Numerous steps have been taken to ensure the reliability of the data collected with the PLAISIR^© system.  A certain number have already been described, yet, given the importance of the subject, we will cover them all in this section.

The key of the evaluation process is of course the nurse-evaluator.  This person generally belongs to the institution under consideration and a priori, has never completed any evaluations with a format or procedures similar to those of the PLAISIR^© system.  She must then be trained to perform the PLAISIR^© evaluations.  This training session has two stages.  The first consists of a group training session (maximum of twelve participants with groups of four or five generally) lasting two to three days.  This session is essentially a review of the FRAN^©, heading by heading.  The reference manual of the system supports this whole process.  The theoretical approach is followed by a practical one in which the "students" evaluate a prototype client using the FRAN^©.

The second step of the process consist of an on-site individual training session which lasts two days.  This training session takes the following form.

Following the 2 to 3 day, "group" training session, the "student"-evaluator returns to her institution and completes 10 PLAISIR^© evaluations.  An instructor will then meet her in her institution.  The instructor randomly chooses 3 clients among the 10 evaluated, independently revaluates them and finally compares her evaluations with the student's evaluations.  Finally, she "reviews" with the student the seven remaining evaluations.  This process lasts a day.  The "student" is then asked to perform ten other PLAISIR^© evaluations.  Once these are done, the instructor returns to the institution, randomly chooses 6 clients among the ten just evaluated, independently assesses them and finally compares her evaluations with those of the student.  This process lasts a day.

The process of desk-reviewing all the evaluations allows - as a byproduct - the identification of "problem" evaluators to which one offers a "follow-up training", in groups or individually.  The format of these follow-ups is variable.  It depends on the types of problems that are detected.  There are also "refresher sessions" which are aimed at trained evaluators who have not performed evaluations for a prolonged period (a year or more).

The number of evaluators is purposely limited.  At the system's inception in 1983, one had a target of training one evaluator per 100 beds to ensure that each one performs a sufficient number of evaluations per year to maintain her skill.  With time, one has trained an average of one evaluator for approximatively fifty beds. This number, in most cases, allows for a good quality of data collection.  The training of a larger number of evaluators than planned is due to the small size of a number of institutions and to the fact that, in certain facilities, management wanted the head-nurses of the respective units to be responsible of the evaluations in their unit.

One of the key elements of data quality assurance is the desk-review of all the evaluations  and their eventual "correction" following the telephone call to the evaluator/care-givers of the client.  To detect the problems, the nurse-desk-reviewers is aided by the computer's program which detects certain anomalies and signals them to the desk-reviewers in the IMPFRANs^©.  The dual character of the PLAISIR^© evaluations is of great importance in the desk-reviewing process.  In fact, the FRAN^© allows one to collect two portraits of the same client :  his bio-psycho-social profile and his profile of nursing care and assistance services required.  An experienced desk-reviewer rapidly detects inconsistencies between these two portraits which allows her to prepare the questions she will ask during the telephone call to the evaluator or care-giver.  It is also important to note that when the IMPFRAN^© of a client is reviewed, it is always compared with the previous IMPFRAN^© if the client had already been evaluated with the PLAISIR^© system.  This comparison is helpful in detecting incongruencies.  Finally, one must note that experience is important in such a system of data verification.  Certain desk-reviewers have reviewed since 1983 several tens of thousands IMPFRANs^©, what helps them considerably in their task.

The process of desk-reviewing and correction is also an ongoing training experience for the evaluators as they are informed of their errors and omissions during the telephone contact with the reviewers.  Furthermore, the nurses responsible of the desk-reviewing are also responsible of the training of evaluators.  They are fully aware of the FRAN^© items that cause the most difficulties to the evaluators and can focus on these items in the training-sessions.  Finally, these same nurse-desk-reviewers pilot the revision/updating process of the PLAISIR^© system (reference manual and FRAN^© form).

In addition to the (microscopic) desk-review  of the IMPFRANs^©, there is the (macroscopic) review  of the PLAISIR^© system outputs.  This last step allows for the eventual detection of systematic bias (inflated evaluations, for example) which may have gone unnoticed in the desk-review of individual evaluations.  During the process of reviewing the outputs, one always compares the current outputs of a program or institution with its last outputs.

Other important elements of the quality assurance of data are the standards of care which were mentioned earlier and which are a necessary condition for the homogeneousness of evaluation and the reference manual which gives accurate and detailed definitions of each of the FRAN^© headings.

Finally, in very rare cases, where all these measures seem insufficient to ensure reliability of the data, an exceptional verification of the evaluations is done on site by nurses not belonging to the institution.  They are generally nurse-instructors-desk-reviewers responsible of the PLAISIR^© system.