Article Text

Download PDFPDF

The utility of a blood culture database to identify patients suitable for outpatient parenteral antibiotic treatment
  1. Mark Melzer1,
  2. Liana Macpherson2,
  3. Catherine Welch3
  1. 1Royal London Hospital, Barts Health NHS Trust, London, UK
  2. 2Royal London Hospital, London, UK
  3. 3Department of Epidemiology and Public Health, University College London, London UK
  1. Correspondence to Dr Mark Melzer, Consultant in Infection, Royal London Hospital, Barts Health NHS Trust, 3'rd floor Pathology and Pharmacy Building, 80 Newark St., London E1 2ES, UK; mark.melzer1{at}


Background The clinical and cost-effectiveness of outpatient parenteral antimicrobial therapy (OPAT) services are well described. We used a blood culture database as a novel approach to case finding and determined its utility in identifying inpatients suitable for OPAT.

Methods From December 2012 to November 2013, consecutive adult inpatients with bacteraemia, and those recruited to OPAT, were prospectively studied. Univariate and multivariate logistic regression analysis were used to investigate the association between bacteraemic patient characteristics and OPAT recruitment.

Results There were 470 bacteraemic and 134 OPAT patients. The blood culture database identified 22 (16.4%; CI 10.5 to 23.6) additional patients suitable for OPAT, 4.7% (95% CI 3.0% to 7.0%) of the total bacteraemic cohort. 20 (90.9%) of these patients had community-acquired bacteraemia. Bacteraemic patients with urinary tract infections (UTIs), 11/157 (7.0%; 95% CI 3.5% to 12.2%) were most commonly recruited to OPAT and Escherichia coli was the most common blood culture isolate. In the E. coli bacteraemic subgroup, extended-spectrum β-lactamase (ESBL) producers were significantly higher in the OPAT group, compared with the non-OPAT group, 9/11 (81.8%) vs 17/192 (8.9%), p<0.001. Among OPAT patients, there were no deaths within 30 days and no significant difference in relapse rates between bacteraemic and non-bacteraemic patients, 1/22 (4.6%) vs 5/112 (4.5%). In logistic regression analysis, there were no patient characteristics in the bacteraemic cohort that predicted recruitment to OPAT. In a subgroup analysis of patients with Gram-negative bacteraemia, ESBL production was strongly associated with OPAT recruitment, OR 5.85 (95% CI 1.94 to 17.58), p=0.002.

Conclusions A blood culture database proved a useful adjuvant to a clinical referral system, particularly for patients with community onset, multidrug resistant UTIs caused by ESBL producing E. coli. All bacteraemic patients recruited to OPAT received treatment safely and had good clinical outcomes.

  • Bacteraemia
  • Outpatient Parenteral Antibiotic therapy
  • Ascertainment
  • Patient identification

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


Outpatient parenteral antibiotic treatment (OPAT) enables patients to receive treatment with intravenous antibiotics in their home or in an ambulatory care setting rather than in hospital. It is widely used in the UK and is associated with admission avoidance, decreased length of inpatient stay, cost savings and high levels of patient satisfaction.1–5 In carefully selected patients, outcomes are comparable with hospital based treatment6 ,7 and national guidelines assist with service delivery, development and good clinical practice. OPAT is also part of the UK government's healthcare strategy of moving hospital services into the community so OPAT is likely to continue to expand over the next 5 years.8

While there is much published literature on the success of OPAT services, there is little on methods used to identify patients. Traditionally, patient identification can be passive, waiting for a referral, or active, targeting specific specialties (eg, orthopaedics or acute assessment units). Often referrals are unsuitable for several reasons. These include patients living outside designated postcodes, inability to attend hospital for treatment on a daily basis, being elderly with multiple comorbidities that necessitate prolonged hospital stay, difficulty with vascular access and having conditions treatable with oral antibiotics. Bacteraemic patients feature in some OPAT populations although it is unclear the site of infection causing bacteraemias and how these patients were recruited.7 ,9–11 Overall, there is little in the literature on the value of microbiological results in patient identification with the emphasis on clinical assessment and the safe delivery of treatment.7 ,12 We used an existing blood culture database as a novel approach for recruitment, and determined its utility in identifying additional adult inpatients suitable for OPAT.

Patients and methods

Study setting

This study was undertaken at the Royal London Hospital (RLH), Barts Health NHS Trust. The RLH serves a diverse population of about 250 000 patients in Tower Hamlets, East London and is a regional referral centre for the North-East London sector. In addition to accident and emergency, general medicine, surgery, paediatric and maternity services, the RLH has 60 high-dependency and critical care beds (including neurosurgical, renal, and obstetric and gynaecological beds), specialist wards for renal transplant and haemodialysis patients, and a high-level intensive care unit.

Study population

From December 2012 to November 2013, consecutive bacteraemic inpatients and patients recruited to OPAT were prospectively studied. Patients aged <16 years were excluded.

Bacteraemia cohort and definitions

In bacteraemic patients, age, gender, inpatient specialty, site of infection, organism, susceptibility profile and mortality related outcomes were recorded. Bacteraemia was considered significant if a blood culture was isolated from a patient with a compatible clinical syndrome that was unlikely to be a skin or environmental contaminant. This was based upon the patient's history, examination, response to antimicrobial therapy and bacterial isolates from other body sites.13 Specialties at the time of bacteraemia were categorised as medicine, surgery (including orthopaedics), critical care, and obstetrics and gynaecology. For hospital-acquired or device related bacteraemia, the Centres for Disease Control and Prevention definitions were used to define the sites of infection14 and for community-onset bacteraemia, sites were defined following clinical, microbiological and radiological assessment. Bacteraemia in patients with an unknown source were classified as undefined.

Microbiology data

Blood cultures were analysed using an automated system BacT/ALERT3D (bioMerieux, Mary l'Etoile, France). Isolates were identified using either the VITEK MS system (bioMerieux, Mary l'Etoile, France, database V.2.0) or Bruker Biotyper (Bruker Daltonic, Leipzig, Germany, software V.3.0) MALDI-TOF MS systems according to the manufacturer's instructions and the laboratory standard operating procedures. Susceptibility testing was performed on the Microscan walkAway system (Siemens Healthcare Diagnostics, Deerfield, Illinois, USA).

OPAT cohort, data collection and ascertainment

Over the same period age, gender, site of infection, medical specialty, presence of bacteraemia, duration of treatment and outcomes were recorded for all patients recruited to OPAT. Where possible, patients were treated with intravenous antibiotics once daily. Patients under 16 years of age were excluded. Patients were recruited by referrals from inpatient teams or general practitioners. or actively sought by attendance at acute assessment unit board rounds or attendance at multidisciplinary team meetings (eg, orthopaedics). In addition, some OPAT referrals were made as a consequence of blood culture results. These results prompted assessment for OPAT suitability by either an infection specialist (including a microbiologist) or the patient's clinical team. Patients received treatment in their homes or via a fast response nursing team in an ambulatory care setting.

Statistical analysis

We analysed the characteristics of bacteraemic patients (age, gender, place of acquisition, inpatient specialty, site of infection and mortality), comparing those who received and did not receive OPAT. For patients recruited to OPAT, we compared patient characteristics and duration of intravenous treatment for bacteraemic and non-bacteraemic patients. We also describe bacteraemic isolates from patients who received OPAT and compared these to patients who did not receive OPAT.

Quantitative data are presented as numbers and percentages. Associations between two categorical variables were tested using the Pearson's χ2 test and continuous variables using t-tests. As patients may present with more than one bacteraemic episode, we used number of patients as the denominator to calculate percentages for patient characteristics and number of bacteraemic episodes as the denominator for infection characteristics.

Univariate and multivariate logistic regression analyses were used to examine the association between age, gender, inpatient specialty and site of infection, and recruitment to OPAT in bacteraemic patients. As all OPAT bacteraemic episodes were community-onset, we could not adjust for place of acquisition because there were no OPAT patients with hospital-acquired bacteraemia. In a subgroup analysis of patients with Gram-negative bacteraemia, we also examined the association between Escherichia coli infection and extended-spectrum β-lactamase (ESBL) production, and recruitment to OPAT. Generalised estimating equations were used to account for dependency between multiple bacteraemic episodes for patients in the univariate and multivariate analyses. Data were analysed using Stata SE (V.13.1).

Clinical governance

The clinical governance audit committee of Barts Health NHS Trust approved the study.


Over the 12-month period, the number of inpatients with bacteraemia and patients recruited to OPAT are illustrated in figure 1.

Figure 1

Inpatients with bacteraemia and patients recruited to outpatient parenteral antibiotic treatment (OPAT) between December 2012 and 2013.

Bacteraemic patients

There were 470 patients with bacteraemia yielding 556 positive blood cultures. Patient demographic, clinical and mortality data are summarised in table 1. Twenty-two (4.7%; 95% CI 3.0% to 7.0%) bacteraemic patients were recruited to OPAT. All patients recruited had community-onset infection (either community-acquired or healthcare-associated). Of these, 20 had community-acquired bacteraemia, equivalent to 20/200 (10.0%; 95% CI 6.2% to 15.0%) of all community-acquired bacteraemic episodes.

Table 1

Demographic and clinical data on 470 patients, with 556 associated bacteraemic or fungaemic episodes, who did and did not receive OPAT

Compared with surgical patients, significantly more medical patients were recruited to OPAT. Outpatient treatment episodes were most common for urinary tract infection (UTI), 11/157 (7.0%; 95% CI 3.5% to 12.2%), skin and soft tissue infection (SSTI), 3/29 (10.3%; 95% CI 2.2% to 27.4%), gastrointestinal tract infection (all enteric fevers), 3/8 (37.5%; 95% CI 8.5% to 75.5%) and biliary tract infection, 3/54 (5.6%; CI 95% 1.2% to 15.4%). There was no statistically significant difference in 30-day mortality between bacteraemic patients receiving OPAT and those not receiving OPAT; 0 vs 24/470 (5.4%; 95% CI 3.3% to 7.5%).

OPAT patients

Of the 134 patients who received OPAT, 22 (16.4%; 95% CI 10.5% to 23.6%) were bacteraemic. All these patients were recruited through the blood culture database and were not referred directly from clinical teams. There was one patient with a bacteraemia and a non-bacteraemic episode so this information was recorded in both groups.

Demographic and clinical data of all OPAT patients are summarised in table 2. UTI, 11/51 (21.6%; 95% CI 11.3% to 35.3%) was the most common cause of bacteraemia in OPAT patients. Unadjusted data demonstrated significant differences in sites of infection, with more upper UTIs in the bacteraemic group compared with the non-bacteraemic group, 9/22 (40.9%) vs 26/123 (21.1%), p=0.046. There were no deaths in either bacteraemic or non-bacteraemic patients and no significant difference in relapse rates at 30 days; 1/22 (4.6%) vs 5/112 (4.5%).

Table 2

Demographic and clinical data on 134 patients who received OPAT (145 treatment episodes), with bacteraemia and without bacteraemia

Total number of days on intravenous antibiotics received out of hospital was 1198, roughly equivalent to the number of bed days saved. The most common drugs administered were either intravenous ceftriaxone once daily or ertapenem. In bacteraemic and non-bacteraemic groups, median (IQR) duration of intravenous treatment was non-significantly different, 7 (5–8) vs 6 (3–8).


For all bacteraemic patients, 377 (67.8%) blood culture isolates were Gram-negative. The most common organisms were E. coli and Staphylococcus aureus (table 3). More patients with E. coli bacteraemia were recruited to OPAT compared with other blood culture isolates and, among these, ESBL production was significantly higher in the OPAT group compared with the non-OPAT group, 9/11(81.8%) vs 17/192 (8.9%), p<0.001.

Table 3

Bacteraemia isolates on 470 patients, with 556 associated bacteraemic or fungaemic episodes, receiving or not receiving OPAT

Bacteraemic patient characteristics and recruitment into OPAT

Univariate and multivariate logistic regression analysis to investigate the association between bacteraemic patient characteristics and those recruited to OPAT is reported in table 4. There were no statistically significant associations in univariate or multivariate logistic regression analysis. In a subgroup analysis of patients with Gram-negative bacteraemia, ESBL production was strongly associated with OPAT recruitment, OR 5.85; 95% CI 1.94 to 17.58, p=0.002 (table 5).

Table 4

Univariate and multivariate logistic regression analyses for all bacteraemic patients to investigate association between patient characteristics and recruitment to OPAT (469 patients, 555 bacteraemic patients)

Table 5

Univariate and multivariate logistic regression analyses for Gram-negative bacteraemic patients to investigate the association between patient characteristics, blood culture isolates and recruitment to OPAT (332 patients, 377 bacteraemic patients)


This is the first paper to examine the utility of a blood culture database to identify patients suitable for OPAT. An additional 16.4%, or 22 extra patients, were recruited to OPAT through the blood culture database. Without this ‘prompt’, clinical assessment of bacteraemic patients for OPAT suitability is unlikely to have occurred. Although a useful adjuvant to a clinical referral system, more patients were probably not recruited due to severity of infection, as bacteraemic patients often require resuscitation in hospital. Also, many patients admitted from the community can be switched to oral alternatives and, in the absence of drug resistance, do not require prolonged intravenous therapy. Our data show that patients with community-onset and multidrug resistant (MDR) infections were most likely to be recruited to OPAT, and none of these patients were directly referred without prompting through the clinical referral system.

The most common sites of infection in patients recruited to OPAT were SSTIs and UTIs. A 2-year retrospective review of patients treated with OPAT in one Scottish centre found the majority of infections were SSTIs, 125 (59%) of 212 episodes.5 All were identified clinically and, in our study, we also found that these patients were predominantly recruited by clinical assessment rather than blood culture findings.

In contrast, the blood culture database was particularly useful in identifying patients with MDR UTIs, a condition that can't be diagnosed clinically, and with these infections a positive blood culture result triggered a clinical assessment which wouldn't have otherwise occurred. One paper retrospectively reviewed the use of OPAT to facilitate early discharge of patients with UTI caused by ESBL producing Enterobacteriaceae. In this small retrospective study, 11 patients with 25 treatment episodes, it is unclear whether any of these bacteraemic patients were identified and recruited following the validation of the blood culture result. The authors concluded that OPAT administration of ertapenem was effective and decreased costs associated with MDR UTIs,15 a finding similar to ours.

The most common bacteraemic isolate in our OPAT cohort was E. coli. Enhanced surveillance of E. coli bacteraemia has been mandatory for NHS acute trusts since June 2011.16 In England, the incidence of bacteraemia is 56/100 000 and ∼12% are ESBL producers or MDR. In 2015, 37 275 bacteraemic episodes were reported and approximately 3000 bacteraemic episodes were either ESBL producers or required intravenous antibiotics due to resistance to all suitable oral agents. If, based on our data, a third of bacteraemic episodes could be treated out of hospital (∼1000/annum) for an average of 5 days, then a possible 5000 hospital bed days could be saved in England each year.

Our blood culture database was also useful in identifying other medical conditions not normally treated with intravenous antibiotics out of hospital, including enteric fevers, biliary tract infections and central venous catheter-associated infections where line salvage was being attempted. In the absence of a blood culture result it is unlikely these patients would have been recruited to OPAT.

There were no deaths in the OPAT group with bacteraemia, despite bacteraemia being a marker of severe infection. Many of these patients were recruited after hospital admission and then stabilised before discharge on an intravenous antibiotic. In addition to reducing length of inpatient stay, our data demonstrate the safety of this approach. Among bacteraemic patients recruited to OPAT, all had community-onset infections. Hospital-acquired infections are generally medical device related or procedure related, and more commonly occur in patients in critical care areas,17 so it is not surprising that the majority of patients recruited had community-acquired bacteraemia.

There were limitations to this study. Before and during this study, we did not record information on the proportion of unsuitable patients referred to OPAT, nor did we have prestudy data on our OPAT mix including bacteraemic patients. Because of our proactive approach, it is possible patients were identified before clinical teams had time to refer (which, in part, is dependent upon awareness of OPAT services within our NHS trust). The numbers of bacteraemic patients recruited to OPAT were small and as there were no deaths in the ‘bacteraemic’ and ‘non-bacteraemic’ groups, a survival analysis was not possible. In logistic regression analysis for bacteraemic patients, we were unable to include place of acquisition in the model as no patients with hospital-acquired infection were recruited. In a larger study, where there were patients with hospital-acquired infection, we could have tested the association between community-onset bacteraemia and OPAT recruitment. The sample size was, however, big enough to demonstrate significant differences in unadjusted and adjusted data.

In summary, our study demonstrates that a blood culture database provided a useful adjuvant to a clinical referral system for OPAT recruitment. The blood culture database was particularly useful for prompting the identification of patients with community-onset infections and MDR UTIs caused by ESBL-producing E. coli. Bacteraemic patients received treatment safely and all had good clinical outcomes. Within the UK, each year, clinical assessment of all patients with ESBL-producing E. coli bacteraemia could identify hundreds of hospitalised patients suitable for OPAT recruitment, with the potential to reduce length of inpatient stay without compromising patient safety and clinical outcomes.

Main messages

  • As a blood culture database provides a useful adjuvant to a clinical referral system, an infection specialist (including a microbiologist) can prompt or make outpatient parenteral antibiotic treatment (OPAT) referrals.

  • This is especially useful for community-onset, extended-spectrum β-lactamase (ESBL)-producing or multidrug resistant Escherichia coli urinary tract infections (UTIs) that cause bacteraemia.

  • All bacteraemic patients recruited to OPAT received safe treatment and had good clinical outcomes.

Current research questions

  • Would a standardised comment, appended to extended-spectrum β-lactamase (ESBL)-producing Escherichia coli blood culture reports, increase outpatient parenteral antibiotic treatment (OPAT) recruitment and safely decrease length of inpatient stay?

  • What would be the national effect on length of inpatient stay if a standardised comment was automatically attached to ESBL-producing E. coli blood culture results in all UK NHS trusts?

  • Other than blood cultures, how useful are significant microbiology results in prompting assessment for OPAT recruitment?


The authors thank the specialist registrars at the Royal London Hospital, fast response team and Tower Hamlet community nurses who helped to collect this data.



  • Contributors MM collected and analysed data in conjunction with LMP. Statistical analysis was performed by CW.

  • Competing interests None declared.

  • Ethics approval The study was approved by the clinical governance committee at Barts Health NHS Trust.

  • Provenance and peer review Not commissioned; externally peer reviewed.