The mean rates of NI and DAI in our adult MS ICU during the study period were much lower than those reported by the INICC as well as for 173 ICUs in developing countries [5
], were similar to those reported by 1,545 hospitals in the US through the CDC NHSN [3
], and were slightly higher than those indicated by 586 ICUs in the German Surveillance System (ICU-KISS) [7
]. Reasons for these high DAI rates in the INICC report and developing countries may include resource limitations, lack of legal enforcement of the infection control program, and poor adherence to infection control guidelines [5
]. The prospective hospital–wide surveillance and infection control program has been established for nearly 30
years in our hospital, which made a great effort to control infection by implementing infection control bundles and educational programs. The increase of device-related infections is not obvious after 2004, except for CAUTI. These strategies showed effectiveness in controlling DAI rates and suggest the necessity of infection control bundles implementation.
The common device–associated pathogens show geographic variation in distribution. A. baumanniiS. aureus
, and P. aeruginosa
were the three most common VAP pathogens in our study, the US CDC NHSN study, and the SENTRY antimicrobial surveillance study, although their percentages differed between studies [18
]. The percentage of isolates of MRSA (p
0.678) and IRPA (p
0.953), but not isolates of IRAB (p
0.001) remained relatively constant. However, any variation in these percentages would not be statistically significant and might rather be due to chance than to an actual variation.
In contrast, higher rates of A. baumannii
and C. albicans
isolation compensated for the relatively low rates of CoNS and Enterococcus
spp. isolation in cases of CLABSI, while C. albicans
was replaced by NAC spp in cases of CAUTI. Differences in clinical setting, institution, study period, target population, and specific infection type might account in part for differences between studies. CoNS was less frequently identified in CLABSI, because our criteria were slightly different from the US CDC definition for laboratory–confirmed BSI. The CDC defined skin contaminant BSI in 1998 and 2004 as 'the common skin contaminant (e.g., CoNS) is cultured from at least one blood culture from a patient with an intravascular line, and the physician institutes appropriate antimicrobial therapy'. However, CoNS BSIs in our study were enrolled if the patient had only one blood culture of CoNS that was positive but then microorganisms cultured from the tip of the intravascular device that were also CoNS. The percentage of CoNS isolates was expected to be at least that of S. aureus
isolates reported in previous studies [15
However, the frequency of A. baumannii
spp. in specimens from patients with CLABSI was also reported to be increasing in other hospitals in Taiwan as well as several Asian countries such as Turkey and Thailand [20
], although the frequency of NAC spp. represented by only one Candida
spp. has also been rising in specimens from patients with CAUTI. Early and empirical usage of broad spectrum antibacterial agents in critically ill patients and preemptive administration of fluconazole are common factors contributing to the increase in frequency of isolation of these relatively resistant pathogens [22
]. Use of indwelling catheters increases susceptibility to those multi–drug resistant pathogens and is associated with biofilm formation [26
The high prevalence of MRSA is a common problem worldwide, and this situation was much more severe in our institute. Our data showed a lower incidence density of S. aureus
but a higher proportion of MRSA. The percentage of MRSA infections was 74.4–84.1% in the INICC report [5
], 54.4–65.2% in the US CDC NHSN report [18
], and, in the Asia–Pacific region, it was 38.2% in the SENTRY Antimicrobial Surveillance Program report (2003–2004) [28
] and 20–85% in the Tigecycline Evaluation and Surveillance Trial (TEST) report [29
]. MRSA rates were decreasing in many European countries but not in USA [18
]. The more severe illnesses of patients and more frequent use of broad–spectrum antibiotics might account in part for the high rate of MRSA isolation from patients in ICU at major teaching hospitals [29
]. Another possible explanation is the clonal spread of resistance genes or resistant strains [31
], but molecular analysis will be needed to prove this hypothesis.
According to the infection control policies in our ICU, when a patient was admitted to the ICU, a multi–drug resistant (MDR) checklist was used to inquire about MDR pathogens including MRSA infection or colonization. If MRSA had been isolated, then contact precautions were implemented. Furthermore, we have promoted hospital–wide hand–washing activity from 2006 to the present. The infection control team also carries on the non-warning investigation of hand-washing and of isolation precautions in each season, and gives feedback of the results to the unit. Infectious disease doctors assist in carrying on the infectious disease treatment and the antibiotic use in the ICU. MRSA infection rates have been reduced by year from 2006. Interestingly, the rates of antibiotic resistance for pathogens other than MRSA was lower at our hospital than those in previously published reports and lower than those for all NIs reported by the TEST and SENTRY antimicrobial surveillance programs [8
However, despite the carbapenems being the most active antimicrobials against Acinetobacter species, the increasing development of significant carbapenem resistance among Acinetobacter species has been reported [3
]. In the present study, the average percentage of A. baumannii
isolates resistant to imipenem was 22.2%. The rate of ICU patients with IRAB DAI has been rapidly rising (from 6.1% to 34.3%). Among Enterobacteriaceae
, Ciprofloxacin–R E. coli
and Ceftazidime–R K. pneumoniae
from 2003, and Ceftazidime–R E. coli
from 2004, had significant increases. This finding revealed that the resistance of gram-negative bacteria has increased, the development of which should require closely monitored.
Aside from the fact that DAI is an important prognostic factor of mortality., several previous studies have shown that the mortality rate attributed to DAI is 1.65–2.75 times higher than that attributed to no infection [13
]. Our study supports the findings of these published reports. In the present study, the multiple regression analysis indicated that patients with DAIs (compared to patients with no HAI) had significantly increased likelihood of mortality (p
0.05). Moreover, the annual 30–day mortality rates of CAUTIs and CLABSIs had significant changes over the period 2000 through 2008. These results may be caused by chance, because this study period did not change substantially in terms of medical care, novelty medical technology, and patient disease severity.
In addition to the above–mentioned findings, we used a multiple regression analysis approach to adjust covariables, in addition to demographics, invasive devices, and laboratory investigations. We also identified severity of illness using APACHE II scores as a predictor of mortality, with the results indicating that the hazard of mortality is associated with increasing scores. Patients who died with DAI infection were usually already severely ill and their existing illness, rather than the DAI, was often the main cause of death. Thus, an important prognostic factor was the severity of their illness, which resulted in an increased likelihood of mortality [14
]. We also found that patients with blood creatinine over 1.5
mg/dL were the highest risk group for dying. Excluding an endogenous effect, the reason may be that many patients in this group were receiving hemodialysis with CVCs inserted.
The rates of DAIs of all types decreased during the period 2005–2006, but this decrease was maintained only for VAP. Similarly, decreased NIs and DAIs were reported by other epidemiological studies [3
]. Effective control of respiratory tract infection, antibiotics stewardship, implementation of traffic control practices, improved adherence to hand hygiene and contact precaution practices as a result of the severe acute respiratory syndrome (SARS) pandemic in 2003 as well as decreased rate of resistance of MRSA, IRPA, and IRAB in 2005 might account for this decrease [37
Some limitations of the present study should be noted. The study was performed at a single medical center. However, the results could be provided to the hospital as a part of the teaching or research mission. This study was a retrospective nine-year survey which might have some potential biases. In the analysis of long–term changes in infection rates or mortality rates, we must consider whether changes in the population, advances in laboratory diagnostic techniques, changes in exposure to risk factors, microbial culture and other factors lead to increased or decreased rates. However, there did not occur any outbreaks of DAIs during the study period, except for the SARS outbreak.