Step 1: Identifying influential variables
A total of 66 variables influencing the pricing of LMPs were extracted from reviewing regulations and expert opinions. These variables were classified into eight categories or domains in order to facilitate the expression of various variables. The domains were selected based on the discussions conducted by researchers of this study. Since statistical analysis is not performed on these domains separately, the displacement of variables within these domains will not have any effect on the results of the study. These domains and the details of extracted variables are presented in
Tables 1 and
2.
Step 2: Designing the questionnaires and assessing validity and reliability
The questionnaire was developed based on variables identified in the qualitative study. This questionnaire included 66 questions and was structured in eight defined domains for more convenience in completion.
Face validity was assessed, and some changes were performed on the structure of the questions so it has more visual appeal. For quantitative assessment of face validity using the item impact method, the participants ranked all questions in terms of importance and all questions were scored more than 1.5 points. This means that in all questions, at least four people gave points 4 or 5 to the questions and the minimum average score for each question was equal to 3, and therefore, all questions remained in the questionnaire.
In content validity assessment, Item Content Validity Index (I-CVI) and Scale Content Validity Index (S-CVI) of each question and the whole questionnaire, respectively, regarding the relevance and clarity exceeded the acceptable level of 78% for I-CVI and 80% for S-CVI. Only one question with both relevance and clarity I-CVI below acceptable level was eliminated by the research team. The interrater agreement index was about 93.3% for relevance and 83.3% for clarity of questionnaire, and both cases were above the acceptance level of 70%. The comprehensiveness of the questionnaire was calculated to be 91.6%.
For assessing test-retest reliability, all 11 participants answered all questions on both test and retest time. Weighted Kappa coefficients were calculated for each question and all achieved the acceptable level. Therefore, there was no need to change any questions of the questionnaire.
Cronbach’s alpha was calculated for the total instrument. This coefficient was equal to 0.873, so it exceeded the acceptable level. Therefore, it concluded that the questionnaire had acceptable internal consistency and all questions were worthy of being remained in the questionnaire.
Step 3: Collecting answers to the questionnaire by pharmaceutical Experts
Among 59 experts selected for this study, 46 experts finally responded to the questionnaire. Thus, the response rate was equal to 78%. The number of participants in each study group is shown in
Table 3.
According to the analyzing method for Likert scale-based surveys, “Mode” and “Range” were calculated for raw data of two groups of study participants (Group A and B) separately so that a comparison can be made between these groups.
Regarding the “range”, as a measure of dispersion in discrete variables, it can be seen that the answers are at most two levels apart. This does not provide us with so much information, because these two levels can be attributed to the difference between the scores 1 or 5 with 3, or whether the difference between 4 and 2.
In order to determine the variables with the greatest effect on pricing LMPs from Group A and Group B viewpoints, variables were ranked based on the number (%) of people who responded 4 and 5 (High effect and very high effect), in each group A and B.
Accordingly, the variables determined to have the highest effect (4 and 5) on LMPs pricing from the viewpoint of more than 50% of participants in Group A and B, are presented in
Table 4. We can see from the viewpoint of Group A, there are 26 variables, and from the viewpoint of Group B, there are 28 variables, which more than 50% of experts in each group identified them as the most influential variables (4 or 5) in LMPs pricing. Sixteen variables were found to be common between these two groups, as shown by “AB” in the “Groups viewpoint” column of
Table 4. Group A and B pointed to 10 and 12 other variables, respectively, which have been considered as the areas of disagreement and shown by letter “A” and letter “B”, respectively, in “Groups viewpoint” column of
Table 4. The percentage of those who gave these variables 4 and 5 scores in the opposing groups was less than 50%. It means that less than 50% in the opposing group considered these variables important in the pricing procedure of LMPs.
The Mann-Whitney U test was performed to determine whether the differences between responses of these two groups were statistically significant. Regarding variables influencing LMPs pricing, in 17 out of 66 questions, there were statistically significant differences between the two groups, and there was no statistically significant difference in the remaining variables.
As can be seen in
Table 4, all variables which were found to be different from the viewpoint of group A and B in primary analysis using “Mode”, showed a significant difference using the Mann-Whitney U test, except 5 variables with the code of D2-9, D4-4, D4-6, D4-11, and D7-8. The p-value of these variables is underlined in
Table 4. The direction of the differences can be obtained from the “Groups viewpoint” Column. It can also be concluded that those variables with the code of D2-9, D4-4, D4-6, D4-11, and D7-8 are among the common variables between groups A and B.
According to these data analysis, variables including being an orphan medicine, the number and variety of production steps, being in a drug shortage situation, number of years of price stability, reference basket price, the cost obtained from Pharmacoeconomic studies, exchange rate and its fluctuations, customs duties, financial costs of the company, and general inflation rate (official rate) are among the effective variables in pricing procedure from the viewpoint of Group A experts (the Commission Members). On the contrary, Group B experts believe that the Commission Members place very little importance on these variables when setting a price.
On the other hand, Group B experts believe that variables like the way of presenting a pharmaceutical pricing file by a pricing expert, conflict of interest of members of the commission, personal tastes and subjective preferences of members of the commission, time of presenting a drug pricing file in the commission, physician’s and some association’s lobbying, presence of MOC representative in the commission, and supplier company advertisement before entering the market are among the most effective variables in pricing procedure in the commission. As can be seen, almost all of these variables are categorized in “conflict of interest, personal tastes and subjective preferences” domain.