Lab Quality Control Software in Nigeria: Levey-Jennings Charts and Westgard Rules Explained

The chemistry analyzer had been running fine for months. Then, gradually, its glucose control values started creeping toward the upper 2SD line — not failing outright, just a little higher each week. Nobody flagged it, because the paper QC sheet only had a pass/fail column and a set of reference ranges taped to the wall. Nothing on that page connected today's control value to last month's. The drift wasn't caught by a supervisor reviewing trends; it was caught three weeks later, when a proficiency-testing panel came back flagged and a clinician called asking why a patient's result didn't match their clinical picture. That's the gap Levey-Jennings charting and Westgard rules exist to close — and it's a gap paper logs are structurally bad at closing, no matter how conscientious the staff filling them in are.

Quick Answer

A Levey-Jennings chart plots daily quality control values against the control material's established mean and standard deviation (±1SD, ±2SD, ±3SD) over time, turning a column of QC numbers into a visual trend an analyst can read at a glance. Westgard rules are a standard set of statistical decision criteria — 1_3s, 2_2s, R_4s, 4_1s, 10x, and related rules — applied together (a "multirule" procedure) to decide whether a run is in control, because a single fixed limit either misses real errors or triggers too many false rejections. Internal QC (IQC) checks day-to-day precision using control materials run alongside patient samples; external quality assessment (EQA/proficiency testing) periodically checks that a lab's results agree with peer laboratories, which IQC alone can't confirm — both are expected for MLSCN-compliant and ISO 15189-aligned quality systems. QC management software makes the trends and rule violations visible automatically instead of requiring someone to notice a slow drift buried in rows of numbers. ClinikEHR Diagnostics lists Levey-Jennings and Westgard rule support on its Professional Lab tier, but this feature is currently rolling out, not fully live — the rest of this article explains the methodology itself, which is useful regardless of which software a lab eventually chooses.

What a Levey-Jennings Chart Actually Shows

A Levey-Jennings chart is a control chart — a specific application of Walter Shewhart's statistical process control concept to the clinical laboratory, first proposed by Stanley Levey and E.R. Jennings in 1950. The mechanics are straightforward once you've seen one, but the value is entirely in what it makes visible over time:

• The x-axis is time or run number — each day's (or each run's) control result plotted in sequence, usually enough points to cover a full month at a glance.
• The y-axis is the control value, scaled around a mean established for that control material and analyte, with reference lines at ±1SD, ±2SD, and ±3SD. Under a roughly normal (Gaussian) distribution, about 68% of control values should fall within ±1SD, about 95% within ±2SD, and about 99.7% within ±3SD — so a point outside ±2SD is already statistically unusual, and a point outside ±3SD is rare enough to demand a hard look.
• Points are connected in sequence, which is the entire point of the chart. A single out-of-range value is one kind of problem. A run of values steadily climbing toward the +2SD line over ten days — each one individually "passing" — is a different, more dangerous kind of problem: a slow analytical drift that a pass/fail checklist will never surface, because every individual day technically passed.

That last distinction is why a chart, not a table, matters. Reading a QC log as a list of numbers tells you whether today passed. Reading it as a chart tells you whether the analyzer is trending toward a failure that hasn't happened yet.

Westgard Rules: Why a Multi-Rule System Catches What One Limit Misses

Once QC values are plotted, the next question is: what counts as "out of control," and who decides? A single fixed limit — reject anything outside ±2SD — sounds reasonable but has a well-documented statistical problem: with a 2SD cutoff, the false rejection rate on a genuinely good run is roughly 9% for 2 control levels, climbing to 14-18% as more control levels are checked per run. That means a lab checking multiple analytes and control levels using a flat ±2SD rule will reject good runs constantly — leading staff to start ignoring warnings, which defeats the point. Move the limit out to ±3SD to cut false alarms, and the lab starts missing real, clinically significant errors instead.

The Westgard multirule procedure, developed by James Westgard, resolves this by applying several rules together rather than one flat limit — a run is rejected if any rule in the set is triggered:

• 1_3s — one control value exceeds the mean ±3SD. A rejection rule with a low false-alarm rate (~1%), sensitive mainly to random error.
• 1_2s — one value exceeds ±2SD. In the classic multirule design this is a warning rule, not an automatic rejection: it triggers a closer look using the rules below, rather than stopping the run outright.
• 2_2s — two consecutive control values exceed the same ±2SD limit (both high or both low). A rejection rule that flags systematic error, such as a shift in calibration.
• R_4s — within the same run, one control exceeds +2SD while another exceeds −2SD. This rule is interpreted within-run only, and it flags an increase in random error (widening imprecision) rather than a directional shift.
• 4_1s — four consecutive values exceed the same ±1SD limit (all on one side). A rejection rule for smaller, sustained systematic shifts that a single ±2SD check wouldn't catch on any given day.
• 10x (with 8x and 12x variants depending on how many control levels are run) — ten consecutive values fall on one side of the mean, regardless of how close to it. This catches a systematic shift so gradual that no single value ever crosses a warning limit — exactly the kind of slow drift described in the opening scenario.

The logic is that each individual rule is tuned to be fairly specific (low false-rejection rate on its own), but together they cover both random error (R_4s, 1_3s) and systematic error (2_2s, 4_1s, 10x) far better than any one limit could alone. Some multi-material QC schemes add analogous rules — 2of3_2s, 3_1s, 6x, 9x — scaled for three control levels instead of two. The underlying principle is the same regardless of exactly which rule set a lab adopts: combine several statistically distinct checks so real errors get caught without drowning staff in false alarms.

Internal QC vs External Quality Assessment — Why a Lab Needs Both

Levey-Jennings charts and Westgard rules operate on internal quality control (IQC) — control materials run alongside patient samples, on the lab's own schedule, checked against the lab's own established mean and SD. IQC is frequent (typically daily, or every run) and tells you whether today's testing is consistent with how the same analyzer and reagent lot performed yesterday and the day before.

What IQC cannot tell you is whether your lab's results agree with everyone else's. A glucose analyzer can be perfectly precise — every control value landing right on the mean, day after day — while still being biased in a way that only shows up when compared against an external reference. That's the role of external quality assessment (EQA), also called proficiency testing (PT): a lab receives blinded samples from an external provider, tests them exactly like patient specimens, and submits results for comparison against peer laboratories using similar methods. EQA is periodic (often quarterly) rather than daily, and it's retrospective — you find out afterward how you compared, not in real time. Because IQC and EQA test different things (internal consistency versus external agreement), accreditation frameworks like ISO 15189 and inspecting bodies expect to see documented evidence of both, not one substituting for the other. This is also why MLSCN and ISO 15189 auditors specifically ask to see EQA participation history going back 12-24 months, alongside daily IQC records — a point covered in more depth in our MLSCN and ISO 15189 readiness guide.

Why Paper QC Logs Make Trend-Spotting Hard in Practice

None of the statistics above are new or exotic — Levey-Jennings charting and Westgard rules have been standard clinical laboratory science for decades, and any trained medical laboratory scientist knows the theory. The practical failure point in a lot of Nigerian labs isn't understanding the method; it's that a paper logbook or a static spreadsheet makes the method exhausting to apply consistently, day after day, analyte after analyte:

• Trend-spotting requires manually re-plotting or eyeballing a column of numbers. A 10x shift is invisible in a list — every value in the run looks unremarkable on its own. Catching it requires someone to actually chart the last ten points and notice they're all above the mean, which rarely happens under routine workload pressure.
• Multirule logic is tedious to apply by hand across many analytes. Checking one control value against six different rule conditions, for every analyte, every run, is exactly the kind of repetitive statistical check that's easy to skip when the bench is busy — which means violations get missed until a much larger failure forces the issue.
• Historical mean and SD get stale. Control ranges should be re-established periodically (new control lot, after major maintenance), but on paper that update is easy to forget, leaving a lab charting against limits that no longer reflect current analyzer performance.
• Retrieving QC history for an audit or investigation is slow. When a result gets questioned, tracing back through weeks of logbook pages to reconstruct whether the analyzer was in control that day is exactly the kind of paper hunt that turns a five-minute question into a half-day search.

None of this means the underlying science is hard. It means paper is the wrong medium for applying it consistently at real-world lab volumes — which is the specific gap QC management software is meant to close.

Where ClinikEHR Diagnostics' QC Management Stands Today

Being direct about this: ClinikEHR Diagnostics lists "Quality control management — Levey-Jennings charts, Westgard rules" on its Professional Lab tier ($76/month, ₦120,000/month), but that feature is explicitly flagged rolling out — it is not a finished, fully live capability yet. If automated Levey-Jennings charting and Westgard rule evaluation is a hard requirement for your lab right now, confirm current status directly before assuming it's available today.

What is fully live, and genuinely useful in its own right for a QC-conscious lab, includes:

• Chain-of-custody timeline and full audit trail on every specimen (Starter Lab and above) — every handoff from registration through result release, timestamped and attributable, which matters when investigating whether a questioned result traces back to a sample-handling issue rather than an analytical one.
• Turnaround-time (TAT) dashboard — per test, per site, per scientist (Professional Lab) — a different but related quality signal: not whether the analyzer is in statistical control, but whether results are moving through the lab on schedule.
• Reagent and consumables inventory with batch/lot tracking and expiry monitoring (Starter Lab and above) — relevant to QC in a direct way, since a reagent lot change is exactly the kind of event that should trigger re-establishing control means and SDs, and lot-level traceability is what lets you connect a QC shift back to a specific reagent batch.
• Advanced audit-log reporting, accreditation-ready exports (Business Lab) — general documentation infrastructure that supports an MLSCN or ISO 15189 audit trail, independent of whether Levey-Jennings charting specifically has finished rolling out.

The honest framing: the QC methodology described above is worth understanding and applying regardless of which software a lab uses, and a mature LIMS should eventually automate it. Where ClinikEHR Diagnostics is today is "rolling out" that specific capability, with several other genuinely live features already supporting the broader quality and audit-readiness picture.

Frequently Asked Questions

What's the difference between a Levey-Jennings chart and a Westgard rule? They're complementary, not competing. The Levey-Jennings chart is the visual format — control values plotted against mean ±1/2/3SD over time. Westgard rules are the decision logic applied to those plotted values (or to the underlying numbers) to decide whether a run should be accepted or rejected. In practice, labs plot a Levey-Jennings chart and then apply Westgard rules against it, rather than choosing one or the other.

Which Westgard rule actually rejects a run, versus just warning? In the classic multirule design, 1_2s (one value beyond ±2SD) is a warning rule — it triggers closer inspection using the other rules rather than an automatic rejection. 1_3s, 2_2s, R_4s, 4_1s, and 10x are rejection rules: if any one of them is triggered, the run is considered out of control and results are held pending investigation, not released.

Why not just use a single ±2SD or ±3SD limit instead of multiple rules? Because each option fails differently at scale. A flat ±2SD limit produces a high false-rejection rate on genuinely good runs — roughly 9% at two control levels, rising to 14-18% as more levels are checked — which trains staff to start ignoring warnings. A flat ±3SD limit lowers false alarms but misses real, clinically meaningful errors that a tighter check would have caught. Combining several statistically distinct rules (each individually fairly specific) catches more real error types while keeping false alarms manageable.

Is ClinikEHR's Levey-Jennings and Westgard rules feature live right now? No, not fully. It's listed on the Professional Lab tier but explicitly flagged as rolling out — meaning it's on the roadmap and in progress, not a finished feature you can rely on today. Confirm current status directly with ClinikEHR before planning around it for an imminent need.

Does internal QC (IQC) alone prove a lab's results are accurate? No. IQC proves your testing is consistent with itself over time — today's control value matches yesterday's and last week's, within statistical limits. It doesn't prove your results agree with other laboratories using similar methods, which is what external quality assessment (EQA/proficiency testing) checks separately, on a periodic, blinded basis.

How often should EQA/proficiency testing be done, versus daily IQC? IQC typically runs daily, or with every batch of patient samples, since it needs to catch problems before results go out the door. EQA is periodic — commonly quarterly — since it's an external, retrospective comparison rather than a real-time check. Accreditation bodies and MLSCN inspectors generally want to see 12-24 months of EQA history alongside continuous IQC records.

Can a small lab apply Westgard rules without dedicated QC software? Yes, in principle — the methodology predates software entirely and can be applied with a manual chart and a checklist. The practical barrier isn't the math, it's consistency: manually checking every analyte's latest value against several rule conditions, every run, without missing a slow trend like a 10x shift, is genuinely hard to sustain by hand at real lab volumes, which is exactly the gap QC software is meant to close once it's fully built out.

What should a lab do about QC management while waiting for automated Levey-Jennings/Westgard tools to fully roll out? Keep running IQC and charting manually or in a spreadsheet in the meantime — the statistical discipline matters regardless of tooling, and MLSCN/ISO 15189 don't require any specific software, just documented, defensible QC practice. Track other genuinely live infrastructure, like reagent lot tracking and chain-of-custody records, since a QC shift is often traceable to a reagent or specimen-handling event those features already capture today.

Conclusion

Levey-Jennings charts and Westgard rules aren't new or exotic methodology — they're decades-old, well-documented clinical laboratory science, and the theory is already familiar to any trained medical laboratory scientist. What's hard in practice is applying that methodology consistently, run after run, analyte after analyte, using a paper logbook that was never designed to make a slow ten-point drift visible. That's the specific gap QC management software exists to close, and it's worth understanding the underlying statistics well regardless of which system a lab eventually adopts.

Key takeaways:

• A Levey-Jennings chart plots control values against mean ±1/2/3SD over time, making trends and drift visible in a way a pass/fail table cannot.
• Westgard rules (1_3s, 2_2s, R_4s, 4_1s, 10x, and related variants) combine multiple statistical checks because any single fixed limit either misses real errors or triggers too many false alarms.
• IQC and EQA test different things — internal day-to-day consistency versus external agreement with peer labs — and accreditation frameworks expect documented evidence of both.
• Paper QC logs make multirule logic and trend-spotting genuinely hard to sustain at real lab volumes, independent of staff competence.
• ClinikEHR Diagnostics' Levey-Jennings/Westgard rules QC management is currently rolling out on the Professional Lab tier, not yet fully live — chain-of-custody tracking, TAT dashboards, and reagent inventory with lot tracking are live today and support the broader QC picture in the meantime.

Explore ClinikEHR Diagnostics to see what's live today across chain-of-custody, TAT tracking, and reagent inventory in detail.

Not sure where your lab stands? Talk to a consultant for free, personalized guidance.

Related Resources:

• MLSCN and ISO 15189 Readiness: How LIMS Software Helps You Pass Audits
• Sample Chain of Custody in Nigerian Laboratories
• How to Reduce Turnaround Time in a Diagnostic Lab in Nigeria
• What Is a LIMS? A Diagnostic Lab Guide for Nigeria
• Top 5 Laboratory Management Software in Nigeria