aInfection and Immunodeficiency Unit, Tayside University Hospitals NHS Trust, Kings ... Faculty of Medicine, University of Dundee; cDepartment of Microbiology, ...
JAC
Journal of Antimicrobial Chemotherapy (2000) 46, 307–309
Community antibiotic therapy, hospitalization and subsequent respiratory tract isolation of Haemophilus influenzae resistant to amoxycillin: a nested case–control study R. A. Seatona*, D. T. Steinkeb, G. Phillipsc, T. MacDonaldb and P. G. Daveyb a
Infection and Immunodeficiency Unit, Tayside University Hospitals NHS Trust, Kings Cross Hospital; b Medical Medicines Monitoring Unit, Department of Clinical Pharmacology and Therapeutics, Faculty of Medicine, University of Dundee; cDepartment of Microbiology, Tayside University Hospitals NHS Trust, Ninewells Hospital, Scotland, UK The study objective was to determine whether recent community antibiotic prescribing and hospitalization are associated with β-lactam resistance in respiratory isolates of Haemophilus influenzae. Data obtained for hospitalization and community prescribing (in the previous 3 months) from 412 adults (>15 years) in whom an episode of respiratory tract infection had been described, during which H. influenzae was isolated, were analysed. Seventy-three (17.7%) isolates of H. influenzae were resistant to amoxycillin. Resistance was associated with recent hospitalization [odds ratio (OR) 3.2, 1.8–5.6] and antibiotic exposure in the community (2.1, 1.2–3.6). These variables were independently associated with amoxycillin resistance [hospitalization (OR 4.5, 1.7–12.5) and community β-lactam antibiotic exposure (3.9, 1.6–9.8)]. Hospitalized patients probably received antibiotics during their admission although aquisition of the organism or the β-lactamase via plasmids from other Gram-negative organisms in the hospital could also be a factor. Control measures to reduce the inappropriate use of antimicrobials in the community and in hospital need to be reinforced.
Introduction Haemophilus influenzae is an important human pathogen but also a commensal of the upper respiratory tract of both adults and children. Approximately one-third of asymptomatic schoolchildren will carry non-capsulate strains.1 Rates in adults are less well defined. Such strains are frequently implicated in exacerbations of chronic obstructive pulmonary disease (COPD) in adults, and pneumonia, bronchitis, otitis media and conjunctivitis in adults and children. As with other bacterial species there is concern over the increasing prevalence of antibiotic resistance in isolates of H. influenzae. The major mechanism for antibiotic resistance is β-lactamase production, which usually occurs as a result of plasmid-mediated TEM-1 β-lactamase production. Less frequently, resistance occurs as a result of a chromosomal mutation leading to reduced affinity of penicillin-binding proteins (PBPs) or a combination of altered permeability and reduced affinity of PBPs.2,3
In the UK, surveillance for resistance in H. influenzae in clinical isolates has been carried out since the 1970s: in surveys of 17 UK laboratories in 1977 and 1981 there was an increase in ampicillin resistance from 1.6% to 6.6%.4 More recently, in the 1995–1996 cold season β-lactamase production was found in 15.1% of UK isolates.5 In this study we investigated potential risk factors for the emergence of amoxycillin resistance in respiratory isolates of H. influenzae in adults, using computerized data linkage between the microbiology laboratory, hospital admission records and community antibiotic prescribing data.
Materials and methods The design was a case–control study nested within a retrospectively defined cohort. The setting was the catchment area for a large teaching hospital in Tayside, Scotland. The cohort was drawn from subjects resident in Tayside. Eligible subjects were adults (15 years) who had sub-
*Correspondence address. Infection and Tropical Medicine Service, Brownlee Centre, North Glasgow University Hospitals NHS Trust, Gartnavel General Hospital, 1053 Great Western Road, Glasgow G12 0YN, UK. Tel: 44-141-211-0292; Fax: 44-141-211-0294.
307 © 2000 The British Society for Antimicrobial Chemotherapy
R. A. Seaton et al. mitted one or more sputum samples for culture and from whose sample H. influenzae was isolated. Within the cohort subjects were divided into cases (H. influenzae resistant to amoxycillin) and controls (sensitive to amoxycillin). Cases and controls were compared for exposure to hospitalization or to antibiotics dispensed from community pharmacies in Tayside in the previous 3 months. Consecutive isolates of H. influenzae cultured from clinical specimens received between 1 March 1993 and 31 December 1995 were identified from the Tayside University Hospitals NHS Trust Department of Microbiology database. Both hospital and community specimens were included. Age and gender of patients and source (community or hospital) were recorded. For each adult only the first (incident) isolate received was included. Non-respiratory tract isolates were excluded. Sensitivities to amoxycillin were determined by the standard Stokes’ disc diffusion method using 2 and 3 g discs on chocolate blood agar (Columbia base, Oxoid, Basingstoke, UK) incubated in CO2. Hospitalization and community prescribing of antibiotics in the 3 months before the isolation of H. influenzae was determined. Hospitalization and prescription of antibiotics in the 3 days before H. influenzae isolation were ignored as this period corresponds to the current clinical episode. Community prescribing was determined from the Medicines Monitoring Unit.6 Antibiotics were then classified as β-lactams (e.g. penicillins or cephalosporins) or other. Hospitalization was determined through the Scottish Morbidity and Mortality Record database. Data were linked with individual microbiology records by a unique 10-digit patient identification number used both in hospital and community medical records. Community prescribing and hospitalization were investigated as risk factors for amoxycillin resistance. Data were analysed using the epi-info 5 statistical calculator. Odds ratios (with 95% confidence intervals) were calculated using the 2 test with the Yates correction. Adjusted odds ratios were calculated by controlling for hospitalization and community prescribing of antibiotics. P values were calculated and regarded as significant if 0.05.
Results Four hundred and twelve subjects (median age 64, range 17–96 years) met the inclusion criteria. Two hundred and forty-three subjects (59%) were male. One hundred and twenty-seven patients (30.8%) had received a β-lactam antibiotic in the previous 3 months and 67 (16.3%) a nonβ-lactam antibiotic. The majority of specimens (351, 85.2%) were submitted from hospital (half of these from general medical or respiratory clinics). One hundred and sixty-nine patients (41%) had been hospitalized in the 3 months before the isolation of H. influenzae. Seventy-three (17.7%) isolates were resistant to amoxycillin, including eight (2%) that were resistant to co-amoxiclav. In a univariate model both hospitalization and antibiotic exposure were associated with risk of amoxycillin resistance (Table I). In a bivariate model, hospitalization (irrespective of community prescribed antibiotics) was associated with amoxycillin resistance and the prescription of β-lactam antibiotics in the community (irrespective of hospitalization) was also associated with amoxycillin resistance (Table II).
Discussion Acquisition of β-lactamase production in H. influenzae (as in other bacteria) is thought to be influenced by antibiotic exposure through selective pressure. Sportel et al.7 identified hospital admission and increased number of recent antibiotic courses as risk factors for β-lactamase production in 50 patients with COPD and Johnson et al.8 identified a recent course of antibiotic as a risk factor in 34 patients with COPD. In the present study the majority of the cohort were in hospital care (in-patient or out-patient) and therefore not representative of the general adult population. By inference it is likely that the proportion of patients with COPD would be higher than in the general population. The population was not a sample as in other studies7,8 but represented the cohort of adult patients with H. influenzae
Table I. Hospitalization and community antibiotic exposure and relationship to resistance to amoxycillin in Haemophilus influenzae isolates (univariate model) Variables examineda Hospitalized (all patients) yes no Community antibiotics (all patients) yes no
Amoxycillin resistance (%)
OR (95% CI)
P value
47/169 (27.8) 26/243 (10.7)
3.2 (1.8–5.6)
0.001
45/194 (23.2) 28/218 (12.8)
2.1 (1.2–3.6)
0.009
a
Hospitalization and community antibiotic prescribing relate to the 3 month period before the isolation of H. influenzae. OR, odds ratio.
308
Amoxycillin-resistant Haemophilus influenzae Table II. Hospitalization and community antibiotic exposure and relation to resistance to amoxycillin in Haemophilus influenzae isolates (bivariate model) Variables examineda
Amoxycillin resistance (%)
Hospitalized (no community antibiotics) yes no Community antibiotics (no hospitalization) yes no Community β-lactam antibiotics (no hospitalization) yes no
OR (95% CI)
P value
19/91 (20.9) 7/127 (5.5)
4.5 (1.7–12.5)
0.001
19/116 (16.4) 7/127 (5.5)
3.4 (1.3–9.2)
0.01
15/71 (21.1) 11/172 (6.4)
3.9 (1.6–9.8)
0.002
a Hospitalization and community antibiotic prescribing relate to the 3 month period before the isolation of H. influenzae. OR, odds ratio.
isolation. Children were excluded as the majority of isolates were from infants. The interactions between mother, antibiotics, hospitalization and infant raised concerns over the introduction of unnecessary bias into the study. Recent hospitalization and the community prescription of antibiotics have been identified as risk factors for amoxycillin resistance in H. influenzae. Data on sampling rates within the community are not currently available although it is possible that general practitioners sample sputa after first-line antibiotics have failed. It is likely that hospitalized patients received antibiotics during their admission, although acquisition of β-lactamase via plasmids from other Gram-negative organisms in the hospital may also be a factor. Whether isolates were pathogenic in the patients studied is uncertain. In many cases it is possible that the isolation of H. influenzae simply reflects carriage. Carriage of penicillin-resistant pneumococci in children in Iceland has recently been demonstrated to correlate with individual consumption of antimicrobials.9 The clinical importance of penicillin-resistant pneumococci in pneumonia has recently been questioned.10 There are currently few data on the outcome of amoxycillin-resistant H. influenzae infections.8 This study adds further evidence of the importance of community antibiotic prescribing in the genesis of antimicrobial resistance. Control measures to reduce the inappropriate use of antimicrobials in the community (as well as in the hospital) need to be reinforced.
philus influenzae. Antimicrobial Agents and Chemotherapy 11, 383–7. 3. Parr, T. R. & Bryan, L. E. (1984). Mechanism of resistance of an ampicillin-resistant β-lactamase negative clinical isolate of Haemophilus influenzae type B to β-lactam antibiotics. Antimicrobial Agents and Chemotherapy 25, 747–53. 4. Philpott-Howard, J. & Williams, J. D. (1982). Increase in antibiotic resistance in Haemophilus influenzae in the United Kingdom since 1977: report of study group. British Medical Journal 284, 1597–9. 5. Felmingham, D., Robbins, M. J., Tesfaslasie, Y., Harding, I., Shrimpton, S. & Gruneberg, R. N. (1998). Antimicrobial susceptibility of community-acquired lower respiratory tract bacterial pathogens isolated in the UK during the 1995–1996 cold season. Journal of Antimicrobial Chemotherapy 41, 411–5. 6. MacDonald, T. M. & McDevitt, D. G. (1994). The Tayside Medicines Monitoring Unit (MEMO). In Pharmacoepidemiology, 2nd edn, (Strom, B. L., Ed.), pp. 245–55. Wiley, Chichester. 7. Sportel, J. H., Koeter, G. H., van Altena, R., Lowenberg, A. & Boersma, W. G. (1995). Relation between β-lactamase producing bacteria and patient characteristics in chronic obstructive pulmonary disease (COPD). Thorax 50, 249–53. 8. Johnson, S. R., Thompson, R. C. F., Humphreys, H. & Macfarlane, J. T. (1996). Clinical features of patients with β-lactamaseproducing Haemophilus influenzae isolated from sputum. Journal of Antimicrobial Chemotherapy 38, 881–4. 9. Arason, V. A., Kristinsson, K. G., Sigurdsson, J. A., Stefansdottir, G., Molstad, S. & Gudmundsson, S. (1996). Do antimicrobials increase the carriage rate of penicillin resistant pneumococci in children? Cross sectional prevalence study. British Medical Journal 313, 387–91.
1. Lerman, S. J., Kucera, J. C. & Brunken, J. M. (1979). Nasopharyngeal carriage of antibiotic resistant Haemophilus influenzae in healthy children. Pediatrics 64, 287–91.
10. Einarsson, S., Kristjansson, M., Kristinsson, K. G., Kjartansson, G. & Jonsson, S. (1998). Pneumonia caused by penicillin-nonsusceptible and penicillin-susceptible pneumococci in adults: a case–control study. Scandinavian Journal of Infectious Diseases 30, 253–6.
2. van Klingeren, B., van Embden, J. D. A. & Dessens-Kroon, M. (1977). Plasmid-mediated chloramphenicol resistance in Haemo-
Received 22 November 1999; returned 22 February 2000; revised 3 March 2000; accepted 3 April 2000
References
309
