TARTARE retrospective microbiological laboratory data from three hospitals, Ethiopia, 2018-2022

Clinical laboratory data were retrospectively collected from bacteriology and parasitology laboratories in three Ethiopian hospitals. The dataset contains stool sample test results and patient metadata for samples tested from January 1, 2018 – December 31, 2022, at Yekatit 12 Hospital (Addis Ababa), University of Gondar Hospital (Gondar), and Hiwot Fana Comprehensive Specialized Hospital (Harar). Samples were tested using traditional laboratory methods and hospital procedures, and data were retrospectively transferred from paper laboratory logbooks into an electronic database (Microsoft Access). Clinical outcomes include raw and clean stool sample results, including codetections. Patient metadata include age, sex, and date of sample submission. Methods for stool sample processing and testing varied by site. In Addis Ababa, four departments ordered submission of patient stool samples. Samples were tested in either the microbiology or parasitology laboratory, but not both. Samples submitted to the microbiology laboratory were only tested for Salmonella and Shigella on routine basis, with testing for Vibrio cholerae occurring during periods of suspected outbreaks. In Gondar, five departments ordered submission of patient stool samples to either the microbiology or parasitology laboratory. Samples submitted to the microbiology laboratory were tested for the same pathogens as in Yekatit 12. In Harar, three departments ordered submission of patient stool samples to the parasitology laboratory for testing. No bacterial testing occurred as the hospital did not have a microbiology laboratory. At all three sites, stool samples were tested using ISO protocol 15189, and parasitology testing was performed using wet mount microscopy. # TARTARE retrospective microbiological laboratory data from three hospitals, Ethiopia, 2018-2022 [doi.org](doi.org) ## Description of the data and file structure A retrospective cross-sectional hospital study was conducted using laboratory data from 2018 through 2022 at three hospitals (Addis Ababa, Gondar, Harar) in Ethiopia. Stool sample outcomes and patient demographics from paper logbooks were transcribed into Microsoft Access and dataset was cleaned for analysis using SAS 9.4. ### Files and variables #### File: hosp_et_2018_2022.csv **Description:** This file was generated by exporting data from Microsoft Access. It contains de-identified patient data related to stool samples tested in hospital parasitology and microbiology (bacteriology) laboratories at three hospitals in Ethiopia.  ##### Variables * ID: unique observation identifier * stool_outcome: diagnostic laboratory outcome from patient sample tested for parasitic or bacterial pathogens * lab_type: type of laboratory sample was sent for testing and analysis (parasitology/microbiology) * hospital: hospital where sample was collected and analyzed * sex: patient sex (male/female) * stool_binary: numeric identified for pathogen detection (Y/N) * month: month of stool sample submission * year: year of stool sample submission * season: season of stool sample submission  * age_cat: categorical age group in years for patient submitting stool sample #### File: hosp_et_2018_2022_Values.csv **Description:** This file contains information on the labels assigned to values in the dataset. Missing data values are indicated as either blank cells (character variables) or with a "." (numeric variables). ## Human subjects data This dataset was derived from retrospective laboratory data collected from hospital sources outside the United States. All data have been de-identified in accordance with applicable ethical guidelines and international data protection standards. Specifically, all direct identifiers (e.g., names, addresses, contact information, medical record numbers) have been removed, and any indirect identifiers (e.g., specific dates, locations, or rare diagnoses) have been sufficiently generalized or excluded to prevent re-identification of individuals. No identifiable personal information is included, and no linkable codes or keys are retained that would allow re-identification. As a result, the dataset does not contain protected health information and is not considered identifiable under relevant regulations.