Article Text

Download PDFPDF
Process performance of a new liquid medication dispensing robot
  1. Fabrice Jean Lagrange1,
  2. Jean Valdomar Lagrange2
  1. 1Department of Pharmacy, Lab testing Unit (BSPharm, PhD,Technology Implementation Engineer), Pierre-Lôo Hospital, GHT58, La Charite-sur-Loire, France
  2. 2Epitech and CSULB (IT Engineer), Le Kremlin-Bicetre Paris and Long Beach, CA, France and USA
  1. Correspondence to Dr Fabrice Jean Lagrange, Department of Pharmacy, Pierre Lôo Hospital, 58405 La Charite-sur-Loire, France; fabrice.lagrange{at}icloud.com

Abstract

Objectives Liquid medications provide an alternative to splitting pills and dosages by measuring an amount of liquid rather than crushing tablets or opening capsules. Special attention should also be paid to specific patient groups with swallowing difficulties or requiring enteral feeding administration. Liquid medicines are also often used to ease withdrawal symptoms for people suffering from addiction. Nevertheless, filling liquid medication cups with the right medication and precise doses may be difficult for healthcare professionals. The Nooddis (‘Nominative Oral Dose Dispenser’—Pierre Lôo Hospital, France and Packinov, France) is a new robotic system for the automated filling of single dose liquid medications. Since the performance of such a complex piece of equipment depends on compliance of the service provider to our building guidelines, the process performance verification is a necessary prerequisite before starting routine production.

Methods The performance of the Nooddis robot (accuracy, precision, and tapering calculation) and its ability to fill medicine cups was evaluated with 18 different liquid medications using an automatic in-line checkweigher. Microbiological testing was also performed.

Results 648 sealed cups were prepared for qualification. The filling accuracy was within the limit of ±10% from 75 µL to 21.25 mL. The repeatability (% relative SD (%RSD) 0.05 to 4.93) and intermediate precision (%RSD 0.01 to 6.59) were validated for all preparations. All medicine cups met the requirements of USP and European Pharmacopoeia acceptance criteria for microbiological quality. Automated tapering calculations allowed for easy production of daily doses for the tapering periods chosen.

Conclusion Since the system met the required quality standards, the Nooddis robot, with automatic in-line tapering system, is regarded as an accurate technology that can fill the exact amount of liquid oral medication in single dose cups. This may promote closer monitoring, which supports medication tapering as well as medication misuse prevention. With a packaging cost similar to current unit dose cup systems, it is a relevant alternative to fractioned or crushed tablets, as well as opened capsules. Further developments for some sterile liquid medications are yet to be studied.

  • validation studies
  • liquid medication
  • tapering
  • automation, packaging

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information. Not applicable.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

Liquid medications allow accurate dosing and titration, and they are also easier to swallow than split or whole pills are. Special attention should also be paid to individual patient needs, like those with enteral feeding administration, mental health disorders, or ones with gradually decreasing doses prescribed during withdrawal protocols for benzodiazepine dependence. Interestingly, there has been a rising concern regarding addictions to some medications (benzodiazepines, analgesics, antidepressants) due to the increase in their prescription, as well as diversion amid the opioid epidemic and the COVID-19 pandemic.1 Tapering medication makes a safe withdrawal possible by utilising liquid medication to make much smaller dosage reductions than can be achieved with standard registered dosages.2 However, handling liquid doses, splitting pills, crushing tablets, or opening capsules remain a manual process inefficiency in hospitals that has yet to be solved with automation. Consequently, healthcare professionals often express a desire for ready-to-use single oral liquid medication doses because of the difficulty in measuring volume or potential drug misuse. Furthermore, splitting or crushing pills may expose nurses to health risks through powder aerosolisation or by giving them access to medications that could be abused.

In 1999, our hospital quality insurance system highlighted alarming practices of preparation and administration of liquid oral medication doses in healthcare settings. Thus, we proposed that the preparation of medication doses for administering to the patients be conducted within our hospital pharmacy.3 We worked towards developing a robotic system to ensure increased safety protocols, enhanced productivity (saving nurses time and resources while reducing time-consuming pharmaceutical manual processes), medication safety for both the patient and employee, and cost considerations. After previous published studies,3 in 2014 we designed and tested a new homemade model to fill microlitre volumes (online supplemental video 1). The authors wrote all the specifications required for a construction contract with a service provider, contributed to the automation, tests and building directives, and performed the pharmaceutical developments of the full scale robot to equip their hospital department. The full scale robot is now called Nooddis (figure 1, figure 2, (online supplemental video 2 and online supplemental video 3).

Supplementary video

to replace online supplemental file 2
Figure 1

Side view of the Nooddis rotary multi-pump head filling machine in our pharmacy lab testing unit (https://www.youtube.com/channel/UClUbsWMpJDshQ-oyApKANzw).

Figure 2

Inline checkweigher automatically drives each peristaltic pump head. Onfremovable amber glass laboratory bottles.

We designed this fourth generation of the robotic system (table 1 and online supplemental figure 1) to comply with industrial and pharmaceutical quality standards. Twenty peristaltic pump heads (314 rapid pump head, Watson-Marlow Fluid Technology Group Wilmington, MA, USA) combined with a qualified checkweigher (weigh module WKC204C 0.1 mg Round SS EN, Mettler Toledo, Columbus, OH, USA) provided continuous volumetric and gravimetric flow control of liquid oral medications. An automated closed-loop weight control system assures that the fill volumes remain within the required tolerances. In doing so, over-filling and under-filling can be avoided. A slow sampling rate at the end of the sampling period allowed for stabilisation of the internal balance. We optimised the filling duration to allow greater throughput from the filling machine.

Table 1

Classification of liquid unit dose packaging machines (see also online supplemental figure 1)

When compared with other systems (table 1 and online supplemental figure 1), our machine is innovative because it can fill volumes from 50 µL to 30 mL, it is connected to the hospital’s electronic prescribing software, it can comply with regulations on electronic records and signatures, and it can edit automated batch reports, quality control graphs and control charts. Furthermore, it is the only one that allows up to 20 different oral solutions loaded in the robotic system to be poured into their individual medication cups in the same run. Depending on the filled volume of the medications, the cadence ranged from 200 to 400 cups filled, sealed, and labelled per hour.

Since the performance of such a complex piece of equipment depends on compliance of the service provider to our building specification, the process performance verification is a necessary prerequisite before starting routine production. The prime objective of the study is to evaluate the performance range of the Nooddis compact automated repackaging line for liquid oral medications.

Materials and methods

Study variables

The dose represents the volume of liquid medication prescribed by the practitioners. Because of different viscosity indexes, ranging from 0.83 for zuclopenthixol (Clopixol 2%) to 8 for potassium 25 mg/mL (H2 Pharma), all of the 18 oral medication solutions prescribed in our hospital were tested. The 18 liquid medications have been tested, each with their own peristaltic pump head, and the filling rates were adapted to optimise speed and weighting. The room was air-conditioned at 73.4°F (23°C) in compliance with the non-sterile drug product repackaging guidelines.4

Accuracy

Accuracy is expressed as % of target dose expressed as mean±relative standard deviation (RSD=SD deviation*100/mean of the data), and acceptance limits were set between 90% and 110% of the target value. For each of the 18 medications, four quality control (QC) volumes were selected, ranging from minimum to maximum based on the drug doses recommended on the product labels. QC1 was the smallest drug dose recommended, QC4 was 85% of the largest drug dose dispensed, QC3 the median between QC1 and QC4, and QC2 the median between QC1 and QC3.

Six measures were conducted for each QC of the 18 medications (ie, 432 cups) between 75 µL to 21.25 mL. Extreme filling capacities were also determined.

Precision

Terms typically used to describe precision are coefficient of variation (CV) and RSD.

Intra-assay precision (repeatability) was determined for every tested volume (ie, QC1, 2, 3, 4) by six measurements observed on the same day for the 18 medications between 75 µL to 21.25 mL.

Since there are no manual processes, the pharmacist was the only operator. Intermediate precision was determined with the 4 QC for 18 medications on three different days (ie, 216 cups). The RSD of both parameters should not exceed 5%.

Tapering calculation algorithm

Once the practitioner has planned the tapering schedule together with the patient, daily doses can be automatically added to the job queue for the chosen tapering period. The operator selects the medication, enters the starting dose in mg or drops, the tapering period, the step dose reduction, and the step duration in days (figure 3 and online supplemental video 4) on our software form. Then, in one click, the automatic tapering calculation algorithm generates production items for all the doses (online supplemental material: automatic tapering calculation algorithm and online supplemental video 4).

Figure 3

View of our software form for automatic tapering calculation.

Closure integrity testing

Turning the sealed cup over and squeezing it lightly between the operator’s fingers provides proof of closure integrity of the sealing and simple leak detection if need be.

Shelf life

The effects of the repackaging process on the shelf-life after repackaging was considered.

All the packaging materials that come in direct contact with the medication are in accordance with the requirements of pharmacopoeias. Removable amber glass laboratory bottles were used to store liquid medications onto the robot (DURAN borosilicate 3.3 European pharmacopoeia type 1 neutral glass, Youtility DWK Life Sciences GmbH, Mainz, Germany). Only chemical resistant inert tubing was used in the peristaltic pumps (Bioprene peristaltic pump tubing UV opaque USP class VI, Watson Marlow Fluid Technology Group, Wilmington, MA, USA).

The cup model is made of US Food and Drug Administration (FDA) compliant polypropylene and the PS/PP (polystyrene/polypropylene) aluminium thermoscellable operculum are food-safe packaging materials. Thus, the beyond-use date (BUD) in this condition can be based on the BUD applicable to non-sterile compounded preparations in USP chapter <795>. The BUD for liquid doses are taken from a manufacturer’s large container and repackaged into smaller containers and are the earliest of the following:

  • 14 days (non-preserved aqueous)

  • the expiration date

  • the period after opening.

Microbiological stability

Microbiological instability in a preparation implies that germs have developed inside it during its shelf life.5 Thus, to be considered in the worst routine conditions, the doses of medications cups were given 10 mL of sterile water by the Nooddis robot and incubated for 5 days at 39.2°F (4°C) and a further 12 hours at 77°F (25°C) before microbiological testing. Two different oral solutions were chosen to overcome the potential intrinsic antibiotic effect of their active ingredients or excipients.6 Thus, two different medicine cups were prepared by the Nooddis (20 drops of cyamemazine 40 mg/mL (Tercian) added to 10 mL of sterile water, 200 drops of levomepromazine 40 mg/mL (Nozinan) added to 10 mL of sterile water). For each, six cups were necessary to prepare the 1 in 10 dilution in buffered sodium chloride-peptone solution pH 7.0.

Then, for each medication, 100 mL of the 1 in 10 dilutions was filtered through three S-Pak membrane filters (0.45 µm 47 mm white gridded HAWG047S6 S-Pak Millipore Burlington, MA, USA). Antimicrobial activity was neutralised by rinsing the filter with 250 mL of buffered sodium chloride-peptone solution pH 7.0. One membrane filter was transferred on a Tryptone soya agar (P05012A Thermo Fisher Scientific Inc, Waltham, MA, USA) for the enumeration of bacteria. The second membrane filter was transferred on a Sabouraud chloramphenicol gentamicin agar (56 594 Bio-Rad, Hercules, CA, USA) for the enumeration of yeasts and moulds. The third membrane filter was transferred on Tergitol 7-lactose-TTC agar (PO5411J Thermo Fisher Scientific Inc, Waltham, MA, USA) to test for Escherichia coli. Tryptone soya agar, Sabouraud chloramphenicol gentamicin agar, and Tergitol 7-lactose-TTC agar were incubated at 86°F (30°C) for 5 days, at 71.6°F (22°C) for 5 days, and 96.8°F (36°C) for 24 hours, respectively, and the isolated colonies were identified by MALDI-TOF MS (matrix assisted laser desorption ionisation time-of-flight mass spectrometry, Biomérieux).

According to the requirements of the European Pharmacopoeia (Ph Eur) 9th edition, section 5.1.4, Microbial Quality of Pharmaceutical Preparations, European Directorate for the Quality of Medicines Council of Europe, the total aerobic microbial acceptance limits count did not exceed 100 colony forming units (cfu)/mL and the total combined yeast and mould counts did not exceed 10 cfu/mL, which meets the requirements of the test for the absence of E. coli in 1 mL.

Results

Overall, 648 medicine cups were filled by 18 different peristaltic pump heads (ie, 432 quality control cups for accuracy and repeatability, plus 216 quality controls cups for intermediate precision).

Accuracy

The accuracy for the volumes is presented in table 2. Accuracy proved satisfactory for all the cups prepared; no preparations were outside the 90–110% confidence interval in the viscosity index range tested (0.83–8). The parameters computed for the volumes of medications (75 µL to 21 250 µL) were 91.3% minimum accuracy to 109.2% maximum accuracy. Extreme capacities were 37 µL for the smallest volume and 30 mL for the largest.

Table 2

Accuracy and repeatability of the medicine cups with different oral solutions (% of target dose expressed as mean±RSD; n=6)

Precision

The repeatability is presented in table 2 and online supplemental figure 2 and intermediate precision in table 3. The parameters computed for the volumes of medications (75 µL to 21 250 µL) were as follows: 0.05% minimum repeatability to 4.93% for maximum repeatability expressed as %RSD and 0.01% minimum intermediate precision to 6.59% for maximum intermediate precision expressed as %RSD. Only the cups filled with 150 µL of tramadol 100 mg/mL (Contramal) and 150 µL of clonazepam 2.5 mg/mL (Rivotril) were slightly outside the specification range for intermediate precision (RSD >5%). All the other tested volumes were repeatable and reliable solutions (RSD <5%).

Table 3

Quality controls levels with their Intermediate precision expressed as RSD% (four QC on three different days)

The dispenser system met the requirements set for volumes ranging from 75 µL to 21.25 mL with the liquid medication tested.

The best results are obtained with fluoxetine 20 mg/5 mL (Arrow) and diazepam (Valium 1%), likely because of a slower sampling rate. Slower sampling rates allow for internal load cell stabilisation. For viscous solutions (eg, potassium 25 mg/mL (H2 Pharma), the results are in line with expectations using larger tubing and needles: 3.2 mm bore diameter instead of 1.6 mm, and 14G instead of 21G respectively.

Tapering calculation algorithm

Ten series of tapering orders were conducted. Automated tapering calculation allowed for easy production of daily doses for the tapering periods chosen. All the doses calculated met the requirements for acceptance. Both the automatic tapering algorithm and the in-line checkweigher assured that the fill volumes remained within tolerances.

Closure integrity testing

The closure integrity showed no leaks when it was tested by turning each sealed dose cup over and squeezing it lightly between its fingers. Multi-layered composition of the consumable materials, combined with our controlled sealing process and dwell time, assures that a safe, sealed, tamper-proof unit dose package is created each time.

Shelf life

Nevertheless, the stability of the doses was also previously controlled by liquid chromatography–mass spectrometry (Laboratory Cerba www.lab-cerba.com) in various temperatures and over various intervals of time. Our results showed that all the doses of drinkable solutions cannot be diluted with sterile water.3 For non-diluted small volumes, the operculum is pierced and the cup dose is reconstituted with 10 mL of water just before oral administration.

Microbiological stability

Total aerobic microbial count (cfu/mL) and total combined yeast and mould count (cfu/mL) were <1 in the cyamemazine (Tercian) and levomepromazine (Nozinan) samples. No sign of specified microorganism (E. coli) growth was observed in the cyamemazine (Tercian) and levomepromazine (Nozinan) samples after the incubation period.

Discussion

Few data are available on the evaluation of a hospital and industrial co-development in pharmaceutical automation engineering. The performance requirements are yet to be made precise by regulations. To our knowledge, apart from the recommendations made for chemotherapy robots, only a few articles deal with the qualification protocols of such appliances.7–10

One study compared the accuracy of manual and robotic compounding.10 However, our dispensing robot was primarily designed to minimise preparation errors by minimising the human factor during preparation, improve working conditions for our technicians and nurses, minimise the number of full time operators needed, improve quality of preparations, and reduce production time for preparing a large number of medicine cups. Thus, the potential for variation in interpretation of oral syringe marking by different nurses and the inherent variability in syringe manufacturing was not studied. However, large percentage variance deviations in gravimetric measurement are generally observed when small volumes are measured, due to variability in syringe marking or manual drop count.

Our experiments with the Nooddis robot bring new data on this subject. With regards to the accuracy of the internal balance, peristaltic dispensing pumps meet this requirement for routine fill volumes as small as 50 µL. Under routine operating conditions, if the dispensed volume value is ever outside the 10% specification, the dose is rejected and an alarm is triggered to warn the operator.

Intermediate precision expresses within-laboratory variations: different days, analysts, equipment, etc.

Automatic mechanical and gravimetric checks throughout the workflow were specially designed to remove interoperator variability. The machine operator is only responsible for the loading of solutions without the reconstitution step, the replacement of consumables before running the batch, collection of produced cups, and quality control checks based on good manufacturing practices and hygiene requirements. Potential errors during use of the Nooddis robot could be:

  • Mechanical faults such as calibration errors with the checkweigher sensor. However, routine controls of the internal balance were performed every day. Furthermore, over a 9 month period in routine use we did not observe any deviation of the internal balance.

  • Automation algorithm faults. However, we conducted tests and gave directives during the developments.

  • Human faults during the filling of a bottle before running the production. However, the filling is regimented by our assisted system implementing multiple automatic technologies and loading steps that must be done in sequence with no deviation (ie, specific scan with medication data matrix, automatic gravimetric verification, automatic RFID (radiofrequency identification) tag verifications for each bottle’s reservoirs (IP68 ID-TAG E80370 ifm eletronic Germany) and each pump-head, data verifications, etc).

Since there are no manual processes, the pharmacist was the only operator of this study.

Compared with chemo robotic compounding,7–10 the checkweigher automatically drives and synchronises each peristaltic pump head until the right weight of the prescribed dose is filled. In doing so, over-filling and under-filling are avoided. Since a continuous gravimetric flow control is provided, the test for weight variation of unit doses was not conducted. In contrast to existing chemo robotic compounding,7 9 10 dose preparation factors like diluent selection or accuracy of reconstitution of drugs in powdered form before preparation does not exist and the system does not mimic a complex human movement. For example, in intravenous compounding, the gravimetric check does not rely on operator visual examination, so the dexterity of pharmacy technicians or nurses performing the oral dose is not taken into account.

Compared with manipulation or creation of compounded sterile preparations in an aseptic environment, the media fill test was not required. Thus microbiological stability was tested with drug solutions instead of media fills.

The dispensing systems, based on the principle of peristaltic pumps, were tested with solutions of viscosity index ranging from 0.83 for zuclopenthixol (Clopixol) to 11.47 for simple syrup EP, and oral suspensions were not included. Since oral suspensions usually need to be shaken before use (eg, paroxetine 20 mg/10 mL oral suspension (Deroxat) or carbamazepine 20 mg/mL oral suspension (Tegretol)) they are yet to be studied. Thus, a stirring sequence with an in-line compact stirrer (FlatSpin Stirrer, Dlab as well as an homemade model) was ordered for the service provider to ensure correct drug suspension before pumping (online supplemental figure 3). Previous studies have shown that carbamazepine 20 mg/mL oral suspension (Tegretol) can form bubbles of gas during relaxation in some dosing pumps.3

Interestingly, a study compared the accuracy of the strength of drinkable suspensions versus that of prepared capsules.11 Through this study, it was made evident that very often, for the compound preparation of capsules, only a uniformity of mass was achieved. However, we can see that even if the mass uniformity test is satisfied, the dose contained in the capsules can vary by up to 25%.11 This can be of interest for healthcare professionals who prepare unit doses for tapering medications.12 Compounding pharmacists may also be required to design and prepare oral suspensions of medications. Thus, the pumping of a solution or suspension made with an oral liquid vehicle guaranteeing the stability (eg, Ora-blend Sugar Free, Paddock Laboratories Inc, Minneapolis, USA) is also yet to be studied.

The Nooddis robot showed satisfactory results for accuracy and precision, but machine performance (ie, minimum of process capability index) could be improved if our model, guidance and recommendations on principles and methods were exactly respected. For example, six or eight roller pump heads that are known to reduce peristaltic pulsations should be tested in order to improve its performance.

Conclusion

Since the performance of such a complex piece of equipment depends on adherence to the building guidelines given to a service provider, the process performance verification is a necessary prerequisite before starting routine production.

As the system met the required quality standards, the Nooddis robot with an automatic in-line tapering system is an accurate technology to fill the exact amount of liquid oral preparations needed in single dose cups. This may promote closer monitoring, which supports medication tapering as well as medication misuse prevention. With a packaging cost similar to current unit dose cup systems, it is a relevant alternative to fractioned or crushed tablets. Further developments for sterile liquid medications are yet to be studied.

What this paper adds

What is already known on this subject

  • Filling liquid medication cups with the right medication and precise doses is difficult for healthcare professionals, especially in gradually decreasing doses during withdrawal protocols.

  • Doses of liquid medications are an alternative to fractioned or crushed tablets, as well as opened capsules.

  • However, handling liquid doses, splitting pills, crushing tablets, or opening capsules remain a manual process; one with inefficiency that has yet to be solved with automation.

What this study adds

  • The Nooddis is a robot assistance system with automatic in-line checkweigher and a medication tapering system developed by hospital pharmacists to meet quality standards.

  • The 100% in-process checkweighing with feedback is an accurate technology that prepares the exact amount of liquid oral medication in sealed single dose cups.

  • This allows for closer monitoring and support vigilance for medication tapering, as well as drug misuse prevention.

  • With a packaging cost similar to current unit dose cup systems, it is a relevant alternative to fractioned or crushed tablets.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information. Not applicable.

Ethics statements

Patient consent for publication

References

Footnotes

  • EAHP Statement 3: Production and Compounding. EAHP Statement 4: Clinical Pharmacy Services. EAHP Statement 5: Patient Safety and Quality Assurance. EAHP Statement 6: Education and Research.

  • Contributors FL and JL conceived and developed the model, performed the computations and theory and verified experimentally in a proof of-concept experiment. FL and JL wrote all the specifications required for a construction contract to the service provider, contributed to the automation and building directives and performed the pharmaceutical developments of the full scale robot bought for their pharmacy department. FL and JL contributed to the design and implementation of the research, to the analysis of the results and to the writing of the manuscript. The author acting as guarantor is FL.

  • Funding Pierre Lôo Hospital-GHT 58 funded the full scale project to optimise health information technology and digital transformation that enhance safe and effective use of medications.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Open material (https://www.youtube.com/channel/UClUbsWMpJDshQ-oyApKANzw)

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.