Nutritional preparations must be sterile and are only prepared in the optimal safety conditions outlined in the Good Preparation Practices published in 2007  in France. A quality assurance system must be established with competent, trained staff and appropriate facilities.
The preparation of parenteral nutrition bags is a complex task due to the large number of contributors involved. It is also a high-risk process due to the sterility requirements and the vulnerable nature of premature babies. Another specific characteristic of the task is the fact that the European Pharmacopoeia requirements for the incubation time of microbiological samples (14 days ) cannot possibly be adhered to. Bags are only cleared for use once a range of production criteria have been met, thus ensuring the sterility of the final preparation. The chemical risk, meaning wrong product or wrong dose, also has to be taken into account.
PRA was carried out at our University Hospital following the recurrence of undesirable events in the production unit (For instance: hospital information technology system failure, robotic device operation error, lack of necessary skills). This measure is in line with the legal obligations prescribed by French law [3–9].
The PRA method was developed in the early 1960s for the aeronautical and military sectors. Its use in the health sector is now increasing [6, 7]. It constitutes an interesting method for assessing the production of parenteral nutrition bags. While Failure Mode, Effects and Criticality Analysis (FMECA) have already been published for parenteral nutrition bags preparation [10, 11], we report here the first PRA applied to parenteral nutrition production. From a theoretical point of view, PRA method processes from identified system risks and is more exhaustive to FMECA that focuses on one failure mode a time .
The PRA system identifies dangerous situations and the vulnerability they present.
PRA scenarios enable to assess the criticality of a residual risk so that risk mitigation measures can be proposed.
The aim of this paper is to use the PRA method to assess and mitigate the risks involved in preparing parenteral nutrition bags. The methodology consists of two stages: an evaluation of the initial risks followed by an assessment of the residual risks once the process has been optimised.
Materials & methods
The study was conducted over a four-month period at the parenteral nutrition production unit at our University Hospital, where bags for premature babies are prepared. The unit opened in 2001 and operates 6 days per week. There are currently 14 pharmaceutical assistants who have been trained to ensure bags production even on Saturdays. The unit is composed of 3 successive clean rooms: a dressing sas (ISO 8), a decontamination room (ISO 7) and the production room (ISO 6). Production is performed in horizontal laminar flow hoods. Since 2008, a commercial production device (EM 2400 Baxter) has been used to prepare the bags. The whole process, from prescription to preparation to administration, is computerised, with no transcription stage. Some activities are outsourced to other hospital units or external companies (e. g. biocleaning, bags check for dosage and sterility, and maintenance of the air conditioning unit). In 2014, about 2,200 binary bags (containing no lipids) were prepared.
Creating the working group
The project group consisted of the chief pharmaceutical assistant, two pharmacy students and the pharmacist responsible for the activity, who was the methodology specialist and also assumed the role of group leader. The chief pharmaceutical assistant was trained to quality approach. The other pharmaceutical assistants were encouraged to participate to brainstorming and in the scenario analysis.
Identifying the scope of the study
The pharmaceutical stage of the production of parenteral nutrition bags is the focus of the PRA. Prescription, bag administration and patient monitoring are therefore all excluded from the scope of this study.
Internal functional analysis using the RELIASEP method 
The process is broken down into different basic functions, with each function subsequently broken down into three sub-functions (Figure 2). The three sub-functions all follow a logical progression:
Gather (receive data)
Process (analyse data)
Transmit (validate data)
Three phases were identified: 1. Field the request, 2. Ensure production, 3. Carry out checks.
Textbook classifications of dangers vary and are more applicable to clinical services. This means that they are not suitable for production. Therefore, a personalised classification of Mortureux, adapted from Desroches , was drawn up based on experience, literature, brainstorming and the 60 undesirable events which had been recorded over the previous 12 months. The team mapped the risks linked to the process, including five general risks (legal, environment, physical, human and managerial) and 14 specific risks. A further 35 potential risks were also identified. Table 1 shows the general risks.
Identification of dangerous situations
For each phase of the process, dangerous situations were analysed for each risk identified. The level of vulnerability was identified alongside the level of priority, which was determined on the basis of a two-tier classification (in Table 6, high vulnerability and priority attention needed 1 in red, medium vulnerability and priority 2 in yellow).
Defining assessment components
Desroches’ recommendations  were used to determine the assessment criteria. A five-tier rating system was chosen for the evaluation.
Level of Severity (S) (Table 2):
This metric enables to assess the impact of undesirable events. At levels S1, S2 and S3, the impact on patient and equipment safetyis of no consequence. At level S4, the impact is reversible whereas it is irreversible at level S5.
Level of Probability (P) (Table 3)
The scale of probability (P) enables the subjective probability of the occurence of a potential risk to be assessed.
Determining decision components
The three-tier criticality scale (C1–C3) and the four-tier effort scale (E0–E3) are decision-making components which assist in measuring the impact and acceptability of a risk in order to determine priority measures to be taken.
Level of criticality and decision-making framework (Table 4)
The decision-making framework shown in Table 4 is based on the levels of severity and probability. It provides an overview of the criticality level of initial risks and forms the basis for determining which response measures are to be taken.
All scenarios with an initial criticality level of 2 or above require action to be taken to reduce the level of risk.
Level of effort (E) (Table 5)
The level of effort (E) required to address the risk, i. e. the financial cost of eliminating, reducing, managing or controlling the risk, is determined by a four-tier classification (Table 5):
Identification & execution of scenarios
In order to craft PRA scenarios, an Excel® 2010 worksheet was used to list incident scenarios, i. e. the cause or initial event which triggers a dangerous situation, its consequences, existing corrective measures and the criticality of the initial and residual risk for each dangerous situation.
Risk mitigation measures were researched for all scenarios with a criticality level of 2 or above. Residual criticality, i. e. the remaining process criticality once risk mitigation measures have been taken, was then determined and residual risk management indicators were proposed.
Results & discussion
The PRA method suggested by the National Authority for Health [6, 7] can be used at any point during a process . This approach is appropriate in the process of producing parenteral nutrition bags. Studies attest to the use of PRA in cancerology, radiotherapy, sterilisation and administration of medicines in hospital pharmacy [13, 17–20] however there is no evidence of it being used for parenteral nutrition.
The PRA system: identifying situations on the basis of general and specific risks and determining their vulnerability
Risks flagged up during the different stages of the process of producing parenteral nutrition bags enabled the identification of 63 dangerous situations (Table 6), of which 77.7 % have a vulnerability level of 2, i. e. significant. This confirms the critical nature of the process and enables priority measures to be drawn up to improve it.
Phase 2, i. e. the production phase, features the highest number of dangerous situations (26) followed by phase 1 (field the request – 19) and phase 3 (checks – 18). This corresponds to the general feeling amongst the team: any product or sterility error during the production phase could be fatal to a premature baby. This is attested to in a paper on ensuring safety during chemotherapy treatment, including the preparation stage, drawn up in 2007 by Hayat .
Phases 1, 2 and 3 feature several dangerous situations relating to task errors. This is due to the numerous tasks that pharmaceutical assistants have to perform to decontaminate apparatus, prepare syringes, set up the robotic device and carry out checks. This observation is also made in Charpentier’s paper on ensuring safety whilst restocking a robotic nominative dispensing unit .
In phase 1, there are also dangerous situations relating to the Hospital Information Technology (HIT) system. A HIT problem (network, software, printer) requires the use of a staggered procedure, which is a potential source of risk. In phase 2, the HIT system (software, network or printer error) and healthcare products (shortage, market change) are potential sources of dangerous situations.
PRA scenarios: Assessment of initial risk criticality, proposals for mitigation measures and evaluation of residual criticality
On the basis of the 71 scenarios described, the percentage share of the initial criticality levels has been calculated for each general risk (Figure 3).
The fact that there are several level 3 criticality risks across the five risk types shows that the process has not been made entirely risk-free. Although the number of scenarios is the highest in the ‘human’ risk category (Figure 3), this category has also seen the most efficient risk mitigation activity (it has the lowest proportion of level 3 criticality risks). One of the contributing factors is initial staff training [22, 23]. Manual additions during the production phase (products added manually following the production of the bag by the robotic device) require a greater level of risk mitigation. This risk is also mentioned in the work carried out by Kotzki et al. . The highest number of level 3 criticality scenarios was in the environmental and managerial risk categories. In the environmental risks category, all scenarios linked to ambient air quality and the performance of the hood, the robotic device and the Hospital Information Service (HIS) are of level 3 criticality. As for the managerial risks, those linked to outsourcing and service quality in particular score highly in terms of criticality. Bags which are non-compliant in quantitative or qualitative terms constitute a potentially lethal risk for the patient.
In order to make the process secure, a range of measures have been put in place and categorised on the basis of the level of effort required. The following measures entail an effort level of 2 or under:
Work with the anti-nosocomial infection unit and the budget unit to identify required skills for maintenance staff, review the biocleaning method, set up visual aids, deliver continuous training courses in-house for staff of external contractors and carry out an annual activity audit for maintenance staff.
Establish charters for sub-contractors.
Further strengthen the skills acquired by pharmaceutical assistants during training courses by implementing a continuous professional development programme.
Perform double checks on all manual additions to bags.
Organise training sessions on safety awareness.
There is one measure with an effort level of 3:
Change the production software and microbiological safety cabinets (ISO 5 for handling activities)
Distribution of residual risk criticality per general risk
Following the implementation of the risk mitigation measures listed above, residual risk criticality was reassessed.
As shown in Figure 4, the PRA method has enabled to suggest risk mitigation measures and reduce system criticality. Initially, there were 4 scenarios classified as C1, 48 as C2 and a further 20 as C3. Now, there are 49 C1 scenarios, 30 C2 scenarios and 0 classified as C3. The most successful risk mitigation measures pertain to physical risks, followed by environmental, human, managerial and legal. It should be noted that the events identified did not cause harm to patients thanks to effective team work.
PRA comparison to other risk analysis methods
Despite PRA has never been published for bags parenteral nutrition production process, other approaches are reported in literature, such as FMECA . For instance, Tours University Hospital reports the results of a FMECA, performed with the objectives to replace the robotic device. The area was quite different from the present PRA, Tours being more focused on prescription step and less on outsourcing step for instance. Nevertheless, it highlights similar critical points: staff formation and accreditation, aseptic process, product error, manual addition to bags. The PRA revealed a more important risk of checking step and outsourcing, Concerning skills of pharmaceutical assistants, it was more critical in PRA, this can be attributed to the important number of pharmaceutical assistants to assume NP production on Saturdays.
However, there are certain limits to the PRA and FMECA methods, such as the subjective nature of risk evaluation. This is all the more relevant given the small size of the working group  and the fact that the evaluation focused on the team’s work at a given moment in time. The results should be interpreted as orders of magnitude which make it possible to assess the team’s perception of risks. It transpires that a mere 15 % of practitioners actually report undesirable events . This emphasises the importance of developing a culture of safety to encourage the reporting of such events.
To avoid the emergence of such problems, specialised teams use complex mathematical approaches (Montecarlo) to quantify risk in certain areas .
The preliminary risk analysis method enabled risks to be assessed and risk mitigation measures to be taken. This can be beneficial for healthcare institutes, as stated by Desroches , but it requires training and time. It also serves as a communication tool for all involved in the process. Thanks to this method, the criticality of risks which could affect the system has been reduced. The establishment of residual risk mitigation indicators will enable the efficiency of the action plan to be evaluated, thus ensuring process safety. Professionalism, team involvement and team work ensure a pro-active approach to risk management.
The authors wish to thank the team of pharmaceutical assistants and the two pharmacy students.
1. Ministère de la santé de la jeunesse et des sports, Agence française de sécurité sanitaire des produits de santé. Bonnes pratiques de préparation. Bulletin Officiel; 2007. Available online: http://ansm.sante.fr/var/ansm_site/storage/original/application/a5d6ae4b3d5fdee013ca463462b7b296.pdf
3. Direction de la recherche des études de l’évaluation et des statistiques. Michel P, Minodier C, Lathelize M, Moty- Monnereau C, Domecq S et al. Les évènements indésirables graves liés aux soins dans les établissements de santé. Résultats des études nationales réalisées en 2009 et 2004. Dossier Solidarité Santé 2010; (17).
4. République française. Arrêté du 6 avril 2011 relatif au mangement de la qualité de la prise en charge médicamenteuse et aux médicaments dans les établissements de santé. Journal Officiel du 16 avril 2011. p. 6687. Available online: https://www.legifrance.gouv.fr/eli/arrete/2011/4/6/ETSH1109848A/jo
5. République française. Loi n° 2009-879 du 21 juillet portant réforme de l’hôpital et relative aux patients, à la santé et aux territoires. Journal Officiel du 22 juillet 2009. p. 12184. Available online: https://www.legifrance.gouv.fr/affichTexte.do?cidTexte=JORFTEXT000020879475&categorieLien=id
6. Haute Autorité de Santé. Manuel de certification des établissements de santé V 2010. Available online: http://www.has-sante.fr/portail/jcms/r_1439924/fr/manuel-de-certification-des-etablissements-de-sante-v2010-revise-avril-2011
7. Haute Autorité de Santé. Manuel de certification des établissements de santé V 2014. Available online: http://www.has-sante.fr/portail/upload/docs/application/pdf/2014-03/manuel_v2010_janvier2014.pdf
8. INSTRUCTION N° DGOS/PF2/DGS/PP2/2015/85 du 20 mars 2015 relative à la gestion des risques liée à l’activité de nutrition parentérale en réanimation néonatale, en néonatalogie et en pédiatrie par la mise en place de bonnes pratiques organisationnelles. Available online: http://circulaire.legifrance.gouv.fr/index.php?action=afficherCirculaire&hit=1&r=39383
9. INSTRUCTION N° DGOS/PF2/DGS/PP2/2015/360 du 15 décembre 2015 relative à l’organisation de la mise en œuvre du diagnostic de territoire relatif aux pratiques de préparation des poches de nutrition parentérale. Available online: http://nosobase.chu-lyon.fr/Reglementation/2015/instruction/15122015.pdf
10. Fauchere C, Rudolf von Rohr T, Fleury S, Bouchoud L, Fonzo-Christe C, Cingria L, et al.. Sécurisation de la fabrication de nutritions parentérales pédiatriques : réévaluation d‘une analyse de risques. Le Pharm Hosp Clin 2015;50(3):317. Doi: . CrossrefGoogle Scholar
11. Boddaert S. Les différentes méthodes d‘analyse des risques en pharmacie hospitalière: application de la méthode AMDEC à la production de nutrition parentérale pédiatrique. Université de Tours: 2012.
12. Haute Autorité de Santé. La sécurité des patients: mettre en œuvre la gestion des risques associés aux soins en établissement de santé - Des concepts à la pratique. 2012. Available online: http://www.has-sante.fr/portail/jcms/c_1239410/fr/mettre-en-oeuvre-la-gestion-des-risques-associes-aux-soins-en-etablissement-de-sante
13. Bonan B, Martelli N, Berhoune M, Maestroni M-L, Havard L, Prognon P. The application of hazard analysis and critical control points and risk management in the preparation of anti-cancer drugs. Int J Qual Health Care 2009;21(1):44–50. doi:. CrossrefWeb of ScienceGoogle Scholar
15. Mortureux Y. Analyse préliminaire de risques, Edition Techniques de l‘ingénieur. Available online: https://books.google.fr/books?id=H4QpGkqQtEsC&printsec=frontcover&hl=fr&source=gbs_ge_summary_r&cad=0#v=onepage&q&f=false
18. Damien Talon. Gestion des risques dans une stérilisation centrale d‘un établissement hospitalier: apport de la traçabilité à l‘instrument. Autre. Ecole Centrale Paris, 2011. Français. l–00680977.
19. Nguyen TD, Devie I, Heusghem M, Gaillot-Petit N, Loiseau M. Cartography and risk management in radiotherapy: A collaborative work of the department of radiotherapy and the department of quality and risk management at the Jean-Godinot Institute. Cancer radiothérapie J la Société Fr radiothérapie Oncol 2010 Jan;14(1):24–8.Google Scholar
20. Mazeron R, Aguini N, Deutsch É. Risk analysis in radiation therapy: state of the art. Cancer radiothérapie J la Société Fr radiothérapie Oncol 2013 Jan;17(4):308–16. quiz 332. Google Scholar
22. Vrignaud S, Le Pêcheur V, Jouan G, Valy S, Clerc M-A. Staff accreditation in parenteral nutrition production in hospital pharmacy. Ann Pharm Fr 2016 Jan 27. doi: . [Epub ahead of print]. Crossref
23. Riquier T, Talon D, Gallet C, Chast F. Les bienfaits de la gestion des risques par la méthode d’Analyse Préliminaire des Risques aux interfaces du système en stérilisation centrale. 2014. Communication affichée [résumé]. Congrès Hopipharm, la Rochelle, 2014.
24. Kotzki S, Minoves M, Robein-Dobremez M, Trouiller P, Luc F. Evaluation des pratiques et des risques à chaque étape du processus de préparation des poches de nutritions parentérales. Le Pharm Hosp Clin 2012;47(S1):S78–S79. Doi : 10.1016/j.phclin.2011.12.190. CrossrefGoogle Scholar
25. Djermoune SO, Oukka M, Bouzid K, Denine R, Bonnabry P. Use of a prospective risk analysis method to improve the safety of the cancer chemotherapy process in a medical oncology service in Algeria. Le Pharm Hosp Clin 2016;51(1):40–50. Doi: . CrossrefGoogle Scholar
About the article
Véronique Le Pêcheur
Véronique Le Pêcheur is a pharmacist working at the pharmacy department of Angers University Hospital since 1990. Between 1995 and 2015, she was head of Pharmaceutical technology department. She implemented cytotoxic drugs preparation at the pharmacy in 1995 and opened the parenteral nutrition production unit in 2002. As a graduate in risk management, she developed a quality approach including staff accreditation and risk management policy applied to pharmaceutic production.
Laurence Spiesser-Robelet is pharmacist at the pharmacy department of Angers University Hospital since 2009. She is currently a doctoral student in public health in the field of perinatal care. She worked in several parenteral nutrition units in pediatric hospitals. Her special interests and research projects concern the pediatric medicines, the proper use of medicines, clinical pharmacy and patient education
Sandy Vrignaud is a pharmacist working at the pharmacy department of Angers University Hospital since 2006. In 2012, she completed her PhD thesis entitled: “Nanoparticular systems for encapsulating hydrophilic drugs”. Since 2015, she is the head of Pharmaceutical technology department. Her special interest and research projects include formulation, physico-chemical characterization and production of pharmaceutical preparations such as topic, eyes drops, injectables or oral forms.
Published Online: 2016-08-06
Published in Print: 2016-06-01
Conflicts of interest statement: The authors state no conflict of interest. They have read the journal’s Publication ethics and publication malpractice statement available at the journal’s website and hereby confirm that they comply with all its parts applicable to the present scientific work.