INTRODUCTION
The World Health Organization (WHO) encourages all sectors of society to help solve the problem of antibiotic resistance.1 It has been estimated that about 10% of antibiotics are prescribed in dentistry, not always in the right way.2-4 In dentistry, antibiotics help control bacterial infections, without replacing dental intervention which is essential to eliminate the focus of infection (e.g. pus from acute abscesses or necrotic pulp). The election of an antibiotic is based on host-related factors (e.g. age, immune status, or allergic history), pharmacological factors (e.g. spectrum of action of the antibiotics, pharmacokinetics, possible pharmacological interactions, etc.) and microbiological factors (e.g. types of bacteria present). Regarding microbiological factors, practitioners make their prescriptions according to possible bacteria that may be causing the infection, bearing in mind that most come from the oral cavity, which is polymicrobian in nature.5,6 In addition, antibiotics are prescribed as a prophylactic measure in dental procedures where there is a risk of endocarditis or other specific conditions.7
In Colombia, Holguin et al, 2013 conducted a study in clinical histories of patients with oral surgery procedures (open method exodontics, closed method exodontics and soft tissue biopsies), showing that only 10.26% of prescriptions or “no” antibiotics as a prophylactic measure were relevant. In addition, they found that 83.8% of cases had over-prescription, showing the existence of antibiotic abuse with no validated indications.8 It is therefore critical to know how dentists are prescribing antibiotics in Colombia, and it is necessary to produce validated and reliable instruments for this purpose. The development and validation of an instrument requires a rigorous process consisting of the search for scientific information, the formation of a focal group, the definition of the construct to be measured, the composition and number of items, and the definition, arrangement and coding of responses.9 In addition, it is also important to submit the instrument to a content validation process with a group of experts, validating the construct using psychometric indicators such as biserial index, discrimination, non-response, and difficulty, in addition to measuring the test’s internal consistency, in order to determine whether it has the ability to evaluate all the dimensions one wants to measure. The aim of the present study was to develop and validate an instrument to determine knowledge, attitudes, and practices (KAP) on antibiotics prescription in the field of dentistry.
METHODS
A qualitative and quantitative study was conducted during the period 2016-2017 in Villavicencio and other municipalities of the department of Meta (Colombia). Data was collected by providing an informed consent to participating dentists, who were selected according to a number of criteria, like being registered with the Secretary of Health. This study was approved by the Bioethics Sub- Committee of the Universidad Cooperativa de Colombia. The study consisted of 5 phases:
Phase 1. Questionnaire construction and focus group
Version 1.0 questionnaire (Figure 1) was created based on scientific literature. This version was analyzed and discussed in a focus group consisting of general and specialized dentists (oral surgery, pediatric dentistry, rehabilitation, pharmacology) and one microbiologist.
Phase 2. Validity of content by expert panel
Content validity by an expert panel is carried out to analyze the items that make up an instrument. One of the criteria for the selection of experts was to be familiar with the prescription of antibiotics in the field of dentistry. Therefore, three general dentists and five dentists specializing in areas such as endodontics, oral surgery, implantology, oral rehabilitation and periodontics were invited, as well as a biochemist (chosen for his knowledge on the basics of antibiotics) and a psychologist (for her experience in the construction of questions). The experts assessed the questionnaire individually, using a form with a standard rating measure of 1 to 5 for the categories of adequacy, clarity, consistency, and relevance. This information determined the concordance among them, using Kendall’s W statistical test at 95% confidence. In addition, they made recommendations on the formulation, terminology and relevance of the instrument.
Phase 3. Pilot test
To confirm the understanding of version 3.0 questionnaire, a test was applied to 7 dentists from the municipality of Acacias, Meta, who did not belong to the study population. The dentists were invited to answer the questionnaire and comment on whether they had difficulties or recommendations.
Phase 4. Quantitative validation
In this phase, the survey was applied to 50 dentists practicing the profession in Villavicencio and 48 in other municipalities of Meta, Colombia. The number of dentists was calculated according to the number of sections of the designed questionnaire, and data collection was done prior informed consent by the participating dentists. The practitioners were invited by visiting their offices in the municipalities of Acacias, Barranca de Upia, Cabuyaro, Cumaral, El Castillo, El Dorado, Guamal, Granada, San Martín, San Luis de Cubarral, Puerto Gaitán y Puerto Lopez, Restrepo. As inclusion criteria, the dentists had to be registered in the Secretary of Health of Meta and have a ReTHUS record. In Villavicencio they were selected randomly, using a list provided by the Secretary of Health. With the collected information and since 98% of the questionnaire contained dichotomic items, the test’s reliability analysis was conducted using the Kuder Richardson’s formula. In addition, the psychometric indices analysis was conducted, including item-total correlation (biserial correlation), discrimination, difficulty index, non-response and unidimensional rates. The criterion for determining unidimensionality was the correlation between the full test and its half, using Pearson’s Correlation Coefficient.
RESULTS
Phase 1. Questionnaire construction and focus group
The focus group showed that most items were relevant to determine how antibiotic prescription is being performed, and specific suggestions were made (Figure 1). The topics covered in the focus group included usefulness of antibiotics, the association of resistance and antibiotic prescription, clinical cases where antibiotics are prescribed, non-clinical factors, and prescription duration. One of the most discussed aspects in the focus group was the issue related to clinical cases and medical conditions in which antibiotics should be prescribed.
Phase 2. Validity of content by expert panel
A good level of agreement among experts was detected in version 2.0 (Kendalle’s: Sufficiency: 0.68, Concordance 0.69, Relevance: 0.72 and Clarity: 0.81) (Table 1), showing an appropriate inter-rater agreement to measure the reliability or reproducibility of the questionnaire. There where recommendations during this assessment, as shown in Figure 1.
Phase 3. Pilot test
During the pilot test conducted with version 3.0 of the questionnaire, dentists stated that the survey was unappealing and too long (Figure 1). Therefore, some of the items in the Attitudes and Practices sections were organized into categories in order to shorten these sections without removing the questions. Finally, with the help of a graphic designer, the instrument’s appearance was improved, resulting in version 4.0.
Phase 4. Quantitative validation of questionnaire
Regarding the characteristics of dentists participating in this phase, 69.7% were male, 59.2% practiced privately and 47.9% had more than 10 years of experience. In addition, 35.5% of dentists had a specialization, with 16.8% in oral rehabilitation, 12.6% in oral surgery (n = 3) and 12.6% (n = 3) in health service management audit.
Internal reliability
In terms of questionnaire reliability, the original questionnaire with 45 items showed an internal consistency of 0.63 (Knowledge: KR20 = 0.44, Attitudes: KR20 = 0.42 and Practices: KR20=0.42 (Table 2a). However, after removing items (n = 9) not complying with the psychometric indicators as explained below, there was an increase in Richardson’s K value (= 0.81), so the instrument’s internal consistency was considered high. The following values were obtained in each section: Knowledge: KR20 = 0.67, Attitudes KR20 = 0.64, Practices KR20 = 0.71 (Table 2b).
Psychometric indexes of items
In version 4.0 of the questionnaire, the validity of questions in the item-total correlation coefficient was “very good” to “acceptable” for most questions (86.6%), but six items (13.3%) did not comply with this index (Table 3a). In the discrimination rate, most items (90.8%) were under the “acceptable” category and 11.1% under “remove” as they were not among the ranges of the index (Table 3b). Regarding non-response rate, the items are suitable for the questionnaire as their values ranged from 0 to 0.15 (Table 3c), and in terms of difficulty index, 52.3% of items (n = 23) are very easy for respondents while 6.8% (n = 3) are very difficult (Table 3d).
Version 4.0 of the questionnaire was therefore modified, eliminating questions that were not among the expected values, and when recalculating the psychometric indices of version 5.0 of the questionnaire, the biserial index showed that 100% of the questions were “very good” to “acceptable”. In the discrimination rate, 100% of questions were in some classification to be accepted (very high, high, moderate, and very low).
In terms of non-response rate, as in version 4.0, all items are adequate, and the difficulty index showed that 55.5% of questions ranked as very easy, while the rest (44.5%) were classified as easy, intermediate, difficult and very difficult.
Regarding the unidimensional analysis of the final questionnaire, significance was 0.93 showing that the test is homogeneous for both the full test and its half (Table 4).
Phase 5. Scale and rating
The scale was constructed using the per- unit index, which corresponds to the test’s percentages of success. This scale ranges from 0 to 1. Finally, to qualify the test, the percentile classification was used in each of the three dimensions (KAP). To this end, the number of items correctly answered was totaled, calculating percentiles 33 and 67 in order to create three rating levels: low, medium and high, based on the right items.
Final questionnaire
The final questionnaire was thus validated to assess knowledge with 6 measurable items on the usefulness and resistance of antibiotics. Attitudes are assessed by 2 categories with 8 measurable items on clinical and non-clinical factors to prescribe antibiotics, and practices are validated by 2 categories with 22 measurable items on clinical cases for prescription and duration. In addition, the questionnaire has a section with five questions about dentists’ general information and a section of “Additional questions” with 6 items about the preference for an antibiotic, the number of patients prescribed in the last week and the interest in updates on the topic (Table 5). The results found during the validation phase show that some dentists consider that antibiotics are to treat virus (11%) or fungal infections (20%), 46.6% do not know exactly what antibiotic resistance is, 56.6% do not consider that antibiotic resistance results from the prescription of these medications, 74.4% state that antibiotic use is common in dental practice and 85.5% use amoxicillin as the first-choice antibiotic.
Table 5
Content of final validated questionnaire
DISCUSSION
Several steps were taken to validate the questionnaire in this study, starting with the construction of the instrument based on the scientific literature. This stage showed that many studies in other countries demonstrated an inadequate prescription of antibiotics but the instruments they used were not subjected to a rigorous validation process..(13-15
Once the instrument was developed, a focus group was held to discuss the questions in it. During the focus group discussion, there were suggestions to eliminate or add questions, improve the wording, or specify and relocate some responses. There were also additional reactions stemming from the participants’ experience, which helped improve the instrument, achieving the purpose of this process. As Escobar and Bonilla-Jiménez pointed out in 2009,16 the main purpose of focus groups is to have participants express attitudes, feelings, beliefs, experiences and reactions regarding the topic and instrument developed. During the focus group in the present study, a great deal of the discussion was about the cases in which antibiotics should be used, demonstrating lack of clarity in this regard due to inconsistent guidelines and protocols.17 Several authors express the need for additional studies to establish protocols for the use of antibiotics.18
During content validity by experts, like in the rigorous validation process of a questionnaire to measure knowledge and attitudes on antibiotics prescription in physicians by López-Vázquez et al, 201610 and Teixeira et al, 2016,11 the expert panel eliminated, added and reformulated questions, in addition to correcting grammar, syntax, organization, and logical sequence.
Later, during the pilot test, participating dentists made observations that helped improve the quality and acceptability of the instrument even more, thus confirming the importance of this phase. Lancaster et al claim that pilot tests ensure that questionnaires are understandable and appropriate, and that the questions are well defined, clearly understood and consistently presented.21
In the present study, reliability was assessed while the test was being applied to respondents, showing low internal consistency in questionnaire version 4.0 (0.63); it also showed that some questions had to be removed. Some items detected in the psychometric test were then removed, showing an increase in internal consistency to 0.81. As in other validation studies, this question elimination process improved questionnaire reliability.10,11 Regarding unidimensionality, the questionnaire proved to be homogeneous, both in full and in half, with a value close to 1 (0.93).
This questionnaire has various strengths as shown by the validation process, with an adequate number of questions that facilitate participation, and a simple scale to grade the adequate items. Finally, it should be noted that, while all the indicators of the various analyses were adequate, a higher quality questionnaire could be achieved with higher reliability and homogeneity values. However, it is important to note that antibiotic prescription is a complex issue due to the various factors that may influence prescription,22 such as lack of clarity of protocols, fear of complications following a dental procedure, and lack of updates on the topic.
This qualitative and quantitative process offers a simple instrument that was developed and validated to assess knowledge, attitudes and practices on antibiotics prescription by dentists. This will help us understand how prescription is done and develop interventions to improve this activity and make a more rational use of antibiotics.