Introduction
Antibiotic resistance (ABR) is recognised as one of the biggest global challenges for future human health and development, as shown by the creation of global action plans by international institutions such as the World Health Organization (WHO), World Organisation for Animal Health (OIE), and the Food and Agriculture Organization of the United Nations (FAO) [
1]. Globally, antibiotic resistant microbes are being recorded at increasing rates [
2], with ABR believed to be caused largely by the inappropriate use of antibiotics in both humans and animals [
3]. Although antibiotic consumption in humans and animals is still typically highest in high-income countries, in recent decades there have been substantial increases within low- and middle-income countries (LMICs) [
2,
4‐
6]. Within LMICs most inappropriate use of antibiotics by humans probably results from the inappropriate prescribing and sale of antibiotics, mainly via poorly regulated private providers, for conditions that do not require antibiotics, particularly uncomplicated upper respiratory tract infections (URTIs) [
7,
8]. Most inappropriate use of antibiotics in animals within LMICs comes from their indiscriminate use as growth promoters in large-scale commercial farming [
2,
6,
9]. However, in LMICs very small-scale, household-based, domestic or “backyard” animal farming is an extensive and important source of food and income across rural communities [
10]. For example, domestic/household poultry farming accounts for the majority of the poultry populations in many LMICs [
11].
Many academic studies and global action initiatives have therefore called for a better understanding of how antibiotics are being used in humans and animals in LMICs, particularly in rural settings where most LMIC populations live [
9]. Looking at both human and animal antibiotic practices is also a key part of a holistic “One Health” approach to tackling ABR, as recommended by the WHO, OIE and FAO [
1]. Currently though, most survey work in this area has only addressed human or animal use separately, most human-focused studies have been facility- or provider-based, and most animal-focused studies have only addressed commercial, large-scale farming (e.g. [
8,
12,
13]). However, only community-based studies can provide a representative understanding of antibiotic practices in humans at the community-level due to the inherent selection bias of facility- or provider-based studies [
14]. Community-based studies also offer the most obvious approach to explore antibiotic practices in small-scale, domestic animal farming, which has received very little attention.
Therefore, in this study we surveyed rural households in villages across a typical district within Bangladesh, a lower-middle-income country [
http://data.worldbank.org/about/country-and-lending-groups]. Our main aim was to understand rural household practices in relation to antibiotic use in humans and households’ domestic animals, to understand the prevalence of key potential drivers of ABR in this context.
Results
Between May and June 2018, across the 14 chosen subdistricts of Cumilla we selected 28 villages (two per subdistrict) and household clusters (one per village). Across the selected households there were 691 eligible female household heads and 672 eligible male household heads, with 682 (98.7%) and 620 (92.3%) female and male household heads agreeing to participate respectively. Female and male respondents’ ages were broadly evenly distributed across the eligible age range, both were mostly married (97%), and a substantial proportion (37%) of both either had no education or had not completed primary education. However, while 74% of men reported having worked for cash or in-kind payment in the last month only 20% of women did so (Table
1).
Table 1
Respondent characteristics (n = 1302)
Sex |
Female | 52% (682) | – | – |
Male | 48% (620) | – | – |
Age |
18–27 | 33% (423) | 21% (128) | 44% (295) |
28–37 | 33% (422) | 33% (203) | 32% (219) |
38–49 | 34% (439) | 45% (276) | 24% (163) |
Education status |
No education/incomplete primary | 37% (484) | 40% (245) | 35% (239) |
Completed primary/incomplete secondary | 43% (564) | 38% (237) | 48% (327) |
Completed secondary | 15% (195) | 16% (99) | 14% (96) |
More than secondary | 4% (54) | 6% (35) | 3% (19) |
Marriage status |
Married | 97% (1249) | 97% (586) | 97% (663) |
Never married | 1% (15) | 2% (15) | 0% (0) |
Widowed/divorced/separated | 2% (24) | 1% (6) | 3% (18) |
Worked for cash/in-kind payment in last month? |
Yes | 45% (589) | 74% (453) | 20% (136) |
No | 55% (707) | 26% (162) | 80% (545) |
Below we highlight the responses to key questions. For questions related to human antibiotic practices we only highlight the overall results unless there were clear differences between groups. For simplicity we sometimes refer to children as “reporting” their responses, but technically their mothers reported on their behalf.
Overall 48% (95% CI: 40, 56%) of women and men said they had heard of antibiotics, and we refer to these individuals as being “antibiotic-aware” from here on. Among these antibiotic-aware women and men 80% (95% CI: 76, 84%) and 76% (95% CI: 68, 82%), respectively, reported having ever taken antibiotics, while 56% (95% CI: 45, 67%) of antibiotic-aware women with children under 15 reported that at least one of their children had ever taken antibiotics. Across all women, men and children who reported taking antibiotics overall 21% (18, 25%) reported having taken antibiotics within the last month. See Table
S1 for full results on reported antibiotic awareness and household member antibiotic use.
We asked those antibiotic-aware respondents who reported that they or any of their children had previously taken antibiotics what illnesses or main symptoms they had ever taken antibiotics for. Overall the most frequently reported condition was undifferentiated “fever” (68% [95% CI: 60, 75%]) followed by “cough/cold” (52% [95% CI: 40, 63%]). For all other illnesses/main symptoms the overall results were substantially less than 50%, but notably “sore throat”, a common symptom of viral URTIs, was reported by 24% (95% CI: 19, 30%). See Table
S2 for full results on reported reasons for obtaining antibiotics for human use. We also asked this same group of respondents where they had ever obtained antibiotics from. Overall by far the most frequently reported source was a pharmacy (79% [95% CI: 73, 84%]), followed by a private medical practitioner (35% [95% CI: 27, 44%]) and a paramedic/village doctor (20% (95% CI: 14, 27%]). Overall only 5% (95% CI: 2, 12%) reported ever obtaining antibiotics from the only local source of free primary care and antibiotics: a Community Clinic. See Table
S3 for full results on reported sources for obtaining antibiotics for human use. This same group of respondents did not frequently report practices when last taking antibiotics that were contrary to the WHO’s recommended best practices [
20], which are: 1) Only use antibiotics when prescribed by a certified health professional. 2) Never demand antibiotics if your health worker says you don’t need them. 3) Always follow your health worker’s advice when using antibiotics. 4) Never share or use leftover antibiotics. Overall the most frequently reported inappropriate behaviours were obtaining an antibiotic without a prescription (30% [95% CI: 24, 37%]) followed by “taking fewer antibiotics than recommended by a health professional” (17% [95% CI: 12, 23%]). See Table
S4 for full results on antibiotic-use related practices that are consistent/inconsistent with the WHO’s recommended best practices.
73% (95% CI: 68, 78%) of households reported owning poultry (mean poultry if any owned = 10 [95% CI: 9, 11]), mainly chickens and ducks, while 32% (95% CI: 26, 40%) reported owning cattle (mean cattle if any owned = 2 [95% CI: 2, 3]), and 9% (95% CI: 6, 13%) reported owning goats and/or sheep (mean goats/sheep if any owned = 3 [95% CI: 2, 4%]). See Table
S5 and
S6 for full ownership results. Due to the low numbers owning goats and/or sheep we only present results for poultry and cattle. Animal husbandry practices that may drive zoonotic disease transmission, thereby increasing the need for antibiotics, were frequently reported. Specifically, 76% (95% CI: 68, 83%) and 35% (95% CI: 28, 43%) of poultry- and cattle-owning households, respectively, reported that they kept their poultry/cattle in their houses at night, while 42% (95% CI: 35, 49%) and 49% (95% CI: 35, 64%) of poultry- and cattle-owning households, respectively, reported that their poultry/cattle sometimes drank from or bathed in the household’s drinking or cooking water source(s). See Table
S7 for full results related to animal husbandry practices relevant to antibiotic resistance and antibiotic use in domestic animals.
Animal illness was frequently reported, particularly for poultry. Among poultry-owning households 65% (95% CI: 53, 75%) reported at least one of their poultry had previously been ill or died suddenly, and among poultry-owning households reporting previous poultry illness 77% (95% CI: 69, 84%) reported that the most recent illness was within the last 6 months. Among cattle-owning households 27% (95% CI: 17, 39%) reported at least one of their cattle had previously been ill or died suddenly, and among cattle-owning households reporting previous cattle illness 57% (95% CI: 39, 73%) reported that the most recent illness was within the last 6 months.
Ill cattle appear more likely to be treated than ill poultry. Among animal-owning households that reported ever previously seeking treatment for ill animal(s) 75% (95% CI: 60, 86%) of these households that owned poultry and 98% (95% CI: 79, 100%) of these households that owned cattle reported that the treatment involved either “western medicines” (in the form of either pills, liquids or injections) or “food with added medicine/drugs”. Among poultry-owning households that reported ever treating their animal(s) when they were ill only 1% (95% CI: 0, 7%) and 4% (95% CI: 1, 18%) respectively, reported having previously used a government or private veterinarian, while the most frequently reported previous source of treatment was a drug seller/pharmacist (48% [95% CI: 35, 60%]). However, among cattle-owning households that reported ever treating their animal(s) when they were ill 25% (95% CI: 14, 41%) and 32% (95% CI: 12, 62%), respectively, reported having previously used a government or private vet, although 14% (95% CI: 5, 31%) still reported previously using a drug seller/pharmacist, and 27% (95% CI: 19, 38%) a village doctor/paramedic/private medical practitioner. See Table
S8 for full results related to domestic animal illness and treatment.
In relation to antibiotics, we asked antibiotic-aware respondents of animal-owning households who reported previously having ill animal(s) treated whether such treatment ever involved antibiotics, and this was reported by 24% (95% CI: 11, 45%) of these households that owned poultry and 36% (95% CI: 13, 67%) of these households that owned cattle. We also asked antibiotic-aware respondents if any of their animals had ever been given antibiotics when they were not ill, and this was reported by 11% (95% CI: 8, 15%) of poultry-owning households and 2% (95% CI: 0, 9%) of cattle-owning households. Additionally, 14% (95% CI: 9, 23%) of antibiotic-aware poultry-owning households said they “sometimes/usually/always” bought commercial feed for their animals, and 21% (95% CI: 7, 50%) of those who responded in this way said that this feed “sometimes/usually/always” contained antibiotics. For antibiotic-aware cattle-owning households the corresponding figures were 63% (95% CI: 41, 81%) and 20% (95% CI: 9, 37%) respectively (Table
S5). Unfortunately, the sample sizes for responses to subsequent questions related to animal antibiotic use, including when healthy and in feed, were too small to be of much inferential use, but reported sources of antibiotics included drug sellers/pharmacists and village doctors/paramedics/private medical practitioners, and reasons for giving healthy animals antibiotics included increasing growth or production of eggs/milk/meat and preventing disease. See Table
S8 for full results on practices related to antibiotic-use in domestic animals. We also asked respondents about domestic animal vaccination and related issues and present the results for interested readers in Table
S9.
Discussion
To our knowledge, our study provides the first comprehensive information from Bangladesh about rural households’ antibiotic practices in relation to humans and domestic animals. We found a substantial lack of awareness about the existence of antibiotics as a distinct class/type of medicine in this context, which necessarily restricted our ability to understand antibiotic practices among non-aware individuals. This problem is faced by all similar surveys of rural LMIC communities, with levels of awareness varying substantially but typically being moderate to low [
30‐
32], although comparisons are difficult. Some similar surveys do not initially check awareness and apparently rely on respondents who are not aware of antibiotics to select “don’t know” type responses [
19,
33,
34], which may lead to increased bias. Therefore, our results, and those of similar studies looking at antibiotic knowledge in more detail, highlight that there is a huge amount of work to be done on educating rural populations in LMICs about antibiotics and appropriate practices.
Like similar surveys from other LMICs, including those in urban settings, our data indicate high levels of antibiotic consumption among rural community members [
17,
33‐
36]. Globally, national-level rates of antibiotic resistant microbes/infections are closely correlated with human antibiotic consumption rates [
37,
38]. Therefore, reducing inappropriate antibiotic use in Bangladesh and other LMICs is required to tackle ABR as overall antibiotic use is expected to continue to rise across LMICs. Probably the most frequent cause of inappropriate antibiotic use for human illness is when antibiotics are taken for URTIs. Globally, URTIs are likely responsible for most visits to formal and informal healthcare providers, but because URTIs are usually viral and self-limiting international guidance advises against routinely treating them with antibiotics unless there are complications or other clear reasons [
39,
40]. However, our results suggest that antibiotic consumption for URTIs is probably common among adults and children in these communities. This is consistent with data from other LMIC community settings as well as formal primary care and informal/private healthcare provider settings in both LMICs and high income countries (HICs) [
13,
41‐
48]. Evidence from both LMICs and HICs shows how important patients’ and caregivers’ beliefs, expectations and demands are in determining whether healthcare providers prescribe/sell antibiotics inappropriately, which highlights that the public have a vital role to play in tackling inappropriate antibiotic use among humans [
48‐
52].
However, despite this clear evidence from across the world that highlights just how problematic inappropriate antibiotic prescribing is within formal and informal primary care settings, appropriately used antibiotics are obviously critical and irreplaceable tools that can substantially reduce mortality from communicable diseases within LMICs, particularly among children. And yet in most LMICs there is still actually often a lack of access to appropriate antibiotics in primary care settings [
8]. There is also limited but compelling evidence that mass, untargeted distribution of some antibiotics can reduce mortality in LMICs [
53,
54]. Therefore, LMICs must somehow find a balance between increasing appropriate access to antibiotics for human use while reducing inappropriate use [
55], but this will be difficult to achieve without affordable and effective diagnostic tools for diseases such as URTIs, appropriate training of healthcare providers, and effective public education. Encouragingly though, in some LMIC primary care settings multi-component interventions that target providers and patients/caregivers, via approaches such as provider training, peer-review of antibiotic prescribing rates, and provider-led patient/caregiver education, have led to substantial and sustained reductions in inappropriate prescribing [
56,
57].
We found that the main reported sources for antibiotics were private providers, particularly pharmacies, which in practice may also mean informal and unregulated drug sellers, again consistent with similar surveys [
14,
32,
35]. It is well established that most medicines in LMICs are obtained from private sources, which are typically far more numerous, easily accessible and often preferred to public facilities, despite often lacking regulation and providing low-quality care, including the frequent sale of antibiotics for inappropriate or incorrectly diagnosed conditions [
58]. We were surprised that very few individuals reported previously obtaining antibiotics from Community Clinics though, given there is approximately one Community Clinic for every 6000 individuals in rural areas of Bangladesh, and given they provide free, basic primary-level care, including providing some antibiotics for non-severe conditions. Despite previous challenges with the quality of care being provided in Community Clinics, a training programme for Community Clinic nurses demonstrated high levels of appropriate antibiotic provision and has now been scaled-up nationally [
59]. Therefore, if Bangladesh can find a way to encourage individuals to use Community Clinics more frequently this could be a cost-effective way to reduce inappropriate antibiotic use within rural communities.
Antibiotic-aware respondents who reported using antibiotics did not report deviating from the WHO’s recommended best practices very frequently. For example, most such respondents reported that they and/or their children purchased antibiotics most recently using a prescription. However, contextual knowledge along with studies from Bangladesh and similar contexts suggest that prescriptions are rarely required to obtain medicines from private sources [
60‐
62]. Similar surveys have asked comparable questions but typically find that inappropriate practices are reported more frequently than here [
17,
19,
63]. However, it is not clear whether this reflects differences in practices or just the effects of reporting bias, and further research is required to explore this question.
Animal-owning respondents frequently reported previous experience of animal illness, and that treatments primarily involved “western medicines” or “drugs”. They also reported seeking treatment for ill cattle far more often than ill poultry, probably reflecting the far higher value of cattle. Evidence from rural areas within another Bangladeshi district suggests such generic terms might sometimes/often reflect the use of antibiotics, but this was based on small numbers of qualitative interviews [
64]. As there is very little other relevant data, the extent of therapeutic antibiotic use in domestic animals within Bangladesh or other LMICs remains unclear. We also found that human healthcare providers were reportedly frequently used to treat ill poultry, and although veterinary care was reportedly used much more frequently for ill cattle human-healthcare providers were still commonly used. This is also consistent with Roess et al. [
64] and studies from other LMICs indicating that human healthcare providers are often used for sick animals [
65,
66], and reinforces the view that LMICs face a huge challenge in providing and facilitating access to appropriate and qualified healthcare for domestic animals.
We also found some evidence of limited use of antibiotics in healthy poultry or cattle, potentially to increase growth or production of eggs/milk/meat or as prophylactics. There is clear and growing evidence linking the indiscriminate use of antibiotics within small- and large-scale agriculture to the development of antibiotic resistant microbes [
67‐
69], and there is an increasing trend in the use of antibiotics for growth promotion and prophylaxis in commercial agriculture across LMICs. However, there has been very little focus on domestic animal husbandry, as opposed to small-scale farming [
69], despite its prevalence in LMICs. For example, in Bangladesh 55% of households own chickens and 33% own cattle, but 76% of chicken owners have ≤10 birds and 83% of cattle owners have ≤3 cattle [
16]. Therefore, while the use of antibiotics in domestic animals for non-therapeutic reasons may not yet be common in this context, given the scale of domestic animal ownership in Bangladesh any increases in such practices would represent clear risks for rural Bangladeshi communities given the inevitable increase in selection pressure for antibiotic resistant microbes/infections [
67‐
69]. Consistent with related studies [
64], we also found risky animal husbandry practices were frequently reported by animal-owning households, like sharing drinking/bathing water sources with animals. Such practices are known to increase the transmission of microbes among household members, and therefore represent pathways by which – potentially increasingly prevalent – antibiotic resistant microbes could infect rural communities.
A key limitation of our study is that we had to restrict our inferences for most questions to the sub-populations who were able to answer them, particularly the sub-population of individuals who had heard of antibiotics. With hindsight we would have been able to estimate the rest of our results (other than the question of antibiotic awareness) with more precision had we planned our sampled size to only include individuals who were aware of antibiotics, but while still collecting information on the key question of antibiotic awareness during an eligibility screening phase, for example. However, this does not affect the level of any bias in our results. A closely related limitation is that we sometimes ended up with impractically small sample sizes for some key questions that only small sub-populations within our sample could respond to. In terms of generalisability, the respondents remaining in our sub-population analyses are very unlikely to be representative of the complete rural communities that we sampled. Therefore, when generalising any of our sub-population results it is important to be clear that they can only be robustly applied to the relevant sub-populations within the sampled target population of adult (aged 18–49) female and male household heads, and the children of those females, within rural communities in Cumilla district. For example, our results on the frequency with which adult female household heads have previously taken antibiotics can only robustly be generalised to the population of adult female household heads living in rural communities who have also heard of antibiotics. Generalisations beyond this group would be questionable. The final obvious key limitation of the study is that we relied on self-reported responses, which are inevitably prone to respondent recall bias and social desirability bias. Obtaining more objective data in such surveys and contexts is very difficult, but future work could test possible methodological innovations [
70].
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.