Methods for computational disease surveillance in infection prevention and control: Statistical process control versus Twitter's anomaly and breakout detection algorithms

Thresholds versus Anomaly Detection for Surveillance of Pneumonia and Influenza Mortality

Computational surveillance of pneumonia and influenza mortality in the United States using FluView uses epidemic thresholds to identify high mortality rates but is limited by statistical issues such as seasonality and autocorrelation. We used time series anomaly detection to improve recognition of high mortality rates. Results suggest that anomaly detection can complement mortality reporting.

Fluorescence spectroscopy for the determination of reconstitution time of an in-vial lyophilised product

Lyophilisation is a prominent technique used to create stabilised, dried forms of biopharmaceutical formulations. Reconstitution of lyophilised parenteral formulations is a key step prior to patient administration. The accurate determination of reconstitution time is a necessity to aid formulation development and support product quality control. Traditional methods for quantifying reconstitution time involve the visual identification of the endpoint, which has led to variable values reported across studies. In this work, the use of ultra-violet (UV) excited fluorescence spectroscopy as an alternative to the visual quantification of the reconstitution time was investigated. Spectrographic information was collected via a bespoke setup that allowed the measurement of the reconstitution time in a standard sealed lyophilisation vial. The spectra were analysed via principal component analysis (PCA) to obtain a time-based representation of the changes in a reconstituting formulation. The analysis was followed by the identification of an endpoint using three techniques ranging from fully automated to manual with regards to the required level of user input. At high protein concentration, the variability of the reconstitution time measurements was reduced from 80.4% relative standard deviation obtained via the traditional method to 8.2% for the instrumental method presented in.

Automated Cluster Detection of Health Care-Associated Infection Based on the Multisource Surveillance of Process Data in the Area Network: Retrospective Study of Algorithm Development and Validation

BACKGROUND

The cluster detection of health care-associated infections (HAIs) is crucial for identifying HAI outbreaks in the early stages.

OBJECTIVE

We aimed to verify whether multisource surveillance based on the process data in an area network can be effective in detecting HAI clusters.

METHODS

We retrospectively analyzed the incidence of HAIs and 3 indicators of process data relative to infection, namely, antibiotic utilization rate in combination, inspection rate of bacterial specimens, and positive rate of bacterial specimens, from 4 independent high-risk units in a tertiary hospital in China. We utilized the Shewhart warning model to detect the peaks of the time-series data. Subsequently, we designed 5 surveillance strategies based on the process data for the HAI cluster detection: (1) antibiotic utilization rate in combination only, (2) inspection rate of bacterial specimens only, (3) positive rate of bacterial specimens only, (4) antibiotic utilization rate in combination + inspection rate of bacterial specimens + positive rate of bacterial specimens in parallel, and (5) antibiotic utilization rate in combination + inspection rate of bacterial specimens + positive rate of bacterial specimens in series. We used the receiver operating characteristic (ROC) curve and Youden index to evaluate the warning performance of these surveillance strategies for the detection of HAI clusters.

RESULTS

The ROC curves of the 5 surveillance strategies were located above the standard line, and the area under the curve of the ROC was larger in the parallel strategy than in the series strategy and the single-indicator strategies. The optimal Youden indexes were 0.48 (95% CI 0.29-0.67) at a threshold of 1.5 in the antibiotic utilization rate in combination-only strategy, 0.49 (95% CI 0.45-0.53) at a threshold of 0.5 in the inspection rate of bacterial specimens-only strategy, 0.50 (95% CI 0.28-0.71) at a threshold of 1.1 in the positive rate of bacterial specimens-only strategy, 0.63 (95% CI 0.49-0.77) at a threshold of 2.6 in the parallel strategy, and 0.32 (95% CI 0.00-0.65) at a threshold of 0.0 in the series strategy. The warning performance of the parallel strategy was greater than that of the single-indicator strategies when the threshold exceeded 1.5.

CONCLUSIONS

The multisource surveillance of process data in the area network is an effective method for the early detection of HAI clusters. The combination of multisource data and the threshold of the warning model are 2 important factors that influence the performance of the model.

Rapid Decline in Online Search Queries for Hip and Knee Arthroplasties Concurrent With the COVID-19 Pandemic

BACKGROUND

In response to the coronavirus disease 2019 (COVID-19) pandemic, US hospitals have canceled elective surgeries. This decline in total joint arthroplasty (TJA) revenue may place financial strain on hospitals. Our goal was to quantify the impact of COVID-19 on the public interest in elective TJA.

METHODS

The Google Search Volume Index (GSVI) identified the terms "knee replacement," "hip replacement," and "orthopedic surgeon" as the most common to describe TJA. The term "elective surgery cancellation" was also analyzed. Weekly GSVI data were extracted between 04-01-2015 and 04-04-2020. Time series analysis was conducted and state GSVI values were compared with COVID-19 prevalence and unemployment claims.

RESULTS

The relative public interest in elective TJA has sharply declined since the WHO declaration of COVID-19 as a global pandemic. Between 03-01-2020 and 03-29-2020, the popularity of searches for "knee replacement", "hip replacement," and "orthopedic surgeon" dropped by 62.1%, 52.1%, and 44.3%, respectively. A concurrent spike was observed for the term "elective surgery cancellation." California, New Hampshire, Maine, and Nevada showed a low relative rate for TJA searches, and the highest increase in unemployment claims.

CONCLUSION

The onset of COVID-19 correlates with declining relative popularity of searches related to elective TJA. Higher volume of COVID-19 cases in certain states may correspond with lower relative search popularity, although this correlation remains unclear. These results portend the possibility of a decline in elective TJA case volume, further straining hospitals. Further research is required to inform stakeholders how best to proceed and determine any sustained effects from the current diminished relative interest in TJA.

LEVEL OF EVIDENCE

Level III.

Hand hygiene compliance surveillance with time series anomaly detection

BACKGROUND

Hand hygiene is the most important intervention to reduce the risk of transmission of pathogens in health care. Assurance of effective hand hygiene improvement campaigns includes adequate data analytics for reporting compliance. Traditional analytical approaches for monitoring hand hygiene compliance suffer from several limitations, including autocorrelation. The objective of this study was to use a novel time series anomaly detection algorithm to analyze routine hand hygiene compliance data.

METHODS

Hand hygiene compliance data were collected daily by trained observers in a large academic medical center. Statistical process control p-charts were used as a comparison method of analysis per facility protocol. Time series anomaly detection was carried out using the seasonal and trend decomposition using LOESS (STL) algorithm.

RESULTS

A total of 34 months of hand hygiene compliance data were analyzed. Traditional statistical process control p-charts identified over 76% of rates as special-cause variation, whereas STL identified 18% of rates as anomalous.

CONCLUSIONS

This study supports the use of time series anomaly detection for the routine surveillance of hand hygiene compliance data. This method will facilitate specific and accurate feedback, helping to improve this critical approach for improving patient safety.

Googling your hand hygiene data: Using Google Forms, Google Sheets, and R to collect and automate analysis of hand hygiene compliance monitoring

BACKGROUND

Hand hygiene is one of the most important interventions in the quest to eliminate healthcare-associated infections, and rates in healthcare facilities are markedly low. Since hand hygiene observation and feedback are critical to improve adherence, we created an easy-to-use, platform-independent hand hygiene data collection process and an automated, on-demand reporting engine.

METHODS

A 3-step approach was used for this project: 1) creation of a data collection form using Google Forms, 2) transfer of data from the form to a spreadsheet using Google Spreadsheets, and 3) creation of an automated, cloud-based analytics platform for report generation using R and RStudio Shiny software.

RESULTS

A video tutorial of all steps in the creation and use of this free tool can be found on our YouTube channel: https://www.youtube.com/watch?v=uFatMR1rXqU&t. The on-demand reporting tool can be accessed at: https://crsp.louisville.edu/shiny/handhygiene.

CONCLUSIONS

This data collection and automated analytics engine provides an easy-to-use environment for evaluating hand hygiene data; it also provides rapid feedback to healthcare workers. By reducing some of the data management workload required of the infection preventionist, more focused interventions may be instituted to increase global hand hygiene rates and reduce infection.