Current antibiotic treatment and outcome for lower respiratory tract infections

A number of national guidelines have been published to aid the antimicrobial management of community-acquired pneumonia. However, data on prescriptions for lower respiratory tract infection (LRTI) indicate considerable variation in the choice of first-line and subsequent therapy at national and local levels. Outcomes research in LRTI, whether based on clinical, economic or patient-focused criteria, is still evolving. Clinical outcomes are best studied for both pneumonia and exacerbation of chronic obstructive pulmonary disease. Economic evaluations often do not encompass all of the coses, for example, time off from work or the economic impact of antibacterial resistance. Duration of hospital stay is a good marker of costs for hospital providers and may be affected by age. marital status and comorbidities. Antibiotic choice may have an impact on the duration of hospital stay by increasing side effects, predisposing patients to hospital acquired infection or reduced clinical efficacy. Patient expectation is largely unstudied in pulmonary infection. monie rapeuLique. economique resistance antibacteriens; hospitalier couts prestateurs hospiLalier facteurs comorbidi Le peut exercer sur la duree du sejour hospitalier en raison I 'augmentation des effets secondaires qui predisposent les patients aux infections nosocomiales ou raison clinique. Les attentes des patients sont largement negligees dans les etudes sur !'infection pulmonaire.

L ower respiratory tract infection (LRTI) is one of the most common indications for antimicrobial chemotherapy, in both the community and hospital. As an aid to prescribing, many national societies of respiratory medicine and antimi-crobial chemotherapy have produced guidelines for the management of pulmonary infection . It is, however, unclear what effect these guidelines may have on day-to-day patient management beyond isolated reports, and it is unknown what ele-

Y•DDNOTCOPY
Antibiotic therapy and outcome ments guidelines should contain to ensure maximum prescriber impact (1). It is difficult to give clear recommendations about the use of individual antimicrobials when data that assess outcomes from a clinical, economic or human point of view are collected poorly. We summarize the basic elements of some of the guidelines on therapy and compare this with the information available on antibiotic use in pulmonary infection at a national and local level. In addition , various outcome measures are discussed and ways in which antimicrobial choice can affect outcome measures are illustrated.
National guidelines for the management of communityacquired pneumonia have been published in the United Kingdom (2), United States (3), Canada (4) and South Africa (5). Most of these guidelines are designed to provide advice on management of patients in the community and hospital, although guidelines for the United Kingdom include only hospital in-patients . Most guidelines stratify patients based on clinical severity, but may also include age (for example, United States , Canada , South Africa). The British Thoracic Society statement recommends the use of named individual antibiotics and discourages the use of others . The North American guidelines concentrate on recommending drug classes rather than individual agents. Evidence-based assessment of background information is largely lacking in the United Kingdom and North American guidelines, but was attempted in the South African guidelines. To date, no guidelines have included an indication as to whether the groups' recommendations are extensively supported by scientific or clinical data , or just opinion. The number of references quoted varies from nine (Canada) to 88 (South Africa). Fein and Neiderman (7) highlight many criticisms of the American Thoracic Society guidelines; most of these criticisms are equally valid for the other guidelines. Perhaps most important for community practice, guidelines do not address the issue of patient education. This is vital because patient expectation about their management has a significant impact on prescribing practices (8)(9)(10).
The failure of most guidelines to recommend specific drugs, compounded by the large number of antimicrobials available for use in LRTI ( Table 1) and exclusion of patient factors , causes prescribing practices to vary widely, even when guidelines are used.
A recent review by Ortqvist ( 12) illustrates this variation. Data collected by Intercontinental Medical Statistics Ltd, London, United Kingdom defined the top four drugs prescribed for treatment of out-patient pneumonia in seven European countries: France, Germany, Italy, the Netherlands, Spain, Sweden and the United Kingdom . In different countries, different groups of prescribers were included; only general practitioners were included in some countries, whereas general practitioners plus specialists in a variety of practice settings were included in others. Furthermore, different countries use different International Classification of Diseases-9 codes to define pulmona1y infection , and the number of antibiotics available varies; however, some broad patterns have emerged ( Table 2). Amoxicillin is among the top four agents used in France, the Netherlands, Spain , Sweden and the United Kingdom. Erythan J Infect Dis Vol 9 Suppl E November/ December 1998 Agents in brackets recently availc1 ble or soon to be ava ilable. Aclaplecl from (11) romycin and tetracycline are among the top four agents prescribed in four of the seven countries studied , and co-amoxiclav is among the top four in three countries. Ceftriaxone is among the top four agents prescribed in Italy and France. Cefuroxime-axetil, penicillin , roxithromycin, ceftazidime, imipenem, cefonicid, cefaclor, cefpodoxine and cefotaxime are among the top four most commonly used agents in any one country (any one of these comes within the first four in ranked order    Amoxicillin, co-a moxiclav, erylh ro mycin, ciprofl oxacin, tetracyclin e (32) Am oxicillin, co-a moxicl av, ofl oxacin, ciprofloxacin (5) Erythromycin, amox icillin, clarithromycin, ciprofloxacin (8) Hospital admissions, duration of stay and hospital cost related with age in patients with lower respiratory tract infection however, data to support the increased use of oral therapy now exist (13,14).
In the United Kingdom, amoxicillin is the antibiotic most commonly used to treat pneumonia, accounting for 67% to 80% of community use (15, unpublished data); however, there is less certainty on the first and second follow-up prescription for those who receive further therapy. Erythromycin is commonly used (43%) at first follow-up , but cephalosporins or cotrimoxazole are used in 20% of patients. On second follow-up, quinolones are also used commonly (15). The authors' experience in a Bristol community health centre, indicated that a wide range of agents are used to treat patients who return within 14 days ofLRTI diagnosis (Table 3). Therefore, despite the development and publication of guidelines for the management of pneumonia, considerable international variation in prescribing practice clearly exists, both in the community and in hospital. Even in countries such as the United Kingdom, where many patients in community practice receive a single drug (amoxicillin) when patients return and receive a further prescription, there is a high degree of variability in the agent chosen. These variations may be related in part to the poor definition of outcomes in LRTI that make comparison difficult, and the use of different patient stratifications and management strategies.
Outcomes can be divided into three areas -clinical, economic and patient-focused (16). The definition of clinical outcomes has long been a problem in assessing antimicrobial chemotherapy in acute infective exacerbation of chronic obstructive pulmonary disease (COPD) . In general, antimicrobial effectiveness can be assessed by using sputum microbiology, speed of recovery from symptoms, time until next exacerbation, subsequent antimicrobial therapy and quality of life. Not all patients with acute exacerbations require antibiotics , and , therefore, it is also important to stratify patients in clinical trials (17)(18)(19) . When attempting to predict repeat consultation with a chest problem , a history of cardiopulmonary disease and four or more previous exacerbations in the past two months were the best predictors ( 19). However, patient expec-  • igni(icant difference P< 0.05 tation of the consultation process and disease management issues plays a major role ( 15). In addition, choice of antimicrobial has an effect on the infection-free period in those with COPD. Ciprofloxacin, ampicillin and bacampicillin have similar infection-free periods, while doxycycline, minocycline, trimethoprim/sulphamethoxazole, cephalexin and cefaclor have shorter infection-free periods (20,21). In patients with community-acquired pneumonia, a variety of end-points have been used to assess outcomes including resolution of clinical signs, mortality, radiology, resolution of laboratory abnormalities, pathogen clearance on microbiology, use of second-line antimicrobial therapy, period of hospital stay and supurative complications. Cost management has become increasingly important in the study of LRTI and has been subdivided into antibiotic costs, office or out-patient visiting, hospital admission and time off work (22). However, it is often difficult to collect all information on the factors related to economic outcome, and important costs such as time off from work are often excluded (23). The impact that therapy has on antimicrobial resistance is almost always unevaluated in economic studies, probably because the costs are too difficult to calculate (24). In the hospital sector, the main provider costs are related to duration of stay. Although this is a crude measure, it probably identifies most of the costs. The more general costs to society and the patient are partially excluded, but factors that affect duration of stay (for example, age, marital status, comorbidity) influence costs and have a major economic impact. Patients with COPD and diabetes have increased hospital stays compared with controls, while asthmatics with LRTI may have shorter stays (25, unpublished data). Because of the higher number of hospital admissions and longer duration of stay in the elderly, age is a main cost determinant (Table 4) . An understanding of the factors related to hospital admission and identifying which factors have an impact on length of stay will help to manage hospital costs in LRTI.
The authors have shown that acquisition of Clostridium djfficile infection in Southmead Hospital is related to the use of cefuroxime for treatment of pulmonary infection (26); the Can J lnfe l Dis Vol 9 Suppl E November/Dece mber 1998 0 Y·DO OTCO Antibiotic therapy and outcome risk being related to both the duration and course of the dose. While the duration of stay of all patients with chest infection admitted to Southmead Hospital has fallen since 1994, those who have C djfficile infection still stay in hospital about twice as long as those without (Table 5) (27) . In April 1996, the authors substituted ciprofloxacin plus penicillin for cefuroxime in the protocol for the treatment of chest infection, a change that is not consistent with the British Thoracic Society guidelines. Ciprofloxacin was selected because quinolones seem less likely to be associated with C djfficile infection. Over the next year, cefuroxime use in patients with chest infection dropped by 90%; ciprofloxacin use increased 10-fold. This was associated with a fourfold reduction in the percentage of patients acquiring C djfficile in hospital and a reduction by onethird in the number of lower respiratory tract bed days occupied by those who contracted C djfficile (Table 6) (26 ,28). Data on how antibiotic choice affects length of stay related to other side effects or difference in clinical efficacy are unclear.
In addition to the better studied clinical and economic aspects , it is now becoming apparent that patient expectation and understanding of the natural course of LRTI are important factors in disease management, especially in those with chronic disease such as COPD. Patient-generated outcome scores based on symptoms, activities and overall well-being can be simple to apply and, while not presently used in pulmonary infection, have definite advantages (29) .

CONCLUSIONS
There are considerable differences between day-to-day practice and antimicrobial practice as recommended by several countries' national guidelines. Several countries' national guidelines often recommend drug classes rather than individ-ual agents, probably because outcome measures in clinical, economic or patient terms do not allow adequate differentiation between classes or individual drugs. If there is a clear relationship between the use of an agent and an adverse outcome, modification of prescribing can have significant favourable effects. However, given the complexity of factors in determining clinical and economic outcomes, such situations are rare. In community-based prescribing for LRTI, patient expectation is an important factor in antibiotic use; however, its role in hospital is unquantified.