Continuous method performance verification (CMPV) is the ongoing, statistically supported review of analytical method performance during routine use. Within the broader framework of Statistical Quality Control, CMPV can be considered an application of Statistical Process Control (SPC), whereas acceptance sampling addresses the separate question of whether a defined lot or batch should be accepted based on sampled results. CMPV uses routine analytical data to determine whether method performance remains stable and subject only to expected, common-cause variation (often estimated from short-term (within subgroup) variation).
Control charts may include both warning and action limits. For continuous data, warning limits are often positioned at approximately two standard deviations (2SD) from the established center line, while action or critical limits are commonly positioned at approximately three standard deviations (3SD). The SD in this case is typically based on the short-term variation estimate of the standard deviation. Depending on the specific requirements and historical performance of the method, laboratories may deviate from these conventional 2SD/3SD limits, provided the selected limits remain scientifically justified.
In alignment with ICH Q14 and USP <1220>, CMPV extends analytical procedure lifecycle management beyond the discrete validation stage. During validation, intermediate precision studies assess variability across factors such as analysts, a limited number of days, instruments, and reagent conditions using a controlled study design. Matrix effects are also evaluated as part of method validation where relevant. Validation therefore establishes that the procedure is suitable for its intended purpose under defined conditions, but it does not fully describe how its performance may behave over months or years of routine use.
A validated method may remain in a validated state while its routine performance shows a statistical shift or drift. A shift is a relatively sudden change in the level or distribution of results often triggered by the introduction of a new reagent lot, an equivalent replacement instrument, a new chromatographic column, or another change in consumables or materials. Conversely, a drift is a gradual change over time and may result from instrument ageing, progressive column deterioration, reagent degradation, contamination through moisture uptake, or another slowly developing condition. Seasonal environmental effects may also contribute, although temperature and humidity controls are expected to mitigate minimize their influence.
Analyst-to-analyst variability should already be represented in the intermediate precision assessment, and the sample matrix would not normally change during routine testing; where such a change does occur, it would typically be addressed through additional validation or revalidation. However, routine monitoring may still identify performance patterns associated with differences in technique, insufficient procedural control, an unexpected change in sample characteristics, or operating conditions that were not adequately represented during validation.
CMPV places these observations within a lifecycle context and helps distinguish common-cause variability, which represents the expected background variation of a stable method, from special-cause variability associated with an identifiable change or emerging performance issue. Under the lifecycle principles described in ICH Q14 and ICH Q2(R2), this approach supports the early detection of shifts and drifts before they develop into out-of-trend or out-of-specification events. Beyond its scientific and regulatory purpose, CMPV serves practical operational functions – for example, flagging when a chromatographic column or other reagent is failing before they compromise method outcomes.
What is continuous method performance verification (CMPV)?
Continuous method performance verification (CMPV) monitors analytical method performance through statistical trending, predefined control criteria, and, where established, formal review intervals and escalation triggers. These reviews may be performed monthly, quarterly, or annually, and may also be initiated when a warning limit, action limit, adverse trend, or defined event is observed. In this way, CMPV supports the ongoing assessment of whether the method remains in a state of control during routine GMP use and provides performance data following the initial validation study. It also captures variability arising from consumables, instrumentation, and environmental conditions, which can inform practical decisions such as when to replace a chromatographic column or certain reagents. Within analytical method lifecycle management, CMPV data can be reviewed alongside validation knowledge and, where relevant, information from routine testing, change control, and investigations.
Common-cause variability is the normal, expected variation inherent to a method when the procedure is working as intended. Special-cause variability is unusual variation associated with a specific event, condition, or change. The appropriate response depends on whether the cause is expected, acceptable, and adequately controlled.
Where a special-cause signal can be linked to a planned or technically justified change, such as the introduction of an equivalent instrument, replacement of a chromatographic column, or another approved change in materials or equipment, an extensive root-cause investigation may not be necessary. Such changes can legitimately alter the process mean, variability, or both. However, the laboratory should perform and document an appropriate impact assessment. The scope and rigor of the risk assessment, and any resulting actions, should be proportionate to the product lifecycle phase and associated risk, rather than uniformly extensive. Depending on the nature and risk of the change, this may include a bridging study, partial or full method validation, method transfer activities, or other measures defined through the change control system.
Where the control chart remains useful after the change, the data may be stratified according to instrument, column, site, reagent lot, or another relevant factor. Stratification prevents data generated under meaningfully different operating conditions from being interpreted as though they came from one homogeneous population. Where the change establishes a new and stable operating condition, new control limits may need to be justified using sufficient post-change data. In practice, the historical control limits may initially be retained after a shift and refined once sufficient post-change data becomes available.
If the special-cause signal cannot be attributed to an expected or acceptable factor, or when the signal is unexplained and could plausibly stem from issues such as reagent failure, reagent degradation, contamination, instrument malfunction, or an uncontrolled procedural change, a formal investigation should be initiated to determine the underlying cause. The investigation should identify the root cause, assess any effect on previously reported results or product quality, and determine whether corrective or preventive actions are required.
CMPV therefore does not treat every special-cause signal in the same way. Expected changes are addressed through impact assessment, change control, and appropriate statistical treatment, while unexplained or undesirable signals require a more thorough investigation.
Why continuous method performance verification matters
Method validation provides the evidence that an analytical procedure is suitable for its intended purpose. As part of validation, intermediate precision studies characterize the precision of the analytical procedure under varying conditions that may include different analysts, days, instruments, and reagent lots. Intermediate precision is therefore an essential component of method validation, but it is designed as a controlled study and is not intended to evaluate method performance over months or years of routine use.
Matrix effects are also evaluated as part of method validation where relevant. The sample matrix would not normally change during routine testing; where such a change does occur, it would typically be addressed through additional validation or revalidation.
The main value of CMPV is its ability to assess analytical performance over an extended period. By reviewing routine data as it accumulates, CMPV can identify gradual trends, drifts, or shifts that may not be observable during a time-limited validation study.
Statistical trending turns routine analytical results into evidence of continued method control. It can show whether performance remains consistent with the variability established during validation and can identify early signs of change before they result in repeated out-of-trend signals, system suitability failures, or out-of-specification results.
CMPV data can also support CAPA and change control decisions and provide useful evidence during quality reviews, inspections, and regulatory discussions.
Key performance parameters monitored during CMPV
The variables selected for trending should reflect how the method behaves during routine use. Typical examples include results from control samples, reference materials, system suitability tests, calibration standards, and other checks that are specific to the method performance. The aim is to choose parameters that can show an emerging change in performance before it potentially develops into a formal failure.
The monitored parameters could include both continuous and discrete data. In the case of continuous data, parameters such as precision, recovery, resolution, retention time, response factor, calibration curve slope, and assay control performance could be monitored. In the case of discrete or event-based data, parameters such as invalid runs, failed plates, repeated injections, system suitability failures, and rejected runs could be monitored. Taken together, these results provide a practical view of how the method behaves in routine use and can highlight changes that would be easy to miss when results are considered separately.
For HPLC and UPLC assay or purity assessments, laboratories often trend retention time, peak shape, resolution, detector response, and relative response factors. In PCR methods, the slope of the standard curve can be used to assess amplification efficiency. The linearity of the standard curve may be evaluated using the coefficient of determination, R², or the Pearson correlation coefficient, r, depending on the intended purpose of the analysis and the laboratory’s established procedure.
When a method operates over a narrow measurement range, a numerically small shift may represent a proportionally larger change relative to that range. Whether such a change is analytically or practically relevant depends on factors such as the expected common-cause variability, whether the shift coincides with a known and expected change (e.g. new column batch), and whether the result falls within or outside the established control limits. Its significance should therefore not be judged by magnitude alone but assessed against the established control limits and any applicable control rules (e.g. Westgard rules), in addition to predefined acceptance criteria agreed upon with the sponsor.
Electrophoresis methods have their own relevant indicators, such as resolution time, ladder migration, and control fragment sizing. Biological assays require a different set of measures. Cell-based assays may monitor four-parameter logistic (4PL) curve parallelism, reference response, and plate variability. ELISA trending may include sample recovery, blank response, standard curve fit, and control precision, whereas for flow cytometry one may follow mean fluorescence intensity (MFI), coefficient of variation (CV), or other method-specific control responses.
Event-based signals should also be included in the CMPV review, because invalid runs, failed plates, repeated injections, system suitability failures and similar events often provide early evidence of performance instability. However, such signals should not be used in establishing the control chart limits given their tendency to widen control limits and hence reduce control chart sensitivity. Reviewed together, these data support continued method verification and show whether the method still fits its validated state.
Drift, shift, and out-of-trend patterns
Trending examines whether routine method variability remains stable over time. Useful indicators include absolute or relative standard deviation, range of control sample results across replicate assay runs, difference between replicates, and, for instance for plate-based assays, plate-to-plate variability. When reviewed alongside the repeatability and intermediate precision data generated during validation, these results show how the method performs over an extended period of routine GMP use.
CMPV may reveal three main types of patterns:
Drift is a gradual change in method performance, seen as a steady increase or decrease in measured values over time. It may develop slowly because of factors such as instrument ageing, chromatographic column deterioration, or buffer degradation.
Shift is an abrupt change in the process mean, variability, or both. A shift may follow an identifiable change, such as the introduction of a new instrument, chromatographic column, or critical reagent lot. Where the cause is expected and documented, the laboratory should assess its impact and determine whether the data should be stratified or new control limits established.
Out-of-trend (OOT) refers to an individual result or pattern that differs statistically from what would be expected based on historical performance, without an immediately identifiable assignable cause. An OOT result can still fall within the established specification or acceptance criteria. Even so, it should be investigated to determine whether it is simply random variation or an early sign of a developing analytical issue.
A method may continue to meet its acceptance criteria while showing a drift, shift, or unusual increase in variability. CMPV helps identify these patterns early and supports a response that is appropriate to the nature of the signal and the associated risk.
System suitability performance
Trending system suitability parameters across runs can reveal shifts or drifts before individual runs fail the predefined SST criteria. SST determines whether an individual run is suitable for its intended analysis. If the SST criteria are not met, the run is rejected; depending on when the failure occurs, samples may not be analyzed, or results generated in that run cannot be accepted. CMPV instead reviews performance data and related events across multiple runs over time to identify emerging patterns. SST acceptance criteria and statistical limits used for CMPV therefore serve different purposes and should not be compared simply as tighter or wider. While CMPV data is typically not used to reject assay runs, CMPV-derived limits may additionally be incorporated into SST criteria, allowing rejection of assay runs that behave abnormally in ways not captured by other SST criteria.
Out-of-trend (OOT) evaluation
OOT evaluation matters in analytical lifecycle monitoring because unusual results may be seen before an OOS result is recorded. A control chart shows where the result sits in relation to the historical mean, expected variability, warning limits, and action limits. This allows the laboratory to judge whether the result reflects common-cause variation or points to special-cause variation that needs further review.
For continuous data, warning limits are often positioned around two standard deviations (2SD) from the established center line, and action limits around three (3SD). The standard deviation is typically estimated from short-term (within-subgroup) variation. Method-specific limits may differ when scientifically justified by the procedure’s requirements and historical performance.
The review should look at both the size of the change and the way it develops over time. A single recovery result near a warning limit may simply reflect random variation. Repeated results of the same kind within a short period, however, may be an early sign of a developing problem. In this context, frequency refers to how often the same type of unusual result or control-chart signal occurs over a defined number of runs or within a defined review period.
A practical review might examine whether retention time has moved steadily in one direction over several batches, whether control recovery has repeatedly approached a warning limit, or whether invalid runs have become more common since a new reagent lot was introduced. For example, a retention time-shift that persists across multiple batches of material points away from the material itself as root cause and rather toward a shared factor such as the chromatographic column. The laboratory should then assess the method risk, check for an assignable cause, and decide whether continued monitoring, an impact assessment, or a targeted investigation is appropriate.
Statistical trending tools in analytical lifecycle management
Statistical tools convert routine analytical data into practical process understanding. Over time, organized trending can show whether a method remains stable, becomes more variable or begins to shift away from historical behavior.
Trend analysis and data visualization
Trend analysis shows how method performance develops over time. For an overview of factors that may influence these patterns, the reader is referred to the section on System suitability performance.
It is important to distinguish method precision from the dispersion observed in routine trending data. Method precision is established during validation through planned studies of repeatability and intermediate precision. In CMPV, dispersion refers to the spread seen in results generated during routine use. Both may be expressed through standard deviation or variance, but they come from different datasets and answer different questions. Validation defines expected performance under controlled conditions, while routine-data dispersion shows whether that performance remains stable during ongoing use.
Line plots and run charts are useful for an initial review because they can reveal trends, unusual patterns, changes in variability, and possible outliers without the need for formal control limits.
JMP for analytical trending and lifecycle monitoring
JMP statistical analysis can support trend visualization, variability analysis, capability assessment and lifecycle monitoring in an interactive way (e.g. via the control chart builder platform). Statistical platforms support CMPV when they are built into a broader lifecycle approach with defined parameters, review frequency, clearly assigned ownership (i.e. who is responsible for performing and interpreting the trending), and escalation paths.
Common challenges in continuous method performance verification
CMPV implementation often faces practical hurdles like inconsistent data collection, limited historical baselines, and the difficulty of setting meaningful control limits that balance false alarms against missed drift. Correctly interpreting trends is another challenge, as normal biological variability can look alarming while special-cause issues require careful tracking. Furthermore, organizations frequently struggle to define optimal review frequencies and establish clear ownership across teams. To be truly effective, CMPV must overcome these operational barriers and integrate seamlessly into existing quality systems such as deviations, CAPAs, and change controls.
Regulatory expectations for analytical lifecycle monitoring
Regulators look well beyond the initial validation report. This is reflected in ICH Q14, ICH Q2(R2), and USP <1220>, which place analytical procedures within a lifecycle framework. The method has to remain understood and controlled after validation, approval or transfer, not only at the point it was first qualified. A practical CMPV program requires documented trending, scientifically justified limits, a defined review frequency, clear ownership and evidence that the method continues to demonstrate suitability in routine use. This also helps with proactive risk management before issues become larger investigations.
Strategic value of CMPV in GMP laboratories
CMPV is most effective when it is treated as a strategic quality and risk-management discipline, rather than as a purely administrative spreadsheet exercise. Earlier issue detection means fewer avoidable investigations, more focused CAPA management and a better-characterized understanding of method robustness over time. It also gives laboratory and CMC teams a clearer feel for the method over time. As patterns become easier to explain, overall process understanding improves. Inspection readiness and GMP compliance are stronger because the data tells a documented story.
Connect with our experts
Anabiotec helps laboratories stay on top of method performance through analytical lifecycle management support, technical expertise, method knowledge and data-driven trending. The goal is simple: confidence in long-term method performance, not only a compliant validation package.
Frequently asked questions
What is continuous method performance verification?
Continuous method performance verification (CMPV) monitors validated method performance during routine use. It uses trending and predefined review criteria to assess whether performance remains stable and to identify potentially deviating signals that require further evaluation.
How does CMPV differ from analytical method validation?
Validation provides initial evidence that the procedure is fit for its intended purpose under defined conditions. CMPV monitors performance during routine use over time and supports the assessment of whether the procedure continues to perform as intended.
What is an out-of-trend result?
An OOT result remains within specification but differs from expected history. It may signal early method drift or a shift.
Why is statistical trending important in GMP laboratories?
Trending turns routine data into quality evidence. It supports early review before OOS events occur. The resulting data can also serve as supporting evidence in investigations or when responding to questions raised during regulatory filing.
How can JMP support analytical lifecycle management?
JMP can support among other things, visualization, trending analysis, variability review, and capability assessment. It is one statistical tool within analytical method lifecycle management.
How does CMPV differ from system suitability testing (SST)?
SST determines whether predefined suitability criteria are met for an individual run. CMPV reviews performance data across runs over time to identify shifts, drifts, and changes in variability. Their respective criteria and statistical limits serve different purposes and are not simply tighter or wider.
What is the difference between short-term (within subgroup) and long-term variability?
Short-term or within subgroup variability is estimated based on smaller subgroups of data or adjacent runs. This estimate of the variability of the method is often used to construct the control limits as it reflects the common-cause variability most accurately. In contrast, long-term variability combines all observations (between subgroup and within subgroup variability), independent of the subgroup, to estimate variability.
