# Data Integrity in the Microbiology Lab: ALCOA+ Principles Data integrity is the degree to which a laboratory record is complete, consistent, accurate, and trustworthy throughout its entire life cycle, from specimen collection to long-term archiving. ALCOA+ is the nine-attribute framework used to achieve that assurance in regulated laboratories, standing for attributable, legible, contemporaneous, original, and accurate, extended by complete, consistent, enduring, and available. In a microbiology laboratory, a record is rarely a single number. A positive blood culture result depends on an instrument flag, a Gram stain observation, an organism identification, an antimicrobial susceptibility panel, and the judgment of the microbiologist who signed the report. If any link in that chain is undocumented, backdated, transcribed incorrectly, or impossible to trace back to the person who made the observation, the clinical value of the result collapses, even if every organism identification itself was analytically perfect. Data integrity in the microbiology lab is therefore both a patient safety issue and a regulatory one, because accreditation standards such as ISO 15189 expect documented [data management](/blog/guides/data-management-basics-principles-processes-and-best-practices) and sample traceability as core elements of the quality system [1]. ## What Is Data Integrity ALCOA+ in a Microbiology Laboratory: Explained Data integrity answers one question: can you trust the record, and can you prove that you should? Data quality answers a different question: is the measurement itself analytically sound? A minimum inhibitory concentration can be measured with an exquisitely validated instrument (high quality) yet be recorded as an unrevisable value in a spreadsheet with no user identity attached (poor integrity). Laboratory quality management treats the two as inseparable, because an untrustworthy record cannot demonstrate quality and a poor-quality result cannot be rescued by careful documentation. The ALCOA acronym originated in United States Food and Drug Administration inspection practice and was later expanded by international regulators into ALCOA+ and ALCOA++ (the latter adding traceability). ALCOA+ is now the working vocabulary of inspections under 21 CFR Part 11 for electronic records in United States-regulated laboratories, under ISO 15189 for medical laboratories, and under ISO/IEC 17025 for testing and calibration laboratories, where a risk-based approach to computerized system assurance is explicitly expected [2]. Studies of laboratory service quality consistently identify documentation practices and adherence to standard operating procedures as key determinants of whether a laboratory delivers reliable results at all [3], which is why data integrity sits at the center of modern laboratory quality management rather than at its margins. ### Attributable Every entry must trace to the person (or system) who performed the action and when. In microbiology terms, the person who read the plate, the person who confirmed the organism identification, and the person who released the susceptibility report must each be identifiable from the record itself. Shared bench logbooks with initials-only entries are a classic inspection finding, because initials are not attributable when two technologists share initials. In electronic systems, attribution means unique user accounts with password policies, no shared logins, and no technologist performing work under a supervisor's open session. Attribution also extends to systems: if an automated blood culture instrument flags a vial as positive, the record should show which instrument, which software version, and which flag, so the observation is attributable to a defined source rather than to a vague "the machine said so." ### Legible A record must be readable at the time of review, months or years later. Legibility failures in the microbiology lab are specific and predictable: copied colony counts between worksheet and LIMS entry, corrected Gram reactions written over with correction fluid, and direct antigen test cassettes photographed in poor light. A practical rule used on the bench is the "stranger test": a qualified reviewer who did not perform the work must be able to reconstruct exactly what was done and what was observed, without asking the original technologist. Illegibility is not a cosmetic issue. An unreadable transposition between aworksheet value and a reported zone diameter is indistinguishable from fabrication, because the reviewer cannot tell them apart. ### Contemporaneous The record must be made at the time the activity happens, not reconstructed afterward. Microbiology is unusually vulnerable to failures of contemporaneity because cultures run on long timelines. A technologist may read thirty urine plates at 8 a.m. and plan to enter all results "at lunch." If the entry happens at 3 p.m. from memory, the record is not contemporaneous, and the gap is where transcription errors and honest misremembering enter the system. Instrument-read methods remove part of this burden because the instrument time-stamps the reading automatically. Manual observations, such as catalase reactions, coagulase tubes, and microscopy fields, need a same-moment entry habit, whether on paper worksheets designed for real-time recording or in a point-of-care application that documents results as the test is performed [4]. ### Original The original record is the first capture of the observation, on paper or in an electronic system. A photograph of a MALDI-TOF target plate annotation sheet, the raw mass spectra file, and the instrument-generated report form a chain of originals. The most common originality failure in microbiology labs is treating a transcription as if it were the record: the value on the worksheet is the original, the value typed into the LIMS is a copy, and any discrepancy between them must be resolvable. Metadata is part of the original record. In genotyping workflows, experts drafting data standards for HLA laboratories have emphasized that sequence data and associated metadata, covering messaging, storage and retention, documentation, reporting, validation, and quality control, must be governed as a complete package, because a sequence without its metadata cannot be re-evaluated later [5]. ### Accurate Accuracy means the record reflects what actually occurred, at the correct value, with corrections handled transparently. In paper systems, a correction is made with a single line through the original entry, the correct value written adjacent, the corrector's initials and the date, and a brief reason. Correction fluid, overwriting, and erasure destroy accuracy because they make the original value unrecoverable. In electronic systems, accuracy is protected by the audit trail: the system retains the original entry, the change, the identity of the person making it, the timestamp, and the reason, with the earlier value remaining visible to authorized reviewers. For example, if a technologist transposes a ceftriaxone minimum inhibitory concentration of 2 mg/L as 4 mg/L, an honest correction preserves both values and the explanation. Silently overwriting the field destroys accuracy. ### Complete A complete record contains everything a qualified reviewer needs to reconstruct the activity, including failed attempts, repeats, deviations, and aborted analyses. This is the attribute that micro labs most often fail, and it fails for sympathetic reasons: a contaminated culture is repeated, the first plate is discarded, and only the repeat is recorded. The record is then incomplete, because a reviewer cannot tell how many attempts preceded the reported result. Completeness applies to quality control with special force. If an internal quality control run for a rapid streptococcal antigen test failed and the run was repeated, both control results belong in the record. Digital quality management platforms that centralize quality control processing exist precisely to capture these complete histories, including documentation, analytics, and approval steps, rather than leaving them on loose bench sheets [6]. ### Consistent The record must be internally consistent across time and across linked documents. Consistency failures in microbiology are chronology failures: a quality control result logged after the patient result that depended on it, an incubation start time recorded as earlier than the specimen receipt time, or an identification confirmed on a date before the subculture was made. The consistency check is arithmetic and chronological. A practical bench exercise is to pick one accession number and lay every document in order. Receipt time, plating time, incubation on and off times, identification and susceptibility dates, and report release must form a single plausible timeline. Electronic systems enforce this automatically with server-side timestamps, which is one reason regulated computerized systems require validated date and time functions that users cannot freely edit [2]. ### Enduring The record must survive, in readable form, for its defined retention period. Endurance failures are mundane: a bench printer with thermal paper that fades within a year, a shared network drive that loses files during migration, an old instrument's proprietary raw files that no current software can open. Enduring means verified readability over time, not simply storage. Retention planning includes periodic checks that archived files still open, migration of records from retiring systems before the system is decommissioned, and preservation of raw data, not only summarized reports. genotyping data standards work has highlighted storage and retention of large datasets as a first-class requirement for clinical laboratories, with recommendations written so that labs can meet them without unsustainable staffing or cost burdens [5]. ### Available The record must be retrievable by the people authorized to retrieve it, within the time the business or regulator needs. An archived box of worksheets in a basement that floods is not available. An audit-trail export that requires a software vendor visit is not available. Availability is tested in practice by timed retrieval exercises: the laboratory selects a random accession from two years ago and measures how long it takes to assemble the complete record, including quality control for that day. Availability also has an access-control dimension, because available does not mean open to everyone, and does not mean locked away from the microbiologist who needs it to answer a clinician's call about a patient from last year. ## Where Data Integrity Fails in a Microbiology Workflow ### Pre-analytical failures The pre-analytical phase generates the identifying data that every later record depends on. Failures here include mislabeled or unlabeled specimens, transcription of patient identifiers onto request forms away from the bedside, and broken chains of custody during transport. Standardized pre-analytical procedures for specimen collection, transport, and storage, combined with sample traceability recommendations, are explicitly built into ISO 15189-aligned laboratory protocols so that every downstream record can be anchored to a verified specimen identity [1]. Assisted reproduction laboratories, which operate under similar traceability demands for gametes and embryos, rely on documentation systems and electronic witnessing to guarantee that the record follows the physical sample through every stage [7], and the same witnessing logic applies to critical microbiology specimens such as tissue cultures from surgical cases. ### Analytical failures The analytical phase is where subjective readings and instrument interfaces concentrate risk. Examples that recur on audits include the following. 1. Transcription errors between worksheet, instrument report, and LIMS entry, particularly for susceptibility values that differ by a single dilution. 2. Instrument-to-LIS interfaces that silently truncate or reformat organism codes, so the reported identification differs from the instrument's original output. 3. Manual observations recorded in batch after the fact, breaking contemporaneity for catalase tests, oxidase strips, microscopy, and rapid antigen cassettes. 4. Repeated runs whose initial failures are never entered, breaking completeness. 5. Quality control data kept on separate, uncontrolled bench sheets rather than in the controlled record, which hides the timeline that consistency requires. Digital quality control platforms address this class of failure by centralizing quality control data processing and embedding documentation, analytics, and approval into one workflow, which removes the uncontrolled bench sheet from the system entirely [6]. ### Post-analytical failures The post-analytical phase covers report release, amendments, and the management of discovered errors. A released report with a wrong organism identification must be corrected through a documented amendment that preserves the original report, states the change, and triggers communication to the clinical team. Surgical pathology laboratories confronting exactly this problem for frozen sections built custom software to monitor discrepancies, increase documentation, categorize the source of error, and facilitate communication when an error is discovered, because unmanaged documentation of errors was the bottleneck [8]. The microbiology analog is the amended susceptibility report: integrity demands that the correction pathway be as controlled as the original release. ## Paper Records Versus Electronic Systems ![Comparison table of paper and electronic laboratory records across attribution, corrections, timestamps, retention and main weakness.](/article-images/illustrations/data-integrity-alcoa-microbiology-laboratory-paper-vs-electronic.webp) Paper and electronic systems fail in different ways, and a laboratory quality management plan has to address both. Paper is fully compliant with ALCOA+ when the system is designed well. Requirements include bound or controlled worksheets, indelible ink, real-time single-line corrections, controlled distribution of blank forms, and archival conditions that keep records legible and retrievable. Paper's weakness is effort: every attribute must be maintained by human discipline, every time. Electronic systems shift the integrity burden from human discipline to system validation. Regulated expectations include the following, drawn from the control framework used under 21 CFR Part 11 and aligned international standards [2]. - Unique, individually attributable user accounts with role-based permissions. - Audit trails that capture creation, modification, and deletion of records, with the identity of the actor and the reason for the change. - Validated date and time stamps that users cannot edit freely. - Controls over record retention, backup, and retrieval that preserve both the record and the audit trail for the retention period. - Risk-based validation and computer software assurance, in which the depth of testing matches the risk the system carries. Modern laboratory platforms increasingly unify the [laboratory information management system](/knowledge/bioinformatics/lims-explained-lab-information-management-systems) and the quality management system in one environment, so that standard operating procedures, quality control checkpoints, and deviation handling share the same controlled data backbone as the patient record [2]. Regional laboratory networks face the same problem at larger scale, and their informatics solutions centralize internal quality control and external quality assessment documentation and approval across multiple sites, so that a technologist at a community station and a supervisor at the central laboratory work from the same controlled record [6]. A second framework, FAIR (findable, accessible, interoperable, reusable), complements ALCOA+ by addressing how data are organized for reuse. An analysis of Food and Drug Administration Emergency Use Authorization diagnostics data found four recurring failure modes: test names too similar or inconsistently applied, inconsistent document identifiers and versioning, heterogeneous data structure and formatting, and poor traceability of amendments [9]. Every one of those failure modes has a microbiology laboratory analog. Local test names that drift from the LIS catalog, documents with no version number, validation summaries stored as free text, and amended reports with no link to the original all make data unfindable and unreusable, even when every ALCOA+ attribute was technically satisfied at capture. FAIR thinking tells the laboratory to govern naming conventions, identifiers, and versioning as data integrity issues, not as housekeeping. ## Microbiology Quality Control Depends on Data Integrity Internal quality control (IQC) is the set of controls run with patient specimens to confirm the analytical system is behaving on that day. External quality assessment (EQA), sometimes called proficiency testing, is the periodic testing of blinded reference materials from an outside body. Both generate data whose integrity requirements are as strict as patient results, for a simple reason: a quality control record that cannot be trusted cannot demonstrate anything. Consider what ALCOA+ demands of a single IQC record for a blood agar plate lot check. - Attributable: the technologist who subcultured the control strain and the one who read it are identified, not "QC done." - Legible: hemolysis reactions and colony morphology are written clearly enough to review at inspection. - Contemporaneous: control readings are entered at the reading, not reconstructed. - Original: the record includes the control strain's passage history and lot documentation, not just a pass or fail. - Accurate: a failed control is recorded as failed, with the corrective action, even if the lot was subsequently rejected. - Complete: failed and repeated controls both appear. - Consistent: the control result predates any patient results released on that lot. - Enduring and available: the record is retrievable for the retention period and readable without the original technologist. External quality assessment extends these requirements across institutions, because the submitting laboratory must be able to prove which specimen produced which result and how the result was transmitted. In a piloted external quality assurance program for SARS-CoV-2 antigen rapid diagnostic tests, all quality control and external quality assessment specimens were distributed to participating laboratories and results were submitted through a dedicated portal using QR code scanning, which tied every submitted result to a traceable specimen identity and transmission event [10]. The same traceability logic underlies microbiology quality control for any test modality, from rapid antigen cassettes to broth microdilution panels. The stakes are institutional as well as analytical. Multi-hospital studies of laboratory service quality identify quality management systems, standard operating procedure adherence, and documentation practice as the determinants that separate reliably performing laboratories from poorly performing ones, alongside equipment maintenance and workflow organization [3]. Data integrity is the mechanism by which those determinants become visible: a laboratory that cannot trust its records cannot measure its own performance, and a laboratory that cannot measure its own performance cannot improve it. ## How Data Integrity Is Audited and Observed in Practice Data integrity is not verified by a single test. It is observed through a structured set of activities, repeated on a schedule and after changes. ### Data flow mapping The laboratory diagrams, in words or a controlled table, every place a data point exists: where it is created, every system it is copied into, who can edit it, and how long each copy is kept. The map almost always reveals uncontrolled copies, the shadow spreadsheet on a shared drive, the printed instrument report filed in a personal folder, the photograph of a cassette on a personal phone. Each uncontrolled copy is a break in the original record's chain of custody. ### Walkthrough of one accession ![Flowchart of a positive blood culture accession from receipt through release and amendment, with the quality control check branching before release.](/article-images/illustrations/data-integrity-alcoa-microbiology-laboratory-accession-walkthrough.webp) Auditors pick a single accession and follow it end to end. A useful worked example is a positive blood culture from receipt to final report. 1. Specimen received, with receipt time and condition documented against the label. 2. Bottle flagged positive by the instrument, with instrument identity and time stamp. 3. Gram stain read and result entered contemporaneously. 4. Subculture plated, with media lot recorded. 5. Organism identified, with instrument output retained as the original. 6. Susceptibility panel performed, with the reading technologist identified. 7. Quality control for the susceptibility system confirmed for that day, chronologically before release. 8. Report released by an identified, authorized person. 9. Any amendment after release preserved with the original visible. Every step must pass all nine ALCOA+ attributes. One missing link, such as a susceptibility reading with no identifiable reader, becomes the focus of the audit. ### Timed retrieval exercises The laboratory randomly selects an old accession and assembles the complete record, including that day's quality control, while a reviewer clocks the process. This tests availability and endurance simultaneously and exposes records that survive only in formats no current system can open. ### Continuous monitoring Between audits, integrity is monitored through the systems themselves. Automated quality assurance software that tracks discrepancies between initial and final diagnoses, categorizes their source, and documents the communication generated in response gives pathology laboratories a living measure of documentation behavior rather than a snapshot [8]. Integrated quality control information systems that combine internal quality control, external quality assessment, documentation, analytics, and approval workflows give supervisors dashboards over the same dimension across a network [6]. Monitoring of point-of-care testing platforms likewise treats documentation and quality assurance as continuous functions embedded in the testing application, not as periodic paperwork [4]. ## ALCOA+, Data Quality, and FAIR: Terms Often Confused | Term | What it actually means | Scope | Most common misconception | |--|--|--|--| | Data integrity | The record is complete, consistent, accurate, and trustworthy across its life cycle, and this can be demonstrated | Applies to every record, paper and electronic, patient and quality control | Believing integrity is an electronic-systems issue that paper records are exempt from | | Data quality | The measurement itself is analytically sound, precise, and fit for purpose | Applies to the analytical result | Treating a validated assay as sufficient, when the record around it can still fail | | ALCOA+ | The nine attributes used to judge whether a record meets integrity expectations | Applies to individual records and the systems that hold them | Reading it as a checklist of software features rather than a description of evidence | | ALCOA++ | ALCOA+ plus traceability of data across systems and versions | Applies where data move between systems | Assuming it applies only to genomics, when any interfaced report needs it | | FAIR principles | Data should be findable, accessible, interoperable, and reusable, with governed naming, identifiers, versioning, and structure | Applies to datasets and documentation packages | Confusing FAIR with integrity, since a perfectly FAIR dataset can still contain untrustworthy entries | | Audit trail | The system record of who changed what, when, and why, with prior values preserved | Applies to electronic records | Believing the audit trail is an optional log, rather than the mechanism that makes electronic corrections honest | ## Common Mistakes and Limitations These are the failures seen most often on the bench and in inspections, stated as practical points. 1. Assuming the LIMS guarantees integrity. The LIMS holds the record, but data integrity is created by the bench behavior around it. A perfectly designed system fed by reconstructed entries produces a perfectly formatted untruth. 2. Shared logins. A single account opened all day at the bench destroys attribution for every entry made under it. This is the single most common electronic failure in small laboratories. 3. Batch entry from memory. Reconstructing a morning of manual readings at the end of the shift converts contemporaneous records into testimony. If a load-shedding event or staffing crunch makes same-moment entry impossible, the delay itself must be documented. 4. Correction fluid and overwriting. Any correction that hides the original value is indistinguishable from falsification. Single-line-through corrections with initials, date, and reason exist for exactly this reason. 5. Discarding failed runs without recording them. The repeated culture, the failed control, the contaminated subculture: all belong in the record. Completeness means the reviewer sees the attempts, not only the success. 6. Treating quality control records as less important than patient records. Regulators and accreditators read them with equal weight, and a released patient result whose control record is missing or late is indefensible no matter how well the assay performed. 7. Ignoring metadata. An organism identification without the method, version, and instrument context is not a complete record. Data standards work in clinical genotyping, where sequence data and metadata must be governed together across messaging, storage, documentation, reporting, validation, and quality control, sets the pattern for all high-complexity microbiology data [5]. 8. Naming drift and versionless documents. Local test names that do not match the catalog, SOPs without version numbers, and amended reports with no link to the original recreate exactly the four failure modes documented in large public diagnostics datasets, including inconsistent naming, inconsistent identifiers and versioning, heterogeneous formatting, and untraceable amendments [9]. 9. Believing that cloud hosting equals endurance. Storage that is not periodically verified for readability and retrievability is a hazard wearing the costume of a solution. Endurance is a tested property, not a purchase. 10. Assuming integrity scales automatically. In a study of determinants of laboratory service quality, adherence to standard operating procedures, documentation practice, and quality management emerged alongside equipment reliability and workflow organization as the factors that decide whether results are trustworthy, which means integrity practices must be resourced and trained like any other core activity, not delegated to goodwill [3]. A note on scope: this article describes record and system practices, and individual patient cases always require interpretation by the responsible clinician together with the laboratory's professional staff. ## Quick Review - ALCOA+ means attributable, legible, contemporaneous, original, and accurate, plus complete, consistent, enduring, and available. - Data integrity is provable trust in the record. Data quality is soundness of the measurement. A laboratory needs both, and neither substitutes for the other. - Contemporaneity is the attribute microbiology fails most, because long culture timelines invite batch entry from memory. - Corrections must preserve the original value. Single-line-through on paper, audit trail in electronic systems. Correction fluid and silent overwrites are integrity failures by definition. - Quality control records carry the same ALCOA+ obligations as patient records, and a control result must be chronologically consistent with any patient result it supports. - Electronic systems must provide unique attribution, audit trails, protected timestamps, and validated retention, consistent with the risk-based controls expected under 21 CFR Part 11 and aligned international standards. - Availability and endurance are tested properties: timed retrieval of an old accession and verified readability of archived files are the practical checks. ## Frequently Asked Questions ### What does ALCOA+ stand for in the microbiology laboratory? ALCOA+ stands for attributable, legible, contemporaneous, original, and accurate, with the plus adding complete, consistent, enduring, and available. Together, the nine attributes describe the evidence standard a laboratory record must meet so that a qualified reviewer, at any later date, can trust and reconstruct the work it documents. ### Is ALCOA+ a regulatory requirement or just good practice? ALCOA+ is both. It began as the vocabulary of regulatory inspection for good manufacturing practice records and is now embedded in the expectations of 21 CFR Part 11 for electronic records, ISO 15189 for medical laboratories, and ISO/IEC 17025 for testing laboratories, so accredited and regulated laboratories must be able to demonstrate it. Laboratories outside formal accreditation still adopt it because it is the most practical available definition of a trustworthy record. ### What is the difference between data integrity and data quality? Data quality describes whether the measurement itself is analytically sound and fit for purpose. Data integrity describes whether the record of that measurement is complete, consistent, accurate, and trustworthy across its life cycle. A validated assay with a reconstructed, unattributable worksheet has quality without integrity, and a beautifully documented record of a mislabeled specimen has integrity of the wrong thing. ### How does an audit trail work in a laboratory information system? An audit trail is a system-generated record of every creation, modification, and deletion of a data record, capturing the prior value, the new value, the identity of the user, the timestamp, and the reason for the change. The original entry is never overwritten or hidden, so an honest correction is transparent and a silent alteration is structurally impossible for authorized users to make. ### Can paper records satisfy ALCOA+? Yes. Paper records satisfy ALCOA+ fully when the system is designed for it, using controlled worksheets, indelible ink, real-time single-line-through corrections with initials, date, and reason, and archives that keep records legible and retrievable for the retention period. Electronic systems make several attributes, especially attribution and the audit trail, automatic rather than effortful, but the standard itself does not require electronics. ### How often should a microbiology laboratory check its data integrity? There is no single mandated interval. A common structure is a self-inspection of data flows and record practices on a defined annual or semiannual cycle, continuous monitoring through audit trails and quality assurance software between inspections, and a focused review whenever a system changes, a deviation occurs, or a new test modality is introduced. What matters is that the checks are scheduled, documented, and acted on, so that integrity is measured rather than assumed. ## Sources 1. [Integrated laboratory protocol for the diagnosis of Sexually Transmitted Infections (STIs): Standardized pre-analytical procedures, rapid screening, hemagglutination, and ELISA methods for use in resource-limited settings.](https://pubmed.ncbi.nlm.nih.gov/42085387/) 2. [A cloud-enabled LIMS-QMS platform with large language model-assisted standard operating procedure generation for GLP/GMP-aligned theranostics laboratory operations.](https://pubmed.ncbi.nlm.nih.gov/42901382/) 3. [Determinants of quality laboratory service provision amongst government comprehensive specialized hospitals in Northwest Ethiopia.](https://pubmed.ncbi.nlm.nih.gov/41923262/) 4. [Feasibility, Preference, and Impact of a Rapid Multiplexed Point-of-Care Digital Innovation (AideSmart!) to Expedite Screening of Sexually Transmitted Blood-Borne Infections in At-Risk Populations in Canada: Cross-Sectional Study.](https://pubmed.ncbi.nlm.nih.gov/39423369/) 5. [Recommendations for HLA Genotyping Data Standards and Clinical Laboratory Staffing Considerations.](https://pubmed.ncbi.nlm.nih.gov/42007733/) 6. [Development and Implementation of an Internal Quality Control and External Quality Assessment Information System for a Regional Medical Laboratory Center: Pilot Design and Implementation Study.](https://pubmed.ncbi.nlm.nih.gov/41364922/) 7. [Quality Control in Assisted Reproduction Centers.](https://pubmed.ncbi.nlm.nih.gov/42441885/) 8. [Software solution for integration of frozen section quality assurance into daily practice.](https://pubmed.ncbi.nlm.nih.gov/39838842/) 9. [Challenges in Structuring SARS-CoV-2 Emergency Use Authorization Diagnostics Data: Leveraging Findable, Accessible, Interoperable, and Reusable Principles for the Next Era of Diagnostic Data Transparency.](https://pubmed.ncbi.nlm.nih.gov/41386638/) 10. [Piloting an external quality assurance program (QAP) for COVID-19 antigen rapid diagnostic tests (Ag-RDTs): findings and lessons learned from Cambodia.](https://pubmed.ncbi.nlm.nih.gov/39762057/)