For a successful introduction of clinical pharmacy services on the wards of a hospital, the support of hospital management is crucial. The acceptance of a clinical pharmacists by other members of the health care team depends mainly on their individual attitude, perception, and personal experience. It is necessary to investigate the needs and expectations of the users before the introduction of any new service, including clinical pharmacy. Clinical pharmacy (CP) service was not used in University medical Centre Ljubljana before 2010. Although the management strongly supported the introduction of clinical pharmacists into health care teams, some of the head physicians doubted it. Before the introduction of clinical pharmacy service in the hospital it was necessary to assess the attitudes of head physicians and head nurses about it, as well as the attitudes of pharmacists as potential providers of the new service. Contemporary society is characterized by plethora of hardly manageable data. The need to store and analyse a huge amount of data and discover useful information hidden in it lead to the development of new methods based on artificial intelligence approaches, statistical methods and machine learning algorithms. Data mining is a process to explore big data warehouses and data bases to discover hidden information and knowledge. Data mining techniques allow automatic prediction of future trends and behaviours, and automatic discovery of hidden patterns in data. In plethora of new methods, we can find some, which are capable to analyse small samples. Medical and pharmaceutical sciences often lack big samples, which is true also in marketing and user satisfaction studies. To analyse small samplesclassification approaches often extract important variables for a given problem. Each variable is evaluated according to the class in question. In practice this means that we can consider only the variables with the strongest impact on the outcome of a given classification problem. Based on these considerations we conducted a survey about the role of clinical pharmacist in University medical Centre Ljubljana. A comprehensive literature search was performed to prepare a list of all possible clinical pharmacy activities. The list was further used to construct a survey questionnaire with Likert measurement scale to conduct a descriptive observational attitude study of physicians’ and nurses’ opinion about the importance of each of the listed activities and competencies of clinical pharmacists. The questionnaire was validated by three experts, a non-hospital pharmacist, hospital head physician, and a human resource manager.
The questionnaire is composed of three types of questions. In the first part of the questionnaire (17 questions), clinical pharmacy activities pertaining to the hospital system are stated, while the second part of the questionnaire (19 questions) contains activities directly connected with individual patient’s care. The third part (16 questions) of the questionnaire deal with clinical pharmacist’s competencies. The participants had to choose the level of agreement on the Likert scale from 1 (I totally disagree) to 5 (I totally agree) with each of the listed affirmative statements in the part one and two of the questionnaire, while indicating the importance (form least important – 1 to very important – 5) of a particular competence in the third part. The questionnaire was sent to 43 physicians – medical directors or heads of departments – and to their head nurses. The survey results were collected in a spreadsheet, which served as a basement for all following analyses. In the first chapter we discuss the results obtained with nonparametric ANOVA tests. We observed considerable differences between pharmacists, physicians and nurses in total mean score of agreement to each statement and to each group of statements. The largest differences exist between the mean scores of pharmacists and nurses. The differences between physicians and pharmacists are also important. The differences between physicians and nurses are smaller. Total mean score of physicians’ agreement with the first and second group of CP activities is 4.28 and 3.73, respectively, while these scores are lower for nurses (3.87 and 3.38 for the first and second group, respectively). Pharmacists’ total mean scores are highest, 4.57 and 4.23 for the first and second group, respectively. Afterwards this preliminary phase we established clinical pharmacy service on designated wards. The satisfaction with the new service was assessed by the head of the pharmacy and the values were entered into the database. This appended database was used to perform analyses with data mining algorithms. In the second Chapter we present the results obtained with OrdEval algorithm, which can determine the importance of each CP activity and their type according to the users’ expectations. The results were used to categorize the activities/competences according to Kano model. Using analysis of individual feature values, we identified six performances, 10 excitements, and one basic clinical pharmacists’ activity. Other feature evaluation algorithms, like ReliefF and MDL, can also be used for the assessment of the importance of the variables. Survey data were analyzed using these two algorithms to identify the most important clinical pharmacists’ activities/competences. The results were compared with the expert estimation of the head of the pharmacy and to the values calculated using OrdEval algorithm. The results are analyzed in the third Chapter. We conclude that using different data mining algorithms is justified. In the case of congruence of the results their reliability is improved. The results on importance of CP activities for both algorithms agreed with the expert estimation for eight CP activities/competences. Finally, in the fourth Chapter we present the pilot case study report on the text analysis of the text of five semi-structured interviews with head physicians about their satisfaction with clinical pharmacist after establishing the collaboration. The text was analysed with QDA Miner text management and qualitative coding software (QDA Miner 4, Provalis research). We identified two reversal CP activities according to Kano model which were not discovered in our previous analyses.
|
