Claims
- 1. In a method of assuring the quality of tablets manufactured in a tabletting machine having an upper and a lower tablet punch to a nominal weight comprising, sampling the actual weight of a plurality of tablets during manufacture, calculating the mean weight of the sampled tablets, comparing the mean weight of the sampled tablets to the nominal weight to determine a deviation, adjusting the tabletting machine by lowering or raising the lower tablet punch according to the deviation in order to adjust the actual weight of the tablets to the nominal weight, the improvement comprising:
- correcting the mean weight of the sampled tablets by corrective computation by means of the Shapiro-Wilks test or Nalimov test when the actual weight of one or several tablets of the sample is found to significantly deviate from the nominal weight and using the corrected mean weight as the mean weight when comparing the mean weight to the nominal weight.
- 2. The method according to claim 1, which further comprises the step of taking and evaluating another sample before the tabletting machine is adjusted, if the weight of one or several tablets of the initial sample is found to significantly deviate from the nominal weight and the entire sample is not used for adjustment.
- 3. The method according to claim 2, wherein, as an additional sample, a sample is taken whose number of tablets is a multiple of the number of the first sample.
- 4. The method according to claim 1, wherein correcting the mean weight of the sampled tablets as well as an automatic corrective computation or determination of a requirement of a second sample is accomplished by a computer whose data output is connected to a device for adjusting the amount of feed stock.
Priority Claims (1)
| Number |
Date |
Country |
Kind |
| 3628757 |
Aug 1986 |
DEX |
|
Parent Case Info
This is a continuation-in-part of application Ser. No. 088,815, filed Aug. 24, 1987, for Method of Quality Assurance In The Manufacture of Tablets, now abandoned.
The present invention relates to a method of assuring the manufactured quality of tablets such as drugs and, more particularly, for assuring that the weight of such tablets are proper.
In the manufacture of drugs in the form of tablets, the weight of the tablets must be maintained within prescribed limits. Thus, the tablets must be weighed for control purposes. Such weighing, however, is a relatively time-consuming operation and, therefore, the weighing of each individual manufactured tablet is not economically possible. Furthermore, even if the tablet weight is found to be incorrect, it would be too late to exercise any influence on the manufacture of the tablets if, for example, a tabletting machine produces 200,000 tablets per hour.
In practice, therefore, tablet weight is controlled indirectly. This method is based on the assumption that with the same volumetric filling of a matrix in which a tablet is compressed or molded, and compression to the same height of the fillet of the tablet, the compression force will be constant. This compression force can be continually measured during the course of manufacture, so that the compression force of each individual tablet can be determined and evaluated. Thus, manufactured tablets whose force of compression exceeds or falls short of the required tolerance can be sorted out individually. In addition, known tabletting machines have adjustment devices by which changes in compression force values are compensated for by altering the amount of the filling. However, determining the weight of tablets solely on the basis of the compression force so measured is inadequate. What is required rather, is to check at certain time intervals the relation between the compression force during manufacture and the weight of the tablets.
With known tabletting machines, this checking of compression force versus tablet weight is accomplished by taking samples of tablets, which are weighed. For this purpose, the actual weight of the tablets taken as samples, for example 20 tablets, is compared with the nominal weight for 20 tablets and recorded. These weights are considered as representative of a production quantity of, for example, 100,000 tablets. If the actual and nominal weights do not correspond, the tabletting machine is adjusted depending on the deviation from the mean value and the weight of the tablets is again adjusted to the nominal value.
However, this procedure, even with the possible sorting out of individual tablets, does not guarantee that the produced tablets do not contain a single tablet with significant weight deviation. The few "runaways", that is, tablets whose weight deviates significantly from the weight cf other tablets, have little bearing on the quality of the production lot of, for example, 100,000 tablets, since they do not significantly influence either the statistical mean value or the relative standard deviation.
Furthermore, it must be kept in mind that monitoring the production of tablets by controlling the production compression force presupposes that the matrices have the same amount of filling and that material with constant quality is used in order to arrive at constant results. In practice, however, variations occasionally occur in the quality of the material, which have no direct effect on the compression force, but do have an effect on the weight of the tablets. Therefore, checking tablet weight by sampling is beneficial for keeping the weight tolerance as low as possible within predetermined limits.
A problem arises if the sample of 20 tablets happens to contain one or several "runaways", since this will result in an incorrect reading for the entire associated production lot of 100,000 tablets and show it to be poorer than it actually is. Furthermore, the disadvantages are particularly serious if the mean value of the weight of the sample tablets is also incorrect as a result of such "runaways" and is used for adjusting the tabletting machine. In this case the assumed error will in fact occur in the next production lot but in a reversed form. For example, if a sample of 20 tablets is taken and this sample contains an error tablet or "runaway" by whose higher weight the mean value of the 20 tablets is caused to be 1% higher than with 20 flawless tablets, and if the tabletting machine is accordingly adjusted to a mean value that is 1% lower than before, all subsequently produced 100,000 tablets will each deviate from the nominal weight by 1%.
It is, therefore, an object of the present invention to provide a method for assuring the quality of manufactured tablets that overcomes the problems associated with the methods indicated above.
This object is accomplished in accordance with the present invention wherein, prior to adjusting the tabletting machine, the weights of the individual tablets of the sample are measured and compared. If the weights of one or several of the tablets of the sample are found to significantly deviate from the nominal weight therefor, then the statistical analysis is corrected by corrective computation and the corrected value used for adjustment of the tabletting machine.
By this procedure, the tablets of the sample lot are controlled by checking such tablets for runaways with weight values significantly deviating from the weights of the other tablets, so that such runaways are eliminated from the calculation and are not taken into account for any adjustment of the tabletting machine that would be inaccurate because of such runaways. Adjustment would therefore take place only if the weight of a plurality of tablets of the sample is found to exceed a certain limit or fall short of a certain limit. Also, adjustment may take place if the tablets have a weight which, between two tolerance limits, closely approaches one limit. Irrespective thereof, the compression force is also continually monitored.
For example, if, as a sample lot, 20 tablets are taken wherein each is to have a weight of 1 gram, and if of these 20 tablets, 19 weigh 1 gram each and one tablet weighs 1.1 grams, this would lead to a mean value exceeding the nominal value by 0.5%, or 5 mg per tablet If this single tablet were not eliminated from the corrective computation, the subsequently produced tablets would each weigh 5 mg below their nominal weight. In other words, disregarding the runaway, that is, an adjustment without correcting the calculation would mean that the subsequently produced tablets would fall short of the nominal weight by 5 mg. The computer used for the evaluation and documentation of the results may disregard the individual "runaway" tablet, or it may take it into account with a fraction of its value in accordance with known methods of statistical calculation. Subsequently, one must determine whether the resulting mean value requires adjustment or not.
If only one or only a few runaways are found in the sample lot depending on the size of the sample, the statistical error that may result therefrom may be corrected by suitable corrective computations. For example, the Shapiro-Wilks-Test or Nalimov-Test, which are known methods of statistical calculation, may be employed and the corrected values used for the adjustment of the tabletting machine and documentation. Otherwise the entire sample is discarded and a new sample is taken immediately, which preferably includes twice as many tablets, and evaluated accordingly. If a relatively high number of runaways are found, a new sample with basically twice as many tablets is taken.
If the sample is twice as large as the preceding one, the probability with respect to the number of runaways present in the sample is greater. In this connection, one must take into account that if many runaways appear within a sample, the cause may be a momentary operating problem or failure such as, for example, a temporary bridging of the outlet hopper or funnel through which the feed stock or fitting is continually admitted into the matrices.
Preferably, controlling the weight and any corrective computations that may be required, or the determination as to whether a second sample is required is automatically handled by a computer. This same computer also carries out the statistical evaluations and the data output thereof is automatically used for adjustment of the tabletting machine, if required, and for a documentation of the values so determined.
US Referenced Citations (4)
Foreign Referenced Citations (3)
| Number |
Date |
Country |
| 0109600 |
Aug 1980 |
JPX |
| 0072811 |
May 1982 |
JPX |
| 0261697 |
Dec 1985 |
JPX |
Non-Patent Literature Citations (1)
| Entry |
| CRC Handbook of Tables for Probability and Statistics, pp. 333-337. |
Continuation in Parts (1)
|
Number |
Date |
Country |
| Parent |
88815 |
Aug 1987 |
|
Patent Grant
4999151
