Journal of Hospital Infection (2009) 71, 108e111
Available online at www.sciencedirect.com
www.elsevierhealth.com/journals/jhin
Antibiotic use: is appropriateness expensive? V. von Gunten a, J.-P. Reymond a, K. Boubaker b, E. Gerstel c, P. Eckert c, ¨thi d, N. Troillet b,* J.-C. Lu a
Department of Pharmacy, Central Institute of the Valais Hospitals, Sion, Switzerland Center for Infectious Diseases and Epidemiology, Central Institute of the Valais Hospitals, Sion, Switzerland c Department of Intensive Care, Mid-Valais Hospital Center, Sion, Switzerland d Valais Health Observatory, Central Institute of the Valais Hospitals, Sion, Switzerland b
Received 3 January 2008; accepted 24 October 2008 Available online 4 December 2008
KEYWORDS Antibiotic use; Appropriateness; Cost
Summary Antibiotics are prone to misuse. In this study, 37% of 600 antibiotic prescriptions in three hospitals were considered unnecessary. When antibiotic therapy was indicated, 45% were considered to be inadequate. In multivariate analyses, the indicated treatments were found to be more expensive than the unjustified ones, probably because the latter were more often oral regimens. However, for indicated treatments, the cost of adequate and inadequate treatments did not differ significantly. ª 2008 The Hospital Infection Society. Published by Elsevier Ltd. All rights reserved.
Introduction Antibiotics represent w30% of acute care hospitals’ drug expenditure as they are prescribed to 20e50% of inpatients. Surveys have shown that antibiotics are particularly prone to misuse, since 22e65% of prescriptions are either not indicated or inadequate to treat the infection. This generates unnecessary costs and side-effects, including
* Corresponding author. Address: Central Institute of the Valais Hospitals, 86 Avenue Grand-Champsec, 1950 Sion, Switzerland. Tel.: þ41 27 603 4790; fax: þ41 27 603 4789. E-mail address:
[email protected]
increased selection pressure for resistant microorganisms and failure in treating infections.1 The present study aimed to measure the relationship between the appropriate use of antibiotics in hospitals and its cost.
Methods The charts of the first 600 patients receiving antibiotics after admission to internal medicine wards and intensive care units of three secondarycare public hospitals in Western Switzerland from November 2002 until April 2003 (200 patients in each hospital) were reviewed by four independent medical experts. These experts were affiliated to
0195-6701/$ - see front matter ª 2008 The Hospital Infection Society. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.jhin.2008.10.026
Cost of appropriate antibiotic use
109
other institutions and blinded regarding the admitting hospital and the prescribing physicians. They were two infectious diseases specialists and two intensive care specialists with 5e12 years’ experience in their fields. Each expert had to decide whether the treatments administered were indicated or not, and if so, whether they were adequate or not with respect to dosage, route of administration, microbial spectrum and duration of administration. No local prescribing guidelines were available at the time of the study and the experts based their evaluations on personal opinions. A given treatment was considered indicated or adequate if more than 50% of the expressed opinions considered it so, provided that at least two experts had answered. The calculation of costs included those of the antimicrobial agents as well as the devices and the nursing time necessary for their administration. Other variables collected included the characteristics of the patient (age, sex and Charlson comorbidity index2), and of the infection being treated (organ involved and microbiological results). Nosocomial infections were defined as those occurring more than 48 h after admission and not incubating at the time of admission. Difficult-to-treat infections were defined as those involving bones or joints, the central nervous system, vascular system (e.g. endocarditis), associated bacteraemia or occurring in a neutropenic patient. Acinetobacter spp., Enterobacter spp., Pseudomonas spp., Serratia marcescens and meticillin-resistant Staphylococcus aureus (MRSA) were considered as difficult-to-treat pathogens. Inter-rater agreement was assessed by using the kappa statistic. Comparisons were made between hospitals using chi-square, Fisher’s exact, or
Table I
Median cost (V) (interquartile range) Treatment indicated, N (%) Treatment adequate, N (%) Median patients’ age, years (range) Female gender, N (%) Charlson comorbidity index, N (%) 0 1e3 >3 Nosocomial infectionsb, N (%) Difficult-to-treat infectionsb, N (%) Difficult-to-treat pathogensb, N (%) a
Results Table I shows the characteristics of the studied population by hospital. Overall, antibiotic treatments were indicated in 376/598 patients (62.9%) and, when indicated, adequate in 207/376 patients (55.1%). According to Landis and Koch, the inter-rater agreement was moderate for the indication (kappa ¼ 0.546) and fair for the adequacy (kappa ¼ 0.264).3 The proportions of the treatments considered indicated by the individual experts were 63%, 69%, 70%, and 78%. They considered them adequate in 47%, 59%, 67% and 71% of the cases respectively. Hospital A had lower costs than the other two, and included fewer patients with a high comorbidity index. The results of the multivariate analyses are presented in Table II. After adjustment, an indicated antibiotic treatment generated a mean cost that was V84 higher than an unjustified one
Comparison of the three hospitals studied
Variable
b
KruskaleWallis tests, as appropriate. Two multivariate linear regression models were developed to analyse the effect of both the indication and the adequacy of antibiotic treatments, on their cost while adjusting for potential confounding factors such as the patients’ characteristics, the hospital to which they were admitted, the presence of difficult-to-treat infections, nosocomial infections or difficult-to-treat pathogens. The costs were logtransformed. Independent covariates were tested for colinearity by using variance inflation factors. All tests were two-tailed. P < 0.05 was considered significant. All analyses were performed with the SAS software (Release 8.1, SAS Institute, Cary, NC, USA).
No. of patients ¼ 200. See text for definitions.
Hospital Aa 51.9 122/199 58/122 72.1 106 83 92 25 10 20 2
(84.5) (61) (48) (14e94) (53) (41.5) (46) (12.5) (5) (10) (1)
Hospital Ba 128.2 128/199 75/128 76.6 111 56 106 38 9 22 8
(184.1) (64) (59) (15e97) (55.5) (28) (53) (19) (4.5) (11) (4)
Hospital Ca 80.8 126/200 74/126 69.4 102 60 101 39 12 20 7
(110.4) (63) (59) (19e97) (51) (30) (50.5) (19.5) (6) (10) (3.5)
P value 3 Nosocomial infectiona Difficult-to-treat infectiona Difficult-to-treat-pathogena a
SE
Model B
Mean cost P value Parameter variation (V) estimate
3.3132 0.6770 e
0.1758 0.0877 e
e 84.17 e
