GLI GLI Quality Tool
GLI Quality Tool — Version 2.0

Using control charts to monitor TB test precision over time

Reliable tuberculosis testing depends on results that remain consistent when the same material is tested repeatedly. A laboratory may have competent staff, validated procedures, and suitable instruments, yet gradual changes in reagent performance, calibration, environmental conditions, or technique can still affect repeatability. Control charts make those changes visible before they become a serious quality problem.

A control chart is a time-based display of internal quality control results. It compares each new control measurement with an established average and expected variation. Used correctly, it helps a TB laboratory distinguish ordinary random fluctuation from a meaningful shift or trend in testing performance.

The method is especially useful when incorporated into a broader quality management system. The GLI Quality Tool provides a practical roadmap for laboratories developing that system across phases and Quality Systems Essentials, including equipment, personnel, documents, assessment, and continual improvement.

Define what precision means for the test

Precision describes how closely repeated measurements agree with one another under defined conditions. It is different from accuracy, which concerns how close a result is to a true or accepted value. A test can be precise but inaccurate if it produces tightly grouped results around the wrong target.

The appropriate precision indicator depends on the TB test. For a quantitative assay, such as a molecular test that reports a numerical concentration or cycle threshold, laboratories may monitor the mean, standard deviation, and coefficient of variation. For smear microscopy, a laboratory may use prepared control slides or blinded rechecking systems, although a conventional numerical chart may be less suitable. For culture or molecular platforms that produce qualitative outcomes, charts can track repeated positive and negative control performance, while invalid rates and contamination rates provide additional surveillance signals.

Before creating a chart, define the measurand, control material, testing frequency, instrument or method, and acceptable response. A control chart is meaningful only when the laboratory knows exactly what each plotted result represents and when the testing conditions are sufficiently consistent.

Select controls and establish a baseline

Internal quality control material should challenge the test system without creating unnecessary risk. For molecular TB testing, laboratories may use manufacturer-provided controls, characterized external materials, or stable in-house materials where appropriate and permitted. The control should be handled in the same workflow as patient specimens whenever possible, while preserving the manufacturer’s instructions and biosafety requirements.

A baseline should be established using enough control results collected under stable operating conditions. The laboratory calculates the average result and the standard deviation from that initial dataset. These values become the center line and the basis for warning and action limits. A small or unstable baseline can produce misleading limits, so laboratories should document how many observations were used and why the dataset was considered representative.

The baseline should not be changed simply to make an outlying result appear acceptable. Recalculate it only after a documented review shows that a sustained process change has occurred, such as a new reagent lot, instrument replacement, method modification, or major maintenance event. Keep the previous baseline available so historical performance remains traceable.

Build a chart that staff can interpret

A basic Levey–Jennings chart places control results on the vertical axis and the date, run number, or sequence of tests on the horizontal axis. The center line represents the established mean. Additional lines are drawn at one, two, and three standard deviations from the mean. For some qualitative assays, the chart may instead display the number of successful control runs, invalid events, or a proportion of expected control outcomes.

The chart should identify the assay, platform, control level, lot number, units, baseline period, and limits. It should also record the operator or testing site when those factors are relevant. Electronic systems can automate calculations, but a paper chart can work well in a resource-limited setting if entries are timely, legible, and reviewed.

Chart signal What it may indicate Appropriate response
One result outside ±2 SD Possible random or emerging error Review the run, control preparation, transcription, and recent changes
One result outside ±3 SD Likely significant problem Reject or hold affected results according to procedure and investigate
Two consecutive results beyond the same ±2 SD limit Systematic shift Check calibration, reagent lot, equipment, and procedure
Six or more results steadily rising or falling Gradual trend Examine deterioration, temperature, maintenance, and operator technique
Eight or more results on one side of the mean Sustained shift Reassess the baseline and investigate a process change
Unusual spread around the mean Increased imprecision Check pipetting, control stability, instrument function, and environmental conditions

These rules are aids to judgment, not substitutes for laboratory procedures. The laboratory should select a manageable set of rules and define them in its quality documents. Applying every possible rule can create excessive false alarms, especially when only a few control results are generated each week.

Investigate signals before releasing results

When a chart rule is triggered, first check for simple recording or clerical errors. Confirm the control identity, result transcription, units, date, instrument, reagent lot, expiration date, and operator. Review whether the control was mixed, stored, thawed, or reconstituted correctly. A problem in preparation can mimic an analytical failure.

Next, examine the run and the surrounding results. Determine whether the event is isolated or part of a shift, trend, increased scatter, or repeated failure. Review calibration status, maintenance records, temperature logs, environmental conditions, pipettes, amplification curves, and instrument error messages as applicable. For a biosafety-sensitive TB laboratory, the investigation must also follow approved procedures for handling potentially infectious materials.

Do not automatically repeat testing until an acceptable result appears. Repeating can conceal the original signal and may waste limited resources. The laboratory should define when results are held, when controls are repeated, when a run is rejected, and when patient specimens must be retested. Every decision should be recorded with the finding, root cause or probable cause, corrective action, and verification that the problem was resolved.

Connect charts with the quality system

Control charts are most valuable when they feed into routine management rather than remaining as isolated graphs. A supervisor or quality officer should review them at a defined frequency, such as each run, weekly, or monthly depending on test volume and risk. Low-volume laboratories may need longer review intervals, but they should still record every control event and avoid treating sparse data as proof of stability.

The chart should connect to other quality indicators. For example, a rise in control variability may occur alongside increasing invalid results, delayed turnaround time, equipment downtime, contamination, or staff changes. Looking at these indicators together can reveal a system issue that a single chart cannot explain.

A strong continuous improvement culture helps staff report signals without fear of blame. The goal is to identify process weaknesses early, learn from them, and prevent recurrence. When several laboratories contribute data to a network, standardized chart definitions and review criteria also make performance comparisons more meaningful.

Adapt monitoring to resource settings

A laboratory does not need specialized software to begin. A ruled paper chart, spreadsheet, or simple quality record can support the essential functions: plotting results, identifying limits, documenting reviews, and tracking actions. Where internet access or automation is limited, establish a backup method and keep records protected from damage or unauthorized changes.

Use separate baselines when conditions differ in ways that affect results. A new instrument, reagent lot, control level, test algorithm, or major procedural change may require verification before routine monitoring resumes. However, splitting data too often can hide a real problem. The decision should be based on documented technical reasoning rather than convenience.

Staff training should include both chart construction and interpretation. Personnel need to know what to do when a warning signal appears, who can authorize result release, and where to document an investigation. Periodic competency assessment can include a simulated control failure, allowing staff to practice recognizing the signal and following the response procedure.

Practical habits that strengthen chart review

Review performance and improve the process

A control chart should be periodically evaluated for usefulness. Ask whether the limits still reflect current stable performance, whether the chart detects meaningful changes, and whether staff receive feedback quickly enough to act. If the chart produces frequent warnings without identifiable problems, the baseline, control material, testing frequency, or decision rules may need technical review.

When a genuine shift or trend is found, look beyond the immediate failed run. A recurring issue may reflect inadequate training, unclear instructions, poor storage, unstable power, insufficient preventive maintenance, or procurement problems. Corrective action should address the cause at the appropriate level and include a measure of effectiveness.

Laboratories can also use chart data during management reviews, internal audits, and external assessments. Trends over several months may support decisions about equipment replacement, refresher training, stock management, or workflow redesign. Feedback from staff using the chart can be submitted through the GLI feedback page, helping improve practical resources for laboratories in varied settings.

Begin with one high-priority TB assay, one clearly defined control, and a documented baseline. Plot every result, review signals consistently, and use investigations to strengthen the process rather than merely explain failures. As the routine becomes reliable, extend the approach to other assays and quality indicators, making precision monitoring a normal part of safe, dependable laboratory service.