Abstract
Background
Nontuberculous mycobacteria (NTM) are emerging pathogens increasingly implicated in healthcare facility-associated (HCFA) infections and outbreaks. We analyzed the performance of statistical process control (SPC) methods in detecting HCFA NTM outbreaks.
Methods
We retrospectively analyzed 3 NTM outbreaks that occurred from 2013-2016 at a tertiary care hospital. The outbreaks consisted of pulmonary
Mycobacterium abscessus complex (MABC) acquisition, cardiac surgery-associated extrapulmonary MABC infection, and a bronchoscopy-associated pseudo-outbreak of
Mycobacterium avium complex (MAC). We analyzed monthly case rates of unique patients who had positive respiratory cultures for MABC, non-respiratory cultures for MABC, and bronchoalveolar lavage cultures for MAC, respectively. For each outbreak, we used these rates to construct a pilot mov ing average (MA) SPC chart with a rolling baseline window. We also explored the performance of numerous alternative control charts, including exponentially weighted moving average, Shewhart, and cumulative sum charts.
Results
The pilot MA chart detected each outbreak within 2 months of outbreak onset, preceding actual outbreak detection by an average of 6 months. Over a combined 117 months of pre- and post-outbreak surveillance, no false positive SPC signals occurred (specificity = 100%). Prospective use of this chart for NTM surveillance could have prevented an estimated 108 cases of NTM. Six high-performing alternative charts detected all outbreaks during the month of onset with specificities ranging from 85.7% to 94.9%.
Conclusions
SPC methods have potential to substantially improve HCFA NTM surveillance, promoting early outbreak detection and prevention of NTM infections. Additional study is needed to determine the best application of SPC for prospective HC FA NTM surveillance in other settings.
