Antibacterial prescribing in primary care

Abstract

Monitoring of general practice antibiotic prescribing is important to allow concordance with prescribing guidelines to be assessed. National Prescribing Analysis and Cost Data are limited by lack of information on the condition for which antibiotics are prescribed. Using the General Practice Research Database, we found that the 10 leading indications for antibacterial prescribing were (in descending order): upper respiratory tract infection (RTI), lower RTI, sore throat, urinary tract infection, otitis media, conjunctivitis, vague skin infections without a clear diagnosis, sinusitis, otitis externa and impetigo. Although for some conditions there appeared to be inappropriately high levels of antibacterial prescribing, the antibiotics chosen were usually those recommended for first-line treatment.

Introduction

Countries with higher community antibiotic prescribing have been shown to have higher antibiotic resistance rates in several pathogens.¹ On the basis of such data, there has been an emphasis on reducing unnecessary prescribing in general practice (GP) through professional education about prudent prescribing and campaigns to educate the public that most respiratory tract infections (RTIs) do not need an antibiotic prescription.² Community antibiotic consumption in the UK increased through the 1980s, but has declined markedly since the mid-1990s.³ The decline has been attributed to increases in GP thresholds for prescribing to patients with RTIs and to an increased use of delayed prescriptions.⁴ Others argue that it reflects declines in RTI consultation rates.⁵ Such declines may represent changes in patient consultation behaviour or a true decline in disease incidence.

The General Practice Research Database (GPRD) is the world's largest primary care database of consultations and prescriptions.⁶ Previous analyses of the GPRD confirm that prescribing rates for RTI have declined markedly.^7,8 However, analyses of the GPRD have not focused on which antibiotics are prescribed for which conditions. As part of its remit to advise on surveillance of antimicrobial resistance, the Specialist Advisory Committee on Antimicrobial Resistance (SACAR) Surveillance Subgroup explored the use of the GPRD to monitor antibiotic prescribing in primary care. This report gives an example of the utility of these data for understanding GP prescribing. The work was conducted as part of a Department of Health-funded project to explore the utility of the GPRD for surveillance of antibiotic prescribing (further details on http://www.idrn.org/antimicrobial_prescribing.php).

Surveillance data

The READ and Oxmis codes were searched to identify all conditions that might result in an antibacterial prescription and were classified into major groups. Code lists for all antibacterials (primarily those in chapter 5.1 of the British National Formulary and also including the antibacterials that are listed in other chapters) were also developed. These codes were used to identify antibacterials and possible indications for antibacterials in 60 GPs that contributed to the GPRD between 1998 and 2001 (the licence used for access to these data did not extend to data beyond 2001). If an antibacterial was prescribed on the same day as a possible indication, it was considered to have been prescribed for that indication.

Figure 1 shows the 20 leading indications for antibacterial prescriptions in 1998–2001 ordered in descending frequency. Table 1 shows the antibacterials prescribed for the top 10 leading indications that accounted for most antibacterial prescribing. The following conditions (in descending order) are the 10 most important causes of antibacterial prescribing: upper respiratory tract infection (URTI), lower respiratory tract infection (LRTI), sore throat, urinary tract infection (UTI), otitis media, conjunctivitis, vague skin infections without a clear diagnosis, sinusitis, otitis externa and impetigo. The proportion of cases receiving antibacterials varied widely according to the condition, with 31% of vague skin infections, 44% of URTIs, 63% of otitis media, 64% of sore throat, 75% of otitis externa, and over 80% of the remaining conditions (LRTI, UTI, sinusitis, impetigo and conjunctivitis) receiving an antibacterial prescription. For URTI, LRTI and otitis media, amoxicillin and erythromycin were the most commonly prescribed antibacterials. For sore throat, penicillin, amoxicillin and erythromycin were the most common prescriptions, whereas for sinusitis, amoxicillin, tetracycline and erythromycin were most commonly prescribed. Flucloxacillin and fusidic acid were the most common antibacterials prescribed for vague skin infections and impetigo. Aminoglycosides and amoxicillin were the drugs most commonly prescribed for otitis externa. Chloramphenicol and aminoglycosides were the antibacterials most commonly prescribed for conjunctivitis. Trimethoprim and cefalexin were the antibacterials most commonly prescribed for UTI.

Figure 2 shows the leading antibacterials prescribed in 1998–2001. Table 2 shows the conditions for which these antibacterials were most commonly prescribed. RTIs (URTI, LRTI, sore throat and otitis media) were the main indications for amoxicillin and erythromycin. Sore throat was the main indication for penicillin. Skin and soft tissue infections were the main indications for flucloxacillin. Otitis externa and conjunctivitis were the main indications for aminoglycosides. Conjunctivitis was the main indication for chloramphenicol. Sinusitis, acne and respiratory infections were the main indications for tetracycline. UTIs were the main indications for trimethoprim and cefalexin.

Discussion

The GPRD is a valuable tool for describing GP prescribing behaviour providing much more detailed information than can be obtained through Prescribing Analysis and Cost Data (PACT data), which provides no information on the condition for which antibiotics are prescribed. As identified in previous studies, GPs continue to prescribe antibacterials for a high proportion of infections. We note that PACT data indicate that antibacterial prescribing in primary care has plateaued since 2001, so we have no reason to suspect that there has been a marked change in the prescribing behaviour since then. Although it is likely to be appropriate that a high proportion of patients with some infections receive antibiotics (e.g. UTI, impetigo, otitis externa and conjunctivitis), it is contrary to guidance for other conditions such as sore throat, otitis media, URTI and sinusitis.⁹ There are a number of potential limitations to the estimates of what proportion of patients receives an antibiotic. First, it may be that not all consultations for minor infection are recorded in the data and that those which result in a prescription are more likely to be recorded. This would tend to lead to an overestimate of the proportion of patients who receive an antibiotic. Some GPs may use delayed prescribing such that although they issue a prescription, they ask the patient not to redeem the prescription unless they fail to improve. This issue is not possible to measure and again may lead to an overestimate of prescribing. Despite these potential biases, it is clear that there remains scope for further reductions in antibacterial prescribing. Previous analyses of the GPRD have not focused on what antibiotics are used when a decision is made to prescribe. This analysis demonstrates that most often the antibiotics chosen are in line with the recommended first-line antibiotics. There is relatively limited use of second-line agents (this use may often be justified by the clinical circumstances). The use of agents that are not recommended at all is very limited.

Analyses of the GPRD have been crucial to our understanding of antibiotic prescribing in primary care. Despite a number of inherent limitations to the data, they remain the only practical way of monitoring GP concordance with prescribing guidelines on a national level. The Surveillance Subgroup recommends that periodic analyses of the GPRD or equivalent data sets should be conducted to monitor condition-specific antibiotic trends.

Transparency declarations

None to declare.

References