Use of a cow-side oestrus detection test for fertility management in Kenyan smallholder dairy herds.

Introduction

In Kenya, as in many lower- and middle-income countries, the use of artificial insemination (AI) has the potential to improve cattle genetics and reproductive performance in smallholder dairy herds (Gicheha et al., 2019). Furthermore, improving access to breeding services and thus increased access to quality genetic material would make smallholder dairy farming more sustainable and economically viable (Murage & Ilatsia, 2011; Odero-Waitituh, 2017). However, the associated high costs, coupled with low success rates, has meant that efforts to improve livestock production via reproductive technologies, including AI, have tended to fail (Lawrence et al., 2015; Omondi et al., 2017; Rademaker et al., 2016). Evidence suggests that dairy farmers in Kenya are keen to use AI services on their farms (Khainga et al., 2018; Omondi et al., 2017) but need to be convinced of its efficacy in order for uptake to be optimised (Murage & Ilatsia, 2011).

A major contributing factor to poor reproductive performance in smallholder dairy farms in Kenya has been poor oestrus detection rates (Mwai et al., 2020; Owen, 2005), with mistiming of AI leading to poor conception rates and increased calving intervals (Mungube et al., 2014). Effective AI requires efficient and accurate detection of ovulation, traditionally signalled by observed oestrus behaviour. Oestrus behaviour typically lasts between six and 30 hours and is dependent on cow and seasonal factors, with the main sign of oestrus standing to be mounted by a bull, or other cows (Ball & Peters, 2004c). Risk factors for poor expression of oestrus can be environmental factors such as nutrition, housing, season and number of herd mates in oestrus simultaneously, or cow factors such as silent or anovulatory oestrus and adverse health conditions (Roelofs et al., 2010). An inability to easily detect oestrus, coupled with poor expression of behavioural signs of oestrus by the cow can hinder insemination at the correct time (Walsh et al., 2011). Even when farm staff are experienced, many oestrus events can still be missed, and ‘false’ oestrus, resulting from abnormal cow behaviour or human error, can also occur. Even in the most sophisticated intensive dairy systems in high-income countries, oestrus detection remains an obstacle to optimal reproductive efficiency (Adenuga et al., 2020).

Calving to first service interval and first service to conception interval are both dependent on the rate of oestrus detection as well as the herd conception rate (Esslemont & Ellis, 1974). The early detection of pregnancy is also essential in managing efficient dairy cow reproduction or rather non-pregnancy so that the appropriate management decision can be implemented e.g. repeat insemination (Fricke et al., 2016). Overall pregnancy rates to AI have been disappointingly low in African countries (Omondi et al., 2017). In Kenya, the estimated national average insemination rate ranges from 1.5 to three per cow per conception (Elving et al., 1979; Ministry of Livestock Development, 2010; Mungube et al., 2019; Mutiga et al., 1994; Odima et al., 1994; PICO-Eastern Africa, 2014).

As progesterone (P4) is a hormonal product of the corpus luteum, its measurement can be applied as a confirmatory test to the two above situations, i.e. to verify accurate detection of oestrus on the day of insemination, or to confirm non-pregnancy approximately 21 days after insemination by the finding of a low P4 concentration in blood or milk. The measurement of bovine P4 has evolved over the past four to five decades from laboratory based radio-immunoassay through enzyme-linked immunosorbent assay (ELISA) techniques to cow-side lateral flow tests or in-milk-line monitoring (Samsonova et al., 2018). The cow-side P4 Rapid lateral flow test was developed in the UK (Ridgeway Science, UK) and studies comparing its performance to other assay methods have been published (Ingenhoff et al., 2016; Kočíková et al., 2019; Muasa, 2020; Omontese et al., 2020; Waldmann & Raud, 2016).

The present study was carried out to evaluate the use of the cow-side P4 Rapid progesterone lateral flow test in confirmation of oestrus, as well as non-pregnancy in dairy cows in Kenya and to determine whether its application was associated with increased effectiveness of AI.

Methods

Study area

The study was conducted in two counties in Kenya (Figure 1), between August 2019 and October 2020. The two counties, Nyandarua and Taita-Taveta, were areas where farmers utilise artificial insemination service providers (AISPs). In Nyandarua, farmers typically practice mixed farming – both intensive (zero-grazing) and extensive (free-range), with some farmers practicing semi-zero (zero-grazing in the morning and free-range in the afternoon). Cattle in this area mostly comprise Ayrshire x Friesian crosses, with very few pure breeds. Herd sizes ranged from three to seven cattle. In Taita-Taveta, zero-grazing is widely practised. Cattle breeds in this area were also mostly Ayrshire x Friesian crosses, very few pure breeds, and some indigenous zebu x Ayrshire/Friesian cross breeds, with herd sizes of two or three cattle.

Figure 1. Location of study areas.

The map on the left shows the position of Kenya (highlighted in blue) within Africa. The map on the right shows the study areas, Nyandarua and Taita-Taveta (highlighted in orange) within Kenya.

Reproductive management

Oestrus was determined by the farmer by observing behavioural signs and oestrous cycle tracking as described by Ball and Peters (Ball & Peters, 2004a). When oestrus was observed in cows in the test group, a milk sample was collected that day, usually at the next milking, and progesterone levels tested by the P4 Rapid test (Figure 2).

Figure 2. Ridgeway P4 Rapid test results and interpretation.

The lower line is the test line (in response to progesterone) and the upper line is the reference/control line (indicates that the test has run correctly). Low P4 is indicated by a stronger test line; medium P4 is indicated by equal lines; high P4 is indicated by a weaker test line.

AI and progesterone measurements were carried out by the AISPs who would normally, in the course of established herd management, inseminate the cows enrolled on the study. Prior to commencement of the study, a detailed training course (see Extended data; Allan, 2022) was provided for the AISP technicians to explain the rationale for the study and the theory behind the P4 Rapid Test, as well as to demonstrate the test procedure.

Results

A total of 527 cows were enrolled in the study, of which 308 were in the test group and 219 in the control group. The outcomes of all enrolled cattle are presented in the Underlying data (Allan, 2022), showing data retained or removed for analysis. In the data subset, a total of 174 cattle remained in the test group (134 removed) and 149 remained in the control group (70 removed).

Days to conception

The days to conception (DTC) were compared between test and control groups. Cows that aborted or had a stillbirth were included in the analysis, and considered to have conceived at their last recorded AI date. For cows that did not conceive during the study, DTC was calculated as the number of days between enrolment into the study (Day 0) and the study end date (1 December 2020). Summary statistics were calculated for test and control groups for DTC (Table 6, Figure 3a and 3b).

Table 6. Summary statistics for days to conception (DTC) for test and control groups.

Group	Number of cattle	Mean	Median	Standard deviation	Interquartile range	Range
Test	174	76.04	0	142.1	0–21	0–453
Control	149	62.11	0	117.0	0–60	0–451

Figure 3. Summary statistic histograms of days to conception for a) test and b) control groups.

Frequency (%) is on the y-axis and number of days on the x-axis.

Discussion

Both cultural and structural challenges constrain cattle breeding productivity in Kenya (DeLay et al., 2020). For example, whilst it has been demonstrated that the early detection of oestrus can reduce calving intervals, the preference for long lactations by some farmers can reduce or eliminate such advantages (Odima et al., 1994) and better record keeping could highlight the importance of fertility problems, abortions and tracking calving intervals (Omore et al., 1998). It is estimated that overall around 16% of smallholder farms in Kenya are using AI (Mutavi et al., 2016) although there are reports in some areas that uptake is as high as 94% (Ajak et al., 2020). In Nyandarua, high costs and inaccessibility of AI have been reported as prohibitive to around 60% of households, with smallholders choosing natural breeding methods instead (Muia et al., 2011), however there are other reports of higher uptakes (Gitau, 2013). The timing of AI is essential to its success, which requires the accurate detection of oestrus, however, oestrus detection failure is a major constraint to reproductive performance (Ball & Peters, 2004b). This has led to the development of technological aids to oestrus detection.

This study assessed the performance of the P4 Rapid progesterone lateral flow test in confirming oestrus and non-pregnancy in Kenyan dairy herds. A total of 745 tests were performed, of which 1.5% (11/745) were inconclusive (medium P4 level). In detecting oestrus, 93.8% (289/308) of all the cows in the test group presented for AI on day 0 had low P4 and were in oestrus. The study protocol required cows that were not in oestrus to be re-presented for AI the next time oestrus signs were visible, however, all repeat visits by farmers were on day 21 and day 42 regardless of day 0 test result. It is therefore not clear whether these cows were showing signs of oestrus on these days and so they were excluded from further calculations. Several possible scenarios associated with the test results were described and worst- and best-case scenarios presented. The true overall accuracy of the test performance likely lies somewhere between the extremes presented in these scenarios i.e. between 87–97%. A previous study, in Estonia, reported 98% accuracy in determining oestrus, and 84% accuracy in determining non-pregnancy, using the P4 Rapid test (Waldmann & Raud, 2016).

Although there was no significant difference in pregnancy rates between the two groups, there was a tendency for higher rates in the test group (83.9%) compared to controls (77.9%). These pregnancy rates were much higher than anticipated, even in controls which may explain why there was no significant difference between tested and control cows. It has been noted previously that any beneficial response to a fertility management intervention is likely to be negatively proportionate to the background fertility of the herd before the intervention (Peters, 1996, citing data from Drew & Peters, 1994) i.e. the poorer the fertility before intervention the greater the beneficial response to the intervention, and vice versa.

For cows with known outcomes, a total of 30 (10.3% of high test results) tests gave high P4 results in non-pregnant cows. It should be remembered that the P4 Rapid is not a pregnancy diagnostic tool. Conversely, there were eight instances where cows were pregnant but gave a low P4 test result (3.7% of low test results).

The study looked at the risk of abortion as a result of insemination of already-pregnant cows. In the test group, three cows were accidentally inseminated despite testing high; one cow was subsequently diagnosed as not pregnant so it is unclear if the cow aborted after AI or if the high P4 level was erroneous; one cow was confirmed as pregnant by ultrasound at 90 days and as the conception date was around the start of the study it was not clear if she was already pregnant, or became pregnant from AI; and one cow was lost to follow up i.e. unknown outcome. In the control group, 2.3% (5/219) of cows were already pregnant when served, based on subsequent calving date and backward calculation of conception date. However, for the cows that aborted, abortion date or foetal age were not always known and therefore it is not possible to determine whether the cow was already pregnant at AI – and AI was a causal factor – or whether the cow became pregnant from AI and then went on to abort. There are no recent published data on the risk of abortion by AI, and only very few older studies available. Vandemark et al. (1952) described 21 pregnant cows, in three groups; the seven cattle in group 1 were inseminated no further than the mid-point of the cervix and all had viable pregnancies; the eight cattle in group 2 were inseminated into the body of the uterus, with one cow aborting and seven resorbing foetuses; and the six cattle in group 3 were inseminated into the body of the uterus with antibiotics given, with four viable foetuses, one resorbing foetus and one partially destroyed foetal membranes but live foetus. Weaver et al. (Weaver et al., 1989) inseminated 57 oestrous cows; 25 were re-inseminated into the uterine body 12–24 days later while not in oestrus. Pregnancy rates were significantly lower in those re-inseminated (4%) compared to those not (40.6%). Sturman et al. (Sturman et al., 2000) observed embryo or foetal loss after insemination in 24% of pregnant cows, compared to 7% spontaneous loss of pregnancy otherwise occurring in the study. Moore et al. (Moore et al., 2005) reported the most important risk factor associated with embryo loss at 21–27 days and 28–35 days to be a second AI just prior or during these periods, with cattle 3.9 and 3.7 times, respectively, more likely to lose the embryo than those not re-inseminated. As well as the risk of iatrogenic abortion, mistiming of AI can also result in wasted semen and its associated expense.

There were obvious limitations in the study. The study was semi-experimental and not randomised, as this was thought to be potentially overly complicated for such a practical study setting on smallholder dairy farms. It was considered that too strict a protocol would reduce operator and farmer cooperation and compliance. This, therefore limited our ability to conclude causal association between the P4 Rapid test intervention and outcome(s). Although the study was planned as a direct comparison of fertility parameters between tested cows and non-tested (control) cows, no significant difference in pregnancy rates was found and thought to be due to the relatively good background fertility management in the selected herds. This suggests that the efficiency of oestrus detection was quite good in these herds and therefore the incremental benefit of confirmation by P4 testing could only be expected to be marginal. In this situation, therefore, routine testing is clearly not warranted on economic grounds but selection of cows for P4 testing should be based on doubtful or problem cases. Further work is planned on the potential economic impact of cow-side progesterone testing in sub-Saharan Africa.

In conclusion, the P4 Rapid test is considered a useful tool to help Kenyan dairy farmers overcome the key challenges in the timing of AI, particularly where the level of oestrus detection is poor. Whilst its accuracy is imperfect, it compares favourably to quantitative progesterone assays but is more suitable and accessible for the smallholder setting, as a cow-side test, without the necessity for refrigeration or laboratory facilities. Additionally, and importantly, the P4 Rapid test may assist farmers in reducing AI-induced abortions by avoiding the insemination of pregnant cattle.