Background

Carbapenem-resistant Klebsiella pneumoniae (CRKP) is a global concern, leading to healthcare-associated infections worldwide. The most important mechanism of resistance is the production of a carbapenemase enzyme that has the ability to hydrolyze carbapenems [1]. Since its first description in North Carolina in 1996 [2], carbapenemase-producing K. pneumoniae (KPC) has been identified in 24 states across the US [3]. CRKP is currently seen all over the world [4–6]. There have been no described outbreaks with CRKP reported from Saudi Arabia or Gulf Cooperation Council (GCC) States.

Healthcare-associated infections (HAIs) caused by CRKP may be linked to increased mortality, length of hospitalization, and increased cost [7]. Data from the National Healthcare Safety Network (NHSN; Centers for Disease Control and Prevention [CDC], Atlanta GA, USA) showed that 8.7% of Klebsiella isolates causing HAIs in 2006–2007 were carbapenem-resistant compared to just less than 1% in 2000 [8, 9]. Since carbapenems are considered as an effective last line of defense in the treatment of severe infections caused by antimicrobial-resistant Enterobacteriaceae, CRKP represent a global threat to the control of HAIs, particularly in the presence of very limited therapeutic options [3]. Moreover, the difficulty of CRKP detection using routine antibiotic susceptibility testing adds to the complexity of the CRKP challenge in the healthcare setting [10]. The CDC recommends aggressive infection control measures to control the CRKP spread in acute care (inpatient) facilities that include early detection and active surveillance [9]. We sought to describe our experience in detecting and containing the first documented CRKP outbreak in Saudi Arabia.

Methods

Setting

The current study was conducted at King Abdulaziz Medical City (KAMC) in Riyadh, Saudi Arabia. KAMC is an approximately 900-bed tertiary care facility that provides healthcare services to about 600,000 Saudi National Guard soldiers, employees, and their families. The care provided ranges from primary and preventive care to tertiary care. KAMC has eight different intensive care units (ICUs) and four step-down units.

Study design

This a retrospective/prospective surveillance study describing the outbreak investigation and containment.

Surveillance and isolation methods

The outbreak was declared after reviewing the first three CRKP isolates at adult ICUs (between March 5th and 13th, 2010). Patients were then placed under strict contact isolation, along with intensified infection control measures and hand hygiene observation. The emerging cluster of CRKP patients were examined prospectively for microbiological and clinical data until discharge or death. The contacts of all clinical cases were screened, minimally through a peri-rectal swab; however, those who were ventilated were also screened through an endotracheal culture and placed under contact isolation until the culture results were available. Included in our study are patients with the first isolate of CRKP related to either a clinical infection or colonization. Patients who were positive by screening surveillance initially and then had invasive disease due to CRKP were reported according to the first positive specimen. The possibility of previously unnoticed CRKP cases was explored through a retrospective review of the microbiology database for any suspected cases of carbapenem resistance among the multiple drug-resistant K. pneumonia isolates over the previous 6 months. Commencing March 24th, weekly active surveillance cultures for CRKP were obtained from all ICU patients using rectal swabs and respiratory specimens. In addition, any new admission to the ICU was screened via peri-rectal swab and placed in contact isolation until the results were available. Weekly active surveillance was replaced by monthly active surveillance after declaring the end of the outbreak. Patients who became ready for transfer from the ICU were maintained on contact isolation in the receiving ward until the results became available. Those who were found to be colonized received weekly surveillance cultures until they became negative or until they were discharged.

Data collection

A standard data collection form was developed and discussed with infection control staff who were responsible for data collection. Index data was defined as the first date of CRKP isolation from surveillance or clinical cultures. Data collection involved using current and previous patient clinical records, microbiology laboratory records, and, sometimes, contact with the treating physician. Data collected included demographic data, location and duration of hospital stay, clinical presentation (clinical healthcare-associated infections or colonization), clinical severity score, comorbid medical conditions, current antimicrobial therapy, potential risk factors for multiple drug resistance, and outcome data (discharge or death) (Table 1 and 2). Potential risk factors at the index date were clearly defined (as shown in Table 3) before the start of data collection. Death was defined as death occurring during the current hospitalization (up to August 31), irrespective of hospital location at the time of death.

Table 1 Demographic and clinical characteristics of confirmed carbapenem-resistant Klebsiella pneumoniae (CRKP) cases at King Abdulaziz Medical City (Riyadh, Saudi Arabia) between September 2009 and August 2010
Full size table
Table 2 Demographic and clinical characteristics of patients with CRKP at King Abdulaziz Medical City (Riyadh, Saudi Arabia) between September 2009 and August 2010
Full size table
Table 3 Potential multiple drug-resistant organism (MDRO) risk factors among patients with CRKP at King Abdulaziz Medical City (Riyadh, Saudi Arabia) between September 2009 and August 2010
Full size table

Microbiological examination and clonal analysis

K. pneumoniae isolates were identified using the standard laboratory methods of the Clinical and Laboratory Standards Institute (CLSI) guidelines [11]. The identification of K. pneumoniae to the species level was done using the MicroScan WalkAway system (Siemens) and then confirmed using API 20E. Antimicrobial susceptibility testing was determined using the MicroScan WalkAway system (Siemens) and confirmed using the disk diffusion method and E-test (AB Biodisk). Minimum inhibitory concentration (MIC) breakpoints of carbapenems (meropenem and or imipenem) were defined according to the 2009 CLSI guidelines [11]. K. pneumoniae isolates found to have elevated MICs of carbapenems or with reduced disk diffusion zone sizes were tested for the presence of carbapenemases by use of the modified Hodge test (MHT) [12]. Carbapenemase production was detected by using meropenem disks; when the test isolate produced the enzyme and allowed the growth of a carbapenem-susceptible strain (E. coli ATCC 25922) towards a carbapenem disk, it was considered as positive. Pulsed field gel electrophoresis (PFGE) was done for the genotyping of K. pneumoniae isolates as previously described [13, 14].

Data analysis

The distribution of K. pneumoniae isolates by carbapenem resistance was plotted over time. Categorical data were presented as a percentage, while continuous data were presented as mean and standard deviation or median and inter-quartile range, as appropriate. SPSS software (release 18.0, SPSS Inc., Chicago, IL, USA) was used for the statistical analyses.

Results

In the year between September 2009 and August 2010, 20 (1.17%) out of 1,706 K. pneumoniae isolates detected at KAMC (Riyadh) were CRKP (Fig. 1). The epidemic curve showed a clear outbreak peak that did not resolve completely, creating an endemic pattern (Fig. 1). CRKP isolates were detected in samples from wound (40%), respiratory tract (25%), urine (15%), and other sites, including blood and the rectum (20%).

Fig. 1
figure 1

Number and distribution of confirmed carbapenem-resistant Klebsiella pneumonia (CRKP) cases at King Abdulaziz Medical City (Riyadh, Saudi Arabia)

Full size image

During March 2010, staff of the infection control and microbiology laboratory noticed the emergence of a cluster of CRKP isolates (n = 7) in mainly adult ICU patients. Patients with CRKP were put under strict contact isolation, along with appropriate infection control measures for staff and visitors. These include intensifying hand hygiene compliance, starting CRKP active surveillance, activating within hospital transfer precautions, enhancing concurrent and terminal cleaning of patients’ rooms, and staff education. Whenever feasible, respiratory therapists, nursing staff, and nursing aids were also cohorted during their shifts and on a rotating basis, to care exclusively for CRKP patients under contact isolation. Careful review of K. pneumoniae isolates over the previous 6 months revealed two previously unnoticed CRKPs; one in late February 2010 and another one in early December 2009. Admission and periodic active surveillance cultures showed a decrease but persistence of CRKP isolates in following months. Reviewing patients’ records during the outbreak period through May 2010 showed that all patients with CRKP but one had stayed for some time in adult ICUs before the index date (Fig. 2). All the study isolates were tested with the MHT and were positive (showing cloverleaf-like indentation). PFGE results were available for 15 (75%) out of 20 isolates. Review of the PFGE results during the outbreak period showed that the majority of outbreak strains were genetically indistinguishable or closely related (Fig. 3). Isolates 1, 6, 8, 9, 10, 14, and 18 are 100% identical and closely related to 16 and 17 (similarity 97.15%), with isolate 7 having 90.91% similarity. Isolates 5 and 15 are 100% identical and closely related to 20 (similarity 97.15%). Isolate 19 is 77% similar to 5, 15, and 20. Isolate 3 is a distant similarity of 67% with the other 14 isolates.

Fig. 2
figure 2

Location and duration of hospitalization and outcome of CRKP at King Abdulaziz Medical City (Riyadh, Saudi Arabia) between September 2009 and August 2010

Full size image
Fig. 3
figure 3

Dendrogram based on pulsed-field gel electrophoresis (PFGE) of DNA restriction fragments of representative CRKP outbreak isolates (n = 15). The dendrogram is shown on the left, with the percentage homology score indicated on top (0–100%). Actual pulsed field gel electrophoresis (PFGE) fingerprints are given on the immediate right of the dendrogram. Computer analysis of Xba I-digested isolates was performed using BioNumerics software (Applied Maths BVBA, Keistraat 120, 9830 Sint-Martens-Laten, Belgium)

Full size image

The majority of patients were initially admitted to emergency rooms (55%) and adult ICUs (30%). About 15% of the patients were transfers from other hospitals. The admitted patients were 55% females and 45% males. The majority of patients (60%) were aged 19 to 59 years (mean ± standard deviation [SD] 50.8 ± 21.8). Their primary admission diagnoses were trauma including motor vehicle accident (25%) and liver or kidney failure (25%). The median APACHE III score was 67. The main comorbid conditions were surgery (50%), hypertension (50%), diabetes (45%), trauma (30%), and gastrointestinal disease, liver disease, and heart disease (25% each). CRKP patients required, on average, 63 days of hospital stay, 27.5 of them before the index date. The majority (85%) of patients required ICU admission during their hospital stay, with a median of 29 days of ICU stay. Two-fifths (40%) of patients presented with clinical infection. Out of those patients who presented with clinical infection, 50% had wound infections and 25% had bloodstream infection.

Many of the potential risk factors for developing multiple drug-resistant organisms (MDROs) were observed in CRKP patients (Table 3). The majority (90%) of CRKP patients spent at least 5 days in the hospital (the majority of which were in the ICU) prior to the index date. The majority (85%) of CRKP patients received antimicrobial therapy (for at least 2 days) during the month prior to the index date. Carbapenem (60%), antianaerobics (50%), colistin (45%), fluoroquinolones (35%), and vancomycin (35%) were the frequently used medications by CRKP patients. Half (50%) of the CRKP patients had previous colonization/infection with MDROs during the month prior to the index date, with Acinetobacter and Pseudomonas being the most frequently encountered MDROs. In addition, the majority of CRKP patients had recent history of indwelling devices, surgical procedures, and mechanical ventilation.

In addition to complete resistance to carbapenems, CRKP isolates had 90–100% resistance to cephalosporins, fluoroquinolones, and aminoglycosides. They were still generally susceptible to colistin (79%) and glycylcyclines (67%). The most frequently used mediations at the index date were carbapenem (55%), colistin (50%), fluoroquinolones (40%), tigecycline (40%), and tazocin (25%). CRKP patients required, on average, 24 days of antimicrobial therapy. By the end of August 2010, 8 (40%) of the CRKP patients died, 6 (30%) were discharged home, and 6 (30%) were still hospitalized (Table 3). The majority of deaths (75%) happened during the outbreak period through May 2010 and only 3 (38%) of the deaths happened within 4 weeks from the index date. Death happened at a median of 32 days from the index date. The most frequent causes of death were septic shock (71%), liver or kidney failure (57%), and cardiopulmonary arrest (57%).

Discussion

In this paper, we described our experience in detecting and containing the first documented CRKP outbreak in Saudi Arabia. As per CDC guidelines [9], we retrospectively reviewed the microbiological and clinical records of all K. pneumoniae isolates 6 months prior to the outbreak, as well as started active surveillance cultures. Active surveillance was shown to be very effective in controlling CRKP outbreak [15]. To early detect new CRKP isolates and to limit within-hospital spread, we have decided to keep CRKP active surveillance cultures to the routine workup done for all new ICU admissions. Unlike the outbreak setting, the effectiveness of active surveillance to limit the spread of CRKP and other multidrug-resistant Enterobacteriaceae in the non-outbreak setting is controversial [16–18]. Considering the cost benefit, we stopped monthly active surveillance cultures 3 months after declaring the end of the outbreak.

As shown in other studies [17, 19], reinforcing infection control measures together with contact isolation of patients colonized or infected with CRKP were successful in controlling the CRKP outbreak in our hospital. Staff education focused on the resistant nature of the CRKP organism and the absolute importance of staff adherence to infection control recommendations. Hand hygiene was an essential element of these measures, and we were able to increase its compliance at adult ICUs from 79.5% in March 2010 to 85–95% during the following months (data not shown). Nevertheless, infection control staff had to deal with some cultural and logistic challenges. Convincing the patient and his/her visitors with standard infection control measures is not always easy, especially when the patient is colonized. Cohorting respiratory therapists, nursing staff, and nursing aids was not always possible due to logistic reasons. Keeping the same level of infection control measures compliance at weekends is another challenge.

CRKP in the current study had 90–100% resistance to most of the commonly used antimicrobial agents, including carbapenems, cephalosporins, and fluoroquinolones, leaving very limited therapeutic options to treat severe hospital infections. Our isolates were, however, still generally sensitive to colistin and, to a lesser extent, to glycylcyclines [20, 21]. Recent reports showing the emergence of colistin-resistant CRKP in some parts of the world are alarming [21–23]. Although it is nephrotoxic, colistin is considered, by far, to be the only therapeutic choice currently available, as glycylcyclines have relatively low peak serum concentration to treat certain severe hospital infections, such as bloodstream infections [24]. Judicious use of antimicrobial agents may be seen as a long-term measure to reduce CRKP and other Gram-negative MDROs. The establishment of a comprehensive antimicrobial stewardship program, on the other hand, is a project worthy of leadership attention and support. As a consequence of CRKP outbreak, we initiated an antimicrobial use and resistance database in our ICUs to monitor carbapenem and other broad-spectrum antimicrobial use.

Quick and adequate detection of CRKP is critical for the success of infection control measures. The detection of CRKP-producing bacteria based only on susceptibility testing is not easy, due to the heterogeneous expression of β-lactam resistance [10]. As per CDC recommendations [9], we used the MHT to confirm suspicious isolates based on the MICs breakpoints of the 2009 CLSI guidelines. Since the test is time-consuming and required reading expertise, securing results at the right time from an infection control point of view was sometimes challenging. The 2010 CLSI guidelines have overcome this problem by lowering the MICs for carbapenems, eliminating the need to perform the confirmatory MHT. The down side to this, however, is the large number of isolates that would mandate a larger number of patients to be placed on contact isolation according to an MDRO isolation policy. Therefore, we decided to continue to use the 2009 CLSI guidelines with all suspicious isolates to be reviewed by the director of the microbiology laboratory in order to decide on the need for the MHT.

The reported mortality in patients with CRKP is striking. It ranged between 40% and 70% [7, 25, 26]. Mortality in the current study was 38%. This crude mortality is probably overestimating the impact of CRKP and is confounded by multiple comorbidities. Only 38% of our CRKPs presented as HAI. It is estimated that about one-third to one-half of crude mortality associated with CRKP were actually attributable to CRKP infection [25, 26]. The high frequency of potential risk factors of MDROs in our CRKP patients is in agreement with previous reports describing CRKP risk factors, including prolonged hospital stay, prior hospitalization, prior antimicrobial intake, current stay in the ICU, recent surgery, and the use of a ventilator or indwelling device [7, 26, 27].

Conclusion

Reinforcing infection control measures together with contact isolation of patients colonized or infected with carbapenem-resistant Klebsiella pneumoniae (CRKP) were successful in controlling the CRKP outbreak, but did not prevent endemicity. Since CRKP has limited therapeutic options and is associated with high mortality, we have decided to focus on limiting the spread by continuing to carry out CRKP active surveillance cultures for all new intensive care unit (ICU) admissions. The need for new effective antibiotics against pan-resistant, Gram-negative, multiple drug-resistant organisms (MDRO) is now more critical than ever before.