AutoMated tool for Antimicrobial resistance Surveillance System version 3.0 (AMASSv3.0)

Release date: 24 April 2024

Below is a list of features that we have incorporated in AMASSv3.0

Main analysis

1. We have added “Annex C: Cluster signals”. This annex shows potential clusters of patients with AMR infection identified using the SaTScan (www.satscan.org).
2. We have left processed (i.e. de-duplicated and/or merged) data files in the folder “Report_with_patient_identifiers” so that users can use the processed data files (e.g. deduplicated and merged data files for each AMR pathogen) for any further analysis and internal use after using the AMASS.;
3. Enterococcus faecalis and E. faecium have been explicitly included in the pathogens under the survey (while Enterococcus spp. are used in the AMASS version 2.0);
4. We have added a few antibiotics in the list of antibiotics for a few pathogens under the survey;

Technical aspects

1. We have added a configuration for “Annex C: Cluster signals” in Configuration.xlsx;
2. We have improved the algorithm to support more several date formats;
3. We have improved the algorithm to translate data files;
4. We have improved Data_verification_logfile report to present local languages of the variable names and values (according to how they were recorded in the data files) in the report;
5. We have improved Annex B: Data indicators to support a larger data set;
6. We have used only Python rather than R + Python (as used in the AMASSv2.0);
7. We have set a default config for infection origin stratification by allowing a specimen collected two calendar days before the hospital admission date and one day after the hospital discharge date into consideration. This config supports the real-world setting that several hospitals in LMICs (particularly Thailand) have patients stay in the hospital (e.g. at ER) due to many limitations before official admissions can be made. The one days after an admission record but the specimen has already taken for pathogenic culturation. The one day after the hospital discharge date supports the real-word setting that some laboratories record specimen-arrival-to-the-laboratory date in their data set rather the specimen-collection-from-the-patient date as specimen date. Therefore, some specimens that were collected on the hospital discharge date (either discharge alive or died) have specimen dates one day after the hospital discharge.
8. We have improved the algorithm to include patients who were not yet discharged from the hospital (i.e. having no discharge dates yet) in the analysis by truncating at the last date of specimen date in the whole data set (usually at the end of the year [e.g. 31 Dec] or the survey period). The change allows us to improve the ‘bed-days at risk’ and ‘bed-days at risk for hospital-origin infection’ to be precisely estimated based on their duration in the hospital. For example, a patient who was admitted on 30 Dec 2023, had blood specimen collected for culture on 31 Dec 2023, and were still in the hospital on 10 Jan 2024. Then, the microbiology data file and hospital admission date file were exported on 10 Jan 2024 for the data of year 2023 [specimen dates and hospital admission dates from 1 Jan 2023 to 31 Jan 2023]. This patient would be included in the analysis using AMASSv3.0, assuming that the days at risk of BSI was from 30 Dec to 31 Dec 2023, and the specimen collected on 31 Dec 2023 would be included in the analysis. This patient would not be included in the analysis of the “Section 6: Mortality” as discharge outcome is still missing. In the AMASSv2.0, this patient would not be included in the analysis due to the missing discharge date and missing discharge outcomes).
9. We have revised text and fixing bugs/typos.

AutoMated tool for Antimicrobial resistance Surveillance System (AMASS) was developed as an offline, open−access and easy−to−use application that allows a hospital to perform data analysis independently and generate antimicrobial resistance (AMR) surveillance reports stratified by infection origin from routinely collected electronic databases. The application was built in Python, which is a free software environment. The application has been placed within a user−friendly interface that only requires the user to double−click on the application icon.

AMASS performs data analysis and generates reports automatically. The raw data files required are hospital admission and microbiology databases. Firstly, the application translates and de-duplicated the microbiology data file, and produces the AMR surveillance reports without stratification by infection origin (Sections 1 and 2 in the report). Secondly, the application then merges the microbiology and hospital admission data files, analyzes the merged data, and produces the AMR surveillance reports with stratification by infection origin (Sections 3, 4 and 5 in the report). Finally, the application then performs a statistical analysis to estimate all−cause mortality of patients following AMR infection (Section 6 in the report). AMASS uses a tier-based approach. For example, in cases that only the microbiology data file with the results of culture positive samples is available, only the AMR surveillance report without stratification by infection origin will be generated.

Further details on how to use the application can be found at https://www.amass.website.