Data Driven Identification of Injury Risk Factors During Expansion on Irish Dairy Farms

ABSTRACT

Objectives

This paper sought to determine risk factors of occupational injury in the Irish dairy farming sector and to estimate the roles of both dairying expansion and the discipline of agricultural extension in influencing reducing injury occurrence.

Methods

Data for this study was obtained via the Irish National Farm Survey (NFS). In total, 260 farm (83.6% of NFS dairy farm sample) workplace injury survey questionnaires were completed by NFS recorders by interviewing principal farm operators for year 2017. Injury survey data was analysed for 48 variables in conjunction with NFS core farm socio-economic, physical and financial data. Additionally, core data from 2010 for 78.5% of farms surveyed in 2017 was included in the study. Data were analysed using a three-step statistical testing process which met all Binary Logistic Regression assumptions.

Results

The study found that dairy farms had a higher farm workplace occupational injury level compared to a previous study. The study data indicates occurrence of elevated injury levels on farms which expanded and which were intensively managed from a milk production perspective. Farm expansion was associated with increased labor units used and increased investment per hectare. The study also found that use of extension services and farm discussion group participation were not associated with injury occurrence.

Conclusions

This study demonstrates how a retrospective farm workplace occupational injury survey of Irish dairy farms, analysed in combination with farm business data can be used to identify injury risk factors, including those associated with production expansion. Irish dairy farms have a heightened farm workplace occupation injury level while dairy production expansion increased injury levels. Extension engagement did not influence on injury levels. Approaches to improve farm safety on dairy farms are outlined.

KEYWORDS:

• Dairy farming
• extension services
• national survey
• output expansion; workplace injury

Introduction

Occupational injuries have the potential to cause pain, suffering and disability among dairy farm workers including farm family members and also have the potential to cause farm economic losses.¹ It has been recently reported in the United States that the occupational injury rate of the dairy farming sector (6.6 per 100 full-time workers) is twice that of the national average across all industries, with the dairying occupational injury rate having remained unchanged as farms increased in scale.² Similarly, dairy farming is the most dangerous farm enterprise in Ireland with higher farm work-related injury rates.³ In Northern Ireland it has been reported that those working on dairy farms exhibit a higher probability of sustaining a farm workplace injury relative to those working on other types of farm enterprises.⁴ Thus, occupational injuries have the potential to adversely affect the sustainability of individual dairy farms. Expansion of dairy farming production as a potential safety risk factor has not been studied extensively.⁵ Therefore, the objective of this study was to investigate the relationship between the expansion of dairy farming and the incidence of occupational injuries in Ireland.

In Ireland, the dairy farming sector has expanded rapidly in production since 2010.⁶ This expansion has been facilitated principally by the phasing out and eventual removal of EU Milk Quotas in 2015. Consequently, the number of dairy cows and milk production both saw significant growth, with a 33.6% increase in cow numbers and a 40.2% increase in milk output from 2010 to 2017. To allow this expansion, significant changes in major areas such as land, livestock, labor, infrastructure, and machinery were required. However, based on nationally weighted data, it is estimated that occupational injuries on Irish dairy farms have almost doubled in frequency from 9.2% to 18% per farm over the period from 2011 to 2017.^7,8 Thus, it is important to seek new knowledge on Irish dairy farming safety risk factors, particularly for an expanding farm production scenario to assist with improving the safety record of the enterprise. In particular, identifying risk factors associated with the balance in factors of production deployed in farming systems is crucial to gain maximum benefit from safety programs.⁵ Hence, this paper seeks to determine risk factors of occupational injury in the Irish dairy farming sector and to estimate the roles of both dairying expansion and the discipline of agricultural extension in influencing reducing injury occurrence which has potential to influence both farm management and development, in influencing reducing injury occurrence.⁹ The study uses a nationally representative survey of Irish dairy farmers who participated in a survey of occupational injuries and where farm socio-economic, physical, and financial data were separately gathered. Prior to outlining the study methods, a brief review of dairy farming in Ireland is undertaken to provide a context for the study for readers.

Dairy farming in Ireland

Spring calving pasture-based systems of milk production predominates (c.90%) in Ireland,⁶ which is based on the cost advantage of feeding stock with grazed grass at pasture with a short indoor winter period (3–5 months) where stock are predominantly fed conserved grass (silage). Dairy cows typically calve during months February to April and go to pasture shortly afterwards. A lower proportion of dairy farms produce “winter milk” with cows calving in September or October to produce milk over the winter months, primarily for human consumption and manufacturing continuity. This system commands a higher milk price to defray higher costs including feeding, indoor housing, and labor costs. The main work tasks on dairy farms include, milking, feeding, animal husbandry, and machinery work such as fertilizer and liquid manure spreading, along with farm maintenance and administration.¹⁰ Milking is a major source of labor demand (29%) of worktime and is mainly done by farmer machine milking (c97%) with a relatively low, but increasing proportion done by of robotic milking. Family labor is predominantly used for dairy farming with the farm operator as the main provider.⁶ Within Ireland, dairy farming is carried out across the country where soil-types allow, but is mainly concentrated in the south, midlands and north east.¹¹

Materials and methods

Participants and procedure

Gaining data on dairy farm workplace occupational injury occurrence (hereafter referred to as “injury”) was the prerequisite requirement to conduct this study. A study has previously been published that reports on the systems in place in Ireland to collect data on both fatal and non-fatal accidents.¹² Data for the current study was obtained via the Irish National Farm Survey (NFS), which operates as part of the pan-European Farm Accountancy Data Network (FADN).¹³ The legal mandate of FADN requires EU Member States to collect farm data that is largely technical and financial in nature (described as core data). However, individual Member States may discretionally collect additional farm data beyond this, such as the farm workplace occupational injury surveys in Ireland. In the case of this study, permission was obtained from Teagasc, the Irish State Agriculture and Food Development Authority, who implement the NFS in Ireland on behalf of FADN, to conduct an additional farm workplace occupational injury survey among a nationally representative sample of dairy farms. Teagasc, as a statutory agency, meets rigorous FADN requirements for conducting both core and additional surveys and use of data collected.

Survey measures

In 2017, 260 farm (83.6% of total dairy farm sample) injury survey questionnaires for dairy farms were completed by NFS recorders based on interviewing the principal person operating a farm referred to as the Farm Operator (F.O.). Questions asked related to farm injuries that occurred on the farm in the previous 5 years and were identical to previous NFS additional survey questions asked.¹³ The principal question asked of the F.O. was as follows: “Did you or anybody in your household or a worker on your farm have an accident causing injury in the last 5 years”. Where the F.O. reported an injury occurrence according to this definition the following additional categorical questions were asked: year of occurrence, person injured; first aid or medical treatment received, work time lost and location of accident. Table 1 provides consolidated information on categories used. Furthermore, core data for farms surveyed in 2017 for 2010 was available for 78.1% (203) of the farms, enabling the study to test independent variables associated with the occurrence of occupational associated with farm changes injuries.

Table 1. Distribution of dairy farm injuries by type, location, medical treatment and estimated work days lost in Ireland (n = 51).

	n	%
Person Injured
Farm Operator	42	82.3
Injury type
Livestock	20	39.2
Machinery (incl.Chainsaws)	21	37.3
Trip/falls, Buildings, Other.	12	23.5
Location
Farmyard	38	74.5
Building & Fields	13	35.5
Injury Treatment
Hospital treatment	39	76.5
Doctor, First Aid, None.	12	23.5
Estimated Work Days Lost
0–3	14	29.7
4–10	10	21.3
11–60	5	21.2
61 or more	13	27.7
Year of Injury
2017	13	25.4
2016	10	19.6
2015	14	27.4
2013/14	14	27.5

At an operational level, farms participating in NFS are visited throughout the year by trained farm data NFS recorders, who with each participating farm operator, assembles core data using FADN accountancy protocols and nationally required additional farm data.

Variables were selected based on a possible association with occupational injury based on previous research,^12,13 while screening out co-related variables. Descriptions of NFS variables as used in 2017 are available in the NFS Report for year 2020. In total 46 variables were used, including 7 F.O. personal variables and 14 farm variables, 15 farm rate variables, which used a combination of two variables to express a rate measure along with 8 farm change variable for the period 2010–2017.

The variables used were as follows:

Personal F.O. variables (n = 7) age; agricultural education level; discussion group membership; extension use (public and private); household size; marital status; profit monitor completion

Farm variables (n = 14): derogation; farm income; farm size; gross output; hours worked; investment in farm; labor units; off farm employment (farm operator); region; standard labor units (SLU); soil class; total hours worked. Machinery costs as a percentage of both total costs and total investment, were also included as farm variables.

Farm rate variables (n = 15): the variables economic size; gross margin; gross output, income and investment, were each divided by farm size, labor units, livestock units, as denominator giving nine farm rate variables.

Farm Change variables (years 2010–2017) (n = 10). Positive change of gross output, land farmed, Labor Units, SLU’s. Percentage increase in gross output, land farmed Labor Units, SLU’s.

Descriptions of variables

Complete descriptions of the NFS variables used in this study are available in the NFS report.¹⁴ Concise descriptions of key variables used in this study are now provided. Agricultural education is categorized as low (having no training, experience only), medium (sub “green certificate”level) or high (full “green certificate” or 3^rd level).¹⁵ Economic size is an economic measure of a farm business size measured in Economic Size Units (ESU). Gross output is the total sales less purchases plus value of farm produce used by household, plus receipts for farm-related services sold and inventory change. Income (referred to as Family Farm Income) is gross output less total net expenses, this represents the total return to the family labor, management, and capital investment in the farm business, but excludes off farm employment income. Investment is the current value of all the assets deployed by a farm including land, farm buildings, plant and machinery, livestock, crops, and financial assets. A labor unit is a measure of the farm labor required to work a farm based on standard time requirements per hectare for different crops, and per head for various categories of livestock. A standard labor unit (SLU) is defined as at least 1,800 hours required for farm work by a person over 18 years-of-age. A full- or part-time farm, respectively, are classified as requiring either less than or equal too and more than 0.75 SLUs. Derogation refers to permission to farm at a higher stocking rate (2.5 versus 2.2 livestock units per hectare) under the EU Nitrates directive at that time. Regions were defined in accordance with the EU NUTS11 classification system¹⁶ with two Regions South and East (SE) and Border, Midlands, West (BMW) used. Soil type is classified by use range as being “good” or “poor” using soil science categorizations.

Statistical analysis

To examine the causal relationship between four categories of independent variables (personal F.O., farm, farm rate, farm change variables) and the occurrence of an injury in the past 5 years (Yes/No), a three-step statistical testing process was conducted to ensure that all Binary Logistic Regression assumptions are met.¹⁷

In the first step, a normality test was performed using the Kolmogorov–Smirnov statistical test on continuous numeric variables that were measured on a ratio/interval scale. The variables subjected to this test included F.O. age, economic size, gross output, household size, income, investment labor units, and farm size. Among these variables, only F.O. age demonstrated normality according to the assumptions. Based on the statistical testing protocol, an independent T-test was employed to explore the relationship between injury occurrence and F.O. age, while the Mann-Whitney U test was used to investigate the relationship between injury occurrence and all other continuous variables that did not meet the assumption of normality. Furthermore, a Pearson Chi-square test was utilized to examine the correlations between categorical variables measured on nominal and ordinal scales. These variables included agricultural education level, discussion group membership, extension use, derogation, region, and the occurrence of an injury. The test compared the proportions of F.O.s who did or did not report an injury to identify any significant associations.

In the second step, the 37 independent variables (with at least 10 observation in each category/cohort) chosen as possibly influential to injury occurrence were analyzed using bivariate cross tabulation and statistical testing. In the third step, seven independent variables that were statistically significant in influencing the dependent variable “injury occurrence” in the second step were selected for inclusion in Logistic Regression Analysis (LRA). This statistical technique was used firstly at a bivariate level to allow calculation of predicted change effect of independent variables on the dependent variable. Variance Inflation Factor (VIF) and Tolerance were measured to test the multicollinearity. Chi-square test was applied to evaluate the goodness of fit of LRA to assess the causal relationship between seven variables.

Odds Ratios (OR) with 95% confidence intervals were estimated to show the causal relationship between independent variables and injury occurrence. LRA was applied initially at a bivariate level and where significant (p < .05) it was followed by multivariate LRA to indicate variables independently associated with the dependent variable “injury occurrence”. As injury data was not normally distributed for LRA analysis continuous variables were recoded into categorical variables by dividing the data above and below the median values. To analyze the statistical tests, SPSS 27 for Windows,¹⁸ was used.

Results

Descriptive analysis

According to the NFS survey, approximately 20% (n = 51) of dairy farms reported experiencing an injury within the previous 5 years. The highest level of these incidents occurred between years 2015 and 2017. Among the victims, farm operators accounted for nearly 82% (n = 42), with the remainder being spouses and family members, workers, and others. Farm operators have consistently experienced injury over the past 5 years (Table 1). The percentage of dairy farm injuries in this study (16.1%) is higher than that of a comparable study NFS study conducted in year 2011 (9.4.%).⁹ The main types of injuries were related to “livestock” (n = 20, 39.2%) and “farm vehicle/machinery” (n = 14, 27.5%). Injuries primarily occurred in farm yards, followed by fields, and farm buildings, making farm yards the most hazardous areas for dairy farm operators in terms of injury risk. In terms of treatment received for an injury, a significant majority (76.5%) were severe, resulting in hospital treatment, with the remainder receiving treatment from a medical doctor, and first aid or no treatment. Concerning the impact of an injury on workdays, 70.3% of injuries led to a loss of 4 or more work days, with 4 or more days being the statutory reporting period. Over 27 (27.7%) suffered an injury resulting in a loss of 61 or more days indicating their potential impact.

Correlation analysis

Personal F.O. variables and injuries

The results of the independent sample t-test analysis revealed no significant difference in the average age of victims who reported injury occurrence compared to those who did not (Table 2). Therefore, age did not appear to be a contributing factor to injury occurrence in dairy farming, as the average age of this cohort of farmers was 53.12 (±10.21) years. The Chi-square analysis indicated no significant relationship between the level of agricultural education of dairy farmers and the prevalence of an injury occurrence. However, there was a slightly higher likelihood of experiencing such injuries among individuals with “medium” and “low” levels of education (Table 2). Participation in farm discussion groups (FDGs) did not show any association with injury occurrence. Similarly, there was no significant correlation between Extension Service use (public and private) and injury occurrence. Therefore, involvement in FDGs and the utilization of advisory services were not found to be associated with the occurrence of an injury. There was no difference in the farm household size between dairy farmers reporting or not reporting an injury. Furthermore, no statistically significant correlations were found between the completion of profit monitoring and the marital status of dairy farmers and injury occurrence. Thus, personal factors related to farm operation were not found to be determinants of injury occurrence and, thus, were excluded from the LRA (Table 2).

Table 2. Correlation and compare mean analysis of injury occurrence on Irish dairy farms with independent variables.

	Injury occurrence (N = 260)
Sets of variables	Yes (50^1, 19.2%)	No (210, 80.8%)	Crude odds of injury	Test	P value
Personal F.O. variables
V1. Age	52.1 ± 11.3	53.3 ± 10	N/A	-0.77^a	0.452
V2. Level of agricultural education				1.261^b	0.531
Low (61, 23.46%)	(13, 21.3%)	(48, 87.7%)	0.27
Medium (114, 43.85%)	(24, 21.1%)	(90, 78.9%)	0.27
High (85, 32.69%)	(13, 15.3%)	(72, 84.7%)	0.18
V3. Farm discussion group participation				1.067^b	0.301
Yes (120, 46.15%)	(24, 20%)	(96, 80%)	0.25
No (139, 53.47%)	(26, 18.7%)	(113, 81.3%)	0.23
V4. Extension’ service use
Yes (216, 83.08%)	(44, 20.4%)	(172, 79.6%)	0.26	1.067^a	0.301
No (44, 16.92%)		(38, 86.4%)	0.16
V5. Farm household size	3.44 ± 1.59	3.36 ± 1.46	N/A	-0.300^c	0.764
V6. Profit monitoring				1.112^a	0.292
Yes (123, 47.31%)	(27, 22%)	(96, 78%)	0.28
No (137, 52.69%)	(23, 16.8%)	(114, 83.2%)	0.20
Farm variables
V7. Derogation				5.337^b**	0.02
Yes (92, 35.38%)	(25, 27.2%)	(67, 72.8%)	0.37
No (164, 63.08%)	(25, 15.2%)	(139, 84.8%)	0.18
Missing (4, 1.54%)
V8. Soil class				2.659^b*	0.05
Good (153, 58.85%)	(34, 22.2%)	(119, 77.8%)	0.29
Poor (106, 40.77%)	(15, 14.2%)	(91, 85.8%)	0.16
Missing (1, 0.38%)
V9.Region				4.877^b	0.02
SE (204, 78.46%)	(45, 22.1%)	(159, 77.9%)	0.28
BMW (56, 21.54%)		(51, 91.1%)	0.10
V10. Work status				0.971^a	0.241
Full-time (240, 92.31%)	(44, 18.3%)	(196, 81.7%)	0.22
Part-time (18, 6.92%)		(13, 72.2%)	0.38
Missing (2, 0.77%)
V11. Gross output (€)	(275718.96 ± 160135.48)	(237597.91 ± 144466.56)	N/A	-1.54^c	0.12
V12. Family farm income (€)	(116422.25 ± 79142.16)	(97694.69 ± 59471.31)	N/A	-1.20^c	0.24
V13. Family size (Ha)	(76.49 ± 43.08)	(67.66 ± 34.23)	N/A	-1.12^c	0.21
V14. Labor units	(1.96 ± 0.86)	(1.67 ± 0.71)	N/A	2.26^c**	0.01
V15. Standard man-days	(661.15 ± 350.46)	(557.28 ± 304.9)	N/A	2.59^c*	0.04
V16. Hours worked	(2485.62 ± 607.41)	(2458.17 ± 537.05)	N/A	0.316^c	0.20
V17. Total machinery cost (1000€)	(8.72 ± 4.18)	(7.74 ± 4.41)	N/A	0.437^c	0.50
V18. Total Investment in machinery (1000€)	(22.55 ± 9.1)	(22.66 ± 10.43)	N/A	0.302^c	0.50
Farm rate variables
V19. Gross Output/Ha	(3642.48 ± 1135.03)	(3533.04 ± 1156.71)	N/A	0.099	0.62
V20. Gross Output/Livestock Unit	(1778.90 ± 301.81)	(1809.25 ± 370.28)	N/A	-0.538	0.59
V21. Gross Output/Labor Unit	(143540.87 ± 57887.92)	(143128.25 ± 68611.92)	N/A	0.476 ^c	0.64
V22. Investment/Ha	(15359.53 ± 9418.62)	(13434.64 ± 7041.78)	N/A	3.56 ^c*	0.04
V23. Investment/Livestock Unit	(11479.01 ± 4691.94)	(12097.39 ± 5070.81)	N/A	0.099 ^c	0.87
V24. Investment/Labor Unit	(1704297.89 ± 1171246.84)	(1491107.56, ±927224.62)	N/A	-0.972 ^c	0.35
V25. Gross Margin/Ha	(2419.42 ± 843.26)	(2319.28 ± 811.37)	N/A	0.778 ^c	0.43
V26. Gross Margin/Livestock Unit	(94949.42 ± 38299.07)	(94109.29 ± 45154.88)	N/A	-0.445 ^c	0.64
V27. FFI/Ha	(1523.74 ± 689.33)	(1473.80 ± 619.69)	N/A	-0.312 ^c	0.75
V28. FFI/Livestock Unit	(11479.01 ± 4691.94)	(12097.39 ± 5070.81)	N/A	-0.383 ^c	0.71
V29. FFI/Labor Unit	(59452.91 ± 27748.04)	(59844.64 ± 31681.27)	N/A	0.396 ^c	0.63
V30. Livestock unit/Ha	(2.05 ± 0.52)	(1.94 ± 0.50)	N/A	1.229^c	0.22
V31. Livestock unit/Labor Unit	(81.14 ± 32.20)	(79.29 ± 35.84)	N/A	1.558 ^c	0.51
V32. Farm Change
V33. Positive change in gross output				0.635a	0.42
Yes (179, 68.85%)	(36, 20.1%)	(143, 79.9%)	0.25
No (26, 10%)		(19, 73.1%)	0.37
Missing (55, 21.15%)
V34. Positive change in Labor Units				4.580^b*	0.03
Yes (99, 38.08%)	(27, 27.3%)	(72, 72.7%)	0.38
No (106, 47.77%)	(16, 15.1%)	(72, 72.7%)	0.18
Missing (55, 21.15%)
V35. Positive change in Standard Man–Days				0.535	0.465
Yes (129, 49.62%)	(25, 19.4%)	(104, 80.6%)	0.24
No (76, 29.23%)	(18, 23.7%)	(58, 76.3%)	0.31
Missing (55, 21.15%)
V36. Positive change in Land Farmed				0.0	0.99
Yes (105, 40.34%)	(22, 21%)	(83, 79%)	0.25
No (100, 38.46%)	(21, 21%)	(79, 79%)	0.27
Missing (55, 21.15%)

¹Out of the 51 dairy farmers who reported injury occurrences, one victim’s information was missing or incomplete. Consequently, 50 victims were included in the correlation and inferential analysis.

^aIndependent samples T-test.

^bChi-square test.

^cIndependent samples Mann-Whitney U test.

Farm variables and injuries

A significant correlation was observed between “derogation” and the occurrence of injury. Dairy farmers who responded “yes” to “derogation” reported an injury at nearly twice as frequently as those who did not. Dairy farmers working on “good” soil class reported significantly higher injuries compared to those working on “poor” soil class. Injuries were nearly twice as high among dairy farmers in the SE region compared to those in the BMW region. There was no significant correlation between the “work status” of dairy farmers and injury occurrence, although part-time dairy farmers experienced slightly higher rates of injuries than full-time dairy farmers. The gross output of farmers who reported an injury was slightly higher than those who did not experience one, but the difference was not statistically significant (Table 2). Despite reporting slightly higher family farm income, dairy farmers who experienced an injury did not show a statistically significant difference compared to those who did not experience one. Similarly, there was no significant difference in farm size between dairy farmers who reported an injury and those who did not. Dairy farmers experiencing an injury had significantly higher labor units than their counterparts (see Table 2). Additionally, dairy farmers who reported an injury had significantly higher “standard man-days” units compared to their counterparts. There was no significant difference in the average number of hours worked between dairy farmers who reported an injury and those who did not. The levels of machinery cost and total machinery investment were nearly identical among dairy farmers who did and did not experience an injury.

In summary, “derogation”, “soil class”, “region”, “labor unit”, and “standard man-days” were significantly correlated with injury occurrence and these variables were inserted into LRA.

Farm rate variables and injuries

Dairy farmers who experienced an injury reported similar levels of gross output per hectare, gross output per livestock unit, gross output per labor unit, investment per livestock unit, investment per labor unit, gross margin per hectare, gross margin per livestock unit, gross output per labor unit, FFI per hectare, FFI per livestock unit, FFI per labor unit, livestock unit per hectare, and livestock unit per labor unit compared to dairy farmers not reporting injury. However, there was a significant difference in the investment per hectare, which was higher among dairy farmers who reported an injury in the past five years. Therefore, “investment per hectare” was included in the LRA.

Farm change variables (years 2010– 2017) and injuries

There was no significant difference in injury occurrence between dairy farmers who reported a “positive change in gross output” compared to other dairy farmers. However, according to the cross-tabulation analysis, dairy farmers who experienced a “positive change in labor units” reported twice the level of injury occurrence compared to other dairy farmers. There was no correlation between a positive change in “ standard man days” and the level of injury occurrence. Similarly, there was no difference in the occurrence of injury between those who had a positive change in land farmed and other dairy farmers. Therefore, only a positive correlation was found between a positive change in labor units and the occurrence of an injury.

Binary logistic regression analysis

The Chi-square test of the regression model yielded a statistically significant result (Chi-square = 13.52, p = .03), indicating that the binary regression model, which incorporated correlated variables such as “derogation,” “soil class”, “region” “labor units,” “standard man–days”, “investment per hectare” and “positive change in labor units,” was a good fit. The findings from the LRA revealed that all seven variables were significantly associated with injury occurrence. Among these variables, a positive change in labor unit emerged as the strongest predictor of injury occurrence. Dairy farmers who experienced a positive change in labor units reported nearly twice injury occurrence compared to other dairy farmers. Similarly, dairy farmers who responded affirmatively to “derogation” had double the number of injuries compared to their counterparts. The findings of this study suggest a clear causal relationship between soil quality and the occurrence of an injury, as dairy farmers working on “good” quality soil reported a significantly higher injury rate. The LRA also demonstrated a positive relationship between “investment per hectare” and injury occurrence, with dairy farmers who invested more per hectare reporting a significantly more injury occurrence. “Standard man days” was identified as a risk factor positively associated with injury occurrence, indicating that dairy farmers with higher “standard man–days” were more likely to experience an injury. Likewise, dairy farms with larger total “labor units” faced a higher likelihood of injury. Finally, the variable “region” emerged as the last contributing factor to injury rate, with dairy farmers from the SE region reporting higher rate compared to their counterparts from BMW (Table 3). Consequently, the expansion of a dairy farm may heighten the likelihood of an injury occurrence among dairy farmers, particularly those who work on “good” quality soils in the SE.

Table 3. Factor affecting injury occurrence in Irish dairy farms.

Independent variables	Dependent Variable: Injury occurrence or non-occurrence Bivariate regression
	OR	95% CI	P
Positive change in Labor Units
Yes	2.109
No	(Indicator variable)	1.056–4.21	.01
Derogation
Yes	2.07
No	(Indicator variable)	1.10–3.88	.029
Soil class
Good	1.73	0.89–3.78	.042
Bad	(Indicator variable)
Region
SE	0.446	0.13–0.92	.035
BMW
Labor units	0.662	0.45–0.96	.032
Standard man-days	0.999	0.99–1	.041
Investment/Ha	1.230	1.1–1.35	.05

Discussion

This paper makes the following two distinct contributions to the literature. Firstly, it indicates how NFS core farm data collected at different time points can be used in association with an additional injury survey to provide data to identify risk factors associated with expansion in output of dairy farms in Ireland. Secondly, the study identifies injury risk factors associated with dairy farming and expansion in output in Ireland in the period 2010–2017. This new knowledge is of value in developing farm health and safety (FHS) programs tailored for dairy enterprises related to both expansion and scale. It is evident from the study data that economic development of dairy farms may result in inferior farm health and safety levels on dairy farms if FHS programs are not developed and implemented. These two contributions are now discussed separately, and conclusions are drawn.

Use of accountancy data to identify risk factors

This paper extends the work of a previous study¹² that demonstrated how the use of data in association with a farm workplace injury survey could identify farm operator injury risk factors. In the current study, a more focused approach was applied to study dairy farming during an expansionary production period using NFS data collected at two points in conjunction with a retrospective farm workplace occupational injury survey to identify injury risk factors. Thus, the data available indicates that where farm socio-economic and accountancy data is routinely collected, it has the potential to be used to study the impact of major changes in farm systems in association with additional injury survey data.

Risk factors associated with dairy expansion in Ireland

This study found that dairy farms in 2017 had higher injury levels compared to a previous 2011 study.⁹ Dairy farm injuries in the current study were associated mainly with livestock, machinery, and trips and falls, which accords with other literature which reported that dairy farming has a wide range of hazardous tasks and where work is conducted at high intensity.^19,20

Our study indicates dairy production expansion has the potential to increase injury risk. Dairy farm expansion was associated with increases in “labor units” and “standard man–days”, which are both NFS measures of increased labor used. High levels of labor use has been reported as associated with increased farm injury.^21,22 Thus, a change in workload relative to the labor available, as a consequence of expansion, may have led to increased injury rate. Previously, it has been reported that Irish farmers consider workplace injury risk is associated with both physical and organizational risk factors,²³ with physical risk being principally associated with machinery and livestock, and organizational risks being associated with work overload, hurry, and tiredness, findings with similarities to the current study.

Specific risk factors found in this study include farm classification as “derogation”, “good soil” class and “SE region”. These risk factors may arise as farms that are intensively farmed on favorable soil types may be most likely to expand production and have a higher workload.

Injury levels in our study were also associated with increased investment per hectare. It is probable that this investment was associated mainly with elements associated with increasing economic production including land, cows, milking facilities, and cow housing and, while the infrastructural component of this investment is likely to be done to a high FHS standard, it may be that overall increased investment led to increased workload per worker.

Overall, the study data point towards elevated injury levels on farms in an expansionary phase, which are intensively managed from a milk production perspective. The study indicates that to reduce injury levels on such dairy farms, enhanced farm management, including labor and occupational health and safety management is required.

The finding that labor level is associated with injury rate presents a challenge for the Irish dairying sector going forward, as family labor is about maximized, and hiring suitable labor has become challenging, due partially to the peak labor demand in spring.⁶ An alternative to dairy farm labor shortage is to maximize the use of technological adoptions that substitute for labor.^24–26The studies cited have demonstrated potential to reduce workload and increase labor efficiency on Irish dairy farms by technology adoption. Regarding implementation of FHS measures on Irish farms, including dairy farms, limited uptake of such measures has been reported,²⁷ indicating potential for increased adoption.

Our study also found that the use of extension services and farm discussion group participation were not associated with dairy farm injury occurrence. This is in accordance with a recent study that found limited focus by discussion groups on occupational health and safety.²⁸ Also, a further recent study indicates that leading farmers, and farmers generally from a business and production perspective, do not emphasize farm safety and accordingly are less likely to adopt FHS risk management.²⁹

Overall our study suggests that, in Ireland, to minimize injury levels on dairy farms, both workload and FHS needs to be managed through technology and practice adoption. The study suggests that guidelines related to worktime use per livestock unit may need to be re-assessed in the light of recent expansion in dairying, and that management strategies related to the balance between technology and labor deployed on dairy farm also may need to be re-assessed. The study also suggests an enhanced role for extension service engagement in both work organization and FHS. This is proposed as the current study indicates heightened injury rates for dairy farming in Ireland and that extension engagement by dairy farmers has not influenced injury levels, while work organization issues including both technology and practice adoption associated with expansion is closely associated with injury rates. An international review of extension engagement in FHS⁹ indicates that extension can be a powerful agent for positive improvement when extension agents (advisors) are trained in FHS and related work organization approaches and engage with farmers on these in both a practical and interactive way.

At an international level, dairy farming technologies and work practices vary greatly from country to country³⁰; however, a necessary first step to prevent farm workplace injuries is to gain suitable data both for injury occurrence and for the technologies and practices in use, to aid development of effective FHS prevention programs. This recommendation mirrors a recent call for more research in the areas of labor use efficiency and FHS on livestock farms.³¹

Strengths and limitations

This study used a large and representative sample of dairy farms where a considerable database of socioeconomic, farm physical, and accountancy data were available for each farm in addition to the injury survey undertaken. The injury data was collected by trained recorders in an interview setting with knowledge of both the farmer and farm. Injury survey data were collected based on farmer self-recall over the previous 5-year period. Re-call as an issue is acknowledged; however, re-call is likely to be biased towards serious injuries, which are the major focus of prevention programs. This assertion is supported by a review of farm workplace injury survey sources, which indicated that a previous NFS injury survey using 5-year recall provided similar estimates to a survey conducted among hospital emergency units and doctors.¹³ A recent report of an NFS farm work place injury survey using a 1-year recall indicates a higher level of injury reporting, but found that 22% reported no lost work time.³² Mandatory reporting of workplace injury for employed workers of 4 or more days of lost work time is a requirement of Irish/EU requirements. Thus, the current study gives robust estimates of mainly serious injuries, in line with current legislative requirements. The study data were analyzed systematically using inferential statistics to establish injury risk factors where statistical testing was not biased by farmer injury re-call.

Conclusions

This study demonstrates how a retrospective farm workplace injury survey on Irish dairy farms, analyzed in combination with farm socio-economic, farm physical and financial data gathered at two time points at an early and later stage in a production expansionary period can be used to identify injury risk factors, including those associated with expansion.

In common with other studies, this study found dairy farms had high injury levels, while dairy production expansion increased injury rates. Injury risks were associated with increased labor use and investment per hectare in factors of production. The study data found elevated injury rates occurred on farms that were intensively managed and in an expansionary phase from a milk production perspective. The study indicated the use of extension services and farm discussion group participation were un-associated with farm injury occurrence.

Overall the study suggests that, in Ireland, to minimize dairy farm workplace occupational injuries, workload needs to be managed and increased uptake of labor-saving and safety-related technologies and practices is required. The study also suggests the role of extension services in assisting with improving farm safety on dairy farms needs to be enhanced.

Acknowledgments

The author wish to acknowledge farmers participating in NFS and data recorders for their input to this study.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Summary statistics can be made available from the corresponding author.

Additional information

Funding

Support of an Irish Department of Agriculture, Food and the Marine Research Grant (BeSafe grant number DAFM: RESL043) for the study is acknowledged.