JSTOR

Methods

In July 2002, trained infection control staff members began to perform active NI surveillance, according to definitions of the Centers for Disease Control and Prevention,¹⁴ in non‐ICU wards of different medical departments (eg, internal medicine, surgery, and neurology) in numerous German hospitals already participating in other KISS components. As in all previous KISS modules, participation in this new component was voluntary. There were no restrictions with regard to hospital size or the number of beds per ward (ie, both university hospitals and community hospitals could participate). Wards were characterized as a special medical department, depending on the majority of patients cared for. If no decision as to a definite type of medical department could be made, wards were grouped as “other.” The minimum participation time in this module was 3 continuous months starting at each quarter of the year, but participation was not time limited. There was no maximum participation time. Participants could decide to perform surveillance of NI for only 1 or for both devices. In addition, the total number of respective use days for CVCs and UCs, as well as the total number of patient‐days, was determined monthly. Peripherally inserted central catheters usually were not in use on these wards. This data were collected by KISS, which calculated, for each ward, specific device use rates (ie, [number of device‐days divided by number of patient‐days ) and device‐associated NI rates (ie, [number of NIs divided by number of corresponding device‐days ‐associated incidence rate) for CVC‐associated bloodstream infections (CVC‐BSIs) and UC‐associated urinary tract infections (CAUTIs), respectively. In addition, on the basis of the complete number of NIs, device‐days, and patient‐days, stratified reference data were provided for all participants twice yearly. The reference data also were published on the KISS Web site (available at: http://www.nrz‐hygiene.de). Of course, all information concerning ward‐specific NI data remains absolutely confidential.

Results

Of 77 wards in 42 German hospitals, 29 wards (14 internal medicine wards, 12 surgical wards, and 3 other wards) reported CVC‐days and performed surveillance for CVC‐BSIs. These 29 wards contributed a total of 181,401 patient‐days, 8,317 CVC‐days, and 36 BSIs to the DEVICE‐KISS reference database. Sixty‐five wards (40 internal medicine wards, 6 surgical wards, 8 neurological wards, and 11 other wards) reported UC‐days and performed surveillance for CAUTIs. These 65 wards contributed a total of 445,536 patient‐days, 65,871 UC‐days, and 447 UTIs to the DEVICE‐KISS reference database.

The average duration of hospital stay was 8.2 days per patient. The reported number of absolute device‐days, calculated device use rates, and device‐associated NI rates for CVCs and UCs among non‐ICU patients are shown in Table 1. Table 2 shows the stratified numbers of wards, device‐days, patient‐days, mean CVC and UC use rates, and the mean device‐associated NI rates for BSIs and UTIs for internal medicine wards, surgery wards, neurology units, and other medical departments. In Table 3, device use rates and device‐associated NI rates for ICU patients included in the ICU component of the KISS database, which contains information from 323 ICUs in 224 hospitals, are compared with rates for non‐ICU patients.

Table 1.  Device Use Rates and Device‐Associated Nosocomial Infection (NI) Rates Among Non–Intensive Care Unit Patients

Open New Window

Table 2.  Device‐Associated Nosocomial Infections Among Non–Intensive Care Unit Patients, by Medical Department

Open New Window

Table 3.  Comparison Of Device Use Rates and Device‐Associated Nosocomial Infection Rates in Intensive Care Unit (ICU) and Non‐ICU Patients

Open New Window

Staphylococcus aureus (24%) and coagulase‐negative staphylococci (16%) were most often recovered in cases of CVC‐BSI. In 13 (36%) of 36 episodes of CVC‐BSI, no causative agent was reported. Escherichia coli (26%) and enterococci were the predominant causative agents of CAUTI. In 36 (8%) of 447 CAUTIs, there were no microbiologic findings reported by the participating wards.

Discussion

In many wards, several infection control measures are performed routinely to reduce the risk of NI, despite the fact that some of the measures have never been shown to be effective for the prevention of NIs. On the other hand, the use of evidence‐based infection control measures might be neglected. This phenomenon leads to a great proportion of preventable NIs. However, those deficits in infection control management usually do not become evident if the median infection rates of other facilities are unknown. Until there are reliable reference data for NIs, it is difficult for medical departments to evaluate their endemic NI rate and realize a possibility for improvement. In addition, to generate valid reference data, stratification of patient populations needs to be performed, because underlying diseases influence the device use frequency, which is a risk factor for NI acquisition.

KISS has generated NI reference data for ICUs since 1997. In addition to an elevated proportion of BSIs without microbiological findings, our data have shown that the distribution of commonly recovered pathogens is comparable to the distribution of causative pathogens recovered from patients in ICUs. However, there are obvious differences between ICU and non‐ICU patients in terms of use rates of medical devices. As shown in Table 3, the frequency of CVC use is 69.2% among ICU patients, compared with 4.6% among non‐ICU patients. The rate of use of UCs is 5 times higher among ICU patients than it is among non‐ICU patients (77.8% vs 14.8%). Trick et al.¹⁵ showed that unjustified CVC‐days were more common among non‐ICU patients, which might contribute to higher NI rates in these wards. This finding becomes particularly important for hospitals in which the total number of used CVCs is greater in non‐ICU wards than in ICUs.¹⁶ In addition, the length of stay differs between ICU and non‐ICU patients. The mean duration of stay in an ICU is 3.6 days (data available at: http://www.nrz‐hygiene.de), but the mean duration of stay in non‐ICU wards is 8.2 days. Therefore, the probability of detecting an NI is greater for non‐ICU patients, whereas some infections in ICU patients become evident after the patients are transferred to general wards. In contrast, despite an increased severity of illness among ICU patients, device‐associated NI rates are higher among non‐ICU patients (Table 3). This finding might be the result of improved training with regard to device handling or to an increased amount of antibiotic treatment given to ICU patients. As a result of these facts, NI reference data generated from ICUs may not be used to judge non‐ICU patient care.

Within the population of non‐ICU patients, our data show that the rate of device‐associated NI varies between different medical departments (Table 2). The rate of CVC‐BSIs among patients in internal medicine and surgical wards is quite similar (4.5 and 4.7 BSIs per 1,000 CVC‐days, respectively). In contrast, the rate of CAUTIs is more than twice as high among patients in internal medicine wards (5.1 CAUTIs per 1,000 UC‐days) as it as among patients in surgical wards (2.0 CAUTIs per 1,000 UC‐days). This finding might be explained by differences in the underlying diseases in patients in different medical departments, which we do not measure in our surveillance system explicitly. For example, the significantly elevated rate of CAUTI among neurological patients (12.8 CAUTIs per 1,000 UC‐days) could be caused by a large proportion of patients with neurogenic bladder dysfunction in neurological wards, which is a risk factor for acquisition of UTI itself.¹⁷

Because non‐ICU patients represent the majority of patients in hospitals, there definitely is a need for valid reference data for these patients. When we searched the literature, we found only a few data for device‐associated incidence rates of NI among non‐ICU patients (Table 4). Furthermore, if no equal definitions of NI in published studies are used (eg, Centers for Disease Control and Prevention definitions) the results of those studies are hardly comparable. To close this gap of information, DEVICE‐KISS has begun to generate a reference database for non‐ICU patient care. These reference data for participating and nonparticipating wards are available on the KISS Web site (available at: http://www.nrz‐hygiene.de). Thus, evaluation of the individual infection control practice turns out to be possible.

Table 4.  Device‐Associated Nosocomial Infection Rates In Non–Intensive Care Unit (ICU) Patients in the Present Study and Other Studies

Open New Window

Device‐oriented surveillance is more cost effective than hospital‐wide surveillance, because it concentrates on patients with devices, who are the minority of the patient population, as demonstrated by our data. However, there are some limitations to this approach. First, selection bias might influence our results. It is likely that personnel in wards with an NI problem are more interested in participating in surveillance than are personnel in other wards. Thus, our data are likely to overestimate the actual NI rate. Second, our overall sample size still is low, and the duration of participation of a few wards may be too short to detect any case of BSI during a 3‐month period. Third, even after stratification by the kind of medical department, there are still great differences within each group. Further differentiation of patients might be necessary when our sample size has increased.

There is no exact rate that serves as a marker for an elevated rate of NI. If personnel in a given ward recognize that the NI rate in the ward exceeds the 75th percentile, several possible explanations have to be taken into consideration. First, frequent routine diagnostic microbiological screening might lead to detection (eg, of a greater number of asymptomatic UTIs). For ICUs, a significant difference in NI rates could be shown for units that perform screening, compared with others that do not.²³ Second, it is possible that the patients cared for are not representative of the category in which they are grouped. Further differentiation and stratification of patients might be required. Third, outlier positions also might occur if an accumulation of NIs had occurred solely by chance. Finally, if none of these explanations applies but NI rates are significantly elevated without another reasonable explanation, the infection control management strategy in these wards might need to be improved. Non–evidence‐based, routine infection control processes may be abandoned in favor of increased compliance with a few evidence‐based prevention measures instead. Education of staff in handling devices according to official recommendations, as well as surveillance of possible transmission of pathogens between different patients, might be reasonable approaches. Even if NI rates are within normal limits, the comparison of device use rates might show potential for reduction of unnecessary use of CVCs or UCs as risk factors for NI acquisition.

Recently, it has been decided to combine the 3 national healthcare surveillance systems of the Centers for Disease Control and Prevention (National Nosocomial Infections System, National Surveillance System for Health Care Workers, and Dialysis Surveillance Network) into a single National Healthcare Safety Network. In the patient safety component of the National Healthcare Safety Network, a device‐associated module will be introduced in which participants may monitor device‐associated infections outside the ICU as well.²⁴ It will be quite interesting to compare our data with future findings of the National Healthcare Safety Network.