Perception of cost-consciousness among medical students : an 11-item scale

| INTRODUCTION: Managing healthcare resources in a cost-conscious way is an important topic in medical education. A validated instrument to measure the perception of cost-conscious attitudes of medical students could be an important tool for university staff in understanding them and to plan and evaluate educational interventions to address them. OBJECTIVES: To validate a scale to measure the perception of cost-conscious attitudes among medical students. Method: Thirteen items from a survey to measure cost-conscious attitudes in medical students were used. Semantic validation and psychometric analysis (exploratory factor analysis, Cronbach alpha, and composite reliability test) were carried out. RESULTS: A sample of 275 participants, of which 62% were female, with an average age of 25±3.1 years, completed the questionnaire. The best structure to measure the perceptions is the one-factor solution, with 11 of the 13 items remaining in the final version. Substantial Cronbach’s alpha for this questionnaire was obtained 0.66 and composite reliability was 0.77. CONCLUSION: The results suggest that the questionnaire has a satisfactory psychometric quality and is, therefore, able to measure the perception of medical students about cost-conscious attitudes.


Introduction
The cost-conscious management of healthcare resources is an important subject in the education of medical students and physicians.The cost of medical tests and procedures should reflect the principles of medical professionalism, from the patient's welfare to respect for the patient's autonomy and social justice 1 .
It is estimated that about 30% of total health costs are unnecessary and could be avoided without harming patient care 2 .Requests by doctors for low-value tests, treatments, and procedures are among the average sources of waste in healthcare 3 .According to Brazilian Ministry of Health data, there are requests for tests after 30 to 50% of outpatient visits to the Unified Health System (SUS), the Brazilian public healthcare system. 4Eliminating or at least reducing the waste from these poorly thought out medical decisions could result in improvements in safety and efficiency in treatment for patients, as well as bring benefits to the sustainability of the health care system.
To encourage more cost-effective medical decisions, it is important to raise awareness among physicians and medical students about this topic daily.The international campaign Choosing Wisely lists situations that occur frequently but that should be avoided 5 .
Educational interventions to promote more costeffective thinking right from medical school are important in changing the behavior of our physicians.In Brazil, the Bahiana School of Medicine and Public Health (EBMSP) has been using an educational approach inspired by the Choosing Wisely campaign on their students and teachers 6 .
Society can benefit from cost-conscious attitudes in healthcare 7 .However, there is a lack of validated instruments to evaluate educational interventions to measure such attitudes.There are some validated cost-conscious perception scales for physicians 2,7-10 , but no validated scales appropriate for medical students have been found.
A survey was applied to 3,395 medical students in the United States to evaluate their perceptions of cost-consciousness 11 .Although the questions were based on other cost-conscious validated scales for physicians 8,9 , the analysis of psychometric characteristics of the survey was not described by the authors.
A validated instrument can be useful to measure the perception of cost-conscious attitudes of medical students before and after educational interventions and to compare students exposed to different scenarios in their training.The purpose of this research is to validate a scale to assess the perception of cost-conscious attitudes among medical students.

The instrument
The questionnaire used in the study was based on a cost-consciousness perception survey originally published by Leep Hunderfund et al 11 .The 13 items categorized as "attitudes" were used in the proposed questionnaire.The questions used a fourpoint Likert scales for scoring, with the following choices: strongly disagree, moderately disagree, moderately agree, strongly agree.Other variables included in the instrument were age, semester of graduation and sex.

Translation and semantic validation of the instrument
The translation and semantic validation were performed according to the international standards 12 .The items, originally written in English, were translated into Portuguese by two independent bilingual Brazilian professional translators.A third translator, who is a native English speaker fluent in the Portuguese language, retranslated the reconciled version from Portuguese to English.The original version and the retranslation were then compared and considered equivalent.
An expert committee, made up of teachers who were familiar with the construct, produced the prefinal version of the translation.Discrepancies were eliminated by consensus on all items.For semantic validation, a group of 11 third-year medical students also assessed the scale in Portuguese.The observations and suggestions made by them were taken into consideration in preparing the final version.

Validation Process:
The final version of the questionnaire was administered to medical students in their fifth and sixth year (the last two years of medical school in Brazil) at the Bahiana School of Medicine and Public Health from December 2016 to March 2017.The recommended sample size for exploratory factor analysis is of at least 10 participants per item in the questionnaire, or a minimum of 200 answered questionnaires 2 .A sample size of 250 answered questionnaires was estimated, considering that some of the responses could be incomplete.

Statistical Analysis
Descriptive analysis of student characteristics was performed, assessing mean and standard deviation of continuous variable and frequencies of discrete variables such as sex and the year of the student on the course.Students that completed less than 80% of the questionnaire items were excluded from the analysis.
For construct validation, we performed exploratory factor analysis (AFE) with orthogonal varimax rotation and extraction by principal components.We performed factorability of the correlation matrix using the Kaiser-Meyer-Olkin (KMO) test and the Bartlett's test of sphericity.Values acceptable for the KMO were > 0.500 and p < 0.05 for the Bartlett test 13 .
Parallel analysis was used to estimate the maximum number of factors by comparing the empirically obtained eigenvalues (from the latent root criterion) to the eigenvalues generated by random matrices 13 .
The factor solutions were tested, and the items with a factor loading of less than 0.3 or conceptual ambiguity were excluded.Composite reliability and Cronbach's alpha were calculated to assess the instrument's precision, acceptable values were considered above 0.7014 and 0.6115, respectively.The software used for analysis was SPSS v21.0 for Windows.

Ethical considerations:
This study was approved by the Ethical Board of Bahiana School of Medicine and Public Health, CAEE number 57164216.1.0000.5544.

Results
The result of the translation of the 13 items is shown in Chart 1.
A total of 500 questionnaires were sent to the medical students, and 275 students answered and returned them.Eight incomplete questionnaires were excluded, and the final sample used for the psychometric analysis was 267 questionnaires.One hundred and sixty-eight (62%) of the students were female, mean age 25±3,1 years old.The distribution of students according to semester were: 22.5% in the ninth semester, 27% were in the tenth semester, 28.8% were in the eleventh semester and 21.7% were in the 12th semester.
One of the requirements for factor analysis is to examine the factorability with the Kaiser-Meyer-Olkin (KMO) which was 0.68 and the sphericity Bartlett test that was significant at the level of p < 0.001.
The parallel analysis was consistent with the retention of a maximum of two factors.However, the twofactor solution obtained through the EFA resulted in a non-empirically sustainable distribution, because four of the items had factor loadings higher than 0.3 in both factors (suggestive of ambiguity).

Discussion
A considerable proportion of healthcare costs can be attributed to medical attitudes 16 , especially overuse and waste of available resources.Most physicians acknowledge the importance of being cost-conscious, but it is difficult for most of them to play an active role in cost containment practices 17 .
Educational interventions to change the behavior of physicians towards more cost-conscious attitudes can be instituted from medical school.A validated measuring instrument can be useful to better plan and assess such educational interventions.Although some researchers are currently working on this subject, we found a lack of validated scales to measure this phenomenon among medical students.
The inclusion criteria for this research was that the participants were medical students in their two final years of medical school.These students are closer to medical activities and practical examples from other physicians.The sample was considered adequate to go ahead with the analysis, as it was representative of students of each period 18 .
The results suggest that the one-factor solution represents the best structure to measure the phenomena.Eleven of the 13 items in the scale had factor loadings higher than 0.3 and remained in the final version.The items number 9 "the cost of a test or medication is only important if the patient has to pay for it out of their own pocket" and number 6 "physicians should provide tests or treatments if a patient requests them" were excluded due to low factor loadings.These two items might be related to some content interpretation bias.Item number 6, for example, can be mistaken with ignoring a patient's request.Even with the exclusion of these two items, it is considered that the content of the scale was not affected because the reduction corresponds to just 15% 19 .
The composite reliability and Cronbach alpha were used to verify the internal consistency of the factors.The Cronbach alpha was considered as substantial 15 .However, few authors report some restrictions to its use as an indicator of consistency.One of them is the possibility of the Cronbach alpha being highly influenced by test length 15 .Literature demonstrates that alpha may be acceptable even when there are recognized problems with the scales and sometimes it is inappropriately used to claim an instrument is unidimensional 20 .Therefore, the composite reliability index was used as it is recognized as being a more robust precision indicator than the Cronbach alpha 21 .The findings related to the scale reliability were not excellent but were considered acceptable.It is possible that the use of 4-point Likert scale could have reduced the scale reliability because in general, scales with few numbers of categories can result in lower variability and reliability 22 .However, it was considered important to maintain the same number of categories as in the original survey so as to facilitate future comparisons.
It is important to note some limitations of this study.Among them is the fact that the students were from a single medical school and selected by convenience.Confirmatory factor analysis could be used to increase the reliability of the findings The results obtained from this study present an important contribution to the subject of costconsciousness in medical education.It is recommended that the instrument developed in this study be used not only in research but also to help to identify problems in specific groups and to assess results after educational interventions.The continuous use of the instrument can also create opportunities to retest it for new groups and institutions, which can provide further evidence for its validity and reliability.

Conclusion
The results suggest that the questionnaire is of a satisfactory psychometric quality and is capable of measuring medical students' perceptions of costconsciousness.

Chart 1 .Chart 2 .
Items of the original Cost-Consciousness Perception questionnaire in English and their respective translation into Portuguese, Salvador, Brazil, 2017 For this reason, an alternate solution composed of just one factor was tested.In the one-factor solution, the items 6 and 9 had low factor loadings and were excluded from the final version of the scale.The factor loadings for the one-factor solution are shown in Chart 2. The 11-item scale underwent a composite reliability test with a value of 0.77 and Cronbach alpha of 0.66.Inter J H Educ, Salvador, 2018 October;2(1):66-72 Doi: 10.17267/2594-7907ijhe.v2i1.1968| ISSN 2594-7907 One-factor solution for the Cost-Consciousness Perception Scale, Salvador, Brazil, 2017