How to read a Levey–Jennings chart and recognise shifts and trends

A Levey–Jennings chart turns internal quality-control results into a visual history of analytical performance. A value that appears acceptable alone may form part of a shift, trend or wider pattern requiring investigation.

Reading it involves more than asking whether today's point is inside a limit. Examine its distance from the mean and relationship with earlier results. Approved QC rules determine whether the pattern is a warning, rejection or observation.

The chart signals that something may have changed. It does not identify the cause or replace professional judgement.

1. Anatomy of the chart

A Levey–Jennings chart normally shows:

• The horizontal axis: date, time, analytical run or control event in sequence.
• The vertical axis: the measured value of one control material for one analyte, often displayed as the actual result or its distance from the target mean.
• The mean: the expected centre of the control results.
• Control lines: commonly the mean plus and minus one, two and three standard deviations.
• Plotted points: each control result, usually connected to make changes over time easier to see.

Use an appropriate chart for each analyte, control level and examination system. Combining different control lots, levels, methods or instruments without a defined transition can create a misleading pattern.

The control lines are not patient reference intervals, clinical decision limits or allowable patient-result ranges. They describe the expected behaviour of the control process and support the laboratory's statistical QC decisions.

2. Mean and control limits

The centre line and standard deviation determine how every point appears and must be established or verified through the approved process.

The mean represents the central value expected for that control under stable operating conditions. Standard deviation describes the observed spread around it. A small standard deviation produces closely spaced control lines; a larger standard deviation produces wider lines.

Manufacturer ranges or peer-group information may help when introducing a control, but are not automatically the laboratory's working limits. Follow the manufacturer's instructions, method requirements and approved target-setting procedure.

Do not widen limits or recalculate the mean to make failures disappear. First assess the control lot, initial dataset, method precision and analytical system.

3. Plotting control results

Plot results in production order. Retain the original even when a repeat is permitted; replacing it with an acceptable repeat removes evidence.

For each point, ask three questions:

1. How far is it from the mean?
2. Is it consistent with the recent spread of results?
3. Does it form a sequence with earlier points?

A point within two standard deviations may complete a sequence above the mean, while one near a limit may be isolated. The chart provides context; approved rules provide the action.

4. Random variation

Random error usually appears as unpredictable scatter. It may produce one unusually high or low point, a large difference between control levels in the same run, or a broader spread without a consistent direction.

Possible causes include pipetting or reconstitution error, bubbles, intermittent aspiration, temperature, electrical disturbance or control-material problems. Investigation should follow the method, instrument and event history.

Do not treat every isolated point as proof of random error, and do not repeat the control until it passes without understanding the reason. The result must be interpreted using the laboratory's accepted rule set and documented troubleshooting process.

5. Recognising a shift

A shift is a sequence of control results that moves to one side of the established mean and remains there. The individual points may still lie within broad control limits, but their sustained position suggests that the system's centre has changed.

A shift may appear suddenly after calibration, a reagent or calibrator lot change, maintenance, control-lot transition or a change in instrument response. The date of the first shifted point should be compared with these recorded events.

The number of consecutive points required to trigger a shift rule is not universal. Different procedures and rule systems may use different sequences. Laboratory staff should not invent a rule while reviewing the chart; they should apply the approved rule consistently.

6. Recognising a trend

A trend is a progressive movement in one direction across successive control results. Each result is higher than the preceding result, or each is lower, over the sequence defined by the laboratory's rule.

Trends may develop gradually because of reagent deterioration, calibration drift, electrode ageing, contamination, environmental change or another evolving condition. A trend can be important before any single point crosses a rejection limit.

Normal random variation can occasionally resemble a short upward or downward run. That is why the laboratory must specify how many successive points constitute a trend signal and what review or action follows.

A shift describes movement to a new level; a trend describes continuing movement in a direction. Both are commonly associated with systematic change, but the chart alone cannot establish the cause.

7. Applying approved QC rules

Levey–Jennings charts are often interpreted using single or multirule QC procedures. Commonly used rules may examine:

• one point beyond a defined standard-deviation limit;
• consecutive points beyond the same limit on one side of the mean;
• a large within-run difference between control results;
• several results on the same side of the mean; or
• a defined sequence that rises or falls continuously.

Named rules such as 1-2s, 1-3s, 2-2s, R-4s, 4-1s and sequence rules are widely recognised, but their purpose and application differ. For example, a 1-2s observation may be treated as a warning within a multirule procedure rather than an automatic rejection. R-4s is designed to identify a large difference within the relevant run, not simply any two distant points on a long chart.

The laboratory should select rules appropriate to the method's performance, number of controls, frequency and analytical quality requirement. Applying every available rule can produce excessive false rejection; applying too few or unsuitable rules can miss relevant error. The current applicable edition of CLSI C24 provides professional guidance on statistical QC design, while the original multirule QC paper explains the basis of the approach.

8. Documenting review and action

Review should remain traceable even when the result is accepted. Useful records connect:

• control result, lot, level, analyte, instrument, date, time and operator;
• mean, standard deviation, limits and rule configuration in effect;
• warning or rejection generated and the sequence that caused it;
• review decision, reviewer and time;
• investigation, repeat testing and corrective action;
• linked calibration, maintenance, reagent or environmental events; and
• assessment of potentially affected patient results and authorisation to resume reporting, where required.

Periodic review should look beyond individual failures. Recurring shifts after calibration, frequent warnings, widening variation or repeated changes after a particular reagent lot may reveal a system issue that daily review alone misses.

ISO 15189:2022 specifies requirements for quality and competence in medical laboratories. The WHO Laboratory Quality Management System handbook provides a practical quality-system foundation. Laboratories must use current controlled standards, manufacturer instructions, approved procedures and applicable accreditation requirements. Indian laboratories should verify the latest relevant NABL documents before publication or implementation.

How a LIMS can help

A LIMS can display control results over time, connect them with control and reagent lots, apply configured rules, record review status and retain investigation and corrective-action history. It may also provide alerts or restrict actions, depending on verified product behaviour and laboratory configuration.

Software does not decide which QC rules are scientifically appropriate or determine the cause of a pattern. These remain laboratory responsibilities.

Frequently asked questions

What is the simplest way to read a Levey–Jennings chart and recognise shifts and trends?

Read each point against the mean and standard-deviation lines, then examine its sequence with earlier results. A sustained move to one side suggests a shift; a progressive rise or fall suggests a trend. Apply the laboratory's approved rules before deciding action.

Which part of the laboratory workflow is responsible?

The chart supports internal quality control of the analytical examination process. Authorised technical staff plot or verify results, while designated reviewers interpret exceptions and approve action according to the laboratory procedure.

What records should remain available for review?

Retain the original control results, chart, material and lot details, targets and limits, rule configuration, flags, reviews, investigations, repeats, corrective actions and any required patient-result assessment and authorisation.

Where can a LIMS help, and where is professional judgement still required?

A LIMS can plot results, identify configured patterns, connect related records and improve traceability. Laboratory professionals must select the rules, interpret signals, investigate causes, assess patient-result impact and authorise action.