Introduction

Antibacterial agents are used to treat a wide range of bacterial infections and are essential lifesaving medicines. They are the most commonly used medicines in Sub-Saharan Africa due to the high prevalence of infectious diseases [1]. Used correctly, they deliver enormous benefits to the health of the population worldwide [2].

Antibacterials are, according to the national drug policy of Uganda, prescription only medicines [3]. However, they are readily accessible and affordable to most patients within the communities in Uganda, not only as prescription medicines as they can often also be obtained over-the-counter especially in private medicine outlets [4]. The relative ease with which communities access these medicines poses several challenges for antibacterial stewardship [4]. The majority of caregivers in low-income countries initiate treatment of their children at home [5]. The use of antibacterials prior to hospital visits is common, especially in low-income countries, and may influence patient treatment outcomes. According to a study in Nigeria, 85% of patients reported to have self-medicated before coming to the health facility and antibacterials were among the most common medicines used [6]. A study in Uganda reported that 62.2% of patients had used antibacterial agents prior to coming to health facility [4]. Another study done in Haiti to assess self-medication among patients presenting at an out-patient department found that 45.5% practiced self-medication with antibacterials [7].

Caregivers’ ability to report antibacterial intake prior to coming to a health facility is crucial for appropriate prescription of medicines at the health facility. Self-reports have been shown to have low validity as they are prone to recall bias and social desirability bias. Respondents normally provide information that conforms to their perceived expectations of the health workers or researchers [5, 8]. A study carried out in Uganda in 2009 reported a limited validity of caregivers’ reports of use of sulfamethoxazole, chloroquine and sulfadoxine in their children prior to arrival to the hospital [5]. Similarly, a study from Tanzania reported that 97% of the children without history of prior chloroquine treatment had detectable levels of chloroquine in blood [9]. Another study in Ghana reported a high prevalence (64%) of antibacterials detected in urine samples of patients compared to the self-reported use (13%) [10].

To our knowledge no study has validated caregivers’ reports of intake of antibacterials in children under five years in rural communities in low resource settings. In this study we describe and validate caregivers’ reported use of antibacterials by their children under five years for treatment prior to seeking care at the healthcare facility.

Materials and methods

Results

Socio-demographic characteristics of the caregivers and children under five years

Of the 385 sampled children, 355 (92.2%) had data on both caregiver’s report on antibacterial use prior to health facility visit and results from urine and blood analysis and were thus included in the analysis. The 30 (7.8%) observations were dropped because they were missing blood analysis data. Over half (53.2%, n = 189) of the children were female. The median age of the children was 29 (IQR: 16–46) months. The majority (96.1%, n = 341) of the caregivers were female. The median age of the caregivers was 25 (IQR: 21–31) years. About half (53.2%, n = 189) of the children attended a healthcare facility located in a rural area. (Table 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Socio-demographic characteristics and prevalence of antibacterial use in children under five years prior to health facility visit as reported by caregivers of children under five years in rural communities of Gulu district, northern Uganda (August, 2019).

https://doi.org/10.1371/journal.pone.0257328.t001

Prevalence of antibacterials detected in blood and urine samples of children under five years

Of the 355 children under five years who were included in the analysis, 226 (63.7%, CI: 58.4–68.7) had detectable levels of antibacterials in urine or blood in the samples taken upon arrival to the healthcare facility (Table 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Prevalence of antibacterials detected in blood or urine samples of children under five years in rural communities of Gulu district, northern Uganda (August, 2019).

https://doi.org/10.1371/journal.pone.0257328.t002

Most commonly used antibacterials.

The most commonly used antibacterials as reported by the care givers were amoxicillin (6.2%, 22/355), cotrimoxazole (2.8%, 10/355), and metronidazole (2.3%, 8/355). The most common antibacterials detected from the laboratory analysis were cotrimoxazole (29%, 103/355), ciprofloxacin (13%, 46/355), and metronidazole (9.9%, 35/355) (Fig 1)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Commonly used antibacterials according to the laboratory analysis.

https://doi.org/10.1371/journal.pone.0257328.g001

Validity of caregivers’ reports of antibacterial intake in children under five years

The sensitivity, specificity, PPV, NPV, agreement and kappa coefficient of the caregivers’ reports of use of antibacterials for treatment of children prior to healthcare facilities visit were 17.3% (12.6–22.8), 90.7% (84.3–95.1), 76.5% (62.5–87.2), 38.5% (33.0–44.2), 43.9% (38.7–49.3%) and 0.06 (0.01–0.12) respectively. The sensitivity, specificity, PPV,NPV, agreement and kappa coefficient varied between the different antibacterials (see Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Validity of caregivers’ reports of antibacterial intake in children under five years in rural communities of Gulu district, northern Uganda (August, 2019).

https://doi.org/10.1371/journal.pone.0257328.t003

Discussion

In this study we demonstrated that the prevalence of antibacterial use prior to health facility visit was high and that caregivers under reported the use of antibacterials in the children under five years prior to coming to the health facility. Antibacterial use prior to healthcare facility visit is a common practice in many resource limited settings globally. Caregivers’ ability to report antibacterial use before coming to the health facility is crucial for appropriate prescription of antibacterial upon reaching health facilities [5]. Appropriate prescription of antibacterials is important because it reduces the emergence of antibacterial resistance, poor clinical outcomes, increased mortality and wastage of financial resources [19].

In the current study, almost two thirds (63.7%) of the samples (blood and/or urine) tested positive for antibacterials. This implies that the prevalence of antibacterial use prior to health facility visit is much higher than what was self-reported (14.4%). This finding is similar to those from other low and middle income countries (LMIC) [4, 5, 10], a study carried out in Ghana reported a prevalence of self-reported antibacterial use prior to health facility visit of 13%, however, analysis of urine samples reported a much higher prevalence of 64% [10]. In Uganda, self-medication with antibacterials is a common practice [1, 4, 20] which is reflected in the high prevalence of antibacterials found in the samples (blood and/or urine) in the current study [1]. Another reason for the high prevalence of antibacterial use in our study is the high prevalence of infectious diseases in these communities. In Uganda, 71% of children under five years attending healthcare facilities do so due to acute respiratory infections [21], however, in the community in this study, the most common diseases in children under five years seeking care at healthcare facilities include; malaria, diarrhea, pneumonia, acute childhood malnutrition and HIV/AIDS [15–17]. High prevalence of antibacterials found in the samples in the current study could also be due to exposure to antibacterials through consumption of water, vegetables and animal products [22, 23]. In the Hong Kong survey to determine the presence of veterinary antibiotics in food, drinking water, and the urine of preschool children, it was found that 13 veterinary antibiotics were detectable in the urine of 77.4% of primary school children with norfloxacin and penicillin having the highest detection rates. Enrofloxacin, penicillin, and erythromycin were the most detected veterinary antibiotics in raw and cooked food [21]. Studies in Uganda, report a high prevalence of veterinary use of antibacterials. The most commonly used antibacterials in veterinary medicine in Uganda include; procaine penicillin, trimethoprim/sulfadiazine, erythromycin sulphate, tylosin tartrate, oxytetracycline hydrochloride [24, 25]. This high prevalence of antibacterial use can lead to increased risk of resistance within the community [26]. A study was carried out in Uganda to determine the epidemiology and antibiotic susceptibility of Vibrio cholerae associated with the 2017 outbreak in Kasese district, and it reported that V. cholerae was highly resistant to the commonly used antibiotics [27].

Most caregivers reported to have given their children amoxicillin, cotrimoxazole and metronidazole. This is consistent with reports from a study in northern Uganda where metronidazole, amoxicillin, ciprofloxacin, doxycycline or cotrimoxazole were reported as the most commonly used antibacterials by patients prior to hospital visit [4]. Metronidazole is commonly used for bacterial gastroenteritis, amoxicillin is used for bacterial chest infections, and cotrimoxazole is used to treat pneumonia, bronchitis, infections of the urinary tract, ears intestines and as prophylaxis against opportunistic infections in HIV [28, 29]. In our study the most commonly detected antibacterials in the laboratory analysis results were cotrimoxazole, ciprofloxacin and metronidazole, similar to findings from a study carried out in Ghana which reported ciprofloxacin, trimethoprim or metronidazole as the most common antibacterials detected in urine samples [30]. Ciprofloxacin is commonly used to treat pneumonia, typhoid fever, infectious diarrhea, skin and bone infections [28, 29]. Amoxicillin was the most commonly reported antibacterial used and yet it was not among the most commonly detected antibacterials from the laboratory analysis. This could be explained by the pharmacokinetics of amoxicillin, which has a very short half-life of about 1 hour and will usually be out of the system within 5 hours. Thus, meaning that for it to be detected in the blood or urine samples, it should have been taken within a few hours before healthcare facility visit [28, 29]. We also observed that the number of children who had cotrimoxazole in their biological samples was higher than those who reported the use. It is possible that some of these children may have tested positive for cotrimoxazole since they could have been receiving it as prophylaxis against opportunistic infections in HIV [31, 32]. The prevalence of HIV/AIDS in northern Uganda as of 2019 when data for this study was collected, was 7.2% in adults and 0.5% in children under five years [33]. Since we were interested in antibacterial use for current illness, for which the children were brought to the healthcare facility, caregivers might not have found it not necessary to report the use of cotrimoxazole as prophylaxis against opportunistic infections in HIV.

The positive predicative value we found for reported use of antibacterials is not high enough to allow caregivers reports to guide treatment. The high specificity values indicate under reporting but the negative predictive value indicate that many children were given drugs that were not reported by caregivers. This study was carried out in rural communities of Gulu district in Uganda where the adult literacy levels are low [1, 20], and the inconsistencies in caregivers’ response to interview questions and laboratory findings, could have been because of caregivers inability to identify medicines taken as antibacterials. Another reason for the inconsistencies in self-reported antibacterial use and laboratory findings could have been due to social desirability bias [34]. The caregivers could have been aware that self-medication is not a good practice, and therefore feared to tell the interviewers the truth. Another reason for the inconsistencies could have been due to consumption of these medicines from diffuse sources such as milk, water or food, studies in Uganda have reported veterinary use of antibacterials [24, 25]. Another worrying explanation for the inconsistencies could be the quality of antibacterial medicines, some of these antibacterials may not contain the actual quantity of the active medicine the manufacturers claim they contain. Although we did not set out to study the quality of antibacterials in this study, high prevalence of substandard antibacterial medicines has been previously reported in developing countries [35]. Furthermore, inaccuracies in self-reports may lead to duplication of therapy, incorrect management of the ill child, failure to appreciate non-compliance leading to exacerbation of chronic medical conditions, or inaccurate research conclusions [36].

We observed a strong association between high self-reported prior antibacterial use and the source of antibacterials being from public health facilities. This could be attributed to the low financial status of the people in these communities [1] forcing them to seek free healthcare from public healthcare facilities. The district-based healthcare system in Uganda consists of level I, II, III, IV and district hospitals [11]. This therefore means that by the time these children were brought to health care level III and IV, they could have already sought care from the lower levels and were referred to these higher levels for further management.

There is need for further research to understand the reasons for caregivers’ poor reports on their children’s prior intake of antibacterials before coming to the health facility. Improved validity could be promoted by encouraging health care workers to carefully explain to the caregivers the medicines they administer to these children when they fall sick. Proper documentation of the medicines given to these children when they are sick could also improve the validity of self-reported medicine use. There is need for the healthcare workers to educate the caregivers about the dangers of using antibacterials without consulting a healthcare worker, and also further research is required to better understand why caregivers initiate antibacterial use at home without consulting a healthcare service provider. This all will allow policy makers to be better informed when planning interventions to reduce the large amount of incorrect antibacterial use in the community.

The results of our study should be considered in light of some limitations. This study could have been affected by recall bias, where antibacterials given may have been forgotten. The study could have also been affected by social desirability bias since the study was carried out in a hospital setting and probably caregivers feared telling the truth because they thought it could affect patient care. Under reporting could have been affected by how the questions were understood by the caregivers. Failure to detect some of the antibacterials in the samples could have been due to the pharmacokinetics of the antibacterials. Factors such as education level of the caregivers could have contributed to the under reporting of antibacterial use prior to healthcare facility visit, however, we didn’t collect this information. This is because adult literacy levels in this community are low [1, 20] and to our knowledge previous studies have not reported associations between self-report and education level [10, 37]. However, there is need for further research to determine if there’s an association between caregivers’ education level and reporting of prior antibacterial use in this setting. The discrepancy between the reported use and the detected antibacterials in blood/urine samples could have also been because the antibacterial could have been given for the management of another condition, such as cotrimoxazole for prophylaxis against HIV related opportunistic infections, but we did not collect this information. Our tool was designed to capture only antibacterial use for the illness for which the children were brought to the healthcare facility. We were unable to report the levels of antibacterials in relation to how far back the antibacterials were taken, this is because we used a qualitative LC-MS/MS method which was developed to report the presence or absence of antibacterials and not to quantify them.

Conclusion

A high proportion of children under five years take antibacterials prior to visiting a healthcare facility in northern Uganda. However, there is low validity of caregivers’ reports on prior intake of antibacterials by these children. There is need for further research to understand the factors associated with under reporting of prior antibacterial medicine use by caregivers of children under five years. In addition, we suggest that health care workers should endeavor to explain the role and names of medicines during dispensing, as well as the importance of reporting correctly on prior medication intake. There is also need to educate the caregivers about the dangers of using antibacterials without consulting a healthcare worker, and also further research is required to better understand why caregivers initiate antibacterial use at home without consulting a healthcare service provider. This all will allow policy makers to be better informed when planning interventions to reduce the large amount of incorrect antibacterial use in the community.

Supporting information

Acknowledgments

We appreciate the tireless effort of the data collection team; Apio Patricia, Ojok Albert, Kagood Francis, Hassan Chakaal and the village local chair persons for their guidance.