Head of Pharmacy
Department of Pharmacy
Charleroi University Hospital
In 2000, the wrap used as a compressive bandage for postcoronarography in our hospital generated so many severe adverse events that the Pharmacy and Therapeutics (P&T) committee was required to select an optimal elastic compression wrap among the six models available on the Belgian market. Selecting such a bandage was difficult because no literature was available and, obviously, some of the desired criteria were controversial. Indeed, it was crucial to avoid bleeding from arterial and/or venous puncture(s) for at least 24 hours, but without provoking pseudoaneurysm, allergy and/or phlyctena, and the bandage could not form strands or twists acting as tourniquets, especially in reclined obese patients.
Because of its major responsibility in the hospital medication use system, the pharmaceutical team was requested to endorse the critical role in this selection. With the help of the nurses concerned, comparative tests in the real hospital setting were designed. From a methodological standpoint, although the team was used to SOJA and Pharma Decision commercial software, they wanted to develop a more comprehensive, easy-to-use, costless but effective approach.(1)
Robustness, a concept developed by Taguchi(2) for the automotive industry, claims that, when faced with complex choices, instead of an excellent product with a major defect or with some criteria with high variability, preference has to be given to a product rated good for all criteria provided these are of very low variability. The team therefore decided to combine classic statistical tests with robustness to select the best wrap on the basis of an ensemble of clinical outcomes (Douchamps J, et al; 34th ESCP Symposium, 2005).
Materials and methods
Six 10cm-wide elastic adhesive wraps were used: Elastoplast-E (Beiersdorf), Fixoma (Stella), Hypolastic (Hartmann), Microfoam (3M), Tensoplast-AB (Smith & Nephew) and Tensoplast-LA (Smith & Nephew).
Six groups of 50 patients (male and female) undergoing coronarography were compared according to an open randomised design with strictly standardised procedures.
After sufficient training, nurses and coronarography technicians served as raters and, in the final phase, as deciders. Six criteria were scored on ordinal subjective Likert’s scales (min=bad, max=excellent): bandaging easiness (0–3); bandage withdrawal easiness (0–1); absence of allergy (0–4); absence of phlyctena (0–1); compression quality (0–3); and absence of glue residues (0–1).
Kruskal–Wallis nonparametric ANOVA with post-hoc tests was used for multiple comparisons and Mann–Whitney U-test served for two-group comparisons (p<0.05).
Graphical representation of quality robustness
The six criteria, which were given identical weights by the raters, were expressed, for each wrap, with a pair of radar-type hexagonal graphs. The six corresponding axes of each graph were graduated in percentages to give them the same heights. The first graph was constructed with mean values scores after conversion of their respective full scales into percentages (0%=origin, 100%=top). The second graph was similarly constructed but with the standard deviations of these scores. In this way, joining the values on the six axes delimited a surface area which, in the first graph, represented quality and which, in the second graph, represented the variability of quality. A big surface in the first graph associated with a small surface in the second graph thus described a robust quality. Beyond the subjective appreciation of the global surface area in the second graph, attention was also given to shape asymmetry, as it could be related to a major defect.
The six patient groups were proven to be of similar age, weight, height, initial cutaneous states, state of mental stress and type of invasive examination. Statistically significant differences were found for all criteria evaluated except for withdrawal easiness (see Table 1). Post-hoc tests for the five other criteria revealed that the statistical results pattern was too complex for easy selection of the optimal wrap.
Six sets of mean and standard deviation graphs were then constructed. Raters were asked to combine intuitively the latter with this graphical approach in order to select a wrap. Hypolastic emerged as the best choice. Figure 1 shows that Elastoplast and Fixoma were respectively excellent and rather good, but both with a major defect, which was not the case for Hypolastic.
Discussion and conclusion
After a further six-month follow-up on 104 patients, Hypolastic has been used exclusively for the last three years (>1,000 postcoronarography patients). So far, no adverse reactions or complaints have been reported. These long-term results tend to support the combination of discrete statistical tests with robustness multicriteria graphics as a worthwhile method for conducting multifaceted, complex decisional processes. On a more general concern, this study confirms that the hospital pharmacists can play a leading role in designing and carrying out comparative tests to help P&T committees in selecting new drugs and medical devices.
Conflicts of interest
The study and the three-year follow-up have been conducted so far free from any grant of any kind from the pharmaceutical companies concerned
- Douchamps J. Qualifying quality and the model of anticipation, realization, control: application to the medication use system in hospital. European PhD thesis in Pharmaceutical Sciences. Rouen University, Faculty of Medicine and Pharmacy; 2002.
- Ross PJ. Taguchi techniques for quality engineering. New York: McGraw Hill; 1988.
European Association of Hospital Pharmacists
European Society of Clinical Pharmacy