Design, setting and subjects: Retrospective analysis of data for January 2006 to December 2010 from an infection surveillance system covering residents of four co-located RACFs, with a total of 150 residential care beds, in Melbourne, Victoria.
Main outcome measures: Number of episodes and incidence of health care-associated infection (HCAI); rate of antimicrobial use; prescribing concordance with McGeer criteria for infection; frequency of clinical specimen collection.
Results: There were 1114 episodes of an infectious syndrome over 267 684 occupied bed-days (OBD), affording an average HCAI rate of 4.16 episodes/1000 OBD annually over 5 years (95% CI, 3.92–4.41). The mean rate of antimicrobial use was 7.07 courses/1000 OBD (range, 6.71–7.84). Around 40% of antimicrobial prescribing was for episodes that did not fulfil the McGeer criteria for clinical infection; this included about half of suspected urinary tract and upper respiratory tract infections (URTI), and about one-third of suspected lower respiratory tract and skin infections. Antimicrobials were routinely prescribed for URTI and bronchitis. Of all episodes treated with antimicrobials, 36% had documentation that a clinical specimen was obtained.
Conclusions: The HCAI rate remained relatively stable over time. Routine surveillance and feedback of infection rates to the facilities did not result in a noticeable decrease of infection burden over time. It is of immediate concern that antimicrobials were being prescribed for a large proportion of suspected infections that did not meet criteria for clinical infection. Opportunities exist to further improve the use of antimicrobials in the RACF setting.
Infection surveillance is a crucial component of an integrated infection control program, providing timely feedback about infection prevention issues that is geared towards decreasing the burden of infection within health care facilities.1 Data from overseas suggest there is a considerable infection burden among aged care residents.2-6 The Aged Care Standards and Accreditation Agency Ltd has highlighted that surveillance is required in Australian residential aged care facilities (RACFs), but there are few published studies describing the epidemiology of health care-associated infection (HCAI) and antimicrobial use.7,8
A recent Australian study highlighted a concern about emergence of antimicrobial-resistant organisms among RACF residents.8 The study, however, did not describe the infection burden and the associated antimicrobial prescribing patterns among this population. Although some overseas studies report extensive prescribing of broad-spectrum antibiotics,9-11 differences in clinical practice are likely in Australia.
Accordingly, we explored the patterns of common infections and use of antimicrobial agents in four Australian RACFs using internationally recognised criteria.12 Recognition of the infection burden among residents in this long-term care setting will help to identify infection prevention priorities and guide antimicrobial stewardship efforts.
This study involved four RACFs (150 residential care beds in total) administered by Caulfield Hospital’s Residential Care Services in Melbourne, Australia. These nursing homes provide 24-hour high-level nursing care to elderly residents with multiple disabilities including dementia and psychological disorders. Most (80%) residential care beds are for permanent residents, but one facility accommodates transitional care residents (30 care beds; maximum length of stay is generally 3 months). Although the four facilities are co-located on the Caulfield Hospital campus of Alfred Health, medical care is provided by general practitioners. The choices of pharmacy and pathology providers are determined by each GP. Nursing services for these RACFs include a dedicated infection prevention nurse consultant. The study was approved by the human research ethics committees of Alfred Health and Monash University.
A systematic infection surveillance system was introduced across the RACFs; initially (from 2003), to monitor the incidence of urinary tract infection (UTI) and conjunctivitis, then later (from mid 2005), other HCAIs were included. HCAI refers to infection occurring at least 48 hours after admission to an RACF with new or acutely worsening symptoms that fulfil criteria for site-specific infection. These definitions, known as the McGeer criteria,12 refer to a set of clinical and other criteria intended specifically for use in long-term care facilities (See appendix). They have been widely accepted as an epidemiological tool for interfacility comparison of infection rates, both locally and globally.3,5,7,13,14
To maintain consistency, infection surveillance activity is coordinated by an experienced infection prevention nurse consultant (IPNC). As routine practice, senior nursing staff from the respective RACFs notified the IPNC, using a standard data collection form, when a resident had possible signs or symptoms of infection. Information such as the resident’s demographic data, signs and symptoms observed and other relevant infection management data (for instance, types of clinical specimens, antimicrobial prescribed and microbiological results) was recorded. The IPNC then reviewed the resident’s medical records to assess whether they met the McGeer criteria. Episodes that were validated based on the McGeer criteria were included in the infection surveillance database, and a monthly feedback on infection rates was provided to the individual RACFs. Data on antimicrobial use were retained for suspected infections that did not fulfil the McGeer criteria, and these were appropriately archived from 2009 onwards.
We analysed the infection surveillance database from 2006 to 2010 to determine the overall prevalence and trends over time of HCAIs. Data collection was based on the validated episodes of HCAI being recorded in the surveillance database over the 5 years. Data on antimicrobial use were only available from 2009 onwards. Accordingly, all episodes of suspected infection (those fulfilling and not fulfilling McGeer criteria) between 2009 and 2010 were reviewed to determine patterns of antimicrobial use.
We used descriptive statistics to summarise the data — number and percentage for categorical variables, and mean for continuous variables. Rate of infection was calculated as the number of episodes of infections per 1000 occupied bed-days (OBD), whereas antimicrobial prescribing was described in courses per 1000 OBD. We calculated the 95% confidence interval (CI) for the infection rate using OpenEpi version 2.3.1 (Open Source Epidemiologic Statistics for Public Health, version 2.3.1; http://www. openepi.com), assuming a Poisson distribution. We used the monthly recorded rates of infection that conformed to the McGeer criteria over the 5-year period to construct a statistical control chart, using IBM SPSS Statistics 19 (IBM Corporation, New York, NY, USA). A proportion (p) control chart was selected to account for the proportion of infections in the studied RACF population over time.15 Infection rates above the upper control limit may indicate a potential outbreak.
Over 5 years, from January 2006 to December 2010, 1114 infections that fulfilled the McGeer criteria were recorded over 267 684 OBD (98% bed occupancy) in the four RACFs studied. The estimated mean HCAI rate was 4.16 episodes per 1000 OBD annually (95% CI, 3.92–4.41). The infection rates across the four RACFs were similar, with yearly incidence ranging from 3 to 5.5 episodes per 1000 OBD. The distribution of validated HCAIs is shown in Box 1.
The four most common infectious syndromes, representing 89% of HCAIs, were UTI, lower respiratory tract infection (LRTI), skin and soft tissue infection (SSTI) and eye infection (1.53, 0.97, 0.61, and 0.57 per 1000 OBD, respectively). Pneumonia (0.07/1000 OBD) and gastroenteritis (0.15/1000 OBD) constituted less than 6%, and only one bloodstream infection was documented.
The monthly trend of infections within the population was assessed using a p control chart (Box 2); the monthly proportion significantly exceeded the upper control limit in September 2010, owing to a scabies outbreak. While the statistical control chart suggested some fall in infection rates at the end of the study (eg, with eight points below the mean, January 2010 to August 2010), the overall trend in the rate of infections showed a small non-significant decrease over the 5-year period.
In 2009–2010, 662 episodes of suspected infection were notified, and patients were prescribed at least one antimicrobial agent in 659 episodes. Of these, 247 episodes (37%) did not fulfil the McGeer criteria for clinical infection. The McGeer criteria were not met in 141/288 reported episodes of UTI (49%), 48/147 episodes of LRTI (33%), 24/47 episodes of URTI and 27/83 episodes of skin infection. For the 141 episodes of UTI that did not fulfil the McGeer criteria, the most frequently reported clinical features were altered mental or functional status and change in urine character. In 20% (28) of these episodes, patients were prescribed antimicrobials for asymptomatic bacteriuria. With respect to the 72 respiratory tract episodes that did not meet the McGeer criteria and for which patients received antimicrobials, the most common symptom was cough, documented for 56% (27/48) of bronchitis episodes and 59% (13/22) of influenza-like illnesses. Fever was documented in only nine of the 34 episodes of influenza-like illness. Fever was a clinical feature in only 13% of all suspected infections (84/662; 17% of validated episodes and 6% of unvalidated episodes).
Between 2009 and 2010, the mean rate of antimicrobial use was 7.07 courses per 1000 OBD (range, 6.71–7.84 courses per 1000 OBD for the four RACFs). A total of 755 courses of antimicrobials were prescribed, of which 86.2% were administered orally, 9.4% were administered topically and 4.4% were given parenterally. Almost all were antibacterials; 1.5% were antiviral or antifungal agents.
Cephalexin, trimethoprim, amoxycillin–clavulanic acid and roxithromycin accounted for 62% of all the courses prescribed (Box 3). UTI was the most common indicator for antimicrobial use (in 41% of courses), followed by LRTI (in 26%), SSTI (in 12%), URTI (in 8%) and eye infection (in 8%). Box 4 shows antimicrobial prescribing patterns for three major sites of infection (ie, UTI, LRTI and URTI combined, and SSTI). Only about 5% of antimicrobials were prescribed for other indications such as ear and mouth infection, gastroenteritis, bloodstream infection and unexplained febrile illness. Around 11% of all antimicrobials prescribed for UTIs were broad-spectrum antibiotics, which included norfloxacin, ciprofloxacin, ceftriaxone and gentamicin. Indwelling catheter-related (IDC) UTI contributed to about 20% of all episodes of UTI; prescribing patterns for both non-IDC and IDC UTIs were similar. Antimicrobials were routinely prescribed for URTI and bronchitis (constituting up to 31% of overall antimicrobial use). Similar prescribing patterns were observed for episodes that fulfilled McGeer criteria and those that did not.
Of 659 episodes for which patients were treated with antimicrobials, only 238 episodes (36%) had documentation that clinical specimens were obtained. Clinical specimens were infrequently collected for most types of infection (Box 5). Conversely, collection of urine specimens was documented for 72.2% of UTI episodes.
The incidence of HCAIs from 2006 to 2010 in four RACFs in Melbourne was comparable to rates reported in the literature, however, despite feedback of infection rates to staff, we did not find evidence of notable improvement in infection burden over time. We found a significant burden of antimicrobial use and identified URTI, bronchitis and asymptomatic bacteriuria as areas where use was likely to be inappropriate.
Our results indicate that systematic infection surveillance can be established and sustained in Australian RACFs using internationally established criteria,12 although there are currently no systematic surveillance systems for infection in RACFs in any Australian jurisdiction. The only two published Australian studies that have examined HCAI rates in this setting were conducted over a 9-year period in a Sydney network and as a 6-month pilot study in a rural Victorian region.7,16 Both studies used the McGeer criteria and reported rates of infection at 3.1 and 2.2 infections per 1000 OBD, respectively. These rates are of the same magnitude as infection rates reported in studies from the United States and European countries of 1.5–9.5 infections per 1000 OBD, based on the McGeer criteria.2-6,14 The average rate and types of HCAI in our study were generally comparable to the reported literature, except for the rate of pneumonia, which was considerably lower than that in some studies.11,17,18 We found that chest x-rays, required to fulfill criteria for pneumonia, were often not performed in this patient population, suggesting that the low rate may reflect differences in clinical practice, rather than a true difference in pneumonia rate. The rate of gastroenteritis (0.15 episodes per 1000 OBD) was lower than that reported in one other local study (0.29–1.42 episodes per 1000 OBD).19
Importantly, the rates of HCAI remained relatively stable over time, with the exception of two outbreaks reported in 5 years, suggesting that routine surveillance and feedback of infection-related issues did not result in noticeable improvement in preventing infections. Routine surveillance without more proactive measures may not effectively improve the burden of HCAIs and antimicrobial use over time. Most infection prevention guidelines are designed for use in hospitalised patients with a high prevalence of invasive devices. Work is required in developing and validating infection prevention policies for the RACF setting.
There was a significant burden of oral antimicrobial use in this population, with a high proportion of treated episodes not meeting infection criteria, reflecting the difficulties in clinically assessing this patient popula-tion. The antimicrobial burden in the RACFs we studied (7.07 courses per 1000 OBD) was mid-range compared with those reported in overseas studies (2–15 courses per 1000 OBD).10,17,20,21 Data from Australian RACFs for benchmarking are lacking. The rate of antimicrobial use cannot easily be compared with hospital-based and community-based studies, owing to differences in surveillance methods. Nonetheless, if the average duration of an antimicrobial course was around 7 days, the use rate could be converted and would approximate 50 defined daily doses (DDD) per 1000 OBD, which is higher than the estimated use in the community (about 25 DDD/1000 population days)22 but much lower than the estimated use in large Australian hospitals (about 1000 DDD/1000 OBD).23
In our study, the considerable proportion of antimicrobials prescribed for episodes that did not meet the McGeer criteria was similar to that reported elsewhere.8,20 Although the McGeer criteria are well established for epidemiological comparison, antimicrobial use in patients not meeting these criteria may not always be inappropriate. Nevertheless, it was noted that all episodes of URTI and nearly all of bronchitis led to routine prescribing of antimicrobials without further diagnostic investigation to confirm bacterial aetiology, contrary to recommendations in the national Antibiotic Therapeutic Guidelines.24 Similarly, the use of antimicrobials for asymptomatic bacteriuria is not recommended. The prescribing of antimicrobials for asymptomatic bacteriuria is of particular concern because the emergence of multiresistant organisms in the RACF setting25,26 is often attributable to extensive or inappropriate use of antimicrobials and may result in the RACFs becoming a reservoir for multiresistant organisms. Barriers to improving the quality of antimicrobial prescribing include the complexities of assessing elderly patients due to atypical illness presentation, lack of timely investigations to guide therapy, and the large number of health care providers involved in the care of RACF residents.
Several sources suggest that fluoroquinolones were the most commonly prescribed antibiotics for common infections in the RACF setting overseas.9,10,27,28 However, our data show that cephalexin and trimethoprim were used to a much higher extent, and the use of fluoroquinolones or other broad-spectrum antibiotics was low. The low use of fluoroquinolones is likely to reflect, at least in part, restricted indications under the Australian Pharmaceutical Benefits Scheme. The high use of chloramphenicol, on the other hand, was due to the widespread prescribing of topical chloramphenicol eye preparations for conjunctivitis. Pleasingly, in most cases, the empirical prescribing pattern for UTIs was in line with the national guidelines.24
Routine ordering of microbiological tests is not always practical in the RACF setting. This reflects difficulties in obtaining specimens, the involvement of multiple laboratories servicing these institutions and the lack of timely results that inform prescribing. In addition to difficulties in clinical diagnosis, the lack of clinical cultures also hinders the monitoring of antimicrobial susceptibility trends among this population. Accordingly, alternative approaches such as periodic surveillance of antimicrobial resistance patterns may be more feasible in this setting.
There were several limitations to our study. The results from four RACFs located on a hospital campus may not be generalisable to other RACF populations. However, we noted that most subjects were residents of the facilities, with little traffic of patients from the hospital. The rates of infection remained relatively constant and were comparable to those reported in another longitudinal study in Australia.7 Although the McGeer criteria are the only internationally recognised epidemiological definitions of infection, they may not be completely specific or sensitive for microbiologically confirmed infection. We were not able to analyse the epidemiology of bacterial pathogens, owing to the small number of cultures performed. Recent point prevalence studies have assisted in defining the prevalence of colonisation with antimicrobial-resistant pathogens,8 but data are also required on their contribution to the burden of infection to inform antibiotic prescribing policies.
Work is required in developing policies to prevent infection among RACF residents, to explore the epidemiology of antimicrobial resistance in clinical infections, and in developing antimicrobial stewardship initiatives appropriate to this primary care setting.
1 Proportion and incidence rate of major types of health care-associated infection in four Melbourne RACFs, January 2006 to December 2010
2 Proportion control chart showing proportion of infections (episodes/OBD) per month in four Melbourne RACFs, January 2006 to December 2010 (n = 1114)
3 Overall antimicrobial use
5 Proportion of episodes with no documentation of collection of a clinical specimen, by type of infection (n = 659)
Appendix The “McGeer criteria”* — definitions of infection for surveillance in long-term care facilities