Rapid control of a hospital‐wide outbreak caused by extensively drug‐resistant OXA‐72‐producing Acinetobacter baumannii

Abstract

Extensively drug‐resistant Acinetobacter baumannii (XDRAb) emerges as an important pathogen of health care–associated infections and outbreaks worldwide. During January and February 2006, there was a hospital‐wide outbreak of XDRAb at a medical center in Taiwan. Without limiting the usage of carbapenems or the closure of any ward, this outbreak was effectively controlled. We investigated the molecular epidemiology and reported the infection control experiences. XDRAb is defined as A baumannii that is resistant to multiple antibiotics but susceptible to tigecycline and polymyxin B. During the outbreak, the clinical and environmental XDRAb isolates were collected and studied by antimicrobial susceptibility testing, pulsed‐field gel electrophoresis, and polymerase chain reaction for Verona integron‐encoded metallo‐beta‐lactamases, imipenemases, and oxacillinases (OXA). Our measures to control the outbreak included private room isolation of patients until there were three successive negative cultures, reinforcement of contact precautions, daily environmental cleansing with room‐dedicated cleaning tools and sodium hypochlorite, and careful auditing of adherence. During the outbreak, 32 clinical XDRAb isolates came from 13 patients who were hospitalized in four intensive care units and three wards. Most (7 of 13, 53.8%) cases were associated with a surgical intensive care unit. The results from pulsed‐field gel electrophoresis study indicated that all isolates were of one genotype. All 32 isolates harbored ISAba1‐bla _{OxA‐51‐like} and bla _OxA‐72 genes. After this outbreak till August 2010, further incidences of XDRAb were sporadic cases of XDRAb with different clones and did not reach the level of outbreak. To our knowledge, this is the first reported hospital‐wide outbreak caused by OXA‐72 carbapenemase–producing A baumannii in the Asia‐Pacific region, with successful and sustained control. Although the source or vehicle of the outbreak was not identified, our results suggest that a hospital‐wide outbreak can be successfully managed with strict infection control measures, and that the limitation of the use of carbapenems and closure of wards may not be necessary.

1. Introduction

Acinetobacter baumannii exists widely in the hospital environment [1], and it can cause serious or fatal illnesses, particularly in immunocompromised patients. Multiple drug‐resistant A baumannii, which is only susceptible to a limited number of antibiotics, has also been increasingly reported worldwide [2]. The emergence of extensively drug‐resistant A baumannii (XDRAb)—defined as A baumannii that is resistant to multiple antibiotics but susceptible to tigecycline and polymyxin B—has recently become a major problem worldwide [3]. When a nosocomial outbreak of XDRAb occurs, the health and lives of hospitalized patients are seriously threatened [[4], [5]]. In Taiwan, there have been outbreaks of multidrug‐resistant A baumannii (MDRAb) that is resistant to more than three classes of antimicrobial agents [[6], [7], [8], [9]]. However, simple and effective infection control measures without the closure of wards or hospitals are preferred to limit the spread of XDRAb in hospital settings.

It could take several months to control XDRAb outbreaks, even with the implementation of infection control measures [[10], [11], [12]]. Infection control measures usually include closure of intensive care units (ICUs) or wards, especially when the source or reservoir is not found [[11], [12], [13], [14], [15]]. Here, we report the first documented hospital‐wide outbreak of XDRAb carrying oxacillinase (OXA)‐72 carbapenemase in the Asia‐Pacific region, focusing on its infection control measures and molecular epidemiology.

2. Materials and methods

2.1. Hospital setting and patient characteristics

Kaohsiung Medical University Hospital is a 1,600‐bed tertiary medical center in southern Taiwan. There are 10 ICUs, one respiratory care center (RCC), and 37 wards. Clinical data, including patient age, gender, days in the hospital, days in the ICU, site(s) of infection, time from admission to XDRAb acquisition, medical comorbidities, and major risk factors (e.g. drain tubes, mechanical ventilation) were recorded for each patient from whom XDRAb was isolated during January and February 2006 (Table 1). Acute Physiology and Chronic Health Evaluation II (APACHE II) scores were calculated at the time of XDRAb isolation. XDRAb was considered to be the cause of pneumonia when there was clinical and radiological evidence of lower respiratory tract involvement [16]. XDRAb‐attributable mortality was determined by subtracting the crude mortality rate of patients with colonization from the crude mortality rate of patients with infection [17]. We defined colonization as the presence of bacteria without clinically relevant signs of illness and infection as the presence of bacteria with clinically relevant signs of illness. The study was performed after approval by the hospital's Institutional Review Board.

Table 1.

Demographic and clinical characteristics of 13 patients with extensively drug‐resistant Acinetobacter baumannii infection or colonization from January to February 2006

Patient no.	Age (yr)/sex	Ward when culture collected	Length of stay (days, before culture/total)	Length of ICU stay (days, before culture/total)	Ventilator days (before culture/total)	APACHE II when culture collected	TGC MIC (μg/mL)	Polymyxcin B MIC (μg/mL)	Status/infection	Outcome
1	75/F	W1	33/64	20/33	11/43	22	3	1	CVC tip infection	Survived
2	86/M	CICU	2/17	2/17	N	20	3	1	Sputum colonization	Survived
3	73/M	SICU	38/65	18/21	37/64	21	3	1	Pneumonia	Died
4	65/F	RCC	50/91	36/41	33/75	16	3	1	Sputum colonization	Survived
5	73/M	SICU	41/53	24/33	22/34	22	4	0.5	Sputum colonization	Died
6	65/M	W2	82/144	19/35	8/8	25	3	0.75	Sputum colonization	Improved
7	76/M	W3	35/68	N	N	21	4	1	Sputum colonization	Died
8	46/M	SICU	42/50	42/50	24/33	21	6	0.5	Wound infection	Died
9	74/M	SICU	41/49	25/33	19/28	17	3	1	Sputum colonization	Died
10	99/F	SICU	8/32	6/29	0/15	36	3	1	Pneumonia	Died
11	63/F	MCU1	54/62	10/18	9/11	30	1.5	1	CVC bacteremia	Died
12	73/M	RCC	21/115	20/20	21/115	20	3	1	Sputum colonization	Survived
13	58/F	NICU	48/100	34/60	6/59	23	3	1	Sputum colonization	Died

APACHE  =  Acute Physiology and Chronic Health Evaluation; CICU = cardiovascular intensive care unit; CVC  =  central venous catheter; F  =  female; ICU  =  intensive care unit; M  =  male; MIC  =  minimal inhibition concentrations; N  =  nil; NICU  =  neurological intensive care unit; RCC  =  respiratory care center; SICU  =  surgical intensive care unit; TGC  =  tigecycline.

2.2. Bacterial isolates

Clinical XDRAb isolates from all 13 hospitalized patients and environmental isolates from surveillance cultures of these patients' immediate environments (including bedrails, monitors, respirators, bedside desks, and bedside sinks) were collected. The hands of personnel taking care of these patients were randomly sampled with a cotton swab moistened with brain heart infusion (BHI) broth during the work time before washing hands [1]. Bacterial identification was performed by conventional methods, using the API 20NE system (bioMérieux, Marcy l'Etoile, France) and recA gene sequencing [18]. Acinetobacter baumannii ATCC 19606 was used as a control strain.

2.3. Antimicrobial susceptibility testing

We used the disk diffusion method, as described by the Clinical and Laboratory Standards Institute [19], to determine the antibiotic susceptibility for the following antibiotics: amikacin, amoxicillin/clavulanic acid, ampicillin/sulbactam, piperacillin, piperacillin/tazobactam, ticarcillin/clavulanic acid, aztreonam, cefotaxime, ceftazidime, ceftriaxone, cefepime, imipenem, meropenem, chloramphenicol, ciprofloxacin, levofloxacin, ofloxacin, gentamicin, tobramycin, and trimethoprim/sulfamethoxazole. The minimal inhibition concentrations (MICs) of tigecycline and polymyxin B were determined by the Etest (AB Biodisk, Solna, Sweden).

2.4. Molecular typing with pulsed‐field gel electrophoresis

Pulsed‐field gel electrophoresis (PFGE) was performed as previously described [20]. ApaI (New England Biolabs, Beverly, MA, USA)–restricted fragments were separated by PFGE in 1% SeaKem Gold agarose gels (Cambres Bio Science, Rockland, ME, USA) using a Bio‐Rad CHEF‐Mapper apparatus (Bio‐Rad Laboratories, Richmond, CA, USA). Band patterns were compared and classified following the criteria of Tenover et al. [21].

2.5. Polymerase chain reaction amplification and gene sequencing

Isolates were grown overnight on BHI agar plates at 37°C, and lysis was achieved by heating for 10 minutes at 95°C. The lysed cells were used on the day of preparation without exception. Polymerase chain reaction (PCR) testing of the clinical isolates for carbapenemase‐encoding genes was performed using primers for OXA, imipenemase (IMP), and Verona integron‐encoded metallo‐beta‐lactamase (VIM) carbapenemase groups [[22], [23], [24], [25]]. For OXA and VIM, the cycle conditions were as follows: initial denaturation at 95°C for 5 minutes; 30 cycles of amplification steps at 95°C for 30 seconds, 52°C for 30 seconds, 72°C for 45 seconds; and a final extension at 72°C for 7 minutes. For IMP, the cycle conditions were as follows: initial denaturation at 95°C for 5 minutes; 30 cycles of amplification steps at 95°C for 30 seconds, 50°C for 30 seconds, 72°C for 1 minute; and a final extension at 72°C for 7 minutes. Primers P1 and P2 were used for sequence analysis of the bla _OxA‐24‐related gene [26]. The adjacency of ISAba1 and the bla _{OxA‐51‐like} gene was determined with primers ISAba1F and OXA‐51‐like R [27].

2.6. Infection control measures

The following intervention measures were implemented to control this outbreak: (1) isolation of patients in private rooms (or private spaces when single rooms were not available); (2) requirements and audits to assure that health care workers washed their hands between contacts with different patients; (3) requirement that health care workers used separate gloves and gowns for each patient; and (4) daily cleansing of objects surrounding patients' beds with a sodium hypochlorite solution and a room‐dedicated cleaning tool. One XDRAb infected/colonized patient was not removed from a private room or private space until there were three successive negative cultures in 1 week during the follow‐up period [28]. Surveillance cultures were performed twice per week.

All laboratory technicians notified physicians through the computer‐assistant critical value alert system when XDRAb was identified from a clinical specimen. If there were two or more XDRAb cases in the same ward, surveillance of the environment and the health care worker was done to prevent local spread and outbreak.

2.7. Statistical analysis

Continuous variables were compared by an unpaired (two‐sample) Student t test. Categorical variables were compared by Fisher's exact test or the Chi‐squared test. A p value less than 0.05 was considered statistically significant.

3. Results

3.1. Epidemiology and patient characteristics

In our hospital, 46,789 patients were hospitalized during 2006, corresponding to 382,341 hospital person‐days. A total of 5,414 patients were discharged from the ICUs, corresponding to 36,468 ICU person‐days, and 261 patients were discharged from the RCC, corresponding to 5,730 RCC person‐days. There had been no clinical isolate of XDRAb before 2006. We did not restrict carbapenem use during or after the outbreak. Meropenem was the only carbapenem used in our hospital from November 2001 to December 2008. The consumption of meropenem, recorded as defined daily dose, varied monthly among 15–30 defined daily dose/1,000 person‐days, and there was no evident trend during 2005–2006. There were 13 XDRAb cases during the outbreak period (January and February of 2006, Fig. 1). After the outbreak ceased, only six additional sporadic cases were observed in the subsequent 10 months (Fig. 2). Two of these sporadic cases occurred in different rooms of the RCC in March and April 2006, and the other four cases occurred in different wards in November and December 2006. There were no more outbreaks during the period when our infection control measures were implemented. In 2006, there were a total of 19 XDRAb cases, corresponding to an incidence of 0.04% for all hospitalized patients.

Figure 1.

Hospitalization courses and positive‐culture days of all 13 colonized/infected patients in the wards/units. CCU  =  cardiovascular intensive care unit; CSCU  =  cardiosurgical intensive care unit; MCU1 and 2  =  medical intensive care units 1 and 2; NSCU  =  neurosurgical intensive care unit; RCC  =  respiratory care center; SICU  =  surgical intensive care unit; wards  =  including different wards.

Figure 2.

Cases of extensively drug‐resistant Acinetobacter baumannii infection or colonization in the outbreak from January to December 2006. ICUs  =  intensive care unit—medical ICU 1, medical ICU 2, cardiac ICU, neurological ICU, cardiovascular surgical ICU, and neurosurgical ICU; SICU  =  surgical intensive care unit; wards: respiratory care center and other wards.

Table 1 summarizes the medical characteristics of the 13 patients infected during the outbreak period. These patients were originally in diverse hospital wards. Cases 1 and 3 had XDRAb cultured 1–2 days after transfer from a surgical ICU (SICU). Case 2 had just been transferred to the cardiovascular ICU from a local hospital. The SICU was the most commonly involved ward (seven cases, 53.8%). Among our 13 cases, five had infection and eight had colonization. The XDRAb isolates from the eight colonized patients were obtained from sputum specimens. The mean APACHE II scores of the infected patients were higher than those of the colonized patients (mean ± standard deviation: 26 ± 6.8 vs. 21 ± 3.0, respectively; p = 0.016). There was no significant difference regarding the underlying diseases between these two groups.

Four of the five infected patients died within 1 month, four (50%) of the eight colonized patients died, and two died within 1 month. Infected patients tended to die sooner than colonized patients (mean ± standard deviation: 16 ± 9.6 days vs. 26 ± 20.4 days, respectively, p = 0.093).

3.2. Bacterial isolates and antibiotic susceptibility

A total of 32 XDRAb isolates were recovered from clinical specimens of 13 patients, and 36 XDRAb isolates were recovered from 187 environmental samples. Among the environmental samples, 10 were from the control panel surface of medical equipments; 19 from surrounding materials, such as pillows, bedrails, and clothes; five from the patient's axilla; and two from the patient's anus. No XDRAb isolates were recovered from 46 health care workers (five doctors and 41 nurses).

All A baumannii clinical and environmental isolates were resistant to the 20 antibiotics and antibiotic combinations mentioned in the “Materials and methods” (antimicrobial susceptibility testing) section, but not to both polymyxin B and tigecycline. Based on the Etest method for tigecycline, the MIC of four isolates was 2 μg/mL or less, and the MIC of the other 28 XDRAb isolates was 3–6 μg/mL (MIC₅₀: 3 μg/mL; MIC₉₀: 4 μg/mL; range: 1.5–6 μg/mL) [29].

Based on the results of Etest to polymyxin B, all 32 isolates' MICs were less than or equal to 1 μg/mL (MIC₅₀: 1 μg/mL; MIC₉₀: 1 μg/mL; range: 0.5–1 μg/mL), indicating susceptibility to polymyxin B.

3.3. PFGE and PCR

All 32 clinical isolates had the same ribotype and indistinguishable pulsotypes, suggesting that this outbreak was caused by clonal spread. The PFGE pattern of our isolates was indistinguishable from, and presumably closely related to, a pulsotype strain isolated in Taiwan, previously named “R5 pulsotype” [30]. PCR testing indicated that none of our 32 isolates had the following four genes: blaVIM, blaIMP, bla _{OxA‐23‐like}, and bla _{OxA‐58‐like}. However, all 32 isolates had bla _{OxA‐24‐like} and bla _{OxA‐51‐like} genes, and all of these bla _{OxA‐51‐like} genes were preceded by ISAba1, indicating hyperproduction of OXA‐51 [27].

Sequencing of the bla _{OxA‐24‐like} gene indicated that it was identical to A baumannii carbapenem‐hydrolyzing beta‐lactamase bla _OxA‐72 (Genebank accession no. EF534256 and AY739646). The clinical isolates from six additional sporadic cases in the subsequent months in 2006 had different pulsotypes, all of which had more than three band differences.

3.4. Infection control measures

We identified the first case (Case 1) when a culture confirmed the isolation of XDRAb. This was the first instance of XDRAb in our hospital. Infection control measures, especially private room isolation, were instituted immediately. However, XDRAb was subsequently isolated from other patients.

XDRAb was also present in 36 of 187 (19.3%) environmental surveillance cultures. Furthermore, we found XDRAb on the control panel surface of an electrocardiogram (ECG) monitor in the SICU, even after cleaning and disinfection after transfer of Case 5 (who had XDRAb colonization) from the SICU to a ward. This led us to institute an intensive infection control program. We reinforced the need for adherence to disinfection protocols and room‐dedicated cleaning tools. After the aforementioned surveillance study and reinforcement of the need for environmental cleansing, frequent audits indicated that there was no XDRAb on the control panel surfaces of the medical equipments.

The use of carbapenem was not restricted because of the widespread presence of extended‐spectrum β‐lactamase‐producing bacteria in our hospital. Our measures appeared to control the outbreak successfully, and there was no need to close any ICU or ward. Only sporadic cases occurred in the subsequent 10 months.

4. Discussion

We identified and controlled the first outbreak of XDRAb that harbored OXA‐72 and ISAba1‐OXA‐51 carbapenemases in the Asia‐Pacific area. The 13 patients with XDRAb were in four ICUs and three wards. An SICU appeared to be the original source of the outbreak, because the first XDRAb was cultured from the index patient who was recently transferred from this unit and because this unit had the greatest number of cases. All 32 XDRAb clinical isolates from these 13 patients belonged to one pulsotype, indicating clonal spread as the mode of transmission. The OXA‐51 carbapenemases, which are intrinsic enzymes in A baumannii, have weak carbapenemase activity in vitro and a controversial role in imipenem resistance [31]. When ISAba1 (which contains a promoter) precedes OXA‐51, greater expression of beta‐lactamase occurs [27]. OXA‐72 belongs to the OXA‐24 group of carbapenem‐hydrolyzing Class D β‐lactamases [26]. The OXA‐72 gene was first identified in A baumannii in Thailand in 2004 [32]. However, there are limited clinical reports of A baumannii producing OXA‐72 carbapenemase [[33], [34], [35]]. Our results, together with another report in a different southern Taiwan regional hospital, indicate that the emergence of OXA‐72 carbapenemase–producing A baumannii is a serious concern in southern Taiwan [33].

Two sequential outbreaks by OXA‐58‐ or OXA‐72‐producing MDRAb occurred in an ICU in France, which was the first documented outbreak of OXA‐72‐carrying strain in the world [35]. This MDRAb outbreak was limited to only one ward, and contact barrier measures could not control the sequential outbreak because of its environmental persistence. Our surveillance investigation found XDRAb on patients' bodies, in their hospital environments, and on equipments. Our initial disinfection protocol was inadequate, as indicated by the positive culture from the control panel surface of an ECG monitor in SICU after an XDRAb‐colonized patient had just been transferred from this unit. Transmission may have been through contact with this EKG monitor, although XDRAb was not isolated from the patient subsequently admitted to the same SICU bed. A previous report suggested transient hand carriage as a mode of transmission [9]. However, we did not culture any XDRAb from the hands of our medical personnel caring for these patients. A previous study suggested that patients, not the environment, are the main reservoirs of A baumannii [36]. Considering that Case 2 was hospitalized without overlapping with other patients, we suspected that the environment reservoir from Case 4 contributed to his colonization, because the same clone was confirmed by PFGE. Although colonization in the environment disappeared after disinfection, A baumannii on patients' bodies could be responsible for pathogen spread with patient's transference. This forced us to implement private room/space isolation until three consecutive cultures were negative for XDRAb. There have been previous reports of hospitals with MDRAb/XDRAb outbreaks in multiple ICUs over the course of more than 1 year [[37], [38], [39], [40]]. The present study is one of several to report rapid containment of XDRAb without the closure of the ICU, even though we did not find the vehicle or personnel responsible for transmission [[41], [42], [43]].

Tigecycline has been reported to have excellent in vitro activity against XDRAb [44]. However, we found that the MIC of tigecycline was around 3–6 μg/mL for 28 of the 32 clinical isolates—higher than the Food and Drug Administration's definition of susceptibility for Enterobacteriaceae (MIC ≤ 2 μg/mL; Food and Drug Administration, 2005) [29] and the interpretation criteria of British Society of Antimicrobial Chemotherapy (tigecycline susceptible and resistant A baumannii, ≤1 μg/mL and >2 μg/mL, respectively). Although Clinical and Laboratory Standards Institute interpretation criteria are still unavailable, we believe that careful tigecycline usage for such XDRAb infections is warranted. On the other hand, the MICs of polymyxin B were lower than 1 μg/mL in all of our 32 clinical isolates. Colistin also remains a treatment option for A baumannii infection, despite its potential nephrotoxicity and neurotoxicity [45].

The mortality rates associated with MDRAb have been reported to be high [[8], [46]]. In Spain, the mortality rate of A baumannii–infected patients (27%) was higher than that of patients with colonization (11%) [12]. A previous study in Taiwan reported the mortality rate of XDRAb infection to be 38.6% [47]. In the present study, the crude mortality rate was 61.5% (8 of 13), and the mortality rate of infected patients (four of five, 80%) was higher than that of colonized patients (four of eight, 50%). The attributable mortality rate of our patients was 30%. The high crude mortality rate of our patients may be attributed to the severity of the underlying conditions, according to their high APACHE II scores.

There have been previous reports of rapid control of MDRAb or pan‐drug‐resistant A baumannii without ICU closure, when there was early recognition of colonized patients or an outbreak within a single ICU [[9], [48], [49]]. Valencia et al. [43] described the effective control of an outbreak in two ICUs at a university hospital in Spain, and emphasized the importance of a multicomponent intervention program. In our case, the outbreak was hospital‐wide and involved seven units. We also implemented multiple infection control measures, including an intensive educational program, private room or space isolation for each patient, strict contact precautions, daily environmental sodium hypochlorite disinfection, ethanol disinfection of equipment surfaces, and weekly culture follow‐ups. During the outbreak, we did not limit the usage of carbapenems and did not close any hospital unit. Daily cleansing with room‐dedicated tools and sodium hypochlorite or detergent may contain the spread of the outbreak pathogen. Our successful experience may mandate that infection control measures should be implemented until there are three successive negative cultures of a previous colonization site [28]. Sporadic cases of different clones were found in different wards in the subsequent months; no limitation of meropenem use or transference of XDRAb cases between long‐term care facilities and hospitals might have contributed to these.

In conclusion, we reported the first hospital‐wide outbreak in the Asia‐Pacific region, caused by the spread of XDRAb, carrying the OXA‐72 gene and ISAba1 preceding OXA‐51‐like gene. Prolonged colonization of patients' bodies and/or failure to adequately perform equipment disinfection was responsible for transmission. Implementation of infection control education, private room isolation with strict barrier precaution, and environment cleansing can halt transmission without the closure of any hospital units.

Acknowledgments

This project was supported by grants from National Science Council (NSC 98‐2314‐B‐037‐042‐MY3) to Po‐Liang Lu and from Kaohsiung Medical University Hospital (KMUH 95‐5D35) to Wei‐Ru Lin. All authors declare no conflicts of interest.