Executive summary

• There is no single ISO 13485 clause-specific standard that dictates “the” sample size for design verification and validation; instead, compliance with clauses 7.3.6 and 7.3.7 is typically demonstrated by referencing broadly accepted sampling standards (for attribute sampling and lot-based testing), device- and study-specific FDA guidances that require statistical sample size justification, and clinical investigation standards that mandate justified sample sizes.
• The most frequently cited frameworks in the analyzed materials are: ISO 2859-1 (and ANSI/ASQ Z1.4) for acceptance sampling plans with AQL/OC-curve foundations; FDA device guidances that explicitly require scientific or statistical sample size justifications and multi-lot sampling; and ISO 14155 for clinical sample size justification in design validation. Process-validation guidance from GHTF SG3 is also commonly used to underpin statistical rationales for sampling in protocols aligned with ISO 13485 ^{22822723023113914021521623623726989909194}.

Core sampling plan standards used for design V&V

ISO 2859-1 and ANSI/ASQ Z1.4 (attribute sampling; OC curves and AQLs)

• The face masks/respirators materials explicitly cite ISO 2859-1:1999 for defining acceptance sampling plans using operating characteristic (OC) curves and Acceptable Quality Limits (AQL); ASQ/ANSI Z1.4 is also named as an alternative attribute sampling standard ^228227230. They emphasize selecting sampling plans via OC-curve analysis and risk assessment; for example, an AQL of 4% or better is recommended in that context, illustrating how risk and AQL drive sample size and acceptance criteria ²³¹²²⁹. The same documents note alignment with ISO 13485:2016 QMS expectations for statistically valid sampling approaches ²³².
• Significance for ISO 13485 V&V: While ISO 2859-1 and ANSI/ASQ Z1.4 are not design-control standards, they provide the statistically valid attribute-sampling schemes frequently used to justify sample sizes and acceptance criteria in design verification plans (e.g., dimensional checks, appearance, pass/fail functional inspections) under clause 7.3.6, and, where appropriate, in certain validation activities that use attribute outcomes ^{228230231229232}.

ASTM F3172-15 (device size and sample selection)

• The DUMI ManipulatOR filing cites ASTM F3172-15 for “design verification device size and sample selection” alongside ISO 2859-1:1999, indicating the use of a product-specific consensus practice to define which sizes/variants and how many units should be included in design verification sampling ²⁴.
• Significance: ASTM F3172-15 provides a structured approach to covering device size ranges and selecting samples for verification, which complements ISO 2859-1 attribute sampling when scoping and justifying a verification matrix under clause 7.3.6 ²⁴.

Cross-cutting guidance that supplies statistical rationales for sampling

GHTF SG3 Process Validation Guidance (GHTF/SG3/N99-10:2004)

• The personalized medical devices materials reference the GHTF SG3 process validation guidance and ISO 13485, highlighting acceptance sampling plans to optimize the number of samples tested and demonstrate conformance, as well as statistically valid techniques such as sampling, design of experiments, Taguchi methods, and response surface studies to determine “how many” measurements are needed with statistical significance ^909189929394.
• Significance: Although centered on process validation, these statistical tools and the emphasis on protocolized statistical methods (including acceptance criteria and sampling rationale) are directly transferable to design V&V protocol development under ISO 13485 clauses 7.3.6 and 7.3.7, particularly when justifying sample size and acceptance criteria in verification plans or validation studies ^899091929394.

FDA device guidance documents with explicit sample size and sampling plan expectations

Peripheral PTA and specialty catheters (2023)

• Recommends “scientific or statistical justification for sample size for each test” and implementing a sampling plan that examines multiple lots (at least three) to assess inter- and intra-lot variability. It also advises testing extremes and appropriate intermediate sizes across the product matrix (“four corners plus intermediate”), and establishing acceptance criteria with scientific rationale—core elements of a defensible verification sampling strategy ²⁷⁴²⁷⁵.
• Significance: This is a clear example of design verification expectations (clause 7.3.6) that pair sample size justification with lot/release representativeness and coverage of variants ²⁷⁴²⁷⁵.

AI-enabled Device Software Functions: Lifecycle Management (2025, draft)

• Calls for a “sample size justification that ensures adequate study power,” with prespecified statistical hypotheses, study success/failure criteria, and analysis plans; it references other FDA guidances (e.g., on pivotal clinical investigations and diagnostic test reporting) as statistical supplements and anchors the design validation activities to design controls under 21 CFR 820.30 ^139140142141. It also points to human factors validation testing guidance for usability-related validation ³⁰¹.
• Significance: Provides a modern framework for validation study planning (clause 7.3.7), where sample size is rooted in power/hypothesis-driven design and representativeness of the intended-use population ^139140142301.

Essential Drug Delivery Outputs (EDDOs) for combination products (2024, draft)

• Recommends risk-based, statistical sampling plans for design verification testing, specifying the number of lots and defining acceptance criteria; recognizes that sampling recommendations in recognized standards may be used and that tested lots should be representative of the commercial manufacturing process ^218220215216.
• Significance: Offers a template for verification protocols that tie sample size, number of lots, and acceptance criteria to risk and recognized standards—consistent with clause 7.3.6 expectations ^215216220.

Blood Glucose Monitoring Systems (2020)

• Provides concrete minimum sample sizes and sampling schemes: for example, precision studies using a minimum of 500 test strips from at least 10 vials and 3 manufacturing lots, with at least 10 meters and 10 measurements per meter per glucose range; it also references CLSI EP6-A for linearity evaluation and cites CLIA Waiver guidance recommending a minimum of 350 patients per sample type for clinical evaluations and additional subgroup targets ^{33323430109111112}.
• Significance: These device-specific expectations illustrate how FDA sets V&V sample sizes and lot coverage to establish precision, linearity, and clinical performance—approaches that can be leveraged to justify sample size and sampling plans for clause 7.3.6 and 7.3.7 activities when similar risks and use-cases apply ^{33323430109111112}.

Points to Consider for Cervical Cytology Devices (1994)

• Requires detailed statistical sampling plans and sample size justifications with prespecified Type I error, power, hypothesized clinical sensitivities/specificities, and prevalence; emphasizes representativeness/generalizability and recommends biostatistical consultation during protocol planning and analysis ^236237233235.
• Significance: A foundational example of hypothesis- and power-based validation sample size justification directly relevant to clause 7.3.7 ^236237233235.

Special 510(k) Program (2019)

• Expects concise summaries of verification/validation activities that include acceptance criteria and quantitative results (mean, standard deviation, range), and leverages recognized standards and risk management; while not prescribing sample size formulas, it places sample size and statistical summaries squarely in the evidentiary package ^203205208.
• Significance: Reinforces that sample size and statistical treatment are integral to verification/validation documentation even in streamlined submissions ^203205208.

Clinical investigation standard relevant to design validation

ISO 14155:2020 (as referenced by MHRA)

• Requires inclusion of sample size with justification in clinical investigation plans; even in first-in-human or pilot/feasibility studies, the proposed sample size must be justified as suitable for the study purpose, with specification of statistical methods, hypotheses, significance levels, and analysis plans ²⁶⁹.
• Significance: Directly supports ISO 13485 clause 7.3.7 for clinical validation activities that are part of design validation ²⁶⁹.

Additional QMS-linked references that reinforce statistical sampling expectations

• Personalized medical device materials align with ISO 13485 terminology and requirements, and emphasize acceptance sampling plans and statistically valid techniques (sampling, DOE, Taguchi, response surface methods) to determine “how many” samples to measure with statistical significance—elements to be specified in protocols with defined analysis methods. The GHTF SG3 guidance is explicitly cited as the basis for these statistical approaches ^929390899194.
• FDA Quality System Regulation under 21 CFR 820.30 is frequently cited as the design control framework anchoring validation/testing expectations in multiple FDA guidances (e.g., AI-enabled device software) ¹⁴¹.

How these sources map to ISO 13485 clauses 7.3.6 and 7.3.7

Clause 7.3.6 – Design verification

• Attribute sampling and acceptance criteria: Use ISO 2859-1 or ANSI/ASQ Z1.4 to select statistically valid lot-by-lot acceptance sampling plans based on OC curves and AQLs; justify the chosen AQLs via risk assessment (example contextual AQL target: 4%) ^{228227230231229}.
• Lot representativeness and manufacturing coverage: Sample across multiple lots (≥3) to assess inter-/intra-lot variability; test extremes plus intermediates; define acceptance criteria with scientific rationale, as articulated for PTA/specialty catheters ²⁷⁴²⁷⁵.
• Variant/size coverage: Employ product-specific practices like ASTM F3172-15 to ensure the device-size matrix is adequately represented in verification testing ²⁴.
• Risk-based planning and recognized standards: For combination products or outputs like EDDOs, create risk-based statistical sampling plans specifying the number of lots and acceptance criteria, leveraging recognized standards where applicable ^215216220.

Clause 7.3.7 – Design validation

• Power- and hypothesis-based sample size justification: Follow FDA guidance for AI-enabled devices (and analogous clinical/performance guidances) to predefine hypotheses, success criteria, and statistical analysis plans with sample size justification that ensures adequate power ^139140142.
• Clinical investigation expectations: Apply ISO 14155’s requirement to include and justify sample size in clinical investigation plans, including in feasibility phases, with explicit statistical methods and significance levels ²⁶⁹.
• Disease performance characterization: Use frameworks like the cervical cytology guidance to set alpha/power, prevalence, and clinical sensitivity/specificity targets; ensure sampling is representative of the intended-use population ^236237233235.
• Device-specific minimums: Where FDA has issued device-specific expectations (e.g., blood glucose systems), adopt those minimum sample sizes and lot sampling schemes for precision/linearity/clinical performance as part of the validation rationale ^{33323430111112109}.

Nuances and limitations observed in the analyzed materials

• Direct clause-by-clause interpretations of ISO 13485 7.3.6/7.3.7 regarding sampling plans are uncommon in the materials reviewed. Many device filings cite ISO 13485 and applicable safety/risk/IEC standards yet do not explicitly identify sampling plan standards or sample size methodologies for design V&V ²³¹.
• Several references (e.g., GHTF SG3) are framed around process validation rather than design V&V, but their statistical principles (acceptance sampling, DOE, response surface methods) are routinely applied to design verification protocol development and sample size justification in practice ^89909194.

Practical application and implications

• For verification protocols, a defensible approach is to combine ISO 2859-1/ANSI Z1.4 attribute sampling (with OC-curve/AQL-based justification) and product-matrix coverage (e.g., extremes/intermediates; ASTM F3172-15 where applicable), and to include multi-lot representativeness. This aligns with the FDA PTA catheter expectations and the risk-based rationale seen in EDDO guidance ^{22823023127427524215216}.
• For validation protocols (clinical or performance), prespecify hypotheses and success criteria, perform power-based sample size calculations, ensure representativeness of the intended-use population, and follow device-specific minimums where FDA has published them (e.g., blood glucose systems, CLIA waiver), consistent with ISO 14155 and FDA AI-enabled/cytology expectations ^{139140269236237111112333234109}.
• By explicitly citing these standards/guidances in your protocols and reports—and by documenting risk-based rationales, AQL/OC-curve decisions, lot selection, and power analyses—you create a clear line of sight to ISO 13485 clauses 7.3.6 and 7.3.7.

Referenced standards and guidance (as identified in the analyzed materials)

• ISO 2859-1:1999; ASQ/ANSI Z1.4 (attribute acceptance sampling; OC curves; AQL) ^{228227230231229}.
• ASTM F3172-15 (design verification device size and sample selection) ²⁴.
• GHTF SG3/N99-10:2004, Process Validation Guidance (acceptance sampling; DOE; statistical protocolization aligned to ISO 13485) ^909189929394.
• FDA PTA and specialty catheters guidance (statistical sample size justification; ≥3 lots; extremes + intermediates) ²⁷⁴²⁷⁵.
• FDA AI-enabled device software draft guidance (sample size justification; power; hypotheses; QSR 820.30; human factors validation) ^{139140142141301}.
• FDA Essential Drug Delivery Outputs draft guidance (risk-based statistical sampling plans; number of lots; acceptance criteria; recognized standards) ^218220215216.
• FDA blood glucose systems guidances and related references (minimum sample sizes; lot sampling; CLSI EP6-A linearity; CLIA waiver sample sizes) ^{33323430109111112}.
• Points to Consider for Cervical Cytology Devices (alpha/power; sensitivity/specificity; prevalence; representativeness) ^236237233235.
• ISO 14155:2020 (clinical investigation plan requires sample size with justification), as referenced by MHRA ²⁶⁹.