List Of Written Questions From The Participants

(Submitted Anonymously)

USP PQM Training Workshop on CAPA Management for Regulator and Pharmaceutical Industry,
Jakarta 31 January-2 February, 2018

Answers from David Buckley (Feb 9, 2018)

NON-BINDING ON USP, USP PQM OR USAID

1. Some personnel really fear they will be blamed for a problem that happened. SO, they keep it
secret until the production process goes to next process, and then the next process finds it.
How to motivate personnel to tell immediately when the problem happened?

ANSWER - This needs an organizational culture change driven by Top Management.

"Drive Fear Out" is Point 8 of Dr Deming's famous 14 Points for Management. “Drive out fear, so
that everyone may work effectively for the company” - W. Edwards Deming1

Fears destroy motivation. Here are some management techniques that will work -

• Don't reprimand people if they make mistakes but teach them to learn from the mistakes
and share the mistakes with other employees to prevent the mistakes from reoccurring.
• Encourage employee's feedback from all levels. This allows management to recognize
and capture problems. The Japanese system of Kaizen (continuous improvements) and
Poka Yoke (stopping unintended mistakes) work well and are commended.
• Not all problems can be rectified immediately. Make this known. Use a Pareto chart to
rank the problems and always focus on solving the top 20%. According to the Pareto
principle the top 20% issues cause 80% of the problems.
• Involve the employees in solving problems. They are the experts at their station. Use their
knowledge to implement sound change. When you involve employees, this naturally
reduces fear. Management needs to encourage and allow time for this.
• Dr Deming’s 14 points
1. Create constancy of purpose for improvement of product and service.
2. Adopt the new philosophy
3. Cease dependence on mass inspection
4. End the practice of awarding business on price tag alone
5. Improve constantly and forever the system of production and service.

1 Deming, W. Edwards (1986). Out of the Crisis. MIT Press.

 6. Institute training
7. Institute leadership
8. Drive out fear
9. Break down barriers between staff areas
10. Eliminate slogans, exhortations, and targets for the workforce.
11. Eliminate numerical quotas.
12. Remove barriers to pride of workmanship.
13. Institute a vigorous program of education and retraining.
14. Take action to accomplish the transformation.

2. How can we make an effective CAPA system with people or personnel who cannot commit
with the CAPA?

ANSWER - I think it is impossible. It is Top Management who should be implementing a culture

change. See above. Furthermore

“1.5 Senior management has the ultimate responsibility to ensure an effective Pharmaceutical
Quality System is in place, adequately resourced and that roles, responsibilities, and authorities
are defined, communicated and implemented throughout the organisation. Senior management’s
leadership and active participation in the Pharmaceutical Quality System is essential. This
leadership should ensure the support and commitment of staff at all levels and sites within the
organisation to the Pharmaceutical Quality System”2

3. Can you give suggestion that all parts of the company commit in the implementation of
CAPA, do not assume that CAPA only QA jobs?

ANSWER - It must be Top Management who should be implementing a culture change. See
above. Furthermore

“1.5 Senior management has the ultimate responsibility to ensure an effective Pharmaceutical
Quality System is in place, adequately resourced and that roles, responsibilities, and authorities
are defined, communicated and implemented throughout the organization. Senior management’s
leadership and active participation in the Pharmaceutical Quality System is essential. This
leadership should ensure the support and commitment of staff at all levels and sites within the
organization to the Pharmaceutical Quality System”3

2 PIC/S PE 009-13 GMP Guide (Part I Basic Requirements for Medicinal Products)
3 PIC/S PE 009-13 GMP Guide (Part I Basic Requirements for Medicinal Products)

Page 2 of 20
4. How to manage so many CAPA from internal and external?

ANSWER - You need to have management commitment for resources. There should be a
CAPA team within the QA section – for large companies this is sometimes one or more full time
positions. See 2. Above.

Also - The correct manufacture of medicinal products relies upon people. For this reason there
must be sufficient qualified personnel to carry out all the tasks which are the responsibility of the
manufacturer. Individual responsibilities should be clearly understood by the individuals and
recorded. All personnel should be aware of the principles of Good Manufacturing Practice that
affect them and receive initial and continuing training, including hygiene instructions, relevant to
their needs. 4

5. How can we arrange the time management to delegate task if we did not have any personal
that capable to do the task?

ANSWER - It is a basic GMP requirement – for example PIC/S GMP Guide PE-009 13

“The attainment of this quality objective is the responsibility of senior management and requires
the participation and commitment by staff in many different departments and at all levels within
the company,…”

6. How to define the root cause if there is no good documentation/record (lack of evidence?)

ANSWER - documentation is the heart of GMP. Remember to have a good written problem
statement –
4W, 2H, 1C:
• What was affected
• Where did the problem take place
• When was the problem discovered
• Who Discovered the problem
2H:
• How much was affected
• How often has the problem occurred

1C: What is the consequence

5 W, 2 H, 1 C .

4 PIC/S PE 009-13 GMP Guide (Part I Basic Requirements for Medicinal Products)

Page 3 of 20
 5W 2H 1C Response

What is the problem? Describe it in a The problem is…

single sentence, so that others will be able
to understand what you mean.

Why is it a problem? What is the pain? This is a problem because…

Where do we encounter the problem? We encounter the problem at (Location)

5W
(Time) when (Specific circumstance)…

Who is impacted? This impacts: (Staff) by…, (Patients)

by…, (Other providers) by … (others)
by…

When did we first encounter the problem? We first encountered this problem…

How did we know there was a problem? The symptoms of this problem are…

2H How often do we encounter this problem? We encounter this problem (x) times and
each encounter is (this big). The problem
is getting (better/worse).

1C What is the consequence. Do not spend too much time on low risk
issues.

Use 5 Why and ask people.

• Use short and simple phrases

• Be as precise as possible, avoid general expressions
• Try to quantify
• Do not stop if you can ask “why” another time. Ask “Why” until the root cause is
uncovered But Stop when the Answer is outside your organization’s control eg “The
Global Economy”
• A root cause has been found if you can put an action on it that will eliminate the
abnormality
• The analysis should be supported by real facts and figures
• Do small logical steps

Remember the general guidelines for 5 Why Analysis.

• Don’t jump to conclusions or assume the Answer is obvious.

• Be objective.

Page 4 of 20
 • Focus on systems and processes, not people
• Path should make sense when read in reverse using “therefore”

How to conduct the 5 Why Analysis.

• Write down the problem statement. ALWAYS.

• Ask ”Why” the problem happened and write down the Answer .
• Repeat until the Root Cause has been identified.

There is also the 5G principal

• Go to the actual place, Go to the spot (GEMBA5)

• Get the Actual Facts – Check facts and figures (GENJITSU)
• Look at Actual Things – Examine the objects (GEMBUTSU)
• Principles – Refer to the proper tools/methods (GENRI)
• Standards & parameters – Follow the standard (GENSOKU)

7. When we operate a compression tablet machine sometimes the weight of tablets tends to go
up or down. IS it consider a deviation that needs CAPA handling? Because we can correct it
by adjusting the machine?

ANSWER - In Process machine adjustments within a predetermined range or specification and

according to batch manufacturing master formula are not deviations so are not CAPA.

In some situations machines monitor their own performance and are making adjustments all the
time within a preset limit. Another example is a purified water system. If a conductivity meter
detects an out of limit result it can activate a by-pass valve before it goes for manufacturing (eg
from the RO to the holding tank). However, if the conductivity is OOL on the return line and the
water could have been used for manufacturing that becomes a deviation subject to CAPA.

8. How can we classify or assess the correction action is a weak corrective action?

ANSWER - If it recurs then the CA was weak or did not address the root cause.

5Gemba Kaizen is a Japanese concept of continuous improvement designed for enhancing processes and reducing
waste. Within a lean context, Gemba simply refers to the location where value is created, while Kaizen relates to
improvements.

Page 5 of 20
9. If there is a CAPA related with registration number, and the manufacturer have not finished
their CAPA, can they get the registration number?

ANSWER - Most highly regulated countries’ national regulators are amenable to a CAPA plan
and are likely to offer qualification based on risk analysis and meeting performance targets. For
example, continuing manufacture with a (well maintained and managed) legacy facility while a
new cGMP unit is being built. You need a risk plan, commitment tracking and have agreed
landmarks with regular updates (by meeting if necessary) to ensure risk is managed

10. Please could you explain what is CAPA lifecycle?

11. In CAPA management, we often delay to meet/fulfill the action plan and the result not meet
the requirement. How many times we can allow to extent the due date to fulfill the CAPA?
(based on your experience)

ANSWER - See above – there is no prescribed limit but your regulator should be involved if the
time line is constantly being extended.

6 http://www.pharmtech.com/avoiding-investigational-failurediscrepancy-related-form-483-observations

Page 6 of 20
12. How long that is suitable to review the effectiveness of the CAPA?

ANSWER - depends on risk analysis. Usually try for soonest follow up after the CAPA has been
closed.

13. We know that identification of problems is a big problem. How we can make same
perspective between inspectors and pharmacy industry to identify the problem or finding
because often the root cause and CAPA did not answer the finding?

ANSWER - work with the regulator – make it as transparent as possible. In some cases, eg
WHO PQ of vaccines, many companies have regular meetings with the WHO team in Geneva in
respect of eg long term building plans to replace legacy facilities.
In some situations there will be differences in interpretation of the GMP regulations from
Inspector to Inspector, for Expert to Expert.

For example, one country’s regulatory authority requires low level air return in the Controlled but
Not Classified (CNC) corridors. There is no GMP requirement for this. In this case if you want to
supply that country’s market it is not CAPA but change control to meet a customer requirement.
Then the organization’s executives need to decide on cost benefit ratio.

14. Could you give me example when should risk assessment use in CAPA? In what condition
the preventive action not needed?

ANSWER - CAPA metrics tend to focus on closure rates, cycle time, number of open CAPAs,
etc. Preventive Actions are, in most cases, longer-term solutions across processes, systems,
product lines, and Quality Systems and will take more time to close.

Use Risk to filter events: Risk Assessment is a great way to do this. Risk matrices will help
make the determination as to the criticality of an event. Very low risk is Correction
• The forklift battery was not charged overnight – Correction – charge the battery. no
product quality impact – low risk. (But how about business risk?)
• The low risk door was open, close the door, record it in the log book.

The higher the risk, the more likely we would like to take Corrective Action, and higher still
Corrective Action and Preventative Action.

ISO has replaced preventive action (ISO uses “preventive” while PIC/S and WHO uses
“preventative” – they mean the same thing!) with the term ‘risk based thinking’, which means to
control and manage your risks to prevent issues from occurring, which is normal risk
management. This of course is what preventive action always was – it’s just that no one
understood it. In fact one could easily argue that the main purpose of a management system is to
prevent things from going wrong, so essentially it is all just a system of preventive action.

Page 7 of 20
 Here are some common preventive actions:

• Installing “alarms” into your process. This might be monitoring of statistics or process
control charts.
• Bar code all instruments and scan them before the test to electronically record instrument
number, and calibration status.
• Biometric security to prevent unauthorized access.
• Audits or reviews of suppliers.
• Any preventative maintenance or calibration programs that ensure your equipment
functions uniformly.
• Regular review of processes or controls to spot any potential issues that could drift into
the area of a non-conformance.
• Have planning in place for disaster recovery (eg server or computer crash, or any
unexpected situations so that your company is prepared.

If you have good preventative action (Kaizen, Poka Yoke), then you may never see Corrective
Action!!!

15. Does the industry have to record all the CAPA?

ANSWER - yes

16. How the best practice to monitoring CAPA since there were so many deviation in industry?
And Are there any tools to evaluate CAPA effectiveness? For example CAPA for solving the
customer complaint related to procedure, which was produced last year. Can we decide that
the CAPA is effective by doing challenge test? Or we should check the result of the CAPA by
period of time? And What exactly the way to evaluate CAPA effectiveness? How to assess
CAPA effectiveness? And how to perform a quantitative CAPA effectiveness?

ANSWER - There are commercial software solutions that make recording and tracking
deviations easier. I increasingly see the Japanese systems of Kaizen and Oka Yoke (and others)
being successfully implemented in the pharmaceutical industry

They are often marketed to fulfill ISO900x but are good for cGMP as well

17. How do we know if the problem that we choose is the root cause? How to prevent choosing
the wrong root cause?

ANSWER - There are no wrong root causes. There can be many root causes that impact the
situation. Some have to be all present, or only one. The example in the work shop was your
friend has hurt their arm. Ask Why?

Page 8 of 20
 How? Slipped and fell Immediate cause (Training is ineffective!)
Correction – mop up the water –
Why did you slip? Water on the floor Underlying cause
but will recur
Corrective Action – fix the tap –
Why the water? Tap was leaking Proximate cause
better
Preventative action – add to
Why leaking tap? No Maintenance Root cause
maintenance program – best
Why no Management
No budget Risk analysis
maintenance? decision
Why no budget? Global economy Stop when it is not under your control

Another example is a fire: it has to have Fuel AND Ignition Source AND Oxygen. There are 3 root
causes and removing just one means no fire.

Some situations are different of course and you have “Or” situations. Eg sterility failure.

Non-sterile product
Fire in market

AND gate
Or gate

Ignition Bad
Fuel (root Oxygen Bad Filter (Many)
source (root Autoclave
cause) (root cause)
cause)

Remove any one, Must correct them all

eliminates the fire

Truth Table:
Fire Fuel O2 Ignition Fire Non-sterile Bad Bad Bad Non-
Root source product - autoclave filter sterility sterile
Cause Root Cause test product
0 0 0 0 0 0 0 0
0 0 1 0 0 0 1 YES
0 1 0 0 0 1 0 YES
0 1 1 0 0 1 1 YES

Page 9 of 20
 1 0 0 0 1 0 0 YES
1 0 1 0 1 0 1 YES
1 1 0 0 1 1 0 YES
1 1 1 YES 1 1 1 YES

For the lack of sterility illustrated above you must treat ALL root causes. These are the basics of Fault Tree
Analysis, one of the important risk analysis tools.

18. To only retrain people as a CAPA is weak. How far is it acceptable? At what kind of case?

ANSWER - If people are making mistakes it is weak CAPA to just retrain them. There must be
an underlying cause.

We try to find errors in system and procedures, sometimes products. The Japanese industry
uses a system of Kaizen and Poka Yoke. We use Ishikawa to find weakness in Machine,
Method, Material etc. The final one should be Personnel. Why are they making errors. Is it the
workplace or documentation etc? Change and correct, then last – retrain staff in the new
procedures

19. If the stability data are not linear, what aspect that should be investigated first? How to
determine if the CAPA of that linear regression data is to perform reformulation?

ANSWER - I’m not sure! But most stability data I see are not linear, but have some variance. If
variance is too high it will impact the shelf life since the ICH rules are that the lower 95%
confidence interval must be used. This will result in a shorter expiry date than if the regression
line is used.

Consider the example below - the lower 95% CI for self life (above 95% label claim at expiry) is
24 months with the 95%LCI and not 36 months if the regression line is used. (In Mintab –
stat>regression>fitted line plot, and enable 95% CI in options menu)

Page 10 of 20
 If the shelf life is too short then consider reformulation, or packaging etc. May need always to be
re-registered.

20. I want to ask about balance stability at weighing room in laminar station. According to WHO
TRS for HVAC system, for weighing the balance must have perforated table to assure
laminarity at weighing process. It’s very difficult to have good stability if balance doesn’t
have fix table and it is impacted by the laminar air flow? What is your suggestion?

ANSWER - As much as possible comply with the WHO but eg for some weighings the balance
must be protected from air flows from the RLAF. Conduct air flow visualization to assess risk and
also use the video for operator training.

21. How do we determine how many batches/ how long times needed as an evaluation of CAPA
effectiveness check? Example : if an OOS of water system monitoring occurs, will 3 times
periodic sampling’s result after the CAPA implemented be sufficient to close out the CAPA
and state them as an effective one? Or do we have to analyses the whole year of trend data
after its implemented
ANSWER - there’s no one good Answer to this, sorry. If the OOS is in the final PW used for
product you may have to go back to Phase 1, then 2 and 3 before closing the CAPA. If it is
upstream water (eg in raw water or pretreatment eg a/carbon filter) then sampling at increased
frequency (say 1 sample per week for 3 weeks) is probably realistic to check CAPA
effectiveness.

Page 11 of 20
22. How about if it’s an environmental monitoring cases. Can we just evaluate the rest 3 EM
result to conclude that the CAPA is effective?

As above

23. Please explain about OOF, what data is related to OOF?

ANSWER - OOF is Out of Frequency. Eg if a calibration event has to occur by Oct 2017 and it
has not, then it is a deviation, OOF. Same for revalidation if it is late. Root Cause could show
lack of human resources or equipment etc. And if the CAPA is not closed by the due date.

24. With the high/strict regulation of GMP, do highly regulated countries like USA or Australia
ever undergo drug shortage? If so, how do they handle it?

ANSWER - yes – it can be a big problem. In the USA the FDA has a regular newsletter that lists
current drug shortages. The FDA works with industry that underpin these shortages and it
sometimes means accepting a long term CAPA plan eg for investment in new technology or
buildings.

See https://www.fda.gov/AboutFDA/ContactFDA/StayInformed/GetEmailUpdates/default.htm

At times the regulator has to take a pragmatic view of health risk (ie no treatment) versus
benefits.

Some Statistical Process and Quality Control (SPC and SQC) questions

1. Microbial limit test OOS. After we conduct the retest of the same sample with the second
analyst (Duplo test method) and found that the result one : 50 cfu/ml and 110 cfu/ml
(specification is max 100 cfu/ml) so how we can make decision? Note : before we already
performed investigation for hygiene of analyst, tools, machine and environment but
overall is clear and not show the non conformities.

ANSWER - retesting samples for microbiological tests is never reliable. The counts can go up
and down as bacteria grow or die in the sample, even if refrigerated. If you are at or near the
regulatory limit the problem is compounded. Set alert, warning and action limits below the
regulatory limit so the product is not impacted by eg adulteration.

Page 12 of 20
 7

Where:
• UAL = Upper Alert Limit
• LAL = Lower Alert Limit
• μ = Mean of distribution
• s = Standard distribution (SD) of distribution (note Poisson distribution so normalize
before calculating sd)
• n = 100 (CFU/mL) = maximum count allowed
• p(bar) = Averaged proportion = averaged count per maximum count allowed
• q(bar) = (1 − pbar )

The value of n = 100 is based on the action limit (100 CFU/mL), which is the maximum number
of colony forming unit (CFU) allowed per mL.

There should be a separate OOS procedure for microbiological results. In some cases, eg
sterility testing, it is prescribed in the harmonised Pharmacopeial methods.

2. If the Cp and Cpk result in the PQR is less than 1.3 but it’s still more than 1.0, do we need
more improvement for our process?

ANSWER - Yes, you should be trying to improve the process. Remember Cp is a measure of
the spread of the results, while Cpk gives us information on centering on the target. CpK is
always =/< Cp. You can improve the process so that it is centered on the target (then Cpk = Cp),
and reduce variance by better control, so Cp will go up. If you reduce variability you increase
quality.

Cp and Cpk above 1.33 is generally acceptable but above or equal to 2 is better.

3. Should we perform Cp & Cpk for 3 validation batches in validation report?

ANSWER - you should determine Cp and Cpk for variables measured during the processing of
each batch. You can then state the process for making that batch or the 2nd batch was capable.
Eg tablet weight, fill volume, potency, individual table assay.

7Cholayudth, Pramote. "Establishing alert limits for microbial counts in Purified Water." Journal of Validation
Technology, Nov. 2006, p. 44+ Also search on other articles by Pramote Cholayudth.

Page 13 of 20
For example here are the mixing studies for 3 batches of a tablet granulation

B1 5 minutes B2 5 minutes B3 5 minutes

504.7 497.8 498.8

509.4 500.6 501.6

508.2 499.7 500.1

505.3 499.6 506.3

504.9 499.2 503.6

504.1 500.6 500.7

505.9 501.6 502.6

504.1 501.4 503.4

506.8 500.8 500.7

508.8 500.2 501.9

502.7 502.1 500.9

504.0 501.7 501.6

505.9 502.2 504.5

It is a mistake to use (for example) the average potency of each batch with only 3 batches and
attempt statistical analysis such as the Six Pack. If using Minitab use the “Subgroups across
rows”.

Page 14 of 20
 Process Capability Sixpack of B1 5 minutes, ..., B3 5 minutes
Xbar Chart Capability Histogram
510
UCL=508.37
Sample Mean

505 _
_
X=502.79
500

LCL=497.22
1 2 3 4 5 6 7 8 9 10 11 12 13 498 501 504 507 510

R Chart Normal Prob Plot

UCL=14.03 AD: 0.538, P: 0.157
Sample Range

10
_
5 R=5.45

0 LCL=0
1 2 3 4 5 6 7 8 9 10 11 12 13 495 500 505 510

Last 13 Subgroups Capability Plot

510 Within
Within Overall
StDev 3.22001 StDev 2.83931
Values

505 Cp 1.55 Pp 1.76

Overall
Cpk 1.26 Ppk 1.43
500 CCpk 1.55 Cpm *
Specs
0.0 2.5 5.0 7.5 10.0 12.5
Sample

For this study the Anderson Darling (AD) P>0.05 so the data are normally distributed and Cp,
Cpk, Pp and Ppk are all acceptable so the mixing study is stable and in control.

For comparison of 3 batches also use ANOVA and a box plot. For example here are the mixing
studies for 3 batches of a tablet granulation

Page 15 of 20
 Set null hypothesis H0: μ1 = μ2 = μ3, then from Excel

SUMMARY

Groups Count Sum Average Variance

B1 5 minutes 13 6574.8 505.7538 4.134359

B2 5 minutes 13 6507.5 500.5769 1.621923

B3 5 minutes 13 6526.7 502.0538 4.029359

ANOVA
Source of P-
Variation SS df MS F value F crit

Between 4.05E-
Groups 184.9112821 2 92.45564 28.3 08 3.3

Within Groups 117.4276923 36 3.26188

Total 302.3389744 38

Conclusion: if F > F crit, we reject the null hypothesis. This is the case, 28.3 > 3.3. Therefore, we reject
the null hypothesis. The means of the three populations are not all equal. At least one of the means is
different. However, the ANOVA does not tell you where the difference lies. You need a t-Test to test
each pair of means.

Page 16 of 20
 Boxplot of B1 5 minutes, B2 5 minutes, B3 5 minutes

510
Data

500

490

B1 5 minutes B2 5 minutes B3 5 minutes

However, each batch is uniformly mixed (from Six Pack above) so the variation may be due to
(acceptable) batch to batch variation, since the regulatory limit in the above case is 485 – 515. You can
conclude that the batches each met the Pharmacopeial limit.

4. Should we perform Cp and Cpk for environment monitoring trending?

You can for EM such as temperature where you have a target and range (eg Temperature) but
not really for limit tests, eg cfu/mL in water.

I’ve never seen it for EM, just for manufacturing processes.

5. I want to ask about the stability program for existing product (on going stability program).
Is it necessary to calculate the Cp and Cpk?

ANSWER - I’ve never seen it. It’s also not really a process you are controlling such as filling
vials

Page 17 of 20
6. When we got some non-normal data (eg Anderson Darling p<0.05) then we transform it
(with Box-Cox analysis) but still not normal, then what can we do with this data? Example:
to know the capability performance of HVAC, and the temperature and delta pressure are
the parameters.

ANSWER - I’ve never seen Cp and Cpk used for HVAC and differential air pressure. If AD
shows the data set is not normal it means it is skewed. There are many, many reasons for skew.
The histogram will show the skew. Eg it could be binomial, outliers, exponential, Poisson, etc.

One drawback: it’s not easy for novice users to visualize the shape of the distribution

The capability calculations are HIGHLY sensitive to skew. Check with a statistician if a regulator
is asking for this but I don’t see the rational. There are other transformations depending on the
skew, eg log normal, Weibull, Johnson, etc., etc. Ask a statistician!

Examples from Jim Colton, Minitab Inc. December 8th 2014. Tampa ASQ Meeting

• Anderson –Darling (AD) Detects Non-Normality well when there is a mixture of

distributions or the distribution is skewed.

• Ryan –Joiner (RJ) Detects Non-Normality well when there are outliers or the distribution
is skewed. Better at detecting Non-Normality than the AD when the measurement
resolution is poor.

Page 18 of 20
7. If there is a shifting in our trend and we have done the investigation from analytical
process and production process, but we can’t find the root cause of that shifting, is it
okay for me to update the action alert level using the updated trend?

ANSWER - EWMA and CUSUM are time based trending tools. They will detect shifts as low as
½ standard deviation. Check Cp and Cpk to see if the root cause is off target or non-centered
process. If it is centered Cp should be = Cpk. As long as Cpk is above 1 and Cp >1.33 it is
generally held to be acceptable. Don’t let Cpk go below 1, then it is trouble!

8. What we suppose to do to involve audit trail as our system if our machine/utility looks old.
There is no HMI or port connector on it? Not of all but its approximately between 10-20
machines. Yesterday and this moment on we combine print out data and logbook. So
what suppose to start to improve all particular element (machine) from logbook become
print out data? Is it mandatory to use print out data?

ANSWER - The new data integrity regulations to stop data fraud were introduced last year. Do a
HACCP and risk analysis to search for Critical Control Points, and Critical Control Parameters.

Where you identify these then take Change Control (not really a deviation as such) to upgrade
the systems. In some situations regulators will allow “4 eyes” eg 2 people to witness.

Also consider photographs with date and time stamp and unalterable meta data. (Some smart
phones imbed date and time stamp as well as phone serial number into the meta data. Always
best to discuss with your regulator an action plan.

9. Sometimes when I do capability analysis (Cp and Cpk) the value for UCL and LCL is
greater than USL and LSL, although the Cp and Cpk value satisfactory. Is it okay? And If
Cp and Cpk > 1.3, but data distribution tend to approach the USL/LSL. How can we
improve?

ANSWER - Best check with a statistician! Sorry.

From WikipediAnswer - In process improvement efforts, the process capability index or process
capability ratio is a statistical measure of process capability: the ability of a process to produce
output within specification limits.

The concept of process capability only holds meaning for processes that are in a state of
statistical control.

Page 19 of 20
Process capability indices measure how much "natural variation" a process experiences relative
to its specification limits and allows different processes to be compared with respect to how well
an organization controls them.

If the upper and lower specification limits of the process are USL and LSL, the target process
mean is T, the estimated mean of the process is µ and the estimated variability of the process
(expressed as a standard deviation) is σ then commonly accepted process capability indices
include:

Page 20 of 20