Guest Essay

We're delighted to have Jagdish Chandarana, MS, contribute a guest article on post-analytical metrics, assessed on the Six Sigma scale. Often the post-analytical phase is considered the least error-prone, but sometimes generalities don't prove true for individual labs.

Extending Six Sigma to the Post-Analytical Phase: A CAPA Bundle — Anchored by a Cloud-LIS Migration — That Eliminated the Hidden “Vital Few”

September 2026
Jagdish Chandarana, BSc — Founder & Managing Director, Vedant Diagnostic Center, an NABL / ISO 15189–accredited clinical laboratory in Rajkot, India

A gap, stated precisely

Six Sigma entered quality management as a way of counting defects, and counting defects is naturally aligned with the pre- and post-analytical phases of the total testing process, where errors present as discrete, countable events. The framework has in fact a longer history contemplating these phases than the analytical phase, where the Sigma metric was adapted to variables data through imprecision and bias. Studies of quality indicators across the non-analytical phases of the total testing process are plentiful, and analytical Sigma-metric articles — including the ones our own laboratory has relied on for years — are plentiful too.

The gap our laboratory encountered is narrower and more specific: fewer studies of post-analytical phase indicators have been assessed on the Six Sigma scale. Yet the total testing process ends only when a correct report reaches the clinician who ordered it, on time, through whatever software and infrastructure stands in between. That last stretch is where our laboratory found its largest hidden error source, and it is where a structured, multi-component CAPA programme, measured on the same DPMO and Sigma scale, was associated with an 83% error reduction over twelve months. This article summarizes the practical lessons; the full study has been peer reviewed and accepted at eJIFCC, scheduled for Issue 5, October 2026.

The setting and the question

Vedant Diagnostic Center is a stand-alone, medium-volume, NABL / ISO 15189–accredited laboratory in western India. Like most accredited laboratories, we had mature analytical-phase quality control: daily IQC, EQAS participation, Sigma-metric review. What we did not have was a quantitative picture of the post-analytical phase — the delivery of verified results to clinicians — expressed on the same Sigma scale we already used everywhere else, with a baseline, a prioritization, and a way to prove that an intervention worked.

Our approach was deliberately unoriginal, and that is the point: we simply gave the post-analytical phase the same treatment the framework was built on — defect counting. Every documented post-analytical error over a 12-month baseline was logged, classified into six failure modes, and expressed as defects per million opportunities against the volume of reports issued. DPMO converts to a Sigma value; a Sigma value identifies whether performance is acceptable; and category-level counts identify the “vital few” failure modes that deserve the intervention budget.

What the baseline revealed: the LIS was the vital few

The baseline classification produced a result we did not expect. The single largest error category — 41.7% of all post-analytical errors — was failures of the laboratory information system itself: our locally maintained, on-premise LIS server. Not transcription, not interpretation, not communication of critical values — the infrastructure carrying the reports.

This finding embarrassed our own risk assessment, and I want to be candid about that, because the lesson is transferable. In our prospective FMEA exercise, LIS failure had been scored as “unlikely.” The retrospective error log told a different story: roughly twenty LIS-related events per year. The gap arose because FMEA probability scores were assigned from team impression, and team impression is systematically kind to infrastructure that fails in small, frequently-recovered ways. Each individual server hang felt like a minor nuisance resolved by a restart; collectively, they were the dominant error source. The corrective is simple and worth adopting anywhere: calibrate FMEA probability scores against a retrospective error log before trusting them, and re-score annually.

The LIS category also displayed a pattern that inflates its true impact: single-event-multiple-error behavior. One server crash does not delay one report; it delays every report in the queue. Infrastructure failure modes therefore punch above their event count, and — the optimistic corollary — infrastructure fixes deliver outsized Sigma gains.

The intervention: a coordinated CAPA bundle, engineered where it mattered most

The intervention was not a single fix. A structured CAPA programme was designed across all six post-analytical failure modes and deployed as a coordinated bundle, so that the post-intervention period reflected steady-state operation of every control in combination — process-level controls on five failure modes, and an infrastructural intervention on the sixth and largest.

That infrastructural component was the principal one, and it followed the human-factors hierarchy: forcing functions and engineered controls outperform training and vigilance. We migrated the LIS from the locally maintained on-premise server to a centrally managed, redundant cloud-based LIS, paired with dual-broadband connectivity and automatic failover, so that neither a server fault nor a single connection outage could stop report delivery. A run chart tracked monthly post-analytical error counts to confirm the effect was sustained rather than a launch-month artifact.

One methodological point deserves candor, because our peer reviewers rightly pressed on it: in a bundled before-and-after design, the headline reduction belongs to the bundle, not to any single component. The place to look for component-specific signals is the category-level Pareto — and that is where the cloud migration earns its billing, as the next section shows.

The result, in Sigma terms

Following the CAPA bundle, and across approximately 42,000 reports, documented post-analytical errors fell from 24 to 4 — an 83% reduction — and post-analytical Sigma performance improved from approximately 4.55 to 5.05. That headline belongs to the bundle as a whole. The component-specific evidence sits at category level: the LIS/software category, the pre-intervention vital few at 41.7% of all errors, went from 10 events to zero and stayed there — the direct signature of the migration and failover architecture. The same arithmetic that laboratories already use for HbA1c or glucose — DPMO, Sigma, category Pareto — measured all of it, before and after, with no new statistical machinery.

Five lessons for laboratories

1. The Sigma scale unifies the total testing process. Defect counting quantifies post-analytical performance exactly as it does elsewhere in the laboratory, and gives the same decision-grade output: a number that says whether you have a problem and whether your fix worked.
2. Infrastructure can be the dominant, invisible error source. In our laboratory the LIS itself was the single largest contributor to post-analytical error. If your error log does not have an infrastructure category, your classification scheme may be hiding your vital few.
3. Prospective FMEA underestimates what it has learned to live with. We scored LIS failure “unlikely” while experiencing it roughly twenty times a year. Recalibrate FMEA probability scores against retrospective error logs — impressions absorb chronic small failures until they disappear from view.
4. Single-event-multiple-error patterns make infrastructure fixes high-yield. One crash delays many reports, so eliminating one infrastructure failure mode buys more Sigma than its event count suggests.
5. A transferable sequence exists — as a bundle, not a silver bullet. Log post-analytical errors retrospectively; classify and Pareto them; deploy CAPA across every failure mode, anchoring the largest with engineered controls — for us, migration to a managed, redundant cloud LIS with independent dual connectivity and automatic failover; sustain process-level controls on the remaining failure modes; and monitor with a run chart.

Closing thoughts

The Six Sigma framework earned its place in the laboratory by making quality measurable, comparable, and improvable. Our experience is that assessing post-analytical indicators on the Sigma scale repays the effort — and that within a structured CAPA programme, the highest-return single component may not be a new analyzer or a new control rule, but the unglamorous infrastructure carrying the reports. When we gave the post-analytical phase its Sigma numbers, they told us exactly where to act — and how much of the credit each action could honestly claim.

Reference

Chandarana J, Butani K. Reducing post-analytical errors through cloud-based information system migration in a stand-alone accredited laboratory. eJIFCC. In press; scheduled for Issue 5, October 2026. Citation details to be updated upon publication.

About the author

Jagdish Chandarana, BSc, is the founder and Managing Director of Vedant Diagnostic Center, an NABL / ISO 15189–accredited clinical laboratory in Rajkot, Gujarat, India. He is first and corresponding author of peer-reviewed studies on laboratory quality management published in Archives of Pathology & Laboratory Medicine, Laboratory Medicine, the Indian Journal of Clinical Biochemistry, and the Journal of the American Nutrition Association, and serves as an invited peer reviewer for eJIFCC. Contact: This email address is being protected from spambots. You need JavaScript enabled to view it. | ORCID 0009-0002-8796-6432.