Strategies to reduce the incidence of hepatopancreatobiliary gram-negative bloodstream infections in England


In 2016, the Secretary of State for Health announced an ambition to significantly reduce the incidence of gram-negative bloodstream infections (GNBSI) in England [Public Health England, “Preventing healthcare associated Gram-negative bloodstream infections: an improvement resource,” 2017]. The current target is a 25% reduction by 2021 [HM Government, “Tackling antimicrobial resistance 2019-2024: The UK’s five-year national action plan,” 2019]. The largest cohort of GNBSI patients are those of urinary origin, who have been targeted with various strategies over the past few years [3, 4]. The second largest cohort nationally are the hepatopancreatobiliary (HPB) cohort [J. Abernethy et al., “Epidemiology of Escherichia coli bacteraemia in England: results of an enhanced sentinel surveillance programme,” Hosp. Infect., vol. 95, no. 4, pp. 365–375, Apr. 2017]. Very little work has been conducted into exploring strategies to reduce GNBSI incidence in this cohort of patients.

A clinical review team from Oxford University Hospitals NHS Foundation Trust have conducted a retrospective case series analysis in order to identify potential strategies to reduce the incidence of GNBSI in this HPB cohort. Herein we present the key findings of our study.


We performed a retrospective analysis of 433 cases of GNBSI coded as being of HPB origin from the Oxford University Hospitals’ Trust database between April 2015 and May 2019. Electronic Patient Records (EPR) and Oxfordshire Care Summary (OxCS) were accessed in order to extract key data relating to each recorded case of GNBSI. These data included:

  • The system of origin of GNBSI
  • The aetiology of the infection (either benign or malignant HPB disease)
  • Whether the patients had any previous documented benign or malignant HPB disease prior to their presentation with GNBSI
  • For patients with previous benign HPB disease:
    • Whether the previous episode was managed surgically (cholecystectomy) or non-surgically (no cholecystectomy)
    • Whether patients who had been treated non-surgically for the previous episode were predominantly managed in primary or secondary care
    • The cause of GNBSI in patients who had been treated surgically
    • In cases of GNBSI caused by retained CBD stones following a prior cholecystectomy, whether intraoperative cholangiography was performed during the cholecystectomy
  • For patients with previous malignant HPB disease:
    • Whether they had a biliary stent in situ at the time of GNBSI
    • The material of any in situ stents
    • Whether patients with stents in situ were intended for surgical or non-surgical management of the HPB malignancy at the time of stent insertion
    • Whether patients with no stents in situ had clinical evidence of malignant biliary obstruction prior to GNBSI.


  • Of the 433 cases of GNBSI initially coded as being of HPB origin, 388 (90%) were truly HPB in origin.

GNBSI caused by gallstone disease

  • 282 of the 388 HPB GNBSI cases (73%) were caused by gallstone disease.
  • Of the 282 gallstone GNBSI cases, 117 (41%) had been previously diagnosed with symptomatic gallstone disease prior to their episode of GNBSI
  • Of the 117 who had a previous diagnosis of symptomatic gallstone disease, 93 (79.5%) had been managed non-surgically compared to 24 (20.5%) who had received a cholecystectomy previously.*
  • Of the 93 managed non-surgically, 81 (87.1%) were managed predominantly in secondary care, 6 (6.5%) in the community, and 6 (6.5%) had no secondary care records of their management.
  • For the 24 who were managed surgically for their previous gallstone presentation, the cause of their subsequent GNBSI was common bile duct stones in 18 patients (75%), bile leak in 1 patient (4%), liver abscess in 1 patient (4%), bile duct stenosis in 2 patients (8%), and unknown in 2 patients (8%).
  • Of the 18 in whom a retained common bile duct stone was the cause of their subsequent GNBSI, 5 operation notes were available to us, none of which mentioned intra-operative cholangiography being performed.

GNBSI caused by malignant HPB disease

  • 70 of the 388 HPB GNBSI cases (18%) were related to malignant HPB disease.
  • Of the 70 malignant HPB GNBSI cases, 54 had a diagnosis of HPB malignancy prior to their presentation with GNBSI.
  • Of the 54 who had a previous diagnosis of HPB malignancy, 37 had at least one biliary stent in situ at the time of GNBSI, while 17 had no stent.
  • Of the 37 cases with a stent in situ, 3 out of 37 stents were plastic, while the remaining 34 were self-expanding metal stents (SEMS). 2 out of 3 (67%) plastic stents were radiologically confirmed to be blocked at the time of GNBSI compared to 14 out of 34 (41%) of the SEMS.
  • Of the 37 cases with a stent in situ, 3 out of 37 were considered by the MDT to have operable cancers but developed GNBSI before a scheduled Whipple procedure. The remaining 34 were considered non-operable and received palliative management.
  • Based upon a prospective trial of fast-track Whipple procedure in Birmingham [K. J. Roberts et al., “A reduced time to surgery within a ‘fast track’ pathway for periampullary malignancy is associated with an increased rate of pancreatoduodenectomy,” HPB, vol. 19, no. 8, pp. 713–720, Aug. 2017.], we estimate that 2 of these 3 operable cases would have been eligible for expedited surgery.
  • Of the 17 cases with no stent in situ, 1 case (6%) had clinical evidence of malignant biliary obstruction prior to GNBSI.
  • Overall, 5 cases of GNBSI in the malignant cohort could have been avoided by use of SEMS instead of plastic stents, earlier stenting of palliative patients with jaundice, and expedited surgical management of operable patients with jaundice.

* We note that the SWORD AUGIS database highlights wide variation in practice across England, with some units performing approximately 5% of their laparoscopic cholecystectomies as emergencies compared to approximately 40% at other units [6].

**Bold indicates these 5 potentially preventable malignancy-related GNBSI cases if alternative management strategies had been implemented.


  • Gallstone disease is by far the most prevalent cause of HPB GNBSI, and should be a significant focus for all NHS trusts in reducing the incidence of GNBSI nationally.
  • In 79.5% of the cases in which there had been a previous diagnosis of symptomatic gallstone disease prior to the GNBSI episode, cholecystectomy had not been performed. This is contrary to NICE guidance relating to symptomatic gallstone disease. In other words, a greater proportion of patients with known gallstone disease go on to develop GNBSI if managed non-surgically.
  • 93 cases (23.9%) of gallstone-related GNBSI were potentially preventable if cholecystectomy had been performed at the index presentation of gallstone disease.
  • Intraoperative cholangiography may reduce the number of GNBSI cases in patients who had been previously surgically managed by reducing episodes of retained CBD stones (18 cases).
  • 5 cases of malignancy-related GNBSI were potentially preventable by ensuring expedited Whipple procedures, use of SEMS and earlier stenting where appropriate.
  • Overall, our analysis suggests up to 30% of cases of HPB GNBSI (116 of the total 388) were potentially preventable by alternative management strategies.

Specific recommendations based on these findings

  • Gallstone disease should be at the forefront of strategies to reduce the overall incidence of GNBSI across the country.
  • NICE guidance should be adhered to and a laparoscopic cholecystectomy offered to all patients presenting to secondary care with symptomatic gallstone disease where they are candidates for an operation. These operations should be carried out expeditiously rather than through an elective waiting list.
  • Patients should be referred on from primary care to secondary care for further diagnostic workup, and if appropriate, considered for a laparoscopic cholecystectomy when presenting with biliary disease in the community.
  • We should seek to reduce the wide variation in practice in England relating to the proportion of laparoscopic cholecystectomies that each unit performs as an emergency compared to those performed electively [Hospital episode statistics data for 2019-20 extracted from the Surgical Workload and Outcomes Database of the Association of Upper GI Surgeons of Great Britain and Ireland].
  • Intraoperative cholangiography should be considered more frequently during laparoscopic cholecystectomy, in order to reduce the incidence of residual bile duct stones after cholecystectomy.
  • Fast-track pathways for operable HPB cancers that omit pre-operative biliary drainage or stenting could reduce GNBSI occurring during surgical delays.
  • Biliary stents inserted for the drainage of malignant biliary obstruction should be SEMS.

Further work

  • Our unit is in the process of collaborating with other HPB units around the country to collate prospective data as to whether increasing the number of “hot” (or early) laparoscopic cholecystectomies leads to reduced incidence of gallstone related GNBSI.

4 March 2020

Document prepared by Dr Cian Wade[1,3] on behalf of the Oxford University Hospitals NHS Trust HPB GNBSI Review Group.

(Mr Mustafa Majeed[2], Mr Harry Ward[2], Ms Lisa Butcher[3], Mr Zahir Soonawalla[3], Mr Giles Bond-Smith[3])

1: Oxford University Clinical Academic Graduate School; 2: Medical Sciences Division, University of Oxford; 3: Oxford University Hospitals NHS Foundation Trust.