Clin Microbiol Infect 2011; 17: 513–518
Little is known about temporal changes in the epidemiology of Staphylococcus aureus bacteraemia. The objective of the present study was to analyse changes in the incidence and mortality of adult S. aureus bacteraemia in Iceland. Individuals 18 years or older with a positive blood culture for S. aureus between 1 January 1995 and 31 December 2008 were identified, with the participation of all clinical microbiological laboratories performing blood cultures in Iceland. Infections were categorized as nosocomial, healthcare-associated or community-acquired. National population statistics and dates of death were retrieved from the National Registry. During the study period, 692 individuals from 19 institutions had 721 distinct episodes of S. aureus bacteraemia. The incidence rose from 22.7 to 28.9 per 100 000 per year during the period (p 0.012). Nosocomial infections comprised 46.3% of cases, 14.6% were healthcare-associated, and 39.1% were community-acquired. The proportion of nosocomial infections decreased during the period (p <0.001), whereas an increase was seen in the proportion of community-acquired infections (p <0.001). All-cause 30-day mortality decreased from 25.0% to 8.1% (p 0.001) and 1-year mortality decreased from 37.0% to 27.9% (p 0.061) between the periods 1995–1996 and 2007–2008. Four cases of bacteraemia caused by methicillin-resistant S. aureus were seen (0.6%), none of which was fatal. In conclusion, there was a significant increase in the incidence of S. aureus bacteraemia in Iceland between 1995 and 2008. Concomitantly, there was a significant reduction in mortality, towards one of the lowest reported. Further studies are needed to understand the basis for these changes.
Staphylococcus aureus is a major cause of both nosocomial and community-acquired bloodstream infections . However, studies on the incidence of S. aureus bacteraemia (SAB) are sparse [1–12]. SAB-associated mortality was over 80% in the pre-antibiotic era, and today the infection still carries a significant risk of morbidity and mortality . Reported mortalities generally range from 11% to 40% in developed countries [2–4,12,14–18]. However, most studies have been institution-based rather than population-based, and have rarely assessed long-term outcome [16,19,20]. Despite the importance of the infection, few data have been published regarding temporal changes in the incidence and mortality of SAB [2–4,9,11,14,17].
The objective of the present study was to evaluate trends in the incidence, short-term mortality and long-term mortality of SAB in Iceland during 1995–2008. A second objective was to assess the proportions of nosocomial and healthcare-associated infections and trends in antimicrobial susceptibility.
Materials and Methods
Study population and protocol
Iceland is an island in the North Atlantic Ocean with an area of 103 000 km2 and a population of just over 300 000. The majority of inhabitants live in the south-west, around the capital. The healthcare system is run by the government, with a university hospital in Reykjavik and a teaching hospital in Akureyri. Smaller regional hospitals are spread around the country.
Cases of SAB occurring between 1 January 1995 and 31 December 2008 were identified retrospectively by clinical microbiological laboratories that perform blood cultures in Iceland. All collected blood cultures in the country are sent to one of two laboratories. Individuals 18 years or older with a positive blood culture for S. aureus were included. Cases were identified from 19 institutions, with the majority of cases being from the hospitals in Reykjavik. Information on admission and discharge dates was obtained from medical records. Information on hospitalizations 3 months prior to the bacteraemia was obtained for those who were diagnosed within 2 days from admission.
Information on the national population and dates of death was available from the Icelandic National Registry. During the study period, the number of people aged ≥18 years increased from 188 888 to 238 587 (26.3%), with their mean age increasing from 30.9 to 33.5 years. The number of people aged ≥50 years rose from 62 360 to 90 459 (45.1%), i.e. from 33.0% to 37.9% of the adult population.
The study was approved by the National Bioethics Committee and Data Protection Authority.
An episode of SAB was defined by the isolation of S. aureus from at least one blood culture bottle. The blood culture systems used at the university hospital were Bactec in 1995, Difco ESP in 1996–2002 and BacT/Alert from 2002, and at the teaching hospital they were Septi-Chek in 1995–1999, Difco ESP in 1999–2008 and BacT/Alert from 2008. All positive blood cultures were considered to represent clinically significant SABs. Repeated isolation of S. aureus within 90 days after the first isolation was thought to represent the same episode unless the isolate was found to have a different antimicrobial susceptibility profile. For those diagnosed in the first 3 months of 1995, it was verified that the SAB was not a relapse of an infection just prior to 1995.
Nosocomial bacteraemia was defined as one for which the sample for the first positive blood culture was drawn more than 2 days after hospital admission. Positive cultures from samples drawn 2 days or less after hospital discharge following a minimum of a 2-day stay were also defined as nosocomial. Healthcare-associated bacteraemia was defined as one that was not nosocomial but that occurred in individuals who had been admitted to hospital for more than 2 days in the 90 days prior to bacteraemia. Community-acquired bacteraemia was defined as neither nosocomial nor healthcare-associated.
The data are given in terms of number of observations, medians, quartiles, and range and means. Linear time trends in incidence and mortality rates were estimated using the chi-squared trend test. Time trends in other data and associations between variables were evaluated by Kendall’s correlation. Pearson’s chi-square test was used for comparison of categorical data. The Mann–Whitney U-test was used to compare two groups, and the Kruskal–Wallis test was used to compare the acquisition groups by age at infection. Kaplan–Meier curves were used to display survival data, and groups were compared by the log-rank test.
The level of significance was set at 0.05. For processing the data, the SPSS 17.0 program package for Windows was used.
In the study period, 692 individuals had 721 distinct episodes of SAB. There were 29 re-infections (4.0% of all episodes) in 27 individuals, with a median time to re-infection of 331 days (range: 50–2632 days). For the whole period, the average incidence of SAB was 24.5/100 000 adults per year. A 27.3% increase in annual incidence could be seen from 1995 to 2008, from 22.7 to 28.9 per 100 000 (p 0.012, linear trend over years). The incidence increased with age, being 6.8/100 000 per year in the age group 18–34 years and 102.1/100 000 per year in the age group 75 years and older (p <0.001, trend for age) (Fig. 1). The mean age at diagnosis was 62.6 years (range: 18–99 years), and the male/female ratio was 1.44.
There were 17.3% episodes with a single positive blood culture in 1995–1999 (34/197, information missing for 19), 19.6% (44/225, information missing for 17) in 2000–2004, and 17.5% (46/263) in 2005–2008. The respective figures for polymicrobial episodes (excluding coagulase-negative staphylococci in one set) with S. aureus growing in more than one blood culture bottle were 5.1% (10/197), 8.0% (18/225) and 5.3% (14/263) in the same periods.
Of 721 infections, 334 (46.3%) were nosocomial, 105 (14.6%) healthcare-associated, and 282 (39.1%) community-acquired. Changes in these proportions were seen during the study period (Table 1). The median age of those with a nosocomial infection was 69.0 years (interquartile range (IQR) 55.0–78.0), that of those with healthcare-associated infections was 62.0 years (IQR 46.5–73.0), and that of those with community-acquired infections was 66.0 years (IQR 48.0–75.0) (p 0.012). Of those with community-acquired infections, 2.5% later had a re-infection; the proportions for healthcare-associated and nosocomial infections were 7.7% and 8.2%, respectively (p 0.022).
(n = 216)
(n = 242)
(n = 263)
|Incidence, per 105 per year|
|Total >18 years||22.4||23.1||28.9||0.012|
|Mean age (years)||63.0||62.3||62.5||NS|
|30-day mortality (%)|
|Total >18 years||22.2||18.7||11.4||0.001|
|365-day mortality (%)|
|Total >18 years||38.9||32.8||28.2||0.061|
|Mortality rate, per 105 per year||5.0||4.3||3.3||0.048|
All-cause 30-day mortality was 17.1% (123/720, one case lost to follow-up) and 365-day mortality was 33.0% (237/719, two cases lost to follow-up). Among SAB episodes with a single positive culture, the 30-day mortality was 17.7% (22/124). Fig. 2 shows changes in mortality during the study period. One-year survival curves for different time periods are shown in Fig. 3. The mortality rate was 4.26 deaths/100 000 adult population per year. It decreased by 64.3%, from 6.6 to 2.4 deaths/100 000 population per year, in the periods 1995–1996 and 2007–2008, respectively (p 0.048, linear trend over years). Increasing age was associated with higher SAB 30-day and 365-day mortalities. These were 0% (0/62) and 3.2% (2/62), respectively, among the patients 18–34 years old, and 27.8% (58/209) and 47.8% (100/209), respectively, among those 75 years and older (p <0.001 for both trends for age). Thirty-day mortality was 15.8% among males and 18.9% among females (p 0.27). Thirty-day mortality was 23.4% (78/334) for nosocomial SAB, 15.2% (16/105) for healthcare-associated SAB, and 10.3% (29/281, one lost to follow-up) for community-acquired SAB (p <0.001). Mortality at 365 days was 41.1% (137/333, one lost to follow-up), 36.2% (38/105) and 22.1% (62/281, one lost to follow-up) for the same groups, respectively (p <0.001). One-year survival curves for the three groups are shown in Fig. 4.
Four cases of bacteraemia caused by methicillin-resistant S. aureus (MRSA) were noted (0.6%) during the study period; all of these patients survived. An 80-year-old male without a known focus for his SAB was diagnosed in 1996, and a 78-year-old male with infected kidney stones as the focus was diagnosed in 2000. In neither case was a travel history available from medical records. Two cases occurred in 2008: a bacteraemic skin infection in a 43-year-old man with psoriasis who had immigrated from eastern Europe 6 months earlier, and an endocarditis in a 47-year-old diabetic sailor who had recently been in England. Susceptibility to penicillin was seen in 18.2% (130/714, lacking information for seven isolates). Resistance to erythromycin and clindamycin was seen in 4.3% (31/721) and 1.4% (10/708, lacking information for 13 isolates), respectively. Temporal changes in susceptibility rates are shown in Table 1. No correlation was observed between antibiotic susceptibility and mortality (data not shown).
The average annual incidence of 24.5 SAB cases/100 000 among adults is lower than that reported in a Danish study (30.5 cases/100 000 per year) . Other studies have reported incidences of 14–41 cases/100 000 per year [1,3–12]; most of these studies have included children, among whom the incidence of SAB is generally low. The present study showed a significant, 27% increase in SAB incidence during the period concerned. A similar trend was seen in two Scandinavian national studies. In Denmark, the incidence rose by 40% from 1981 to 2000, and in Finland it rose by 55% from 1995 to 2001 [2,3]. A recent study from Minnesota, however, did not report a statistically significant increase in SAB incidence from 1998 to 2005 . The noted increase in SAB incidence could be attributed to an increase in predisposing risk factors for S. aureus infections, resulting in a larger population at risk. The population is getting older, higher numbers of individuals are being diagnosed with malignancy, and more people are living with chronic diseases such as diabetes and obesity [21,22].
The proportion of nosocomial infections (46%) is somewhat lower than generally observed [2,4,12,15,16], but identical to that of the study from Finland . Most studies have not included healthcare-associated infections as a special entity; however, Laupland et al.  reported a ratio of 36%. Because a limited definition was used, as a result of complete clinical information not being collected, it is likely that the proportion of healthcare-associated infections (15%) is underestimated in the present study. SAB episodes in patients receiving ambulatory intravenous treatments or undergoing haemodialysis were considered to be healthcare-associated only if those patients were admitted to a hospital in the 90 days prior to bacteraemia. The reason for the low and decreasing proportion of nosocomial infections in our study is not clear. It could, in part, be a consequence of the population-based nature of the study, whereby a higher number of cases from smaller hospitals were included. Also, there is a possibility that improved infection control measures after the year 2001 could be having an effect on reducing the risk of transmission of S. aureus. These included the following: the establishment of a department of infection control in the university hospital in 2001; an MRSA-targeting search and destroy policy, focusing on screening and isolation of at-risk patients along with eradication of MRSA in positive individuals, starting in 2001; and a hand hygiene promotional campaign starting in 2004, emphasizing the use of alcohol hand sanitizers at the university hospital.
The all-cause 30-day mortality of 17.1% for the whole period is low in comparison with other studies, and our 8.1% mortality rate for 2007–2008 is among the lowest reported. To the knowledge of the authors, the lowest mortality rates are those noted by Melzer et al.  of 5.1% and 11.8% for nosocomial bacteraemia caused by methicillin-sensitive S. aureus (MSSA) and MRSA, respectively, in two hospitals in London (including all cases with a positive blood culture for S. aureus, as was done here). With regard to other national studies, SAB mortality was 22% in Denmark from 1996 to 2000 and 17% in Finland from 1995 to 2001 (both studies included all positive cultures) [2,3].
The 1-year all-cause mortalities of 27.9% in 2007–2008 and 33.0% for the whole study period are lower than the 56% in 1988–1992 reported in a study from Israel (excluding probable contaminations), the 37.6% in 1997–2000 in Germany reported by Fätkenheuer et al. (including all positive cultures), and the 46.2% in a recent study on MRSA bacteraemia [16,19,20]. In the German cohort, there was an unusually high proportion of nosocomial infections (77%), which could have contributed to a higher mortality rate . With regard to those who died within 1 year, it is interesting that 52% (123/237) of the patients died in the first 30 days and 48% (114/237) in the next 11 months; these proportions are almost identical to those reported by Fätkenheuer et al. . All-cause 1-year mortality may reflect both deaths from serious underlying comorbid conditions and long-term sequelae of SAB. This indicates that it may be important to look not only at the short-term mortality of patients with SAB, and calls for further studies to assess long-term outcome.
The 68% decrease in 30-day mortality (p 0.001) and the 25% decrease in 365-day mortality (p 0.06) are of special interest. In-hospital mortality was reported to have decreased by 37% in Denmark from 1981 to 2000, and by 14% in a Swiss study from 1980 to 2002 [2,14], whereas other studies have not shown a change in outcome over time [3,4]. The reasons for the low and decreasing mortality in Iceland are not clear. A decreasing mortality rate for bloodstream infections in general has, however, also been observed, and has in part been thought to reflect improvements in supportive treatment . Additionally, the low mortality rate could be influenced by the relatively low rates of antibiotic resistance, including the low rates of MRSA bacteraemia usually associated with poor outcome as compared with MSSA bacteraemia [4,25].
The rate of MRSA bacteraemia (0.6%) reported here is much lower than those in most previous studies but similar to that in the national Finnish and Danish studies [2,3,26]. Nosocomial SAB is often seen being caused by MRSA, but this study demonstrates that MSSA is a significant cause of nosocomial SAB. The reasons for the low level of MRSA could be the geographical location of Iceland and the search and destroy policy against MRSA implemented in hospitals in Iceland as well as in the other Nordic countries and The Netherlands . The rate of penicillin susceptibility (18%) in this study is higher than is generally observed among S. aureus isolates [28,29], possibly making broader sets of treatment effective. It is nevertheless close to the penicillin susceptibility of MSSA isolates reported elsewhere [12,30]. Although increasing levels of resistance against clindamycin and erythromycin were observed, they were considerably lower than generally observed [12,28,29].
The study has several limitations. All-cause 30-day mortality is probably an overestimate of mortality resulting from SAB. Because S. aureus was never considered to be a contaminant of blood cultures, possible contaminations have probably been missed, leading to an underestimation of the association between true SAB and mortality. As previously mentioned, another limitation is the limited definition of healthcare-associated infections used. Strengths of the study are that it is based on a whole nation rather than on a limited population or institution, and that there was long-term follow-up of the patients in this setting.
In conclusion, this population-based study demonstrates a significant increase in the incidence of SAB during 1995–2008 in Iceland. Concomitantly, it reveals a significant decrease in mortality, towards one of the lowest reported. Further studies are needed to aid in our understanding of the basis of these changes, in order to further reduce the incidence and to improve the outcomes of SAB.
The authors thank M. L. Helgadottir, biomedical scientist at the Department of Microbiology, Landspitali University Hospital, for her great help with data collection.
This study was financially supported by a grant from the Scientific Trust of Landspitali University Hospital. All authors report no conflicts of interest.