Of all 23 clinical microbiology laboratories in Sweden, 19 provided data for this study; those laboratories collectively serve 8.78 million persons and correspond to ≈90% of the country’s population. The laboratories contributed data on cases within a 2- to 5-year period (depending on the time of introduction of MALDI-TOF mass spectrometry for Aerococcus spp. identification), yielding a total observation of 39.6 million person-years.

We identified a total of 591 episodes of aerococcal BSI in 581 persons. Among those episodes, 3 were considered relapses (<90-day interval) and excluded from our incidence calculations. The average incidence of aerococcal BSI was 1.48/100,000 person-years (95% CI 1.37–1.60/100,000 person-years). Most aerococcal BSIs occurred in older persons; the median age was 82 (interquartile range 74–87) years, and 78.6% (452 cases) of infected patients were men (Figure 2, panels A, B). Other than 2 cases that occurred in children <1 year of age, the age-stratified incidence was ≈0 in persons <60 years of age for both sexes, after which a substantial increase in the number of cases occurred, especially in men (maximum incidence 62.8/100,000 person-years [95% CI 46.9–78.6/100,000 person-years] in men 90–94 years of age) (Figure 2, panel C).

Most (n = 422 [71.7%]) aerococcal BSI episodes were caused by A. urinae, whereas 61 (10.4%) episodes were attributed to A. sanguinicola and 91 (15.5%) episodes were caused by other or unspecified aerococci; 14 (2.4%) patients had 2 concomitant species of aerococci in their blood cultures. Additional species of bacteria were identified in blood cultures from 201 (34.2%) cases, including 62 (10.5%) cultures containing coagulase-negative staphylococci, 35 (6.0%) with Escherichia coli, 26 (4.4%) with Actinotignum schaali, 17 (2.9%) with Enterococcus faecalis, and 15 (2.6%) with Staphylococcus aureus.

In this nationwide population-based study in Sweden, we identified associations between aerococcal BSI and urogenital diagnoses in general and urologic malignancy in particular, consistent with previous observations. We also identified associations with neurologic conditions, such as dementia, which has been infrequently reported. Our findings support a considerable predilection for elderly men to acquire aerococcal BSIs and also support the hypothesis that aerococci enter the bloodstream through the urogenital tract and renal mucosa (24). That hypothesis is strengthened by the observation that previous prescriptions of antimicrobial drug classes commonly administered for urinary tract infections were positively associated with aerococcal BSI.

We noted a higher incidence of aerococcal BSI in this study than in previous reports on A. urinae and A. sanguinicola, although the number of studies estimating rates has been low (5,6,12,13,29). As in previous studies, we observed sex differences in the tendency toward developing aerococcal BSIs; men had a greater incidence (5–7,12–14,18,29), but the mechanisms of this differential pattern of infection remain unknown. We found male urogenital tract disorders were more common in patients with aerococcal BSI than in controls. We speculate that the disproportionate male prevalence of aerococcal BSI can be explained in part by the preponderance of prostate disorders in elderly men.

The mechanisms of susceptibility to aerococcal BSI in male and female patients warrant further study. Because aerococcal BSIs are more frequent in elderly men, Aerococcus spp. have been proposed to enter the bloodstream through the prostate (3,5–7,12–14,18,29,30). Our findings indicate prostate cancer increases the risk for aerococcal BSI, whereas the association between infection and benign prostate hyperplasia was modest compared with conditions in other structures of the urogenital system. Our results indicate that clinical conditions in the kidneys and ureters were also associated with a higher risk for aerococcal BSI than those in the bladder, prostate, or urethra. A link between neurologic conditions in general, particularly dementia and paresis, and aerococcal BSI has been suggested in a small case series from Denmark, although only for A. sanguinicola (31). It is possible that a connection between neurologic diseases and aerococcal BSI is mediated by immobility issues in some patients (e.g., cerebrovascular insults leading to immobility) and, in the case of paresis, through catheterization of the urinary tract. Rheumatologic disease and aerococcal BSI had a negative association in multivariable analysis but had a positive crude association; however, the association was relatively modest and borderline significant and might have been a spurious effect of multiple comparisons.

In women, the levels of A. urinae and A. sanguinicola in vaginal microbiota increase during hormone replacement therapy (32). We did not, however, observe an increased risk for aerococcal BSI in women who had been prescribed estrogen medication, although the validity of that finding is limited by the low number of female study participants. One possible explanation is that urogenital carriage of aerococci is less likely to lead to bacteremia in women. That interpretation is consistent with smaller studies in Scotland and southern Sweden in which a larger proportion of men compared with women had Aerococcus spp. in blood but not urinary cultures (11,12,33,34).

The main strengths of this study are that it is population-based, examined the incidence of and risk factors for aerococcal BSI, and had a follow-up of ≈40 million person-years and a matched control group from the source population, enabling high-precision estimates and robust control for confounding. Access to healthcare in Sweden is not restricted by having private insurance, which might have otherwise introduced bias and possibly caused underestimation of aerococcal BSI incidence and limited the generalizability of our results to persons in lower socioeconomic strata or vulnerable subgroups (25). Second, in contrast to previous studies in which species were identified by using varied methods (often involving hospital-based or local inclusion of patients), the participating laboratories used MALDI-TOF mass spectrometry throughout the study period, and we included persons from multiple regions in a population-based manner (12–14). MALDI-TOF mass spectrometry has been shown to reliably identify Aerococcus spp. and has rectified the misclassification of A. sanguinicola as A. viridans and of aerococci in general as streptococci (35). Third, using population-based selection of matched controls enabled relevant comparisons with aerococcal BSIs among persons of the same age, sex distribution, and county of residence. That setup promoted a more efficient analysis of risk factors other than age and sex and reduced potential biases from variabilities in the structure or care registration between counties in Sweden. Moreover, considering the low incidence of aerococcal BSI, we consider it highly unlikely that a substantial proportion of the control population would have had an aerococcal BSI outside of the study uptake area or before the study period. Finally, the use of structured backward selection, according to variance likelihood ratio and variance inflation, enabled additional control for confounding and multicollinearity.

The first limitation of our study is that, despite our approach to control for confounding, residual confounding from factors that were not included in the registers cannot be ruled out. Despite efforts to limit potential sources of bias, claims on causality cannot be made by using a register-based study design. We also acknowledge that our findings might not be generalizable to children and young adults because of the limited number of those participants in this study. Second, information on over-the-counter drugs and drugs given during inpatient care was unavailable; however, except for topical estrogen, no drug reported in this study was available over the counter. Third, iatrogenic immune suppression caused by drugs (other than corticosteroids), especially chemotherapeutics, was not possible to define operationally by using the available registers because those drugs are often administered directly at the clinic and not filled by prescription in a pharmacy outside of the hospital. Thus, the relative effects of treatments for malignancy versus treatments for aerococcal BSI were not possible to evaluate in this study. Fourth, information on intermittent catheterization and the use of medical devices, including permanent urinary catheters, was lacking. Therefore, it was not possible to study the extent to which urinary catheterization might explain the link between neurologic conditions and aerococcal BSI. Finally, A. urinae has recently been proposed to comprise a cluster of related species (36). Because our study data were collected before that publication appeared, potential clustering was not possible to include in our analyses.

Previous validation studies of inpatient registry data in Sweden have shown that 85%–95% of diagnoses are correct, inasmuch as the ICD-10 code corresponds to the diagnosis entered into the medical records (27). However, diagnoses made during primary healthcare are unavailable in the patient register. That lack of accessibility could lead to selection bias, wherein each predisposing condition would only be represented by the stratum of disease severity that justified specialized care, potentially causing low power and inflated OR estimates because milder disease would be undetected. To minimize that risk, we combined registered ICD-10 codes with drug prescription data to capture the conditions of interest more robustly, which is especially critical for conditions such as diabetes mellitus, benign prostatic hyperplasia, and less severe infectious diseases that are generally managed during primary care in Sweden.

We defined aerococcal BSI according to the detection of aerococci in a blood culture. In ≈33% of cases, >1 bacteria species grew in the blood culture. We do not believe that polymicrobial growth reflects contamination of blood cultures by aerococci to any appreciable extent, because they are not known to routinely colonize the skin. It was not possible to determine the relative significance of aerococci when identified with other bacteria; however, we consider the presence of Aerococcus spp. in a patient’s blood to represent true bacteremia. A high degree of co-isolation with other established uropathogens, including E. coli, E. faecalis, and A. schaali, also suggests bacteremia originated from a urinary tract infection. Furthermore, prescriptions of antimicrobial drugs commonly taken for urinary tract infections were strongly associated with the development of aerococcal BSI. Whereas that association might reflect a chronic predilection for urinary tract infections, it is also possible that antimicrobial drug treatment promotes ecologic dysregulation, causing aerococcal overgrowth and invasive disease.

In conclusion, our findings substantially strengthen the proposed link between urologic conditions and aerococcal BSI. Neurologic conditions were also strongly associated with aerococcal BSI, although this association might be mediated by bacterial entry through the urothelium. Further study of the mechanisms that underly the associations between urologic and neurologic conditions and aerococcal bacteremia might be especially valuable in light of our findings. Awareness of Aerococcus spp. in patients, especially elderly men, will be needed to manage invasive infections.