Trends in nosocomial bloodstream infections following health care restructuring in Alberta between 1999 and 2005

1Division of Infectious Diseases, Department of Medicine, University of Alberta Hospital; 2Division of Infectious Diseases, Department of Pediatrics, Stollery Children’s Hospital, Edmonton, Alberta Correspondence: Dr Mao-Cheng Lee, Division of Infectious Diseases, Department of Medicine, University of Alberta Hospital, #200, 10150–102 Street, Edmonton, Alberta T5J 5E2. Telephone 780-451-3702, fax 780-454-2845, e-mail Mao-Cheng.Lee@albertahealthservices.ca Hospital-acquired (nosocomial) infections are the most common complication of health care delivery (1). Bloodstream infections (BSIs) are especially serious, and, because they are readily detectable through prospective surveillance, they are useful markers of long-term trends in infection rates. Occurrences of health care-acquired infections are known to be sensitive to processes of care (2,3). The 1993 to 1996 restructuring era of public health care delivery in the province of Alberta involved a dramatic budgetary funding cut for existing health care services, and a significant shifting of certain types of patients from acute tertiary hospitals to community-based hospital settings. This also resulted in a significant reduction in the operational budget of the University of Alberta Hospital and the Stollery Children’s Hospital (located within the University of Alberta Hospital) in Edmonton, Alberta, and a marked concomitant increase in patient acuity and in primary and secondary BSI rates (4). In the present follow-up study, we sought to analyze the trends of both hospital activity and patient acuity as well as BSIs during the subsequent seven years following the previous report (4).

CONCLuSIONS : Selon ces données, au fil du temps, les hôpitaux peuvent graduellement s'adapter aux modifications à la situation des soins aux patients et, dans cet exemple, aux complications infectieuses de contrôle pour la prestation des soins. neurology, general surgery, internal medicine and its subspecialty services, renal dialysis, emergency and solid-organ transplant (SOT) for both the pediatric and the adult population. Obstetrics/ gynecology and ophthalmology services are not included. There are seven intensive care units ([ICUs] -coronary, adult cardiac surgery, adult neurosurgery, adult medical-surgical, burns, pediatric and neonatal), which make up 14.4% of rated beds. No change in services provided occurred during the study period.

Hospital volume and acuity
To assess the trends in hospital patient care volume, the number of annual patient-days (overall) and admissions in both hospitals and in the seven ICUs were obtained. To assess changes in the prevalence of risk factors for BSI, facility-wide days-of-use of total parenteral nutrition (TPN) supplied by the hospital pharmacy (5,6), central venous catheter (CVC) days-of-use by ICUs (excluding neonatal ICUs) (7)(8)(9) and annual number of all on-site SOTs (kidney, liver, heart, lung and heart-lung) were determined. Data on the number of hemodialysis runs performed annually by the hemodialysis unit, the extent of mechanical ventilation use in the ICUs or CVC days-of-use outside of the ICUs for the period of the current study were not available for use as additional indicators of acuity.

Hospital-acquired bSIs and rates
Prospective surveillance for hospital-acquired BSIs in all units has been performed in the facility since 1986 (except 1998) by prospective monitoring of clinical microbiology laboratory reports as well as review of patient charts by the hospitals' infection control practitioners, as previously described (10,11). The Centers for Disease Control and Prevention (USA) criteria for infection are used (12,13). Infection rates are calculated using patient-days, patient admissions and catheter-days as denominators where applicable -all supplied by the hospitals' information systems department and the respective ICUs. ICU CVC-associated BSI rates were calculated using CVC line-days as denominators, which were supplied by the ICU departments (9,14). Primary BSIs are attributable to CVCs only, which may include a small number of transient in-ICU hemodialysis and peripherally inserted central catheter lines, and are exclusive of all peripheral lines. Secondary BSIs are attributable to all other body sources and exclusive of all intravenous devices -central or peripheral. Infections and infection rates occurring in the seven years (1999 to 2005) following the end of the 1993 to 1996 Alberta health care restructuring were analyzed.

Statistical calculations
To detect any significant monotonic, linear temporal trends in the rates, a nonparametric Mann-Kendall test (15) was performed. In the Mann-Kendall test, a negative (positive) value of S, where S is the Kendall trend score, indicates a downward (upward) trend. The magnitude of the linear trend in terms of slope Q was then estimated using a nonparametric Sen's method (15). P value shows statistical significance of the trend analysis. The calculations were performed using an Excel (Microsoft Corporation, USA) macrotemplate MAKESENS (Mann-Kendall test for trend and Sen's slope estimates [16]).

RESuLTS
Hospital volume and acuity (Table 1) Over the 10 years following the restructuring era, there was a steady rise in the number of total admissions at the University

Hospital-acquired secondary bSIs and rates (Table 3)
Although the rate of secondary BSIs in the adult and pediatric groups of patients each showed a positive downward trend during this study period (S=-11; P=0.06 for both; Q slope=-0.04 and -0.10, respectively), they did not reach statistical significance individually. The combined rate resulted, however, in a significant downward trend overall (S=-15; Q=-0.03; P=0.016). The rates of pneumonia-related BSI displayed a significant reduction over the duration of the study period (S=-15; P= 0.017).

DISCuSSION
In the present study, we have found that the overall combined BSI rates, both primary and secondary, among adult and pediatric populations served by the University of Alberta Hospital/ Stollery Children's Hospital fell significantly following the end of the restructuring era between 1993 and 1996. By 2005, the hospitals' BSI rate had returned to the baseline rate seen immediately before the restructuring era (analysis data not shown). The fall in the primary BSIs, particularly among adults and occurring in most circumstances as a complication of intravascular catherization, was reflected by the marked decline in Gram-positive bacteremia, and S aureus in particular, while bacteremia involving other common pathogens remained mostly unchanged. Over the same period, the hospital's patient volume and acuity, as marked by patient admission volumes, length of stay (expressed in patient-days), and BSI risk factors such as CVC use, TPN use and number of SOTs, increased substantially.
Our current findings are counter to the trends observed during the restructuring period in Alberta (1993 to 1996), which saw a period of rapid health care downsizing and shifting care and resources from acute tertiary hospitals to community-based hospital settings (4). During that era, bed numbers at the University of Alberta Hospital/Stollery Children's Hospital fell by 10%, and volumes declined following a 19% drop in the annual number of admissions and a 17% drop in the annual patient-days parameter. Acuity soared concomitantly. Markers of hospital acuity such as ICU use of CVCs, number of TPN days, and number of hemodialysis runs reflected that trend, with increases seen at 41%, 25% and 9%, respectively (4). Concurrently, some services within tertiary centres saw a marked increase in patient-days (neurosurgery 49%, nephrology 30%, orthopedic surgery 24% and general surgery 16%). During this same period, hospital-acquired BSIs saw a dramatic increase such that toward the end of the restructuring period, there was an annual 31% increase in the number of hospital-acquired BSIs and a 60% increase in BSI rate compared with that observed in the prerestructuring era.
In spite of the soaring of facility volumes and acuity since the end of the restructuring era, the present study shows that BSI rates have steadily declined in the 10 years that followed to near prerestructuring baseline. During the same period, Census Canada data revealed that the city of Edmonton saw a steady population growth of 18.5%, 8.1% of which occurred before the beginning of the current study. Such concomitant population growth before and during our study may have accounted in part for the increased hospital activity and volume we have observed. In contrast, it is unlikely that a technological improvement in care is solely responsible for the decline in BSI rates. Improved care of CVC lines and the introduction of antibiotic-impregnated catheters (which occurred midpoint of the study period) may have accounted for some of the observed trend, especially in the decline of the primary BSI rate. However, introduction of antimicrobialimpregnated catheters would be expected to have a singlestep effect on BSI rate rather than a gradual downward trend.
No systematic institution-wide process to improve the quality of catheter care was implemented during the study period. Furthermore, the reductions seen were widespread, involving pediatric and adult patient groups, and secondary and primary BSIs. Thus, a single change in the process of care is unlikely to have resulted in such widespread and progressive decline in BSI rates. Rather, the most likely explanation for our observed BSI rate decline, considering all observations gathered, would be that a collection of potentially heterogeneous health care process improvements occurred within the University of Alberta Hospital/Stollery Children's Hospital during our study period. Examples of such care processes would include improved CVC care and use, mouth care, skin antisepsis and hand hygiene, among others. Unlike BSIs involving Gram-positive organisms, BSIs with Pseudomonas aeruginosa very infrequently involve CVCs as primary sources (17,18). Rather, bacteremia with Pseudomonas aeruginosa is often secondary, originating most commonly from primary sites such as the respiratory system (eg, ventilator-associated pneumonia), and, to a lesser extent, urinary and surgical wound sites (17). This offers a plausible explanation for the concurrent decline in the rates of Pseudomonas aeruginosa BSIs and that of pneumonia-derived secondary BSI source in the present study. Although multifactorial in cause, enhanced quality of care through improved oral hygiene and ICU-assisted respiratory support would invariably account for the observed reduction in pneumonia rates at our site over the study period.
Thus, our data were most compatible with a gradual improvement in quality of care, returning BSI rates to the prerestructuring baseline. These findings are not surprising. As stated by Pittet and Donaldson (3) in their launch of the Global Patient Safety Challenge and WHO World Alliance for Patient Safety, adverse events in health care settings are not just a series of unconnected one-off incidents. Multiple observations have substantiated that institutional level factors of health care organizational culture, leadership, staff education and morale, and staffing levels are critical to controlling health care-associated infections (2,19). Our data imply that over time, hospitals can gradually adjust to changing patient care circumstances and, in this example, control BSI rates as representing infectious complications of health care delivery.