LL Part 4: Outbreak in a New NICU

Build it and the pathogens will come...

Following on from our previous post (here)

This story begins in February 2013 as water treatment, system flushing and extensive water testing declares the plumbing in a newly build neonatal intensive care unit (NICU) fit and ready. Between the completion of this testing and mid-May, when the first staff and patients began to take up residence, the process of ‘fitting out’ and ‘setting up’ the NICU occured.

In early June, not even a month after the new NICU had been opened, a few babies began to appear unusually unwell. While some babies appeared to decline gradually, a couple were seen to decline from seemingly healthy to seriously unwell quite rapidly. However, it would not be until early September before hospital clinicians and management would recognise that some sort of outbreak had occured and commence an investigation. But by that time whatever it was had already taken hold of their unit.

The hospital declared an outbreak, notified the health department, and closed their recently opened unit to new admissions by diverting them to another NICU.

The Health Department called for outside assistance, and epidemiologists and other experts were provided who observed the infection control processes in the NICU, conducted hygiene and contact precautions audits, and collected extensive swab samples for laboratory testing.

The samples were drawn from the NICU environment, for example: hand basins, faucets, drains, freshwater samples, surfaces around the ward and from ventilators, breast pumps and incubator humidity outlets. Pseudomonas was identified on 21 separate samples collected from the NICU environment, and screening swabs from 10 of the neonates - 5 were classified as clinical cases (those with both a positive test result and symptoms) and 5 were classified as surveillance cases (those with either symptoms or a positive test result, but not both). Tests of all neonates between September and December identified 15 cases, 13 clinical and 2 surveillance. Even with extensive treatment 2 of the clinical cases died during the period. The case fatality rate among clinical cases during the first outbreak was 15.4% (2/13).

Infection control teams identified issues with inconsistent handwashing and the washing of breast pump parts under tap water without following the manufacturer’s guidelines for daily steralisation. However, the most curious finding of all was that the source of the outbreak appeared to be the fresh water coming from the faucets in the unit. The bacteria had colonised the clean water plumbing while it sat idle between when tests were carried out in February 2013, and when the unit was opened to patients in May 2013.

Extensive remediation and mitigation work followed. Aerators were removed and disposable membrane filters were installed ahead of the complete replacement of every faucet. NICU surfaces, basins and faucets were cleaned and disinfected, and hyperchlorination flushes were run through the building’s fresh water pipes. Greater emphasis was placed on healthcare worker hand hygiene training and auditing, and a policy for the use of alcohol based hand rub was implemented. Gloves were mandated during every contact with the neonates.

Sampling continued, but between late December 2013 and May 2014 no additional positive swabs occured.

The sink faucets were replaced in May 2014 with models thought less likely to become contaminated however, in mid June 2014, a new clinical case was identified. Fearing Pseudomonas was in the pipes again, the hospital again flushed the pipes with hyperchlorinated solution. The second outbreak in mid-2014 identified a total of 16 cases, 4 clinical and 12 surveillance. A third infant died and the NICU was again closed and new patients diverted to other facilities. The case fatality rate among clinical cases during the second outbreak was 25% (1/4)

Risk Factors

Low birthweight was a significant factor for infection, because while only 10% of infants admitted to the NICU during 2013-2014 were <1000g, 52% of infected neonates had birthweights <1000g. Almost all (95%) of the infected infants had a peripherally inserted central catheter (PICC) and/or were ventilated (95%).

Conclusion

There are several conclusions to be drawn from this example.

First, the case fatality rates are low compared to many other studies. We believe this is because hospital staff recognised fairly quickly that something was amiss, and their rapid response meant that they were testing for and reacting to cases as they happened. In many of the other studies we have reviewed while researching this issue recognition and the implementation of surveillance and mitigations occurred much later - resulting in more cases and a greater number of fatalities.

Second, just because the hospital ward or NICU is brand new does not mean it is pathogen free. Further, the fact that extensive surveillance and expensive mitigations have recently been undertaken does not guarantee the ward will remain pathogen free.

Third, hospital wastewater does not have to be present as the colonisation mechanism for a hospital ward. It is possible for the clean water - water that the adult nurses, doctors and parents of the neonates are probably drinking, to be the source of a deadly infective organism. This water is often used to wash healthcare workers hands, rinse hospital equipment that comes into contact with the neonates, and bathe the tiny infants.

Finally, it is possible for neonates to return a positive test result for the infective organism and not show symptoms (potentially a silent infection or false positive test result), as well as to have symptoms while not returning a positive test result (potentially a false negative test result). Nurses and doctors must be trained to be aware of the posibility for both outcomes.

Kinsey, C. B., Koirala, S., Solomon, B., Rosenberg, J., Robinson, B. F., Neri, A., ... & Gould, C. V. (2017). Pseudomonas aeruginosa outbreak in a neonatal intensive care unit attributed to hospital tap water. infection control & hospital epidemiology, 38(7), 801-808.

The next post in this series can be found (here)