Association between Livestock Ownership and Malaria Incidence in South-Central Ethiopia: A Cohort Study

INTRODUCTION

Zooprophylaxis is the use of animals to divert biting mosquitoes away from humans to prevent malaria infection. It was one of the recommended environmental measures for malaria vector control many years ago. It was considered an effective strategy in areas where the vectors that transmit malaria prefer to feed on animals (zoophilic) rather than on humans (anthropophilic). In addition, it could be implemented in areas where domestication of cattle, sheep, goats, or other livestock is common practice.¹

However, the effect of zooprophylaxis remains debatable.^2–8 Some researchers are in favor of it, whereas others reported it not to be beneficial, implying the need to evaluate its effect context-wise.⁹ It was argued that the presence of animals avails a blood meal source for mosquitoes and in effect contributes to their longevity, and thus their potential infectivity could be prolonged—the condition referred to as zoopotentiation; this might neutralize (even surpass) the desired zooprophylactic effect.^10,11 Accordingly, to counterbalance such an undesired effect, strategies were proposed to enhance zooprophylaxis by, for instance, treating the animals with insecticides or separating their dwellings from humans and by other vector control measures, including the use of long-lasting insecticidal nets (LLINs) and indoor residual spraying with insecticides (IRS).^12–14

Quantifying and projecting the effect of zooprophylaxis has been difficult, and concluding whether zooprophylaxis is a helpful strategy or not just by changes in entomological parameters makes it more complicated.¹⁵ For example, a study documented that, among collected mosquitoes, none was found to have the malaria parasite, making entomological inoculation rate (EIR) calculations impossible,¹⁶ even though its parallel epidemiological study confirmed the presence of cases of malaria.¹⁷ Such discrepancies could arise from (but are not limited to) mosquito sampling variations that complicate extrapolation of the effect of zooprophylaxis measured on entomological parameters to its impact on the overall incidence of malaria among humans. Although measuring the effect of zooprophylaxis using entomological parameters remains the best strategy, whenever parameters like EIR cannot be obtained, investigating the effect on human infections may be an option.

In Ethiopia, with an altitude < 2,000 m, the malaria parasite prevalence is generally low, at 1.2% in areas considered malarious.¹⁸ However, the heterogeneous nature of the transmission leads to high disease rates of epidemic potential in specific foci. The country is embarking on elimination, but maintaining the optimal benefits of the existing vector control mechanisms remains challenging.¹⁹ For example, a recent large cluster-randomized control trial investigated if combining interventions (LLINs with IRS) has an added value to either LLINs or IRS implemented alone. The result showed no added value. Surprisingly, it also documented no effect of either the combined or single intervention compared with the group that received neither. Consequently, residual malaria transmission and the low malaria incidence in the study area were speculated as possible reasons for not observing any effect of these proven prevention tools,¹⁷ but these results merit further investigation.

A review of the literature has suggested that an essential factor to consider when using zooprophylaxis as a malaria prevention strategy is whether the vectors are zoophilic or exophilic.^9,13 Interestingly, the principal malaria vector in the study area was Anopheles arabiensis, which is zoophilic,^14,20 biting often outdoors (exophilic) and before bedtime (before people are protected by bed nets).¹⁶ The bovine blood index of mosquitos collected outdoors was high (68%).²¹ This may imply the potential role of zooprophylaxis in the study area; this requires investigation. In addition, the fact that Ethiopia has the largest livestock population in Africa (65, 40, 51, and 49 million cattle, sheep, goats, and chickens, respectively)²² may encourage assessing this huge potential for malaria epidemiology in the country.

Therefore, taking advantage of the rich entomological information generated from the study area showing the exophilic and zoophilic behavior of the vector, the presence of data on distance from a potential malaria vector breeding place, demographic and economic characteristics, and the allocation of study participants to different trial arms for a large cluster-randomized controlled trial,^16,17,20,23 this study aims to assess the effects of livestock ownership on the incidence of malaria in a large cohort of residents followed for 2 years in the Adami Tullu district in Oromia Regional State, Ethiopia.

MATERIALS AND METHODS

This is a secondary analysis of the effects of livestock ownership on the incidence of malaria using a dataset from a large cluster-randomized controlled trial conducted to study the effect of a combination of IRS and LLINs versus a single intervention conducted from October 2014 to January 2017. This analysis was conducted in a framework of prospective cohort study design. The study area is located in the Adami Tullu district of the East Shewa Zone of the Oromia region in Ethiopia—in the Great Rift Valley with an altitude ranging between 1,500 m and 2,300 m. The capital of the district, Batu town, has a latitude of 7°56′N and a longitude of 38°42′E. Lake Ziway and Bulbula river are the main water bodies in the area. Farming and livestock rearing are the main means of subsistence in the area. Background information was collected at the beginning of the study, and a weekly follow-up visit to each household (HH) was conducted for a total of 121 weeks. Detailed descriptions of the main study setup, the dataset, and the results of the trial are provided elsewhere.^17,23,24

RESULTS

A total of 27,471 residents had complete follow-ups. The majority (87.5%) of study participants lived in a HH that owned livestock. The types of livestock included cow/ox, donkey/horse/mule, goat, sheep, and chickens. The majority of HHs with livestock owned a cow/ox (92%), with a median (IQR) of 4 (2–6). The majority of the HHs owned more than one type of livestock, and nearly one-fourth owned two, three, or four types of livestock. Slightly more than half (54%) of the study livestock owners had nine or more livestock of any type (Table 1).

The distribution of livestock ownership was almost similar across categories of variables initially considered as potential confounders (including the study arms of the main trial) except the location/Kebele of the HH and the distance of the HH from the potential malaria vector breeding place, for which the proportion ranged from 75.2% to 95.6% and median distance of 1.19 km versus 1.81 km, respectively (Supplemental Table 1).

A total of 1,013 of 27,471 cases of malaria were documented. The incidence of malaria was almost equal across different categories of potential confounders. Like the exposure variable distribution, the only exceptions observed were for the location/Kebele of the HH and the distance of the HH from the potential malaria vector breeding place, whereby we see a varying proportion that ranged between 0.8% to 13.9% and median distance of 1.76 km versus 1.33 km, respectively (Supplemental Table 2).

Among the available variables investigated for potential confounding effect, all variables except the location of the HH (Kebele and distance from vector breeding place) showed very little or no effect on the exposure variable estimate. The percentage change resulting from the inclusion of the location of the HH with reference to the potential malaria vector breeding place or the Kebele was almost 10% for the risk ratio, and this is the case only for the distance variable for the rate ratio analysis (Table 2). Because both Kebele and distance from the vector breeding site indicate the physical location of the HH, to be precise, distance from the vector breeding site of each HH was considered to adjust for the effect of the livestock ownership on malaria risk or rate.

DISCUSSION

There was a 24% reduction in the risk of acquiring malaria infection if the HH owns livestock after adjusting for proximity of the household to vector breeding habitat. The protective effect increased to 34% for those having nine or more livestock. Similarly, there was a 17% reduction in the rate of malaria among the exposed. Living with many livestock showed a further reduction of the malaria rate by 27%. The absolute reductions (1.5% or 4.5 cases per 1,000 person-years) may seem negligible, but because the overall malaria risk (3.7%) and rate (14.7 cases per 1,000 person-years) in the study area are low, the absolute reductions observed could be considered proportionally sizable and of public health importance.

This study showed that the greater the number of livestock, the lower the risk of malaria. This may imply that an increased animal blood meal availability might have resulted in an effective diversion of malaria mosquitoes away from humans. The increase in the number of domestic animals may also accompany a risk of increasing the lifespan of the vector; however, if the vectors were attracted to animals and prefer to be fed outdoors, having livestock around may not always increase the malaria risk (zoopotentiation). A model by Sota and Mogi⁵ showed that to have the expected benefit of zooprophylaxis, there must be an increase in the number of domestic animals, given the malaria infection rate remained low. In line with this notion, it was interesting to see the relative abundance of livestock to the household size. The result was consistent in that the greater the livestock-to-human ratio was, the greater the protective effect.

In the main trial, both indoor malaria prevention tools (IRS with LLINs or each alone) did not show a difference in reducing malaria incidence compared with the control arm. The plausible reason provided was residual malaria transmission. Likewise, the entomological finding from the same study setup indicated that the principal malaria vector in the area demonstrated exophilic behavior and early biting (before bedtime), leading to speculation that an outdoor malaria transmission potential is more likely.¹⁶ In such a circumstance, we may consider livestock ownership to have played at least some role in malaria transmission dynamics, although the overall human blood index of An. arabiensis was higher compared with the bovine blood index (69% versus 39%) and the bovine blood index of those collected outdoors was higher than that indoors (68% versus 27%).²¹ This may indicate that livestock was around to feed the zoophilic mosquitoes outdoors in close vicinity of the owners. The pits that were used as outdoor mosquito resting shelters (and their subsequent collection spot) were also dug in a compound of the main house.^20,21 In addition, the more abundant and exophilic vector species in the area, An. zeimanni,¹⁶ was reported to feed more on cattle than on humans, with a bovine and human blood index of 67% and 50%, respectively.²¹ Therefore, if residual confounding could not explain the observed effect of livestock ownership on the risk or rate of malaria infection, given the feeding preferences of the An. arabiensis (principal) and also An. Zeimanni (potential and more abundant) vectors of malaria in the locality, the common practice of domestication of animals (88%) in this location might have contributed to the suppression of human malaria infections to a greater extent.

Meanwhile, LLINs are among the approved malaria prevention tools. In the study area, the proportion of individuals sleeping under the net the night before the interview never surpassed 60%, even at the start of the study, despite universal coverage being achieved. The coverage of LLINs decreased from 100% to as low as 4% in 2 years; the goal was that the HHs would preserve the LLINs in a functional state for 3 years, but most of the distributed LLINs were either thrown away or used for unintended purposes.^26,27 Therefore, in such a setup where we could not maintain optimal coverage and use of LLINs and where the predominant malaria vector has a zoophilic tendency, considering zooprophylaxis may be sensible. The proposed treatment of cattle with insecticides may need to be revitalized to maximize the benefit of zooprophylaxis.

This study has limitations. Most importantly, the exposure measurement was limited to the availability and count of livestock at the start of the study period, whereas malaria incidence was measured longitudinally. Because there were no data on whether the HH kept the livestock for the whole period of the study, the conclusions from this study should be taken with caution as the HH may not keep the livestock until the end of the follow-up period, and each member of the household may not get similar and sustained protection. In addition, the place to keep the livestock also matters for the effective diversion of mosquito bites, but there were no data on it. Therefore, it was impossible to report the effect of the distance of the animal quarters from or having the livestock inside or outside of the human dwelling. Nevertheless, with all such limitations, this study may shed light on the potential of this long-standing but poorly understood vector control mechanism in areas where the context favors it. Further research is warranted with better exposure measurement.

Supplemental Materials

Download PDF