A total of 480 urban and peri-urban dairy farmers were interviewed about their milk and meat consumption habits as part of a wider cross-sectional study of the epidemiology of bTB in Ethiopia. The study was carried out in the five study sites shown in Fig. 1. The study areas contained both urban and peri-urban settings where intensive dairy-farming activities took place.
Milk consumption
Per capita milk consumption. The average per capita milk consumption per day for our sample was found to be 0.25 litres (SD = 0.26). Farmers were asked about their consumption on a ‘daily’ and ‘monthly’ rather than on a yearly basis, as the latter would be difficult to recall and estimate and could also be erroneous due to there being many fasting days in the Ethiopian orthodox church calendar when believers do not consume milk. In this sample the orthodox Christians made up 83.3%. This average per capita milk consumption figure is statistically significantly higher (t = 16.09; P = 0.000) than the national average of 19 liters per year, which corresponds to ~ 0.05 liters per day. The mean per capita milk consumption among the sample was not found to be statistically significant between sexes, religions, literacy statuses, or study sites (Supplementary Table S1).
Raw milk consumption. Farmers were asked about their habits of raw (unboiled and unpasteurized) milk consumption. As shown in Table 2, 77.5% of the respondents (n = 371) indicated that they never drank raw milk while about 20.4% drank raw milk but with varying degrees of frequency. Only 8.1% (n = 39) stated that they were regular drinkers of raw milk, drinking it at least once a day. Although the majority of the sample farmers indicated that they did not drink raw milk, about 82% of the respondents did actually drink fermented milk, a yoghurt commonly called ergo in the Amharic language, which is usually made from non-pasteurized/unboiled milk.
Table 2
Consumption frequency of raw and processed milk reported by dairy farmers
|
Raw milk
|
Pasteurised milk Boiled milk
|
|
% (n)
|
Cum. %
|
% (n)
|
Cum. % % (n) Cum. %
|
Everyday
|
8.1 (39)
|
8.1
|
0.6 (3)
|
0.6 25.5 (122) 25.5
|
3–6 times a week
|
6.3 (30)
|
14.4
|
1.9 (9)
|
2.5 46.3 (222) 71.8
|
Once/twice a week
|
2.9 (14)
|
17.3
|
1.3 (6)
|
3.8 17.3 (83) 89.1
|
Once/twice a month
|
3.1 (15)
|
20.4
|
1.9 (9)
|
5.7 3.8 (18) 92.9
|
On Special occasions only
|
2.1 (10)
|
22.5
|
5.4 (26)
|
11.1 2.3 (11) 95.2
|
Not at all
|
77.5 (371)
|
100
|
88.9 (426)
|
100 4.8 (23) 100
|
Total
|
100 (479)
|
|
100 (479)
|
100 (479)
|
We also investigated raw milk consumption in relationship to gender, literacy, study site, religion, and age and we found that it was not related to any of these socioeconomic variables, except for study site (data not shown). A statistically significant systematic relationship between study site and raw milk consumption habit was established (likelihood-ratio chi2 (4) = 28.70, P = 0.000) with only 5% of the dairy farmers from Mekele (in Northern Ethiopia) indicated that they consumed raw milk as compared to 37% of those from Hawassa (in the South).
The majority of respondents (88%; n = 424) indicated that they knew that drinking raw milk can cause diseases while only 5.6% (n = 27) indicated that it does not cause any diseases. Some 6.0% (n = 29) indicated that they did not know about such risk (data not shown), while 78% indicated that they thought drinking raw milk to be unhealthy or very unhealthy (Table 3).
Neither general training on zoonosis transmission mechanisms (Fisher's exact = 0.415), nor specific training on bTB bore any relation to raw milk consumption frequency (Fisher's exact = 0.680). Moreover, we observed no difference in the frequency of raw milk consumption between those farms whose animals were tested for bTB before our survey and those which were not. This indicates that acquiring knowledge of the bTB status of the cattle at farm level has not generally led to change in raw milk consumption behaviour of these farmers.
Table 3
Farmers’ perception of the healthiness of drinking raw milk
How healthy is drinking raw milk?
|
|
% (n)
|
Cum. %
|
Very healthy
|
4.8 (23)
|
4.8
|
Healthy
|
10.1 (48)
|
14.9
|
Do not know
|
7.4 (35)
|
22.3
|
Unhealthy
|
50.1 (238)
|
72.4
|
Very unhealthy
|
27.6 (131)
|
100
|
Total
|
100 (475)
|
|
Despite their knowledge of the possible risk of disease transmission, a considerable number of our sample (20.4%; n = 98) consumed raw milk frequently and on a regular basis (Table 2). Among participants drinking raw milk, 47% had bTB positive animals in their herd but there was in fact no statistically significant difference in the raw milk consumption habits between farms with bTB positive and negative cattle. Interestingly however, we did find that there was a statistically significant relationship between raw milk consumption habit and occurrence of TB disease in the farm household in the last three years before the survey (likelihood-ratio chi2(1) = 12.09; P = 0.001). Among those farm households which reported a TB case in the last three years, 41% indicated that they were in the habit of consuming raw milk, compared to only 20% among those farmers who reported no TB cases. This result warrants further clinical epidemiological investigation to establish whether the confirmed TB cases might be attributable to zoonotic or human TB, caused by M. bovis or M. tuberculosis, respectively.
Pasturized Milk Consumption. As shown in Table 2, nearly 89% of the surveyed farmers did not drink pasteurized milk. Only 38% (n = 181) knew the benefits of pasteurization, while 54% (n = 259) did not and 8.1% (n = 39) had never heard about pasteurization. Only 1.5% of the respondents indicated that their main source of milk was pasteurized milk. Pasteurized milk was ranked second by 7.6% respondents, 15 % ranked it third and 19% as fifth (data not shown).
Investigation of the relationship between literacy and pasteurized milk consumption frequency showed no systematic relationship (Fisher's exact = 0.690). Since the majority of the large dairy farms and pasteurization plants in Ethiopia are located in the capital Addis Ababa and its surrounding towns, it was logical to expect regional differences in the use of pasteurized milk consumption frequency. However, contrary to what was expected, no relationship was found between study sites and pasteurized milk consumption (Fisher's exact = 0.480). Similarly, there was not a statistically significant relationship between gender and frequency of pasteurized milk consumption (Fisher's exact = 0.156). It needs to be added though that the generally low levels of consumption of pasteurized milk among the surveyed farmers could be because they have easier access to unpasteurized milk than the average consumer. On top of this, our data show that there is statistically significant relation between bTB status of the farmers’ herd and knowledge about pasteurization/pasteurized milk (Chi(2) = 7.19 and P = 0.007) i.e. among those farmers who had bTB positive cattle, 55% had no knowledge of pasteurization and among those farmers who did not know about pasteurization, 42% had bTB positive animals. These results are alarming given the high prevalence of zoonotic diseases in the area, including bTB.
Boiled milk consumption. Although the vast majority of respondents said they did not drink raw milk (78%) or pasteurized milk (89%), 89% (n = 427) of the respondents drank boiled milk at least once a week (Table 2), while only 4.8% (n = 23) indicated that they never drank boiled milk. The frequency of boiled milk consumption was found to be dependent on study site (likelihood-ratio chi2(8) = 21.62; P = 0.006), with those in Hawassa (87%) and in Addis Ababa (75%) were drinking boiled milk more frequently than those in Gondar, in Amhara region (71%), in Mekele, in Tigray region (68%) and in the Oromia towns surrounding Addis Ababa (65%).
Determinants of raw milk consumption frequency.
The results of this analysis indicate that the independent variables in our model are good predictors of the frequency of raw milk consumption (LR chi square of 109.2 significant at 1% confidence level (p value of 0.000 and Pseudo R-square of 0.184). Among the variables entered into the model, we found Study site, gender of the household head, Previous animal bTB testing in farm, knowledge of zoonotic risk of milk consumption, household size, and er capita milk consumption levels were important determinants of frequency of raw milk consumption among the studied dairy farm households (Table 4).
The results of the model suggest that, as compared to farm households in Addis Ababa, being in the Oromia towns surrounding Addis Ababa increased both the probability of raw milk consumption as well as its frequency. Being from Gondar decreased the probability of a respondent consuming raw milk.
The gender of the household head was also found to be an important determinant of raw milk consumption habits. Our results indicate that a household head being male increased the probability of raw milk consuming, but not the frequency of that consumption.
Awareness of bTB due to previous testing of cattle at farm decrease both the probability of raw milk consumption and its frequency. Its effect was more pronounced on decreasing the frequency of consumption, indicating that although there has been change in behaviour regarding raw milk consumption due to awareness of a farm’s bTB status, this change seems to have impacted more in decreasing the frequency at which raw milk was consumed, rather than halting its consumption altogether. In addition, knowledge of the possible zoonotic risk associated with raw milk consumption had its own effect on the raw milk consumption behaviour of farmers, in that it was found to significantly decrease the frequency of raw milk consumption but not the probability that raw milk would be consumed at all.
Household size and per capita milk consumption were found to be important determinants of raw milk consumption habits. With an increase in household size, the probability of raw milk consumption and the frequency of raw milk consumption were found to increase significantly. Similarly, with higher per capita milk consumption, both the probability of raw milk consumption and its frequency were found to increase significantly. The reasons behind this is not clear but could be due to the probable increased costs associated with boiling more milk or purchasing more pasteurized milk consumed by more people per capita.
-
No raw milk consumption;
-
Moderate level of raw milk consumption (only occasional consumption of raw milk); and
-
High frequency of raw milk consumption (at least once a fortnight)
-
No raw meat consumption;
-
Moderate level of raw meat consumption (less than once a fortnight); and
-
High frequency of raw meat consumption (at least once a fortnight)
Table 4
Generalized Ordered Logit Estimates of raw milk (Eq. 1 and Eq. 2) and meat (Eq. 3 and Eq. 4) consumption frequency among dairy farmers in Ethiopia.
|
Raw milk consumption frequency
|
Raw meat consumption frequency
|
VARIABLES
|
Eq1 (A vs B + C)
|
Eq2 (A + B vs C)
|
Eq3 (D vs E + F)
|
Eq4 (D + E vs F)
|
Oromia towns around Addis Ababa
|
0.508*
|
0.923**
|
0.584**
|
0.395
|
|
(0.296)
|
(0.391)
|
(0.297)
|
(0.304)
|
Gondar
|
-0.978**
|
-0.372
|
-0.257
|
1.211***
|
|
(0.491)
|
(0.630)
|
(0.395)
|
(0.410)
|
Mekele
|
-17.45
|
21.66
|
-2.146***
|
-2.324***
|
|
(1,318)
|
(5,068)
|
(0.439)
|
(0.574)
|
Hawassa
|
0.794
|
0.555
|
0.644
|
-0.221
|
|
(0.483)
|
(0.566)
|
(0.528)
|
(0.565)
|
Sex (1 = male; 0 = Female)
|
0.628*
|
-0.176
|
0.743***
|
0.448
|
|
(0.33)
|
(0.436)
|
(0.281)
|
(0.328)
|
Meat preference (1 = beef; 0 otherwise)
|
-
|
-
|
0.507**
|
0.212
|
|
-
|
-
|
(0.251)
|
(0.264)
|
Literacy (1 = Literate; 0 otherwise)
|
0.0833
|
0.929
|
0.339
|
0.315
|
|
(0.555)
|
(0.733)
|
(0.469)
|
(0.558)
|
Can raw meat cause TB (1 = yes; 0 otherwise)
|
-
|
-
|
-2.287***
|
-1.120**
|
|
-
|
-
|
(0.694)
|
(0.521)
|
Raw milk cons. has zoonotic risk (1 = Yes)
|
-0.412
|
-1.745***
|
-
|
-
|
|
(0.413)
|
(0.531)
|
-
|
-
|
Previous bTB Test (1 = Yes; 0 otherwise)
|
-0.728**
|
-1.575***
|
-0.401
|
-0.373
|
|
(0.307)
|
(0.481)
|
(0.260)
|
(0.284)
|
Know benefits of pasteurization (1 = Yes)
|
-0.117
|
0.0879
|
-
|
-
|
|
(0.277)
|
(0.340)
|
-
|
-
|
Had any zoonosis training (1 = yes; 0 otherwise)
|
0.0656
|
0.698
|
0.0626
|
-0.593**
|
|
(0.305)
|
(0.442)
|
(0.269)
|
(0.290)
|
Herd size in number
|
-
|
-
|
-0.00170
|
-0.00929
|
|
-
|
-
|
(0.00503)
|
(0.00678)
|
Number of milking cows
|
-0.00236
|
0.000149
|
-
|
-
|
|
(0.00174)
|
(0.00280)
|
-
|
-
|
Per capita milk consumption per day
|
0.768*
|
1.086**
|
-
|
-
|
|
(0.435)
|
(0.488)
|
-
|
-
|
Meat per capita consumption per month in kg
|
-
|
-
|
0.206
|
0.828***
|
|
-
|
-
|
(0.138)
|
(0.145)
|
Age of the household head
|
-0.0152
|
-0.00135
|
0.0265
|
0.0908*
|
|
(0.0101)
|
(0.0147)
|
(0.0505)
|
(0.0537)
|
Age squared
|
-
|
-
|
-0.000424
|
-0.000939*
|
|
-
|
-
|
(0.000484)
|
(0.000523)
|
Household size
|
0.106*
|
0.190**
|
0.0203
|
0.159**
|
|
(0.0602)
|
(0.0751)
|
(0.0591)
|
(0.0631)
|
Constant
|
-0.633
|
-2.401*
|
1.266
|
-4.167***
|
|
(0.968)
|
(1.431)
|
(1.471)
|
(1.496)
|
Observations
|
440
|
440
|
417
|
417
|
Standard errors in parentheses; *** p < 0.01, ** p < 0.05, * p < 0.1 |
Meat Consumption
Per capita meat consumption. The published national average of per capita meat consumption in Ethiopia is 5.3kg per annum [46], which corresponds to less than 0.5kg per month. However, the corresponding consumption for urban areas is 11.5kg (~ 1kg per month). As shown in Table 5, the latter figure is much in line with the mean rate of per capita monthly meat consumption of 1.25kg (Std. dev = 1.44) among the dairy farmers in our sample, with Hawassa having the lowest (0.88kg) and Addis Ababa having the highest (1.37kg) consumption. Interestingly, the mean per capita meat consumption per month for male-headed households was found to be 1.35kg, which was statistically higher (t=-2.43, P = 0.015) than the 0.95kg per month for female-headed households. We found no statistically significant difference in per capita meat consumption between study sites, religions (Christians and Muslims), or between households with illiterate and literate heads.
Table 5
Average monthly meat consumption per capita (in kg) among dairy farmers in each study site.
|
Mean
|
Std. Dev.
|
n.
|
Addis Ababa city
|
1.4
|
1.87
|
164
|
Oromia towns around Addis Ababa
|
1.1
|
1.06
|
135
|
Gondar
|
1.4
|
1.46
|
66
|
Mekele
|
1.3
|
0.95
|
57
|
Hawassa
|
0.9
|
0.49
|
25
|
Total
|
1.3
|
1.44
|
447
|
Meat type preference. For the whole sample, when respondents were asked to rank their meat type preference it was found that 48% (N = 220) ranked beef as first choice, followed by mutton 32% (n = 144), chicken 11% (n = 48), and goat meat 9.4% (n = 43). However, these figures varied by region with a statistically significant association between preferred meat-type and study site (likelihood-ratio chi2(12) value of 135.5 and P = 0.000). The responding dairy farmers in Addis Ababa (62%) and in the towns of Oromia (64%), as well as in Hawassa (44%) tended to significantly prefer beef, while the majority in Gondar (75%) and Mekele (56%) in the Northern regions prefered mutton over any other meat. A similar association was observed between frequency of meat consumption and meat type preference, with Fisher's Exact test of 0.029 being significant at the 0.05 level; i.e. those households which preferred beef tended to be more frequent meat consumers (79%). A one-way analysis of variance in terms of mean age of the respondents and their preferred meat type showed a statistically significant difference (F value = 5.49; P = 0.001). Those households which preferred chicken meat had the lowest mean age of 41 years (SD = 11) and this was significantly different from the mean age of 50 years (SD = 15) among those who preferred mutton and the mean age of 46 years (SD = 15) among those who preferred beef. Education, which was measured as a dummy variable with the two categories being literate or illiterate, showed no relationship to meat type preference (Fisher's exact test P value = 0.830). We also tested for a relationship between gender of the household head and meat type preference but found no significant correlation.
Meat consumption frequency. As shown in Table 6, only 1% of the responding farmers ate meat every day while the majority of them (57%) consumed meat 2–5 days a week. Only 0.6% (3 individuals) indicated that they did not consume meat at all. There was no statistical difference between study site and meat consumption frequency, and gender of the household head and meat consumption frequency. Also, no significant difference in mean age of respondents was observed between those households which frequently consumed meat and those who did so less frequently (t = -0.2779 and P = 0.7812). On the other hand, literacy level and frequency of meat consumption were found to be associated (Fisher's exact value P = 0.001); 56% of the illiterate household heads indicated a high frequency of meat consumption while about 83% of the literate households consumed meat at high frequency. 63.9% of the respondents habitually consumed raw meat (mainly beef) and about 20% were in the habit of consuming raw meat either every day or 2–5 times a week. However, more than a third of the respondents (36.1%) indicated that they had never consumed raw meat (Table 6).
Table 6
Frequency of meat consumption among sampled dairy farmers
|
General meat consumption
|
Raw meat consumption
|
|
|
% (n) Cum. %
|
% (n) Cum. %
|
Everyday
|
|
1.0 (5) 1.0
|
0.4 (2) 0.4
|
2–5 days a week
|
|
56.6 (271) 57.6
|
20.0 (96) 20.4
|
Once every fortnight
|
|
22.8 (109) 80.4
|
10.7 (51) 31.1
|
Once a month
|
|
13.8 (66) 94.2
|
13.8 (66) 44.9
|
Only for holidays
|
|
5.2 (25) 99.4
|
19.0 (91) 63.9
|
Never
|
|
0.6 (3) 100.0
|
36.1 (173) 100.0
|
Total
|
|
100 (479)
|
100 (479)
|
An investigation was conducted into the relationship between raw meat consumption frequency and demographic factors of which statistically significant associations were found between frequency and study site (likelihood-ratio chi2(20) = 120.6; P = 0.000) and religion (likelihood-ratio chi2(1) = 13.34; P = 0.0000), respectively. Muslims tended to avoid raw meat, with 75% (15 out of 20) of the surveyed Muslims indicating that they had never consumed raw meat. Among the study sites surveyed, the proportion of dairy farmers who consumed raw meat more frequently (at least once in a fortnight) was 66% in Addis Ababa, 77% in Oromia, 66% in Hawassa, and 67% in Gondar. However, only 25% in Mekele had a habit of frequent raw meat consumption. No relationships between raw meat consumption frequency and gender, literacy, or age were found in this survey.
Meat source preference. As shown in Table 7, the most preferred source of meat across the sample of farmers was found to be butchery (53.9%), followed by home slaughter (36.6%) and then communal slaughter (9.5%).
Table 7
Ranking order of meat source by sample dairy farmers
|
Rank 1
|
Rank 2*
|
Rank 3*
|
|
% (n)
|
|
% (n)
|
|
% (n)
|
|
Butchery
|
53.9 (256)
|
|
24.0 (109)
|
|
17.8 (58)
|
|
Home slaughter
|
36.6 (174)
|
|
42.2 (191)
|
|
27.8 (91)
|
|
Communal slaughter
|
9.5 (45)
|
|
33.8 (153)
|
|
54.4 (178)
|
|
Total
|
100 (475)
|
|
100 (453)
|
|
100 (327)
|
|
* Not all respondents gave a second or third ranking |
The relationship between meat source ranking and variables such as gender of the household head, literacy status, religion, study site, and age were examined. We found a statistically significant association between study site and the main source of meat (likelihood-ratio chi2(8) = 126.4; P = 0.000). Most farmers from Addis Ababa city (76%) indicated that butchery was their primary source of meat, compared to only 6.9% among farmers from Gondar in the Amhara region. Also mean age and the primary source of meat were found to be statistically different (F = 4.15; P = 0.016) between those households using butchery (45 years) as a primary source and those using home slaughter as a primary source (49 years). The results indicated that none of the other socioeconomic factors were associated to the primary meat source of a household.
Knowledge of zoonoses. The interviewed farmers were also asked if they think that eating raw meat can cause diseases. The result shows that the vast majority of farmers (92.9%) believed that consumption of raw meat can cause diseases and about 40% had actually experienced disease symptoms which they attributed to eating raw meat (Table 8). Many respondents reported that they had experienced diseases and symptoms after eating raw meat, such as abdominal discomfort, tape worm, amoeba, gout and even TB.
Table 8
Farmers’ views about risks of getting disease while eating raw meat
Do you think eating raw meat can cause diseases?
|
Have you ever experienced diseases due to eating raw meat?
|
|
% (n)
|
|
% (n)
|
|
No
|
7.1 (34)
|
|
59.9 (287)
|
|
Yes
|
92.9 (445)
|
|
40.1 (192)
|
|
Total
|
100 (479)
|
|
100 (479)
|
|
Farmers were also asked if they knew that TB can be transferred from animals to humans through consumption of raw meat. Out of 477 respondents, 62.3% (297) indicated that they thought eating raw meat could do so, while 23.1% (n = 110) indicated that they did not know whether this was the case. Only 6.3% (n = 30) stated that TB cannot be transferred from animal to human by eating raw meat.
We also investigated wheather a relationsip exists between having attended training on zoonotic diseases and bTB transmission pathways, and farmers’ meat consumption behaviour. The results of this test indicated that there is a statistically significant relationship (likelihood-ratio chi2 (2) = 7.72; P = 0.021). Among the 348 farmers who had not undertaken training on zoonoses provided by local government extension services, 33.3% indicated that they consumed raw meat frequently. In contrast, only 24.3% of the 123 farmers who undertook training on zoonoses indicated that they consumed raw meet frequently. Our data also show that there was a statistically significant relationship between raw meat consumption habit and past occurrence of TB in the family (likelihood-ratio chi2 (2) = 5.68; P = 0.017). Out of the 48 farm households who reported that there has been a confirmed human TB case in the last three years in their farm, 20.8% indicated that they have the habit of raw meat consumption while the 79.2% for those farm households reported no TB case in the past three years.
Determinants of raw meat consumption habits. The results of the analysis indicate that some of the independent variables in our model are good predictors of the frequency of raw meat consumption (LR chi square = 156.3, p-value = 0.000 (SD 99%)). Study site, gender of household head, knowledge about zoonotic risks associated with raw meat consumption, training on zoonoses, age squared, household size, and per capita consumption of meat (of all types, either raw or cooked) were all found to be significant variables when predicting the frequency of an individual’s raw meat consumption, i.e. which of the three stated categories (D-F) they would fall into.
In terms of study site as a predictor, respondents based in the Oromia towns surrounding Addis Ababa were more likely to consume raw meat than those in Addis Ababa city, however, the frequency of such consumption was not significant. As compared to respondents from the capital, an average household based in Gondar did not consume more raw meat, but dairy households in Gondar tended to consume raw meat more frequently than those in Addis Ababa. However, the data also indicates that being based in Mekele reduced both the probability of consumption of raw meat as well as its frequency.
The gender of the household head was also found to affect the probability of raw meat consumption, but not the frequency of that consumption; a household head being male significantly increased the probability of members of that household consuming raw meat. Training on zoonotic disease transmission risks was found to have an effect on the frequency, rather than the probability of consumption, meaning those households which had access to zoonosis training tended to report a lower frequency of raw meat consumption as compared to those who did not have access to zoonosis training.
The result showed that age of dairy farmers had a positive effect on raw meat consumption frequency up to some limits but the effect of age on the frequency of raw milk consumption turned to be negative as farmers got older. However, age did not have a significant effect on the probability of raw meat consumption. Young farmers tended to have higher raw meat consumption frequency and as farmers got old they tended to decrease the frequency of raw meat consumption
Interestingly, having ‘knowledge of the effects of raw meat consumption on the risk of zoonotic transmission of diseases’ had a statistically significant effect on both the probability of raw meat consumption and its frequency, and with a higher impact on the former. This means having knowledge about the risks involved in consumption of raw meat negatively affected both the decision to consume raw meat as well as its frequency, but it affected the former much more than the later.
Our data also suggests that the effect of meat type preference, i.e. a farmer who preferred beef meat, also had positive and significant effect on the probability of eating raw meat; however, no effect on the probability of the raw meat consumption frequency was seen. This might be due to the suitability of beef meat for raw meat based meals such as kitfo, kurt, and gored gored, their local names in the Amharic language.
In the model, Herd size was entered as a proxy variable to capture the effect of wealth on raw meat consumption habits. The result indicated that the habit of consuming raw meat was similar across the different wealth categories in Ethiopia and there was no statistically significant difference between the behaviour of the rich and the poor in this regard.
the data showed that high consumption of meat in general (expressed as ‘Per capita meat consumption’) did not affect the probability of raw meat consumption but it did increase the probability of doing so more frequently. Increased family size was also found to be linked with increased frequency of raw meat consumption.