Yulia Rosa Saharman

In, we found that carbapenem-non-susceptibility of these three species was largely based on the production of β-lactamases, but these three species did not share the same genes coding for the carbapenemases. Hospital-associated infections (HAI) outbreaks caused by pathogenic strains have dramatic repercussions. In order to tackle this problem, infection-control teams use a broad range of typing techniques to help trace bacterial sources, to define the mode of dissemination of pathogenic clones and, ultimately, to hinder or preferably stop microbial spread. Most routine medical microbiology laboratories across the world currently use molecular biological techniques for this purpose. For the last decades Pulsed-Field Gel Electrophoresis (PFGE; based on the fragmentation of the bacterial genome by a specific enzyme) has been one of the most commonly used techniques in diagnostic laboratories across the world. In addition, Multi-locus variable number of tandem repeats (VNTR) analysis (MLVA), multi-locus sequence typing (MLST, based on polymorphisms in a limited number of housekeeping genes) are employed. During the last few years, whole-genome sequencing (WGS) has become a preferred tool for the molecular epidemiological surveillance of infectious diseases.(6) In the beginning of the study, we also used Raman spectroscopy as a rapid phenotypic bacterial typing method.(10) We were able to apply this technique to differentiate strains of Klebsiella pneumoniae and Acinetobacter baumannii, but the technique could not be applied to P. aeruginosa due to interference by pyocins that are intrinsic to this species.

Typing with Raman spectroscopy revealed five major clusters of CNAB, the largest cluster (designated CIPTO-31) consisted of isolates obtained from patients (screening and clinical specimens) and isolates from the environment throughout the study period. MLST, performed for a subset of 14 isolates, revealed the presence of multiple sequence types (STs), which corresponded closely to the Raman spectroscopy clustering. Four previously identified STs (STN9S, ST208, ST2N8, and STS42) as well as several new STs, and a new allele for the gpi gene were found in this study.(2)

Raman spectroscopy analysis also revealed the presence of multiple types of CNKP. There were three major clusters, the largest cluster (CIPTOKPN24) consisted of isolates obtained from patients (screening and clinical specimens) and were present in both ICUs throughout the study period, whereas other clones seemed to wax and wane with time.(4) A total of 97 clinical and 1 environmental isolate were further analyzed using MLVA genotyping, identifying 30 different genotypes. Clustering of strains by Raman spectroscopy into three dominant groups was concordant with clustering by MLVA, e.g. the 20 Raman CIPTOKPN24 strains all belonged to a single MLVA clonal complex. Likewise, the four Raman CIPTOKPNPO strains belonged to a single MLVA clonal complex as did CIPTOKPN27 isolates.(4)

MLST revealed four major clusters of CNPA (ST2PS, ST82P, ST44S and STPS7) as well as several new sequence types. By MLVA, five major clusters were distinguished, two belonging to ST2PS and the others corresponding to ST82P, ST44S and STPS7. Most isolates belonged to ST2PS (patient and environmental isolates), of which 22 isolates harbored blaIMP, 24 isolates harbored blaGES-5 but no isolates contained blaVIM. All ST82P isolates harbored blaVIM.

We further evaluated five molecular typing techniques using an epidemiologically well-characterized set of CNPA. They were Multi-Locus Variable Number of Tandem Repeat (VNTR) Analysis (MLVA), in-silico seven-loci multi-locus sequence typing (MLST), core and whole genome MLST (cg/wgMLST), and core Single Nucleotide Polymorphism (SNP) analysis (cgSNP). Our findings show that the three latter techniques (cgMLST, wgMLST and cgSNP) provide the highest level of resolution allowing detailed epidemiological analysis of local outbreaks and international dissemination. MLVA is a suitable alternative for accurate typing of P. aeruginosa, useful in settings where the transition towards WGS is currently not feasible.

Thus, the overall impression is that carbapenem-non-susceptible strains of the three species targeted by our studies show endemicity of a few clones that are circulating among ICU patients and the ICU environment.

Objective 2 of the study: To develop an intervention - feasible to be applied in a low-resource ICU setting - that may significantly reduce the risk of acquisition and infection by carbapenem-non-susceptible A. baumannii, P. aeruginosa and K. pneumoniae.

A large percentage of HAI are preventable through effective infection prevention and control (IPC) measures.(11) After phase 1, we introduced a multimodal bundle of IPC interventions that initially consisted of the following measures (12) and see chapter 8:

1. A multifaceted hand hygiene improvement program. We based the hand hygiene programs on the WHO’s Five Moments for Hand Hygiene guidelines and tools. The hand hygiene improvement program included education with pre- and post-questionnaires testing of knowledge and attitudes, performance feedback and reminders, interviews, and role models as described before.(12)
2. A single round of an environmental disinfection campaign involving the whole environment of both ICUs using 1:100 sodium hypochlorite solution. This disinfectant solution was applied to walls, floors, doors, beds (mattresses and bed rails), sinks, overbed tables, infusion and suction pumps and stands, monitors and ventilators including connecting lines, surrounding counter tops including the adjacent cleaning service room. In addition, all curtains between beds were exchanged for clean ones.
3. Routine environmental disinfection was introduced with 1:100 sodium hypochlorite solution that included the floors, beds, and immediate surrounding of the patients. This was done twice daily. In case of visible dirt, this was first removed with a brush and water, before the application of the sodium hypochlorite solution. The intensive procedure as described above was repeated every two weeks in this phase. The curtains between beds were refreshed every 1-2 months or immediately after visible soiling.
4. Enforced antibiotic stewardship (including daily evaluation of all antibiotic prescriptions on weekdays).
5. All patients found positive for one or more carbapenem-non-susceptible A. baumannii-calcoaceticus complex, K. pneumoniae, or P. aeruginosa were cohorted in one dedicated corner of the ICU. HCWs donned mask, gown, and gloves when approaching and providing care for cohorted patients.
6. Daily total body-washing with cloths soaked in a chlorhexidine gluconate 2% solution. These cloths were prepared and pre-packaged individually in sealed plastic bags by the hospital pharmacy.
7. Introduced a 2% chlorhexidine gluconate solution for decontamination of the oropharynx. Bottles containing this solution were also prepared and provided by the hospital and used per patient. Oral decontamination was performed 4 time daily.

Objective 3 of the study: To apply and determine the efficacy of the intervention (developed as specified above/ under 2) in a low-resource ICU setting.

We measured HH knowledge and HH compliance before (at baseline) and directly after a multifaceted improvement program (post-intervention) and performed a re-evaluation three years later. The multifaceted improvement program included education, feedback, reminders, interviews and the use of role models. There was a statistically significant improvement in the median overall HH knowledge score at post-intervention. The overall HH compliance was 27% at baseline and significantly improved to 77% post-intervention. For all five HH moments, the compliance of nurses and physicians separately improved significantly from the baseline phase to the post intervention phase, except for ‘moment 3’ (after body fluid exposure), for which baseline rates were already high. Most of the compliance rates were significantly lower in both groups of healthcare workers upon follow-up three years later, they essentially had fallen back to pre-intervention levels. Overall, the HH compliance of the nurses was significantly better than the physicians’ compliance.(12) Thus, maintaining high levels of HH compliance requires continuous monitoring and regular interventions.

We evaluated the effect on the acquisition of carbapenem-non-susceptible Acinetobacter baumannii-calcoaceticus complex (CNAB), Klebsiella pneumoniae (CNKP), and Pseudomonas aeruginosa (CNPA). Using a quasi-experimental before-and-after design study, for all three species taken together there was a significant step change, from phase 1 to phase 3 in the rate of acquisition of carbapenem-non-susceptible strains. This significant decrease in the overall acquisition rate of carbapenem-non-susceptible strains of the three species was mainly caused by a decrease in the acquisition of carbapenem-non-susceptible A. baumannii-calcoaceticus complex and, to a lesser extent, K. pneumoniae. Interestingly, the acquisition rate of P. aeruginosa was little affected by the multimodal intervention. Within each of the two phases there was no major downward or upward trend observed in the rate of acquisition of resistant strains for any of the three species separately nor for the three species taken together, although the risk of acquisition was increasing slightly in phase 1.(see chapter 8)

Using WGS in combination with clinical data, we were able to closely track and trace the endemic spread of isogenic carbapenem-non-susceptible strains of Pseudomonas aeruginosa over a 3-year period in the ICUs. We found that the number of CNPA transmissions and acquisitions by patients was highly variable over time but that, overall, the rates were indeed not significantly reduced by the intervention. Environmental sources were involved in these transmissions and acquisitions. Four high-risk international CNPA clones (ST2PS, ST82P, STPS7, and ST44S) dominated, but the distribution of these clones changed significantly after the intervention was implemented.(6) Thus, the multimodal intervention may have altered the clonal composition of endemic carbapenem-non-susceptible P. aeruginosa but it did not eradicate the sources or niches nor affected the transmission of such strains in the ICU environment.

We conclude that a multimodal intervention aiming to prevent acquisition of resistant strains of important ICU pathogens is essentially feasible and may be quite effective in ICUs in lower-middle income countries. However, even multimodal interventions may not be equally effective for all species harbouring antibiotic-resistant strains.

RECOMMENDATION FOR FURTHER RESEARCH

Our study was conducted in two ICUs in a single center, the national reference hospital in Jakarta, Indonesia. Since Indonesia is a large country and the fourth most populous in the world, the data presented in this thesis do not represent the whole of Indonesia. In the future, a national surveillance program for carbapenem-non-susceptible Gram-negative bacilli should be established, including all provinces of Indonesia, with collection of epidemiological data, phenotypic and genotypic resistance mechanisms, and analysis of clonal relatedness. An alert system should be set up for the recognition of “high-risk” clones or unnoticed transmission routes. For example, our finding that some patients already carried carbapenem-non-susceptible isolates on admission to the ICU, could be suggestive for transmission between hospitals when patients are transferred or referred, or even suggestive for the existence of reservoirs in the community. These possible sources and routes should be further explored.

We found a limited variety of carbapenemase genes in each species, CNAB only had blaoxa-23 and blaoxa-51 genes, CNKP only carried blaNDM and CNPA carried blaVIM, blaGES-5 and blaIMP. Do we also find the same carbapenemase-coding genes among ICU isolate of these three species in the other provinces in Indonesia? That still questionable. A surveillance system would generate data on this.

We concluded that a multimodal intervention aiming to prevent acquisition of resistant strains of important ICU pathogens is feasible and may be quite effective in ICUs in lower-middle income countries, but not for CNPA. Environmental cleaning seemed to be a very important part of the intervention with CNPA. Pseudomonas aeruginosa is a unique pathogen, this organism can survive over a long-term period and live in moist niches. There are two likely transmission pathways for CNPA, i.e. via exposure to untreated (waste)water and via contaminated sinks/taps in the hospital. This hypothesis needs to be explored by multicentre collaborative research efforts.

We found in our study, that the use of antibiotics, especially carbapenems, in ICUs is a major independent risk factor for the acquisition of carbapenem-non-susceptible strains. We, therefore, suggest that antimicrobial stewardship should be introduced in ICU daily practice and that its effects on the epidemiology of antimicrobial resistance be studied in this setting.

Also, the application of the intervention was not monitored except for the hand hygiene compliance. Future studies should closely monitor environmental cleaning (by observation of cleaning practices, and by much more environmental culturing), monitor the use of chlorhexidine body and oral decontamination) and last, but not least, it should be monitored whether patients with positive cultures are actually cohorted in the designated space or room. This should all be registered in future studies, essentially providing accurate data on the implementation of multimodal interventions.

Finally, when building new ICU facilities, even in low-resource settings, one should pay much more attention to avoid creating typical environmental niches for ICU pathogens and provide more structural barriers to the survival and spread of micro-organisms in general.