GLI GLI Quality Tool
GLI Quality Tool — Version 2.0

Designing a TB laboratory data system that tracks results over time

A tuberculosis laboratory data management system should do more than store a final positive or negative result. It should preserve the history of each specimen, test method, quality control event, result review, amendment, and report issued to a clinician or surveillance program. When these links remain intact, laboratories can identify delayed reporting, recurring test failures, changing resistance patterns, and weaknesses in workflow.

The design must fit the laboratory’s operating environment. A national reference laboratory may use a laboratory information management system with instrument interfaces, while a smaller facility may depend on controlled spreadsheets, paper registers, or a shared local database. The technology can differ, but the principles of traceability, confidentiality, accuracy, availability, and controlled access remain the same.

The GLI Quality Tool offers a practical framework for building these principles into a quality management system. Its phased roadmap and Quality Systems Essentials can help a laboratory connect data practices with document control, personnel responsibilities, equipment management, assessment, and continual improvement.

Define the questions the system must answer

Begin with decisions rather than software. Identify what the laboratory needs to know about patients, specimens, tests, results, and service performance. Typical questions include whether a specimen was received, when testing began, which platform was used, whether a result was verified, when it was reported, and whether a corrected report was later issued.

For longitudinal tracking, define the unit of record carefully. A patient may provide several specimens during one diagnostic episode and may return for follow-up testing months later. A single specimen may undergo microscopy, molecular testing, culture, identification, and drug susceptibility testing. The system should connect these events without replacing one result with another.

Establish a data dictionary before creating forms or database fields. Define terms such as “not detected,” “invalid,” “contaminated,” “no growth,” “rifampicin resistant,” and “testing not performed.” Record which fields are mandatory, which values are permitted, and who is responsible for entering or validating them. Clear definitions prevent different facilities from treating similar outcomes as different indicators.

Build a traceable data model

A useful structure separates patient information, specimen information, testing events, reported results, and quality records. A patient identifier should not be the only key. Each specimen needs a unique accession number, and each test event should retain the method, instrument, reagent lot where relevant, operator, date, and review status.

The model should preserve relationships across time. For example, a culture isolate may be linked to the original specimen and later to identification or susceptibility testing. If a result is amended, the original value, reason for amendment, authorizer, date, and new value should remain visible in an audit history. Deleting the earlier result makes it impossible to investigate reporting errors or explain changes to surveillance data.

Use structured fields wherever possible. Drop-down values and controlled codes improve consistency, while free-text comments can capture unusual circumstances. Date and time fields should use a consistent format, and the system should distinguish collection time, receipt time, test start time, verification time, and report time. These timestamps make turnaround-time analysis meaningful.

Choose technology that matches the workflow

Select the simplest platform that can meet the laboratory’s requirements reliably. A low-resource setting may begin with a protected workbook or offline database, provided it has unique identifiers, validation rules, user permissions, version control, backups, and a documented review process. A larger service may need an LIS connected to analyzers, barcode printers, electronic request systems, and national reporting platforms.

Before implementation, map the actual workflow from request to report. Observe how specimens arrive, how labels are checked, where results are recorded, and how urgent findings are communicated. This often reveals practical issues that a software demonstration misses, such as intermittent connectivity, shared user accounts, duplicate registers, or instruments that export incomplete data.

Design area Minimum control Useful enhancement
Identification Unique patient and specimen identifiers Barcode-based accessioning
Result entry Required fields and permitted values Direct analyzer interface
Review Named verifier and approval timestamp Electronic authorization workflow
Amendments Reason and history of every change Locked audit trail
Security Role-based access and password control Single sign-on and access monitoring
Continuity Scheduled backup with restore testing Redundant servers and offline recovery
Reporting Standard reports and export rules Automated dashboards and alerts

Test the system with realistic scenarios before going live. Include repeat specimens, rejected specimens, invalid molecular tests, contaminated cultures, mixed growth, delayed results, instrument downtime, amended reports, and patients with multiple episodes. A system that works only for uncomplicated negative results will create hidden workarounds and unreliable records.

Protect data quality, privacy, and continuity

Data quality should be checked at the point of entry and through routine review. Validation rules can prevent impossible dates, missing specimen types, incorrect identifier formats, or a resistance result entered without the corresponding test method. Supervisors should periodically compare electronic records with source documents, instrument printouts, request forms, and issued reports.

Assign responsibilities using a written data management procedure. Analysts, bench staff, supervisors, information technology personnel, and quality officers need clearly defined roles for entry, verification, correction, backup, access approval, and incident response. Training should cover both the technical steps and the reason each field matters to patient care and surveillance.

Confidentiality controls are essential because TB records may contain sensitive clinical and identifying information. Use individual accounts, least-privilege access, screen privacy, secure storage, and defined retention periods. Avoid sending identifiable result files through unsecured personal messaging channels. Where data are shared externally, document the purpose, recipient, fields included, and authorization.

Continuity planning should address power interruptions, connectivity loss, hardware failure, malware, and staff absence. Decide how work will be recorded during downtime and how backlogged entries will be reconciled without creating duplicates. Backups should be scheduled, protected from unauthorized access, stored separately when feasible, and tested through an actual restoration exercise.

Monitor performance across time

A longitudinal system becomes valuable when it supports regular analysis. Track the interval from specimen collection to receipt, testing, verification, and reporting. Stratify turnaround time by facility, test type, specimen category, urgency, and month. A rising delay may indicate transport problems, reagent shortages, staffing gaps, instrument downtime, or a bottleneck in result authorization.

Monitor quality indicators alongside results. Useful measures include specimen rejection, invalid molecular tests, contamination, recollection rates, amended reports, missing identifiers, transcription errors, and the proportion of results released within the target time. Trends are more informative than a single monthly percentage, particularly when volumes vary.

Interpret trends with operational context. A sudden increase in rifampicin resistance findings may represent a true epidemiological change, a new testing algorithm, improved referral, or a coding error. Data review should therefore bring together laboratory staff, clinicians, surveillance personnel, and quality managers when appropriate. Every indicator needs a defined numerator, denominator, data source, reporting frequency, and person responsible for action.

Govern changes and verify the system

Software updates, new assays, revised algorithms, instrument replacements, and changes to report wording can alter the meaning of historical data. Use change control to document what will change, why it is needed, who approved it, how users were trained, and how the transition will affect trend analysis. Retain old codes or create a crosswalk when definitions change.

Review user access at scheduled intervals and whenever staff move roles or leave the laboratory. Investigate unusual edits, unexplained missing records, duplicate accessions, and late data entry. Internal audits should test whether the system follows approved procedures, while management reviews should use findings to assign corrective actions and deadlines.

Peer review can strengthen the procedures around data entry, result verification, amendments, and reporting. Guidance on reviewing TB laboratory SOPs can help laboratories compare written instructions with actual practice and identify inconsistent steps before they affect patient results.

Put the system into routine practice

Implementation works best as a controlled rollout rather than a one-time installation. Start with a limited set of tests or one work area, measure errors and turnaround times, then adjust fields and procedures before expanding. Keep a register of issues raised by users, decisions made, and changes introduced during the pilot.

Training should be role-specific and practical. Bench staff need to register specimens and record testing events; reviewers need to authorize or amend results; supervisors need to interpret indicators; and administrators need to manage access and backups. Competence can be assessed through observed tasks and review of completed records, not attendance alone.

Use available quality management checklists to organize implementation evidence and routine assessments. Keep approved forms, data dictionaries, access lists, backup logs, audit findings, and corrective-action records together so that the laboratory can demonstrate control of its information processes.

The strongest system is one that staff can use consistently during busy periods and unusual events. Define a small set of essential fields, automate checks where possible, and review the system’s workload as testing volume changes. Each month, select a few records and follow them from request through final report, confirming that the data tell a complete and accurate story.

A well-designed TB data system turns individual test results into dependable evidence for patient management, laboratory improvement, and public health action. Begin by mapping the workflow, approving the data dictionary, assigning responsibilities, and testing a realistic pilot. Then use routine trend review and corrective action to keep the system accurate as methods, technologies, and service needs evolve.