GxP-compliant calibration: webinar Q&A

Here are some answers to questions we did not have time to answer during our webinar:
"GxP compliant calibration learning from form 483s and warning letters"

The full transcript of the video interview is below.

[00:00:00] Today one of the webinar presenters, Heather Wade from Heather Wade Group, joins us to answer questions we didn’t have time to answer during the webinar.

Let's go to our first question: Does an internal Q.C. lab for a pharma company need ISO 17025 certification if they do not claim conformity to this standard?

[00:00:54] No, they don't need it because it's a voluntary standard, unless it’s required otherwise. And I'd also like to differentiate between certification and accreditation. It's a common confusion. Certification is for personnel or products, or you might be ISO 9000 registered, or certified because you received the certificate of registration. Accreditation is different and it's a more robust assessment of technical competence. So you want to be proud that if you have ISO 17025 accreditation, that you refer to it as 17025 accreditation.

And then adding on to that [regarding] accreditation; Almost all of the accreditation bodies in the world, those are the organizations that would assess, as a third party, [calibration] labs for 17025 accreditation and those you can easily script search their Web sites for a company name, or the parameters that a company is claiming accreditation for. So, you want to make sure that you go to that accreditation body’s Web site, or contact that accreditation body for the most current and up-to-date and correct accreditation scope for a company. Whether it is for testing, or for calibration under 17025. That's because unfortunately, sometimes on a company's Web site, they may have an outdated or sometimes even a falsified scope of accreditation, or claims that they're accredited. And the accreditation bodies work very hard to track these companies; they do that so that they can say: “Hey, this company is not accredited. Don't believe anything that they say.”

So, when you get your calibration certificate, you may get the stamp of accreditation logo, which is the evidence of metrological traceability. That's the other thing to remember. And as you're looking at the calibration certificate or the test report, anything that isn't on their scope of accreditation, they must say and denote what measurements are not part of their accreditation scope. So that's also really important thing to know. So if you're paying good money for a calibration certificate and you get that and you review it and you see, hey, there's something around here that have asterisks that aren't in their scope of accreditation, I need that. So that's something to be aware of as well.

[00:03:47] Ok, great. Next question. Who defines the calibration intervals for instruments or devices?

[00:03:55] The owner of the equipment is responsible for defining [calibration intervals], and in 17025 only the owner of the equipment can specify it. It can't be specified by the lab doing the calibration. There can be conversation and discussion to say: “Hey, this is often what people will use.” But really that decision has to come from the owner of the equipment because they know how their equipment is used and they know the risks that they might be taking. So that is why it's up to the owner of equipment to set their own calibration intervals.

[00:04:33] Ok, so this next question has two parts. Part one. What factors and data should be used to justify a calibration interval when a recommended manufacturer's frequency is not available? Part two: Is it acceptable to perform it at different intervals than recommended by the equipment manufacturer?

[00:04:55] I'm going to start with the second question. It is absolutely acceptable to have calibration intervals that differ from a manufacturer’s recommended interval. You may need to calibrate every three months where the vendor may say: “We recommended year…” yet, you're doing it every three months because the risk of the amount of product that could be recalled due to a failed calibration is just too significant.
It would just devastate your company, not only [due to] the time to investigate, but also to do all the recalls and the corrective actions and preventive actions. So, it is a balance of that risk versus the expense to do the calibration. Additionally, you may have data over time that show you this instrument is stable and could be calibrated every three years. And instead of a manufacturer's recommended [interval] of nine months. But you would want to have those data and you would still want to have some type of verification in that time just for your own protection.

Now, the other question, what factors in data should be used to justify that calibration interval, especially when that interval is not available from a manufacturer? There is one guide. There are many guides. There is one that I was easily able to find through NIST. It's called NIST GMP 11, “Good measurement practice for assignment and adjustment of calibration intervals for laboratory standards”. So, the laboratory standards would be what you would use to then calibrate other equipment. There's also a recommended practice from NCSLI: “RP-1: Establishment and Adjustment of Calibration Intervals”. There are a number of good practice guides, and that is something that you can take some training in understanding how to do that. I have gone through training and have experience in adjusting and setting and evaluating calibration intervals.

[00:07:16] We received many questions during the webinar on repeatability and rather than itemize all the different questions on that, do you want to say a few words about it?

[00:07:29] Sure. In in relation to repeatability, in regards to balances, there were a number of questions that came through and some people reiterating their understanding and correct understandings at that, of calibrating a balance following USP <41>. You are calibrating using approximately 5 % of the balances range and 100 % of the balances range, as well as points in between. You don't need to go lower than 5 %. And in that, you are also going to be checking the eccentricity, how the balance pan or grid [performs] if you're weighing at different points around the pan as compared to a center load. And then as well as linearity by addition, linearity by substitution. And then the other part is repeatability.

Repeatability is where you have measurements that you do over and over in a short period of time using the same person, the same equipment, the same reference standard, in the same environment. So you are getting that on and off, on and off, on and off measurement. And that is where, as one person pointed out, that point, measurement tool, or that mass doesn't need to be calibrated because it's just the same object going on and off.

In one NIST tutorial that I [took part] in, the tutorial was about calibrating balances and scales. They said you could use a rock if you wanted to, in order to calibrate for repeatability because it's the same thing. But you would not want to just do a go grab a stone out of the driveway and have it all filthy and get your equipment all filthy.

But that repeatability is what you are then going to calculate the standard deviation on. And that standard deviation is what also goes into the measurement uncertainty. And those together are compared to your accuracy specification for your use of the balance. That is how you determine the minimum sample weight or minimum weight. That minimum weight can very easily be lower or smaller than that 5 percent of the total range of your balance. I hope that helps clear that up!

[00:10:06] Does the label need to be placed over the adjustment point?

[00:10:22] Ok. So, in the webinar we talked about calibration labels and the information that needs to go on them. And I am going to approach this question in a couple of ways. The calibration label, if we talk about it as the label that identifies the idea of the equipment, the calibration period, it's due date and who calibrated that piece of equipment, that [label] should be easily viewable on the piece of equipment.

There is no requirement that it needs to go over the adjustment port; not all equipment has adjustment ports. For example, with a liquid-in-glass thermometer, there is no way to adjust [that type of] thermometer. So, you could put that label on in a way that you can still read the thermometer in the range that you need it.

But let us look at something that does have an adjustment port, like a pressure gauge. Then you would want something like a tamper evidence seal that you would place either over the adjustment port, or over the halves that screw together. This is so that if anything were breached on that pressure gauge, it would be easily visible, because those tamper evidence seals are very delicate. But they are also hard to clean off, and that is what you want. You want to know if the equipment may have been breached, you want to know if there has been unintentional or intentional adjustments.

[00:12:05] So another question regarding calibration stickers: does a calibration sticker need to include the range of measurement that was calibrated?

[00:12:13] It is not required, depending on the standard or regulation that you're that you're using. It is handy. It is useful information, especially in the cases of limited calibrations. A limited calibration would be like that liquid-in-glass thermometer. The thermometer has a total measurement range from zero to 100 Celsius. But let us say it is only calibrated from zero to 50 Celsius. In that case, it would be useful to have that calibration sticker that it's a limited range, that it's calibrated from zero to 50°Celsius, and then you know it's not to be used beyond that.

Other things that are often useful to have on a calibration label could be the accuracy. So using that same liquid-in-glass thermometer, the resolution of that thermometer maybe a half a degree Celsius, or 1 degree Celsius, but maybe your accuracy is to 5°Celsius. So by having that accuracy on there, the people that use it have a better understanding of what they can claim, and can also compare it if they’re trying to use it to see if it’s the appropriate thermometer for that test.

[00:13:43] Next question: “What about stateside labs who are not ISO 17025 accredited but claim to be covered under their parent company overseas?”

[00:14:03] In that case, whether it's in the United States or Europe or Asia or anywhere else, if a parent company is  in a different country, any lab under their accreditation would need its own scope of accreditation. There are some examples of this in labs that have the same name but are located in different states or even different countries. They are accredited under the same accreditation body. They might have the same accreditation number with a sub designation like XYZ123.1, XYZ 123.2, and XYZ 123.3 and so on. So, each of those physical locations would have their own sub-identifier: .1, .2, .3; or even their own individual scope of accreditation. That is for ISO 17025- each physical location has to have its own scope of accreditation.

[00:15:11] Next question: What you said about the increasing levels of uncertainty reminds me of doubling and redoubling video with each generation, the quality decreases like the number of instruments in between your instrument and an S.I. standard, results in less certainty. Is this a correct analogy?

[00:15:34] The analogy is correct in that the measurement uncertainty will increase and but hopefully the quality of each measurement is going to still be the same; it's still a high-quality measurement. But as you move further and further away in those successive comparisons and in that unbroken chain of calibration, the measurement uncertainty will increase in size. [The uncertainty] can never get smaller as it goes further down that chain.

That is really important to know. As an ISO 17025 assessor, that's one of the things we have to check is, is anybody claiming a measurement uncertainty on their scope that is smaller than something at the national or international or primary standards level. That is something we look for.

[00:16:31] Next question: For temperature data loggers, my calibration contractor cannot calibrate to my desired temperature range, but they tell me the device is capable of measuring to my temperature range. Is this acceptable?

[00:16:46] What the calibration vendor may be saying is: “Yes, your instrument can measure beyond what we can calibrate to.” That is often referred to as the operating range, and that lab may only be accredited to, or capable of calibrating up to a certain temperature. Maybe they can only calibrate up to 200°C. But you need to use your instrument at 300°C. You cannot extrapolate and say because it is good at 200 °C, it's going to be good at 300°C for the temperature. You absolutely need to have calibration data that show you it is good at that 300 degrees.

Let us say it was 275 degrees. It is not necessary to have calibration points at each of the temperatures that you measure. But the range should encompass that temperature range. I hope that that that makes sense. Before we go on to the next question, it may be that they need to find a different calibration company that's accredited for the ranges that they need, if they are not able to get that from that vendor.

[00:18:21] Next question: Regarding dimension and weight calibration, can zero point be used as a calibration point?

[00:18:31] So I am going to assume that when they said weight, that they meant balance. When you are calibrating a balance, you want to check to see with no load on the balance, the reading is zero. It is the same with dimensional equipment such as a micrometer, or a caliper, or something that you can move to check over that range; you want a zero point. So that when you completely close it down, it goes back to zero. Same with the balance; when you take the weight off, does it go back to zero?

There are other pieces of dimensional equipment, like a gauge block, would be considered like a weight for a calibration weight. It is what it is, and it is being compared to find out what that is. And in those cases, there is not a zero point because it's always going to be…2 centimeters or whatever the size is for that gauge.

[00:19:37] Next question: If calibration is a comparison against a reference, then what is the term when one has to tweak an instrument to bring it within an acceptable range?

[00:19:50] The term would be adjustment. But not everything can be adjusted. So the other thing we want to take into account is— [sometimes we cannot] adjust a piece of equipment. We will go back to that liquid-in-glass thermometer. You cannot adjust a liquid-in-glass thermometer and still have a usable liquid and blast the thermometer afterwards. But you could apply a correction factor. So if you are applying a correction factor of, let's say of 3°C , it has to be clear what that correction factor means. Are you adding 3°C to the display to get your temperature? Or are you subtracting 3°C? How you write the correction factor (and preferably write the correction factor on the label),  if it's something that is necessary for the final measurement, everybody must understand what that correction factor is, and how to apply it.

[00:20:55] Next question: If a calibrated balance is moved from a lab bench to inside a biosafety cabinet for weighing a biohazardous material, does a full calibration need to be made or is a verification enough?

[00:21:14] It depends on the balance. There are many balances that are robust and stable enough to be moved on a countertop, or even move from a countertop into a bio-safety cabinet. And in that case, that has to be proven and documented. And it is also determined by what sort of risk mitigation you need to practice. [For example], a single-point verification at full scale of the balance before you move it. And then after you move it and set it up in the bio-safety cabinet, making sure it is level and that it's accurate when you put that weight on. For other balances, such as an analytical balance, they're so sensitive that sometimes just a minor bump to the balance can throw that balance out of tolerance.

So those [analytical balances] are really sensitive instruments that (for a good reason) you do not want to move until they have been calibrated or verified. [This will] mitigate the risks that you have in following procedures that you have determined to say: “this is how much we need to do before we remove a balance.” [Once a balance is] set up in its new location, then you follow your procedure again, to calibrate it and adjust it to the accuracy that you need.

[00:22:48] Next question: Are original equipment manufacturers (OEM) required to notify their customers if their equipment has a calibration out-of-tolerance on a standard used for customer equipment calibration?

[00:23:03] The OEM, or any company, would have to follow the contractual obligations as far as communicating in those instances. So, that is something that it is worthwhile to check when setting up under contract review. If a company is accredited to ISO 17025, then yes, they would be required to notify you if their calibration standard was found out-of-tolerance, and they had used it to calibrate your equipment in the range or parameter that was out of tolerance. If they are not accredited, it would still be good business for them to notify you so you could do your impact analysis of the impact on your equipment.

[00:23:51] One last question. What if you use internal staff versus an outside certified calibration service for performing calibration? What should you have in place?

[00:24:06] Using inside staff can be a great money saver on the cost of calibration. But understand the costs of setting that up internally. There might be the upfront costs in the beginning, but then your operating costs may be lower overall. To set up an internal calibration program, you would need people who are competent to do the calibration. You would need the equipment to be able to do the calibration. You'd need an environment to do the calibration. And you'd need procedures that are validated for those calibrations. You would need measurement uncertainty; a procedure to calculate the measurement uncertainty, as well as the actual uncertainty budgets that are calculated. You would need to look at the intervals—how frequently does calibration need to be performed? Control charting is a great way to track that.

You would need that supporting quality system of internal audits. So, ISO 17025 is a great standard to follow, even if you are not going to be accredited to it. You can use it to be compliant to it. But think of the basics you need the people, you need the place, you need the things. And those things are the equipment, the procedures and something to monitor those environmental conditions to.

[00:25:41] Heather, thank you so much for doing our webinar with us and taking time to answer the questions that we didn't get time to answer during the webinar. You can watch the webinar recording with Heather Wade and Paul Daniel at any time.  

[00:26:26] Thank you. I really appreciate the opportunity. And we will still be reaching out to people that sent in questions that we maybe haven't answered yet or even the ones that we have in case you don't see this Q&A. So again, thank you. Come see me at www.heatherwadegroup.com/ and we look forward to helping you.