Search control section

The MyHospitals website is moving on 31 March 2020

Don’t worry – from December 11 2019 you can find the latest information about your local area on the MyHospitals webpages on the AIHW website, along with many more reports and data on a range of health and welfare topics.

In some cases, the way you find information has changed. If you need help finding anything, please contact the AIHW.

Once the website has moved, you will be able to access old archived versions of the previously published data through Trove, the National Library of Australia’s web archive. Please note the interactive content will not work in the archived version.

Hospital Performance: Healthcare-associated Staphylococcus aureus bloodstream infections in 2012–13, In Focus

Left hand navigation section

Appendix 1: Review of peer groups

For the publication, Hospital Performance: Healthcare-associated Staphylococcus aureus bloodstream infections in 2011–12, a new classification system was developed for comparing hospitals based on size, percentage of immunosuppressed patients and exposure to treatments or procedures that increase the opportunity for infection.

Using information from literature review, clinical advisory groups and statistical analyses, the Performance Authority identified 10 hospital-level risk factors in two categories:

  1. Percentage of patient bed days of hospital care for immunosuppressed patients, including those admitted for:
    • Bone marrow transplant
    • Burns
    • Disorders of the immune system
    • HIV/AIDS
    • Oncology, including haematological malignancy
    • Transplant.
  2. Percentage of patient bed days for patients having treatments or procedures that increase the opportunity for infection, including those patients with:
    • Acute renal failure
    • Acute spinal injury
    • Surgery (including cardiac surgery)
    • Venous catheterisation.

For more information see Hospital Performance: Healthcare-associated Staphylococcus aureus bloodstream infections in 2011–12, Technical Supplement.

The Authority reviewed these hospital peer groups for the 2012–13 report. This appendix summarises the findings of this work.

Review of literature

A review of literature was conducted to identify any additional work on risk factors for healthcare-associated S. aureus bloodstream infection or peer grouping hospitals since the publication of Hospital Performance: Healthcare-associated Staphylococcus aureus bloodstream infections in 2011–12. A total of 1,324 unique articles were identified and were reduced to 36 articles following a review of abstracts. Full text review of these 36 articles resulted in 22 articles found to be directly relevant to risk factors for S. aureus bloodstream infections. No new published research was found on development of peer groups relevant to healthcare-associated infections.

Based on the review of literature, the Authority selected the following additional risk factors to be reviewed for inclusion to determine peer group allocation (see Table 1 for further details):

  • High Body Mass Index (obesity)
  • Alcohol and drug abuse
  • Diabetes
  • Intensive Care Unit (ICU) admissions.

Table 1: Additional risk factors reviewed for inclusion to determine peer groulp allocation for the report Hospital Performance: Healthcare-associated Staphylococcus aureus bloodstream infections in 2012–13

Risk factor identified in literature review Identification in data Rationale to inclusion to determine peer group allocation
High Body Mass Index ICD-10-AM codes for obesity: E66 E236 and E039 High BMI is a predictor of S. aureus bloodstream infection for patients across many different surgical procedures.
Alcohol and drug abuse (including chronic liver disease and cirrhosis) ICD-10-AM codes for alcohol and drug problems: F10-F19, T519, Z502, Z503, Z715 and Z721 The articles identified in the literature review suggest that poor nutrition and hygiene in this population predisposes patients to S. aureus bloodstream infection.
Diabetes ICD-10-AM codes for diabetes: E09-E14 Literature demonstrates that diabetes is a predictor of S. aureus bloodstream infection. As such, this risk factor was included.
Intensive Care Unit (ICU) admissions ICU hours reported- Possible through use of IHPA data Activity File data Identified as an important risk factor according to literature and the report clinical advisory committee.

Findings of data analysis

Investigation 1: Effect of adding additional risk factors on peer group allocation

For the 2011–12 report, the Performance Authority calculated the percentage of vulnerable patients at each hospital using the Admitted Patient Care (APC) National Minimum Data Set, 2011–12. Hospitals were sorted by 1) AIHW peer group classifications, then 2) descending order of the percentage of vulnerable patients. Using a histogram of hospitals in each AIHW peer group, four analysts independently identified the division point between more and fewer vulnerable patients for major, large, medium and small hospitals peer groups then jointly agreed this division point for each group. For the major hospitals group (sometimes referred to as ‘principal referral hospitals’), hospitals with more than 35% vulnerable patients were allocated to the ‘major hospital, more vulnerable patients group’ (36 hospitals) and those with equal to or less than 35% vulnerable patients were allocated to the ‘major hospital, fewer vulnerable patients group’ (40 hospitals).

Table 2 shows the effect of adding the risk factors included for analysis in Table 1 to the risk factors used in the 2011–12 report. The 36 hospitals allocated to the ‘major hospital, more vulnerable patients group’ in the Hospital Performance: Healthcare-associated Staphylococcus aureus bloodstream infections in 2011–12 report are shaded for each combination of risk factors reviewed. It is important to note that this table does not take into account that the division point between more and fewer vulnerable patients might change in each comparison but provides preliminary evidence for comparison. Obesity and alcohol/drug problems were analysed together because of their prominence in the literature review.

Special attention was given to the effect on the 10 hospitals named in the 2011–12 report with the highest rates of healthcare-associated S. aureus bloodstream infection in each of the major hospital peer groups. Seven of these hospitals had rates over 2.0 per 10,000 patient days (highlighted in orange in Table 2) and another three had lower rates but still the highest for their peer group (highlighted in blue). Of all these hospitals, only one had the potential to move from the more vulnerable patient group to the fewer vulnerable patient group. The addition of obesity and alcohol/drug problems to the existing risk factors moves this facility from 26th on the list to 36th, the highest place on the list to still be considered a more vulnerable patient hospital. Addition of obesity, alcohol/drug problems and diabetes result in this hospital being reclassified to the fewer vulnerable patient group. If this hospital did move to the fewer vulnerable patient group, it would still be named as a hospital in the highest 10% of results for its peer group.

This analysis also sought to better understand the data quality of the four potential risk factors, for example:

  • The diagnosis codes (ICD-10-AM) for obesity were recorded in only 0.5% of all episodes. Because the incidence in administrative data is so low, and further lowered by overlap with existing risk factors, obesity has a very small impact on the percentage of vulnerable patients
  • Alcohol/drug problems were considered for the 2011–12 report but, at the time, there was not clarity about how this could be a causative relationship. Since then, papers and consultation have agreed that poor nutrition and hygiene in this population makes them predisposed to healthcare-associated S. aureus bloodstream infection. Approximately 3% of patients have this code and few of these episodes are already included in the risk factors (13%)
  • Generally, the effect of adding the percentage of episodes for diabetes to obesity and alcohol/drug problems was marginal. There is a known relationship between obesity and type 2 diabetes which may account for part of the small effect; overlap with risk factors already being used accounts for most of this increase
  • The addition of ICU admissions was also marginal, most likely because the number of patients admitted to ICU is small compared to all admitted patients. This would have decreased further due to many ICU patients already covered by existing risk factors such as surgery, venous catheterisation or immunosuppression. However, the Clinical Advisory Committee for this work recommended that the inclusion of ICU admission to the risk factors would greatly increase clinical acceptance of these methods. Further investigation showed that this risk factor cannot be used because this data field is not currently available in a nationally consistent format; one state did not use the ICU admissions flag in 2011–12.

Decision: Risk factors will not change for Hospital Performance: Healthcare-associated Staphylococcus aureus bloodstream infections in 2012–13. Introducing additional risk factors resulted in minimal changes to peer group allocation.

Table 2: Major hospitals ranked by the percentage of vulnerable patients, showing the effect of adding additional risk factors (2011–12 data)

2011–12 report risk factors plus additional risk factors
Hospital name Peer group in 2011–12 report 2011–12 report risk factors + ICU + obesity & alcohol/ drug + obesity, alcohol/drug & diabetes + ICU, obesity, alcohol/ drug & diabetes
Number of additional risk factors - 0 1 2 3 4
Hospital 1 Major, more vulnerable* 1* 1* 1* 1* 1*
Hospital 2 Major, more vulnerable* 2* 4* 5* 3* 3*
Hospital 3 Major, more vulnerable* 3* 2* 4* 9* 4*
Hospital 4 Major, more vulnerable* 4* 6* 9* 6* 9*
Hospital 5 Major, more vulnerable* 5* 3* 6* 8* 5*
Hospital 6 Major, more vulnerable* 6* 5* 8* 7* 7*
Hospital 7 Major, more vulnerable* 7* 9* 3* 4* 6*
Hospital 8 Major, more vulnerable* 8* 7* 11* 11* 11*
Hospital 9 Major, more vulnerable* 9* 8* 2* 2* 2*
Hospital 10 Major, more vulnerable* 10* 12* 7* 5* 8*
Hospital 11 Major, more vulnerable* 11* 10* 10* 10* 10*
Hospital 12 Major, more vulnerable* 12* 22* 13* 14* 31*
Hospital 13 Major, more vulnerable* 13* 11* 21* 22* 16*
Hospital 14 Major, more vulnerable* 14* 19* 14* 17* 22*
Hospital 15 Major, more vulnerable* 15* 17* 12* 12* 12*
Hospital 16 Major, more vulnerable* 16* 13* 20* 19* 13*
Hospital 17 Major, more vulnerable* 17* 14* 17* 21* 18*
Hospital 18 Major, more vulnerable* 18* 21* 23* 26* 26*
Hospital 19 Major, more vulnerable* 19* 15* 19* 18* 15*
Hospital 20 Major, more vulnerable* 20* 20* 24* 20* 19*
Hospital 21 Major, more vulnerable* 21* 25* 25* 24* 24*
Hospital 22 Major, more vulnerable* 22* 24* 26* 28* 29*
Hospital 23 Major, more vulnerable* 23* 26* 30* 34* 33*
Hospital 24 Major, more vulnerable* 24* 27* 22* 23* 23*
Hospital 25 Major, more vulnerable* 25* 28* 27* 27* 30*
Hospital 26 Major, more vulnerable 26* 29* 36* 38 38
Hospital 27 Major, more vulnerable 27* 23* 28* 31* 28*
Hospital 28 Major, more vulnerable 28* 18* 41 43 27*
Hospital 29 Major, more vulnerable 29* 16* 32* 33* 17*
Hospital 30 Major, more vulnerable 30* 30* 16* 16* 21*
Hospital 31 Major, more vulnerable 31* 37 34* 29* 35*
Hospital 32 Major, more vulnerable 32* 33* 39 44 46
Hospital 33 Major, more vulnerable 33* 35* 33* 30* 34*
Hospital 34 Major, more vulnerable 34* 39 15* 13* 14*
Hospital 35 Major, more vulnerable 35* 32* 40 40 37
Hospital 36 Major, more vulnerable 36* 36* 29* 25* 25*
Hospital 37 Major, fewer vulnerable 37 38 47 56 56
Hospital 38 Major, fewer vulnerable 38 47 31* 32* 36*
Hospital 39 Major, fewer vulnerable 39 48 53 50 57
Hospital 40 Major, fewer vulnerable 40 45 43 46 47
Hospital 41 Major, fewer vulnerable 41 44 35* 42 45
Hospital 42 Major, fewer vulnerable 42 40 55 53 51
Hospital 43 Major, fewer vulnerable 43 31* 54 61 48
Hospital 44 Major, fewer vulnerable 44 46 18* 15* 20*
Hospital 45 Major, fewer vulnerable 45 41 49 35* 32*
Hospital 46 Major, fewer vulnerable 46 34* 50 52 43
Hospital 47 Major, fewer vulnerable 47 52 37 37 40
Hospital 48 Major, fewer vulnerable 48 53 45 47 55
Hospital 49 Major, fewer vulnerable 49 43 48 48 44
Hospital 50 Major, fewer vulnerable 50 42 51 59 54
Hospital 51 Major, fewer vulnerable 51 50 52 55 53
Hospital 52 Major, fewer vulnerable 52 49 56 51 50
Hospital 53 Major, fewer vulnerable 53 51 42 45 42
Hospital 54 Major, fewer vulnerable 54 57 38 39 49
Hospital 55 Major, fewer vulnerable 55 55 44 41 41
Hospital 56 Major, fewer vulnerable 56 56 61 62 65
Hospital 57 Major, fewer vulnerable 57 60 60 57 61
Hospital 58 Major, fewer vulnerable 58 61 58 58 63
Hospital 59 Major, fewer vulnerable 59 54 66 66 62
Hospital 60 Major, fewer vulnerable 60 63 62 64 67
Hospital 61 Major, fewer vulnerable 61 58 64 60 59
Hospital 62 Major, fewer vulnerable 62 59 59 65 60
Hospital 63 Major, fewer vulnerable 63 65 63 36* 39
Hospital 64 Major, fewer vulnerable 64 69 65 67 72
Hospital 65 Major, fewer vulnerable 65 64 70 71 70
Hospital 66 Major, fewer vulnerable 66 66 46 49 52
Hospital 67 Major, fewer vulnerable 67 62 69 69 66
Hospital 68 Major, fewer vulnerable 68 67 67 68 68
Hospital 69 Major, fewer vulnerable 69 70 57 54 58
Hospital 70 Major, fewer vulnerable 70 68 71 73 69
Hospital 71 Major, fewer vulnerable 71 72 72 70 71
Hospital 72 Major, fewer vulnerable 72 71 68 63 64
Hospital 73 Major, fewer vulnerable 73 73 73 72 73
Hospital 74 Major, fewer vulnerable 74 74 74 74 74
Hospital 75 Major, fewer vulnerable 75 76 75 75 75
Hospital 76 Major, fewer vulnerable 76 75 76 76 76

Hospitals named in the 2011–12 report with healthcare-associated S. aureus bloodstream infection rates in the highest 10% of results for its peer group (rate >2.0 per 10,000 patient days under surveillance).

* The 36 hospitals with the highest percentages of vulnerable patients are shaded for each combination of risk factors.

Hospitals named in the 2011–12 report with healthcare-associated S. aureus bloodstream infection rates in the highest 10% of results for its peer group (rate <2.0 per 10,000 patient days under surveillance).

Investigation 2: Peer group stability over time

The Performance Authority investigated stability of the current peer group allocation over time and determined the percentage of vulnerable patients at each hospital across five years of data. Over the five years of data, very few hospitals moved across the 35% divider identified for the 2011–12 report (Figure 1). Large changes in the percentage of vulnerable patients between years were observed in a small number of major hospitals, driven primarily by the percentage of admissions receiving surgery.

Decision: Based on the investigation of allocation to peer groups over time, the Authority is satisfied that peer groups are stable and these groups are appropriate to use again for the 2012–13 data.

Figure 1: Major hospitals changing between more and fewer vulnerable patients over five years, 2007–08 to 2011–12

Figure showing major hospitals changing between more and fewer vulnerable patients over five years from 2007-08 to 2011-12

Investigation 3: Weighting of risk factors

The Performance Authority has received feedback that it should weight risk factors to account for differences in the degree to which each factor contributes to risk of infection. While this approach has merit, the calculation of weights at the patient- or hospital-level requires an information system that has complete and accurate patient-level data. This type of data is not currently available in a nationally consistent format.

Decision: Weighting of risk factors have not be applied to assign hospitals into peer groups on the basis of risk.