Quality control that catches a bad run before it reaches a patient
The most dangerous result a lab can produce is a wrong one that looks right. Quality control is the safety net that catches the bad run before a clinician ever trusts it.
Judging a control by a single limit either floods the bench with false alarms or misses a slow drift entirely. The fix is a set of rules, evaluated automatically, on a chart nobody has to draw.
Every laboratory that runs quality control faces the same dilemma. Set the acceptance limit too tight and the bench is flooded with false rejections, every ordinary day-to-day wobble triggers an alarm, and staff learn to wave the alarms away. Set it too loose and a slow, steady drift sails through unnoticed until the analyser is reading meaningfully wrong. A single limit cannot do both jobs at once. It either cries wolf or it sleeps through the real problem.
The discipline that solves this is well established, and it has two parts. A Levey-Jennings chart plots each control result over time against the mean and the standard deviations around it, so a run can be read in context rather than as an isolated number. Westgard rules are a set of statistical tests applied to that chart: instead of one crude pass-fail line, a combination of rules distinguishes a genuine error from harmless random variation. The problem has never been the method. The problem is doing it by hand.
On paper, this discipline is hard to sustain. The realities of a busy bench work against it.
So the method that was designed to catch real error gets reduced to a ritual: plot the dot, glance at it, move on. The statistical power is on the page, but it is not being used.
Veona Lab Quality Control runs the full Westgard multi-rule set automatically as each control is captured. The familiar rules all fire on their own evidence: a single point beyond two standard deviations as a warning, a point beyond three as a rejection, two consecutive points on the same side beyond two deviations, a range spanning four deviations across levels, four results trending one-sided, and ten consecutive points on one side of the mean. Each control is plotted on its Levey-Jennings chart per analyte and level the instant it arrives, and the verdict comes back as a clear accept or reject, with the exact rule that fired named. Nobody draws the chart. Nobody carries the rules in their head. The warning is told apart from the rejection, and a slow trend is caught long before any single point breaks a limit.
The rules were never the hard part. Applying all of them, correctly, to every control, in real time, was. That is the part a system should do.
The most useful thing the multi-rule approach buys you is nuance. A single point drifting past two standard deviations is a warning, a signal to look closer, not necessarily a reason to throw the run out. A true rejection requires stronger evidence: a point far out, a pair confirming each other, a trend that cannot be coincidence. By telling these apart, the laboratory stops over-reacting to ordinary variation and stops under-reacting to genuine drift. That balance is the whole reason the method exists, and it is the reason a single limit can never replace it. When a rule does signal a true rejection, the run is held and the follow-through begins, which we cover in corrective actions that close the loop.
For a Nigerian laboratory working toward ISO 15189, or climbing the SLMTA ladder toward measurable quality improvement, an assessor will look closely at how quality control is evaluated. Hand-plotted charts and a single acceptance limit are not just laborious; they are weak evidence. An assessor wants to see that the full multi-rule logic is applied consistently, that warnings and rejections are distinguished, and that the evaluation happens in real time rather than retrospectively. A laboratory where every control is charted and rule-evaluated automatically can show exactly that, on demand, for any analyte, on any day. The statistical rigour the standard expects is simply how the bench already works.
That same rigour is why a result released on an ordinary morning can be trusted: the run behind it passed the full rule set, not a single forgiving line. We make the broader safety case in catching a bad run before a patient.
See the full Westgard set fire on a live Levey-Jennings chart, with the rule that caught it named. Book a demo and we will show you QC evaluated the moment the control lands.
The most dangerous result a lab can produce is a wrong one that looks right. Quality control is the safety net that catches the bad run before a clinician ever trusts it.
Turnaround is the complaint every lab hears and few can answer with a number. Here is how per-test targets and breach monitoring turn it from anecdote into something you can actually manage.
Catching a failed control is the easy half. The hard half is proving you found the cause, fixed it, and verified the fix. That is what a corrective-action workflow is for.
We will tailor a demo to how your hospital, clinic, or lab actually runs, offline behaviour, payments, reporting, and all.