Regulations and compliance · Practical guide

FDA CSA: assessing process risk from the effects of failure

Under FDA CSA, process risk concerns the consequences of a possible software failure within a particular production or quality management system process. Even a small automated step can have a significant effect. Trace a functional failure through to its potential effect on quality and safety, then select an assurance approach that addresses it.

Table of contents

At a glance

How do we assess the consequences of failure in a production or QMS software process?

Under FDA’s CSA guidance, process risk is high when a function’s failure could create a quality problem that foreseeably compromises safety. Describe the function, a reasonably foreseeable failure and the effect of existing controls. For high process risk, assurance follows medical device risk. Otherwise, it follows process risk and still needs appropriate evidence. This distinction applies to medical device production and QMS software and does not replace the device risk assessment.

  • Assess potential error consequences, not just suspected error rates.
  • Link process risk and potential medical device effect together.
  • Consider only justified impacts when making decisions.

1. Limit the process prior to evaluation

The CSA Guidance dated February 3, 2026, addresses software used in production and quality management within the medical device sector. It is a non-binding FDA recommendation within the applicable regulatory framework. It is neither a general point system applicable to all GxP applications nor a comprehensive guide for conducting a risk assessment of a medical device.

Name a specific function rather than a full product name. "ERP" is too broad. "The system selects the appropriate material inventory level for a manufacturing order" is verifiable. Capture input data, processing, decision-making, and the subsequent process step. If multiple functions have different effects, their evaluation may vary accordingly.

Sources: [1] [2]

2. Describe a traceable failure sequence

Start with the conceivable functional failure. Then inquire about the potential quality issues that may arise and whether these could compromise safety. This chain of reasoning is more important than an isolated risk score. Without it, it is difficult to justify why a function has been classified as high or low process risk.

The FDA refers to reasonably predictable errors. This is not limited to errors that the team deems particularly likely. An infrequent but plausible error should not be dismissed solely because of a low estimated occurrence rate. Similarly, a purely theoretical assertion without reference to actual operational practice should not unilaterally determine the overall evaluation.

Sources: [1]

3. Keep the terms distinct

The following internal work aid distinguishes between errors, quality issues, safety effects, and controls. These elements can be documented within an existing risk analysis. Two separate, mandatory analysis documents are not derived from this. What is important is the traceable linkage between perspectives.

The classification into high and non-high process risk serves to guide the selection of appropriate assurance measures. It is neither a quality judgment regarding the supplier nor a compliance certificate. Within an organization, further gradations may be justified, provided their significance remains clear and the relevant error consequences are not obscured by the scale.

Terms that support a reasoned assessment
TermQuestion to askInadequate abbreviation
system failure What can software fail to do or do incorrectly? The software is complex.
Impact on quality How does a failure affect product or process decision? There could be an issue.
Safety effect Can the quality issue, if predictable, compromise safety? Every quality defect is automatically considered critical.
Control What does the sequence detect or prevent under real-world conditions? An employee might notice it.
Assurance Which evidence addresses the relevant trust question? For this system class, we always perform the same test.

Sources: [1]

4. Critically evaluate existing controls

An existing control can alter the error chain. Describe the information it uses, when it takes effect, and who responds to a result. An independent verification of the actual material should be evaluated differently from a repeated review of the same material, which may reflect a false system indication.

Consider the conditions under which the control may fail. Is it designed for a single sample only? Can a deviation be overlooked? Is the responsible person unavailable during a shift handover? Such questions transform a general control assertion into a assessable control measure. The guidance permits the consideration of additional process controls, not their unverified assumption.

Sources: [1]

5. Example: Material Assignment and Status Display

In the example, a medical device manufacturer has two functions within the same application. Function A automatically selects a material batch for a critical manufacturing step. In the assumed workflow, there is no additional effective verification prior to use. Function B displays the progress of an administrative task; the actual material selection occurs independently of this display.

For A, it is explicitly assumed that an incorrect material assignment could lead to a quality issue with foreseeable safety implications. Under these assumptions, a high process risk is plausible. For B, it is assumed that an incorrect display would only delay administrative processing. These conditions must be substantiated in an actual project; the function names alone are insufficient.

Example: completed process risk matrix
FunctionErrors and ConsequencesControls as per the assumed procedure flowRational inference
A: Select material status An ineligible version is used, affecting a product characteristic relevant to safety No further effective testing prior to use is accepted. High process risk; align assurance to the relevant device effect
B: Display processing status Status outdated; administrative responsibility delayed visible Material usage is not dependent on this display. Not high process risk on these assumptions. A suitable process check is still needed.
After the planned additional control False mapping should be detected before use Control previously only planned, not validated Do not reduce assessment solely on the basis of planned measures
B following intended use change The display now controls the actual material usage. Current independence is no longer applicable. Evaluate again; do not accept old classification unchanged

Sources: [1]

6. Choose assurance activities based on the consequences of failure

For Function A, targeted cases should evaluate selection rules, boundary conditions, and inappropriate material states. Additionally, review of interfaces and master data may be required. For Function B, other activities may suffice, provided they adequately support the intended use. This represents a justified methodological decision, not an automatic waiver of testing.

The FDA requires assurance of the relevant medical device performance for high-process risk scenarios; for low-process risk scenarios, such assurance remains aligned with the process risk. Scripted, unscripted, or automated testing are tools. A rigid schema—“high always implies fully scripted, low always implies never scripted”—would be overly simplistic and fail to address the specific evidentiary question.

Sources: [1]

7. Make unclear assumptions visible

When it is unclear whether Function B actually influences a material selection decision, a critical foundational basis is missing. Clarify the actual workflow instead of closing the gap through a favorable assessment. Discussions with end-users, process observation, and review of existing documentation can reveal where the software is truly being used.

Document assumptions in such a way that subsequent changes can be identified. The introduction of a new interface or the elimination of a manual control may alter the previous rationale. For small teams, a concise and well-structured functional overview is often more effective than an extensive, one-time table whose prerequisites are no longer known after project completion.

Sources: [1]

8. Checklist prior to risk decision

The decision should also be understandable to a colleague who did not attend the meeting. Can he trace the path from the functional failure to the potential consequences and to the selected assurance approach? If only a number or color remains, the majority of the rationale is typically missing.

Our checklist supports professional discussion. It does not prescribe a fixed evaluation formula, a universal point scale, or a specific signature sequence. Existing procedures may be utilized if they adequately represent the relevant decision and its underlying conditions.

  • Specific production or quality management system application described.
  • Predictable functional failure identified.
  • Quality and potential safety implications associated.
  • Evaluates controls in terms of their actual effectiveness and limitations.
  • Unclear assumptions and additional knowledge gaps evident.
  • Methodology and depth derived from error root cause analysis.
  • Trigger for a renewed assessment is documented.

Sources: [1]

Project experience and collaboration

Assess the consequences of a function failing

With process owners, Quality Management and IT, we examine the individual function. What could happen on a reasonably foreseeable failure, and which existing controls actually limit the consequences? An administrative forwarding step may warrant a different assessment from a function with a direct safety impact. We document the reasoning alongside the intended use. Within the applicable FDA medical device scope, this informs which assurance activities will adequately answer the open question.

Frequently asked questions about Assessing process risk under FDA CSA

Is process risk the same as medical device risk?

No. The Guidance distinguishes between potential impairment of a manufacturing or quality management system (QMS) process and the resulting device- or safety-related impact. These perspectives must be integrated. This does not imply a requirement for two separately maintained documents, provided the interrelationships are sufficiently understandable.

Sources: [1]

Must we consider only probable software defects?

No. The Guidance addresses reasonably predictable errors. A plausible error with serious consequences must not be dismissed solely because its frequency is estimated to be low. The specific error chain and the effectiveness of existing controls must be evaluated.

Sources: [1]

Is a CAPA system always not a high process risk?

Typically not. The FDA identifies CAPA routing, automated complaint capture, change control, and process management as examples of functions that generally do not pose a high process risk. The key factor remains the individual function: if its failure could cause a quality issue that adversely affects safety and is foreseeable, the FDA defines this as a high process risk. A system name such as CAPA does not substitute for this assessment.

Sources: [1]

Does high process risk imply that validation is not required?

No. The function may still be subject to relevant validation requirements. The level of assurance will be selected appropriately based on process risk. A low risk does not imply that data quality, reliability, or applicable documentation obligations are insignificant.

Sources: [1] [2]

Can manual checks influence the assessment?

Yes, if they are actually appropriate to prevent or detect the relevant consequence. Describe the procedure, information, timing, and response. Simply expected employee attention is less reliable than a defined and demonstrably effective control.

Sources: [1]

Does a risk number from an FMEA suffice?

A number may support an existing method but does not replace the rationale. It must remain evident which functional failure could lead to a particular consequence and why the controls are necessary. Different failure chains may result in the same numerical value yet require distinct validation evidence.

Sources: [1]

Does a high-process-risk always require exclusively scripted tests?

Not mandatory. The Guidance allows for an appropriate combination of methods. What is essential is that the controls are aligned with the relevant failure mode. Exploratory or automated checks can contribute, without automatically replacing targeted validation for critical rules.

Sources: [1]

When must we review the assessment again?

If the intended use, function, data flow, or underlying controls change, the previous justification may be affected. New findings may also serve as a basis for review. Assess the specific impact; an unchanged product name does not guarantee an unchanged risk profile.

Sources: [1]

Does this classification also apply to pharmaceutical GMP?

The FDA-CSA Guidance has a specific medical device scope. Its terminology must not be issued as general pharmaceutical regulatory requirements without proper contextual interpretation. For pharmaceutical GMP, the requirements outlined therein must be independently determined in conjunction with an appropriate risk assessment methodology.

Sources: [1]

How do we avoid endless risk discussions?

Work with a specific function and a traceable failure scenario. Identify which assumption is in question, and proactively obtain the missing information. Discussions about abstract system classes often take longer without bringing clarity to the actual decision required.

Sources: [1]

Sources and review date

Source date: 7 October 2026. Regulatory requirements, guidelines, and industry practices must be categorized separately. Tables, figures, and case examples are to be considered standalone reference materials; they do not constitute completed product testing.

  1. FDA: Computer Software Assurance for Production and Quality Management System SoftwareFinal guidance of 3 February 2026. Non-binding recommendations for production and quality management system software in the medical device sector.
  2. 21 CFR Part 820 – Quality Management System RegulationQMSR, effective since 2 February 2026: §§ 820.1, 820.7 and 820.10. The CSA-Guidance refers for software validation to the included ISO-13485:2016 subsections 4.1.6, 7.5.6 and 7.6 as well as for records on 4.2.5; § 820.35 contains additional recording requirements. The ISO-Verweiseen are documented here via the FDA-Guidance, no independent full text comparison of the ISO standard.

Sources checked: .

Daniel Herrmann, Daniel Herrmann Consulting

Daniel Herrmann Consulting

Jointly develop the specific evidence process

We support you in the traceable evaluation of specific software functions and in selecting the appropriate Assurance activities accordingly.

  • GAMP 5 consulting →

    Understand how GAMP 5, EU GMP Annex 11 and 21 CFR Part 11 apply to your systems. We translate the relevant requirements into proportionate validation work and evidence your team can defend in an audit.

Send a project enquiry Customer testimonials →