A method for evaluating the risk of misjudgment in quantitative test of medicine microorganism

Abstract

This invention discloses a method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals, belonging to the field of pharmaceutical quality control and microbial testing technology. The method includes obtaining the quantitative microbial test report results of the target test sample, evaluating the standard synthetic measurement uncertainty of the laboratory for the sample test, calculating the coefficient of variation (COP) to describe the excessive dispersion of microbial count data, setting microbial limit standards, and using the report results, the COP, and the microbial limit standards as inputs. Through a parameterized negative binomial distribution model, the probability that a single observation result does not exceed the microbial limit standard is calculated. This probability is then used as a quantitative assessment result of the risk of misjudgment when determining the conformity of the test result. This invention provides a complete technical solution for a quantifiable microbial laboratory risk assessment tool, which helps to promote the standardization process of risk assessment in the industry.

Description

Technical Field

[0001] This invention relates to the field of pharmaceutical quality control and microbial testing technology, specifically to a method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals. Background Technology

[0002] Microbiological testing of pharmaceuticals is a crucial step in ensuring drug safety. Due to the uneven distribution of microorganisms in samples and technical fluctuations in the testing process, microbial count results inherently exhibit significant variability (typically manifested as "overdispersion" with variance greater than the mean). This leads to a significant risk of misjudgment (i.e., false negatives or false positives) when laboratory determinations of "pass" or "fail" are close to the quality standard limits.

[0003] Currently, quality risk management in pharmaceutical microbiology laboratories primarily relies on general frameworks such as ICH Q9, lacking dedicated, quantifiable risk assessment tools tailored to the characteristics of quantitative microbial count data. Conventional methods like FMEA (Failure Mode and Effects Analysis) depend on subjective scoring, failing to directly calculate specific misjudgment probabilities. While the Poisson distribution can be used for ideal microbial count models, its assumption of "mean equals variance" is significantly inconsistent with reality, potentially underestimating risk. Furthermore, "measurement uncertainty," a routine laboratory assessment, is a crucial error metric, but current technologies have failed to effectively translate it into an intuitive risk probability that can be used for result determination.

[0004] The published literature, Wang Ganggang, Wang Sijin, Ma Shihong. Risk study on the selection of test quantity in monitoring the bioburden of intermediate drug solutions [J]. China Pharmaceutical Industry Magazine, 2025, 56(1):130-135, discloses the use of Poisson distribution and negative binomial distribution models to quantitatively analyze the probability of misjudgment under different test quantities. Summary of the Invention

[0005] The purpose of this invention is to provide a method that can fully utilize existing laboratory quality data (such as measurement uncertainty) and a mathematical model that fits the statistical characteristics of microbial count data to scientifically and quantitatively assess the risk of misjudgment of microbial test results, and provide data-driven decision support for the determination of critical results, the optimization of testing strategies (such as reducing the amount of testing and selecting dilution levels), and the formulation of production process monitoring limits.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: A method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals uses a parameterized negative binomial distribution model to calculate the probability that a single observation result does not exceed the microbial limit standard. The model is as follows: in: For gamma function, It is a limit standard. This is the result of a quantitative test report. These are the discrete coefficients. It is a random variable. This is a result of a single observation.

[0007] In the above technical solution, the calculation of the probability includes the following steps: S1: Obtain the results of the quantitative microbial test report for the target test sample; S2: Assess the standard composite measurement uncertainty of the laboratory for the test of the sample; S3: Based on the standard synthetic measurement uncertainty, the reported results, and the sample dilution factor during testing, calculate the coefficient of variation used to describe the overdispersion characteristics of the microbial count data; S4: Set a microbial limit standard, take the reported result, the coefficient of variation and the microbial limit standard as input, and calculate the probability that a single observation result does not exceed the microbial limit standard through a parameterized negative binomial distribution model; S5: The probability is used as a quantitative assessment of the risk of misjudgment when determining the conformity of the test result.

[0008] In the above technical solution, the expression for calculating the discrete coefficients is: in: For technical uncertainty, For matrix uncertainty, This refers to the sample dilution factor used during testing. For the quantitative test report results, 5.3020 is the natural constant. Approximate value.

[0009] In the above technical solution, the measurement uncertainty The square of the corresponding value is determined by the technical uncertainty. matrix uncertainty And Poisson uncertainty The sum of the squares of the three corresponding values ​​is determined; The formula for calculating measurement uncertainty is: In the above technical solution, the technical uncertainty is evaluated through in-laboratory reproducibility experiments, specifically as follows: Test several laboratory samples; Each laboratory sample was tested twice independently; The difference between the two test results corresponding to each laboratory sample is calculated after logarithmic transformation. The root mean square of the differences between all laboratory samples is calculated as the technical uncertainty. The formula for calculating technical uncertainty is: in, The number of laboratory samples used for evaluation. and For the first The value of two independent test results of a laboratory sample after logarithmic transformation to base 10.

[0010] One application scenario of the above technical solution is to apply it to the microbial counting results that are close to the limit standard value to quantify the risk of false negatives or false positives.

[0011] One application scenario of the above technical solution is to assess whether the risk of misjudgment is within acceptable limits when the amount of test samples is lower than the statutory requirements.

[0012] One application scenario of the above technical solution is to evaluate and compare the differences in the risk of misjudgment caused by testing with different dilution levels.

[0013] In the above technical solution, one application scenario is to apply historical microbial monitoring data based on the production line to calculate and set the alert limits and action limits of the process.

[0014] This invention discloses a misjudgment risk assessment system for quantitative microbial testing of pharmaceuticals, comprising: The data input module is used to obtain the results of the quantitative microbial test report for the target test sample and to assess the standard synthetic measurement uncertainty of the laboratory for the test of the sample. The discrete calculation module is used to calculate the coefficient of variation, which describes the overdispersion characteristics of the microbial count data, based on the standard synthetic measurement uncertainty, the reported results, and the sample dilution factor during testing. The probability calculation module is used to set the microbial limit standard. It takes the reported result, the coefficient of variation, and the microbial limit standard as inputs and calculates the probability that a single observation result does not exceed the microbial limit standard through a parameterized negative binomial distribution model. The output visualization module is used to quantify the probability as the risk of misjudgment when determining the conformity of the test result.

[0015] The present invention also provides a computer storage medium storing instructions, which, when executed on a computer, cause the computer to perform the method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals as described in the present invention.

[0016] Compared with the prior art, the present invention has the following advantages and beneficial effects: This invention is the first to combine the measurement uncertainty assessment framework in the ISO 19036:2019 standard with the negative binomial distribution model that can accurately describe the overly discrete data of microbial counts, so that the risk assessment is based on a solid metrology and statistics foundation, and the assessment results are more accurate and reliable.

[0017] The evaluation model designed in this invention requires parameters (technical uncertainty, matrix uncertainty, quantitative test report results, and sample dilution factor during testing) that can be obtained through routine laboratory quality activities (such as reproducibility tests and historical data review), without the need for complex additional experiments, which greatly reduces the application threshold.

[0018] This invention transforms the abstract concept of "uncertainty" into the intuitive concept of "probability of misjudgment," providing a quantitative scientific basis for laboratory managers to handle critical results, optimize testing procedures (such as justifying the reduction of testing quantity and selecting appropriate dilution levels), and set process warning limits and action limits based on historical data. This enables a shift from a "compliance culture" to a "risk-based quality culture."

[0019] This invention provides a complete technical solution for introducing quantifiable microbial laboratory risk assessment tools into the Chinese Pharmacopoeia and related technical guidelines, which helps to promote the standardization process of risk assessment in the industry. Attached Figure Description

[0020] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings: Figure 1 This is a schematic diagram of the operational characteristic curve (OC curve) of the scheme model in this embodiment; Figure 2 A flowchart for assessing technical uncertainty; Figure 3 This is a flowchart for assessing matrix uncertainty. Figure 4 This is a schematic diagram of the system structure. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0022] Example 1 This embodiment addresses technical uncertainty. and matrix uncertainty It is carried out using standard industry procedures.

[0023] like Figure 2 As shown, for technical uncertainty, the same laboratory sample is evaluated twice under different test conditions, resulting in two independent test results. These results are then logarithmically transformed to base 10, and the transformed values ​​are used to calculate the technical uncertainty, such as: in, The number of laboratory samples used for evaluation. and For the first The value of two independent test results of a laboratory sample after logarithmic transformation to base 10.

[0024] like Figure 3 As shown, for matrix uncertainty, the process involves multiple evaluations of the same laboratory sample. Generally, the same sample is evaluated at least 10 times under the same conditions, using the same test conditions. The test results are then evaluated using the pooled sample standard deviation method to assess random error.

[0025] In the verification process of this embodiment, the results may be highly discrete due to multiple verifications, making it impossible to accurately evaluate the final result. Therefore, the key to this embodiment is to obtain an accurate coefficient of dispersion.

[0026] This embodiment uses the results of quantitative testing of laboratory samples. Technical uncertainty, matrix uncertainty, and sample dilution factor during testing. Estimate the coefficients of variation, such as: Set a limit standard for microorganisms. Using the obtained coefficients of variation, limit standards, and quantitative test report results as input conditions, a standard negative binomial distribution model is used for fitting, resulting in the parameterized negative binomial distribution model of this embodiment: in, This is the gamma function.

[0027] Using this negative binomial distribution model, it is possible to calculate the results of a single observation. The probability of not exceeding the limit standard .

[0028] This embodiment provides an evaluation system based on the evaluation method designed above, for executing the evaluation method, such as... Figure 4 As shown, it includes: The data input module is used to obtain the results of the quantitative microbial test report for the target test sample and to assess the standard synthetic measurement uncertainty of the laboratory for the test of the sample. The discrete calculation module is used to calculate the coefficient of variation, which describes the overdispersion characteristics of the microbial count data, based on the standard synthetic measurement uncertainty, the reported results, and the sample dilution factor during testing. The probability calculation module is used to set the microbial limit standard. It takes the reported result, the coefficient of variation, and the microbial limit standard as inputs and calculates the probability that a single observation result does not exceed the microbial limit standard through a parameterized negative binomial distribution model. The output visualization module is used to quantify the probability as the risk of misjudgment when determining the conformity of the test result.

[0029] For the final calculated structure of the model, OC curves are used for output to intuitively demonstrate the evaluation process and results, such as... Figure 1 As shown.

[0030] This embodiment also provides a computer storage medium storing instructions, which, when executed on a computer, cause the computer to perform each step of the method for assessing the risk of misjudgment in the quantitative microbial testing of pharmaceuticals according to the present invention.

[0031] Example 2 Based on Example 1, the reliability of the test results is assessed (the risk of misjudgment for a specific result is calculated).

[0032] Scenario: The total aerobic bacteria count of a certain oral granule product is 2300 CFU / g, while the standard limit (AL) is 2000 CFU / g. The risk of a false negative in determining this batch of products as "qualified" needs to be assessed.

[0033] (1) Parameter acquisition: Reported result λ = 2300 CFU / g; Standard limit AL = 2000 CFU / g; Technical uncertainty determined by the laboratory through reproducibility testing ( ) = 0.0158log 10 Based on the characteristics of the drug, the matrix uncertainty is taken as ( Take 0.1 log 10The test uses a 1:10 solution with a dilution factor d=10.

[0034] (2) Calculate the coefficient of variation k: k = 5.3020 × (0.0158) 2 +0.1 2 )+(10-1) / 2300≈0.05826.

[0035] (3) Calculate the probability of misjudgment: Substitute λ=2300, k=0.05826, AL=2000 into the cumulative distribution function of the negative binomial distribution for calculation. The result is P(X≤2000|λ=2300,k=0.05826)=31.52%.

[0036] Conclusion: The model of this invention quantitatively assesses that even if the actual contamination level of the sample is 2300 CFU / g (exceeding the standard), there is a 31.52% probability that a single test result will be lower than 2000 CFU / g and thus incorrectly judged as qualified. This specific risk probability provides a clear quantitative basis for laboratories to decide whether to conduct retesting, carry out investigations, or release the sample directly.

[0037] Example 3 Based on Example 1, the dilution level of the test sample was evaluated.

[0038] Scenario: The total aerobic bacteria count of an oral granule is tested using the pour plate method. Assuming the true level of microbial contamination is 2300 cfu / g and the standard deviation of laboratory reproducibility is 0.0158 log10, if 1:10 and 1:100 test solutions are used respectively, and the microbial count is performed using the pour plate method (1 ml / plate), evaluate the probability of misjudgment for the two methods.

[0039] (1) Parameter acquisition: Reported result λ = 2300 CFU / g; Standard limit AL = 2000 CFU / g; Technical uncertainty determined by the laboratory through reproducibility testing ( ) = 0.0158log 10 Based on the characteristics of the drug, the matrix uncertainty is taken as ( Take 0.1 log 10 The test uses a 1:10 solution with a dilution factor d=10; the test uses a 1:100 solution with a dilution factor d=100.

[0040] (2) Calculate the coefficient of variation k: k(d=10)=5.3020×(0.0158) 2 +0.1 2 )+(10-1) / 2300≈0.05826;k(d=100)=5.3020×(0.0158 2 +0.1 2)+(100-1) / 2300≈0.09739.

[0041] (3) Calculate the probability of misjudgment: Substituting λ=2300, k(d=10)=0.05826, k(d=100)=0.09739, and AL=2000 into the cumulative distribution function of the negative binomial distribution, the calculation results are: P(d=10)(X≤2000|λ=2300,k=0.05826)=31.52%; P(d=10)(X≤2000|λ=2300,k=0.05826)=37.05%. Conclusion: The model calculation results of this invention show that using a 1:100 test solution ratio with a 1:10 test solution ratio increases the probability of misjudgment of the counting results by 5.53%.

[0042] Example 4 Based on Example 1, an evaluation was conducted on setting warning limits and action limits for the production process. Scenario: A sterile preparation production line needs to scientifically set alert limits and action limits based on historical microbial monitoring data of the pre-filtration solution.

[0043] (1) Parameter calculation: Monitoring data of all qualified batches in the past 3 years (617 in total) were collected, and the historical average bioburden μ = 9 CFU / mL and variance σ were calculated. 2 =370. The coefficient of variation k is calculated directly using the formula: k = (σ...) 2 -μ) / μ 2 =(370-9) / 81≈4.1849.

[0044] (2) Percentile Calculation: Input μ=9, k=4.1849 as parameters into the negative binomial distribution model of this invention. Calculate the 95th percentile (i.e., the pollution level corresponding to the cumulative probability P=0.95) and the 99th percentile of this distribution.

[0045] (3) The calculation results are: 95th percentile = 45 CFU / mL; 99th percentile = 91 CFU / mL.

[0046] Limit settings: The 95th percentile (45 CFU / mL) is set as the warning limit. When the monitoring result exceeds this value, it indicates that the pollution control trend may be abnormal, requiring vigilance and enhanced monitoring; the 99th percentile (91 CFU / mL) is set as the action limit. When the monitoring result exceeds this value, it indicates that the effectiveness of pollution control measures may be seriously insufficient, and immediate investigation and corrective measures must be taken.

[0047] Conclusion: This invention provides a method for scientifically calculating internal control limits based on historical process data and using a negative binomial distribution model. This method is more effective than traditional methods such as "mean ± 2 or 3 times standard deviation" in reflecting the excessive dispersion of microbial counts, and the set limits are more reasonable and sensitive.

[0048] Example 5 Based on Example 1, when the amount of test samples is lower than the statutory requirement, assess whether the risk of misjudgment is within acceptable limits.

[0049] Scenario: A sterile preparation production line has a low batch output (e.g., 200mL of drug solution before filtration). The specified testing volume is 100mL, and the acceptance standard is 10CFU / 100mL. If the manufacturer reduces the testing volume to 10mL, assess what control measures should be taken.

[0050] (1) Parameter acquisition: Assume the manufacturer's risk acceptance level is 5%; when the test volume is 100 mL, the acceptance standard is 10 CFU / 100 mL; when the test volume is 10 mL, the acceptance standard is 1 CFU / 10 mL; the technical uncertainty is determined by the laboratory through reproducibility tests. ) = 0.0158log 10 Based on the characteristics of the drug, the matrix uncertainty is taken as ( Take 0.1 log 10 The original sample solution was used as the test solution, with a dilution factor d=1.

[0051] (2) Calculate the coefficient of variation k: k = 5.3020 × (0.0158) 2 +0.1 2 )+(1-1) / 2300≈0.05434.

[0052] (3) Calculate the upper limit of the results for different test values ​​at a 5% risk acceptance level: Set P=95%, k=0.05434, AL 100mL =10, AL 10mL Substituting 1 into the cumulative distribution function of the negative binomial distribution, the upper limit of the count result λ is calculated when the test volume is 100mL. 100mL =20 CFU, with a test volume of 10 mL, the upper limit of the count result λ 10mL =5 CFU.

[0053] Control Measures: Calculations show that with a test volume of 100 mL and a 95% cumulative distribution probability, if the sample contains 20 CFU of microorganisms, the test result will be 10 CFU; with a test volume of 10 mL and a 95% cumulative distribution probability, if the sample contains 5 CFU of microorganisms, the test result will be 1 CFU. Reducing the test volume increases the risk of uncertainty in the results. Manufacturers should adopt stricter control measures, such as reducing the maximum filtration volume of the filter, to control the increased risk due to reducing the test volume.

[0054] Conclusion: This invention provides a method for assessing the increased control risk caused by insufficient testing volume using a negative binomial distribution model. For monitoring the bioburden of the drug solution before filtration, when the testing volume is reduced from 100 mL to 10 mL, the risk of misjudgment of the test results increases by 2.5 times.

[0055] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for evaluating the risk of misjudgment in a quantitative test for microorganisms in a pharmaceutical product, characterized by, The probability that a single observation result does not exceed the microbial limit standard is calculated using a parameterized negative binomial distribution model. The model is as follows: ， in: For gamma function, It is a limit standard. This is the result of a quantitative test report. These are the discrete coefficients. It is a random variable. It is a single-pass result.

2. The method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals according to claim 1, characterized in that, The calculation of the probability includes the following steps: S1: Obtain the results of the quantitative microbial test report for the target test sample; S2: Assess the standard composite measurement uncertainty of the laboratory for the test of the sample; S3: Based on the standard synthetic measurement uncertainty, the reported results, and the sample dilution factor during testing, calculate the coefficient of variation used to describe the overdispersion characteristics of the microbial count data; S4: Set a microbial limit standard, take the reported result, the coefficient of variation and the microbial limit standard as input, and calculate the probability that a single observation result does not exceed the microbial limit standard through a parameterized negative binomial distribution model; S5: The probability is used as a quantitative assessment of the risk of misjudgment when determining the conformity of the test result.

3. The method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals according to claim 1 or 2, characterized in that, The expression for calculating the discrete coefficients is as follows: ， in: For technical uncertainty, For matrix uncertainty, This refers to the sample dilution factor used during testing. This is the result of a quantitative test report.

4. The method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals according to claim 2, characterized in that, The square of the measurement uncertainty is determined by the sum of the squares of the corresponding values ​​of the technical uncertainty, matrix uncertainty, and Poisson uncertainty.

5. The method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals according to claim 3, characterized in that, The technical uncertainty is evaluated through in-laboratory reproducibility experiments, specifically: Test several laboratory samples; Each laboratory sample was tested twice independently; The difference between the two test results corresponding to each laboratory sample is calculated after logarithmic transformation. The root mean square of the differences between all laboratory samples is calculated as the technical uncertainty.

6. The method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals according to claim 1, characterized in that... Application: Quantifying the risk of false negatives or false positives in microbial counts that are close to the limit standard values.

7. The method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals according to claim 1, characterized in that... Application: When the sample quantity for testing is lower than the legal requirement, assess whether the risk of misjudgment is within acceptable limits.

8. The method for assessing the risk of misjudgment in the quantitative microbial testing of pharmaceuticals according to claim 1, characterized in that... Application: To evaluate and compare the differences in the risk of misjudgment when using different dilution levels for testing.

9. The method for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals according to claim 1, characterized in that... Applications include: calculating and setting alert limits and action limits for processes based on historical microbial monitoring data from the production line.

10. A system for assessing the risk of misjudgment in quantitative microbial testing of pharmaceuticals, characterized in that, include: The data input module is used to obtain the results of the quantitative microbial test report for the target test sample and to assess the standard synthetic measurement uncertainty of the laboratory for the test of the sample. The discrete calculation module is used to calculate the coefficient of variation, which describes the overdispersion characteristics of the microbial count data, based on the standard synthetic measurement uncertainty, the reported results, and the sample dilution factor during testing. The probability calculation module is used to set the microbial limit standard. It takes the reported result, the coefficient of variation, and the microbial limit standard as inputs and calculates the probability that a single observation result does not exceed the microbial limit standard through a parameterized negative binomial distribution model. The output visualization module is used to quantify the probability as the risk of misjudgment when determining the conformity of the test result.

Images