Original article
Modelling and forecasting antimicrobial resistance and its dynamic relationship to antimicrobial use: a time series analysis

https://doi.org/10.1016/S0924-8579(99)00135-1Get rights and content

Abstract

To investigate the relationship between antimicrobial use and resistance in our hospital, we collected antimicrobial susceptibility and use data from existing microbiology laboratory and pharmacy databases for the period July 1st, 1991–December 31, 1998. The data was analyzed as time series and autoregressive integrated moving average (Box–Jenkins) and transfer function models were built. By using this method, we were able to demonstrate a temporal relationship between antimicrobial use and resistance, to quantify the effect of use on resistance and to estimate the delay between variations of use and subsequent variations in resistance. The results obtained for two antimicrobial–microorganism combinations: ceftazidime–gram-negative bacilli and imipenem–Pseudomonas aeruginosa, are shown as examples.

Introduction

Antimicrobial resistance is a problem in the majority of hospitals worldwide and treatment of infections due to resistant microorganisms is often difficult because many strains are resistant to most available antimicrobial agents [1]. In hospitals, antimicrobials used to treat infected patients are generally prescribed empirically before the results of microbiological samples are available. To make these empirical prescriptions, physicians refer to their infectious disease knowledge, clinical experience and knowledge of the local microbial ecology. In most hospitals, computerized data on the antimicrobial susceptibility patterns of clinically relevant hospital microorganisms are available from the microbiology laboratory information system. Current guidelines for the control and the prevention of antimicrobial resistance in hospitals recommend the implementation of a surveillance system and the rapid reporting of trends and significant changes to prescribing physicians [1], [2], [3]. Empirical antimicrobial prescriptions should be based on the information presented in these reports and some hospitals have included such summarized information on the antimicrobial order form to facilitate empiric prescriptions [4]. When implemented, success of such a system depends greatly on accessibility and regular update of antimicrobial resistance surveillance reports. However, although it is recommended to monitor trends and changes on a monthly or quarterly basis, or at a frequency appropriate to the volume of isolates when analyzed, data are generally analyzed from 1 year or semester to another, not taking into account variations at shorter time intervals.

Because local antimicrobial use data are difficult to obtain from hospital pharmacies, they are rarely reported and linked to antimicrobial resistance data. Analysis of the scientific literature shows indirect evidence of a relationship between antimicrobial use and resistance in hospitals [5]. Concomitant variations, i.e. changes in antimicrobial use leading to parallel changes in the level of resistance, probably represent the most convincing type of evidence since, unlike other types, they take into account the influence of time in the relationship. Several studies have reported such variations [6], [7], [8], [9], [10], [11], [12]. However, pooled data are generally used to analyze temporal associations observed between one period, generally 1 or several years, and another period [6], [7], [8], [9]. Applied to longitudinal data, this type of analysis cannot take into account small variations in antimicrobial resistance and use observed for shorter periods of time, e.g. months. Additionally, it does not take into account the fact that the level of resistance observed 1 month in a hospital is necessarily correlated to the level of resistance observed during the preceding month(s). Finally, these methods do not allow for quantification of the magnitude of the effect of antimicrobial use on the level of resistance.

Time series analysis corresponds to a group of techniques aimed at adjusting a mathematical model to a series of observations taken over time, or time series, for the purpose of predicting future behaviour of the series based in its historical behaviour, and trying to explain its characteristics as well as other factors influencing the series. Unlike usual statistical methods which assume that observed data are realizations of independent random variables, time series analysis takes into account the possible relationship existing between consecutive observations [13]. This method is appropriate when data are measured repeatedly at equal intervals of time. In 1976, Box and Jenkins provided a practical method for constructing autoregressive integrated moving average (ARIMA) models that analyze the temporal behaviour of a variable as a function of its previous values, its trends and its abrupt changes in the near past [14]. Since this date, time series analysis has been used in various fields such as econometry, engineering, meteorology, water resources research and more recently in medical specialities such as endocrinology, cardiology, environmental medicine and for the study of chronic diseases [13], [15]. Additionally, ARIMA modelling permits assessment of the relationships between one or more time series representing the influential factors by using an extension of the method called transfer function [13]. A practical method for constructing transfer function models was provided by Haugh [16]. For example, such models have been used to study the relationship between weather parameters and mortality [17], or influenza and mortality [18].

In the present study, we used time series analysis of monthly antimicrobial resistance and antimicrobial use data to predict for the level of antimicrobial resistance in our hospital and to investigate the relationship between antimicrobial resistance and use. Ceftazidime resistance in gram-negative bacilli and imipenem resistance in Pseudomonas aeruginosa isolates were chosen as examples.

Section snippets

Setting

Hospital Vega Baja is a tertiary care facility comprising various medical and surgical specialities, a paediatric unit, a maternity unit and a general intensive care unit. It was opened in 1990 as the referral hospital for a healthcare area of approximately 200 000 inhabitants, plus another 50 000 tourist residents. Since its opening, the size of the hospital has increased from 250 to 400 beds.

The VIRESIST Project

In 1995, the hospital infectious diseases committee decided to investigate the problem of empiric

Ceftazidime-resistant/intermediate gram-negative bacilli and ceftazidime use

Between July 1991 and December 1998, the clinical microbiology laboratory isolated 6244 non duplicate gram-negative bacilli from 5604 hospitalized patients. These gram-negative bacilli had the following distribution: Escherichia coli 46.5%, P. aeruginosa 9.9%, Proteus spp. 7.9%, Salmonella spp. 6.9%, Klebsiella spp. 6.7%, Enterobacter spp. 5.1%, and other gram-negative bacilli 17.0%. The average observed monthly percentage of ceftazidime-resistant/intermediate gram-negative bacilli was 3.3%

Discussion

Since its description in 1970, time series analysis has been used in various fields including medicine. However, it is to our knowledge the first time that it is applied to antimicrobial resistance and use data. By using this method, we were able to model several antimicrobial resistance and antimicrobial use series and to demonstrate the temporal relationship between hospital ceftazidime use and the percentage of ceftazidime-resistant/intermediate gram-negative bacilli, and between hospital

Acknowledgements

This project is supported by grant no. 97/1297 of Fondo de Investigación Sanitaria, Ministerio de Sanidad, Spain.

References (28)

  • L.L Snyder et al.

    Antibiotic sensitivity and the prescribing information sheet: assisting the prescribing physician

    Am. J. Infect. Control

    (1990)
  • B.F Crabtree et al.

    The individual over time: time series applications in health care research

    J. Clin. Epidemiol.

    (1990)
  • D.A Goldmann et al.

    Strategies to prevent and control the emergence and spread of antimicrobial–resistant microorganisms in hospitals. A challenge to hospital leadership

    J. Am. Med. Assoc.

    (1996)
  • Report of the ASM task force on antibiotic resistance. Antimicrob Agents Chemother...
  • D.M Shlaes et al.

    Society for Healthcare Epidemiology of America and Infectious Diseases Society of America Joint Committee on the Prevention of Antimicrobial Resistance: guidelines for the prevention of antimicrobial resistance in hospitals

    Infect. Control Hosp. Epidemiol.

    (1997)
  • J.E McGowan

    Is antimicrobial resistance in hospitals microorganisms related to antibiotic use?

    Bull. NY Acad. Med.

    (1987)
  • R.J Courcol et al.

    A seven year survey of antibiotic susceptibility and its relationship with usage

    J. Antimicrob. Chemother.

    (1989)
  • M.O Frank et al.

    Decrease in expenditures and selected nosocomial infections following implementation of an antimicrobial-prescribing improvement program

    Clin. Performance Quality Health Care

    (1997)
  • D.N Gerding et al.

    Aminoglycoside resistance and aminoglycoside usage: ten years of experience in one hospital

    Antimicrob. Agents Chemother.

    (1991)
  • R.P Mouton et al.

    Relationship between antibiotic consumption and frequency of antibiotic resistance of four pathogens — a seven-year survey

    J. Antimicrob. Chemother.

    (1976)
  • J.K Møller

    Antimicrobial usage and microbial resistance in a university hospital during a seven-year period

    J. Antimicrob. Chemother.

    (1989)
  • S Natsch et al.

    Use of amoxicillin-clavulanate and resistance in Escherichia coli over a 4-year period

    Infect. Control Hosp. Epidemiol.

    (1998)
  • P Søgaard

    The epidemiology of antibiotic resistance in three species of the Enterobacteriaceae and the relation to consumption of antimicrobial agents in Odense University Hospital

    Dan. Med. Bull.

    (1989)
  • U Helfenstein

    Box–Jenkins modelling in medical research

    Stat. Methods Med. Res.

    (1996)
  • Cited by (0)

    View full text