease transmission caused by daycare and school closures and other COVID-19 mitigation efforts. Overall decreases across all antibiotic classes may also be due to fewer healthcare encounters for non-COVID-19–related conditions.

Further decreases in outpatient antibiotic prescribing are uncertain. Although many states began reopening efforts in May, we observed that the number of patients prescribed antibiotics continued to decrease in May 2020 relative to previous years. This ongoing decrease appeared to be driven by primary care prescriber specialties. Among many medical subspecialty and surgical specialties there was an increase in May compared with previous months in the number of patients with antibiotic prescriptions, likely correlating with resumption of elective and nonurgent subspecialty care and surgeries. Ongoing surveillance will be important as the pandemic continues and to evaluate long-term trends in outpatient antibiotic use.

Trends for azithromycin differed temporally and geographically from trends for other antibiotics, with implications for patient safety and antimicrobial resistance. Similar to trends reported for hydroxychloroquine [2, 3], nationally, azithromycin prescribing increased slightly from February to March 2020 and then decreased. Azithromycin, particularly in combination with hydroxychloroquine, was suggested as a potential therapy for COVID-19 among hospitalized patients based on extrapolations from in vitro studies of other respiratory infections [16, 17], medical reports [18], and widely reported press conferences [19]. Compared with historic levels, increases in patients dispensed azithromycin prescriptions in March and April were highest in New York and New Jersey, which experienced high COVID-19 case counts during these months.

Increases in outpatient azithromycin prescriptions are concerning for several reasons. First, limited clinical data are available on the effectiveness of azithromycin for the management of COVID-19. Second, azithromycin is associated with QTc prolongation, a potentially life-threatening cardiac adverse event, particularly in older adults [20, 21]. Because of the risk of adverse events, if azithromycin is combined with hydroxychloroquine, close monitoring by healthcare professionals is recommended [22]. Third, azithromycin use for COVID-19 may result in antibiotic resistance and diminished effectiveness for recommended indications (eg, sexually transmitted infections [23], community-acquired pneumonia [24]).

Outpatient antibiotic stewardship efforts will continue to be important for ensuring patient safety during the COVID-19 pandemic and beyond. Prior to the COVID-19 pandemic, at least one-third of outpatient antibiotic prescriptions were considered unnecessary [25, 26]. It is unclear how much of the decline observed in our study may be decreases in unnecessary antibiotic prescribing because of reduced healthcare visits and how much may be due to reduced disease transmission. It is also unknown how much of this decrease may represent needed care that was not received. Ongoing surveillance of complications of common bacterial infections (eg, mastoiditis as a complication of acute otitis media) and antibiotic-associated adverse events may be needed to evaluate the impact of reductions on both necessary and unnecessary antibiotic prescribing. Additionally, ongoing surveillance of vaccination rates and vaccine-preventable disease incidence may inform antibiotic stewardship efforts. Encouraging routine vaccination uptake, especially influenza vaccination, may be especially important during the COVD-19 pandemic to prevent increases in non-COVID respiratory infections [27].

Large healthcare delivery changes resulting from COVID-19 may create new challenges and opportunities for antibiotic stewardship in outpatient settings. The Centers for Medicare & Medicaid Services has substantially expanded coverage of telehealth opportunities during the public health emergency [28]. A market research company estimated that telehealth visits increased by 50% during the pandemic and are expected to top 1 billion by the end of 2020, with primary care visits projected at 200 million [29]. Continued coverage of telehealth services for low- and intermediate-acuity healthcare needs will be relevant to antibiotic stewardship because these types of encounters may include evaluation and management of common outpatient infections, such as sinusitis and acute otitis media. Telehealth has been used during the pandemic to triage patients to home monitoring or appropriate medical settings. Similar approaches may be beneficial for triaging common outpatient respiratory infections and identifying cases that may require additional diagnostic evaluation or in-office testing (eg, acute otitis media, pharyngitis) versus those that may be more appropriate for home care without antibiotics (eg, viral upper respiratory infections). Antibiotic use in telemedicine has not been widely studied, and it is not clear from previous studies how well guideline-concordant care is delivered in telemedicine compared with in-person visits [30–32]. Concerted stewardship efforts may be needed in this new and growing setting to ensure appropriate antibiotic use and patient safety.

Our study has limitations. First, because these data do not include visit information, we cannot determine the relative contributions of factors such as decreased infection incidence, decreased healthcare utilization, or shifts in healthcare settings (eg, telemedicine vs in-person) to reductions in antibiotic dispensing. Second, we could not evaluate the impact of prescribing changes on disease complications or adverse health outcomes. Third, the IQVIA TPT dataset does not contain diagnosis information; therefore, we could not evaluate indications or appropriateness of antibiotic prescriptions included in this study. Fourth, for prescriber specialty and state, the IQVIA TPT dataset does not contain aggregated data for all antibiotic classes. Therefore, we selected commonly used agents to examine major trends. Fifth, in this dataset, nurse practitioners, physician assistants, and osteopathic medicine physicians are categorized by provider type rather than specialty. Sixth, this analysis did not include antibiotics dispensed by mail order; however, we expect few antibiotics to be dispensed via mail-order pharmacy, because most are 1-time prescriptions for an acute illness. Finally, we were unable to separate whether changes observed in our study were due solely to the COVID-19 pandemic or influenced by other secular trends, such as general decreases in outpatient antibiotic prescribing [33]. However, a study of national outpatient antibiotic prescriptions showed only a 5% decrease in dispensed prescriptions from 2011 to 2016 [34], suggesting that changes observed in 2020 exceed previously observed secular trends.

CONCLUSION

During the early period of the COVID-19 pandemic, the number of outpatients with antibiotic prescriptions dispensed from retail pharmacies decreased substantially more than would be expected because of seasonal trends alone. Decreases were greatest among children and agents frequently prescribed for respiratory infections, dentistry, and surgical prophylaxis. Trends in azithromycin prescriptions varied from other antibiotic classes, likely related to transient interest in azithromycin in COVID-19 management. Ongoing surveillance of antibiotic prescribing, antibiotic-associated adverse events, antibiotic-resistant infections, and complications of common infections is needed to evaluate the long-term effects of these changes in outpatient antibiotic trends. Shifts in healthcare delivery, particularly the use of telehealth, are likely to be sustained, and should prompt assessment of effective antibiotic stewardship interventions [35, 36] to optimize antibiotic prescribing practices and ensure patient safety.

Supplementary Data

Supplementary materials are available at Clinical Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Financial support. This study was funded by the US Centers for Disease Control and Prevention. The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.

Potential conflicts of interest. L. M. K. is employed by Chenega Enterprise Systems and Solutions and is assigned to the US Centers for Disease Control and Prevention as part of a contract covering multiple tasks and positions. N. S. is employed by Lantana Consulting Group and is assigned to the US Centers for Disease Control and Prevention as part of a contract covering multiple tasks and positions. All other authors report no conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.