J. Dairy Sci.

93:624–627
doi:10.3168/jds.2009-2689
© American Dairy Science Association®, 2010.

Short communication: Repeatability of measures

of rectal temperature in dairy cows
O. Burfeind,*† M. A. G. von Keyserlingk,* D. M. Weary,* D. M. Veira,‡ and W. Heuwieser*†1
*Animal Welfare Program, Faculty of Land and Food Systems, The University of British Columbia, 2357 Main Mall, Vancouver, BC,
V6T 1Z4, Canada
†Clinic for Animal Reproduction, Faculty of Veterinary Medicine, Freie Universität Berlin, Koenigsweg 65, 14163 Berlin, Germany
‡Agriculture and Agri-Food Canada, Pacific Agriculture Research Station, Agassiz, British Columbia, Canada

ABSTRACT The method most commonly used to identify sick

dairy cows in the postpartum period is rectal tem-
Although taking body temperature by rectal ther- perature (Smith and Risco, 2005). Measuring rectal
mometer is the method most commonly used to iden- temperature for 5 to 10 d after calving has received
tify sick cows in the postpartum period, no data on the attention in the past because of ease of implementation
repeatability of this measure are available. The overall and low cost (Kristula et al., 2001) and has been incor-
objective of this study was to evaluate variability of porated into standard operating protocols for fresh cow
rectal temperatures in dairy cows considering differ- management and disease intervention. Several studies
ent factors (intra- and interinvestigator repeatability, demonstrating the efficacy of antibiotic treatment of
different thermometers, penetration depth into the postpartum metritis used rectal temperature thresholds
rectum, and defecation). High coefficients of correla- as inclusion criteria (Drillich et al., 2001, 2006; Zhou
tion (r = 0.98) and small differences between values of et al., 2001). Among researchers, veterinarians, and
rectal temperatures (observer A = 39.3 ± 1.0°C and producers, there is agreement that monitoring rectal
observer B = 39.4 ± 1.0°C) provide evidence that rectal temperatures during this critical period is a successful
temperature was a repeatable measure in dairy cows. tool for the management of diseases (Smith and Risco,
Testing was carried out using 4 different digital ther- 2005).
mometers: GLA M750 (GLA Agricultural Electronics, Most authors define fever as a body temperature
San Luis Obispo, CA), MTI8101 (SES Scala Electron- exceeding a predefined threshold and consider a single
ics, Stahnsdorf, Germany), MT1831 (Microlife AG, instance of a temperature above the threshold value
Widnau, Switzerland) and Domotherm TH1 (Uebe an indication of illness. Threshold values of 39.4 and
Medical GmbH, Wertheim, Germany). Thermometers 39.7°C (Smith and Risco, 2005; Benzaquen et al., 2007;
were inserted into the rectum to a certain depth (GLA Wagner et al., 2007) distinguished between healthy
M750 and MTI8101 = 11.5 cm; MT1831 = 8.4 cm; cows and cows suffering from an infectious-related dis-
Domotherm TH1 = 7.7 cm) and a measure was finished ease process.
when a visual or acoustic signal was emitted by the Despite the common use of rectal temperatures, there
thermometer. The measures could be influenced by the is a lack of science-based information on the value or
procedure itself (up to 0.5°C), type of thermometer (up significance of measuring the rectal temperature as a
to 0.3°C), and the penetration depth (11.5 cm or 6.0 diagnostic tool to identify infectious diseases in dairy
cm in one of the experiments) into the rectum (up to cattle in the postpartum period (Benzaquen et al.,
0.4°C difference between a penetration depth of 11.5 2007). Type I (fever when the animal is actually healthy)
cm and 6.0 cm in one of the experiments). Differences and type II (no fever when the animal is actually sick)
in rectal temperature before and after defecation were errors were reported (Kristula et al., 2001; Sheldon et
minor (<0.1°C). These results indicate that some care al., 2004; Wagner et al., 2007). A type I error can cause
is required in generalizing rectal measures of body tem- financial losses to the producer because a healthy ani-
perature. mal is unnecessarily treated. A type II error leaves a
Key words: dairy cow, rectal temperature, repeat- beneficial treatment effect on health and performance
ability unrealized and might constitute an animal welfare issue
because a sick animal is not being treated.
No published study documents the repeatability of
body temperature measured by rectal thermometer in
Received September 2, 2009.
Accepted October 28, 2009. dairy cattle. The objective of this study was to evaluate
1
Corresponding author: [email protected] variability of rectal temperatures in dairy cows consid-
624
 SHORT COMMUNICATION: REPEATABILITY OF RECTAL TEMPERATURE 625

ering different factors. Specifically, the objectives were certificate 5–2253–81A) as a gold standard. The water
1) to determine intra- and interinvestigator variability bath was adjusted in 1.0°C increments from 35.0 to
of rectal temperature measurements; 2) to evaluate dif- 42.0°C. This procedure was repeated 25 times per ther-
ferences among 4 thermometers; and 3) to determine mometer to provide 200 measurements per thermom-
the effects of penetration depth into the rectum and eter. The measurement was terminated and the values
defecation on measured body temperature. of the thermometers recorded when a visual or acoustic
Five experiments were conducted between January signal was emitted by the digital thermometer.
and April 2009 at the University of British Columbia During the in vivo phase, a single investigator
Dairy Education and Research Centre (Agassiz, British measured the rectal temperature of 37 dairy cows,
Columbia, Canada). Cows were managed according to including 10 with temperatures above 39.4°C Rectal
the guidelines set by the Canadian Council on Animal temperature was measured twice per cow with each of
Care (1993) and were housed in postpartum group pens the 4 thermometers. The GLA M750 and the MTI8101
with sand-bedded freestalls. Cows were milked (ap- thermometers had an angle probe (42°). The distance
proximately 0700 and 1700 h) and fed (approximately from the tip of the sensor to the angle was 11.5 cm
0600 and 1600 h) twice daily. All measurements were for both the GLA M750 and the MTI8101. The Do-
performed between 0900 and 1100 h. motherm TH1 and MT1831 were thermometers with
In experiment 1, rectal temperature was measured straight probes. The distance from the tip of the sen-
multiple times per cow by a single investigator using sor to the display was 7.7 and 8.4 cm for Domotherm
the same digital thermometer (GLA M750, GLA Ag- TH1 and MT1831, respectively. The probe on the GLA
ricultural Electronics, San Luis Obispo, CA) equipped M750 was made entirely of metal, whereas only the tip
with an angle probe (11.5 cm, 42°). The measurement of the probe was metal on the MTI8101, Domotherm
was terminated when either a visual or acoustic signal TH1, and MT1831 thermometers. All 4 thermometers
was emitted by the digital thermometer, and the probe were equipped with digital displays. The GLA M750
was retracted from the rectum. The thermometer was and the MTI8101 thermometers were made specifically
reset, the probe reinserted into the rectum, and the for veterinary use, the MT1831 was recommended for
next measurement initiated. This general procedure use in both animals and humans, and the Domotherm
of measuring rectal body temperature was the same TH1 was made for humans. For the measurements, the
in all 5 experiments. Any differences in protocol (e.g., probe was inserted into the rectum such that the angle
penetration depth) are mentioned in the descriptions of of the probe (located at 11.5 cm) disappeared into the
the individual experiments. A total of 10 measures were anus (GLA M750, MTI8101) or up to the edge of the
completed within 5 ± 1 min/cow. For this experiment display (located at 8.4 cm for the MT1831 and at 7.7
33 cows were used, including 16 cows with a tempera- cm for the Domotherm TH1). All 4 measurements were
ture above 39.4°C. For all measures the probe was in- completed within 4 ± 1 min/cow.
serted into the rectum such that the angle of the probe In experiment 4, 1 investigator measured the rectal
(located at 11.5 cm) had disappeared in the anus. temperature of 33 cows (9 cows >39.4°C) with the
In experiment 2, the rectal temperature of 38 cows MTI8101 thermometer, inserting the probe either 11.5
was measured by 2 investigators; 15 cows had tempera- or 6.0 cm. Both measurements were conducted twice;
tures above 39.4°C. The digital thermometer and the all 4 measurements were taken within 2 ± 0.5 min/
procedure used were identical to experiment 1. Both cow.
investigators measured the temperature twice and in- Experiment 5 tested the effect of defecation on rectal
dependently from each other. All 4 measurements were temperature. Temperature was measured twice before
completed within 2 ± 0.5 min/cow. and twice after defecation in 20 cows using the GLA
Experiment 3 consisted of in vitro and in vivo phases M750 thermometer. The temperature was measured
to test the variation among different digital thermom- first while the cows were lying. Cows were then gently
eters: GLA M750 (GLA Agricultural Electronics), encouraged to stand up. If cows defecated upon stand-
MTI8101 (SES Scala Electronics, Stahnsdorf, Germa- ing, the temperature was measured again. All 8 mea-
ny), Domotherm TH1 (Uebe Medical GmbH, Wertheim, surements were completed within 4 ± 1 min/cow.
Germany), and MT1831 (Microlife AG, Widnau, Swit- All data were recorded in Excel (version 2003, Micro-
zerland). The in vitro phase used a water bath (GFL soft Office, Redmond, WA) and statistically analyzed
Gesellschaft für Labortechnik m.b.H., Burgwedel, with MedCalc software (version 10.1.3.0, MedCalc,
Germany) filled with 16 L of water with a calibrated Mariakerke, Belgium). Each rectal temperature was
liquid-in-glass thermometer (validated by the Office of obtained by taking an initial measurement followed im-
Legal Metrology of the State of Hessen, Germany; type mediately by a second measurement. These 2 measures
E 0.1/0/50 DIN 12775; serial number 2253; verification were averaged and the resulting mean used for further
Journal of Dairy Science Vol. 93 No. 2, 2010
626 BURFEIND ET AL.

Table 1. Correlation and differences of paired measurements between a calibrated liquid-in-glass thermometer
and 4 digital thermometers using a water bath between 35.0 and 42.0°C1

Coefficient of correlation Difference of paired measurements

Thermometer2 r P-value Mean ± SD (°C) Paired t P-value

GLA M750 0.99 <0.001 0.0 ± 0.1 5.39 <0.001
MTI8101 0.99 <0.001 0.0 ± 0.1 4.55 <0.001
MT1831 0.99 <0.001 0.0 ± 0.0 2.39 <0.05
Domotherm TH1 0.99 <0.001 0.1 ± 0.1 11.42 <0.001
1
The water bath was adjusted in 1.0°C increments from 35.0 to 42.0°C. This procedure was repeated 25 times
per thermometer to provide 200 measurements per thermometer.
2
Thermometer compared with a calibrated liquid-in-glass thermometer. GLA M750: GLA Agricultural
Electronics, San Luis Obispo, CA; MTI8101: SES Scala Electronics, Stahnsdorf, Germany; MT1831: Microlife
AG, Widnau, Switzerland; Domotherm TH1: Uebe Medical GmbH, Wertheim, Germany.

analysis. To determine intrainvestigator reliability, the mometers were small (Table 1). The measures from the
minimum, maximum, differences, and coefficient of vari- 4 different thermometers were highly correlated (r =
ance were calculated for each of the 33 cows. Pearson 0.94– 0.96; P < 0.001; Table 2) during the in vivo stage
coefficient of correlation was calculated to determine of experiment 3. The mean difference between ther-
the reliability between the 2 investigators. Relation- mometers varied from 0.1 to 0.3°C (Table 2; GLA M750
ships between the 4 thermometers and the influence of = 39.2 ± 0.7°C; MTI8101 = 39.3 ± 0.6°C; MT1831 =
the 2 penetration depths into the rectum on tempera- 39.0 ± 0.7°C; Domotherm TH1 = 38.9 ± 0.7°C). Agree-
ture were evaluated by Pearson coefficients of correla- ment was high when both thermometers compared had
tion. Differences between investigators, thermometers, either a short (MT1831 = 39.0 ± 0.7°C; Domotherm
penetration depths, and temperatures before and after TH1 = 38.9 ± 0.7°C) or a long (GLA M750 = 39.2 ±
defecation were tested using paired t-tests. 0.7°C; MTI8101 = 39.3 ± 0.6°C) probe (mean difference
Repeated measures by a single investigator (experi- = 0.1°C). The effect of insertion depth was confirmed
ment 1) were consistent (mean ± SD = 39.5 ± 0.1°C; in experiment 4; measures taken at the 2 penetration
coefficient of variance = 0.2%), indicating a high re- depths were highly correlated (11.5 cm = 39.2 ± 0.8°C;
peatability. Within the 10 measurements on the same 6.0 cm = 38.8 ± 0.7°C; r = 0.95; P < 0.001; n = 33),
cow, the maximum difference varied between 0.1 and but the temperature was higher when the probe was
0.5°C. The difference between lowest and highest rec- inserted deeper into the rectum (0.4 ± 0.2°C; paired t
tal temperature was 0.5 and 0.4°C in 2 and 5 cows, = 10.33; P < 0.001).
respectively. For the remaining 26 cows, the difference Differences in rectal temperature before (38.6 ±
between lowest and highest rectal temperature was 0.2°C) and after (38.5 ± 0.2°C) defecation were minor
0.3°C or lower. Correlation between measures taken (i.e., experiment 5; 0.1 ± 0.2°C; paired t = 1.14; P =
independently by 2 investigators (experiment 2) was 0.27). In 2 cows the measured temperature was 0.3°C
high (r = 0.98; P < 0.001). The mean difference of the or higher before defecation, whereas in 3 cows it was
investigators was 0.1 ± 0.2°C (paired t = 2.77; P < 0.3°C or higher after defecation. In the remaining 15
0.01). Overall, these data demonstrate a high repeat- cows, the difference before and after defecation was
ability of rectal temperature measurements within and less than 0.3°C. The consistency of the manure was not
between investigator(s). However, for some cows, the systematically determined; however, all cows used had
difference between lowest and highest rectal tempera- a normal manure.
ture was considerable (0.5°C in 2 cows and 0.4°C in 5 The results indicated that rectal temperature can be
cows). The reason for these differences remains specula- measured repeatably in dairy cows. In some, albeit few,
tive. It was assumed that these differences were caused instances, the procedure itself can influence the result
by the procedure of taking the temperature and not by (up to 0.5°C). More importantly, the type of thermom-
a change of the body temperature of the animal. eter (up to 0.3°C) and the penetration depth into the
Measures from all 4 digital thermometers were cor- rectum (up to 0.4°C) can bias the results. These factors
related to the calibrated liquid-in-glass thermometer may help explain why 1) a significant percentage of
(experiment 3 in vitro component; r = 0.99; P < 0.001) healthy cows occasionally have temperatures greater
when measuring the temperatures of a water bath than 39.5 (26%) or 39.7°C (9%) in the first days post-
between 35 and 42°C. The mean differences between partum (Smith and Risco, 2005; Benzaquen et al., 2007;
the liquid-in-glass thermometer and the 4 digital ther- Wagner et al., 2007), and 2) not all cows diagnosed

Journal of Dairy Science Vol. 93 No. 2, 2010

 SHORT COMMUNICATION: REPEATABILITY OF RECTAL TEMPERATURE 627
Table 2. Correlation and differences between paired measurements from 4 digital thermometers recorded on 37 dairy cows in experiment 2

Coefficient of correlation Difference of paired measurements

Thermometer comparison1 r P-value Mean ± SD (°C) Paired t P-value

GLA M750 vs. MTI8101 0.96 <0.001 0.1 ± 0.2 2.44 <0.05
GLA M750 vs. MT1831 0.94 <0.001 0.2 ± 0.2 4.12 <0.001
GLA M750 vs. Domotherm TH1 0.96 <0.001 0.2 ± 0.2 7.45 <0.001
MTI8101 vs. MT1831 0.96 <0.001 0.2 ± 0.2 7.50 <0.001
MTI8101 vs. Domotherm TH1 0.96 <0.001 0.3 ± 0.2 10.84 <0.001
MT1831 vs. Domotherm TH1 0.96 <0.001 0.1 ± 0.2 2.48 <0.05
1
GLA M750: GLA Agricultural Electronics, San Luis Obispo, CA; MTI8101: SES Scala Electronics, Stahnsdorf, Germany; MT1831: Microlife
AG, Widnau, Switzerland; Domotherm TH1: Uebe Medical GmbH, Wertheim, Germany.

with metritis have an instance of an elevated tempera- (Milner, British Columbia, Canada), Pfizer Animal
ture when measured once per day (Benzaquen et al., Health (Kirkland, Québec, Canada), BC Cattle Indus-
2007). Alternatively, endotoxins can be absorbed by the try Development Fund (Kamloops, British Columbia,
postpartum uterus in dairy cows (Peter et al., 1990). Canada), the BC Milk Producers (Burnaby, British
Escherichia coli, a common cause of metritis (Sheldon Columbia, Canada), BC Dairy Foundation (Burnaby,
et al., 2008), might be a possible cause for endotoxi- British Columbia, Canada), BC Dairy Education and
cosis, which can cause elevated rectal temperatures Research Association (Abbotsford, British Columbia,
in dairy cows. Elevated rectal temperatures were not Canada), and Alberta Milk (Edmonton, Alberta,
always observed in endotoxic cows (Andersen, 2003), Canada).
thus providing a possible explanation as to why cows
diagnosed with metritis failed to exhibit an elevated
REFERENCES
temperature when measured only once daily (Benzaquen
et al., 2007). To develop individual-standard fresh-cow Andersen, P. H. 2003. Bovine endotoxicosis: Some aspects of relevance
monitoring programs, treatment protocols at the herd to production diseases. Acta Vet. Scand. 98(Suppl.):141–155.
Benzaquen, M. E., C. A. Risco, L. F. Archbald, P. Melendez, M.-
level, and experimental protocols requiring body tem- J. Thatcher, and W. W. Thatcher. 2007. Rectal temperature,
peratures, veterinarians and dairy scientists should rec- calving-related factors, and the incidence of puerperal metritis in
ommend measuring rectal temperature with the same postpartum dairy cows. J. Dairy Sci. 90:2804–2814.
Canadian Council on Animal Care. 1993. Guide to the Care and Use of
thermometer and at the same penetration depth as part Experimental Animals. Vol. 1. E. D. Olfert, B. M Cross, and A.A.
of the standard operating procedures. Furthermore, the McWilliam, ed. CCAC, Ottawa, Ontario, Canada.
results of this study confirm the necessity of additional, Drillich, M., O. Beetz, A. Pfutzner, M. Sabin, H. J. Sabin, P. Kutzer,
H. Nattermann, and W. Heuwieser. 2001. Evaluation of a systemic
reliable indicators of illness besides a single tempera- antibiotic treatment of toxic puerperal metritis in dairy cows. J.
ture value above a certain threshold, as suggested by Dairy Sci. 84:2010–2017.
Kristula et al. (2001) and Sheldon et al. (2004). Further Drillich, M., M. Mahlstedt, U. Reichert, B. A. Tenhagen, and W.
Heuwieser. 2006. Strategies to improve the therapy of retained
research should elucidate which combination of clinical fetal membranes in dairy cows. J. Dairy Sci. 89:627–635.
signs, temperature threshold, and frequency of mea- Kristula, M., B. I. Smith, and A. Simeone. 2001. The use of daily
surements per day can improve diagnostic performance postpartum rectal temperatures to select dairy cows for treatment
with systemic antibiotics. Bovine Pract. 35:117–125.
in identifying sick cows postpartum. Peter, A. T., W. T. K. Bosu, and R. O. Gilbert. 1990. Absorption of
endotoxin (lipopolysaccharide) from the uteri of postpartum dairy
ACKNOWLEDGMENTS cows. Theriogenology 33:1011–1014.
Sheldon, I. M., A. N. Rycroft, and C. Zhou. 2004. Association between
postpartum pyrexia and uterine bacterial infection in dairy cattle.
We thank the staff of The University of British Vet. Rec. 154:289–293.
Columbia Dairy Education and Research Centre and Sheldon, I. M., E. J. Williams, A. N. A. Miller, D. M. Nash, and S.
the University’s Animal Welfare Program (Vancouver, Herath. 2008. Uterine diseases in cattle after parturition. Vet. J.
176:115–121.
British Columbia, Canada). Onno Burfeind was funded Smith, B. I., and C. A. Risco. 2005. Management of periparturient
by a scholarship from Tiergyn Berlin e.V. (Berlin, Ger- disorders in dairy cattle. Vet. Clin. North Am. Food Anim. Pract.
many). This project was funded by Freie Universität 21:503–521.
Wagner, S. A., D. E. Schimeck, and F. C. Chend. 2007. Body
Berlin (Germany) and by Canada’s Natural Sciences temperature and white blood cell count in postpartum dairy cows.
and Engineering Research Council (Ottawa, Ontario, Bovine Pract. 42:18–26.
Canada) Industrial Research Chair Program, with in- Zhou, C., J. F. Boucher, K. J. Dame, M. Moreira, R. Graham, J.
Nantel, S. Zuidhof, L. Arfi, R. Flores, G. Neubauer, and J. Olson.
dustry contributions from the Dairy Farmers of Canada 2001. Multilocation trial of ceftiofur for treatment of postpartum
(Ottawa, Ontario, Canada), Westgen Endowment Fund cows with fever. J. Am. Vet. Med. Assoc. 219:805–808.

Journal of Dairy Science Vol. 93 No. 2, 2010