Risk Factors for Mortality Associated with Respiratory Disease among Menz and Horro Sheep in Ethiopia

Risk Factors for Mortality Associated with RespiratoryDisease among Menz and Horro Sheep in Ethiopia

M. TIBBO�, E. MUKASA-MUGERWA�§, M. WOLDEMESKEL�{ and J.E.O. REGE�

�International Livestock Research Institute (ILRI), Animal Genetic Resources, PO Box 5689, Addis Ababa, Ethiopia; �Departmentof Pathology and Parasitology, Faculty of Veterinary Medicine, Addis Ababa University, P.O. Box 34, Debre Zeit, Ethiopia

SUMMARY

A retrospective case-control study was conducted on 6718 sheep of two breeds (2772 Horro and 3946 Menz)on risk factors for mortality associated with respiratory disease (MARD) in Ethiopia, based on data collectedbetween October 1993 and December 1997. Potential risk factors examined were breed, gender, age, month,and air temperatures.

Fifty-four per cent of total deaths in the flock could be attributed to respiratory disease and the annual MARDrate ranged between 6.3 and 19.0%. There was significant breed (P < 0:0001) and gender (P < 0:0001)difference in MARD. The Horro breed had a higher (P < 0:0001) annual MARD than the Menz breed(16:5 � 0:18 vs. 12:4 � 0:15%). A higher (P < 0:0001) proportion of males suffered than females(15:1 � 0:23% vs. 13:8 � 0:13%). Age was also an important risk factor for MARD: there was a strong poly-nomial relationship (R2 ¼ 0:91, P < 0:0001) between MARD and age; the risk of being young if a sheep was aMARD case was high. MARD was high between October and March but relatively low between the months ofApril and September. There was a significant (P < 0:05) negative cubic relationship (R2 ¼ 0:49) betweenmonthly MARD and monthly average minimum air temperatures. There was also a significant (P < 0:01)positive exponential relationship (R2 ¼ 0:61) between monthly MARD and average monthly daily deviationbetween maximum and minimum air temperatures.

Timely health and management interventions focusing on these factors are necessary to alleviate lossesfrom MARD. Understanding variations in MARD risk within a population can enhance early response topotential outbreaks, reducing losses.

� 2002 Elsevier Science Ltd. All rights reserved.

KEYWORDS:EYWORDS: Epidemiology; respiratory disease; mortality; risk factors; sheep; Ethiopia.

INTRODUCTION

The Ethiopian sheep population is estimated at 21.7million (FAO, 1994), three-quarters of which arelocated in the highland region, more than 1500 mabove sea level. Sheep are a major source of animalprotein for the human population but current off-take rates for this purpose are constrained by nu-trition and disease factors. Depending on the level

of management, annual mortality rates may rangefrom 7 to 44% (Agyemang et al., 1985; Njau et al.,1988; Bekele et al., 1992; Mukasa-Mugerwa et al.,2000; Tibbo et al., 2001). Moreover, respiratory dis-eases account for up to 54% of overall mortalityrates in the Ethiopian highlands (Mukasa-Mugerwaet al., 2000). Pneumonia has been noted to be themost prominent infectious cause of mortality, bothon-farm and on-station, in Ethiopian highlandsheep (Bekele et al., 1992; Mukasa-Mugerwa et al.,2000). A number of bacterial, viral, parasitic, andmycoplasmal causes have been identified (Bekeleet al., 1992; Mukasa-Mugerwa et al., 2000; Ayeletet al., 2001; Tibbo et al., 2001; F. Roger, personalcommunication). Environmental and/or manage-mental factors can act as stressors and hamper the

The Veterinary Journal 2003, 165, 276–287doi:10.1016/S1090-0233(02)00184-3

1090-0233/02/$ - see front matter � 2002 Elsevier Science Ltd. All rights reserved.

Correspondence to: Dr. MarkosTibboDVM, InternationalLivestock Research Institute (ILRI), Animal Genetic Resources.P.O. Box 5689, AddisAbaba, Ethiopia. Fax: +251-1-46-1252;E-mail: [email protected]

§ Present address: P.O. Box 3987; Kampala, Uganda.{ Present address: Department of Comparative Medicine, Col-

lege of Veterinary Medicine, University of Tennessee, P.O. Box1071, TN 37901-1071, USA.

immune response (Kimberling, 1988) and these,combined with increased exposure to pathogens,may lead to respiratory infection (Rook et al., 1990).An alternative to costly treatments for pneumonia isprevention through adjusting the managementroutine to reduce the risk of disease development(Alley, 1991).

Although losses due to pneumonia have beenwidely reported to be high in Ethiopian highlandsheep (Njau et al., 1988; Bekele et al., 1992; Mukasa-Mugerwa et al., 2000; Ayelet et al., 2001; Tibbo et al.,2001), risk factors predisposing sheep to pneumo-nia have not been systematically studied. This study,therefore, was conducted to: (1) determine mortal-ity associated with respiratory disease (MARD) pat-terns using data collected from a large experiment;and (2) identify and quantify the risk factors likelyto influence animal susceptibility to pneumonia.

MATERIALS AND METHODS

Study area and animalsThe study was conducted at the International Live-stock Research Institute (ILRI) Debre Berhan Re-search Station, located in the central highlands ofEthiopia, 120 km NE of Addis Ababa, at 2780 m al-titude, between 9�360N latitude and 39�380E longi-tude. Average annual rainfall from 1993 to 1997 was896.8 mm, most (75%) of which came during thelong rains (June to September). The short rainsoccur from February to April/May and the long dryseason occurs from October to January. Averagemonthly minimum ambient temperatures duringthe study period ranged from 1.3 �C in November(the lowest record was )8.5 �C in November 1994)to 8.0 �C in April and August. The mean monthlymaximum temperatures ranged from 18.6 �C inAugust to 23.4 �C in May. The mean relative hu-midity was 49.7% during the same period.

The total number of animals studied was 6718(monthly average flock size ranged between 1780 to2693) and consisted of 41% Horro and 59% Menzsheep. The Menz breed is native to the study areaand is distributed between longitude 30� and 40� E,latitude 10� and 11� N and, altitude 2500–3000 mabove sea level. The Horro breed was introducedinto the station from its natural mid-altitude regionof 35–38� E and 6–10� N at 1400–2000 m above sealevel (Galal, 1983).

Sheep management and healthDuring the day, animals grazed natural pastureslargely dominated by Andropogon longipes grass with

varying proportions of Trifolium spp. They wererotated among five paddocks totalling 138 ha. Allanimals had free access to hay at night in properlysheltered pens. In addition, 200 g head�1 day�1 ofconcentrate – a mixture of wheat bran (66%), Nougcake (Guizotia abyssinica) (33%), and salt (1%),containing 30–35% crude protein – was fed to ewes.The allowance was increased to 400 g head�1 day�1

during the third trimester of pregnancy. Water andmineral lick were available ad lib. The lambs had noaccess to feed other than what was provided to theirdams until weaning at 90 days (Rege et al., 2002).

A substantial proportion of the animals studiedhere was part of a large experiment on the station inwhich animals were monitored for genetic resis-tance to gastro-intestinal parasites (Rege et al.,2002). The breeding of animals for genetic resis-tance to gastrointestinal parasites in small rumi-nants at the station was controlled: mating wasplanned mainly for dry season lambing to occurbetween October and December and wet seasonlambing between June and August. However, a largenumber of lambs, progeny of ewes not on the mainexperiment, including the station �multiplicationflock,� was also born in other months of the year.The present study looked at all sheep on the stationduring the study period.

As part of general flock health management, ani-mals were drenched for trematodes with oxycloza-nide at 15mg kg�1 body weight (Zanil, CoopersAnimal Health) in August and triclabendazole12mg kg�1 body weight (Fasinex 10%, Ciba Geigy) inNovember, December, and January of each year.They were also drenched against nematodes with ei-ther fenbendazole 5mg kg�1 body weight (Panacur,Hoechst) or levamisole HCl 7:5mg kg�1 body weight(Nilverm Super, Coopers Animal Health). The twodrenches were used in alternate years to avoid thedevelopment of nematode resistance to the ant-helminthics. Lambs were also treated against tape-worms with niclosamide whenever this was diagnosed.

The drenching regime for animals in the studyexamining genetic resistance to gastro-intestinalparasites in small ruminants was based on wormeggs per gram (epg) counts. Faecal samples col-lected monthly were used to determine worm eggcounts and individual lambs with >2000 epg weretreated with fenbendazole 5mgkg�1 body weight(Panacur, Hoechst) or levamisole HCl 7:5mgkg�1

body weight (Nilverm Super, Coopers AnimalHealth) prior to weaning. At weaning, all lambswere drenched. Between weaning and one year ofage, all lambs were drenched only when the mean

RISK FACTORS FOR MORTALITY ASSOCIATED WITH RESPIRATORY DISEASE IN SHEEP 277

faecal count for a group of monitor lambs exceeded2000 epg. The drenching regime against trematodesfor this group of animals was the same as that for theother flocks of the station.

A resident veterinarian performed necropsies onall dead animals to ascertain causes of death. Whena diagnosis could not be made on gross lesions,tissue samples were submitted to the Faculty ofVeterinary Medicine and National Veterinary Insti-tute, Debre Zeit, Ethiopia.

Data collectionA retrospective study was conducted on post mortemrecords of all dead sheep between October 1993 andDecember 1997. Monthly number of cases of MARDand average flock size between October 1993 andDecember 1997 are presented in Table I. The flockbook was the source of information for gender,breed, birth, and mortality dates. Age was calculatedfrom birth dates.

A conservative case definition was used with adeath classified as MARD if written necropsy reportsrecorded that over 25% of lung tissue showed dis-ease involvement and was described as red- and grey-hepatised, consolidated, abscessed, dark-red, orshowing the presence of fibrin or adhesions.

Meteorological information was obtained fromdata collected at the station on standard meteoro-logical data collection sheets during the study period.

Statistical analysesA preliminary comparison of analyses of MARD bycategorical data analysis procedure (CATMOD)

(based on a logit model) and by general linearmodel procedure (GLM), both of the StatisticalAnalysis System (SAS, 1998), yielded nearly identicalresults. Thus, invoking the Central Limit Theorem(Winer, 1971) based on the large data size involved,all the analyses were done by GLM. A dichotomousvariable RISK, coded on individual post mortem casesas 1 (death due MARD) and 0 (for live or dead dueto other causes) as response variable with the fixedeffects of breed, gender, and age was analysed byGLM procedure to generate least squares means(LSM) and standard errors (SE) of MARD. The RISKwas defined as the risk of MARD in any one month.

Data were analysed on monthly records for eachyear. Least square means of annual MARD was ob-tained by calculating from total MARD for the yeardivided by total animal-months at risk by fitting thefixed effect of breed, gender, and age. The numberof animals alive in each month was determined byadding new entries (either as live births or pur-chases) to the total live animals and subtracting thenumber of animals dying and slaughtered duringthe previous month according to Putt et al. (1988).Age category (66, >6 to 612, >12 to 624, >24 to636, and >36 months) was obtained as a differencebetween mortality and birth dates for animals thatdied in a particular month or as a difference be-tween adjusted date (the last date of that month)and birth date for live animals.

Odds ratios (OR) computed using LOGISTICprocedure of Statistical Analysis System (SAS, 1998)were employed to determine case-control OR forestimating the level of risk for breed, gender and

Table IMonthly number of cases of MARD and average flock size between October 1993 and December 1997

Month Year

1993 1994 1995 1996 1997

January 72/2143 56/2039 46/2264 33/2400

February 17/2020 42/1937 50/2176 35/2270

March 17/1973 22/1909 38/2108 30/2240

April 13/1877 18/1941 12/2001 25/2143

May 14/1780 24/1847 11/2035 21/2055

June 25/2151 32/2468 10/2512 9/2195

July 27/2098 33/2418 16/2526 10/2318

August 8/2096 17/2354 19/2462 9/2264

September 20/2086 28/2328 22/2389 13/2214

October 14/2109 10/2133 20/2333 54/2468 45/2176

November 8/2268 24/2266 49/2539 141/2693 25/2023

December 37/2216 39/2197 66/2394 43/2495 15/1944

Numerators are MARD cases whereas denominators are flock size in the month.

278 THE VETERINARY JOURNAL, 165, 3

age subclasses. OR, the ratio of the odds in favour ofexposure among cases to the odds in favour of ex-posure among non-cases (Last, 1988), as applied byYapi et al. (1990), was computed for animals of agiven subclass. OR with a value equivalent to 1.0 wasconsidered to show no association between eachexposure category and reference category.

Relationships of MARD with months of a year,mean age, average monthly minimum air tempera-ture, and monthly average deviation of daily mini-mum temperature from maximum temperature

were obtained by regressing average monthly MARDon these factors.

RESULTS

MARD as a proportion of total deaths accounted for54.4% as depicted in Table II. Least squares meansof annual MARD rate was 14:4 � 0:13 ranging be-tween 6.3 and 19.0% (Table IV).

The overall monthly MARD (%) by year is pre-sented in Table III. MARD (%) was high betweenOctober and March (range: 0.3–4.6%) whereas rel-atively low between April and September (range:0.5–1.9%) (Table V–IX). Fig. 1 clearly shows thistrend. The pattern of MARD had a U-shape, withhigh levels in the early part of the year, a substantialreduction between the months of April and Sep-tember followed by another steep increase in thelast third of the year.

Breed differencesAnnual MARD are presented in Table IV. Horrobreed had higher (P < 0:0001) overall annualMARD than the Menz breed (16:5 � 0:18 vs.12:4 � 0:15%). The annual MARD was significantlydifferent by breed except for year 1994.

Monthly MARD (Tables V–IX) also indicates thata higher proportion (P < 0:05 to P < 0:0001) ofHorro sheep died from respiratory disease eachmonth than the Menz sheep. For example, monthlyMARD for Horro and Menz was 3.0–5.8% vs. 0.4-3.3% in January, 1.7–4.0% vs. 0.3–1.8% in October,2.2–6.5% vs. 1.1–2.2% in November, and 2.0–4.6%vs. 1.0–2.7% in December, respectively.

Table IIMARD as a proportion of all deaths (%) in the flock

during the 4.25 years

Source All deaths

(%)

MARD as a proportion

of all deaths

Overall 100.0 54.4

Breed

Horro 49.2 25.0

Menz 50.8 29.4

Sex

Female 75.7 40.7

Male 24.3 13.7

Age group (months)

66 34.0 17.0

>6 to 612 12.9 7.8

>12 to 624 16.7 8.6

>24 to 636 12.3 6.6

>36 24.1 14.4

Table IIILeast-square means (±SE) of monthly MARD (%)

Month Year

1993 1994 1995 1996 1997

January 4:58 � 0:46 3:68 � 0:43 2:69 � 0:37 1:70 � 0:29

February 1:23 � 0:25 3:53 � 0:41 2:56 � 0:40 2:05 � 0:32

March 1:25 � 0:26 1:51 � 0:32 2:71 � 0:36 1:62 � 0:31

April 0:60 � 0:25 1:82 � 0:28 1:10 � 0:23 1:57 � 0:30

May 1:05 � 0:29 1:72 � 0:39 0:54 � 0:23 1:57 � 0:30

June 1:29 � 0:28 1:90 � 0:29 0:60 � 0:16 0:53 � 0:18

July 1:66 � 0:29 1:47 � 0:31 0:77 � 0:20 0:47 � 0:18

August 0:48 � 0:16 0:85 � 0:23 0:73 � 0:23 0:89 � 0:17

September 1:43 � 0:26 1:28 � 0:29 0:79 � 0:25 0:68 � 0:22

October 0:89 � 0:24 0:75 � 0:19 0:97 � 0:25 2:00 � 0:38 2:90 � 0:40

November 0:29 � 0:17 1:04 � 0:35 2:14 � 0:40 4:35 � 0:60 1:85 � 0:39

December 2:06 � 0:35 1:97 � 0:33 3:65 � 0:40 2:28 � 0:33 1:50 � 0:28

RISK FACTORS FOR MORTALITY ASSOCIATED WITH RESPIRATORY DISEASE IN SHEEP 279

Gender differencesThere was a significant (P < 0:0001) gender differ-ence in MARD (Table IV). Males had higher annualMARD than females (15:1 � 0:23% vs. 13:8�0:13%). The annual MARD was significantly differ-ent by gender except for year 1996.

Monthly MARD (Tables V–IX) also showed that ahigher proportion of males died from respiratorydisease than females (P < 0:05 to P < 0:0001) rang-ing from 0.4 to 6.7% vs. 0.2 to 3.6%, respectively.

Age variationThe annual MARD was 24.1%, 15.4%, 10.1%, 10.8%,and 11.7% for age classes 66, >6 to 612, >12 to

624, >24 to 636, and >36 months, respectively(Table IV).

Monthly MARD for each age category is given inTables V–IX. Monthly MARD was higher (P < 0:05to P < 0:0001) in lambs younger than six monthsthan in older animals.

The highly significant (P < 0:0001) relationshipbetween average monthly MARD and age had apolynomial form, the linear component of whichhad a negative slope (R2 ¼ 0:91, Fig. 2). As can bededuced from the tables and the figure, themonthly risk decreased with age up to 2.5 years. Aslight increase in monthly MARD was observed afterthe age of 3.5 years and remained insignificant forolder animals.

Effects of air temperatureThere was a negative correlation (r ¼ �0:59,P < 0:01) between average monthly MARD andmonthly average minimum air temperature. Therelationship between the two was cubic the qua-dratic component having a negative slope(R2 ¼ 0:49, Fig. 3). In addition, there was a strongpositive exponential relationship (R2 ¼ 0:61,P < 0:01) between average monthly MARD andmonthly average daily deviation of minimum tem-perature from maximum temperature (Fig. 4) witha strong positive correlation (r ¼ 0:69, P < 0:001)between the two.

Fig. 1. Average monthly MARD pattern by months of ayear over the 4.25 years study period.

Table IVLeast-square means (±SE) of annual MARD (%) by major risk factors between 1994 and 1997

Year Overall Breed Sex

Horro Menz Female Male

1994 18.98� 0.30 18.54� 0.47a 19.41� 0.37a 19.72� 0.34a 18.23� 0.52b

1995 17.58� 0.29 19.95� 0.40a 15.21� 0.34b 16.00� 0.27b 19.16� 0.51a

1996 14.41� 0.24 17.64� 0.34a 11.17� 0.29b 14.13� 0.25a 14.68� 0.43a

1997 6.29� 0.17 8.05� 0.24a 4.53� 0.19b 5.27� 0.18b 7.31� 0.30a

Overall 14.43� 0.13 16.49� 0.18a 12.37� 0.15b 13.80� 0.13b 15.07� 0.23a

Age group (months)

66 >6 to 6 12 >12 to 624 >24 to 636 >36

1994 26.93� 0.54a 21.37� 0.63b 13.50� 0.54e 17.59� 0.69c 15.49� 0.65d

1995 25.22� 0.47a 18.26� 0.60b 13.69� 0.53d 13.50� 0.69d 17.23� 0.50c

1996 29.78� 0.44a 12.65� 0.53b 7.96� 0.45d 10.70� 0.59c 10.94� 0.41c

1997 12.01� 0.32a 6.71� 0.40b 4.45� 0.34c 3.44� 0.34d 4.83� 0.26c

Overall 24.12� 0.23a 15.41� 0.28b 10.14� 0.24d 10.77� 0.30d 11.71� 0.22c

a;b;c;d;eWithin variable row means with common superscript do not differ (P > 0:05).

280 THE VETERINARY JOURNAL, 165, 3

Table VLeast-square means (±SE) of monthly MARD (%) by major risk factors for year 1993

Month Breed Sex

Horro Menz Female Male

October 0:13 � 0:40b 1:66 � 0:26a 0:45 � 0:24b 1:34 � 0:38a

November 0:31 � 0:28a 0:26 � 0:19a 0:21 � 0:18a 0:36 � 0:26a

December 2:40 � 0:60a 1:71 � 0:37a 1:11 � 0:39b 3:01 � 0:55a

Age group (months)

66 >6 to 612 >12 to 6 24 >24 to 636 >36

October 0:09 � 0:31d 3:38 � 0:64a 1:85 � 0:37b 0:65 � 0:57c 0:00 � 0:046e

November 0:40 � 0:22a 0:02 � 0:47a 0:72 � 0:26a 0:03 � 0:47a 0:26 � 0:32a

December 3:49 � 0:60a 2:91 � 0:62b 1:08 � 0:57d 0:88 � 1:07d 1:92 � 0:69c

a;b;c;d;eWithin variable row means with common superscript do not differ (P > 0:05).

Table VILeast-square means (±SE) of monthly MARD (%) by major risk factors for year 1994

Month Breed Sex

Horro Menz Female Male

January 5:82 � 0:76a 3:33 � 0:52b 2:48 � 0:48b 6:68 � 0:77a

February 2:01 � 0:40a 0:45 � 0:29b 0:72 � 0:25b 1:74 � 0:43a

March 1:44 � 0:42a 1:06 � 0:30a 0:65 � 0:26b 1:85 � 0:45a

April 0:47 � 0:40a 0:73 � 0:29a 0:70 � 0:24a 0:50 � 0:44a

May 1:14 � 0:45a 0:96 � 0:34a 0:81 � 0:30a 1:28 � 0:49a

June 1:76 � 0:43a 0:82 � 0:35a 1:16 � 0:28a 1:41 � 0:48a

July 1:18 � 0:46a 2:13 � 0:37a 0:98 � 0:30b 2:34 � 0:51a

August 0:41 � 0:25a 0:55 � 0:21a 0:21 � 0:16a 0:74 � 0:29a

September 1:89 � 0:41a 0:97 � 0:34a 0:91 � 0:27a 1:94 � 0:46a

October 0:73 � 0:28a 0:77 � 0:23a 0:53 � 0:19a 0:97 � 0:32a

November 1:27 � 0:47a 0:81 � 0:39a 0:66 � 0:36a 1:42 � 0:50a

December 2:30 � 0:51a 1:64 � 0:42a 1:75 � 0:34a 2:19 � 0:58a

Age group (months)

66 >6 to 6 12 >12 to 624 >24 to 636 >36

January 8:61 � 0:88a 7:08 � 0:87b 1:32 � 0:87d 2:34 � 1:08c;d 3:55 � 0:97c

February 3:22 � 0:46a 1:06 � 0:47b;c 0:44 � 0:48c 0:08 � 0:58d 1:35 � 0:52b

March 1:99 � 0:54a 1:33 � 0:45a 1:33 � 0:50a 0:77 � 0:59a 0:87 � 0:53a

April 1:35 � 0:58a 1:41 � 0:39a 0:05 � 0:50a 0:00 � 0:52a 0:31 � 0:49a

May 2:09 � 0:91a 1:01 � 0:40a 0:44 � 0:52a 0:47 � 0:57a 1:23 � 0:54a

June 2:42 � 0:62a 1:04 � 0:48b 0:20 � 0:68d 0:75 � 0:60c 2:42 � 0:48a

July 2:38 � 0:51b 3:90 � 0:66a 0:77 � 0:51c 0:91 � 0:71c 0:32 � 0:64d

August 0:87 � 0:28a 0:38 � 0:36a 0:70 � 0:29a 0:28 � 0:40a 0:16 � 0:36a

September 0:86 � 0:45b;c 3:57 � 0:64a 0:99 � 0:43b;c 0:07 � 0:64d 1:65 � 0:56b

October 0:33 � 0:27a 1:92 � 0:55a 0:68 � 0:31a 0:20 � 0:39a 0:60 � 0:40a

November 1:26 � 0:36a 0:11 � 1:36a 1:24 � 0:45a 0:69 � 0:56a 1:89 � 0:60a

December 2:46 � 0:63ab 3:38 � 0:70a 1:42 � 0:73b 0:49 � 0:73c 2:08 � 0:78a;b

a;b;c;d;eWithin variable row means with common superscript do not differ (P > 0:05).

RISK FACTORS FOR MORTALITY ASSOCIATED WITH RESPIRATORY DISEASE IN SHEEP 281

DISCUSSION

Respiratory disease mortality was the most impor-tant infectious disease problem in the flock ac-counting for 14.4% of annual deaths, its share of thetotal mortality being 54.4%. Bekele et al. (1992)found MARD to account for 20.9% of the overallmortality in sheep of all age classes in the highlandsof Ethiopia. Mukasa-Mugerwa et al. (2000) reported54% of the overall mortality to be attributable toMARD in lambs up to one year of age in the samearea. Mourad et al. (2001) ascribed 23% of the lambmortality to pneumonia and diarrhoea in Djallonk�eesheep in Guinea.

There was a significant breed difference inMARD, the Horro being more susceptible than theMenz (Tables IV and X). This is in agreement with

the previous report by Mukasa-Mugerwa et al. (2000)in lambs under one year of age that more Horrothan Menz lambs died from pneumonia. While bothbreeds studied are indigenous to Ethiopia, only theMenz sheep are native to the study area. The highersusceptibility of the Horro sheep at the altitude(2780 m above sea level) of the present study is mostlikely a result of ill-adaptation to higher cold stressat this altitude compared to its natural mid-altitude(1400–2000 m above sea level) habitat in the westernpart of Ethiopia. Moreover, the Menz sheep havecoarse wool whereas the Horro have only short hair.The short hair coat of Horro sheep is a disadvantagein a relatively cold climate of the higher altitude ofthe study area, and this might have predisposedthem to pneumonia. Between-breed differences in

Table VIILeast-square means (±SE) of monthly MARD (%) by major risk factors for year 1995

Month Breed Sex

Horro Menz Female Male

January 4:97 � 0:62a 2:40 � 0:54b 2:34 � 0:44b 5:04 � 0:78a

February 4:05 � 0:57a 3:01 � 0:49a 1:86 � 0:41b 5:20 � 0:43a

March 1:71 � 0:43a 1:31 � 0:38a 0:97 � 0:30a 2:05 � 0:56a

April 1:91 � 0:38a 1:73 � 0:34a 0:71 � 0:27b 2:93 � 0:50a

May 2:18 � 0:49a 1:25 � 0:48a 0:87 � 0:34b 2:56 � 0:68a

June 1:93 � 0:39a 1:87 � 0:35a 1:06 � 0:27b 2:75 � 0:50a

July 1:70 � 0:41a 1:23 � 0:36a 1:05 � 0:29a 1:88 � 0:53a

August 0:86 � 0:31a 0:84 � 0:27a 0:36 � 0:22b 1:34 � 0:39a

September 1:77 � 0:40a 0:79 � 0:35b 1:11 � 0:28a 1:44 � 0:51a

October 1:65 � 0:33a 0:29 � 0:30b 0:45 � 0:24b 1:49 � 0:43a

November 3:18 � 0:49a 1:10 � 0:49b 1:37 � 0:39b 2:91 � 0:63a

December 4:63 � 0:57a 2:67 � 0:48b 2:84 � 0:40a 4:45 � 0:71a

Age group (months)

6 6 > 6 to 6 12 > 12 to 6 24 > 24 to 6 36 > 36

January 5:10 � 0:82a 5:18 � 0:89a 3:33 � 0:78a 2:37 � 1:03a 2:44 � 0:84a

February 6:79 � 0:86a 4:53 � 0:76b 1:93 � 0:70d 2:70 � 1:04c 1:70 � 0:71d

March 2:24 � 0:64a 1:84 � 0:57a 1:16 � 0:53a 0:97 � 0:82a 1:34 � 0:53a

APril 4:59 � 0:63a 0:93 � 0:45b 1:27 � 0:49b 1:03 � 0:72b 1:28 � 0:46b

May 0:97 � 1:09a 2:62 � 0:49a 2:63 � 0:70a 1:02 � 0:81a 1:33 � 0:56a

June 1:69 � 0:42b 4:13 � 0:64a 1:30 � 0:48b 0:20 � 0:68c 0:75 � 0:60c

July 2:91 � 0:43a 1:59 � 0:71b 1:42 � 0:59b 0:68 � 0:70c 0:73 � 0:53c

August 1:25 � 0:32a 0:08 � 0:61a 1:28 � 0:40a 0:75 � 0:51a 0:89 � 0:39a

September 2:58 � 0:42a 2:02 � 0:75b 0:31 � 0:53d 0:42 � 0:64d 1:07 � 0:50c

October 2:03 � 0:38a 0:08 � 0:67d 0:88 � 0:43c 0:39 � 0:56c 1:51 � 0:41b

November 4:29 � 0:45a 2:45 � 1:34b 1:35 � 0:65c 1:27 � 0:83c 1:34 � 0:60c

December 10:98 � 0:81a 2:26 � 0:70b 1:59 � 0:77c 1:30 � 0:98c 2:10 � 0:72b

a;b;c;d;eWithin variable row means with common superscript do not differ (P > 0:05).

282 THE VETERINARY JOURNAL, 165, 3

MARD have also been reported in studies elsewhere(Yapi et al., 1990; Nash et al., 1997).

That male sheep had higher MARD than femalesis consistent with results of other studies. Mukasa-Mugerwa et al. (2000) found OR of 1.32 while Nashet al. (1997) reported OR of 1.9 for male vs. femalelambs for MARD and concluded that males weremore susceptible to MARD than females. Norwak(1990) reported, in twin-born lambs, that recogni-tion ability of dams by female lambs is superior totheir male counterparts after separation. In largeflocks, where group mating and lambing occurs, aswas the case in the present study, with births of up to450 lambs a month, starvation–missmothering–exposure complex has been reported to be animportant cause of lamb loss (Mukasa-Mugerwa etal., 2000). Under these circumstances even a small

advantage by female lambs to recognize their damsthan their male counterparts might have contrib-uted to the observed difference. Moreover, severalreports indicate that birth weight as a risk factor forlamb survival has an inverted U-shaped relationship(Meyer & Clarke, 1978; Dalton et al., 1980; Scales etal., 1986; Yapi et al., 1990; Mukasa-Mugerwa et al.,1994) implying that lamb mortality is expected to behigh in heavier lambs (possibly males) at birth forunexplained reasons.

Differences amongst age groups in MARD ex-hibited similar patterns to those reported in otherstudies (Rook et al., 1990; Nash et al., 1997), andMARD was, in general, highest in sheep underone year of age and decreased as age increased.Consequently, a strong polynomial relationship(R2 ¼ 0:91, P < 0:0001) between MARD and age

Table VIIILeast-square means (±SE) of monthly MARD (%) by major risk factors for year 1996

Month Breed Sex

Horro Menz Female Male

January 3:70 � 0:52a 1:69 � 0:43b 2:01 � 0:36b 3:37 � 0:65a

February 3:16 � 0:57a 1:96 � 0:46a 3:08 � 0:40a 2:04 � 0:70a

March 3:48 � 0:51a 1:95 � 0:41b 2:51 � 0:36b 2:92 � 0:62a

April 1:58 � 0:31a 0:61 � 0:27b 0:87 � 0:23a 1:32 � 0:40a

May 0:60 � 0:30a 0:49 � 0:27a 0:23 � 0:24a 0:85 � 0:38a

June 0:78 � 0:22a 0:43 � 0:19a 0:24 � 0:17b 0:97 � 0:26a

July 0:91 � 0:28a 0:63 � 0:24a 0:62 � 0:22a 0:92 � 0:34a

August 1:06 � 0:31a 0:40 � 0:26a 1:09 � 0:24a 0:36 � 0:38a

September 0:61 � 0:35a 0:98 � 0:29a 0:66 � 0:28a 0:93 � 0:42a

October 2:58 � 0:52a 1:41 � 0:45a 2:14 � 0:42a 1:85 � 0:63a

November 6:48 � 0:77a 2:22 � 0:69b 3:62 � 0:65a 5:07 � 0:91a

December 2:87 � 0:46a 1:68 � 0:40b 2:22 � 0:35a 2:34 � 0:58a

Age group (months)

66 >6 to 612 >12 to 6 24 >24 to 636 >36

January 5:17 � 0:81a 4:53 � 0:62b 1:60 � 0:69c 0:88 � 0:85d 1:28 � 0:64c

February 8:12 � 0:92a 2:95 � 0:67b 0:89 � 0:77c 0:08 � 0:88d 0:91 � 0:68c

March 8:79 � 0:86a 3:39 � 0:61b 0:41 � 0:67c 0:47 � 0:78c 0:51 � 0:61c

April 3:03 � 0:42a 1:47 � 0:38b 0:34 � 0:42c 0:24 � 0:49c 0:42 � 0:49c

May 0:12 � 0:71a 0:98 � 0:33a 0:20 � 0:40a 0:65 � 0:49a 0:98 � 0:33a

June 0:47 � 0:25a 1:01 � 0:27a 0:41 � 0:27a 0:28 � 0:40a 0:85 � 0:26a

July 0:61 � 0:31a 1:15 � 0:59a 1:06 � 0:34a 0:50 � 0:50a 0:53 � 0:33a

August 1:82 � 0:35a 1:33 � 0:68a 0:46 � 0:39b 0:02 � 0:51c 0:01 � 0:37c

September 1:56 � 0:40a 0:46 � 0:73a 0:90 � 0:43a 0:10 � 0:61a 0:95 � 0:40a

October 5:66 � 0:55a 1:54 � 1:17b 0:96 � 0:68c 1:24 � 0:91b;c 0:57 � 0:60d

November 13:58 � 0:70a 2:05 � 2:11b;c 1:95 � 0:99b;c 2:75 � 1:33b 1:40 � 0:87c

December 3:72 � 0:61a 3:58 � 0:74a 0:70 � 0:62c 2:53 � 0:84b 0:86 � 0:55c

a;b;c;d;eWithin variable row means with common superscript do not differ (P > 0:05).

RISK FACTORS FOR MORTALITY ASSOCIATED WITH RESPIRATORY DISEASE IN SHEEP 283

was obtained (Fig. 2). Mukasa-Mugerwa et al. (2000)reported decreasing trend in overall mortality: 20%,18%, 10%, and 5% for lambs below 3, 3–6, 6–9, and9–12 months of age, respectively. That the rela-tionship between overall mortality and age in thestudy by Mukasa-Mugerwa et al. (2000) was similar tothat between age and MARD in the present studywas not surprising given the fact that MARD ac-counted for a large part of mortality in both studies.Although early immune response to infections byyoung animals is variable, resistance in young ani-mals is generally known to be low because theirimmune system is underdeveloped (Warner et al.,1987; Eide et al., 1992; Lunden, 1995). This explainsthe higher MARD in young animals. Appropriateflock health management requires particular care inthe high-risk age classes, that is, in lambs below 12

months of age. Further research is needed on theunderlying causes during these periods with a viewof minimising losses ascribed to MARD.

Month was an important risk factor for MARD(Table III and Fig. 1). Two distinctive MARD pat-terns were observed: high MARD in the typicallycold and dry months (i.e. October to March) and alow MARD during warm and semi-dry (April toMay), and warm and wet (i.e. June to September)months. The higher MARD recorded between Oc-tober and March was, at least in part, due to thehigh number of cases of pneumonia in the dryseason. In particular, the high MARD in the twomonths of the short rainy season (February andMarch) mirrors the high number of cases in late dryseason, which subsequently died in these monthsafter protracted clinical course of pneumonia

Table IXLeast-square means (±SE) of monthly MARD (%) by major risk factors for year 1997

Month Breed Sex

Horro Menz Female Male

January 3:03 � 0:42a 0:37 � 0:34b 1:91 � 0:31a 1:49 � 0:51a

February 2:41 � 0:45a 1:68 � 0:38a 1:57 � 0:34a 2:52 � 0:57a

March 2:12 � 0:43a 1:12 � 0:36a 1:78 � 0:32a 1:46 � 0:54a

April 2:34 � 0:42a 0:80 � 0:35b 1:30 � 0:30a 1:84 � 0:54a

May 2:60 � 0:41a 0:55 � 0:35b 1:51 � 0:30a 1:63 � 0:54a

June 0:61 � 0:25a 0:45 � 0:20a 0:46 � 0:20b 0:61 � 0:31a

July 0:42 � 0:25a 0:53 � 0:20a 0:48 � 0:20a 0:46 � 0:30a

August 1:09 � 0:25a 0:69 � 0:19a 0:36 � 0:19b 1:41 � 0:29a

September 1:10 � 0:31a 0:26 � 0:24b 0:94 � 0:24a 0:41 � 0:37a

October 3:96 � 0:58a 1:84 � 0:44b 2:69 � 0:44a 3:11 � 0:69a

November 2:20 � 0:53a 1:50 � 0:40a 1:10 � 0:41b 2:60 � 0:63a

December 2:03 � 0:40a 0:97 � 0:30b 1:24 � 0:30a 1:77 � 0:50a

Age group (months)

6 6 >6 to 612 >12 to 624 >24 to 6 36 >36

January 4:70 � 0:54a 1:38 � 0:65b 0:90 � 0:55c 0:91 � 0:76c 0:61 � 0:48d

February 6:60 � 0:63a 0:36 � 0:69d 0:82 � 0:58c 1:28 � 0:85b 1:17 � 0:51b

March 3:74 � 0:61a 2:55 � 0:67b 0:06 � 0:55d 0:57 � 0:79c 0:66 � 0:48c

April 3:31 � 0:60a 1:53 � 0:61b 0:97 � 0:61c 0:71 � 0:68c 1:32 � 0:46b

May 4:48 � 0:77a 1:40 � 0:51b 0:73 � 0:56c 0:19 � 0:65d 1:44 � 0:44b

June 0:62 � 0:35a 0:61 � 0:49a 0:63 � 0:35a 0:52 � 0:34a 0:31 � 0:27a

July 0:41 � 0:30a 0:51 � 0:49a 0:91 � 0:37a 0:19 � 0:34a 0:32 � 0:27a

August 0:27 � 0:30c 1:31 � 0:48b 1:73 � 0:36a 0:52 � 0:35c 0:59 � 0:26c

September 1:01 � 0:37a 1:32 � 0:59a 0:49 � 0:47a 0:21 � 0:43a 0:37 � 0:32a

October 7:42 � 0:70a 3:69 � 1:11b 1:10 � 0:82c 1:23 � 0:82c 1:05 � 0:59c

November 3:06 � 0:57a 1:69 � 1:25a 1:60 � 0:73a 1:35 � 0:66a 1:56 � 0:48a

December 4:37 � 0:63a 0:29 � 0:57d 1:35 � 0:64b 0:77 � 0:53c 0:73 � 0:39c

a;b;c;d;eWithin variable row means with common superscript do not differ (P > 0:05).

284 THE VETERINARY JOURNAL, 165, 3

ranging from 17 to 34 days (M. Tibbo, unpublisheddata). In addition, the beginning of the short rainyseason (February/March) coincides with the timewhen all grazing is depleted resulting into inade-quate nutrition, which may predispose animals toinfections. Lambs were not supplemented untilweaning except that they shared what was providedfor their dams. This, plus the fact that supplemen-tation regime of dams was not strictly linked toquantitative assessment of nutrition available frompasture, may have inadvertently subjected lambs toundernutrition, thus predisposing them to MARD.Mukasa-Mugerwa et al. (2000) reported that lambsborn in the wet season experienced higher mortalitythan those born in the dry season. This was becauselambs born in the wet season were exposed to thecold weather of the dry season and this, togetherwith weaning stress in the dry season, could havecontributed to the observed MARD pattern in thepresent study. As stated earlier, the dry months(between October and January) are also associatedwith lowest minimum air temperatures at the stationand this could have been a significant predisposingstressor to animals, eventually resulting in highMARD during these months.

Diurnal variation in ambient temperature re-sulted in high MARD (Fig. 4). This is in agreementwith the report by Nash et al. (1997). Furthermore,the contribution of other predisposing factors ofMARD such as increased parasite challenge (e.g.fasciolosis) between the months of October andFebruary (Tibbo, 2000) should be examined. In

Fig. 2. Trend of MARD plotted against mean age(years).

Fig. 3. Relationship between average monthly MARDand average monthly minimum air temperature (�C).

Fig. 4. Exponential relationship between averagemonthly MARD and average monthly diurnal temperaturedifference (�C).

Table XAdjusted odds ratios of risk factors for MARD

Source Odds ratio 95%

confidence

interval

Chi-square,

P-value

Breed 323.5a

Horro vs. Menz 1.38 1.33–1.43

Sex 16.9a

Female vs. Male 0.92 0.88–0.96

Age group ðmonthsÞb 2234.8a

66 2.49 2.37–2.61

>6 to 612 1.42 1.34–1.50

>12 to 624 0.88 0.83–0.93

>24 to 636 0.93 0.87–0.99

>36a Chi-square: P <0.0001.b Each age group compared to the oldest (>36 months) agegroup.

RISK FACTORS FOR MORTALITY ASSOCIATED WITH RESPIRATORY DISEASE IN SHEEP 285

contrast, during the months between April andSeptember, diurnal fluctuation in air temperaturewas small and there was good grazing. This mayexplain the low MARD in these months. A strongcubic relationship between monthly MARD andmonthly air minimum temperature with a negativeslope of the quadratic component (Fig. 3) indicatedthe importance of air temperature as a cold stressorpredisposing animals to pneumonia.

A comparison of differences between age classesin MARD in different months of the year may bequite instructive with regard to animal manage-ment, including timing of breeding, to reduce ani-mal losses. Although animals below 12 months ofage were clearly the most vulnerable, there wereinteresting differences amongst age groups fromone month to another. Thus, mating should betimed to ensure that lambing occurs in late shortrainy season (March/April) so that lambs areweaned at the beginning of the rainy season (June/July) in order to reduce losses from MARD.

Sheep losses varied greatly among years with de-creasing trend across the four years of study: thehighest record was in 1994 and the lowest in 1997(Table IV). Although the ranking of months withregard to MARD did not change from year to year(Tables V–IX), there were differences across yearswith regard to absolute MARD figures for eachmonth as well as magnitude of differences betweenmonths. This indicates that year to year variation inweather pattern may be substantial enough to war-rant close monitoring of the flock in order to makestrategic management decisions. That is, completereliance on an annual management calendar, whileuseful, may not be adequate to address annualvariations in key predisposing factors.

CONCLUSIONS AND RECOMMENDATIONS

Losses due to MARD were shown to be substantialand could have considerable effect on overall flockproductivity. Menz and Horro sheep are known tohave good reproductive ability (Mukasa-Mugerwa etal., unpublished data), but high overall mortality.These results suggest that it may not be advisable tointroduce the Horro breed at this altitude. Breed,gender, age, season, and year appear to have im-portant influence on MARD and should be takeninto account when developing a programme to ad-dress potential losses.

Good management routines such as flock healthand nutritional management, and housing (e.g.

providing bedding) for sheep in the cool highlandsare recommended during critical seasons, especiallyfor lambs below 12 months of age, to minimise los-ses. An annual management calendar should con-sider such factors as timing of lambing in relation topasture availability, timing of vaccination and cull-ing of animals positive for maedi-visna (Ayelet et al.,2001). In addition, because seasons tend to varyacross years in terms of factors predisposing animalsto MARD, there is need for close flock monitoringto be able to make adjustments to the managementcalendar as necessary. Such adjustments couldbenefit from close monitoring for, and properidentification of, pneumonia, while strategic inter-ventions may include raising of lambs on heat-trea-ted colostrum whenever this is possible. Futureresearch in the highlands should consider possibleuse of satellite imagery and modelling to predictrisks from MARD by making use of informationgenerated from area-specific climatic data (e.g. airtemperature), nutritional information (e.g. pastureavailability and quality) and epidemiological data(e.g. number of cases of pneumonia in theflock, pathogens involved, age of sheep at risk, etc.)to make recommendations for specific sets ofcircumstances.

ACKNOWLEDGEMENTS

The authors wish to acknowledge the contributionmade by the staff of the ILRI Debre Berhan Station,Faculty of Veterinary Medicine of Addis AbabaUniversity and National Veterinary Institute in datacollection, laboratory and computer analyses. ILRIanimal genetics research is principally funded byprogramme grants from the United Kingdom, Ja-pan, the European Union, Ireland, and France andcore funding from the CGIAR investors.

We are grateful to anonymous reviewers whosesuggestions have greatly improved the valueand significance of the results presented in this pa-per.

REFERENCES

AGYEMANGGYEMANG, K., AKALEWORKKALEWORK, N., VOORTHUIZENOORTHUIZEN, K. & ANDER-NDER-

SONSON, F. M. (1985). A rapid survey of sheep productionin the traditional sector of Debre Berhan, Ethiopianhighlands. In: Small Ruminants in African AgricultureProceedings of a Conference held at ILCA, Addis Ababa,Ethiopia, eds. R. T. Wilson & D. Bourzat, pp. 174–85.

ALLEYLLEY, M. R. (1991). Pneumonia in sheep. VeterinaryAnnals 31, 51–8.

286 THE VETERINARY JOURNAL, 165, 3

AYELETYELET, G., ROGEROGER, F., TIBBOIBBO, M. & TEMBELYEMBELY, S. (2001).Survey of Maedi-Visna (MV) in Ethiopian highlandsheep. The Veterinary Journal 161, 208–10.

BEKELEEKELE, T., WOLDEABOLDEAB, T., LAHLOU-KASSIAHLOU-KASSI, A. & SHERINGTONHERINGTON,J. (1992). Factors affecting morbidity and mortality on-farm and on-station in the Ethiopian highland sheep.Acta Tropica 52, 99–109.

DALTONALTON, D. C., KNIGHTNIGHT, T. W. & JOHNSONOHNSON, D. L. (1980).Lamb survival in sheep breeds in New Zealand hillcountry.New Zealand Journal of Agricultural Research 23, 167–73.

EIDEIDE, D. M., �AADNOYDNOY, T. & LARSENARSEN, H. J. (1992). Selectionfor immune response in goats: the effects of immuni-sation procedure on antibody response to Diphtheriatoxoid and human serum albumin. Journal of AnimalScience 70, 1432–8.

FAO. (1994). Fao Production Yearbook. Vol. 47. FAO:Rome. Italy.

GALALALAL, E. S. E. (1983). Sheep germplasm in Ethiopia. UNEP/FAO. Animal Genetic Resources. Newsletter 1, 4–12.

KIMBERLINGIMBERLING, C. V. (1988). Diseases of the respiratorysystem. Third Jensen and Swift’s Diseases of Sheep pp. 173–81, Philadelphia: Lea and Febiger.

LASTAST, J. M. (1988). A Dictionary of Epidemiology pp. 91–92,2nd edn. New York: Oxford University Press.

LUNDENUNDEN, A. (1995). Immune responses in sheep afterimmunisation with Toxoplasma gondii antigens incorpo-rated into iscoms. Veterinary Parasitology 56, 23–35.

MEYEREYER, H. H. & CLARKELARKE, J. N. (1978). Genetic andenvironmental effects on incidence and cause of lambmortality. Proceedings of the New Zealand Society of AnimalProduction 38, 181–84.

MOURADOURAD, M., GBANAMOUBANAMOU, A. & BALDEALDE, I. B. (2001).Performance of Djalonk�ee sheep under an extensivesystem of production in Faranah, Guinea. TropicalAnimal Health and Production 33, 413–22.

MUKASA-MUGERWAUKASA-MUGERWA, E., SAIDAID, A. N., LAHLOU-KASSIAHLOU-KASSI, A., SHE-HE-

RINGTONRINGTON, J. & MUTIGAUTIGA, E. R. (1994). Birth weight as arisk factor for perinatal lamb mortality, and the effectsof stage of pregnant ewe supplementation and gesta-tion weight gain in Ethiopian Menz sheep. PreventiveVeterinary Medicine 19, 45–56.

MUKASA-MUGERWAUKASA-MUGERWA, E., LAHLOU-KASSIAHLOU-KASSI, A., ANINDONINDO, D., REGEEGE,J. E. O., TEMBELYEMBELY, S., TIBBOIBBO, M. & BAKERAKER, R. L. (2000).Between and within breed variation in lamb survivaland the risk factors associated with major causes ofmortality in indigenous Horro and Menz sheep inEthiopia. Small Ruminant Research 37, 1–12.

NASHASH, M. L., HUNGERFORDUNGERFORD, L. L., NASHASH, T. G. & ZINNINN, G. M.(1997). Risk factors for respiratory disease mortality inlambs. Small Ruminant Research 26, 53–60.

NJAUJAU, B. C., KASALIASALI, O. B., SCHOLTENSCHOLTENS, R. G. & MESFINESFIN, D.(1988). Field and laboratory studies of causes of sheepmortality in the Ethiopian highlands. ILCA Bulletin 31,23–6.

NORWAKORWAK, R. (1990). Development of mother discrimina-tion by single and multiple newborn lambs. Developmentin Psychobiology 22, 833–45.

PUTTUTT, S. N. H., SHAWHAW, A. P. M., WOODSOODS, A. J., TYLERYLER, L. &JAMESAMES, A. D. (1988). Veterinary Epidemiology andEconomics in Africa. A manual for use in the designand appraisal of livestock health policy. ILCA ManualNo. 3. Addis Ababa: ILCA.

REGEEGE, J. E. O., TEMBELYEMBELY, S., MUKASAUKASA-MUGERWAUGERWA, E., SOVANIOVANI,S., ANINDONINDO, D., LAHLOUAHLOU-KASSIASSI, A., NAGDAAGDA, S., BAKERAKER,R. L. (2002). The effect of breed and season onproduction and response to infections with gas-tro-intestinal nematode parasites in sheep in thehighlands of Ethiopia. Livestock Production Science (Inpress).

ROOKOOK, J. S., SCHOLMANCHOLMAN, G., WING-PROCTORING-PROCTOR, S. & SHEAHEA, M. E.(1990). Diagnosis and control of neonatal losses insheep. Veterinary Clinicians of North American Food AnimalPractice 6, 531–62.

SAS, 1998. SAS User�s Guide for Personal Computers,Version 8 edn. Under windows 2000, SAS Institute.Cary, New York, USA.

SCALESCALES, G. M., BURTONURTON, R. N. & MOSSOSS, R. A. (1986). Lambmortality, birth weight, and nutrition in late preg-nancy. New Zealand Journal of Agricultural Research 29,75–82.

TIBBOIBBO, M. (2000). The effect of triclabendazole (Fasinex)on acute fasciolosis in sheep in central highland ofEthiopia. Bulletin of Animal Health and Production inAfrica 48, 87–92.

TIBBOIBBO, M., WOLDEMESKELOLDEMESKEL, M. & GOPILOOPILO, A. (2001). Anoutbreak of respiratory disease complex in sheep incentral Ethiopia. Tropical Animal Health and Production33, 355–65.

WARNERARNER, C. M., MEEKEREEKER, D. L. & ROTHSCHILDOTHSCHILD, M. F.(1987). Genetic control of immune responsiveness:a review of its use as a tool for selection for diseaseresistance. Journal of Animal Science 64, 394–406.

WINERINER, B. J. (1971). Statistical principles in experimentaldesign, 2nd edn., pp. 4–54, New York: McGraw-HillBook.

YAPIAPI, C. V., BOYLANOYLAN, W. J. & ROBINSONOBINSON, R. A. (1990). Factorsassociated with causes of lamb preweaning mortality.Preventive Veterinary Medicine 10, 145–52.

(Accepted for publication 12 June 2002)

RISK FACTORS FOR MORTALITY ASSOCIATED WITH RESPIRATORY DISEASE IN SHEEP 287