Executive Summary

This report analyses meat inspection data collected by veterinary officers at licensed abattoirs across Kenya. The dataset comprises 234 records spanning 11 counties, 12 named abattoirs, and 7 species groups. Data covers ante-mortem inspection (animals presented for slaughter) and post-mortem inspection (carcass and organ condemnation).

Here are some of the few points gathered from prior analysis:

Data completeness is the primary surveillance risk. 59.4% of records carry a valid reporting date, meaning 40.6% of national slaughter activity cannot be placed in time. Every month with a missing record is unmonitored meat entering the food chain.
The national carcass condemnation rate is 0.27%. Across 1,286,563 slaughtered animals, 3,440 carcasses were condemned as unfit for human consumption. This rate is not uniformly distributed ie. specific counties and abattoirs account for a disproportionate burden.
3.42% of inspection records flagged C. bovis (beef tapeworm). This is a direct zoonotic risk: Taenia saginata larvae in beef infect humans. Geographic hotspots are identifiable and actionable.

1 Data Overview and Quality Assessment

1.1 Background and Dataset Scope

This report analyses the consolidated dataset comprising 234 records across 11 counties and 12 named abattoirs. The dominant species group is Goat (34.6% of records). A total of 116,710 animals were presented for slaughter across the dataset period.

1.2 Analytical Confidence Groups

All columns in this dataset have been classified into four confidence groups based on their completeness. No finding in this report is drawn from a column below its stated group.

Group	Completeness	Key columns
Complete	100%	county, species_clean, total_presented, carcasses_condemned, all 17 binary disease flags, partial_condemnation, organs_condemned_flag
High	85–99%	slaughtered_approved, number_dead, number_rejected, condemnation_rate_pct, source_of_animals, mortality_rate_pct, rejection_rate_pct, abattoir
Moderate	30–60%	reporting_date (59.4%), liver_condemned_n (48.7%), lung_condemned_n (39.3%), no_of_females (37.2%), condemn reason fields
Unusable	<15%	septicaemia_no (0.4%), c_bovis_no (3.4%), emaciation_n (5.6%), breed (5.1%), latitude/longitude (12.8%), process flags (13.2%)

Data Completeness Report
Column Name	Missing (%)	Completeness (%)
septicaemia_no	99.57	0.43
septicaemia_partially_condemned	99.57	0.43
septicaemia_fully_condemned	99.57	0.43
septic_peritonitis_no	99.57	0.43
septic_peritonitis_partially_condemned	99.57	0.43
septic_peritonitis_fully_condemned	99.57	0.43
multiple_abscesses_no	99.57	0.43
partially_condemned	99.57	0.43
fully_condemned	99.57	0.43
other_reason_for_partial_condemnation	99.57	0.43
skin_crop_condemn_reason	99.57	0.43
weight_condemned_skin_crop	99.57	0.43
weight_condemned_breast_muscle	99.57	0.43
breast_muscle_condemn_reason	99.57	0.43
stomach_intestines_condemn_reason	99.57	0.43
boilers_number	99.15	0.85
stomach_intestines_condemned_n	99.15	0.85
number_condemned_heads	98.72	1.28
head_tongue_condemnation_reason	98.72	1.28
jaundice_n	98.72	1.28
jaundice_partially_condemned	98.72	1.28
jaundice_fully_condemned	98.72	1.28
other_condemnation_reason	98.72	1.28
number_condemned_kidneys	98.29	1.71
kidney_condemnation_reason	98.29	1.71
condemnation_reason	98.29	1.71
weight_condemned_gizzards	97.86	2.14
gizzards_condemnation_reason	97.86	2.14
c_bovis_no	96.58	3.42
c_bovis_partially_condemned	96.58	3.42
c_bovis_fully_condemned	96.58	3.42
broilers_number	96.15	3.85
animals_with_other_condition	95.73	4.27
condemnation_led_no	95.73	4.27
observed_conditions	95.30	4.70
breed	94.87	5.13
no_of_pregnant_females	94.87	5.13
heart_condemned_n	94.44	5.56
heart_condemn_reason	94.44	5.56
emaciation_n	94.44	5.56
emaciation_partial_n	94.44	5.56
emaciation_full_n	94.44	5.56
partial_condemn_reason	92.31	7.69
carcass_condemn_reason	87.61	12.39
latitude	87.18	12.82
longitude	87.18	12.82
myotic_disease	87.18	12.82
arthritis	86.75	13.25
ascites	86.75	13.25
bruising	86.75	13.25
contamination	86.75	13.25
folliculitis	86.75	13.25
inadequate_bleeding	86.75	13.25
overscalding	86.75	13.25
peritonitis	86.75	13.25
pigmentation	86.75	13.25
lung_condemn_reason	68.38	31.62
liver_condemn_reason	63.25	36.75
no_of_females	62.82	37.18
lung_condemned_n	60.68	39.32
liver_condemned_n	51.28	48.72
reporting_date	40.60	59.40
report_year	40.60	59.40
report_month	40.60	59.40
report_quarter	40.60	59.40
report_yearmon	40.60	59.40
organs_condemned	11.54	88.46
no_of_males	10.68	89.32
rejection_rate_pct	9.83	90.17
mortality_rate_pct	8.97	91.03
abattoir	5.98	94.02
condemnation_rate_pct	2.56	97.44
source_of_animals	1.28	98.72
number_dead	1.28	98.72
number_rejected	1.28	98.72
number_isolated	0.43	99.57
slaughtered_approved	0.43	99.57
source_dataset	0.00	100.00
county	0.00	100.00
subcounty	0.00	100.00
ward	0.00	100.00
geo_valid	0.00	100.00
species_clean	0.00	100.00
sex_category	0.00	100.00
has_pregnant	0.00	100.00
total_presented	0.00	100.00
carcasses_condemned	0.00	100.00
total_condemned	0.00	100.00
partial_condemnation	0.00	100.00
multiple_abscesses	0.00	100.00
emaciation	0.00	100.00
septicaemia	0.00	100.00
septic_pleurisy	0.00	100.00
septic_peritonitis	0.00	100.00
jaundice	0.00	100.00
c_bovis	0.00	100.00
any_condition_flagged	0.00	100.00
organs_condemned_flag	0.00	100.00
total_organs_condemned	0.00	100.00
cbovis_positive	0.00	100.00
other_conditions_flag	0.00	100.00
flag_condemned_gt_slaughtered	0.00	100.00
flag_missing_date	0.00	100.00

Important

Approximately 60 columns fall below 15% completeness. These include all quantitative disease count fields (septicaemia_no, c_bovis_no, emaciation_n, jaundice_n), all organ-specific reason fields, the process failure flags (contamination, bruising, inadequate_bleeding etc. at 13.2%), and GPS coordinates (12.8%). None of these are used for inference in this report. The binary disease flags (100% complete) are the correct and sufficient signal for disease burden analysis.

1.3 Field Completeness Profile

1.4 Reporting Completeness by County

Counties are ordered by mean completeness across fields. Dark green = fully reported; red = rarely reported.

1.5 Species Distribution

Data confidence: (100% complete). species_clean is one of the most analytically valuable fields in the dataset and forms the stratification variable for all sections that follow.

Distribution of inspection records by species. All rate comparisons in this report are stratified by species.

The dataset is dominated by Goat (34.6% of records). This species imbalance means that county-level condemnation and mortality rates are heavily influenced by the mix of species slaughtered in each county. Counties where a minority species accounts for a disproportionate share of condemnations may show inflated or deflated aggregate rates.

1.6 Reporting Date Completeness by County

Proportion of records with a valid reporting date by county. Counties below 50% cannot support time-series analysis.

Nationally, 59.4% of records carry a valid reporting date, leaving 40.6% of slaughter activity temporally unanchored. This is not merely a statistical inconvenience because each undated record represents meat that entered the food supply without a timestamped veterinary inspection record.

2 Ante-Mortem Findings

Ante-mortem inspection assesses live animals before slaughter. Elevated mortality and rejection rates indicate that sick, injured, or distressed animals are being transported to abattoirs. This problem originates upstream at farms, livestock markets, or during transit.

Data confidence: (High). mortality_rate_pct and rejection_rate_pct are 92.3% complete; number_dead and number_rejected are 98.7% complete.

2.1 National Summary

Metric	Value
Total Animals Presented	116,710.0
Total Dead on Arrival / in Lairage	1,858.0
Total Rejected at Ante-Mortem	2,082.0
Total Isolated (Suspect)	50.0
Mean Mortality Rate (%)	2.94%
Mean Rejection Rate (%)	9.74%

Across 116,710 animals presented for slaughter, 1,858 were found dead on arrival (mean mortality rate: 2.94%) and 2,082 were rejected as unfit (mean rejection rate: 9.74%). These figures originate upstream of the abattoir and point to animal welfare and transport management as a key intervention point.

2.2 Species-Stratified Mortality and Rejection Rates

Data confidence: species_clean (100%) combined with mortality_rate_pct and rejection_rate_pct (92.3%).

Mean ante-mortem mortality rate (left) and rejection rate (right) by species. Species groups have fundamentally different physiology, transport distances, and market chains — aggregate county rates without species stratification are misleading.

2.3 County Maps of Mortality and Rejection Rates

County-level mean mortality rate (left) and rejection rate (right) at ante-mortem inspection. Note: these aggregate rates reflect the species mix in each county, not solely animal welfare or transport conditions.

2.4 Pregnant Female Slaughter Events

Data confidence: (100% complete). The has_pregnant binary flag is fully populated. Note: no_of_pregnant_females is 94.9% missing and is not used here.

Proportion of slaughter events where pregnant females were present, by county. Elevated rates are a potential indicator of distress-driven herd liquidation linked to drought or economic hardship.

The counties recording the highest rates of pregnant female slaughter events are: Bungoma (n=67, 7.5%); Kiambu (n=119, 5.9%); Kajiado (n=36, 2.8%); Baringo (n=1, 0%); Embu (n=2, 0%); Garissa (n=1, 0%); Homa Bay (n=3, 0%); Isiolo (n=2, 0%); Marsabit (n=1, 0%); Nairobi (n=1, 0%); Uasin Gishu (n=1, 0%).

3 Post-Mortem Findings (Carcass Condemnation)

Post-mortem inspection is the last line of food safety defence. A condemned carcass is one deemed entirely unfit for human consumption. The condemnation rate is the primary food safety performance indicator in this dataset.

Data confidence: carcasses_condemned and total_condemned are 100% complete; condemnation_rate_pct and slaughtered_approved are 98.7% complete.

3.1 National Condemnation Summary

Metric	Value
Total Carcasses Slaughtered and Approved	1,286,563.0
Total Carcasses Fully Condemned	3,440.0
National Full Condemnation Rate (%)	0.27%
Mean Facility-Level Condemnation Rate (%)	3.91%
Inspection Events with Partial Condemnation	21.0

The national full carcass condemnation rate stands at 0.27% , meaning 3,440 of 1,286,563 slaughtered carcasses were condemned as entirely unfit for human consumption.

3.2 Partial Condemnation Analysis

Data confidence:(100% complete) partial_condemnation is 100% complete. Note: partial_condemn_reason is only 7.7% complete and is not mapped or summarised as an estimate.

Proportion of inspection events recording a partial condemnation, by county. Partial condemnation referes to where specific organs or body parts are condemned rather than the whole carcass.

4 Disease Flag Analysis

The binary disease flag columns are 100% complete and represent the most analytically reliable layer of the entire dataset. Each flag records whether a specific condition was detected during inspection. Note: the associated quantitative count columns (e.g. septicaemia_no, c_bovis_no) are 0.4–5.6% complete and are not used for inference, this is something to be looked into. Is it a data quality problem?

Data confidence: All binary flag columns are 100% complete.

4.1 National Condition Prevalence

Proportion of inspection records where each condition was flagged.

Across the full dataset, 12.39% of inspection records flagged at least one condition. The most commonly detected disease conditions are emaciation (5.56% of records) and septicaemia (0.43% of records). C. bovis was detected in 3.42% of records , a figure that carries direct public health significance given its route of human infection.

4.2 C. bovis : Zoonotic Risk Priority

Cysticercus bovis is the larval stage of Taenia saginata, a tapeworm that infects humans through consumption of undercooked beef containing viable cysts. Every detection at an abattoir is a direct public health event, not merely an animal health indicator.

Proportion of inspection records where C. bovis was detected, by county. This is a direct proxy for human Taenia saginata infection risk through beef consumption.

Nationally, C. bovis was flagged in 3.42% of all inspection records. Given the 100% completeness of this flag, this figure is reliable at both national and county level. Counties with sustained detection rates above the national average of 3.42% should be treated as active zoonotic risk zones and are candidates for: county public health notification, source farm traceability investigations using the source_of_animals field, and targeted cattle deworming and sanitation programmes.

5 Organ Condemnation Analysis

Organ condemnation data is available at two confidence levels: organs_condemned_flag and total_organs_condemned are 88.5% complete and provide reliable headline figures. Specific organ counts (liver_condemned_n at 48.7%, lung_condemned_n at 39.3%) are valid for national estimates.

5.1 Organ Condemnation Summary

Organ condemnation summary.
Organ	Completeness	Observed value	Estimated true total
Any organ condemned (flag)	88.5%	88.5% of records	N/A or TRUE/FALSE
Total organs condemned (all types)	88.5%	2,973	3,359
Liver condemned (count)	48.7%	1,721	3,534
Lung condemned (count)	39.3%	282	718
Heart condemned (count)	5.6%	957	Unreliable (<10% complete)
Kidneys condemned (count)	1.7%	7	Unreliable (<10% complete)

Organs condemned flag data (88.46% of inspection events recording at least one condemned organ) represents the most reliable organ-level figure available.

5.2 Organ Condemnation by Species

Data confidence: organs_condemned_flag (88.5%) cross-tabulated with species_clean (100%).

Proportion of inspection records flagging any organ condemnation, by species. Based on the 88.5% complete organs_condemned_flag.

6 Temporal Analysis

Temporal analysis is restricted to the 59.4% of records carrying a valid reporting date.

Data confidence: The reporting date column is 59.4% complete.

6.1 Sensitivity Check: Dated vs Undated Records

Before presenting temporal trends, it is important to establish whether the 59.4% of records with valid dates are representative of the full dataset. If dated records systematically come from different counties or abattoir types than undated records, the trend analysis reflects a biased sample.

Comparison of key metrics between dated (59.4%) and undated (40.6%) records. If bars are substantially different, temporal trends drawn from dated records may not represent the national picture.

6.2 National Weekly Condemnation Trend

National weekly carcass condemnation rate over time. Solid red line = weekly mean; dashed grey = smoothed trend. Based on dated records only.

The temporal record spans May 27, 2025 to June 03, 2026, comprising 22 distinct weekly periods across the 59.4% of records with valid dates. The sensitivity check above should be consulted before drawing strong conclusions from this trend.

6.3 Seasonal Pattern with Kenya Rainfall Calendar

Mean condemnation rate by month, annotated with Kenya’s rainfall calendar. Seasonal peaks during dry periods may reflect herd liquidation, nutritional stress, or increased animal movement.

7 Abattoir-Level Analysis

Abattoir-level analysis identifies specific facilities requiring regulatory attention. This section uses abattoir (94% complete) combined with condemnation and disease data.

7.1 Condemnation Rate League Table

Top abattoirs by condemnation rate (minimum 5 records). Abattoirs with zero recorded condemnations across all records are flagged separately as potential surveillance gaps.
County	Abattoir	Records	Slaughtered	Condemned	Rate (%)	Mortality (%)	C. bovis (%)
Bungoma	Naitiri Slaughter Slab	11	15	6	40.00	0.00	54.55
Kiambu	Kenchic Ltd	10	1210480	3334	0.28	0.00	0.00
Kajiado	Kiserian Slaughter House	26	1744	2	0.11	3.89	0.00
Kiambu	Juja International Abattoir	108	73038	10	0.01	0.03	0.93
Bungoma	Webuye Slaughter House	52	328	0	0.00	0.00	0.00
Kajiado	Keekonyokie	5	205	0	0.00	0.00	0.00

8 Data Quality Flags

Proportion of records triggering logical consistency flags.

1.3% of records have an internal contradiction where the number of condemned carcasses exceeds the number recorded as slaughtered. These records should not be used in any rate calculation without correction. 40.6% of records carry the missing date flag.

9 Kiserian Slaughterhouse (Pilot Site)

About this section. Kiserian Slaughterhouse (Kajiado County) is the primary pilot site for the meat inspection digitisation and food safety improvement programme. This section provides a detailed facility-level analysis covering data quality, throughput, ante-mortem findings, post-mortem condemnation, disease burden, and organ condemnation, to serve as the baseline against which future programme improvements will be measured.

The Kiserian dataset comprises 26 records from Kajiado County. Of these, 3 records (11.5%) carry a valid reporting date ,meaning 88.5% of Kiserian’s slaughter activity is temporally unanchored. This poses a significant challenge for any time-series analysis or trend monitoring.

9.1 Statistical Caution: Baseline Validity

Before presenting Kiserian-specific metrics, it is important to establish the statistical reliability of findings based on 26 records, of which only 3 are dated.

Warning

Baseline validity. With 26 total records and 3 dated records, the statistics below should be treated as indicative baseline estimates, not definitive performance measurements. The 95% confidence interval on Kiserian’s condemnation rate of 0.11% spans 0.03% to 0.42%.

9.2 Data Quality at Kiserian

Field completeness for Kiserian records compared with the national average. Fields where Kiserian performs worse than the national average by more than 10 percentage points are priority improvement targets for the pilot.

9.3 Throughput Overview

Data confidence: total_presented, carcasses_condemned, total_condemned (100%); slaughtered_approved, number_dead, number_rejected (98.7–99.6%).

Stage	Metric	Value
Ante-mortem	Total animals presented	1,775
	Mean mortality rate (%)	3.89%
	Mean rejection rate (%)	3.89%
General	Inspection records	26
	Slaughtered and approved	1,744
	Dead before slaughter	2
	Rejected at ante-mortem	2
	Isolated (suspect)	6
Post-mortem	Carcasses fully condemned	2
	Full condemnation rate (%)	0.11%
	Events with partial condemnation (%)	11.54%

Kiserian’s full condemnation rate stands at 0.11% (95% CI: 0.03%–0.42%). The mean ante-mortem mortality rate of 3.89% is the most concerning single metric at this facility. The national mean mortality rate is 2.94% — Kiserian’s rate warrants investigation of transport conditions and lairage management.

9.4 Ante-Mortem Benchmark

Metric	Kiserian	National mean	National Q1 (best 25%)	National Q3 (worst 25%)	N facilities in benchmark
Mortality rate (%)	3.89	0.65	0	0.02	6
Rejection rate (%)	3.89	12.94	0	15.40	6
Note:
Benchmark based on 6 abattoirs with 5+ records nationally. Q1 = best-performing quartile; Q3 = worst-performing quartile.

Kiserian’s mortality rate of 3.89% is benchmarked against 6 abattoirs nationally with five or more records. The national Q3 threshold (worst-performing 25% of facilities) is 0.02%.

9.5 Condemnation Rate Over Time

Daily carcass condemnation rate at Kiserian over the available dated period. Dashed line = national mean.

Based on 3% dated records (11.5%of Kiserian total). 88.5% of Kiserian records have no date and cannot be placed in this timeline.

9.5.1 Organ Condemnation at Kiserian

Data confidence: MODERATE liver condemned counts are ~47% complete nationally; lung ~37%. The analysis below is valid but represents a subset of Kiserian records. The organs_condemned_flag (88% complete) and total_organs_condemned (88% complete) provide the higher-confidence headline numbers.

Organ field	Completeness at Kiserian (%)	Status
Liver Condemned N	88.5	Usable
Lung Condemned N	69.2	Partial
Heart Condemned N	3.8	Unreliable
Total Organs Condemned	100.0	Usable
Organs Condemned Flag	100.0	Usable

Total organs condemned by type at Kiserian Slaughter House.

9.6 Disease Flag Profile : Kiserian vs National

Data confidence:(100% complete). All binary disease flags are fully populated.

Prevalence of each flagged condition at Kiserian (bars) versus national average (grey dots). Saturated colours indicate Kiserian exceeds the national rate.

9.6.1 Species Breakdown

Breakdown of slaughter records by species at Kiserian.

9.6.2 Sex Composition and Pregnant Female Flags

9.7 C. bovis at Kiserian

Warning

C. bovis at Kiserian: C. bovis was detected in 0% of Kiserian inspection records, compared with a national average of 3.42%.

Cysticercus bovis is the larval stage of Taenia saginata, a tapeworm infecting humans through consumption of undercooked beef containing viable cysts. Every positive detection constitutes a direct public health event requiring notification. Recommended actions where Kiserian’s rate is above the national average:

Trace the source farms from the source_of_animals field and schedule inspections
Consider a targeted deworming and sanitation programme for source herds

Monthly C. bovis detection rate at Kiserian.

9.8 Pilot Improvement Scorecard

This scorecard summarises Kiserian’s current performance against placeholder programme targets.

Category	Indicator	Kiserian (current)	Target	Status
Food Safety	Full carcass condemnation rate (%)	0.1	<2%	On target
	Ante-mortem mortality rate (%)	3.9	<1%	Needs attention
	Ante-mortem rejection rate (%)	3.9	<5%	On target
	Records with partial condemnation (%)	11.5	Monitor	Monitor
Zoonotic Risk	C. bovis detection rate (%)	0.0	<0.5%	On target
Data Quality	Records with valid reporting date (%)	11.5	>90%	Needs attention
	Records with condemned > slaughtered (%)	0.0	<1%	On target
	Records with missing date flag (%)	88.5	<5%	Needs attention
	Liver condemned field completeness (%)	88.5	>80%	On target
	Lung condemned field completeness (%)	69.2	>80%	Needs attention