Introduction

The occupational hazards associated with farming are well recognized and can have impacts not only on physical health [1], but also mental health. For example, farmers in the United Kingdom (UK), United States (US), Australia, Finland, and Norway are reported to experience high levels or risk of stress [2,3,4,5], anxiety [6], depression [6,7,8], psychological distress [9], and psychological morbidity [10]. Regrettably, estimates of mental health outcomes in Canadian farmers are lacking. Given the important role farmers play in Canadian economy and international trade [11], and the global need for increased and sustainable agriculture to meet the challenge of feeding 9 billion people by 2050 [12], this represents a serious gap in knowledge.

Resilience is the ability to “thrive in the face of adversity” and is a learned process that can help to protect against occupational stresses and mental illness [13; p. 76]. However, very little research has been conducted on resilience in farmers, representing another important area for investigation.

Addressing the gaps in understanding of mental health and resilience amongst farmers is key to supporting agriculture and food systems, informing resource and training development, and allocating resources and service delivery. Hence, the objectives of this study were to estimate the prevalence of stress, anxiety, depression, and resilience amongst Canadian farmers.

Methods

Study design

A cross-sectional study was conducted from September 12, 2015 to February 9, 2016, using an online questionnaire administered via Qualtrics (Provo, UT, USA). We contacted provincial and national agricultural organizations across Canada by email and/or telephone to ask them to share news of the study and distribute the survey web link to their members via emails, listserves, newsletters, magazines, and social media (predominantly Twitter). Inclusion criteria were: being 18 years of age or older, able to read and write in English, and being a self-identified farmer from any Canadian agricultural commodity group. Informed written consent was obtained prior to starting the questionnaire, and the data were collected anonymously. Participants were given the opportunity to submit their email address (separate from their questionnaire responses) to be entered into a draw to win one of three $250 cash prizes. The Research Ethics Board at the University of Guelph approved the study (15-JN-007).

Questionnaire

The questionnaire contained the following three validated, self-report, mental health scales:

  • Perceived Stress Scale (PSS): this commonly used, 10-item scale, with established reliability and validity [14], measures “a global level of perceived stress”, “the degree to which situations in one’s life are appraised as stressful” and how “unpredictable, uncontrollable, and overloading” respondents found their lives to be in the previous month [15; p. 387]. Items are measured on a 5-point Likert scale (0–4). Responses were analyzed as per the scale instructions; all items were summed to produce an overall score (range 0–40), where greater scores signify higher levels of perceived stress.

  • Hospital Anxiety and Depression Scale (HADS): this is a 14-item measure with two distinct subscales: anxiety (“restless, anxious moods, and thoughts”) and depression (“state of loss of interest and diminished pleasure response”) [16; p. 2] in the past week. Both subscales contain seven items; responses are scored on a 4-point Likert scale (0–3) yielding a score range of 0–21 [16]. As per scale instructions, the anxiety and depression subscales were scored separately by summing the items and then classifying “caseness” as: Normal (0–7), Mild (8–10), Moderate (11–15), and Severe (≥ 16) [16]. Other authors have also defined “possible” (cut-point score of ≥ 8) and “probable” (cut-point score ≥ 11) case categories for anxiety and depression [17]. The HADS has been extensively validated [18] and has been used in studies of farmers in England [2], France [19], and Norway [6, 8]. Nevertheless, there has been some question as to whether the HADS is able to consistently differentiate between anxiety and depression constructs (e.g. [20]), leading to investigation of the HADS-total score (i.e., the sum of the scores from HADS-A and HADS-D) as a measure of overall psychological distress [20, 21]. A review of the HADS included 16 different articles with cut-point scores for HADS-total ranging from ≥ 8 to ≥ 21 [18]. Pallan and Tennant reported the HADS-total with cut-point score of ≥ 12 to be a viable means of classifying psychological distress [21]; hence, we also used this approach here.

  • Connor–Davidson Resilience Scale (CD-RISC): this 25-item scale is used to measure resilience as “successful stress-coping ability” [13; p. 77]. It is reported to have sound psychometric properties and good internal consistency and test–retest reliability [13, 22]. As per scale instructions, responses to 5-point Likert scale (0–4) were summed to produce an overall resilience score (range 0–100), with higher scores indicative of higher levels of resilience.

Questions were also included to collect participant demographic data, including age, self-identified gender, and farming commodity. Two questions (“Are you currently on medication for mental illness (e.g., depression, anxiety, schizophrenia, bipolar disorder)?” and “Have you ever suffered from a mental illness in the past?”) were used to provide insight into self-reported history of mental illness. The full questionnaire was designed to take approximately 15–20 min to complete.

Sample size

As there were no published prevalence estimates of the mental health outcomes of interest amongst Canadian farmers for use in sample size calculations, an a priori estimate of 50% was used conservatively to maximize the required sample size. A minimum required sample size of 385 farmers was calculated using an allowable error of 5% and confidence level of 95% [23].

Statistical analyses

Descriptive statistics [means, medians (M), interquartile ranges (IQR), and percentages] were used to examine the data. Person mean imputation was used to account for missing data if only one value was missing per scale (PSS, CD-RISC) or subscale (HADS), as has been reported elsewhere [17, 24]. If an observation had two or more missing values per observation per mental health scale or subscale, it was deemed invalid and dropped from that analysis. To provide some insight for the potential for non-response bias, we applied the extrapolation method of comparing early-to-late responders, as described by Lindner et al. [25]. In this analysis, late responders (defined here as participants responding in the last third of the study period) are used as a proxy measure of non-responders, and statistical comparisons are made between early and late responders on primary variables of interest as a means of addressing non-response error [25]. Chi square tests were used to assess whether differences in mental health outcomes existed between early and late responders. All statistical analyses were conducted using Stata v.15 (StataCorp, College Station, TX, USA).

Results

Study population

A total of 1132 participants returned a questionnaire. As no questions were mandatory, response percentages and sample sizes per question vary, as noted. The average participant age (n = 980) was 46.6 years (SD 13.3; IQR 28–64; range 19–88), which is lower than the average age reported for farm operators in Canada (55.0 years) [26]. Of participants responding to the open-ended question on gender (n = 977): 69.4% identified as male, 30.4% identified as female, and 0.2% identified as gender-queer. For reasons of maintaining anonymity and given small group sample size in the gender-queer category, data for these participants were not reported separately, and were instead reported with the missing gender category. The male and female breakdown observed here is close to the sex breakdown reported for farm operators in 2016 in Canada (71.3% male and 28.7% female) [27]. In identifying their relationship status (n = 986), most participants were married (77.8%) or in a committed relationship (9.9%); the remainder were single (9.0%), separated/divorced (2.0%), widowed (0.7%), or identified their relationship status as “other” (0.5%). One-quarter (242/970) of participants self-reported having a previous mental illness; 9.1% (88/969) reported currently taking medications for mental illness.

All provinces in Canada were represented (Table 1), although Ontario participants were the majority (72.5%). The percentage of participants by province is compared to 2016 national census data of the number of farm operators per province [28] in Table 1. Participants (n = 1132) were asked to select from a list all of the commodity groups to which they belonged. The represented commodities are presented in Table 1 and compared to 2016 national census data for farm type [29]. A total of 708 (62.5%) participants identified as crop growers; of these, 529 (46.7% of all participants) grew crops and were involved in at least one other commodity, and 179 (15.8% of all participants) only grew crops. Approximately half of all participants were involved in more than one commodity (53%; 600/1132).

Table 1 Breakdown of farm type and province in the study sample compared to 2016 Canadian census data (Statistics Canada, 2016 Census of Agriculture—Table 32-10-0403-01 Farms classified by farm type and Table 32-10-0440-01 total number of farms and farm operators)
Full size table

Perceived stress

The average PSS score was 18.9 (n = 1127, SD 4.9, M = 19, IQR 15–22). Of these, 972 participants self-identified their gender: amongst men (n = 677), the average score was 18.3 (SD 4.9, M = 18, IQR 15–22); amongst women (n = 295), it was 20.1 (SD 4.9, M = 20, IQR 16–24). For the 153 participants who did not report gender or identified as gender-queer, the average score was 18.7 (SD 4.8; M = 19, IQR 15–22).

Anxiety and depression

The results from the HADS are presented in Table 2. Approximately 57% and 33% of all participants were classified as possible and probable cases for anxiety, respectively. Approximately 34% and 15% of all participants were classified as possible and probable cases of depression, respectively. Nearly two-thirds of participants (61.9%) had HADS-total scores ≥ 12, indicating psychological distress. Anxiety, depression, and HADS-total scores, and the percentage of probable cases of anxiety, depression, and psychological distress, were higher in females than in the overall sample and males.

Table 2 Hospital Anxiety and Depression Scale (HADS) subscale and total scores, and the prevalence of anxiety, depression, and psychological distress caseness amongst farmers in Canada (September 2015–February 2016)
Full size table

Resilience

Amongst the 968 participants who provided responses for the CD-RISC, the average resilience score was 71.1 (SD 15.6; M = 72; IQR 61–90). Amongst males (n = 659), the average score was 71.8 (SD 15.5; M = 73; IQR 62–83), and amongst females (n = 292), it was 69.5 (SD 15.7; M = 70; IQR 59–81). Amongst the participants who responded to the CD-RISC and did not specify gender (n = 15) or identified as gender-queer (n = 2), the average resilience score was 70.1 (SD 12.3; M = 71; IQR 66–78).

Discussion

To the authors’ knowledge, this is the first published study to quantify resilience in farmers and the first to investigate farmer mental health in Canada using validated psychometric tools. These results shed light on mental health issues affecting the agricultural community in Canada, which is particularly concerning given the importance of agriculture to Canadian economy, trade, and employment [11]. Comparisons of the present results with studies that used the same psychometric scales follows below; however, comparisons may be hindered by different study periods. Where specific Canadian data were unavailable, we drew on the international literature for interpretation of results. On-going qualitative work in the lead author’s research laboratory explores Canadian farmers’ perceptions of stressors that lead to negative impacts on mental health, and the resulting impacts this has on their farm production and home lives.

The mean PSS scores reported here for males (18.3) and females (20.1) are notably higher than reported normative data means from the US general population (12.1 and 13.7 for males and females, respectively) [30]. The PSS scores for males here are also higher than those reported for male grain farmers in Ohio, US (calculated average 10.9) [7] and rural men (not necessarily farmers) in Australia (15.0) [31]. Reasons for high stress amongst farmers have been reported to include: unpredictable weather; animal disease; economic stresses; overwork; burden of paperwork/bureaucracy; media criticism; and social isolation [2, 32]. High stress amongst farmers is associated with increased risk of farm injury [1, 33] and depression [7], while stress in general is associated with mental (e.g., depression and anxiety) and physical (e.g., cardiovascular disease and decreased immune function) health ramifications [34].

The percentage probable cases (scores ≥ 11) reported here for anxiety (males: 28.8%; females: 42.8%) and depression (males: 12.5%; females: 18.9%) were notably higher than reported general population norms for anxiety (males: 12.5%; females: 19.0%) and depression (males: 6.9%; females: 6.9%) in the UK [35]. Compared to farming populations, the anxiety and depression caseness percentages here are notably higher than reported for farmers in France [19]. The overall, male, and female HADS scores and possible caseness percentages are considerably higher than those reported for farmers in two studies in Norway [6, 8], and in the UK [2]. The HADS scores and percentage of participants classified with anxiety and depression are very similar to those reported for farmers in the UK [32]. Nearly two-thirds of participants were classified as having experienced psychological distress (HADS-total ≥ 12). We were able to find only one other study that reported HADS-total in farmers for comparison; our HADS-total score results were higher than those reported by Eisner et al. for a group of farmers after the 1996 ‘beef crisis’ in the UK [36]. The impact of anxiety, depression, and psychological distress on farm productivity is not well investigated; however, anxiety and depression have been reported to increase both absenteeism and presenteeism (i.e., working while ill), and presenteeism has been reported to be associated with work impairment and productivity loss [37]. Hence, in addition to important personal ramifications, poor mental health could have productivity implications for farmers as well.

Resilience is a “dynamic process wherein individuals display positive adaptation despite experiences of significant adversity or trauma” and is not “a personality trait or an attribute of the individual” [38; p. 858]. The average resilience scores of farmers in the present study (71.1) were lower than population norms reported for several general populations in the US (range 75.7–83.0) [39]. This is especially concerning given that resilience helps guard against the negative impacts of stress, and protects against negative mental health outcomes [13]. Many of the stresses that farmers face would be difficult, if not impossible, to eliminate (e.g., weather and disease outbreaks). Yet, resilience is a process that can be learned, and, thus, could help to protect farmers from negative stress and better thrive in their demanding occupations [40]. Greenhill and colleagues identified eight themes that influenced farmer resilience; several of which are reasonably actionable and could be used to promote farmer wellness. For example, being “more than a farmer” (e.g., hobbies and sport, and valuing other social roles like community leaders, parents), having “opportunities to disengage” (e.g., getting off-farm, and time with children, family, friends, community), and self-care (exercise, time to self, and time with friends) [40; p. 322]. Indeed, social support and sense of community have been reported to be positively correlated with male farmer well-being [31, 41]. Strategies to provide resilience training to farmers, and help them identify realistic, practical wellness activities could help farmers to better cope with the multitude of occupational stresses which they face.

Compared to males, female participants scored less favorably on all of the mental health outcomes studied here. This supports Booth and Lloyd [2] where female farmers in the UK reported higher scores for stress, depression, and anxiety than their male counterparts, and Brumby et al. [42] where female farmers in Australia scored higher in psychological distress than male farmers, as well as gender differences in mental health outcomes in the general population [43]. Females are reported to face numerous unique challenges in farming that can negatively impact their wellness [4, 40]. Two individuals identified as gender-queer, which regrettably, is too small a sample for analysis. This may serve as a reminder of the importance of gender diversity in questionnaires (indeed, both individuals thanked us for using an open-ended question for gender) and may inform lines of future research inquiry. Although this study was not designed to explore reasons for gender differences in farmer mental health, our results suggest that such investigation would be useful. Females and gender-queer individuals may represent a priority area for future research and interventions.

In addition to the implications of poor mental health on individual well-being and on farm productivity discussed above, another important consideration is the potential impact on animal welfare. The interdependence of farmer well-being and animal welfare has been described [44], and animal neglect cases have been reported to be associated with mental health and addiction issues, in addition to social problems, amongst the farmers that care for them [45, 46]. This concept of one welfare, which “recognizes the interconnections between animal welfare, human well-being, and the environment” [47; p. 412], represents an important area for further study.

Limitations

This cross-sectional study investigated mental health outcomes at one point in time, and did not use random sampling; thus, it represents only a ‘snap-shot’ of the mental health of participating Canadian farmers and limits the ability to infer causation. A French version of the survey was not offered, which limited our ability to reach French speaking members of our agricultural community; this oversight will be corrected in future work. The resulting limitations on generalizability of the study findings should be considered.

The sample of 1132 farmers represents approximately 0.6% of all farm operators in Canada [29]. While we did not have data on non-responders, there were no significant (p ≤ 0.05) associations between the mental health indices and time of survey completion (data not shown), with one exception: late responders were less likely to be classified in the severe anxiety category of HADS than people who responded early and mid-way through the data collection period (χ2 = 28.2; p = 0.021). As such, severe anxiety may be over-represented in our sample. However, it has been reported that people experiencing poor mental health are less likely to participate in surveys, which would underestimate the prevalence of mental illness [48]. As a broad approach to sampling was used (i.e., participants were not recruited proportionally by commodity type or by province), our sample is not proportional to the commodity structure or provincial breakdown in Canada (Table 1). The use of formal sampling frames (e.g., membership lists of agricultural organizations) and random sampling procedures proportional to commodity structure would help to avoid issues with sampling bias in future studies.

Our networking with farmers and the agricultural industry during survey development underscored the importance of having a succinct survey for this population. As such, several variables of interest (e.g., alcohol and drug use, suicide ideation, and suicide) were dropped from the questionnaire. Still, we observed a tendency for participants to stop answering questions near the end of the survey (e.g., approximately 1100 responses at the beginning of the survey vs. 965 toward the end). This highlights a methodological challenge of survey research and affirms that questionnaire length, completion time, and response effort are important considerations in designing assessment tools for this population. The use of quantitative research methods also limits the depth of exploration; qualitative approaches would complement these findings and enhance our understanding of farmer mental health.

Public health implications

The results on perceived stress, anxiety, depression, and low resilience amongst Canadian farmers underscore concerns and illustrate a critical need for research and interventions related to farmer mental health. Indeed, farmer mental health has recently been identified as a priority by Canada’s House of Commons agricultural committee [49]. In addition, a bipartisan bill, The Stemming the Tide of Rural Economic Stress and Suicide (STRESS) Act, recently introduced to the US government, calls for increased mental health services for farmers and ranchers [50]. The findings reported here can help to guide policy makers, physicians, and public and mental health service providers in terms of decision-making, policy recommendations, resource allocation, and mental health service and training provision.