TL;DR
This study uses a large UK primary care dataset to analyze antibiotic prescribing patterns in COPD patients. It reveals that while severe COPD patients receive more antibiotics per capita, the majority of antibiotics are prescribed to those with mild to moderate COPD, highlighting a key area for antibiotic stewardship interventions.
What Problem Does This Paper Address?
Antibiotic resistance is a growing global concern, and inappropriate prescribing is a major driver. In the UK, General Practitioners (GPs) are the primary prescribers, and while stewardship efforts have focused on common infections, patients with chronic diseases, particularly Chronic Obstructive Pulmonary Disease (COPD), receive a disproportionately high number of antibiotics. Specifically, COPD patients are prescribed antibiotics three times more often than the general population, mainly for acute exacerbations (AECOPD). However, there's uncertainty about which COPD patients truly benefit from antibiotics, as only about half of AECOPD are thought to be bacterial. This makes it challenging for GPs to balance the need for treatment with the risks of resistance and side effects. This paper aims to shed light on these prescribing patterns to inform targeted antibiotic stewardship.
Methods
The researchers conducted an observational cohort study using electronic health records (EHRs) from the Clinical Practice Research Datalink (CPRD) in England. They analyzed data from a large, representative sample of primary care patients over a 12-month period.
Key methodological aspects include:
- Data Source: The CPRD, a pseudonymized database of UK primary care records, linked to hospital admission and census data for a subset of practices.
- Patient Identification: Patients were identified with a COPD diagnosis using Read codes.
- COPD Severity Measurement: Multiple measures were used:
- Forced Expiratory Volume in 1 second (FEV1) as a percentage of predicted (categorized by GOLD stages).
- MRC dyspnoea scale (measuring breathlessness impact).
- Number of AECOPD in the year preceding the study, identified through a combination of prescribing (antibiotics, oral corticosteroids), symptoms, and diagnosis codes.
- Covariate Measurement: Factors like asthma, chronic heart disease, diabetes, obesity, smoking status, and influenza vaccination were also recorded.
- Antibiotic Prescribing Analysis: Prescriptions for systemic antibiotics (excluding anti-tuberculosis and anti-leprotic drugs) were analyzed. Indications were categorized based on national guidance (first-line, second-line, prophylactic for AECOPD) and proxies for non-respiratory infections. Prescription durations (acute first/second course, continuous <6 months, continuous ≥6 months) were also assessed.
- Statistical Analysis: Negative binomial regression models with random intercepts for general practices were used to estimate crude and adjusted rates of antibiotic prescribing, accounting for various patient characteristics. Mosaic plots were used to visualize relationships. Multiple imputation was employed for missing FEV1 and MRC dyspnoea scale data.
Datasets
- Clinical Practice Research Datalink (CPRD): A large, pseudonymized, retrospective database of primary care records from the UK. This dataset was linked to NHS hospital admissions and census data for English practices.
Key Insights and Ideas
- Disproportionate Prescribing: While patients with severe COPD have higher rates of antibiotic prescribing per person, the majority of antibiotics are prescribed to patients with mild to moderate COPD. This is due to the sheer number of patients in the mild-to-moderate category.
- Severity Matters, But So Does Volume: Antibiotic prescribing increases with COPD severity (FEV1, MRC scale) and the frequency of exacerbations. However, the sheer volume of mild-to-moderate COPD patients means they account for a larger overall share of prescriptions.
- Comorbidities and Risk Factors: Beyond COPD severity, several comorbidities (asthma, chronic heart disease, diabetes, heart failure) and influenza vaccination are associated with higher antibiotic prescribing rates. Current smokers, interestingly, received fewer antibiotics.
- AECOPD vs. No AECOPD: Even patients without a recorded AECOPD received significantly more antibiotics than the general population, suggesting a broad pattern of over-prescribing in COPD patients.
- Long-Term Antibiotics: While a small proportion of patients received continuous antibiotic treatment (≥6 months), this accounted for a notable percentage of prescriptions, particularly in those hospitalized for AECOPD.
Key Results
- Overall Prescribing: COPD patients accounted for 11.5% of all antibiotic prescriptions in primary care.
- Average Rate: The mean antibiotic prescription rate per COPD patient per year was 2.88.
- Mild-to-Moderate COPD: Patients with mild to moderate COPD and zero or one AECOPD received one to three prescriptions per year but accounted for 42.5% of all prescriptions.
- Severe COPD: Patients with severe COPD and frequent AECOPD (six to nine prescriptions per year) accounted for 13.0% of antibiotics.
- Comorbidity Impact: Asthma, chronic heart disease, diabetes, and heart failure were independently associated with increased antibiotic use.
- Influenza Vaccination: Patients vaccinated against influenza had higher rates of antibiotic prescribing (RR 1.23).
- Smoking: Current smokers received 9% fewer antibiotics on average (RR 0.91).
- Non-COPD Population: The rate of prescribing in a matched group without COPD was significantly lower at 0.60 prescriptions per year.
Limitations and Future Work
- Data Completeness: The indication for antibiotic prescription was not always recorded, and data on exposures like sputum color or C-reactive protein testing (which could aid decision-making) were not consistently available or analyzed.
- Influenza: The poor recording of influenza in primary care limited the ability to investigate its confounding effect on prescribing, especially for severe influenza.
- COPD Severity Measurement: While multiple measures were used, the paper notes the challenge of accurately capturing COPD severity and its progression over time within EHRs. The use of MRC grade was preferred over the number of AECOPD in the prior year due to its perceived completeness.
- "Rescue Packs": The analysis couldn't distinguish between prescriptions for current AECOPD and "rescue packs" for future events.
- Dispensing vs. Prescribing: CPRD records prescriptions, not actual dispensing, meaning some prescriptions might not have been taken.
- Single Year Analysis: The analysis was limited to a single year to ensure data consistency, though replication with 2013-2014 data showed similar findings.
- Hospital Prescribing: Prescriptions issued in hospitals were not captured, potentially underestimating total antibiotic use for patients with severe disease who are frequently hospitalized.
Future Work: The authors suggest that future research should quantify the risk of adverse events associated with prolonged and heavy antibiotic exposure in patients with severe COPD to ensure they have ongoing access to effective treatments. They also highlight the need for better information on prescribing patterns to support GPs in identifying patients who might benefit from delayed or withheld antibiotics, or the use of rapid diagnostic testing.
Summary updated