1、E2810-196-95 Lower Bound(区Z*ULS/Wn,ULS)Acceptance Region(0=100)Acceptance Region(n=10Confidence Interval(0=10)8486889092949698100102104106108110112114116MeanNorE 1-All points below the lower bound contour have higher than a 95 chance of passing UDU test if mean and standard deviation are known.All p
2、oints below the acceptance region contours pass the associated acceptance limit table for n=100 and n=10.ULS is the upper confidence limit for o.Z is a standard normal critical value.FIG.2 Example of Simultaneous Confidence Interval with 95%Lower Bound and Acceptance Regionscurve,as depicted as dott
3、ed lines in Fig.2.Acceptance limit5.1.6.1 Table 5 is provided at C=90 and LB=95 forcurves are shown for n=10 and n=100,illustrating that thecomparison with Table 3 to demonstrate the effect of a loweracceptance limits approach the acceptable parameter regionconfidence level.with increasing sample si
4、ze.5.1.5 Computer programs have been developed for generat-NorE 1-Tables can also be generated for other choices for ranges ofing acceptance limit tables,but these may not be available tomeans,such as 85.1 to 114.9%LC,or for other sample sizes.all practitioners.This practice contains four acceptance
5、 limit5.1.7 A representative sample of n dosage units is takentables for many practical use situations.from the batch and the active content is measured on each unit,5.1.6 Using the Acceptance Limit Tables in This Practice-taken as a random sample or as a systematic sample(selectingVarious acceptanc
6、e limit tables are given in Tables 2-5.In eachand testing one unit at equally spaced time intervals oftable acceptance limits on the standard deviation are given formanufacture across the batch).Other sampling plans aremeans ranging 90-110 of LC in increments of 0.2%LC foroutside the scope of this p
7、ractice.sample sizes ranging from n=10 to n=500.In all tables,thetarget is set at T=100%LC,so the acceptance limits for5.1.8 The sample average(X)and sample standard deviationstandard deviations are symmetrical around 100%LC.This(s)are calculated.Enter the table row with the sample averagetarget is
8、also required for interchangeability across the ICHand the table column with the sample size to find theregions(6).At the confidence level of C=95 used often inacceptance limit on the standard deviation.If the samplethe regulatory arena,three levels of the probability lowerstandard deviation is at o
9、r below the acceptance limit then itbound are provided:LB=90(Table 2).LB=95(Table 3)can be asserted at the C%confidence level that the batch willand LB=99(Table 4).These provide 90%95%and 99have an LB chance of passing the UDU test.Some interpo-coverage,respectively,of the population of dosage units
10、 underlation may be required for sample sizes and sample means notconsideration.The usual coverage is 95%.listed in the table.E2810-19TABLE 2 Acceptance Limits on Sample Standard Deviation(%LC)forT=100%LC,C=95%,LB=90%LCSample AverageSample Size(n)(%LC)103040506080100120150200500100.02.914.364.654.84
11、4.995.195.335.435.545.665.9399.80r100.22.884.314.594.794.945.145.285.385.505.625.9199.6or100.42.844.264.544.744.895.095.245.345.455.585.8899.40r100.62.814.214.494.694.835.045.185.295.405.535.8499.20r100.82.774.164.434.634.774.985.135.235.355.485.7999.0or101.02.744.104.384.574724.925.075.175.295.435.
12、7498.80r101.22.704.054.324.524.664.865.015.15.235.375.6998.6or101.42.674.004.274.464.604.804.94055.175.305.6398.40r101.62.633.954.214.404.544.744.884.95.105.245.5698.2or101.82.603.894.344.484.684.824.925.045.175.4998.0or102.02.563.844.104.284.424.624.754.864.975.105.4397.80r102.22.533.794.054.224.36
13、4.554.694.794.905.035.3597.60r102.42.493.743.994.174.304.494.624.724.844.9752897.40r102.62.463.6834.114.244.564.664.774.905.2197.20r102.82.423.634.054.184.504.594.704.835.1397.0or103.02.393.58233.994.124.304.434.534.634.765.0696.80r103.22.353.533.773.934.064.244.374.464.564.694.9996.6or103.42.323.48
14、3.713.874.004.184304.394.504.624.9196.40r103.62.283.42.653.813.944.114.234.334434.554.8496.2or103.82.243.373.603.763.884.054.174.264.364.484.7796.0or104.02.213.323.543.703.823.994.104.194.294.414.6995.80r104.22.17326.483.763.924.044.134.234.344.6295.60r104.42.143.21.433.703.863.984.064.164.274.5495.
15、40r104.62.103.163.373.523.633.803.913.994.094.204.4795.20r104.82.073.113.313.463.573.733.843.934.024.134.3995.0or105.02.033.053263.403.513.673.78954.0643294.80r105.22.003.003.203.353.453.613.713.793.883.9942494.60r105.41.962.953.153.293.393.543.653.733.823.924.1794.4or105.61.932.890g33.233.333.483.5
16、83.663.753.854.1094.20r105.81.892.843.03.173.273.5233683.784.0294.0or106.01.862.792.983.113.213.453.533.613.713.9593.8or106.21.822.742.923.053.153.293.393.463.543.643.8793.60r106.482.68283.093.233.323.393.573.8093.4or106.62.633.033.163253.333.403.503.7293.20r106.81.722.582.752.872.973.103.193263.343
17、.433.6593.0or107.01.682.522.812.903.033.123.193.273.363.5792.Bor107.21.65472.642.752.842.973.063.133.203.283.5092.60r107.42.4222.702.782.912.993063.133.213.4292.4or107.62.36522.642.722.842.932.993.063.143.3592.20r107.81.542.312472.582.662.782.862.922.3.072803.2792.0or108.02.26412.522.602.7186293.003
18、.2091.80r108.292.202.32.462.542.652.732.792.82.933.1291.60r108.41.432.152.302.402.482.592.662.722.792.863.0591.4or108.61.402.102.242.342.522.602.652.722.792.9791.2or108.82.042.182.282.462.532.592.652.722.8991.00r109.01.331.992.132.222.292.392.472.522.582.652.8290.8or109.21291.942.072.162.232.332.402
19、.452.512.582.7490.60r109.41.261.892.012.102.172.272.332.392.442.512.6790.4or109.61.221.831.962.042.112.202.272.322.372.442.5990.20r109.81.191.781.901.982.052.142.202.252.302.372.5290.0or110.01.151.731.841.931.992.082.142.182.242.302.445.1.8.1 Example 1-A sample of n=60 dosage units pro-acceptance li
20、mit=4.18+(4.36-4.18)(80-70)/(80-60)vided dosage content statistics of X=98.6%LC and=4.18+(0.18)(10)/八20=4.27%LCs=3.91%LC.Since s was below the upper limit of 4.41%LC(using Table 3),we can state with 95 confidence there is atSince s 4.29%LC was above the upper limit of 4.27%LC,least a 95 probability
21、that a future sample taken from thewe cannot state with 95 confidence there is at least a 95%batch will meet the UDU test.probability that a future sample taken from the batch will meet5.1.8.2 Example 2-A sample of n=70 dosage units pro-the UDU test.vided dosage content statistics of X=97.8%LC and5.
22、1.8.3 Example 3-A sample of n 60 dosage units pro-s=4.29%LC.Since n was not provided in the Table 3,wevided dosage content statistics of X=98.6%LC andmust interpolate between entry n=60 with the upper limits=3.91%LC at time of manufacture.At the end of shelf life4.18%LC and entry n=80 with the upper limit 4.36%LC asthe dosage units are predicted to have lost potency such that thefollows:effective X=96.2%LC.Since s=3.91%LC was above the5