Water purification systems installed in two ice machines in a Boston hospital were supposed to make the water taste and smell better for patients on a surgery floor—but it ended up killing three of them, an investigation found.

The purification systems inadvertently stripped chlorine from the municipal tap water, allowing bacteria normally found at low levels to flourish and form biofilms inside the machines. This led to infections in four vulnerable cardiac-surgery patients who had prolonged stays on the hospital floor. Three of them died of their infections.

Researchers detailed the case cluster and ensuing investigation in a study published Monday in the Annals of Internal Medicine.

"Our cluster demonstrates the risk for unintended consequences associated with systems designed to improve hospital water," the researchers—led by Michael Klompas, who works at the hospital and is a professor in Harvard's department of population medicine—concluded.

Common source

Identifying the cluster and the culprit took some significant sleuthing; the four cases occurred sporadically between March 2017 and October 2018 at Brigham and Women's Hospital in Boston. And it wasn't immediately obvious that they were linked.

Doctors became suspicious after taking note of three cases in 2018, all of which had invasive Mycobacterium abscessus infections. M. abscessus infections are rare, but they're known to be associated with health care—called a nosocomial infection—which hinted at a common source of infection somewhere in the hospital. This led the researchers to search hospital records dating back to 2015 for other possibly linked cases. That's when they identified the fourth case, which occurred in 2017.

Whole genome sequencing of clinical isolates from the four identified cases revealed almost perfect matches of their M. abscessus isolates. The bacterial isolates' genomes had just one to three different point mutations among them; fewer than 20 suggest an epidemiological link. So, the researchers got to work trying to find a common source.

Of the four cases, only three had surgery at the hospital, and they were done in different operating rooms, with different cardiopulmonary bypass heater-cooler devices, which help control a patient's temperature during surgery. Only one patient required hemodialysis, ruling that out as a common source. The researchers also ruled out mechanical ventilation as a source. Although the patients were all on the same cardiac surgery floor of the hospital and each occupied multiple rooms during their stay, there was only one room in common for two of the four patients.

The most obvious similarity between the patients' stays was that they were all prolonged. While the mean length of stay for similar patients in the cardiac surgery floor was just over two weeks, the four patients had stays between 42 days and 131 days before their M. abscessus were identified. And nurses anecdotally noted that the patients seemed to consume more ice than others.

Pure and deadly

M. abscessus is a water-loving bacterium known to occur at low levels in tap water, which is not sterile. As such, the researchers turned to the hospital's water sources, surveying sinks, showers, and ice and water machines on the floor. Samples from two of the 14 sinks and showers on the floor had low levels of Mycobacteria contamination, between 10 to 50 colony-forming units per milliliter. But samples from the two ice and water machines were highly contaminated—between 2,000 and 8,000 colony-forming units per milliliter. And from those samples, the researchers could fish out genetic sequences unique to the M. abscessus isolates that had infected the patients, strongly suggesting this was the common source.

Although hospital records indicated that the machines had been maintained and cleaned per the manufacturer's instructions, the researchers noted visible bacterial biofilms on the inside components of the machines.

Alarmed by the finding, the researchers began surveying sinks, showers, and ice and water machines in other hospital areas—but nothing else had such high levels of bacterial contamination as the two machines on the cardiac surgery floor. When they checked chlorine levels, this made more sense. Chlorine levels were in acceptable ranges for all of the samples, except from the two machines.

On closer inspection, the researchers noted that the two machines had been fitted with commercial water purification systems that included a 5.0-micron carbon filter (designed to improve taste and remove odors) and an ultraviolet disinfection unit. Both carbon filters and ultraviolet light are known to decrease chlorine concentrations. Sampling from the input and output pipes confirmed that the purification system was stripping the acceptable input chlorine concentration of 2.5mg/L to 0mg/L in the output.

The ice machines were removed from service in October 2018 and no other M. abscessus infections were identified afterward.

"Our experience suggests the potential danger of well-intentioned measures designed to improve water quality in health care facilities," the researchers wrote in their study. They outlined various strategies to help prevent such infections in the future, including switching to sterile or distilled water for patient care, regular sampling, and chlorine monitoring. "Hospitals must be particularly attentive to the threat of water-based infections," they wrote.