Knowledge, attitudes and practices relating to antibiotic use among community members of the Rupandehi District in Nepal

A cross sectional quantitative survey of community members was conducted in the Rupandehi district of Nepal. At the time of developing the study design for this research, the administrative re-structure of Nepal had not been fully implemented. As per the earlier structure, Nepal was divided into five developmental regions and subdivided into 75 districts [40]. The districts were further divided into village development committees (VDCs) and municipalities, which were divided into wards as the basic administrative units. Districts are spread across three geographic regions, high-hill, hill and low-land, with approximately half of the population living in low-land regions [41].

According to the 2011 census, the total population of Nepal was 26,494,504 with 3.3% (880,196) of the population living in the Rupandehi district [41]. Almost two-thirds (66.0%, 580,688) were adults. The district is situated towards the central southern part of the country. As per the earlier structure, the district was divided into six municipalities and 42 VDCs. Municipalities and VDCs were aggregated in seven electoral areas [42].

We used guidelines developed by the World Health Organization (WHO) [43, 44] for the selection of households. Following these guidelines, public health facilities were selected as the basis of household selection. We chose six of the seven electoral areas to survey. Two were purposively selected: one that includes the largest hospital in the district and the other was the area with the lowest socio-economic status. An additional four areas were selected randomly from the remaining five electoral areas as recommended by the guidelines. One public health facility was selected from each survey area, consisting of all types of health facilities i.e. two each of hospitals (the largest as discussed above plus another), primary health care centres and health posts. The additional hospital plus other facilities were randomly selected. The VDCs and municipalities in which the public health facilities were located were used as the sampling area for selection of households. Of the total of six areas selected, four consisted of VDCs and two were municipalities.

A cluster sampling technique was applied to identify households to survey within the selected survey areas. Based on the WHO manual [45], we identified 20 clusters from the selected municipalities and VDCs. The smallest administrative unit, the “ward”, was considered as a cluster. Four clusters per municipality and three clusters per VDC were selected randomly. The sample size of 220 was based on an estimated prevalence of 33.7% of the population lacking knowledge on antibiotics and their role [4], a 95% confidence interval, a precision effect of 10%, a design effect of two to account for heterogeneity between clusters and an adjustment of 25% to allow for non-response [12, 46].

A list of households in each cluster was obtained from the records of respective municipalities and VDC offices. This list was verified after visiting each cluster and updated by deleting any duplicate households and adding any households missing from the records. Using the updated list of households in each cluster, an equal number of subjects (eleven) was selected from each cluster applying simple random sampling techniques.

The head of household was the preferred respondents for the study. However, if the head of household was absent at the time of interview, the most senior member of the household, who was 18 years and older, was interviewed.

A structured questionnaire was developed by adapting related questionnaires including one from the United States Agency for International Development (USAID) module “Antimicrobial resistance module for population-based surveys” [47] and those used in previous studies [48, 49] (Additional file 1).

A set of questionnaires was pre-tested with 30 respondents in urban and rural areas of the Nawalparasi district, Nepal (a neighboring district of Rupandehi), to ensure the cultural appropriateness, any problems with question wording, layout and understanding or a respondent’s reaction. As a result, minor adjustments were made to the final questionnaire based on the pre-test results. With a few people not knowing what the word “antibiotics” was, the questionnaire was amended to ask if they had heard of widely used antibiotics such as penicillin or metronidazole before being asked the main questions. Similarly a few respondents were unsure of the difference between “good” and “bad” bacteria present in our bodies so this difference was explained before they answered the question. Following explanation, issues with language did not appear to cause ambiguity that might impact on interpretation of the survey and the ensuing results. The final questionnaire included twelve questions relating to knowledge, eight questions to attitudes and six questions to practices. The reliability coefficient of responses to the final questionnaire was calculated using the Cronbach’s alpha score with the following results recorded: knowledge (0.63), attitudes (0.65) and practices (0.67).

The questionnaire comprised four sections: socio-demographic characteristics of respondents and a section on each of knowledge, attitudes and practices relating to antibiotics and their use. Questions about knowledge were divided into four domains, namely “identification of antibiotics” (Q1-Q3), “knowledge on the role of antibiotics” (Q4-Q6), “side-effects of antibiotics” (Q7-Q9) and “antibiotic resistance” (Q10-Q12). The questions on attitudes were divided into three domains: “preference for use of antibiotics” (Q13-Q15), “antibiotic resistance and safety” (Q16-Q18), and “attitudes to doctor’s prescribing of antibiotics” (Q19-Q20). The six questions relating to practices (Q21-Q26) were not divided into domains. The English version of the questionnaire was translated into Nepali and back translated into English to ensure the accuracy of the translated text.

Interviews were conducted in the Nepali language by two trained research assistants from September 2017 to December 2017. The training of research assistants covered the objectives of the study and familiarising them with the data collection techniques. A flow chart for the recruitment of respondents and consent process was provided to the research assistants and used in the data collection process. The average duration per interview was 20 min. Ten households were replaced in the original sample due to refusal to participate (n = 7) and no one at home at the time of interview (n = 3).

All respondents were informed of the nature of the study and written consent was sought to interviews being conducted. The study was approved by the Human Research Ethics Committee, Curtin University (HRE2017–0394) and the ethics committee of the Nepal Health Research Council (Reg. no.189/2017).

Data were collected via paper-based questionnaires and the data were entered and analysed using the Statistical Package for the Social Sciences (SPSS) version 25.0 for Windows (IBM Corp., Armonk, NY, USA).

Demographic variables and responses to the knowledge, attitudes and practices questions were analysed using descriptive statistics. Responses to the five-point Likert scale for the knowledge and attitudes questions were combined into three groups: ‘strongly agree’ and ‘agree’, ‘strongly disagree’ and ‘disagree’, and ‘uncertain’. The three groups are referred to as “Yes”, “No” and “Don’t know”, respectively [50]. Questions relating to practices were assessed using the five-point Likert scales scoring scheme of ‘never’, ‘seldom’, ‘sometimes’, ‘often’ and ‘always’.

Regression analysis was conducted to identify demographic factors associated with knowledge, attitudes and practices. Responses to the knowledge and attitudes questions were given a score of “1” for a correct response and “0” for an incorrect or uncertain response, and scores summed for respondents across each of the domains. For the practices questions, responses were given a score based on the five-point Likert scale, ranging from “5” for the most appropriate answer to “1” for the least appropriate answer, and summed. The median score based on responses to questions in each of the knowledge, attitudes and practices sections was used as the cut-off to dichotomize the continuous variable for use as the dependent variable in multiple logistic regression analysis. Respondents scoring higher than the median were assessed as having “better knowledge”, “more appropriate attitudes” and “better practices” relating to antibiotic use [51]. The significance level (α) was set at 0.05 for all statistical tests. Spearman’s rank order correlation coefficient was used to describe the strength and direction of the relationship between responses to the knowledge, attitudes and practices questions.

The sample consisted of 220 households (Table 1), with a response rate of 97% (n = 210). Compared to the adult population of the Rupandehi district, the sample included slightly more females (54% vs 52%) and respondents from rural areas (60% vs 51%). The mean age was 38.5 years (SD 11.5). Most respondents had achieved a level of education of primary/secondary school level (31.4%) or high school/intermediate level (30.0%). The mean monthly income of respondents was Nepalese Rupees (NPR) 42,491 (SD 16,835), compared with an estimated average monthly household income for Nepal of NPR 30,121 in 2015 [52].

Respondents had relatively good knowledge about three of the four knowledge domains: “knowledge about the role of antibiotics (Q4-Q6)”, “side-effects of antibiotics (Q7-Q9)”, and “antibiotic resistance (Q10-Q12)” (Fig. 1). While the majority of responses to questions in the three domains were correct, for questions on “side-effects of antibiotics” and “antibiotic resistance” a relatively high percentage of responses to five of the six questions fell in the “don’t know” category (16–27%). Statements for which the “don’t know” response was highest included that “antibiotics can cause secondary infections after killing good bacteria present in our bodies” (25.0%), “many infections becoming increasingly resistant to treatment by antibiotics” (25.0%) and “misuse of antibiotics leading to antibacterial resistance” (27.7%).

Respondents had relatively less knowledge in regard to “identification of antibiotics (Q1-Q3)”. More than two-thirds (67.7%) did not answer correctly to the question that “amoxicillin is an antibiotic” and nearly one-third (32.7%) did not know that “antacid is not an antibiotic”. However, most respondents (94.1 and 84.1% respectively) answered correctly that “antibiotics are useful for killing germs” and “antibiotics are not often needed for cold and flu illness” while more than two-thirds (71.5%) knew paracetamol was not an antibiotic.

The level of knowledge about antibiotics use was better for respondents who lived in urban compared to rural areas (X² = 16.257, p = < 0.001), for younger respondents (X² = 30.696, p = < 0.001) and those with higher levels of education (X² = 72.264, p = < 0.001) (Table 2).

Responses to questions about attitudes to antibiotics reflected varying points of view (Fig. 2). In terms of having a “preference for use of antibiotics (Q13-Q15)”, most respondents were aware that they did not need to take antibiotics for a cold to prevent them getting a more serious illness (77.3%) and did not want to take an antibiotic if they did not need one (78.2%). However, almost half (47.7%) thought antibiotics would help them to get better more quickly if they had a fever.

In the domain of “antibiotic resistance and safety (Q16-Q18)”, half of respondents (50.9%) were uncertain if skipping doses would not contribute to the development of antibiotic resistance and almost one fifth (17.3%) were uncertain about if taking an antibiotic contributed to the development of antibiotic resistance. Most respondents (81.8%) agreed antibiotics should not be commonly used.

Attitudes to prescribing of antibiotics by doctors were somewhat ambivalent. Most respondents (61.8%) were not less satisfied with a doctor’s visit if they did not receive an antibiotic; however, the majority (88.2%) indicated if they were not prescribed an antibiotic when they thought one was needed, they would go to another doctor.

Attitudes to antibiotic use was significantly associated with areas of residence (X² = 5.197, p = 0.023) and education level (X² = 27.306, p = < 0.001) (Table 2). Respondents living in urban areas and those with higher levels of education had more appropriate attitudes than those living in rural areas and those with lower levels of education.

In responding to questions about practices the majority always or often consulted a doctor before starting an antibiotic (94.5%), checked the expiry date of antibiotics before using them (85.8%) and completed the full course of treatment (81.3%), and never or seldom preferred to obtain antibiotics from the pharmacy (76.8%). However, in contrast to good practice reflected with these questions, the majority (84.6%) reported at least sometimes preferring to take an antibiotic when they have a cough or sore throat and almost a third (31.8%) to using antibiotics as a prophylaxis (Fig. 3).

Practices in relation to antibiotic use were significantly associated with gender (X² = 5.984; p = 0.014), area of residence (X² = 13.996, p = < 0.001), age group (X² = 17.921, p = 0.001) and education level (X² = 42.452; p = < 0.001) (Table 2). Female respondents, those who lived in urban areas, were younger and those with a higher level education reported better practices in regard to antibiotic use compared to males, respondents living in rural areas, older respondents and those with lower levels of education.

In multivariable logistic regression analysis (Table 3), after adjusting for other variables, education level was found to be significantly associated with each of knowledge, attitudes and practices, with respondents with a level of education of Bachelor degrees and above having better knowledge, more appropriate attitudes and better practices. Area of residence was significantly associated with better knowledge on antibiotics use, with rural respondents being less likely to have better knowledge compared to urban residents, and females being more likely to have better practices than males.

Spearman rank order correlation revealed a positive association between each pair of the knowledge, attitude and practice scores for respondents (p = < 0.001) (Table 4). The correlation was good between knowledge-attitudes and fair between knowledge-practices and attitudes-practices [54].

However, comparing responses to questions in different domains, a few inconsistencies were noted. For example, although most respondents (84.1%) correctly answered the question in the knowledge section that antibiotics are not often needed for cold and flu illness (Fig. 1), in the practices section the majority (84.1%), answered that sometimes or often they preferred to take an antibiotic when they had a cough or sore throat (Fig. 3). Another example was respondents seemingly having good knowledge about antibiotic resistance (correct answers of between 70.0 to 74.5% for relevant questions) (Fig. 1), however more than one-third (35.0%) were less satisfied with a doctor’s visit if they did not receive an antibiotic and the majority (88.2%) would go to another doctor if a doctor did not prescribe an antibiotic when one was needed (Fig. 2).