The Unknown Sources of Error in the Pharmaceutical Weighing Process

Thomas Caratsch ,  Head of Laboratory Division, Mettler Toledo

There are two conventional processes involved in the preparation of an analytical reference standard. These processes both assume that there is a standard operating procedure or work instruction for the preparation of a nominal concentration of reference standard and therefore the amount of reference material to be weighed is known. If there are no instructions available, then an analytical scientist is required to prepare a reference solution from first principles. In this case an additional calculation step is required to determine the amount of standard to be weighed and the volume that the substance must be dissolved in. This is not included in the two process flows and would prove to be more time-consuming. The processes also describe a vessel in which the reference standard is weighed. This can be a weighing boat or directly into a volumetric flask depending on the working practices of an individual laboratory.

PROCESS 1: WEIGHING AND RECORDING BY OBSERVATION

This is the process flow for weighing the reference standard where the values are recorded by observation from the balance screen and written into a laboratory notebook by an analyst. The subsequent calculations are performed using a hand-held calculator with the results also written directly into the analyst’s laboratory notebook. The volumetric flask is labelled by hand to identify the solution, preparation information, and the expiry date of the solution.

The process begins by writing up the work to be done in the laboratory notebook and checking that the correct reference standard to be prepared has been selected. Next the balance is checked with an external calibration mass. The balance and calibration standards used are recorded in the lab notebook by the analyst. Then the weighing vessel is weighed and the balance tared.

The reference material is weighed on the analytical balance and the reading on the screen is observed and recorded in the lab notebook by the analyst. When completed, the vessel is removed and the balance is cleaned and tidied ready for the next user. The reference material is transferred to the appropriate size volumetric flask and liquid added to prepare the reference solution.

Then the flask is inverted to dissolve the analyte and then made up to volume. The flask is labelled by hand with the standard identification number, substance information, calculated concentration, analyst who prepared the solution, storage conditions, and the dates of preparation and expiry. The analyst uses any factors such as purity or water content to calculate the actual concentration of the reference standard solution. The calculations and conversion factors used are recorded in the lab notebook. The calculation is performed using a hand-held calculator; the analyst reads the final value from the calculator display and transcribes

it into their lab notebook. The analyst checks the data and results, including repeating the hand-held calculation, and if correct signs the relevant pages of the lab notebook. If there any deviations from the procedure or instructions, the analyst must record them in their laboratory notebook. A second person reviews the data and procedure to confirm that all data are correct and then signs to approve the work. If there are any corrections, the first analyst will do this and return for approval.

Areas Susceptible to Errors

Transcribing Data and Transcription Error-Checking—Quality standards (e.g. ISO 17025) and pharmaceutical industry regulations use the four eyes principle when performing work: one person performs the work and a second independent person reviews it. This principle is based on the idea that four eyes are better than two. However, as the process is performed by humans it is error-prone in its own right. So this manual process is not perfect and typographical errors could be missed in the second person review, especially if the individual is under pressure with other tasks to perform.

Lack of a Data Audit Trail—In essence there is no paper audit trail; this process relies on the ability of the analyst performing the work to accurately record the values displayed on the balance and calculator, and transcribe these results into the laboratory notebook without error. Only the calculator result can be replicated by keying in the data and performing the calculation again. Unfortunately, as the process is operated by humans, even highly trained humans, it is subject to errors. What the brain thinks it has seen and recorded may not be the actual value on the balance or calculator. The second person cannot check the actual balance reading, which is the major failure point of this process.

Failure to Meet GxP Regulatory Requirements— For laboratories that must comply with GxP regulations it is important to have records or documented evidence that can be checked by a second person and is also available for inspection.

PROCESS 2: WEIGHING WITH A PRINTER ATTACHED TO THE BALANCE

In this process, the main changes are the addition of a thermic printer to record results contemporaneously from the printer, and the substitution of a validated spreadsheet for the handheld calculator. This allows the process to be more efficient but, more importantly and less error-prone than process 1 with the added bonus of a regulatory or quality paper trail.

In the same way as the first process started, the lab notebook is written up for the work to be performed and the correct reference standard is selected. Next, the balance is checked and the weighing vessel is weighed and tared. The values are printed out on the printer together with the date and time of the activity. Then the reference standard is weighed on the analytical balance and the reading is recorded on the printout. Adding the printer avoids the need for the analyst to record this information in the lab notebook.

When completed, the vessel is removed and the balance is cleaned and tidied ready for the next user. The reference material is transferred to the appropriate size volumetric flask and the standard solution prepared. The balance can print a label for the reference solution containing all quality and regulatory information to save the analyst performing this task manually.

The balance printout is removed and the analyst inputs the actual weight of the reference substance into the spreadsheet together with any correction factors such as purity or salt to base conversion to calculate the concentration of the reference standard solution automatically. The spreadsheet is then printed out and the analyst pastes the balance and the spreadsheet printouts into the lab notebook.

The analyst checks the data and results. This check is more complete than process 1 as all the data are available on the two printouts. The calculation need not be repeated or verified as the spreadsheet is validated. Once the data checks show the data are correct, the analyst signs the relevant pages of the lab notebook.

The second person review becomes more relevant as there is now a complete data trail to follow and demonstrate that the procedure was followed and that all data are correct. The reviewer signs the lab notebook. This second person check is faster than the previous process as there are no hand-held calculations to be performed.

Areas Susceptible to Errors

Manual Data Entry to the Spreadsheet— Transcription error checking is not eliminated as the balance result and any conversion factors have to be entered into the spreadsheet. These figures must be checked by the analyst and the reviewer to ensure that they are correct.

Paper Based Process— This process is paper based. There are two printouts that are produced and pasted into the laboratory notebook to form the analytical record. This, together with the preparation and summary of the work that needs to be written by the analyst, results in a slow process. To improve the process further and to eliminate the transcription check we need to consider working electronically.

SUGGESTED PROCESS

To avoid erroneous occurrences, Mettler Toledo has presented a third and error-free electronic process of weighing by the utilisation of the LabX Server Software.

In this process the lab notebook, the spreadsheet and the associated printouts have all been replaced by METTLER TOLEDO’s LabX software. The LabX software has the technical controls for ensuring data integrity to comply with GxP regulations for electronic records and electronic signatures. However, the function of more importance to regulated and non-regulated laboratories alike is the ability to sign records electronically. Whilst a non-regulated laboratory is not concerned about compliance with pharmaceutical industry regulations, cost-consciousness and improvements to analytical laboratory processes are still relevant topics, and this is where LabX software can help.

Implementing LabX allows the process to be made fully electronic. It eliminates the last source of transcription error, improves data integrity and quality, speeds up the overall process, and reduces time needed to perform a task. LabX is a configurable software application and the SOP for preparing a standard solution can be incorporated in an electronic process and validated, thus enforcing compliance with the written procedure.

The LabX electronic process starts by the analyst selecting the reference standard to weigh and then logging onto LabX at the terminal of the analytical balance. There is no need to log onto a separate workstation to access LabX. Balance checks must be performed if prompted by LabX. Otherwise the analyst takes the weighing vessel and tares the boat, then weighs the reference material. No results are recorded by the analyst as LabX does all the work: actions and weights are recorded in the database against the user’s identity, together with a time and date stamp.

Dissolving of the reference material in the volumetric flask and making up to volume is carried out by the analyst. A printer attached to LabX can produce a label for the volumetric flask containing the requisite quality or regulatory information such as identity, concentration, expiry date, etc. When complete, the validated process simply needs the analyst to electronically sign what has been done. The reviewer’s tasks in an electronic system are greatly simplified as the process is enforced by the software. No checks for transcription or calculation errors are required, as the whole process and records are held within a single system. Once the data are checked they can be electronically signed. There is an option to print out the record, although this is not strictly necessary unless required by local procedures or practices.

Advantages of the Electronic Process

There are a number of advantages to the electronic process. Some of them are:

Elimination of Manual Data Entry: There is no manual data entry in the process; all data are captured via LabX server.

Elimination of Transcription Errors: All transcription errors have been eliminated, which improves on the situation with the other two processes shown in Figure 1 and Figure 2. Now there is a single process controlled by LabX with automatic data capture that eliminates any human recording of data. The analyst can focus on scientific work rather than clerical tasks.

Single System Log-On: Interfacing an analytical balance to LabX turns its display screen into a terminal for interaction with the software. A user logs on to the system via the balance screen rather than at a separate workstation; no separate terminal is necessary for operating LabX.

Fast, Efficient Process: The electronic process is faster than the paper based ones and therefore saves laboratory time and effort.

Conclusion

The focus needs to be centred on reduction of errors, improving data integrity and quality as well as laboratory efficiency when using an analytical balance. Starting with a purely manual process we have seen the problems that can occur and the issue of traceability due to a lack of documented evidence. With the addition of a printer, errors can be reduced, data integrity and compliance can be raised and some improvements in laboratory efficiency can be accomplished. However, for the greatest productivity improvements and best data integrity, an electronic process with on-line data capture, data processing plus review and approvals using electronic signatures is highly recommended. Technology using Radio Frequency Identification (RFID) also enables the new range of Excellence Plus analytical balances to improve traceability and integrity of data whilst increasing productivity for pipette calibration and titration analysis.

The Unknown Sources of Error in the Pharmaceutical Weighing Process

Thomas Caratsch has been the Head of Laboratory Division since December 2007 at the company's headquarters in Greifensee, Switzerland. Prior to joining the company, Caratsch held various management positions within the Hoffmann La Roche group in the U.S. and Switzerland, including General Manager of Roche Instrument Centre AG and Head of Disetronic Medical Systems AG. He began his career as a development engineer at Sodeco-Saia in Geneva.

TOP