surveillance of surgical site infections in nhs hospitals in england 2013/14 · 2014. 12. 12. ·...

Surveillance of Surgical Site Infections in NHS Hospitals in England 2013/14

Surveillance of Surgical Infections in NHS hospitals in England 2013/14

2

About Public Health England

Public Health England exists to protect and improve the nation's health and wellbeing,

and reduce health inequalities. It does this through world-class science, knowledge and

intelligence, advocacy, partnerships and the delivery of specialist public health services.

PHE is an operationally autonomous executive agency of the Department of Health.

Public Health England

133-155 Waterloo Road

Wellington House

London SE1 8UG

Tel: 020 7654 8000

http://www.gov.uk/phe

@PHE_uk

Suggested citation

Public Health England. Surveillance of surgical site infections in NHS hospitals in England,

2013/14. London: Public Health England, December 2014. Available from: www.gov.uk/phe

Main author: S Elgohari with the assistance of M Mihalkova, C Wloch, A Saei, R Harris, A

Charlett and T Lamagni. We are grateful for additional comments from Prof Alan Johnson, head

of the Department of Healthcare Associated Infections and Antimicrobial Resistance. We

extend our thanks to: P Harrington for additional comments and and the administrative staff at

PHE’s Surgical Site Infection Surveillance Service, technical staff within the PHE Software

Development Unit and the considerable contribution made by NHS Trusts in England who have

devoted time and effort in collecting these data. For queries relating to this document, please

contact the Department of Healthcare Associated Infections and Antimicrobial Resistance at:

[email protected]

© Crown Copyright 2014

Published December 2014

PHE gateway number: 2014557

This document is available in other formats on request. Please call 020 8327 7018

or email [email protected]



mailto:[email protected]

mailto:[email protected]


3

Contents

Key points 4

Section 1 Introduction and surveillance methods 5

1.1 Introduction 5 1.2 Data collection and feedback 5 1.3 Definitions 6

1.4 Participation in international surveillance 6 1.5 Changes to surgical categories under surveillance 7 1.6 Analyses presented in this report 7

Section 2 Overview 8

2.1 Hospital participation and surgical volumes 8 2.2 Data quality 9

2.3 Patient and operation-related characteristics 12

Section 3 Rates of surgical site infection 14

3.1 Inpatient and readmission rate by surgical category 14 3.2 Risk factors for SSI 16

Section 4 Trends in the rate of SSI 18

4.1 Orthopaedic categories 18

4.2 Hospital-level trends – orthopaedic categories 19 4.3 Non-orthopaedic categories 20

Section 5 Characteristics of surgical site infections 22

5.1 Type of SSI 22 5.2 Causative micro-organisms 24

Section 6 Variation in rates of SSI between NHS hospitals 26

6.1 Box and whisker plots 26

6.2 Funnel plots 27

Section 7 Conclusions 29

7.1 Trends in the rate of SSI 29 7.2 Variation in the SSI rate between participating hospitals 29

7.3 Optional post-discharge surveillance 30 7.4 Microbial aetiology 30 7.5 Guidelines for the prevention of SSI 30 7.6 Future directions 31

Section 9 Glossary 32

Section 10 References 33


4

Key points

Data on 594,855 procedures and 8,458 inpatient and readmission surgical site infections (SSIs)

from 17 surgical categories were collected by 234 NHS hospitals and NHS independent sector

Treatment Centres between April 2009 and March 2014.

In 2013/14, a total of 198 NHS hospitals represented by 148 NHS Trusts along with eight NHS

Treatment Centres participated in the national SSI surveillance.

For the mandatory orthopaedic surveillance, the unit of analysis is at NHS Trust-level. In

2013/14, 143 NHS Trusts and eight NHS Treatment centres participated in the surveillance,

contributing data on 102,570 procedures. One eligible NHS Trust did not participate this year.

The proportion of hospitals undertaking continuous surveillance increased in 2013/14 for hip and

knee prosthesis (62% each). This was the highest proportion of all surgical categories.

Data completion for key risk factors (age, patient sex, duration of operation, date of admission

and procedure code) remained very high (100%) in 2013/14. BMI completion was much lower at

47% although this was an increase over the previous year.

Between 2008/9 and 2013/14, a significant decrease in the inpatient/readmission SSI incidence

was found for repair of neck of femur and reduction of long bone fracture, reaching 1% in each

category in 2013/14. No overall trends for hip and knee prosthesis were found and the incidence

remained low (<1%) in these categories.

Analysis of hospital-level trends for the orthopaedic categories showed that the majority of

centres exhibited stable or decreasing trends over time.

Among non-orthopaedic categories, a significant increasing trend in SSI was found for patients

undergoing spinal surgery with an SSI rate of 1.3% in 2013/14. A significantly decreasing trend

in was found for patients undergoing bile duct/liver/pancreatic and gastric surgery.

In 2013/14, 10 NHS Trusts were identified as high outliers for the mandatory orthopaedic

surveillance with an incidence of SSI higher than expected nationally. An additional seven NHS

Trusts were identified as low outliers. All 17 NHS Trusts have been contacted and encouraged to

review their practices.

S. aureus as a reported cause of inpatient SSIs accounted for 16% of cases in 2013/14. This

followed a decreasing trend observed from 2006/07 due to concurrent decreases in methicillin-

resistant S. aureus (MRSA). MRSA decreased from 26% in 2004/5 to 4% by 2013/14.

Enterobacteriaceae increased from 2008/9 and accounted for 26% of cases by 2013/14.

In 2013/14, S. aureus was the predominant organism in orthopaedic and spinal surgery

accounting for ≥40% of cases. Coagulase-negative staphylococci and Enterobacteriaceae were

predominant in coronary artery bypass graft and large bowel surgery respectively.


5

Section 1 Introduction and surveillance

methods

1.1 Introduction

This report is a summary of data on surgical site infections (SSIs) collected by NHS hospitals and

independent sector NHS Treatment Centres in England participating in one of 17 surgical

categories of surveillance between April 2004 and March 2014. The results include orthopaedic

data submitted by hospitals as part of the mandatory requirement for the surveillance of healthcare

associated infections (HCAI) introduced by the Department of Health in April 2004[1]. The national

mandate is directed at NHS Trusts and requires all NHS Trusts undertaking orthopaedic surgical

procedures to carry out a minimum of three months’ surveillance in each financial year in at least

one of four categories (hip prosthesis, knee prosthesis, repair of neck of femur or reduction of long

bone fracture). The orthopaedic data from participating hospitals are aggregated to Trust level for

public reporting purposes. Trusts with very small volumes are exempt from the mandatory

surveillance but are expected to undertake surveillance in a category of their choice that reflects

their surgical activity. This report includes data on 13 other categories submitted voluntarily.

1.2 Data collection and feedback

Data are collected at hospital level. All hospitals participating in Public Health England’s (PHE)

national SSI surveillance are required to follow the surveillance protocol outlining the case

definitions and follow-up methods[2]. Each hospital collects data prospectively on all eligible

patients in a surgical category over a 3 month period. The post-operative follow-up period is set at

30 days for non-implant procedures and one year for implant procedures. A set of demographic

and operation-related data are collected for each eligible procedure and submitted via a secure

web-based data entry screen.

Since July 2008, hospitals are required to have systems in place to identify patients who are

included in the surveillance and readmitted to hospital with an SSI. SSIs identified on readmission

at a hospital other than the one where the original operation took place are assigned to the original

hospital. Two other post-discharge surveillance methods were introduced in 2008 but remain

optional: a) systematic review of patients attending outpatient clinics or at home by clinical staff

trained to apply the case definitions and b) wound healing questionnaires completed by patients 30


6

days after the operation[2]. Data derived from these optional methods are not currently included in

the national benchmarks or used for outlier assessment.

After each completed quarter, participating hospitals can download automated confidential reports

accessed securely from the SSI web application to disseminate within their Trust. These reports

detail the hospitals’ crude SSI incidence and the corresponding national benchmark for the relevant

surgical category. A range of reports are available from the web application providing stratified

rates by key risk factors. Data quality reports are also available based on data completion for key

data items collected.

PHE analyses submitted data at quarterly intervals to identify hospitals whose SSI incidence falls

above the 90th or below the 10th percentiles nationally for a given surgical category. PHE alerts

these hospitals of their outlier status and encourages them to investigate possible reasons.

1.3 Definitions

SSIs are defined according to a standard set of clinical criteria for infections that affect the

superficial tissues (skin and subcutaneous layer) of the incision and those that affect the deeper

tissues (deep incisional or organ-space) defined by the Centers for Disease Control and

Prevention (CDC), USA [3]. One of the key risk factors collected by hospitals is the ASA score. This

is a classification system developed by the American Society of Anesthesiologists to classify a

patient’s pre-operative physical status on a scale from one to five with higher scores indicating

severe systemic disease. This along with the level of wound contamination and duration of surgery

>75th percentile constitute the three components of the National Healthcare Safety Network

(NHSN) Risk Index[4].

1.4 Participation in international surveillance

PHE shares anonymised SSI surveillance data with HAI-Net at the European Centre for Disease

Control and prevention (ECDC) on an annual basis using ECDC’s standardised protocol (also

based on CDC definitions)[5]. The data cannot be traced back to individual patients, surgeons or

named hospitals. ECDC collates SSI data from other European member states and publishes

comparative analyses including trends. These analyses provide an opportunity to examine variation

in the SSI incidence between European countries and to improve understanding of how these

infections may be prevented.


7

1.5 Changes to surgical categories under surveillance

Changes were made to the following categories during the period covered by this report:

Discontinuation of hip hemiarthroplasty and introduction of repair of neck femur in July 2008

(comprising hip hemiarthroplasty and reduction of fractured neck of femur using open

fixation).

Discontinuation of open reduction of long bone fracture and introduction of reduction of long

bone fracture in July 2008 (comprising open or closed reduction procedures).

Spinal surgery was introduced in July 2008.

Breast, cranial and cardiac (non-coronary artery bypass graft [CABG]) surgery were

introduced in April 2010.

1.6 Analyses presented in this report

Data on surgical procedures carried out between April 2004 and March 2014 collected by NHS

hospitals and independent sector NHS Treatment Centres were extracted on 23rd September 2014

for this report. To evaluate trends, inpatient and readmission SSI data from 2008 were analysed.

For benchmarking, cumulative five-year data were used (April 2009 to March 2014).

The cumulative SSI incidence (%) presented in this report was based on SSIs detected during the

hospital stay combined with SSIs identified on readmission following the initial operation. Where

appropriate, inpatient SSIs were analysed separately for meaningful interpretation. To take into

account the variation in the length of follow-up, the incidence density was calculated using the

number of cases in the numerator and the total number of days of patient follow-up (from inpatient

surveillance) in the denominator giving the number of SSIs per 1,000 patient days of follow-up.

Funnel plots were constructed for each orthopaedic category using the inpatient/readmission

cumulative SSI incidence per NHS Trust during 2013/14. These plots identify any Trust whose

incidence falls within the expected variation and any outliers (whose incidence is significantly

higher or lower than expected nationally). Independent sector NHS treatment centres are also

included in these plots. An additional supplement to this report presents the cumulative SSI

estimates in 2013/14 by NHS Trust for the mandatory orthopaedic surveillance:

https://www.gov.uk/government/publications/surgical-site-infections-ssi-surveillance-nhs-hospitals-

in-england

https://www.gov.uk/government/publications/surgical-site-infections-ssi-surveillance-nhs-hospitals-in-england



8

Section 2 Overview

2.1 Hospital participation and surgical volumes

Figure 1 shows trends in hospital participation and surgical volume in the national SSI surveillance

scheme split by the mandatory orthopaedic and the voluntary non-orthopaedic modules. Overall, a total

of 198 NHS hospitals represented by 148 NHS Trusts and eight NHS Treatment Centres participated in

2013/14.

For the orthopaedic module, the number of participating hospitals and surgical volume increased by

95% and 342% respectively between April 2004 and March 2014. Between 2012/13 and 2013/4, the

number of participating hospitals remained stable (n=192 NHS Hospitals/NHS Treatment Centres in

both years) although surgical volume increased by 6% (from 96,470 to 102,570). Overall, 184 hospitals

representing 143 NHS Trusts along with eight independent NHS Treatment Centres participated in the

orthopaedic surveillance in 2013/14 with an average of 25,600 records submitted per quarter across all

sites. One eligible NHS Trust did not participate in the orthopaedic surveillance in 2013/14. Of 143 NHS

Trusts that did so, 88% (n=126) fulfilled more than the minimum required for the mandatory surveillance

(one surveillance category for a minimum of three months).

For the voluntary non-orthopaedic categories, the number of participating hospitals and surgical volume

increased by 76% and 408% respectively between April 2004 and March 2014. In 2013/14 a mean of

7,714 non-orthopaedic procedures were submitted per quarter. Hospital participation in the voluntary

scheme increased slightly from 2012/13 to 2013/14 (from 69 to 73 NHS hospitals respectively). Surgical

volume increased by 13% between these two years (from 27,287 to 30,857).

After excluding seven of 17 surgical categories that had a small number of participating hospitals in

2013/14 (<10), the remaining categories showed considerable variation in the proportion of hospitals

undertaking continuous surveillance in 2013/14. This proportion was highest in hip and knee prosthesis

(62% each), a further increase from 2012/13. Coronary artery bypass graft (CABG) had the next highest

proportion of hospitals doing continuous surveillance (53%) followed by spinal surgery and repair of

neck of femur (47% each) and reduction of long bone fracture (42%).


9

Figure 1: Trends in the number of procedures and participating hospitals by surveillance

quarter, NHS hospitals in England

2.2 Data quality

Data completion for key SSI risk factors is essential for risk stratification purposes, enabling hospitals to

investigate factors underpinning unusual deviations in their overall estimate from the SSI national

benchmark. Table 1 shows the proportion of records submitted to the national database with completed

fields for key data items by surgical category during 2013/14.

Data completeness varied by data item and category but was above 99% for all risk factors used in the

NHSN risk index except ASA score at 95% overall (61% to 98% depending on category). Data

completeness was considerably lower for height and weight used to calculate body mass index (BMI) at

47% overall although this increased from 32% observed in 2012/13. BMI data completion was highest in


10

CABG at 83%, cardiac non-CABG at 75% (rounding directly from the original value), and knee

prosthesis surgery (57%).

Table 1: Proportion of submitted surveillance records with complete data for key data

items, NHS hospitals in England, 2013/14

Total No.

operations Age

ASA score

Duration of operation

Wound class

BMI* Patient

sex Date of

admission OPCS code

Abdominal hysterectomy 708 100.0 85.6 100.0 99.7 55.2 100.0 100.0 100.0

Bile duct, liver and pancreatic surgery 654 100.0 96.5 100.0 99.7 40.3 100.0 100.0 100.0

Breast surgery 2,553 100.0 91.6 100.0 100.0 18.0 100.0 100.0 100.0

Cardiac (non-CABG) 2,968 100.0 61.2 100.0 99.9 75.5 100.0 100.0 100.0

Cholecystectomy 233 100.0 96.6 100.0 100.0 28.8 100.0 100.0 100.0

Coronary artery bypass graft 6,694 100.0 78.8 100.0 100.0 82.6 100.0 100.0 100.0

Cranial surgery 2,131 100.0 92.1 99.7 99.6 64.2 100.0 100.0 100.0

Gastric 240 100.0 97.1 100.0 100.0 0.0 100.0 100.0 100.0

Hip prosthesis 40,202 100.0 96.8 99.9 99.9 50.2 100.0 100.0 100.0

Knee prosthesis 40,565 100.0 97.4 99.9 100.0 56.7 100.0 100.0 100.0

Large bowel surgery 3,442 100.0 95.2 100.0 100.0 47.3 99.9 100.0 100.0

Limb amputation 324 100.0 80.9 100.0 95.7 22.7 100.0 100.0 100.0

Reduction long bone fracture 2,959 100.0 97.7 99.9 99.9 17.0 100.0 100.0 100.0

Repair of neck of femur 18,844 100.0 94.7 99.9 100.0 16.4 100.0 100.0 100.0

Small bowel surgery 1,009 100.0 94.8 100.0 99.5 42.3 100.0 100.0 99.9

Spinal surgery 8,303 100.0 96.0 99.8 100.0 31.1 100.0 100.0 100.0

Vascular surgery 1,598 100.0 96.7 99.9 99.9 32.2 100.0 100.0 100.0

Total 133,427 100.0 94.6 100.0 99.9 46.8 100.0 100.0 100.0

*based on patients ≥16 years

A choice of optional post-discharge surveillance methods were available to hospitals in addition to the

inpatient and readmission surveillance required in the protocol as described in Section 1.

The proportion of procedures where patient post-discharge questionnaires (PDQs) were given to

patients to identify potential problems with wound healing within 30 days of surgery increased from 32%

in 2009/10 to 45% in 2013/14 (Figure 2). The patient response rate based on PDQs returned increased

from 63% of questionnaires given in 2009/10 to 73% in 2013/14. Surveillance activity based on

outpatient follow-up was lower, being used in 20% of procedures with little change around this level over

the last five years.


11

Figure 2: Trends in the proportion of procedures with post-discharge questionnaire

activity, NHS hospitals in England, April 2009 to March 2014*

*excludes breast, cranial and cardiac surgery; † PDQ given as a percentage of operations; ‡ PDQ completed as a

percentage of PDQ given


12

2.3 Patient and operation-related characteristics

Key patient and operation-related characteristics derived from surveillance records for patients

undergoing surgery in 2013/14 are shown in Table 2 (all surgical categories) and in Table 3

(orthopaedic related only).

The median age (Table 2) varied by surgical category being lowest in abdominal hysterectomy (50

years) and highest in repair of neck of femur (84 years). The duration of operation was highest in bile

duct/liver and pancreatic surgery, cardiac, CABG and gastric surgery, with all having median durations

in excess of 200 minutes. The categories of surgery with the highest proportion of patients with an ASA

score ≥3 were in cardiac (non-CABG) (97%), CABG (96%) and limb amputation surgery (78%). Small

bowel, large bowel and limb amputation surgery had the highest proportion of records with a wound

classified as contaminated or dirty (41%, 17% and 11%, respectively).

Knee prosthesis, cholecystectomy, abdominal hysterectomy, limb amputation and hip prosthesis had

the highest proportions of patients classed as obese i.e. BMI ≥ 30kg/m2 (Table 2). The categories with

the lowest proportion of obese patients were repair of neck of femur followed by reduction of long bone

fracture. However, owing to low BMI data completion in some surgical categories, these results should

be interpreted with caution. The proportion of operations performed on an emergency basis (defined as

procedures that are immediate, unplanned and life-saving or those that are performed immediately after

resuscitation) were most common for cranial and large bowel surgery (8% and 7%, respectively).

The proportion of patients receiving surgical antimicrobial prophylaxis was ≥ 98% in three categories of

orthopaedic surgery (hip prosthesis, knee prosthesis and reduction of long bone fracture), bile duct/liver

and pancreatic, cardiac (non-CABG), CABG, cranial, small bowel and spinal surgery. The high

proportions reflect the current recommendations aimed at defined surgical groups[6; 7].

The proportion of patients where discontinuation of inpatient surveillance was due to death was highest

in repair of neck of femur (6.3%) followed by cranial and vascular surgery (3.6% both). These should not

be interpreted as case-fatality rates as the data only capture deaths within the inpatient stay.

Table 3 shows primary indication for surgery in patients undergoing hip prosthesis, knee prosthesis and

repair of neck of femur. In hip and knee prosthesis, osteoarthritis accounted for majority of primary

indications (81% and 90%, respectively). The second most common indication for these procedures was

revision (11% and 6%, respectively). For repair of neck of femur, the most common primary indication

was trauma (99%).


13

Table 2: Patient and operation-related characteristics by surgical category, NHS hospitals in England, 2013/14

Table 3: Primary indication for surgery, orthopaedic procedures, NHS hospitals in England, 2013/14

Abdominal hysterectomy 50 90 3 8.3 0.3 43.2 N/A 0.6 N/A 97.7 0.0 87.0 0.0

Bile duct, liver or pancreatic 66 260 8 33.3 3.8 28.1 52.3 0.5 1.1 99.1 2.5 87.0 1.7

Breast surgery 59 75 1 9.7 0.1 31.9 1.9 0.0 2.0 83.0 11.2 86.9 <0.1

Cardiac surgery 68 235 8 96.8 0.5 27.6 62.8 2.5 4.1 99.8 91.4 96.3 2.6

Cholecystectomy 62 170 6 23.6 0.9 47.8 46.4 0.0 N/A 91.1 0.4 92.7 0.0

Coronary artery bypass graft 69 230 7 95.6 <0.1 32.8 80.3 1.3 0.2 99.4 75.1 87.5 1.5

Cranial surgery 58 100 5 42.7 0.9 26.5 54.9 7.9 0.2 99.1 26.5 62.7 3.6

Gastric surgery 61 217 8 41.2 2.5 - 57.9 0.0 2.9 96.7 3.3 96.6 1.3

Hip replacement 70 83 4 22.2 0.1 36.6 39.9 0.1 9.2 98.5 100.0 82.3 0.3

Knee replacement 70 78 4 21.2 <0.1 54.6 42.3 <0.1 5.2 99.0 100.0 82.4 0.2

Large bowel surgery 68 163 8 34.7 16.7 24.8 50.5 7.3 N/A 97.6 1.8 84.1 2.6

Limb amputation 68 45 8 78.2 11.0 41.1 73.5 5.6 N/A 77.5 1.5 76.0 3.4

Reduction of long bone fracture 62 83 4 31.4 2.2 18.8 42.7 0.5 N/A 98.0 98.3 51.0 1.6

Repair of neck of femur 84 65 12 69.0 <0.1 9.0 29.0 1.6 0.2 96.3 100.0 52.2 6.3

Small bowel surgery 61 117 8 42.2 41.1 19.9 52.1 4.7 0.4 99.7 1.1 77.4 3.1

Spinal surgery 54 117 3 20.8 0.2 33.6 46.7 2.8 45.4 99.0 47.0 87.6 0.2

Vascular surgery 72 130 5 74.5 0.5 27.2 70.1 4.7 0.1 95.1 57.6 95.1 3.6

* patients ≥16 years old

Discontinuation of

inpatient

surveillance due to

death (%)

Antibiotic

prophylaxis

(%) Implant (%)

Lead surgeon

Consultant

(%)Male

(%)

Emergency

(%)

Revision

(%)

BMI ≥ 30

(%)*

Median length

hospital stay

(days)

Median age

in years (IQR)

Median

duration

operation in

minutes

(IQR)

ASA ≥ 3

(%)

Contaminated/

dirty incision

(%)

No. operations

Hip replacement 39,268 31,794 (81.0) 268 (0.7) 313 (0.8) 345 (0.9) 333 (0.9) 3,227 (8.2) 539 (1.4) 4,444 (11.4) 1,568 (4.0) 881 (2.2)

Knee replacement 39,551 35,685 (90.2) 402 (1.0) 19 (0.1) 61 (0.2) 340 (0.9) 1,733 (4.4) 299 (0.8) 2,433 (6.3) 170 (0.4) 842 (2.1)

Repair of neck of femur 18,759 97 (0.5) 3 (<0.1) 1 (<0.1) 18 (0.1) 4 (<0.1) 17 (0.1) 11 (0.1) 50 (0.3) 18,512 (98.7) 96 (0.5)

Trauma

No. (%)

Other

No. (%)Unknown

No. (%)

All

No. (%)

Fracture

No. (%)

Infection

No. (%)

Other

No. (%)

Revision

Osteoarthritis

No. (%)

Inflammatory joint

disease No. (%)

Avascular necrosis

No. (%)


14

Section 3 Rates of surgical site infection

3.1 Inpatient and readmission rate by surgical category

Table 4 shows the volume of surgical procedures, the number of SSI cases and the cumulative SSI

incidence (%) by surgical category for surgery undertaken between April 2009 and March 2014. Over

this period a total of 594,855 procedures on 17 surgical categories were submitted by 224 participating

hospitals representing 154 NHS Trusts. An additional 10 NHS Treatment Centres (participated in the

surveillance (this excluded one NHS Treatment Centre that became part of an NHS Trust in 2012/13

and was counted as part of the NHS group).

The cumulative incidence (%) of SSI varied by surgical category depending largely on the inherent

likelihood of microbial contamination at the operative site. The highest incidence (risk) was observed in

gastro-intestinal procedures, with large bowel surgery having the highest risk at 10.2% over this five

year period. The lowest risk was observed in hip and knee prosthesis surgery (0.7% and 0.6%

respectively).

SSIs detected through readmission accounted for a relatively high proportion of inpatient and

readmission cases in surgical categories with a relatively short length of hospital stay such as breast

surgery (86%), knee prosthesis (68%), abdominal hysterectomy (62%) and hip prosthesis (58%). In

categories with relatively longer length of hospital stay (gastro-intestinal, repair of neck of femur, CABG,

cardiac (non-CABG) and limb amputation surgeries), readmission SSIs accounted for ≤30% of all these

SSIs detected.

To take into account the variation in the length of patient follow-up during their hospital stay, the

incidence density was computed. This metric adjusts for SSIs being more likely to be detected in

patients with a longer length of hospital stay. After taking into account the length of inpatient follow-up,

the variation between surgical categories is less pronounced when compared to the inpatient-based

cumulative incidence. However, large bowel surgery remains as having the highest SSI incidence and

knee surgery as having the lowest incidence (Table 4).


15

Table 4: SSI incidence by surgical category, NHS hospitals in England, April 2009 – March 2014

No.

operations

No.

hospitals

No.

Inpatient

No. inpatient &

readmission

Inpatient &

readmisson % 95% CLs

Median time

to infection

(days)**†

Incidence

density/1,000

inpatient days‡ 95% CLs

Abdominal hysterectomy 4,512 28 25 65 1.4% 1.1 - 1.8 9 1.4 0.9 - 2.0

Bile duct, liver and pancreatic surgery 2,544 6 130 152 6.0% 5.1 - 7.0 8 4.5 3.8 - 5.4

Breast* 7,634 26 10 71 0.9% 0.7 - 1.2 12 0.7 0.3 - 1.3

Cholecystectomy 967 7 38 45 4.7% 3.4 - 6.2 7 4.7 3.3 - 6.4

Coronary artery bypass grat 30,838 24 1,027 1,400 4.5% 4.3 - 4.8 12 3.5 3.3 - 3.7

Cardiac (non-CABG)* 9,465 14 81 116 1.2% 1.0 - 1.5 12 0.7 0.6 - 0.9

Cranial* 4,963 6 31 72 1.5% 1.1 - 1.8 17 0.7 0.5 - 1.0

Gastric 1,281 8 23 25 2.0% 1.3 - 2.9 8 2.0 1.3 - 3.0

Hip prosthesis 180,852 198 525 1,246 0.7% 0.7 - 0.7 15 0.5 0.4 - 0.5

Knee prosthesis 188,974 190 367 1,145 0.6% 0.6 - 0.6 16 0.3 0.3 - 0.4

Large bowel 17,924 50 1,556 1,824 10.2% 9.7 - 10.6 8 8.1 7.7 - 8.5

Limb amputation 2,134 19 52 68 3.2% 2.5 - 4.0 12 1.7 1.3 - 2.3

Reduction of long bone fracture 15,021 34 94 176 1.2% 1.0 - 1.4 16 0.7 0.6 - 0.9

Repair of neck of femur 83,339 126 913 1,196 1.4% 1.4 - 1.5 14 0.7 0.6 - 0.7

Small bowel 4,105 19 247 275 6.7% 6.0 - 7.5 8 5.1 4.4 - 5.7

Spinal 33,053 30 177 378 1.1% 1.0 - 1.3 14 1.0 0.9 - 1.2

Vascular 7,249 29 139 204 2.8% 2.4 - 3.2 11 2.1 1.7 - 2.4

Total 594,855 234 5,435 8,458

* introduced in April 2010; ** based on inpatient or readmission SSsI; † two SSIs omitted from analysis due to missing onset date;

‡eight records (of which one had a SSI) omitted due to missing date of operation or date of hospital discharge

Surgical Site Infections


16

3.2 Risk factors for SSI

Table 5 shows the cumulative incidence of SSI according to given risk factors for each surgical

category. For some types of surgery, the risk of SSI increased in younger patients (<45 years) as in

cardiac (non-CABG surgery). For bile duct/liver/pancreatic, cholecystectomy, coronary artery bypass

graft, gastric, reduction of long bone fracture, and spinal surgery the reverse was observed with the risk

being higher in older patients although in knee prosthesis the risk was similar in all age categories.

Patients with ASA score of ≥3 had an increased SSI risk except in breast, cardiac (non-CABG), cranial,

gastric, and vascular surgery where the reverse pattern was observed. Duration of operation greater

than the NHSN 75th percentile for that category had an increased risk of SSI compared to procedures

below this threshold across all categories with the exception of breast, spinal surgery and limb

amputation where the opposite was seen.

The risk of SSI increased among patients who were obese (BMI ≥30kg/m2) in the majority of surgical

categories with the exception of cholecystectomy where the reverse was observed. Having an

emergency procedure increased the risk of SSI in cardiac (non-CABG), CABG, large bowel, vascular

surgery and limb amputation. It should be noted that the underlying sample for these categories was

very small and as such, the risk estimate for this group should be interpreted with caution.

The risk of SSI increased if wound class was contaminated/dirty in bile duct/liver/pancreatic, large

bowel, limb amputation, reduction of long bone fracture, small bowel and vascular surgery. The reverse

pattern was observed in the remaining surgical categories. However the sample for contaminated/dirty

wound class across all categories was very small (with the exception of large bowel surgery) and the

risk estimate for this wound class group therefore imprecise.

For hip prosthesis, the risk of SSI was highest for surgery undertaken for revision due to a previous

infection followed by revision due to a fracture (3.9% and 2.6% respectively). In knee prosthesis the risk

of SSI was highest for revision due to a fracture followed by revision due to a previous infection (3.3%

and 2.1% respectively). For repair of neck of femur the primary indication carrying the highest risk was

revision due to a fracture followed by ‘revision (other)’ (11.1% and 5.9%, respectively) but these

estimates were based on very small sample sizes. In this surgical category, revision (aggregated data

for all reasons) carried the highest SSI risk at 4.3%.


17

Table 5: The incidence of SSI stratified by key risk factors, by surgical category, NHS hospitals, England 2013/14

Risk factor No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI No. ops

%

SSI

Age

<45 197 0.5 70 4.3 447 1.1 523 1.7 42 0.0 123 2.4 591 1.7 37 0.0 1,443 0.8 422 0.5 405 10.9 29 3.5 886 0.7 220 1.4 253 5.1 2,834 1.0 51 0.0

≥45-64 387 1.8 241 3.3 1,198 0.6 727 1.1 99 4.0 2,263 3.4 786 1.7 101 0.0 11,189 0.5 12,226 0.5 994 8.7 113 6.2 706 0.7 1,139 1.3 339 8.6 2,968 1.3 333 3.0

≥65 124 0.8 343 7 908 0.8 1,718 0.9 92 4.4 4,308 5.2 754 0.5 102 2.9 27,569 0.7 27,916 0.5 2,043 9.1 182 2.8 1,367 1.2 17,484 1.1 417 7.2 2,501 1.8 1,214 2.5

Patient sex

Male N/A 342 7.3 49 0.0 1,863 1.3 108 3.7 5,376 4.0 1,170 1.2 139 1.4 16,053 0.6 17,145 0.6 1,738 9.0 238 5.0 1,262 1.4 5,469 1.0 526 7.2 3,879 1.2 1,120 2.4

Female 708 1.3 312 3.2 2,504 0.8 1,104 0.7 125 3.2 1,318 6.5 961 1.4 101 1.0 24,144 0.7 23,409 0.5 1,702 9.3 86 1.2 1,697 0.7 13,373 1.2 483 7.0 4,422 1.5 478 2.7

ASA score

<3 556 0.7 421 4 2,112 0.9 59 1.7 172 2.9 235 3.0 1,125 1.4 137 1.5 30,305 0.4 31,134 0.4 2,143 8.2 57 1.8 1,984 0.8 5,535 0.9 553 5.4 6,307 1.1 394 2.5

≥3 50 6.0 210 8.1 227 0.4 1,758 1.0 53 5.7 5,040 5.2 837 1.3 96 1.0 8,622 1.4 8,371 0.9 1,137 10.2 205 3.9 908 1.2 12,316 1.2 404 9.4 1,660 2.4 1,151 2.4

Duration of operation

≤T-time 554 1.1 409 3.9 2,409 0.8 2,358 1.1 90 2.2 5,763 4.0 1,857 1.0 99 0.0 34,498 0.5 36,849 0.5 2,085 7.7 213 4.2 2,265 0.4 15,348 1.0 724 6.6 6,413 0.9 1,112 1.9

>T-time 154 2 245 7.8 142 0.7 609 1.2 143 4.2 930 7.4 266 3.0 141 2.1 5,682 1.5 3,693 0.9 1,356 11.4 107 3.7 691 2.8 3,482 1.7 284 8.5 1,864 3.0 484 3.9

Wound class*

C/C-C 704 1.1 627 4.2 2,550 0.8 2,950 1.1 231 3.5 6,692 4.5 2,105 1.3 234 1.3 40,151 0.7 40,540 0.5 2,866 7.8 276 4.0 2,890 0.9 18,840 1.1 591 6.4 8,283 1.3 1,588 2.3

C/D 2 0.0 25 28.0 3 0.0 16 0.0 2 0.0 1 0.0 18 0.0 6 0.0 29 0.0 18 0.0 576 16.3 34 5.9 66 4.6 3 0.0 413 8.0 16 0.0 8 25.0

BMI

<30 217 0.9 184 4.9 310 1.0 1,474 1.0 33 6.1 3,692 3.4 937 1.3 12,599 0.5 10,370 0.4 1,171 7.8 43 7.0 343 2.6 2,414 0.9 293 7.2 1,562 0.9 358 1.7

≥30 169 1.8 74 6.8 146 1.4 574 1.9 32 0.0 1,814 6.7 354 2.0 7,388 1.1 12,542 0.7 401 10.0 30 13.3 89 3.4 278 2.9 80 10.0 809 2.1 140 3.6

Admission type

Elective 704 1.3 651 5.4 2,553 0.7 2,893 1.0 233 3.4 6,606 4.5 1,962 1.3 240 1.3 40,168 0.7 40,549 0.5 3,190 9.0 306 3.6 2,945 1.0 18,533 1.1 962 7.2 8,068 1.4 1,523 2.4

Emergency 4 0.0 3 0.0 0 0.0 75 4.0 0 0.0 88 4.6 169 1.2 0 0.0 33 0.0 15 0.0 252 11.9 18 11.1 14 0.0 309 0.3 47 6.4 235 0.4 75 4.0

31,661 0.5 35,632 0.4 96 2.1

Inflammatory joint disease 263 0.0 398 0.5 3 0.0

309 0.3 18 0.0 1 0.0

Infection 330 3.9 339 2.1 4 0.0

Fracture 341 2.6 60 3.3 18 11.1

Other 3,209 1.3 1,730 1.5 17 5.9

Unknown 537 0.4 297 0.3 11 0.0

All 4,417 2.1 2,426 1.8 50 4.3

1,548 1.0 167 1.2 18,117 1.1

877 1.3 841 0.7 93 2.2

*C/C-C=clean or clean-contaminated, C/D= contaminated or dirty; ** coronary artery bypass graft; ‡ data not available

‡

Primary indication

(Hip, Knee and Repair of neck of

femur)

Osteoarthritis

Avascular necrosis

Fracture

Vascular surgeryHip prosthesisGastric surgeryCranial surgery Knee prosthesis

Large bowel

surgery Limb amputation

Repair of neck of

femur

Small bowel

surgery Spinal surgery

Reduction of

long bone

fracture

Cardiac (non-

CABG) surgery Cholecystectomy CABG**

Other

Revision

Abdominal

hysterectomy

Bile duct/

liver/pancreatic

surgery Breast surgery


18

Section 4 Trends in the rate of SSI

4.1 Orthopaedic categories

Figure 3 shows trends in the crude cumulative incidence of SSI in the mandatory orthopaedic categories

in the 10 year period between 2004/5 and 2013/14. For each category, three types of trends are shown:

one based on inpatient SSIs, one based on readmission SSIs and one based on combined inpatient and

readmission SSIs. For all these categories of surgery inclusion of readmission infections from July 2008

increased the SSI risk estimates compared to the inpatient SSI estimates.

To estimate the average annual change in SSI, a univariable logistic regression model was used with a

linear assumption. Trends in inpatient and readmission SSIs in the six year period from July 2008 to

March 2014 were evaluated as this indicator formally superseded the inpatient-based indicator from July

2008. Since some categories exhibited variation in the annual crude incidence, estimates from the linear

model were compared against those from an independent-year effect model (with the initial year as the

baseline) using a likelihood ratio test (LR) to determine if model fit improved with the linear model. The

Odds Ratio (OR) with the 95% confidence interval (CI) is reported throughout.

Over the six year period, no clear trend was evident in hip or knee prosthesis. The SSI incidence for hip

prosthesis was at its peak in 2008/9 at 0.9% reducing to 0.6% in 2013/14 with small fluctuations in the

intervening period. For knee prosthesis the incidence was at its lowest in 2008/09 at 0.5% and remained

as 0.5% in 2013/14 with small increases in the intervening period. The LR test showed evidence of non-

linearity in both categories (p=0.0035 and p=0.0002 respectively). For hip prosthesis there was

evidence of a decrease compared to the baseline in only three specific years (2009/10, 2011/12 and

2013/14), of which 2011/12 exhibited the largest decrease (OR: 0.73, 95% CI: 0.59 – 0.88, p=0.002;

OR: 0.70; 95% CI: 0.58 – 0.85; p<0.001 and OR: 0.73; 95% CI: 0.61 - 0.88; p=0.001 respectively). For

knee prosthesis, evidence of an increase occurred in only three specific years (2010/11, 2011/12 and

2012/13) of which 2011/12 exhibited the largest excess risk (OR: 1.41; 95% CI: 1.14 – 1.74; p=0.002).

For repair of neck of femur and reduction of long bone fracture surgery, a general decline in the crude

inpatient and readmission incidence occurred reaching 1.1% and 0.9% in 2013/14 respectively. The

univariable model found significant decreases in both categories (OR: 0.93; 95% CI: 0.90 – 0.96;

p<0.001 and OR: 0.91; 95% CI: 0.83 – 0.99; p=0.034 respectively). In both categories, the LR test did

not reject the assumption of linearity meaning that the data were described well by the linear model

(p=0.2672 and p=0.7577 respectively).


19

Figure 3: Trends in the annual cumulative incidence of SSI (%) in the orthopaedic

surveillance categories, with lower and upper 95% CIs, NHS hospitals in England

4.2 Hospital-level trends – orthopaedic categories

Trends in the incidence of inpatient and readmission SSIs from July 2008 to March 2014 were examined

by hospital and orthopaedic category. Hospitals with less than four quarters of surveillance data during

the study period were excluded. Hospitals that experienced 0% SSI incidence throughout the study

period lacked the variation needed to assess trends and were therefore also excluded from the trend

analyses (data from such hospitals led to model convergence problems in the statistical analysis). A

total of 506,597 orthopaedic procedures and 4,106 SSIs submitted by 199 hospitals were available for

analysis. A univariable model based on a generalised linear model was used for evaluating trends.

Figure 4 shows that a high proportion of participating hospitals experienced either stable or a

decreasing incidence for each surgical category during the study period, this proportion being highest in

repair of neck of femur (94%) and lowest in reduction of long bone fracture (70%).


20

Figure 4: Hospital-level trends in the incidence of SSI, orthopaedic surveillance

categories between July 2008 and March 2014, NHS hospitals in England

4.3 Non-orthopaedic categories

Figure 5 shows the trends in the crude inpatient SSI incidence, the readmission SSI incidence and the

combined inpatient and readmission SSI incidence for the non-orthopaedic categories. Surgical

categories introduced after July 2008 are presented last.

For most of the 13 voluntary surgical categories the crude cumulative SSI estimates varied substantially

between years. The linear and year-effects model were compared to evaluate trends based on inpatient

readmission SSIs from July 2008 to March 2014. Of thirteen categories examined, three showed

evidence of a clear trend. Bile duct/liver/pancreatic and gastric surgery showed evidence of a significant

decrease while spinal surgery showed evidence of a significant increase.

For bile duct/liver/pancreatic surgery the inpatient/readmission SSI incidence decreased in general from

the peak observed in 2008/9 at 9.6% to 5.4% in 2013/14. There was no evidence against linearity

(p=0.2336) indicating that the data were described well by the linear model; the average annual

decrease was 13% (OR: 0.87; 95% CI: .79 – 0.94; p=0.004). For gastric surgery, the incidence was at

its peak in 2008/09 at 8.6% declining sharply to 1.3% by 2013/14. There was no evidence against

linearity in this category (meaning that the data were described well by the linear model) although this

reached borderline significance (p=0.0512); the average annual decrease was 22% (OR: 0.78; 95% CI:


21

0.61 – 1.00; p=0.046). For spinal surgery a general steady increase in the crude incidence occurred

from 1.1% in 2008/9 to 1.3% in 2013/14 with small fluctuations in the intervening period. There was no

evidence against linearity (p=0.2445) indicating that model fit was improved by the linear model. The

average annual increase was 12% (OR: 1.12; 95% CI: 1.04 – 1.20; p=0.001).

Figure 5: Trends in the annual cumulative incidence of SSI (%) in the non-orthopaedic

surveillance categories with lower and upper 95% CLs, NHS hospitals in England


22

Section 5 Characteristics of surgical site

infections

5.1 Type of SSI

Figures 6 and 7 show the distribution of SSI type indicating the site and depth of infection for inpatient

cases compared to that for the inpatient and re-admission cases combined (from April 2013 to March

2014) by surgical category.

Overall, information on SSI type was available for all but two of the 1,694 SSIs detected in 2013/14. SSI

type data for three mandatory orthopaedic categories were analysed. Reduction of long bone fracture

was excluded since the sample involved only 28 inpatient/readmission SSIs. Five of the remaining 13

remaining surgical categories were also examined on the basis of having ≥ 40 inpatient and readmission

SSIs (the others excluded due to smaller sample sizes). Data from the nine surgical categories

examined yielded a total of 1,545 inpatient/readmission SSIs with SSI type data.

The proportion of inpatient-detected SSIs that were superficial incisional infections varied by surgical

category ranging from 39.0% in spinal surgery to 73.3% in reduction of long bone fracture. The

observed proportions are affected by differences in the length of post-operative hospital stay between

surgical categories.

Re-admission SSIs involve more serious wound complications. Inclusion of these readmission cases

tended to increase the proportion of SSIs that are deep or organ-space, particularly in surgical

categories where the median length of hospital stay was relatively short. For example, in hip and knee

prosthesis surgery, the proportion of inpatient and readmission detected cases that were either deep or

organ-space increased to 64.8% and 60.0% respectively compared to the proportions based on

inpatient- detected cases. In most other voluntary categories included in this analysis, the median length

of hospital stay was longer than that for hip or knee prosthesis but typically the proportion of inpatient

and readmission SSIs that were deep or organ-space was broadly similar to the corresponding

proportion for inpatient-detected SSIs.


23

Figure 6: Distribution of SSI type in inpatient cases compared to combined cases, orthopaedic surgery, NHS hospitals in England, 2013/14

Figure 7: Distribution of SSI type in inpatient cases compared to combined cases, non-orthopaedic surgery, NHS hospitals in England, 2013/14


24

5.2 Causative micro-organisms

Figure 8 shows the trends in organisms reported as causing inpatient SSIs from 2004/05 to

2013/14 based on data from all surgical categories except breast, cardiac (non-CABG), cranial

and spinal surgery as these modules were introduced more recently. Inpatient SSIs were

analysed to remove the effect of readmission surveillance introduced in July 2008. Overall, of

10,208 inpatient SSIs detected over this 10-year period, 68% (6,940/10,208) had data on micro-

organisms reported as causing SSI.

S. aureus as a reported cause of inpatient SSIs accounted for 16% of cases in 2013/14

following a decreasing trend commencing from 2006/7 where it accounted for 39% of cases and

signalled the first occurrence of a decreasing trend. The reduction in S. aureus was driven by a

concurrent decrease in methicillin-resistant S. aureus (MRSA). By 2013/14 MRSA accounted for

4% of cases. Methicillin-susceptible S. aureus (MSSA) however showed marginal changes and

coagulase-negative staphylococci (CoNS) exhibited marginal increases. Enterobacteriaceae as

reported causes of SSIs accounted for 26% of cases in 2013/14 following a markedly increasing

trend observed from 2008/9.

Figure 8: Trends in key micro-organisms reported as causing SSI (inpatient), all

surgical categories*, NHS hospitals in England

*excludes breast, cranial, cardiac (non-CABG) and cranial surgery


25

Table 6 shows the distribution of organisms reported to cause SSIs in 2013/14 by surgical

category. Categories with organism data on more than 80 SSI cases were analysed. Categories

with the lowest proportion of SSI cases with microbiology data were large bowel surgery at 45%

(141/316) and CABG surgery at 58% (176/302). Repair of neck of femur had the highest

proportion of SSI cases with microbiology data at 84% (177/211). Polymicrobial SSIs (cases

with more than one organism reported as causing SSI) were most frequent in CABG at 39%

(68/176) and lowest in spinal surgery at 17% (14/82).

For monomicrobial SSIs (one organism reported as causing SSI), those reported as having S.

aureus aetiology accounted for the majority of SSIs in knee prosthesis (45/97) and repair of

neck of femur (55/119) at 46% each, followed by spinal surgery at 44% (30/68). SSIs reported

as having Enterobacteriaceae aetiology was highest in large bowel surgery at 57% (51/90)

followed by CABG surgery at 25% (27/108). It should be noted that organisms reported as

‘coliforms’ still account for a relatively high proportion of all Enterobacteriaceae at 21% (32/154)

in 2013/14. CoNS were highest in CABG surgery at 31% (33/108).

Among SSIs reported as having polymicrobial aetiology, Gram-positive and Gram-negative

combinations were the predominant type across all six categories analysed.

Table 6: Distribution of micro-organisms reported as causing SSI (inpatient/readmission), by surgical category, NHS hospitals in England, 2013/14

Reported causative organism No. % No. % No. % No. % No. % No. %

Monomicrobial Methicillin-sensitive S. aureus 49 35.3% 42 43.3% 31 26.1% 25 23.1% 9 10.0% 30 44.1%

Methicillin-resistant S. aureus 7 5.0% 3 3.1% 24 20.2% 4 3.7% 3 3.3% 0 0.0%

Coagulase-negative staphylococci 27 19.4% 28 28.9% 18 15.1% 33 30.6% 1 1.1% 16 23.5%

Enterobacteriaceae 26 18.7% 13 13.4% 26 21.8% 27 25.0% 51 56.7% 11 16.2%

Pseudomonas spp. 4 2.9% 1 1.0% 5 4.2% 8 7.4% 8 8.9% 5 7.4%

Streptococcus spp. 5 3.6% 5 5.2% 1 0.8% 1 0.9% 8 8.9% 2 2.9%

Enterococcus spp. 13 9.4% 2 2.1% 8 6.7% 3 2.8% 1 1.1% 2 2.9%

Other bacteria 8 5.8% 3 3.1% 4 3.4% 7 6.5% 7 7.8% 2 2.9%

Fungi including Candida spp. 0 0.0% 0 0.0% 2 1.7% 0 0.0% 2 2.2% 0 0.0%

Total monomicrobial 139 100% 97 100% 119 100% 108 100% 90 100% 68 100%

Polymicrobial Gram-positive combinations only 21 31.3% 12 34.3% 11 19.0% 9 13.2% 0 0.0% 3 21.4%

Gram-negative combinations only 15 22.4% 4 11.4% 11 19.0% 16 23.5% 7 13.7% 0 0.0%

Gram-posItive and Gram-negative combinations 30 44.8% 17 48.6% 33 56.9% 36 52.9% 25 49.0% 8 57.1%

Other combinations* 1 1.5% 2 5.7% 3 5.2% 7 10.3% 19 37.3% 3 21.4%

Total polymicrobial 67 100% 35 100% 58 100% 68 100% 51 100% 14 100%

Total monomicrobial and polymicrobial cases (% of total SSI cases) 206 78.9% 132 64.1% 177 83.9% 176 58.3% 141 44.6% 82 73.9%

55 21.1% 74 35.9% 34 16.1% 126 41.7% 175 55.4% 29 26.1%

261 100% 206 100% 211 100% 302 100% 316 100% 111 100%

*combination that include 'anaerobic bacilli' 'anaerobic cocci' ,‘other bacteria’ or fungi

Total cases

SpinalHip Prosthesis Knee Prosthesis CABG Large bowelRepair of neck of femur

Total cases with unknown microbial aetiology (% of total SSI cases)


26

Section 6 Variation in rates of SSI between

NHS hospitals

6.1 Box and whisker plots

Figure 9 is a box and whisker plot showing the distribution of individual hospital SSI estimates (%)

against five percentiles (10th, 25th, 50th, 75th and 90th) by surgical category. Each percentile represents a

value below which a proportion of the total observations lie. The box plots show inter-hospital variation

within each surgical category inter-category variation. SSI estimates that lie at extreme ends (above the

90th percentile or below the 10th percentile) may be indicative of a problem (either excess or failure to

detect SSIs respectively).

Figure 9: Distribution of the cumulative SSI incidence (%) at hospital level* by

surgical category, NHS hospitals in England, April 2009 to March 2014

* hospitals with <95 operations in hip, knee or abdominal surgery are excluded; in the remaining categories, hospitals with <45 operations are

excluded. For categories with an insufficient number of hospital participants (<20), the distribution without a box plot is presented.


27

6.2 Funnel plots

In 2013/14, data on 96,909 orthopaedic procedures and 696 inpatient/readmission SSIs were submitted

by 184 NHS hospitals representing 143 NHS Trusts. An additional eight independent sector NHS

Treatment Centres contributed data on 5,661 orthopaedic procedures and 10 SSIs. One NHS Trust did

not contribute orthopaedic data in 2013/14 as part of the mandatory surveillance requirement for

healthcare associated infections.

Figure 10 shows funnel plots displaying variation in the Trust SSI 2013/14 by orthopaedic category. The

cumulative incidence of SSI per 100 operations is plotted against the number of procedures for each

participating NHS site. The 95% control limits (dashed black lines) define the ‘limits’ of expected

variation in the Trust estimates. Any data point lying outside these limits is considered an outlier. These

outliers are described as having ‘special cause variation’ and should be used as triggers for further

investigation as the results from funnel plots are designed to be indicative not confirmatory. The margin

of error associated with 95% control limits is 5%. The 99% control limits (solid red lines) are also

presented for information. The funnel plots do not adjust for case-mix so the results should be

interpreted with caution.

The funnel plots identified a total of 17 individual NHS Trusts that were statistical outliers with SSI

estimates beyond the 95% CLs in 2013/14. Of these 17 Trusts, 10 were high outliers and seven were

low outliers. Of 10 Trusts identified as high outliers, three were high outliers in more than one surgical

category. Of seven Trusts identified as low outliers, one was a low outlier in more than one surgical

category. Five of 17 outlier Trusts were also outliers in the previous year.

Although a low outlier may reflect a high standard of infection control practice and surgical skill, other

factors should be considered to eliminate methodological bias. For example failure to implement a

systematic readmission alert system may miss SSI cases. All high and low outlier Trusts have been

contacted to make them aware of their outlier status and encouraged to review their practices including

their surveillance methodology.

The individual Trust SSI estimates are published separately in an accompanying supplement to this

report:

https://www.gov.uk/government/publications/surgical-site-infections-ssi-surveillance-nhs-hospitals-in-

england




28

Figure 10: Cumulative incidence of inpatient/readmission SSI plotted against the number

of operations by NHS Trust and mandatory orthopaedic surgical category in England,

2013/14


29

Section 7 Conclusions

7.1 Trends in the rate of SSI

Between July 2008 and March 2014, significant decreases in the incidence of inpatient and readmission

SSIs were observed at national level in repair of neck of femur and reduction of long bone fracture. For

hip and knee and prosthesis surgery, no trends were detected and the SSI incidence remained low in

these categories at <1%. The incidence of SSI in hip and knee prosthesis was similar to the overall

European estimate based on aggregated data from other European countries including the UK[8].

However direct comparisons with individual countries are difficult due to variation in follow-up methods,

the interpretation of case definitions, patient demographics, the mix of high and low risk procedures and

differences in healthcare systems.

Trends at hospital-level based on NHS hospitals or NHS Treatment Centres contributing data in four or

more surveillance periods showed that the majority of hospitals experienced either a reduction or a

stable incidence of inpatient and readmission SSI between July 2008 and March 2014. Overall, the

results for the orthopaedic categories are encouraging. More in-depth analyses are needed to evaluate

the inpatient and readmission trends taking into account key risk factors and follow-up time.

In the non-orthopaedic modules, significant decreases were found for bile/duct/liver/pancreatic and

gastric surgery. However a significant increase was found for spinal surgery. Further study is needed to

identify possible causes for these increases. For example, the increase may reflect a shift towards more

complex patients requiring open rather than laparoscopic surgery, the latter being excluded from the

national surveillance.

7.2 Variation in the SSI rate between participating hospitals

Variation in SSI incidence between hospitals can be expected to occur and provides a trigger for

identifying possible means to reduce infection rates. Funnel plots were used to examine the extent of

such variation among NHS Trusts and NHS Treatment centres participating in the mandatory

orthopaedic categories. A total of 17 individual NHS Trusts were identified as outliers; 10 of which were

high outliers and seven low outliers. These individual Trusts have been contacted and encouraged to

investigate further.


30

7.3 Optional post-discharge surveillance

Post-discharge surveillance in outpatient settings or via patient post-discharge questionnaire provides

trusts with valuable additional data on SSIs that occur after discharge. Based on national aggregated

data, we showed that the proportion of PDQs used for surveillance and the proportion completed as a

percentage of those given, have been increasing since they were introduced in 2008. The results for

PDQ surveillance activity in England are encouraging. However there can be considerable variation in

the proportion of procedures with completed follow-up. This variation has been found to show some

correlation with the SSI rate[9]. High levels of patient coverage and patient response rates are needed

in the future before benchmarking based on PDQs could be implemented.

7.4 Microbial aetiology

S. aureus as a reported cause of SSI reduced substantially between 2004/5 to 2013/14. It accounted for

16% of inpatient cases in 2013/14 following year-on-year decreases commencing from 2006/7. These

decreases were explained by concurrent decreases in MRSA, from 25% in 2006/7 to 4% of cases in

2013/14. The low occurrence of MRSA is likely to reflect the impact of infection control initiatives

directed at controlling MRSA[10]. Enterobacteriaceae as reported causes of SSIs increased from

2008/09 onwards and accounted for 26% of cases in 2013/14.

At surgical-category level, data for 2013/14 showed that among monomicrobial infections, S. aureus

was the predominant organism in orthopaedic and spinal surgery accounting for between 40% and 46%

of cases depending on the category. CoNS were the predominant organisms in CABG surgery that may

reflect the use of implants. Enterobacteriaceae were the predominant organisms in large bowel surgery

reflecting the expected organisms at the operative site.

Among polymicrobial infections, Gram-positive and Gram-positive combinations were the most frequent

type of across all surgical categories. Although this combination affected a small set of patients, this

may have implications for choice of surgical antimicrobial prophylaxis at local level.

7.5 Guidelines for the prevention of SSI

Factors underpinning SSI risk are related to patient characteristics and peri-operative factors. Although

many patient-related risk factors are non-modifiable there are several aspects in the peri-operative

phase that can be optimised to minimise the risk of SSI. Implementation of key national guidelines and

standards provides a means of achieving a patient safety culture within surgery. These guidelines cover

key aspects of surgical antibiotic prophylaxis practices (choice, timing of agent, frequency and, dose),


31

glucose control and skin asepsis[6;7]. NICE’s recent publication of the Quality Standards for SSI

provides a simple evidence-based checklist of pre-operative, peri-operative and post-operative

standards of care designed to drive improvements in patient outcome[11].

7.6 Future directions

Further to the 2013 survey of NHS Trusts’ future priorities for SSI surveillance, results are currently

being finalised and will be disseminated over the coming months to all NHS Trusts. Collated results are

being used to inform the development of the national SSI surveillance programme.

The existing SSI web-based application, first developed in 2004 and then in 2008, is currently

undergoing a re-design to provide a more efficient and intuitive system for hospitals to enter data and

access reports. Hospitals’ feedback has been pivotal in developing the new design. Beta testing (or

user acceptance testing) of the SSI application is planned for early 2015; any hospitals interested in

assisting with this should contact the SSI team ([email protected]).


32

Section 9 Glossary

Cumulative Incidence

The total number of surgical site infections as a proportion of the total number of patients undergoing a

procedure in the same category of surgery expressed as a percentage.

Incidence Density

The total number of SSIs (identified through inpatient surveillance) divided by the total number of days

of inpatient follow-up expressed as the number of SSIs per 1,000 days of patient follow- up.

Confidence Interval and Control Limits (CL)

The 95% confidence interval provide a guide to the precision of the estimate based on the number of

procedures. Given the level of confidence, they indicate that the ‘true’ incidence’ could lie anywhere

between the lower and higher confidence limits. Control limits (CL) used in funnel plots are equivalent

to exact binomial confidence limits at 95% (warning) and 99% (action). These define the limits of

expected variation. The probability of a cumulative incidence lying above the 95% CL by chance alone

is less than 5% (there is a 1 in 20 chance that the observed estimate is due to chance alone in a defined

period); if above the 99% CLs, this is less than 1%. The 99% control limits also define the limits of

expected variation and are computed based on the same ‘sample size; the confidence range is wider

but this is offset with a lower margin of error (1%). The 99% control limits are presented for information

since the conventional level of significance is 5%.

P-value

This measures the probability of an estimate (e.g. a risk) being as or more extreme than that observed

in a study population occurring by chance alone if the null hypothesis is true (the assumption that there

is no difference between different groups). The significance level for the P-value is conventionally set at

<0.05. This means that we are prepared to accept no more than a 5% chance of being wrong in

claiming that our observed result is true. If the p-value computed around an estimate is >0.05, this

indicates that the data do not provide sufficient evidence to reject the null hypothesis.

Odds Ratio This is the odds or likelihood of an event e.g. an infection in an ‘exposed’ group of subjects compared to

another group (‘non-exposed’). An Odds Ratio of one indicates that there is no difference in the odds of

developing an SSI between the two groups. An Odds Ratio of >1 indicates that the odds or likelihood of

developing an SSI is greater in the ‘exposed’ group compared to the ‘non-exposed’. An Odds Ratio of

<1 indicates that the odds or likelihood of developing an SSI is higher in the ‘non-exposed group’.


33

Section 10 References

[1] Chief Medical Officer. Surveillance of Healthcare Associated Infections PL CMO 2003(4).

Department of Health. 2003 (available at

http://www.dh.gov.uk/en/Publicationsandstatistics/Lettersandcirculars/Professionalletters/

Chiefmedicalofficerletters/DH_4003782)

[2] Public Health England. Protocol for the surveillance of surgical site infection. Version 6

June 2013. Public Health England. 2013 (available at https://www.gov.uk/government/publications/surgical-site-infection-surveillance-service-

protocol-procedure-codes-and-user-manual)

[3] Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG. CDC definitions of nosocomial

surgical site infections, 1992: a modification of CDC definitions of surgical wound

infections. Am J Infect Control 1992; 20:271-274.

[4] Culver DH, Horan TC, Gaynes RP, Martone WJ, Jarvis WR, Emori TG et al. Surgical

wound infection rates by wound class, operative procedure, and patient risk index.

National Nosocomial Infections Surveillance System. Am J Med 1991; 91:152S-157S.

[5] European Centre for Diseased Prevention and Control. Surveillance of surgical site

infections in European hospitals - HAISSI Protocol. Version 1.02. Stockholm: ECDC; 2012

[6] National Collaborating Centre for Women's and Children's Health. Surgical site infection -

prevention and treatment of surgical site infection. Clinical Guideline. NICE. 2008

(available at http://www.nice.org.uk/nicemedia/pdf/CG74FullGuideline.pdf)

http://www.dh.gov.uk/en/Publicationsandstatistics/Lettersandcirculars/Professionalletters/Chiefmedicalofficerletters/DH_4003782

http://www.dh.gov.uk/en/Publicationsandstatistics/Lettersandcirculars/Professionalletters/Chiefmedicalofficerletters/DH_4003782

http://www.nice.org.uk/nicemedia/pdf/CG74FullGuideline.pdf


34

[7] Scottish Intercollegiate Guidelines Network. Antibiotic prophylaxis in surgery: A National

Clinical Guideline. Scottish Intercollegiate Guidelines Network. 2008 (available at

http://www.sign.ac.uk/pdf/sign104.pdf)

[8] European Centre for Disease Prevention and Control. Surveillance of surgical site

infections in Europe, 2010-2011. European Centre for Disease Prevention and Control.

2013 (available at http://ecdc.europa.eu/en/publications/Publications/SSI-in-europe-2010-

2011.pdf)

[9] Wilson J, Wloch C, Saei A, McDougall C, Harrington P, Charlett A et al. Inter-hospital

comparison of rates of surgical site infection following caesarean section delivery:

evaluation of a multicentre surveillance study. J Hosp Infect 2013; 84:44-51

[10] Department of Health. Screening for Meticillin-resistant Staphylococcus aureus (MRSA)

colonisation - a strategy for NHS Trusts: a summary of best practice. 15-11-2006.

London, Department of Health.

[11] National Institute for Clinical Excellence. Surgical Site Infection Quality Standards, QS49.

2013. London, National Institute for Health and Care Excellence .

http://www.sign.ac.uk/pdf/sign104.pdf

http://ecdc.europa.eu/en/publications/Publications/SSI-in-europe-2010-2011.pdf

http://ecdc.europa.eu/en/publications/Publications/SSI-in-europe-2010-2011.pdf

surveillance of surgical site infections in nhs hospitals in england 2013/14 · 2014. 12. 12. ·...

Documents