Part 1 of this three-part series provided a procedure for quickly recording acceptable fluorescence spectra with classic commercial spectrofluorometers for most samples in common 1-cm-pathlength cuvettes. Part 2 of the series describes instrument-specific concerns that require modifications to our initial procedure if we want spectra that can be accurately reproduced in other laboratories. Part 3 will discuss how the sample itself can make reproducibility between instruments challenging.

In Part I of this series, we described a simple procedure to select conditions under which fluorescence excitation and emission spectra are acquired with a commercial spectrofluorometer for the most common type of sample—a solution of emitting molecules in a common 1-cm-pathlength cuvette. This procedure is satisfactory for routine samples, particularly when the results are for relative or qualitative purposes. However, when comparisons are needed between different laboratories or different instruments, the differences between instrument responses become important to understand and rectify.

There are many reasons that spectrofluorometry of a routine sample might be inconsistent from instrument to instrument. This part of the series describes at a high level some of the reasons for these inconsistencies even when you “collect data right the first time,” and provides a procedure for how to correct for instrumental responses and artifacts.

Figure 1 in Part 1 provided a schematic diagram of a conventional spectrofluorometer with a continuous light source, and defines horizontal and vertical directions with respect to light polarization. For reproducible spectrofluorometry, we need to reproduce the instrument settings and filters at a minimum. However, even with these factors kept constant, there are reasons fluorescence data recorded on different instruments or at different times on the same instrument might be inconsistent with one another.

Beginning with the lamp and ending with the emission detector, every element of the instrument that produces, detects, or interacts with light can result in instrument-specific effects on a measured spectrum. Nearly all interactions can affect the light intensity or polarization to a greater or lesser degree.

Some of the instrumental factors that can affect the apparent spectrum and that were not discussed in Part I include the following:

The lamp intensity and spectrum can vary from instrument to instrument and over time

Transmission of light from the light source into the excitation monochromator can vary with wavelength and over time

The excitation spectrometer optics can affect the overall throughput and the polarization of light reaching the sample

The emission spectrometer optics will affect the throughput and polarization of fluorescence reaching the detector

The reference detector and its beam splitter have overall relative efficiencies that vary with wavelength and polarization of the beam

Any optical filters used in the system will introduce their spectral features, which may vary with the degree of collimation of the beam

Colored glass filters used in the spectrofluorometer can fluoresce and add to the background of the measurement (1)

Any interference-type filter used in the system will affect horizontal and vertical polarizations differently if not used at normal incidence

The cuvette affects reflection and transmission losses as a function of its material

The point of intersection of the excitation and detection optical paths inside the sample may vary, affecting the relative amount of inner filter effects that are observed

The cuvette positioning affects the point of overlap of the excitation and detection volumes inside the sample

A poorly chosen or poorly cleaned cuvette may have a measurable auto-fluorescence of its own, or may have a measurable absorption spectrum that interferes with the spectroscopy of the sample it contains (for example, optical glass in the UV)

Some cuvettes may not be intended for precision spectroscopy because of heterogeneities in the material

Using different sample geometries (90-degree, 180-degree, square or triangular cuvettes) will greatly affect self-absorption, polarization, and the sampling volume

The detectors have wavelength-dependent light detection efficiencies

Both monochromators may exhibit wavelength offset errors

Motor drive mechanisms could be malfunctioning leading to poor repeatability of the wavelength

Each monochromator may exhibit errors in spectral bandwidth (SBW)

The monochromator slits may not move evenly, causing the wavelength offset errors to vary with slit width

Gear mechanisms for moving gratings in both monochromators may exhibit hysteresis

Recorded spectra may have a background or intensity offset that varies with wavelength.

Reference detectors are commonly used to reduce the dependence of the spectrum on many of these factors in the excitation system. Pre-installed calibration files for the main detector may be used to largely correct for a number of these factors in the emission system. However, these compensations are not found universally, and even with them, there are factors that are not fully compensated or that change over time and require recalibration. As a result, we still observe wavelength- and polarization- dependent factors that vary between different instruments, or even between measurements made on the same instrument if parts have been replaced, if the alignment changes, if calibration files have been updated, removed, added, lost, or if components of the instrument have aged. In addition, slow phosphorescence of the sample could make a spectral profile dependent on how quickly the overall measurement is made, which could also vary between users or instruments.

Procedure: Evaluate → Collect → Correct

In Part 1, we described the method for choosing initial conditions for measurement. In Part 2, we focus on the process of modifying our initial conditions if necessary and correcting measurements for instrument response and artifacts. The procedure we are about to introduce is appropriate for situations in which you are trying to generate data for routine samples that will be reproducible on other instruments and in the future. We recommend that certain aspects (wavelength errors, SBW errors, range of linearity, polarization) of the instrument performance be determined before data is collected. By doing so ahead of data collection, final throughput correction procedures based on fluorescence standards and calibrated detectors can be applied properly.

We need to evaluate our system for wavelength errors in the excitation and emission monochromators, SBW errors, and detector linearity as a starting point to collecting data for interlaboratory comparisons. We also need to evaluate the need for polarization compensation in our instrument. These factors cannot be readily repaired after our corrected spectrum is created in the end.

Evaluation of the wavelength precision, SBW precision, and detector linearity are adapted from the literature (2), which provides much more information and is recommended to the reader.

A gear mechanism moves optical components of most monochromators. To avoid the effects of gear backlash or hysteresis on wavelength precision, all measurements should be recorded in the same scan direction (either increasing wavelength or decreasing wavelength). To ensure the wavelength of the first points in the spectrum are correct, we recommend over-scanning by a few nanometers. For example, if we are going to make our measurements by scanning downward in wavelength starting from 600 nm, we might want to begin the scan a few nanometers higher in wavelength so that gear backlash is removed by the time the monochromator reaches 600 nm. If scanning upward, we might start the scan a few nm lower in wavelength than the planned starting point for the same reason. Commercial instruments may automatically overscan the wavelength axis to avoid this problem.

Small atomic (mercury, neon, argon, krypton, xenon) lamps that are approximately the size and shape of pencils and that provide well-known reference wavelengths are common tools in spectroscopy. There are various manufacturers of these inexpensive lamps with different brand names. If a lamp of this type is placed in the position of the sample so that the emission monochromator entrance is illuminated, and if the entrance slit is narrowed to give the highest resolution, the apparent emission wavelengths of the atomic lines from these lamps can be measured and used to correct the emission monochromator wavelength axis. Any filters placed in the spectrofluorometer need to be removed before beginning.

These lamps are difficult to use to correct the excitation wavelength. One common approach to wavelength correction of the excitation monochromator is to use the emission monochromator as a reference. It requires the emission monochromator to have its wavelength axis determined first, and is only as good as the emission monochromator wavelength correction.

The first step in this procedure is to select an emission wavelength, place a cuvette containing an aqueous dilute scattering solution of your choice into the sample compartment, and record an excitation spectrum. Large peaks in the excitation spectrum will appear at the selected wavelength of the emission monochromator. Large peaks can also appear at 1⁄2 of that wavelength because of higher orders of the emission monochromator grating if those wavelengths are produced by a lamp. A few well-chosen emission monochromator settings can potentially provide several points for correction of the excitation monochromator wavelength axis. The accuracy of the emission wavelength correction is ±0.1 nm or better (2). The excitation wavelength correction done this way is not quite as good (±0.2 nm). Because of gear backlash, wavelength measurements should be recorded while scanning in the same direction that will be used in spectrofluorometry of the target compound.

The motors that drive the monochromators and slits are typically stepper motors, which require a driver to sequence the phases of the motor. If a phase of the motor fails, or if the driver board becomes intermittent, or if a connection becomes intermittent, it is possible for the driver to step inaccurately. A symptom of this failing would be a change in the sound of the instrument as it changes wavelengths, or an inability to return to the same wavelength. Some instruments may combine motors with an indexer that reads the position of the motors independently; if so, then the indexer may be susceptible to mechanical or electrical failures, also leading to poor repeatability. Repeatability of the monochromators can be tested by repeating the calibration runs several times and overlaying the scans to look for “jitter” in the peak positions. Poor repeatability is cause for an immediate call to the manufacturer.

The slit width is the main variable used to determine the SBW of a monochromator. The slits are formed as a space between two metal plates that each resemble knife-edges; the slits are each supposed to travel the same distance from the closed position so that the center wavelength is always the same regardless of SBW. The same pencil-type lamp used above can be used to determine the SBW of the emission monochromator by assuming the observed width of isolated atomic lines are solely because of the SBW of the monochromator. In this case, SBW is determined as full-width at half maximum (fwhm) of a measured peak. The excitation monochromator again requires a different procedure. By setting the SBW of the emission monochromator to its lowest setting, the SBW of the excitation monochromator can be estimated by recording an excitation spectrum of a dilute scattering solution and repeating the fwhm procedure. These procedures are thought to be subject to an uncertainty of ±0.5 nm in SBW (2). This test should be performed at a few different wavelengths to understand how any SBW error varies with wavelength. If there is any significant error in SBW, and particularly if the SBW error is a function of SBW, one possibility is that at least one side of the mechanical slit is not moving properly. If no fix or repair is possible, the wavelength error determined may be found to be a function of SBW; it is also likely the SBW will not be repeatable if the slit is not moving properly. As with wavelength unrepeatability, any indication of a SBW unrepeatability is cause for an immediate call to the manufacturer.

Even the best detectors become nonlinear at sufficiently high light intensities. Therefore, it is important to know the linear range over which your instrument operates. One approach is to measure a series of known concentrations of a fluorophore whose emission wavelength is similar to your target compound. To avoid nonlinearity from inner filter effects, the concentration of these standards should be kept very low (below 0.05 absorbance units). The peak excitation and emission wavelengths of the standards can be used, and the series of measured intensities can be plotted and fit to ensure they are consistent with the blank and show no systematic deviation from a straight line. If the highest detected signal is still linear, the calibration measurements can be repeated with an excitation SBW up to the limit described in Part I of this series, and an even wider SBW for the emission monochromator, to give a wide intensity range for the standard concentrations. The linearity of a detector, such as a photomultiplier tube, may change for different accelerating voltages, so care should be taken to use the detector as intended for your study. When changing the contents of a cuvette, the cuvette should be emptied, rinsed with clean solvent, rinsed with the next sample, emptied again, and then refilled before the next measurement. This procedure avoids cross-contamination and dilution of the sample.

If the fluorescence of a sample in the 90-degree sample geometry is at least partially polarized, there will be effects on its measured intensity and probably at least its measured excitation spectrum. Counterintuitively, this is true even if there are no polarization artifacts in any of the optics. Fortunately, polarized fluorescence is not generally a problem for small molecules in low viscosity solvents. However, even small molecules can exhibit polarized fluorescence if their fluorescence lifetime is short enough (3), whereas large molecules and viscous solvents invite this problem. If you are concerned that you might need to eliminate fluorescence polarization as a complication, then you can add a “vertical” polarizer in the same location as the excitation filter, and insert another polarizer in the same location as the emission filter to serve as a polarization analyzer. This orientation of the emission polarizer should be set to the 54.7° magic angle to make the measurement independent of the fluorescence polarization of the sample (4).

The conditions for measurement can be determined using the methods of Part I, or they can be optimized from that starting point. The following points are specific to measurements made for publication and reproducibility:

The conditions of measurement must be controlled and recorded. This includes making sure the instrument scan range, step size, spectral bandwidth, number of points per unit wavelength, and integration or dwell time, are appropriate for the sample and that they are all recorded and preserved. It also means that any steps taken to evaluate the wavelength accuracy, spectral bandwidth accuracy, photometric accuracy, and to prevent dependence on sample polarization, should be recorded and preserved

The clean sample cuvette should be handled with gloves and carefully wiped on all its optical surfaces with a fresh lint-free paper or cloth to remove dust or surface contaminants before being placed into the spectrofluorometer for measurements

If your instrument has an automated shutter to block the excitation beam, this should be engaged any time a measurement is not being made

A background spectrum of the pure solvent should be recorded under the same conditions as the sample. Ensure that the same cuvette is used for the background and for the sample, and that it is placed into the instrument in the same orientation. This procedure helps to account for any stray light, dark signal, amplifier switching, or filter fluorescence that might take place inside the spectrofluorometer, and can be subtracted from the measured spectrum before corrections are applied.

At this point, all concerns about critical aspects of the instrument that cannot be easily corrected post-measurement have been addressed. What remains is the part of the instrument performance that affects the throughput of the excitation source and the emission of their respective monochromators. On the excitation side, the reference detector, if present, corrects for many of the wavelength-dependent behaviors of the instrumentation. Without the reference detector, fluorescence excitation spectra could be so heavily convolved with the instrument response they would be almost uninterpretable without correction. However, the response of the reference detector itself and the effects of optical components after the reference detector still require throughput corrections.

There are several alternate procedures for implementing these throughput corrections and we describe only the simplest. At present, excitation and emission spectra require different approaches for their correction. The National Institutes of Standards and Technology (NIST) and the Federal Institute for Materials Research and Testing in Germany (BAM) are both excellent sources for details on instrument concerns and corrections that we recommend to the reader (1,4).

For most users, certified reference materials or secondary emission standards are the simplest approaches to correcting emission spectra. An excellent alternative is to use a calibrated lamp, but the standards offer the opportunity to use a cuvette-type sample.

A certified reference material is either a solution or glass that can be placed into a spectrofluorometer cuvette holder, and that has known excitation or emission properties. These materials are certified by, for instance, NIST in the United States and equivalent organizations in other countries. As an example, NIST has certified a series of standard reference materials that include solid glass, cuvette-shaped standards that can be placed directly into a spectrofluorometer (SRM 2943 and SRM 2944 for the spectral ranges 350–640 nm and 530–830 nm, respectively). Certified reference materials are relatively expensive compared to available secondary emission standards (5), though both are used in the same way for the same purpose.

Whether a certified reference material or a secondary standard, what these reference materials have in common is a defined emission spectrum under defined measurement conditions. The measured emission spectrum of the reference material in your instrument is compared to its certified or reported spectrum to determine a correction factor over the range of wavelengths for which it has been evaluated. This correction factor is then used to correct the measured spectrum of your target fluorescence sample.

There are not currently any primary fluorescence excitation spectroscopy standards like those for fluorescence emission spectroscopy. The simplest available approach to correcting fluorescence excitation spectra is to use a calibrated detector. In this case, the detector is placed in the location of the cuvette to intercept the excitation beam, and its corrected signal is recorded as a function of excitation wavelength. The reference signal of the spectrofluorometer is recorded at the same time. The ratio of these two responses then becomes a correction factor for the wavelength range over which the detector has been calibrated. If there is no reference channel, the corrected signal of the calibrated detector is used to form the correction factor by itself.

Although this article covers spectrofluorometry only at a very high level without much detail, it is still restricted to only one measurement case (solution cuvettes) in steady-state spectrofluorometers. The wide field of spectrofluorometry includes many more sample types and configurations, along with a wide variety of instrumentation, all suited to particular purposes. If the authors could leave the reader with two important messages, the first would be this: Read the user manual for your instrument, read literature about your measurement type, and read about your sample type to prepare yourself for making your own measurements. Having this background knowledge ahead of time will enable you to avoid as many errors as possible and enable you to compare your own measurements to those in the literature to assess whether your measurements are reasonable or not. The second message is that there are many parameters that should be recorded and transmitted about spectrofluorometry measurements in addition to the basics of the instrument settings, without which reproducing your spectra could become very difficult for others.

(2) P.C. DeRose, “Standard Guide to Fluorescence – Instrument Calibration and Validation,” NIST Interagency/Internal Report (NISTIR) – 7458 (National Institute of Standards and Technology, Gaithersburg, MD, 2008, accessed September 27, 2021). https://doi.org/10.6028/NIST.IR.7458

, 33–41 (1969). https://doi.org/10.1016/0003-2697(69)90238-3

Standardization and Quality Assurance in Fluorescence Measurements I

 (Springer-Verlag, Berlin, 2008), vol. 5.

, 1179–1189 (1998). https://doi.org/10.1366/0003702981945192.

 are with the University of South Carolina, in Columbia, South Carolina. Direct correspondence to: 

Articles

Columns

In the second part of a three-part series, instrument-specific concerns that require modifications to reproduce fluorescence spectra accurately are addressed.

Challenges of Spectrofluorometry, Part 2: Instrument-Specific Concerns

A tutorial and spreadsheet for the validation and uncertainty evaluation for ICP-MS analysis was successfully applied to determine multiple elements in a nasal spray.

Instrumental methods of analysis are frequently used to determine the level of a compound, class of compounds, or other chemical property on the studied item. The ability to adequately quantify measurement result uncertainty is crucial for the objective that triggered the chemical analysis. A tutorial and spreadsheet for the validation and bottom-up uncertainty evaluation of quantifications performed by instrumental methods of analysis based on linear weighted calibrations was presented by Ricardo J.N. Bettencourt da Silva of the University of Lisbon in Lisbon, Portugal, and his colleagues. This software tool was successfully applied to the determination of the mass concentration of cadmium, lead, arsenic, mercury, cobalt, vanadium, and nickel in a nasal spray by inductively coupled plasma–mass spectrometry (ICP-MS) after samples dilution and acidification. Bettencourt da Silva spoke to 

 about applying this software tool and the implications for a better understanding of quantitative analytical results.

Your recent paper (1) describes a tutorial and spreadsheet for the validation and bottom-up uncertainty evaluation for quantitative instrumental methods based on a linear weighted regression model. Why do you believe that there is a need for such a tool? What sort of information or insight does it provide that currently available methods or tools do not? 

The developed tool has two merits: It has the ability to make accessible complex uncertainty evaluations, and it can evaluate the uncertainty from weighted calibrations without the need to collect many replicate signals of calibrators.

The paper includes a user-friendly Microsoft Excel file that can be used to assess the assumptions and evaluate the uncertainty from linear weighted calibrations of instrumental methods of analysis. It is provided with metrological and statistical tools to test the quality of the calibrator and the instrumental response linearity where different weighting factors can be selected according to how precision varies in the calibration interval. The disadvantage of weighted regression models, requiring the reliable determination of signal precision in each calibration to better set the weighting factors, is overcome by defining precision models that can be updated with the residual standard deviation estimated from daily calibrations based on a few replicate signals.

This tool is particularly useful because most instrumental methods of analysis are calibrated in intervals where the signal varies linearly with the studied quantity, typically the concentration of an analyte, and the precision of the signal varies in that interval. Before the publication of this paper, such a tool was not available that can be easily integrated into the daily activities of test laboratories.

Your spreadsheet tool was successfully applied to the quantitative determination of the mass concentration of cadmium, lead, arsenic, mercury, cobalt, vanadium, and nickel in a nasal spray by ICP-MS after sample dilution and acidification. What made you decide to choose this particular application to test your tool? What benefits did the utilization of this spreadsheet bring to the analysis? 

This application was selected because of the relevance of the quality control (QC) of medicines and because these analyses do not include a complex sample preparation that constitutes an additional relevant uncertainty component. In these analyses performed by experts from the Palacky University in Olomouc (Czechia) in their Good Manufacturing Practice (GMP)-certified laboratory, the measurement uncertainty is estimated by combining calibrators’ values uncertainty with the uncertainty from the regression model used to define the calibration curve. Therefore, the experimental application of the developed tool to the analysis of nasal sprays spiked with studied analytes allowed a detailed assessment of the developed uncertainty models.

Briefly discuss your findings and their implications for a better understanding of quantitative analytical results. 

The paper presents experimental evidence of the adequacy of the developed measurement uncertainty models and constitutes a relevant contribution to the democratization (making the tool accessible to everyone). of complex measurement uncertainty evaluations. The quantification of measurement uncertainty is crucial in guaranteeing that performed analyses are fit for purpose and have known qualities. Measurement results are only valid for a specific purpose if their uncertainty is adequately low. The measurement uncertainty is also crucial for the sound and objective interpretation of the analytical information.

What are the biggest challenges that you have encountered in developing this tool? What options or alternative developments are available to overcome these challenges or to improve this spreadsheet approach?

The biggest challenges we have faced in developing this tool were making it user-friendly and avoiding the need to collect many replicate signals for the daily calibration of an instrumental method of analysis based on the linear weighted regression model. Some authors have developed software for evaluating the uncertainty of instrumental quantifications that require demanding and not supervised collections of input data. A spreadsheet for the Monte Carlo simulation of instrumental quantification uncertainty was also developed but requires some computational resources for the simulation.

What sort of feedback did you receive from this paper and the study results?

We have received very positive feedback from the pharmaceutical industry that is becoming more interested in measurement uncertainty evaluations to guarantee conformity assessment reliability. However, the developed tool can be applied to any instrumental method of analysis.

What are your next steps regarding this research? Are there any additional analytical approaches or methods where this tool might be beneficial?

We are developing equivalent tools for nonlinear regressions and integrating complex sample preparation in analytical methods based on instrumental quantification. These research efforts will involve the development of new metrological tools and their democratization through tutorials and user-friendly software.

(1) T. Pluháček, D. Milde, J. Součková, and R.J.N. Bettencourt da Silva, 

, 122044 (2021). https://doi.org/10.1016/j.talanta.2020.122044

Determination of Elemental Impurities in a Nasal Spray by ICP-MS

The new PIC/S data guidance document was published in July 2021. The guidance will be reviewed in two parts, starting with the scope, data governance, and paper records.

Blue whales have a gestation of 12 months, whereas elephants take 22 months; in contrast, the gestation period for the Pharmaceutical Inspection Co-operation Scheme (PIC/S) PI-041 data integrity guidance has been five years. The first draft was issued in April 2016; draft 3 was issued for public comment in November 2018, with the final version issued in July 2021. The formal title is 

Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments

, with a document number PI-041 (1), but the focus is data integrity. The document scope covers all activities under the remits of good manufacturing practice (GMP) and good distribution practice (GDP) regulations. This column only focuses on laboratories working to GMP regulations. The review of this guidance document will be in two parts:

Part 1, here, discusses the scope of the guidance, data governance, and paper records

Part 2 will focus on computerized systems, outsourcing, and regulatory and quality oversight.

Another DI Guidance? You Cannot Be Serious!

The rationale for another regulatory guidance document on data integrity is the continued focus of health authorities on numerous issues of data falsification and poor data management practices, and the basic inability of companies globally to comply with GMP regulations. The PIC/S guidance joins other data integrity guidances from other regulatory authorities, such as the World Health Organization (WHO) (2), the U.K. Medicines and Healthcare Products Regulatory Agency (MHRA) (3,4), and the U.S. Food and Drug Administration (FDA) (5), who have issued guidance documents on the topic. Recently, the Organization of Economic Cooperation and Development (OECD) also issued a good laboratory practice (GLP) data integrity guidance (6).

In addition, there are data integrity publications from scientific societies and industry bodies such as the GAMP Forum (7–9), the Parenteral Drug Association (10), and the Active Pharmaceutical Ingredient Committee (APIC) (11).

You may be forgiven for asking: Do we really need another data integrity guidance document, and who or what is PIC/S anyway? Let’s start with the second part of the question first: Who or what is PIC/S?

Pharmaceutical Inspection Co-operation Scheme

, but documents still have both names on them—hence the abbreviation of both to PIC/S. PIC/S is a global organization that is essentially a club of GMP inspectors from more than 50 regulatory authorities, from Europe, the United States, Canada, Japan, Australia, South Korea, and Singapore. A regulatory authority applies to join the scheme, and, after an assessment by other PIC/S members, the authority is admitted if it meets the organization’s criteria. PIC/S has its own GMP regulations, which are adopted by most of their members, the exception being the U.S. FDA, which still use 21 

In addition, PIC/S publishes regulatory guidance available on its website (www.picscheme.org) for both regulators and industry to download. When a document is to be written, an expert circle is convened, consisting of inspectors with expertise in the subject, who work on the various drafts. Document naming is important:

PIC/S External (PE) documents written by inspectors for industry. 

PIC/S Internal (PI) documents written for inspectors, some of which are made available for industry, one of which is PI-041 on data integrity.

PIC/S PI-041: Good Practices for Data Management and Integrity

PIC/S PI-041 has been written to help with onsite inspections of GMP facilities; however, in this two-part article, we focus on analytical development and quality control laboratories. The overall structure of the new data integrity document is shown in Figure 1, and it shows the scope of the guidance for data integrity within a regulated organization:

Section 5: Data Governance within a Pharmaceutical Quality System.

Section 6: Organization Issues, such as staff values, quality, ethics, and conduct.

Section 7: General data integrity principles.

Section 8: Considerations for paper records, including the control of blank forms and master templates.

Section 9: Considerations for computerized systems, including hybrid systems.

Section 10: Outsourcing and data integrity.

Section 11: Regulatory actions in response to data integrity findings.

Section 12: Remediation of data integrity failures.

I’ll not bother you with Sections I–4, as they only cover the document history, introduction, purpose and scope.

The only other regulatory guidance with equivalent breadth is the WHO guide issued in 2016 (2). Although there are common sections between the PIC/S and WHO guidances, the best discussion of the ALCOA (attributable, legible, contemporaneous, original, and accurate) criteria for data integrity is found in Appendix 1 of the WHO guidance. Although the PIC/S guidance discusses all nine ALCOA+ criteria (the “+” refers to “complete, consistent, enduring, and available”), it does not provide the in-depth coverage that is found in WHO guidance.

The final version of PI-041 guidance document is now 63 pages, compared with 53 of the 2018 draft. It is easily the longest regulatory data integrity guidance. We examine below whether the wait for the final version was worth it.

If you read any recent FDA guidance for industry, you will see the definition of the word “should”:

In general, FDA’s guidance documents do not establish legally enforceable responsibilities. Instead, guidances describe the Agency’s current thinking on a topic, and should be viewed only as recommendations, unless specific regulatory or statutory requirements are cited. The use of the word “should” in Agency guidances means that something is suggested or recommended, but not required (5).

Paul Smith has performed an analysis of the four main regulatory authority data integrity guidance documents comparing the use of must and should, as shown in Figure 2. First, there are very few instances of must; and the majority of them are in the FDA and MHRA guidances. Second, the clear winner is the PIC/S guidance with over 500 times where should is used, and there are no instances of the word must. One reason the word 

 appears so many times is simply the length of the document, which is nearly twice that of the WHO and three times that of the MHRA and FDA guidance documents. This, of course, contrasts with the use of the word 

 in regulations, which can be interpreted simply as “Do it, or you’re toast.”

Data governance (DG) has been a regulatory expectation since the MHRA presented the UK pharmaceutical industry with an early Christmas present in December 2013. The agency posted on its website the requirement for internal audits to cover data governance, and, by the way, added that it would start inspections for this in January 2014 (so have a Happy New Year)! The requirement for data governance has been reiterated in both the MHRA GMP and GXP guidance documents (3,4). In contrast, PI-041 states in 5.1.1 that there is no current GMP regulation to implement a data governance system, but offers help to a laboratory to define, prioritize, and communicate DG risk management activities (1).

The problem with the guidance is that it can appear paranoid; the title of the section is “DG System,” but the heading of section 5.2 refers to “DG Systems” in plural (1). This situation is also repeated in individual clauses in section 5.2, where the clauses flip between singular and plural. Which is it, especially, given that there is no regulatory requirement for DG? Also, the definition of DG (no system and apparently singular) in the glossary refers simply to all arrangements to ensure the integrity of any GMP records throughout the data lifecycle (1).

The main plank of data governance is to use computerized systems to automate processes and enable technical controls to ensure and enforce the integrity of data, coupled with validation to demonstrate that the system works as expected. However, data governance does not forget the role an organization has to play, through procedures, training, data review, and quality oversight, as shown in my data integrity model that has been presented in several publications of mine over the past few years (12–15).

We must wait until clauses 5.2.4 and 5.3.1 before we mention the key group accountable for data integrity in an organization—senior management (1). When earlier sections of the guide discuss the pharmaceutical quality system (PQS), they do not mention senior management. We must wait until Section 6 , which delves into organizational influences on successful data integrity management, including a detailed discussion on the role of senior management. For example (1):

Leading, resourcing, and funding the data integrity and data governance initiative in an organization. This is a program that will take more than an email from the CEO stating that from now on the company will ensure data integrity; rather, it requires a concerted and prolonged effort over time. Data integrity is a journey, not an event.

Ensuring that there is a culture and ethos for honesty, and allowing people to admit mistakes.

Establishing an open culture with employee empowerment to raise integrity issues through the PQS. The guide does not acknowledge that there may be cultural issues that may make this difficult to achieve, and that there will be reliance on quality oversight to achieve the goal.

Establishing and maintaining a quality culture.

Setting and maintaining the ethos and values for data integrity.

Establishing policies and procedures with associated training for all aspects of data integrity.

Ensuring assessments of processes and systems are carried out to identify and classify data vulnerabilities. Carry out both short term remediation (quick fixes) as well as implementing long term solutions providing business benefit.

Setting up metrics for various data integrity activities and management review.

Conducting regular and meaningful review (this is not a once-a-year, box ticking exercise) of data integrity projects.

Setting up of a protected whistleblower process to raise data integrity issues to senior management for their attention.

If you look at the structure of Sections 5 and 6 of PI-041, shown in Figure 3, there is no subsection specifically dealing with senior management and its role in ensuring data integrity. This is surprising, given that ICH Q10 (16) and EU GMP Chapter 1 (17) for pharmaceutical quality systems (PQS) both hold senior management accountable for the PQS in any GMP organization, and for all work carried out under it. From the regulations, and in my view, there should be a specific section early in the document focused on senior management, rather than dispersing it in different sections of the guidance. A similar situation occurred where senior management was omitted from the draft of the FDA data integrity guidance (18), and added as an afterthought in the background section of the final document (5).

Risk management and its role in data integrity and data governance is covered in Sections 5.3 through 5.5. As that well known data integrity expert, George Orwell, noted, “All data are created equal, but some are more equal than others.” Therefore, it is the role of risk management to identify systems and processes that create and manage data that are more critical, so that they can be assessed first to identify data vulnerabilities and mitigate them.

Candidates at the front of the queue are processes that are complex, inconsistent, or open ended, and generate data used in product submissions or batch release. Also, manual, paper based processes can be problematic, as we shall see later in this article. Ideally, short-term remediation (mainly procedural with some simple technical fixes) should be applied for quick fixes, with longer-term solutions implemented (automation with technical controls) to have a process or system to ensure data integrity.

Organizational Influences on Data Integrity

The scope of Section 6, shown in Figure 3, covers some critical areas for data integrity (1). These include the following:

Expectations are set for staff ethics and behavior with respect to recording, interpreting, calculating, and reporting data that are clearly communicated to all.

Management needs to ensure that staff are aware that breaching ethics and behaviors could result in disciplinary procedures.

Unacceptable behaviors must be identified, documented in policies, and communicated clearly to staff, along with the range of company actions if they commit unacceptable behavior.

Quality culture is the responsibility of management to establish and foster. In my view, this is the most difficult part of any data integrity program, and requires management to lead by example

Section 6.4 deals with modernizing the PQS so that it is able to detect and correct weaknesses that could lead to data integrity lapses. Particular areas for a laboratory are second-person review, quality oversight, and the purchase of instrumentation and systems to ensure data integrity. This latter topic will be covered in Part 2 of this review of PI-041.

This subhead is the title of Section 7 of the PIC/S guidance, and it emphasizes good documentation practices (GDocP) and discusses the nine ALCOA+ criteria (1). Section 7.5 is essentially a table covering the criteria, and the requirements for each one.

Appendix 1 of the WHO guidance (2) provides more comprehensive coverage of ALCOA principles with:

a table showing the expectations of paper and electronic records side by side for each term

a presentation of special risk factors for each term.

However, later sections of the PIC/S document provide further information on ALCOA+, but it is not highlighted. My suggestion is to use the definitions of ALCOA+ in section 7.5 of PIC/S PI-041 as an overview (1), but to use Appendix 1 of the WHO guidance (2) for more detailed reference and in your data integrity program.

Section 7 finishes with guidance on how to create a true copy of a record, and the limitation of remote review of data in summary reports.

Section 7.7.2 states that it is conceivable for raw data (later we will discuss the fact that the definition of raw data in this document is wrong) generated by electronic means to be retained in an acceptable paper or PDF format. It then mentions that, as well as the data, all direct and in- direct metadata also need to be printed (1). Having the option for companies to justify printing to paper in a guidance document is wrong, as most will ignore the caveats in this section and just do it. My opinion is to ignore this section.

PI-041 notes in section 7.8 that analytical summary reports from contract laboratories are limited, and that supporting data and associated metadata are often not included, as the original data cannot be reviewed (1). Thus, the outsourcing laboratory must conduct due diligence and assess the contract facility’s PQS and data integrity program, as well as being able to conduct on-site audits when appropriate.

Data Integrity Considerations for Paper Records

Covering approximately 13 pages of the guidance, there is an in-depth treatise of data integrity considerations for paper records needed for following GDocP and meeting the ALCOA+ criteria (1), as shown in Figure 4. First up, the old chestnut of control of master templates and blank forms for which there is accountability, coupled with a discussion of the importance of controlling records. Although ignored by many laboratories, control of blank forms is not new and has been required by the FDA since 1993 (19) and reiterated in the data integrity guidance documents from WHO, MHRA, FDA (2–5), and now PIC/S PI-041(1).

If management fully implemented the requirements of this and other guidance documents on blank forms, they would realize the overall cost and fund automation projects rapidly to remove paper-based processes from the laboratory. Chris Burgess and I wrote a “Questions of Quality” column on this subject where more detail can be found (20). I also refer you to a recent article of mine on the “Hidden Factory in Your Laboratory,” whose only product was paper printouts (21).

Section 8.4 brings the first of many tables on specific subjects, in this case expectations for the generation, distribution, and control of records (1). The table presents:

Expectation: All documents should have an unique identifier, and uncontrolled documents should be prohibited.

Potential risk of not meeting expectation: It is easier to falsify uncontrolled documents, and increases the potential for loss of official records by discarding or destruction.

This approach is very good and clear. My only concern is that I hope that analysts read and understand the whole of Section 8 to understand the requirements for paper records. Reading through this section, you find requirements that original from EU GMP Chapter 4 on documentation (22). Perhaps there should be a few more instances of must in this section of the guidance?

Sections 8.5 to 8.7 are tables of basic GDocP requirements for paper records, from control of records at point of use to how to complete a record to how to make corrections. While these sections are basic GMP requirements, the content is the basis for a good training course in GDocP; all it needs to complete the course are pictures of right and wrong examples. Second-person review of paper records is covered in Section 8.8—another good source of training material.

The last topic in Section 8.9 that I will review here is direct printouts from electronic systems, such as standalone analytical balances and pH meters, containing measurements performed by an analyst. PI-041 notes that the original record should be signed and dated by the person generating the record, and information to ensure traceability (attribution, perhaps?). These original records should be attached to batch processing or testing records (1). There are a few points that need commenting:

Older analytical balances and similar instruments may not have the function to have user accounts, and so the analyst must (not “should,” as stated in PI-041) sign or initial and date the printout.

Newer balances have time and date stamps, as well as individual user accounts and passwords; each printout from the instrument will have the name of the analyst, as well as the date and time on it.

Surprisingly, there is no mention of an instrument log that should be completed in parallel with the analysis being performed. Logs provide critical correction, and are essential for ensuring data integrity. Instrument log books were discussed in an earlier “Focus on Quality” column (23).

Then, to finish this first part review of PIC/S PI-041, “moan” mode is turned on with the volume cranked up....

The Wrong Definition of Raw Data —Again!

There is a major inability of GMP inspectors to understand and grasp the meaning of the term raw data, and this is perpetuated in PIC/S PI-041. The term raw data is defined in the glossary as the original record (data) which can be described as the first-capture of information (1). Wrong, wrong, wrong! As we have discussed in this column previously (24,25), raw data is a GLP term that is defined in 21 

 58.3(k) as “the result of original observations and activities of a nonclinical laboratory study and are necessary for the reconstruction and evaluation of the report of that study. (26).” This definition is mirrored in OECD GLP regulations (27). Raw data should be considered the same as complete data as used in 21 

 211.194(a) (28), and means all data and records without selection, filtering, or deletion.

Interestingly, the new OECD GLP data integrity guidance, (6) which is the MHRA GXP guidance (4) fancied up for GLP, actually gets the raw data definition right!

This is the first of a two-part review of the new PIC/S PI-041 guidance on Good Practices for Data Management and Integrity in Regulated GMP/GDP Environments, and has focused on the scope, data governance, and paper records. In Part 2, we will look at what lies in store for computerized systems, and provide an overall summary of the guidance.

I would like to thank Paul Smith for Figure 2 and his analysis of data integrity guidance documents.

 (Pharmaceutical Inspection Convention/Pharmaceutical Inspection Cooperation Scheme, Geneva, Switzerland, 2021).

Annex 5 Guidance on Good Data and Records Management Practices

. (World Health Organization, Geneva, Switzerland, 2016).

MHRA GMP Data Integrity Definitions and Guidance for Industry, 2nd Edition

 (Medicines and Healthcare Products Regulatory Agency: London, United Kingdom, 2015).

MHRA GXP Data Integrity Guidance and Definitions

 (Medicines and Healthcare Products Regulatory Agency, London, United Kingdom, 2018)

FDA Guidance for Industry Data Integrity and Compliance With Drug CGMP Questions and Answers

OECD Series on Principles of Good Laboratory Practice (GLP) and Compliance Monitoring, Number 22, Advisory Document of the Working Party on Good Laboratory Practice on GLP Data Integrity

 (Organization of Economic Cooperation and Development, Paris, France, 2021).

GAMP Good Practice Guide: Data Integrity - Key Concepts

 (International Society for Pharmaceutical Engineering, Tampa, FL, 2018).

GAMP Good Practice Guide: Data Integrity by Design

 (International Society for Pharmaceutical Engineering, Tampa, FL, 2020).

Columns: Focus on Quality

In the first article of a two-part series, the new PIC/S guidance is discussed, with particular focus on the scope, data governance, and paper records.

Are You Ready for the Latest Data Integrity Guidance? Part 1: Scope, Data Governance, and Paper Records

By now, it is well known that Raman spectroscopy has the potential to be used for a variety of analytical applications, and that the hurdles to using the technique 30–40 years ago have been largely overcome. However, a remaining impediment to exploiting the technology is that very few analysts have the ability to know what they are looking at. Even if there is experience in infrared (IR) spectroscopy, which looks at similar molecular vibrations, making the Raman measurements and interpreting the results will be quite different. There are databases and searching programs that can aid the analyst, but using these resources effectively also requires a learning curve. What I am going to try to do in this article is provide some advice for dealing with those situations where the searching program does not provide a definitive result.

When using a database searching program (1–4), you need to keep in mind that the quality of the result will depend on whether spectra of your compounds are in the database. It will also depend on the quality of the spectra; in comparison to IR spectra, Raman spectra tend to be noisier. However, when a problem can only be solved with Raman spectroscopy, we need to know how to achieve the required information. Because the time available to an analyst is limited by multiple pressures in the industrial environment, I thought it would be useful to summarize some of the things I have been learning in the last few weeks. I have been analyzing spectra leached from polymers, as a follow-up to the work that I did two years ago on extractables and leachables (5). A previous column (6) discussed an earlier foray into data base searching. In this column, I am going to be more specific in differentiating different classes of organic molecules.

Optimal Cases Where the Library Searching Provides an Answer

The first example that I will show is the results from a deposit leached into water at 100 °C from a polymer container. Figure 1 shows our spectrum (in black) overlaid with the match from the database (in red). Inspection of the figure indicates a near perfect match, except for maybe the relative intensities. Note that the hit quality index (HQI) is 81—good, but not fabulous. Remember that the HQI is based on calculating the correlation coefficient (that is, the dot product of the query spectrum with the library spectrum). In this case, because the relative intensities do not match, the value for the HQI is somewhat degraded. But this supports the user applying human judgment on the results. After deciding that this result is a likely fit, one question needs to be asked—does it make sense chemically? In fact, one would not expect to see a metal oxide in a polymer sample unless there was residual metal oxide from the polymerization process that was carried with the polymer, or perhaps some other source of contamination.

The next example is a deposit collected from a solution from a polypropylene (PP) container. Figure 2a shows our spectrum of the deposit as well as the spectrum of the PP container. You can see that the bands do overlap, but there is a significant difference in the bandwidths and relative intensities of some of the sharp bands. Figure 2b shows the results of the database search; we have included the first and the 13th hits, which are respectively ethylene propylene ethylidene norbornene, and amorphous PP (aPP). The respective HQI values are 83 and 80, which is not much different. But because I knew that the composition of the container was PP, I felt that aPP is a more likely choice and some places in the spectrum confirm that. However, keep in mind that the database may not always be 100% accurate. The best practice is to combine what you know about your sample and its spectrum with what the database tells you.

The next spectrum that I show here has a background of carbonaceous material. Many spectra of organic unknowns that come from contamination or some kind of degraded species often have the carbon signature superimposed on the spectrum of the unknowns. It is important for the analyst to know this signature or many errors can occur; however, it is fraught with difficulty because the carbon Raman spectrum is quite variable. The carbons are related to graphite, which has a strong, reasonably sharp band called the G band at 1583 cm

. When it is disordered, a second band appears somewhere in the mid-1300 cm

 region; I am purposely imprecise about its position because the center of the band depends on the excitation wavelength. This band is called the D band, which over the years has meant disorder or diamond-like. In fact, the D band is not allowed in the strictest sense of solid-state physics, but it appears because of the double resonance with electronic transitions when the crystals are small or there is a defect, a phenomenon that is unique in graphite. Reich and Thomsen explained this phenomenon in great detail, and their explanation of it is in the literature (7). The correspondence between the non-zero phonon wavevector and the two electron transitions (incident and scattered photon) leads to the conclusion and observation that the phonon frequency will depend on the excitation wavelength, which is observed.

While doing analytical spectroscopy of chemical materials, it is not necessary to understand all of this information, but you must be aware of it because the presence of a carbon background will definitely have an impact on the ability of a database to match your spectra. You might think that it should come up with at least a two-part mixture of the carbon and the target analyte. But once you realize that the carbon spectrum first depends on the excitation wavelength, and then that the relative intensity and widths of the D and G bands depend on the physical state of the sample, you can realize that a database may have a difficult time finding the precise carbon spectrum to fit your spectrum.

Figure 3 shows spectra of carbon fibers (provided years ago by Herman Noether, who was retired from Celanese) that provide an idea of how much variability there is. The fiber made from pitch is the closest to graphite that has no band in the 1330–1370 cm

 region. The two fibers made from polyacrylonitrile (PAN) were made by different processes that accounts for the different spectra reflecting different microstructures. In addition, the top two spectra show an intense overtone at 2665 cm

, almost exactly twice the frequency of the D band. The spectral behavior of carbons is quite complex and interesting, which is why I am suggesting that you get some familiarity with these spectra (8).

In principle, library searching programs should be able to determine that a spectrum represents a mixture of an organic with a carbon. Figure 4 shows an example where library searching did work. Figure 4a shows the spectrum that I recorded with a micrograph indicating the material from which this spectrum was generated. Figure 4b shows the results of database search. Note that the gray regions in Figure 4b were deselected from the search so that those regions where there is no information did not contribute noise to the “result.”

What happens when the search does not easily produce something reasonable? In the course of searching many spectra over the past weeks, I have noticed that surprisingly, the CH region can be used for classification. Figure 5a shows the CH region for a number of organic materials. Olive oil is a triglyceride of mostly oleic acid, which has one double bond; the CH band at 3008 cm

 is an analytical band for olefinic CH. Stearic acid is representative of the class of saturated free fatty acids (FFAs), which are solid at room temperature. Any CH band below 3000 cm

 is normally attributed to a saturated hydrocarbon; whereas the CH bands in this region overlap, the overall structure of the spectrum is quite different and is indicative of liquid or amorphous hydrocarbons compared to solid hydrocarbons. The next two spectra represent two proteins—skin and tendon. Note how different the envelope is of the CH bands below 3000 cm

 from that of the lipid. This spectrum, peaked at ~2930 cm

Innovative database search technology can help Raman spectroscopists identify molecular vibrations; here, we show how to use these tools more effectively.

November 2021

Challenges of Spectrofluorometry, Part 2: Instrument-Specific Concerns

Determination of Elemental Impurities in a Nasal Spray by ICP-MS

Are You Ready for the Latest Data Integrity Guidance? Part 1: Scope, Data Governance, and Paper Records

A Simple Introduction to Raman Spectral Identification of Organic Materials

H-Classic Papers in Atomic Spectroscopy: An Integrative Literature Review

A Highly Selective Colorimetric and “Turn-Off” Fluorimetric Probe Based on Unsymmetrical Naphthalene Formaldehyde Derivatives for the Detection of CN− in Aqueous Media

The Infrared Spectra of Polymers III: Hydrocarbon Polymers