Article Text

A managed multidisciplinary programme on multi-resistant Klebsiella pneumoniae in a Danish university hospital
  1. Stig Ejdrup Andersen1,
  2. Jenny Dahl Knudsen2,3,
  3. for the Bispebjerg Intervention Group
  1. 1Department of Clinical Pharmacology, Copenhagen University Hospital, Bispebjerg, Copenhagen NV, Denmark
  2. 2Department of Clinical Microbiology, Copenhagen University Hospital, Hvidovre, Denmark
  3. 3Infection Control Organisation, Copenhagen University Hospital, Bispebjerg, Copenhagen NV, Denmark
  1. Correspondence to Dr Stig Ejdrup Andersen, Department of Clinical Pharmacology, Copenhagen University Hospital, Bispebjerg, Bispebjerg Bakke 23, Copenhagen NV DK-2400, Denmark; sand0011{at}


Background Bacteria-producing extended spectrum β-lactamase (ESBL) enzymes are resistant to commonly used antimicrobials. In 2008, routine monitoring revealed a clonal hospital outbreak of ESBL-producing Klebsiella pneumoniae (ESBL-KP).

Methods At a 510-bed Danish university hospital, a multidisciplinary change project inspired by Kotter's Eight Steps of Change was designed. In addition to revision of antimicrobial guidelines and restriction of selected antimicrobials, the complex, managed, multi-faceted intervention comprised training and education, enhanced isolation precautions, and a series of actions to improve the infection control measures and standardise procedures across the hospital. A prospective interrupted time series design was used to analyse data collected at hospital level from January 2008 through December 2011.

Results Though overall antimicrobial consumption remained unaffected, the intervention led to intended, immediate and sustained reduction in the use of cefuroxime, and an increase in the use of ertapenem, piperacillin/tazobactam and β-lactamase sensitive penicillin. Moreover, a postintervention reduction in the rate of ESBL-KP in diagnostic samples and in the incidence of ESBL-KP infections was observed. The intervention may also have reduced the need for isolation precautions and may have shortened each isolation period.

Conclusions The results indicate that an immediate and sustained change in the antimicrobial consumption and a decreasing rate of ESBL-KP are achievable through the application of a managed, multi-faceted intervention that does not require ongoing antibiotic stewardship.

  • Antibiotic management
  • Infection control
  • Leadership
  • Quality improvement

Statistics from


Globally, antimicrobial resistance is an increasing challenge1–4 and an important example is extended spectrum β-lactamase-producing Klebsiella pneumonia (ESBL-KP), which is mostly involved in nosocomial infections.5 Resistant to cephalosporins and other commonly used antimicrobials, infections with these bacteria are often not covered by empiric antimicrobial treatments. Moreover, ESBL-producing bacteria cause increased morbidity and mortality, and increase the hospital costs.5–11 Lately, an increasing number of Danish hospital outbreaks of ESBL-KP have occurred12 and the CTX-M-15 ESBL enzyme is now a frequent cause of resistance to third generation cephalosporins.12 ,13

For decades now, efforts have been made to improve the use of antimicrobials, often in the framework of an antimicrobial stewardship programme (ASP).4 ,14 Although not well defined, ASP often refers to a programme to optimise antimicrobial use within a hospital and may employ many different strategies.14–16 ASP may focus on patient outcome, toxicity and cost, but mitigating antimicrobial resistance could also be an objective. Although transmission of ESBL-producing bacteria occurs by direct and indirect patient-to-patient contact,17 antimicrobial exposure is a predictor of the selection of ESBL-producing bacteria18 and, therefore, ASP comprising antimicrobial restriction may decrease the prevalence of these microorganisms.2 ,19

Managing an ASP that potentially interferes with the relationships between diverse departments, however, can be complex and challenging to hospital leadership17 ,20 and the aim of this paper is to describe an approach that did not require ongoing antimicrobial stewardship but used other interventions to control antimicrobial use and improve infection control to prevent spread of ESBL-KP.

Assessment of the problem

Bispebjerg Hospital is a 510-bed teaching hospital with a catchment population of approximately 400 000 people. In 2010, the hospital had 100 900 outpatient contacts and 42 300 admissions to 12 clinical departments. Having no department of clinical microbiology (DCM) or microbiology lab, a university hospital situated 10 km away provides clinical microbiology service including blood cultures and other lab services.

In 2008, the DCM revealed a rapidly increasing number of diagnostic samples with ESBL-KP resistant to cefuroxime. Consequently, the efforts in infection control and surveillance were increased and hypochlorite cleaning was re-introduced.

The need for new first-line antimicrobial regimens became evident when the Drug and Therapeutic Committee subsequently unveiled a cefuroxime consumption of approximately 200 defined daily doses per 1000 occupied bed-days (1000 OBD). Thus, a first project draft of a strategy to optimise the use of antimicrobials was compiled and agreed with the hospital vice-directors.

The next step was involvement of the quality department (comprising nurses, physiotherapist and academics) which arranged a workshop on ESBL for all members of the clinical departments’ multidisciplinary quality groups. Potential contributing problems and barriers were identified by brainstorming and refined by discussion. For example, it was recognised that inhospital transmission of antimicrobial resistance also needed to be addressed. These proposals were incorporated into the final project plan.

A project organisation was set up comprising a steering committee, a coordination group and four project groups: (1) work processes and isolation; (2) internal communication and guidelines; (3) diagnostics and antimicrobials; and (4) cleaning and infection control. A clinical pharmacologist, a clinical microbiologist, clinicians, the infection control team and staff members from the quality department joined the project groups.

Strategy for change

Having created urgency and formed the project organisation, we developed a changing process for the hospital inspired by the next steps of Kotter's framework for change21 (create a vision for change, communicate the vision, remove obstacles and create short term wins). We emphasised a clear vision of the intended achievements and aimed at explaining each subintervention to those influenced by it. Moreover, we attempted to support and encourage front-line staff to adopt appropriate procedures, and to soften barriers22–24 that could hinder the intended development.

The steering committee created a set of empirical goals for the project. The verbalised expectations were that if a 95% and 50% reduction in the consumption of cephalosporins and quinolones could be achieved, a 50% reduction of nosocomial ESBL infections could be expected. A reduced need for isolation precautions and hypochlorite cleaning should also be achieved, which is important to the nursing staff.

A complex, stepwise intervention was then developed. Having no final blueprint at the beginning, the intervention evolved gradually in response to new challenges and, throughout the project period, the project groups met regularly to exchange information, promote coordination, monitor the project, solve emerging problems and discuss the need for corrective actions.

The plan was to work simultaneously on different levels using actions aimed at leaders, clinicians and other groups of staff, and to manage these activities we used selected project management tools such as critical pathway analysis for identifying related and interdependent activities, for example, tasks that had to be started or completed before others. Gantt charts with break points were used for identifying all elements of the intervention, defining tasks and responsibilities, and for scheduling the project activities. Stakeholder analysis was used to identify people in and outside the hospital that could support or impede the project.

The intervention

Table 1 shows the elements of the intervention (for chronology, please see online supplementary appendix). To take into account the limited resources available for the project and be able to tailor the intervention to the different specialities, a series of scenarios were set up in collaboration with a pharmacist and the economical consequences were estimated.

Table 1

Overview of the intervention

Cephalosporin restriction may lower the incidence of ESBL infections.19 ,25–29 Moreover, carbapenems or β-lactam/β-lactamase inhibitor combinations are the recommended initial treatment against serious infections with ESBL-producing bacteria.3 ,5 ,6 But because carbapenems and piperacillin/tazobactam combinations were highly expensive, it was decided to restrict cephalosporins in general but still recommend ceftriaxone plus ampicillin for meningitis, and cefuroxime for surgical prophylaxis. Piperacilline/tazobactam was recommended for empiric treatment of sepsis, pneumonia or fever of unknown origin whereas ertapenem was recommended when penicillin allergy was suspected or documented. Finally, meropenem was recommended for severe sepsis/septic shock, or whenever required for patients with diagnosed infections.

To ensure timely standardisation across the clinical departments, all local antimicrobial guidelines were collected and revised before the new recommendations were put into force by 18 January 2010.

No preauthorisation to receive an antimicrobial was introduced, but the pharmacy technicians were instructed that non-surgery wards should no longer store cefuroxime unless specifically requested.

Awareness of antimicrobial resistance plays an important role in controlling antimicrobials30 but varies among the physicians.31 Thus, we developed a mandatory brush up programme on rational use of antimicrobials and arranged several training sessions with the clinicians on rational use of antimicrobials.

Although the use of traditional contact precaution measures was well established, the discussion at the multidisciplinary workshop revealed that the isolation-precaution policy and coordination with sections that provide transverse services needed to be improved. The new measures came to include a self-adhesive sign for the bedpost, a leaflet for the front page of the patient files, new ordering procedures with prompt questions about the patient's isolation requirements, and rapid involvement of the infection control nurses who started visiting the wards the same day a new ESBL case had been detected.

A communication plan was also prepared and at several meetings, staff and ward managers were informed about the intervention and the intended achievements. Moreover, newsletters and diagrams of the resistances rates and, later, cumulative sum control charts (CUSUM charts) of the bacterial resistance were published regularly on the hospital intranet and distributed to all heads of departments as feedback. The CUSUM charts were chosen because they are easy to interpret and have a high sensitivity to detect smaller changes.32 The direct impact of these measures was not measured.

Finally, to produce data that could sustain the involvement of the leaders and maintain the sense of urgency, two pilot studies were carried out in three departments revealing an inpatient ESBL carrier rate between 3% and 31%.

Effect of change

Study design

We used a prospective interrupted time series design to analyse data at the hospital level from January 2008 through December 2011. The hospital pharmacy provided data on antimicrobials (Antatomical-Therapeutic-Chemical code J01 ( To adjust for variation in the activity, the monthly drug consumption in defined daily doses ( was divided by the OBD. Since the intervention was fully implemented on 18 January 2010, data from this month were omitted from the analyses leaving 24 points before and 23 points after for the analyses.

The magnitude of the intervention on antimicrobial consumption was assessed with before-and-after analyses and segmented regression analyses of interrupted time series data.33 For the latter, the break point was set to January 2010. Including no lag time, the analyses were controlled for seasonality as described by Högberg et al34 and the models were fitted using least squares regression. Removing the least significant parameters first, all non-significant autoregressive parameters were removed from the models. While a significant change in level is indicative of a short-term effect, a significant change in slope indicates a long-term effect. Data on hypochlorite cleaning from January 2009 through December 2011 were analysed in a similar way.

The DCM provided data on K pneumonia in diagnostic samples from inpatients. A failure chart was constructed by plotting the cumulative rates of samples of ESBL (failures) against the total number of samples of K pneumoniae. A 100% ESBL rate would produce a 45° upward slope, while no ESBL would produce a horizontal line. In 2008, the accepted failure rate was set to 18%. The unaccepted failure rate was set to 25%. An upper boundary ‘alarm’ line was added using type 1 and 2 error rates of 0.05 and 0.2, respectively.

Statistical process control charts (u-charts) with 3 σ-control limits were constructed to report the incidence of ESBL-KP and non-ESBL-KP infections as a secondary outcome.

Data on the use of isolation precautions were available from January 2010 through May 2011. Comparing the periods from July through December 2010 and January through May 2011 with the period January through June 2010, we used repeated-measures MANOVA to assess the effect on the number of isolated patients, the total number of isolation days and the length of each isolation period. Data were divided into ESBL and non-ESBL groups and if there was a group by time interaction, the simple effect of time was analysed in each group.

All analyses used a two-tailed α level of 0.05 and were performed using JMP 9.0 with SAS V.9.1 add-in (AUTOREG) for JMP (SAS Institute Inc., Cary, North Carolina, USA).

Antimicrobial consumption

Table 2 summarises the antimicrobial consumption and figure 1 shows the consumption of cefuroxime, ciprofloxacin, ertapenem and piperacillin/tazobactam over the study period. Although the overall consumption remained unaffected by the intervention, the regression analyses revealed significant changes, which were most pronounced for cefuroxime, ertapenem, piperacillin/tazobactam, and β-lactamase sensitive penicillin. Thus, the cefuroxime consumption dwindled by 74.5%. A statistically insignificant reduction of 8.9% (p=0.70) in ciprofloxacin was also observed. Only the cephalosporin (first and third generation), ertapenem, other flouroqinolones and piperacillin/tazobactam postintervention slopes differed slightly but statistically significant from zero.

Table 2

Monthly antimicrobial consumption (defined daily doses/1000 occupied bed-days)

Figure 1

The proportion of the overall antimicrobial consumption in defined daily doses that are made up by cefuroxime, ciprofloxacin, ertapenem and piperacillin/tazobactam over the 48-month study period.

Resistance surveillance

Figure 2 shows a cumulative failure chart for ESBL production in diagnostic samples of K pneumoniae from inpatients. From 2008 to 2011, the rate of isolated ESBL-KP decreased from 39.5% to 22.5%.

Figure 2

Cumulative failure chart constructed by plotting the yearly cumulative rates of samples of extended spectrum β-lactamase (ESBL)-producing Klebsiella pneumoniae (failures) against the total number of samples of K pneumoniae. On average, 35.5% of all isolated K pneumoniae in 2008 were ESBL-producing, and since we aimed at a 50% reduction, 18% was chosen as the accepted failure rate (grey line).

In contrast to the common cause pattern of the non-ESBL-KP infections (figure 3), the incidence of infections with ESBL-KP shows a special cause pattern (non-random variation) indicative of a decrease over the study period.

Figure 3

Control charts (u-charts) with 3 σ-limits plotting the incidence of non-extended spectrum β-lactamase (ESBL) Klebsiella pneumonia (upper panel) and ESBL K pneumonia infections (lower panel). The latter shows a special cause pattern (non-random) indicative of a shift to a lower level.

Hypochlorite cleaning

The average number of hypochlorite cleaning episodes of 32.0 per 1000 OBD (95% CI 27.7 to 36.2) was unaffected by the intervention (p=0.34 for level and 0.84 for slope).

Use of isolation precautions

The number of isolated patients per 1000 OBD declined from 0.94 (95% CI 0.74 to 1.14) to 0.65 (95% CI 0.43 to 0.87), p=0.021, for ESBL and did not change for non-ESBL causes (6.8 (95% CI 5.6 to 8.0) patients per 1000 OBD). The number of isolation days per 1000 OBD decreased from 13.8 (95% CI 8.6 to 19.0) to 7.1 (95% CI 3.4 to 10.8) for ESBL, and from 42.8 (95% CI 30.8 to 54.7) to 28.6 (95% CI 22.0 to 35.3) for non-ESBL, p=0.0032. No group by time interaction was detected indicating similar changes in the two groups. The length of each isolation period decreased from 14.6 (95% CI 9.8 to 19.3) to 10.6 (95% CI 7.9 to 13.2) days for ESBL, and from 6.3 (95% CI 5.4 to 7.1) to 5.2 (95% CI 4.3 to 6.0) days for non-ESBL p=0.021 with no group by time interaction.

Lessons and messages

In this project, we implemented change at the institutional level by the use of a pragmatic ASP approach. The intervention led to immediate and sustained changes in the antimicrobial consumption and a decreasing rate of ESBL-KP in diagnostic samples and a lower incidence of ESBL-KP infections. Two years postintervention, no apparent compensatory increase in other antimicrobials in response to the cephalosporin restriction was detected. Moreover, the intervention may have reduced the use of isolation precautions.

The project involved all clinical departments and from the beginning the hospital vice-directors and the Quality Board were involved. This support and endorsement were central to the project since the hospital leadership has the ultimate responsibility for dedicating the necessary resources, demonstrating commitment and propagating the ideas throughout the organisation.14 ,20 ,24 ,35

Change involves learning and unlearning well-integrated skills and procedures and one of the important concerns was to soften barriers that could hinder the intended development. These barriers include predisposing factors (eg, knowledge), enabling factors (eg, availability of services) and reinforcing factors (eg, opinion of others).36 Strategies taking into account all levels seem the most successful23 and the desired change may have occurred due to the application of a multi-factorial intervention that targeted patients, professionals, team leaders and the entire organisation. Still, several factors may facilitate or hinder change and knowledge of which are most decisive is still lacking.23

Gaining acceptance of an ASP is difficult, especially if there is perceived loss of autonomy.35 In this project, support of front-line staff was a major issue. Front-line staff might be aware of the problems they encounter, but also unaware of their own inappropriate actions and reluctant to speak up about errors.37 Nevertheless, improving cross-compartmental collaboration relies on cooperative front-line staff and their willingness to speak up about problems in processes that go across inter-connected departments. These problems may otherwise be hard to identify since present in one setting, they may have negative consequences elsewhere even with delay. Furthermore, the responsibility of commencing antimicrobial prescribing remains primarily with the front-line junior doctors who are also the most mobile workforce. That is why ASPs have been criticised for not encouraging the nurses to take a more practical role to support rational antimicrobial prescribing and best infection control practice.38 Besides involving nurses directly in the project groups, our response to these challenges was the multidisciplinary workshop at the beginning of the process where the different staff groups were encouraged to discuss the practical problems they faced. In addition to revealing process defects important to the project planning, this approach proved an important collective learning opportunity and may have facilitated collaboration and system thinking.39

Changing established processes require cooperation of the right people and coordination of the activities, and must be organised and be able to provide a leading. Still, involving an entire hospital in large-scale change also means risk of unpredictability. Despite planning and directing the process, our intervention evoked unforeseen responses. One example occurred when nurses, who are not allowed to prescribe, challenged the new antimicrobials regimens, arguing that piperacillin/tazobactam required time-consuming preparation and was therefore an inappropriate choice. Involvement of the pharmacist and purchase of a more easy to handle drug container solved the problem. As such, the project developed into a series of directed subinterventions in response to feedback; a dynamic interaction that is perhaps better described by Leavitt's model proposing that whenever change is imposed on one of the four interdependent variables in an organisation, structure, technology, people and task, one or more of the other variables will also be impacted.40

Comparison with the literature

Simultaneous employment of multiple different interventions makes it difficult to compare the effect of our approach with the literature. Furthermore, the nature of antibiotic resistance is multi-factorial41 and a wide range of factors may contribute to successful infection control such as IT capabilities, culture, communication and coordination.42 Excluding traditional ASP core strategies such as formulary restriction and audit with feedback,15 our approach comprised strategies that are often considered supplementary.35

Though several strategies to control resistant bacteria have been developed and a number of organisational frameworks have been proposed, no universal means seem to exist.14 ,15 ,35 ,38 ,41 ASP is increasingly recognised for its ability to sustain benefits over time,30 ,43 but this study indicates that a managed change project can produce similar results. Thus, we accomplished an instant and sustained change in the consumption of cefuroxime, piperacillin/tazobactam and β-lactamase sensitive penicillin which might be an important step towards lessening the burden of ESBL-producing K pneumonia.25 ,27 ,28 ,43 Thus, several studies have reported that cephalosporin restriction is associated with a reduction in ESBL-producing Enterobacteriacea.19 ,28 ,43

Cohort isolation is useful against ESBL-producing bacteria44 and the present study comprised several measures to improve the isolation precaution. Data from the first part the intervention period indicate that the length of isolation may have declined for all patients but most pronounced for patients with ESBL-infections who were isolated for approximately 15 days. Although only indicative, these results are important since isolation puts patients under pressure and coping with the isolation standards are cumbersome to the nursing staff. Thus, less use of isolation precautions appeared a very relevant goal for the nurses, and might in this quality have been an important incentive.

ESBL infections are associated with longer and more costly hospital stays11 ,18 ,45 and infection related length of stay seems to be the main driver of costs.11 Although we report no cost data, a reduction in the length of each isolation period is an important result that may have a significant impact on the hospital costs.

Strengths and limitations of the study

The strengths of the study were prospective collection of standardised objective data covering all phases of the intervention, collection of detailed antimicrobial procurement data and an appropriate analysis to identify any changes. Interrupted time series is considered the method of choice for the analysis of before-and-after quasi-experimental study designs14 and to evaluate the impact of interventions to change prescribing.33 ,46

An important limitation to the study is the non-randomised and unblinded design. Furthermore, we relied on historical controls influenced by unknown confounders. Though randomised controlled studies are preferable, cross-contamination and other issues make it difficult to conduct a blinded, randomised study within a single hospital. Other limitations include simultaneous employment of multiple strategies making it impossible to conclude which is most effective, and the use of data on bacteria resistance and infections instead of the more important data on mortality. Finally, another important potential drawback needs attention. In response to the new regimens, resistance to another class of antimicrobials may occur (‘squeezing the balloon’ phenomenon28), but this is under constant surveillance and has not emerged yet.


We describe an alternative approach to contain a problem with resistant bacteria that does not require ongoing antimicrobial stewardship. Within the framework of a change project, we developed and implemented procedures that resulted in immediate and sustained changes in the antimicrobial consumption and a decreasing rate of ESBL-KP. Additionally, the use of isolation precautions tended to decline in the project period.

Our experiences may be useful to other hospitals challenged by resistant bacteria and the project illustrates the value of establishing a close relationship between the quality organisation and the different boards of specialist for identifying and responding adequately to a highly complex quality problem. A similar model in which external departments of microbiology collaborate with local hospitals around specific infection control targets might be of use to other settings, especially if they cannot have permanent antibiotic stewardship. But it is important to emphasise that the project required a lot of development and it is hardly likely that adoption of our complex intervention as a whole is appropriate. Other settings and other infection control targets such as MRSA might require interventions that look different. Still, this project demonstrates how to create an environment where such interventions can be developed, change actually can take place and sustained improvements can be accomplished.

Increased awareness and improved infection control measures may have prepared our hospital better for the next resistant bacteria. With the emergence of carbapenemase-producing Gram negative bacteria,47 which are serious threats, both to the patients and to the health system, this may soon prove useful. The problem with resistant bacteria has no correct, let alone final solution.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:


  • Collaborators The Bispebjerg Intervention Group: Stig Ejdrup Andersen (Department of Clinical Pharmacology, Copenhagen University Hospital, Bispebjerg. Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark); Jenny Dahl Knudsen (Department of Clinical Microbiology, Copenhagen University Hospital, Hvidovre. Kettegaard Alle 30, DK-2650 Hvidovre, Denmark, Infection Control Organisation, Copenhagen University Hospital, Bispebjerg. Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark); Marie Stangerup (Department of Clinical Microbiology, Copenhagen University Hospital, Hvidovre. Kettegaard Alle 30, DK-2650 Hvidovre, Denmark); Jeannette Havstreym (Department of Clinical Microbiology, Copenhagen University Hospital, Hvidovre. Kettegaard Alle 30, DK-2650 Hvidovre, Denmark); Dorrit Langsted Olsen (Department of Clinical Microbiology, Copenhagen University Hospital, Hvidovre. Kettegaard Alle 30, DK-2650 Hvidovre, Denmark); Steen Werner Hansen (Board of Directors, Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark); Mette Christensen (Board of Directors, Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark); and Janne Elsborg (Board of Directors, Bispebjerg Bakke 23, DK-2400 Copenhagen NV, Denmark).

  • Contributors Both the authors contributed to the conception and design; acquisition, analysis and interpretation of data; drafting, critically revision and final approval of the submitted manuscript.

  • Competing interests None.

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

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.

Linked Articles