Abstract
Background
Antibiotic use remains common for the treatment of lower respiratory tract infections. The objective of this study was to evaluate the impact of antibiotic use on the duration and severity of acute lower respiratory tract infection (LRTI).
Participants
Adult patients presenting to US primary or urgent care sites with a chief complaint of cough and symptoms consistent with LRTI.
Main Measures
Collected data included demographics, comorbidities, symptoms, and 48 viral and bacterial respiratory pathogens by PCR. Severity of signs/symptoms was reported for up to 28 days using diaries and text messages. Interpolation was used where data were missing.
Key Results
Of 718 patients with baseline data, 29% had an antibiotic prescribed at baseline. The most common antibiotics were amoxicillin-clavulanate, azithromycin, doxycycline, and amoxicillin in 85% of patients. Provision of an antibiotic had no effect on the duration or overall severity of cough, including in patients with viral, bacterial, and mixed infections. Receipt of an antibiotic did reduce the likelihood of a follow-up visit (14.1% vs 8.2%, aOR 0.47, 95% CI 0.26–0.84), perhaps by removing the motivation of getting an antibiotic at a follow-up visit. However, they were also more likely to receive a systemic corticosteroid (31.9% vs 4.5%, p < 0.001) and were also more likely to receive an albuterol inhaler (22.7% vs 7.6%, p < 0.001). Patients believed that receiving an antibiotic would reduce the duration of their illness by nearly 4 days.
Conclusions
In this large prospective study in the US primary and urgent care setting, antibiotics had no measurable impact on the severity or duration of cough due to acute LRTI. Patients had unrealistic expectations regarding the duration of LRTI and the effect of antibiotics which should be the target of antibiotic stewardship efforts.
Supplementary Information
The online version contains supplementary material available at 10.1007/s11606-024-08758-y.
KEY WORDS: antibiotics, lower respiratory infections, cough
INTRODUCTION
Cough is the most common illness-related reason for an ambulatory care visit, accounting for 2.7 million outpatient visits and more than 4 million emergency department visits annually.1 Cough is a symptom; acute bronchitis is inflammation of bronchial epithelium and is a commonly used diagnostic term, while lower respiratory tract infection (LRTI) is a major cause of both inflammation and cough and can be either viral or bacterial. Conservative estimates suggest that about half of all antibiotic prescriptions for acute respiratory conditions are unnecessary.2 Before the pandemic, it was shown that antibiotics were prescribed about 70% of the time when a diagnosis of uncomplicated cough and LRTI (often called “acute bronchitis”) is made.3,4 A viral pandemic did not change this practice; a meta-analysis of 130 studies found that 78% of COVID-19 patients were prescribed an antibiotic.5
Antibiotics are essential in treating severe bacterial infections, but like all interventions, they have potential harm to the individual. Frequent but usually mild acute harms include dizziness, nausea, diarrhea, and rash.6–14 Rare but serious adverse effects include anaphylaxis, Stevens-Johnson syndrome, and Clostridioides difficile–associated diarrhea, occurring in up to 4% of individuals.15 Antibiotics also have societal harms, most significantly antimicrobial resistance (AMR), which the World Health Organization (WHO) recognizes as a major public health concern.16 Infections caused by antibiotic-resistant organisms can be very serious and even life-threatening. According to the Centers for Disease Control and Prevention (CDC), antimicrobial resistance adds a 20 billion dollar surplus in direct healthcare costs in the United States.17 This has led to a push to curb inappropriate usage of antibiotics for LRTIs.
Thus, it is known that antibiotics have personal and societal harms and that most LRTIs are caused by viral infections. Despite decades of medical education and other efforts to decrease inappropriate use,18 antibiotics are still prescribed for the majority of visits for LRTIs. Is it possible that clinicians are seeing patients with higher levels of acuity, that we have underestimated how often bacterial infections are causing LRTIs, or that patients are truly improving faster when prescribed antibiotics? The Enhancing Antibiotic Stewardship in Primary Care (EAST-PC) study is the largest study to date of LRTI in adults in primary and urgent care settings in the United States. It set out to describe what treatments patients are receiving, specifically looking at the association between antibiotic prescribing and the duration and severity of acute cough. We also examined whether the type of pathogens detected were associated with antibiotic usage and resolution of symptoms.
METHODS
EAST-PC was a prospective observational study of adults in the outpatient setting presenting with acute cough. It was approved by the Western Institutional Review Board (study number 1253415) and the IRBs of each participating institution (University of Georgia, Georgetown, University of Wisconsin). The study was sponsored by the federal Agency for Healthcare Research and Quality (grant number R01HS025584).
Participants
Participants between 18 and 75 years were recruited from primary care and urgent care clinics in the Washington, D.C., Madison, Wisconsin, or Athens, Georgia metro areas. All reported a cough for no more than 14 days accompanied by at least one of the following: shortness of breath, sputum or phlegm production, body or muscle aches, chest discomfort with cough, chest congestion, measured fever or felt feverish, chills or sweats. Patients who had taken an antibiotic, antiviral, or corticosteroid in the previous 28 days, those with HIV disease or other serious immunodeficiency, those currently on cancer chemotherapy, or those taking systemic steroids or other immunosuppressive drugs were excluded. Patients with chronic lung disease other than mild to moderate asthma were also excluded based on feedback from the human subjects committee. We began data collection in June 2019 and ceased data collection in April 2023.
Baseline Data Collection
At baseline, detailed information about patient demographics, the presence and severity of respiratory symptoms, current medications and comorbidities, and any tests or referrals were recorded. In addition, any medications, including antibiotics and antivirals, prescribed at the index visit were recorded. Participants had two swabs taken, one mid-turbinate or anterior nares and one pharyngeal. These were combined and stored in a viral transport medium at − 40 to − 80C prior to shipping. The samples were then processed by the Pneumonia Response and Surveillance Laboratory of the Respiratory Diseases Branch at the Centers for Disease Control and Prevention.
Microbiologic results were classified as only bacteria detected, only viruses detected, both viruses and bacteria detected, and no organism detected. We further created a subclassification of bacteria likely to be pathogenic based on previous research (acknowledging that they can also be commensal), including Moraxella catarrhalis, Haemophilus influenza, Bordetella pertussis, Strep pneumoniae, Pseudomonas aeruginosis, Mycopasma pneumonia, Group A Feel strep, and Chlamydia pneumoniae.
Determining the Duration and Severity of Cough
Patients were asked to report the duration of cough prior to the index visit. If that was not known, they were asked to estimate the date of onset of cough. The severity of cough was recorded at baseline, in daily paper diaries, and on days 3, 7, 14, 21, and 28 via text message link to an online survey. Where data were reported by both diary and online surveys, the diary value was used if the two differed. We interpolated missing diary data by using the illness course for patients with a similar severity of cough at the beginning and end of the missing period but who had complete data.
Severity of their acute LRTI at baseline was estimated using the previously validated Bronchitis Severity Score (BSS).19 It assigns 0 to 3 points for increasing severity of cough, sputum, chest pain with coughing, dyspnea, and impact on usual activities (range 0 to 15 points). The overall severity of cough used interpolated diary data, assigning 1 point for mild, 2 for moderate, 3 for severe, and 4 for very severe cough, and summing the severity across the entire cough illness (“area under the symptom curve”).
Medication Use and Follow-up Visits
Any new medications prescribed at the index visit were recorded by a research assistant. Any return visit to the same or another primary or urgent care center, visit to the emergency department, or hospitalization was recorded, as well as whether antibiotics were prescribed at the follow-up visit.
Analysis
For bivariate analysis, differences between groups for continuous variables like duration of symptoms were evaluated using Student’s t-test, and for dichotomous variables using the Pearson chi-square test. For multivariate analysis, logistic regression was used for the dichotomous outcomes of follow-up visit and antibiotic prescription at baseline. Linear regression was used for the continuous outcome variables cough duration and overall severity. All analyses used Stata version 18 (StataCorp, College Station, TX).
RESULTS
A total of 718 adults with acute LRTI were enrolled in the study; approximately half (n = 355) were enrolled in 2022. Valid specimens for PCR testing were obtained for 618 participants, 443 provided usable diary data regarding symptom severity, and 401 had both usable diary data and a valid PCR specimen.
Patient Characteristics
Characteristics of participants are shown in Table 1. The mean age of participants was 39 years, with a range of 18 to 74 years, with 65.3% female. Most participants were recruited in the Washington DC area. Overall, 78.1% were White race, 11.0% Asian/Pacific Islander, 8.2% Black, and 1.0% American Indian or Alaska Native. Most had a household income over $78,000 and 62.1% had completed college or graduate education. Among the 443 asked about COVID vaccination status once vaccines were available, 25.3% had received the primary series only and 67% had received the primary series plus a booster. Characteristics were generally similar between the full population of 718 participants and the 618 with a valid PCR specimen.
Table 1.
Characteristics of Participants, Stratified by Antibiotic Prescription at Index Visit
| Patient characteristic | All symptomatic patients, number (%) | Antibiotic recipient, number (%) | No antibiotic, number (%) | p |
|---|---|---|---|---|
| Total participants | 718 | 207 (28.8%) | 511 (71.2%) | |
| Mean patient age | 38.9 | 40.9 | 38.1 | 0.02 |
| Sex | ||||
| Female | 469 (65.3%) | 139 (67.1%) | 330 (64.8%) | 0.397 |
| Male | 243 (33.8%) | 68 (32.9%) | 175 (34.4%) | |
| Other | 4 (0.6%) | 0 (0%) | 4 (0.8%) | |
| Race + | ||||
| White | 561 (78.1%) | 168 (81.2%) | 393 (76.9%) | 0.185 |
| Asian/Pacific Islander | 79 (11.0%) | 18 (8.7%) | 61 (11.9%) | |
| Black | 59 (8.2%) | 15 (7.3%) | 44 (8.6%) | |
| American Indian/Alaska Native | 7 (1.0%) | 4 (1.9%) | 3 (0.59%) | |
| Did not respond or missing | 26 (3.6%) | 5 (2.4%) | 21 (4.1%) | |
| Site of data collection | ||||
| Washington DC | 432 (60.2%) | 89 (43.0%) | 343 (67.1%) | < 0.001 |
| Madison, Wisconsin | 196 (27.3%) | 56 (27.1%) | 140 (27.4%) | |
| Athens, Georgia | 90 (12.5%) | 62 (30.0%) | 28 (5.5%) | |
| Moderate or worse severity at baseline | ||||
| Cough | 539 (75.1%) | 171 (82.6%) | 368 (72.0%) | 0.003 |
| Sputum | 302 (42.1%) | 95 (45.9%) | 207 (40.5%) | 0.186 |
| Chest pain or ache with cough | 254 (35.4%) | 76 (36.7%) | 178 (34.8%) | 0.633 |
| Usual activities affected | 463 (64.5%) | 139 (67.1%) | 324 (63.4%) | 0.342 |
| Fever, chills, or sweats | 204 (28.4%) | 111 (53.6%) | 273 (53.4%) | 0.961 |
| Colored sputum (green or bloody) | 228 (31.8%) | 87 (42.0%) | 141 (27.6%) | < 0.001 |
| Shortness of breath | 154 (21.4%) | 52 (25.1%) | 102 (20.0%) | 0.127 |
| Fatigue | 485 (67.6%) | 154 (74.4%) | 331 (64.8%) | 0.013 |
| Myalgias | 265 (36.9%) | 79 (38.2%) | 186 (36.4%) | 0.657 |
| Coryza | 388 (54.0%) | 129 (62.3%) | 259 (50.7%) | 0.005 |
| Wheezing | 192 (26.7%) | 66 (31.9%) | 126 (24.7%) | 0.048 |
| Headache | 290 (40.4%) | 95 (45.9%) | 195 (38.2%) | 0.056 |
| Presence of specific symptoms | ||||
| Cough causing shortness of breath or lightheadedness | 358 (49.9%) | 122 (58.9%) | 236 (46.6%) | 0.003 |
| Felt warm or feverish on every day of illness | 334 (46.5%) | 90 (43.5%) | 244 (48.1%) | 0.259 |
| Cough causing nausea or vomiting | 312 (43.5%) | 103 (49.8%) | 209 (41.02%) | 0.037 |
| Double-sickening* | 286 (39.8%) | 94 (45.4%) | 192 (37.9%) | 0.065 |
| Duration of cough prior to presentation in days (mean, 95% CI) | 5.1 (4.9–5.4) | 6.2 (5.7–6.6) | 4.7 (4.4–5.0) | < 0.001 |
*Generally defined as an initial viral illness that resolves and is followed suddenly by new onset of symptoms, often believed to be bacterial
+ Self-identified, could choose more than one group
At their index visit, 75% had a moderately bad or worse cough, 72% felt moderately unwell or worse, and 67% had at least moderate fatigue. Moderate or worse feverishness was reported by 28.4%, and moderate or worse chills or sweats by 23.4%. Detailed symptom severity data are shown in Table 1. Patients receiving an antibiotic were significantly more likely to have moderate to severe cough, colored sputum, coryza, wheeze, and fatigue than those who did not receive one. Cough causing shortness of breath or lightheadedness and cough causing nausea or vomiting were also seen more often in patients receiving an antibiotic.
The mean duration of cough prior to presentation was 5.1 days and was longer in patients who received an antibiotic at the index visit (6.2 vs 4.7 days, p < 0.001). Patients expected that their duration of cough would be significantly shorter if given an antibiotic (9.7 vs 13.4 days, p = < 0.001).
Antibiotic Usage
In our cohort of 718 symptomatic outpatients with LRTI, 29% received at least one antibiotic, while 7.1% received an antiviral agent. Specific agents prescribed are summarized in Table 2. Amoxicillin-clavulanate, azithromycin, doxycycline, and amoxicillin were most commonly prescribed and accounted for 85% of antibiotic prescriptions. Patients receiving an antibiotic were significantly more likely to receive a systemic corticosteroid (31.9% vs 4.5%, p < 0.001) and were also more likely to receive an albuterol inhaler (22.7% vs 7.6%, p < 0.001). Patients were significantly less likely to receive an antibiotic after the onset of the COVID-19 pandemic than before (22.2% vs 40.9%, p < 0.001). Antibiotic prescribing varied by site, occurring in 68.9% of visits in Athens, Georgia, 28.6% at the University of Wisconsin, and 20.6% in Washington, D.C. (p < 0.001).
Table 2.
Medications Prescribed to 718 Symptomatic Outpatients with LRTI
| Medications prescribed | Number |
|---|---|
| Antibiotics | |
| Amoxicillin-clavulanate | 63 |
| Azithromycin | 62 |
| Doxycycline | 35 |
| Amoxicillin | 15 |
| 1st generation cephalosporin (cephalexin, cephradine) | 2 |
| 2nd generation cephalosporin (cefzil, cefprozil, cefuroxime) | 10 |
| 3rd generation cephalosporin (cefdinir, cefpodoxime) | 3 |
| Augmentin + azithromycin | 2 |
| Levofloxacin | 1 |
| Azithromycin + oseltamivir | 1 |
| Amoxicillin + oseltamivir | 1 |
| Azithromycin + ceftriaxone injection | 1 |
| Unknown* | 11 |
| Any antibiotic | 207 (28.8%) |
| Antivirals | |
| Oseltamivir | 30 |
| Nirmatrelvir-ritonavir (Paxlovid) | 21 |
| Any antiviral | 51 (7.1%) |
* “Unknown” means that the clinician checked a box indicating that an immediate antibiotic was prescribed, but did not record its name
Patients who received antibiotics presented to the clinic further along in their illness, with 6.3 versus 4.7 days of symptoms (p < 0.001). The mean Bronchitis Severity Score (range 0 to 15) at baseline was 7.1. Patients receiving an antibiotic had a higher Bronchitis Severity Score than those who did not (7.6 versus 6.8, p = 0.005). Additionally, those who got an antibiotic had a higher overall disease severity during the entire course of their illness as measured by the area under the symptom curve than those who did not (26.0 vs 22.6, p = 0.02).
Of 84 patients with a clinic follow-up visit during the study period, 30 patients received an antibiotic: 23 during week 1, 6 during week 2, and 1 during week 3. Five of the patients who received an antibiotic at the follow-up visit had also received one during their index visit.
Duration of Cough
The duration of cough overall and by type of pathogen detected is summarized in Table 3. Receiving an antibiotic was associated with a small but significant increase in the duration of cough overall (17.5 vs 15.9 days, p = 0.05). The number of days until cough resolution for the subset of patients with bacterial infection who received an antibiotic prescription was the same as for those who did not receive an antibiotic (17.3 versus 17.4 days). As expected, cough resolution if one had a virus was similar whether or not they were prescribed an antibiotic (14.5 days with antibiotic versus 15.4 days without). Interestingly, for those that had a mixed infection (both viral and bacterial), receiving an antibiotic was associated with a significantly longer time to resolution (21.2 vs 15.8 days, p = 0.009). Finally, patients presenting with a moderate or severe cough at baseline had a significantly longer overall duration of cough than those with milder symptoms (17.5 vs 12.7 days, p < 0.001).
Table 3.
Duration of Infection by Whether Antibiotic Was Prescribed
| Population | Number given antibiotic/total (%) | Duration of cough (days) | P value (t-test) | |
|---|---|---|---|---|
| Antibiotic | No antibiotic | |||
| All participants | 138/34 | 17.5 | 15.9 | 0.05 |
| 1 or more viruses detected and no bacteria | 29/135 | 15.4 | 14.5 | 0.59 |
| 1 or more bacteria detected and no viruses | 40/111 | 17.4 | 17.3 | 0.96 |
| Mixed: both viruses and bacteria detected | 19/82 | 21.2 | 15.8 | 0.009 |
| No virus or bacteria detected | 13/55 | 19.0 | 17.8 | 0.66 |
| 1 or more pathogenic bacteria detected* | 51/164 | 18.8 | 16.8 | 0.12 |
| Only pathogenic bacteria detected* | 36/99 | 17.8 | 17.6 | 0.88 |
| 1 or more pathogenic bacteria in mixed infection | 15/65 | 21.1 | 15.9 | 0.02 |
*Pathogenic bacteria included Moraxella catarrhalis, Haemophilus influenza, Bordetella pertussis, Strep pneumoniae, Pseudomonas aeruginosis, Mycopasma pneumonia, Group A Feel strep, and Chlamydia pneumoniae
Follow-up Visits
A total of 84 participants (11.7%) had a follow-up visit to the primary or urgent care setting, 7 (1.0%) had a follow-up to the emergency department, 2 (0.3%) were hospitalized, and 89 (12.4%) had any follow-up visit or visits. One hospitalized patient had influenza A detected, while the other had rhinovirus. Among patients with an unscheduled visit to the ED for LRTI, 3 had H. influenzae (all types) detected, 1 each had human metapneumovirus, influenza A, and SARS-CoV-2, and one had no pathogens detected. Patients who did not receive an antibiotic were slightly more likely than those who did to have a follow-up visit of any kind (14.1% vs 8.2%, p = 0.03) in the univariate analysis. Characteristics of patients with and without a follow-up visit are summarized in Appendix Table 1.
Multivariate Analysis
Results of a multivariate analysis with duration of cough and overall cough severity as the dependent variable are shown in Table 4. The strongest independent predictors of greater overall severity of cough were having moderate or severe cough (beta coefficient 7.66) and shortness of breath (beta 4.38) on presentation, and a longer duration of cough prior to presentation (beta 1.23). The strongest predictors of a longer duration of cough were moderate or severe cough (beta 3.3), impaired usual activities (beta 1.87), and an increasing number of days of cough prior to the index visit (beta 0.98). In neither analysis was provision of an antibiotic independently associated with the duration or severity of cough. In separate models (Appendix Tables 2 and 3) that replaced individual symptoms by the overall BSS at baseline, the BSS was an independent predictor of both the overall duration of cough (beta 0.50, 95% CI 0.24–0.76) and overall severity of cough (beta 1.69, 95% CI 1.25–2.14).
Table 4.
Linear Regression for Total Cough Severity and Total Cough Duration as Dependent Variables. Bold Face Indicates a Statistically Significant Association with the Dependent Variable at p < 0.05
| Beta coefficient (95% CI) | ||
|---|---|---|
| Independent variables | Dependent variable = total cough severity | Dependent variable = total cough duration |
| Antibiotic prescribed | − 0.084 (− 2.74 to 2.91) | − 0.59 (− 2.27 to 1.90) |
| Patient age (years) | 0.077 (− 0.004 to 0.158) | 0.04 (-0.0067 to 0.090) |
| Cough duration prior to index visit in days | 1.23 (0.86 to 1.60) | 0.98 (0.76 to 1.20) |
| Type of infection by PCR | ||
| Viral | − 2.23 (− 5.96 to 1.50) | − 1.84 (− 4.06 to 0.66) |
| Bacterial | 0.41 (− 3.34 to 4.142) | − 0.89 (− 3.14 to 1.37) |
| Mixed | − 0.54 (− 4.61 to 3.52) | −0.16 (− 2.58 to 2.26) |
| Symptom at least moderate baseline severity | ||
| Sputum | 0.18 (− 2.43 to 2.80) | − 0.86 (− 2.40 to 0.66) |
| Cough | 7.66 (4.56 to 10.76) | 3.30 (1.48 to 5.11) |
| Chest pain | 2.84 (-0.026 to 5.71) | 0.55 (− 1.14 to 2.25) |
| Shortness of breath | 4.38 (1.31 to 7.44) | 0.67 (− 1.14 to 2.49) |
| Impaired usual activity | 2.62 (− 11 to 5.35) | 1.87 (0.25 to 3.48) |
| Sputum colored | − 0.53 (− 3.34 to 2.28) | 0.026 (− 1.63 to 1.69) |
| Fever, chills, or sweats present | 1.77 (− 0.74 to 4.29) | 0.55 (− 0.96 to 2.05) |
Table 5 summarizes the results of logistic regression for the outcomes of follow-up visits and antibiotics prescribed at baseline. For follow-up visit occurred as the dependent variable, provision of an antibiotic at the index visit reduced the likelihood of follow-up (aOR 0.46, 95% CI 0.24–0.88). Patients were more likely to receive an antibiotic if they were older (aOR 1.02, 95% CI 1.00–1.03 per year), and had a longer duration of cough prior to the index visit (aOR 1.13, 95% CI 1.07–1.20 per day). They were less likely to be given an antibiotic if a viral pathogen was detected by PCR (although note that physicians did not have access to PCR results). When replacing individual symptoms with the BSS (Appendix Tables 4 and 5), it was a significant predictor of antibiotic prescription (aOR 1.09, 95% CI 1.02–1.17) but not a follow-up visit occurring (aOR 1.020, 95% CI 0.94–1.11).
Table 5.
Logistic Regression for Follow-up Visit Occurred and Antibiotic Prescribed at Baseline Visit as Dependent Variables. Bold Face Indicates a Statistically Significant Association with the Dependent Variable at p < 0.05
| Adjusted odds ratio (95% CI) | ||
|---|---|---|
| Independent variables | Dependent variable = follow-up visit occurred | Dependent variable = antibiotic prescribed |
| Received antibiotic at baseline visit | 0.46 (0.24 − 0.88) | NA |
| Patient age (years) | 1.02 (0.99 − 1.03) | 1.02 (1.01 − 1.03) |
| Cough duration prior to index visit in days | 1.04 (0.96 − 1.12) | 1.13 (1.08 − 1.20) |
| Type of infection by PCR | ||
| Viral | 0.56 (0.25 − 1.24) | 1.14 (0.61 − 2.10) |
| Bacterial | 1.33 (0.64 − 2.78) | 1.78 (0.97 − 3.25) |
| Mixed | 1.24 (0.57 − 2.73) | 1.283 (0.66 − 2.47) |
| Symptom at least moderate baseline severity | ||
| Sputum | 1.30 (0.77 − 2.20) | 0.81 (0.53 − 1.23) |
| Cough | 1.50 (0.77 − 2.95) | 1.33 (0.79 − 2.20) |
| Chest pain | 0.98 (0.55 − 1.75) | 0.99 (0.63 − 1.56) |
| Shortness of breath | 0.97 (0.52 − 1.81) | 1.26 (0.78 − 2.03) |
| Impaired usual activity | 1.17 (0.66 − 2.08) | 1.18 (0.75 − 1.85) |
| Sputum colored | 0.69 (0.38 − 1.23) | 1.46 (0.95 − 2.25) |
| Fever, chills, or sweats present | 0.90 (0.54 − 1.52) | 1.50 (0.99 − 2.26) |
DISCUSSION
In this large prospective observational study of adults with LRTI in the US outpatient primary and urgent care setting, the use of antibiotics was not independently associated with the duration or severity of cough. Actually, those receiving an antibiotic had a small but significant increase in length of illness overall (17.5 vs 15.9 days, p = 0.05) although this is largely due to patients with a longer duration of illness prior to the index visit being more likely to receive an antibiotic. Importantly, the pathogen data demonstrated that the length of time until illness resolution for those with bacterial infection was the same as for those not receiving an antibiotic versus those receiving one (17.3 vs 17.4 days).
However, it appears that clinicians prescribed antibiotics when they felt patients were sicker, as patients who received antibiotics presented to the clinic further along in their illness, 6.3 versus 4.7 days, had higher Bronchitis Severity Scores (mean 7.6 versus 6.8 points), and had a higher overall disease severity as measured by the area under the symptom curve. This is important information for public health authorities as it demonstrates where clinician education may be necessary and important to decrease unnecessary antibiotic usage.
Independent predictors of worse overall severity were having at least a moderate to severe cough, chest pain, or shortness of breath on presentation, as well as fever, chills, or sweats. Predictors of a longer duration of cough included moderate to severe cough and impairment of usual activities at baseline, as well as increasing age. Patients presenting with these factors could be counseled to expect a longer, more severe duration of cough; those presenting with moderate or severe cough had a mean 5 days longer cough than those with mild cough.
Our findings are consistent with findings from a large European study of 2714 adults with LRTI, which also found that the rate of recovery was similar whether or not antibiotics were prescribed.19 In the current study, the strongest predictors of greater severity and longer duration were greater baseline severity of cough, a longer duration of cough prior to the index visit, and increasing patient age. Patients with a longer duration of illness prior to the index visit were also more likely to receive an antibiotic.
We did find a statistically significant association between receiving an antibiotic and not requiring a follow-up visit (aOR 0.47). However, the absolute reduction in having a follow-up visit was small (5.9%) and of the 9 patients who had an ED visit or hospitalization, only 3 had a bacteria detected (H. influenzae, all types) while 5 had a viral pathogen. Neither of the hospitalized patients had a bacterial pathogen detected. It is likely that having received an antibiotic at the index visit removed an important patient motivation for a follow-up visit (e.g., to obtain an antibiotic).
Research by Little and others has shown that prescribing antibiotics for infections where patients are unlikely to benefit “medicalizes” that illness and makes it more likely that they will seek care and an antibiotic with subsequent infections.20 However, meeting these expectations for an antibiotic does not necessarily result in a more favorable patient assessment of the care that they received.21 It is also important that physicians clarify for patients the expected duration of an acute lower RTI (an average of 2 or more weeks) in order to avoid not only initial but requests for second antibiotic prescriptions.22 In fact, previous research has shown that prescriptions for a second antibiotic in patients with RTI peak at 7 days after the initial visit.23
Antibiotic prescribing for acute LRTI varied by site and was associated with longer duration of illness and higher symptom severity in the univariate analysis. Patients receiving antibiotics also had a much higher likelihood of receiving a systemic corticosteroid, resulting in further medicalization of what is almost always an uncomplicated, self-limited illness. Also worrisome is that we found that the type of pathogen detected did not affect the duration of illness or the likelihood of receiving an antibiotic.
The study also revealed that patients had unrealistic expectations about the duration of cough and the effectiveness of antibiotics, which may be informative for future efforts to educate both patients and physicians about what to expect when they have a LRTI. On average, patients expected that receiving an antibiotic would reduce the duration of their LRTI from a total of 13.4 days to 9.7 days. In fact, there was actually no reduction in duration and the average duration of all coughs was over 2 weeks regardless of the type of pathogen and whether or not an antibiotic was prescribed.
As in all observational cohorts, there are some inherent limitations, such as lack of control group and clinicians and patients knowing they were being surveyed. The overall rate of antibiotic prescribing was lower in our study than in many others. This may be due to a Hawthorne effect, as physicians knew that their prescribing decisions were being recorded, as well as the fact that patients who had serious respiratory comorbidities or immunosuppression were excluded, leaving a somewhat healthier than average population. The potential for confounding by indication or other factors that influenced antibiotic prescribing is a possibility. Also, we only recruited from three centers and 78% of the patients were White and the household income was higher than the average US income.
This study has several strengths, including the large and diverse sample of patients with acute LRTI, the use of PCR testing to identify the causative pathogens, the collection of detailed symptom data using a validated diary, and a relatively diverse population at 3 geographically separated sites. It is the largest study of its type to date in the US outpatient setting.
CONCLUSIONS
The results of this relatively large study have important implications for clinical practice and public health. Suggesting that antibiotic prescribing for acute LRTI continues to be unnecessary and may be harmful, as there was no independent association of receipt of an antibiotic with shorter duration or lower severity of illness. We found much lower antibiotic usage than most previous studies, suggesting that clinicians were being more careful about whom to prescribe antibiotics. However, even with stricter usage, we found little benefit of antibiotics. Furthermore, two different multivariate analyses failed to demonstrate that prescribing an antibiotic was associated with the duration or severity of cough. Further research can help better understand if there is a limited role for antibiotics in patients with cough, i.e., maybe based on age and baseline severity, but at the present time clinicians need to use antibiotics less frequently as the evidence for their benefit does not exist.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements:
We would like to thank all the clinics who participated in enrollment and the patients who were kind enough to share their data.
Funding
This research was funded by the AHRQ Grant RO1 (grant number: R01HS025584).
Data Availability:
Once our primary series of analyses are completed, data will be made available upon reasonable request for relevant collaborative research.
Declarations:
Conflict of Interest:
The authors declare that they do not have a conflict of interest.
Footnotes
This research has not been presented previously.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.National Hospital Ambulatory Medical Care Survey: 2011 outpatient department summary tables. http://www.cdc.gov/nchs/data/ahcd/nhamcs_outpatient/2011_opd_web_tables.pdf. Accessed 28 Nov 2023
- 2.Fleming-Dutra KE, Hersh AL, Shapiro DJ, et al. Prevalence of Inappropriate Antibiotic Prescriptions Among US Ambulatory Care Visits, 2010-2011. JAMA. 2016;315(17):1864-1873. 10.1001/jama.2016.4151. 10.1001/jama.2016.4151 [DOI] [PubMed] [Google Scholar]
- 3.Linder JA, Stafford RS. Antibiotic treatment of adults with sore throat by community primary care physicians: a national survey, 1989-1999. JAMA. 2001;286(10):1181-1186. 10.1001/jama.286.10.1181 [DOI] [PubMed] [Google Scholar]
- 4.Ebell MH, Radke T. Antibiotic use for viral acute respiratory tract infections remains common. Am J Manag Care. 2015;21(10):e567-575. [PubMed] [Google Scholar]
- 5.Malik SS, Mundra S. Increasing Consumption of Antibiotics during the COVID-19 Pandemic: Implications for Patient Health and Emerging Anti-Microbial Resistance. Antibiot Basel Switz. 2022;12(1):45. 10.3390/antibiotics12010045. 10.3390/antibiotics12010045 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Turck D, Bernet JP, Marx J, et al. Incidence and risk factors of oral antibiotic-associated diarrhea in an outpatient pediatric population. J Pediatr Gastroenterol Nutr. 2003;37(1):22-26. 10.1097/00005176-200307000-00004. 10.1097/00005176-200307000-00004 [DOI] [PubMed] [Google Scholar]
- 7.Mitchell DK, Van R, Mason EH, Norris DM, Pickering LK. Prospective study of toxigenic Clostridium difficile in children given amoxicillin/clavulanate for otitis media. Pediatr Infect Dis J. 1996;15(6):514-519. 10.1097/00006454-199606000-00008. 10.1097/00006454-199606000-00008 [DOI] [PubMed] [Google Scholar]
- 8.McCarty JM, Phillips A, Wiisanen R. Comparative safety and efficacy of clarithromycin and amoxicillin/clavulanate in the treatment of acute otitis media in children. Pediatr Infect Dis J. 1993;12(12 Suppl 3):S122-127. 10.1097/00006454-199312003-00006. 10.1097/00006454-199312003-00006 [DOI] [PubMed] [Google Scholar]
- 9.Elstner CL, Lindsay AN, Book LS, Matsen JM. Lack of relationship of Clostridium difficile to antibiotic-associated diarrhea in children. Pediatr Infect Dis. 1983;2(5):364-366. 10.1097/00006454-198309000-00006. 10.1097/00006454-198309000-00006 [DOI] [PubMed] [Google Scholar]
- 10.McFarland LV. Epidemiology, risk factors and treatments for antibiotic-associated diarrhea. Dig Dis Basel Switz. 1998;16(5):292-307. 10.1159/000016879. 10.1159/000016879 [DOI] [PubMed] [Google Scholar]
- 11.Beaugerie L, Flahault A, Barbut F, et al. Antibiotic-associated diarrhoea and Clostridium difficile in the community. Aliment Pharmacol Ther. 2003;17(7):905-912. 10.1046/j.1365-2036.2003.01531.x. 10.1046/j.1365-2036.2003.01531.x [DOI] [PubMed] [Google Scholar]
- 12.Young VB, Schmidt TM. Antibiotic-associated diarrhea accompanied by large-scale alterations in the composition of the fecal microbiota. J Clin Microbiol. 2004;42(3):1203-1206. 10.1128/JCM.42.3.1203-1206.2004. 10.1128/JCM.42.3.1203-1206.2004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Finkelstein JA, Davis RL, Dowell SF, et al. Reducing antibiotic use in children: a randomized trial in 12 practices. Pediatrics. 2001;108(1):1-7. 10.1542/peds.108.1.1. 10.1542/peds.108.1.1 [DOI] [PubMed] [Google Scholar]
- 14.Finkelstein JA, Metlay JP, Davis RL, Rifas-Shiman SL, Dowell SF, Platt R. Antimicrobial use in defined populations of infants and young children. Arch Pediatr Adolesc Med. 2000;154(4):395-400. 10.1001/archpedi.154.4.395. 10.1001/archpedi.154.4.395 [DOI] [PubMed] [Google Scholar]
- 15.Goldenberg JZ, Yap C, Lytvyn L, et al. Probiotics for the prevention of Clostridium difficile-associated diarrhea in adults and children. Cochrane Database Syst Rev. 2017;12(12):CD006095. 10.1002/14651858.CD006095.pub4. [DOI] [PMC free article] [PubMed]
- 16.Antimicrobial resistance. https://www.who.int/news-room/fact-sheets/detail/antimicrobial-resistance. Accessed 4 Dec 2023.
- 17.Antibiotic resistance threats in the United States, 2013. https://www.cdc.gov/drugresistance/pdf/ar-threats-2013-508.pdf. Accessed 4 Dec 2023.
- 18.Core Elements of Antibiotic Stewardship | Antibiotic Use | CDC. Published September 7, 2023. https://www.cdc.gov/antibiotic-use/core-elements/index.html. Accessed 4 Dec 2023.
- 19.Butler CC, Hood K, Verheij T, et al. Variation in antibiotic prescribing and its impact on recovery in patients with acute cough in primary care: prospective study in 13 countries. BMJ. 2009;338:b2242. 10.1136/bmj.b2242. 10.1136/bmj.b2242 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Little P, Rumsby K, Kelly J, et al. Information Leaflet and Antibiotic Prescribing Strategies for Acute Lower Respiratory Tract Infection: A Randomized Controlled Trial. JAMA. 2005; 293: 3029-35. 10.1001/jama.293.24.3029 [DOI] [PubMed] [Google Scholar]
- 21.Hamm RM, Hicks RJ, Bemben DA. Antibiotics and respiratory infections: are patients more satisfied when expectations are met? J Fam Pract. 1996; 43: 56-62. [PubMed] [Google Scholar]
- 22.Ebell MH, Lundgren J, Youngpairoj S. How Long Does a Cough Last? Comparing Patients’ Expectations With Data From a Systematic Review of the Literature. Ann Fam Med. 2013; 11: 5-13. 10.1370/afm.1430 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ebell MH, Radke T. Antibiotic use for viral acute respiratory tract infections remains common. J Managed Care. 2015; 21(10):e567-75. [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Once our primary series of analyses are completed, data will be made available upon reasonable request for relevant collaborative research.