Population group differences in subjective importance of meat in diet and red and processed meat consumption

Red and processed meat (RPM) consumption associates directly with several unfavorable health outcomes and with environmental impact of diet. RPM consumption differs between certain population groups, and moreover, encompasses various subjective meanings. Literature on determinants of subjective importance of meat in diet (SIM), however, is scarce. Aims of this study were to determine which sociodemographic and -economic characteristics associate with SIM and RPM consumption. The study was based on the FinHealth 2017 Study. The sample comprised 4671 participants aged 18-74 years. SIM was asked with a question including five response options from "not important at all" to "very important". Habitual dietary intake including RPM consumption was studied with a food frequency questionnaire. RPM consumption level grew in parallel with SIM categories. RPM consumption was high and SIM prevailing in men, those living in rural areas, and those with low education. Women living in household with children consumed more RPM than other women but did not find meat more important. Conversely, men living in household with children found meat more important but did not consume it more than other men. Domain analyses considering individuals within the highest RPM consumption quintile revealed that the oldest age group found meat significantly less important than the youngest group. In order to be able to lower RPM consumption at population level and to move towards healthier and climate-wiser diets, it is important to identify subgroups that consume much meat but also subgroups that find meat especially important. Such dietary transition may be especially challenging to subgroups that consume much meat and also consider it important. Actions to support the dietary transition in different population groups should be developed.


Introduction
Red and processed meat (RPM) consumption has been shown to directly associate both with poorer health outcomes (Ekmekcioglu et al., 2018;Yip, Lam, & Fielding, 2018) and with greater environmental impact (Clune, Crossin, & Verghese, 2017). For instance, risk of developing type 2 diabetes, cardiovascular disease or several types of cancer (Ekmekcioglu et al., 2018;Neuenschwander et al., 2019;Yip et al., 2018) increases in parallel with RPM consumption. Moreover, diets rich in RPM, especially beef and other ruminants' meat, have been shown to cause more greenhouse gas emissions and land use than diets with more plant-based foods (Hallström, Carlsson-Kanyama, & Börjesson, 2015;Hyland, Henchion, McCarthy, & McCarthy, 2017). Thus, reducing RPM consumption at population level has been suggested to aid both in tackling public health burden of non-communicable diseases and in fighting climate change (European Public Health Association., 2017;Hemler & Hu, 2019;Willett et al., 2019).
Meat, being a source of e.g. high quality protein, iron with good bioavailability and vitamin B12, has constituted an important part of human diet in many food cultures (Clonan, Wilson, Swift, Leibovici, & Holdsworth, 2015;Clonan, Roberts, & Holdsworth, 2016). Nowadays, reasons for meat consumption include, for instance, affordability, meat enjoyment, eating routines, health concepts and difficulties in preparing vegetarian foods (Clonan et al., 2016;Pohjolainen, Vinnari, & Jokinen, 2015;Verbeke, Perez-Cueto, Barcellos, Krystallis, & Grunert, 2010). Lea and colleagues (2006) concluded that a large proportion of an Australian study population was not yet ready to consume a plant-based diet. Those who were not ready did not recognize benefits of consuming plant-based diet, but experienced various barriers to it. Another Australian study showed that a general lack of awareness exists on implications of meat Abbreviations: FFQ, food frequency questionnaire; MJ, megajoule; OR, odds ratio; RPM, red and processed meat; SES, socioeconomic status; SIM, subjective importance of meat in diet. consumption on human and planetary health, and that assumed health reasons were the most common reason to consume meat (Marinova & Bogueva, 2019). Neff and colleagues (2018) reported that the most common reason for not reducing meat consumption was a belief that meat is necessary for a healthy diet. Meat-eating has been strongly associated with masculinity, performance, power and wealth (Bogueva & Marinova, 2020;Kildal & Syse, 2017;Leroy & Praet, 2015;Marinova & Bogueva, 2019;Rothgerber, 2013;Rozin, Hormes, Faith, & Wansink, 2012). Even though this association has generally abated, and not all men associate meat with masculinity (De Backer et al., 2020) in certain subgroups the ideal of meat-eating may still be prospering (Kubberød, Ueland, Rødbotten, Westad, & Risvik, 2002).
Some studies have examined the reasons behind willingness to reduce meat consumption. In a representative American study conducted in 2015, most common motives for reducing meat consumption were cost and health while environmental and animal welfare-related reasons were found less significant (Neff et al., 2018). A Finnish study concurred with this with healthiness being the most focal reason for changing consumption habits and environmental and animal welfare lagging (Latvala et al., 2012). Also a study on vegetarian motives concluded that health was the most common motive for non-vegetarians to consider a vegetarian diet (Hopwood, Bleidorn, Schwaba, & Chen, 2020).
In addition to meat consumption, meat attachment has been studied with a Meat-Attachment Questionnaire (MAQ) including 16 questions concerning hedonism, entitlement, affinity and dependence in regard to meat consumption (Graca, Calheiros, & Oliveira, 2015). Meat attachment has been shown to inversely relate to willingness to reduce meat consumption (Lentz, Connelly, Mirosa, & Jowett, 2018). In addition, Pohjolainen et al. (2015) found that barriers to consuming plant-based diet were more prevailing among men, younger individuals, those having a low level of education, those having children in the family, and those living in the rural areas. In the same study, frequency of meat consumption proved to be the strongest predictor of barriers to consuming plant-based foods.
Awareness and knowledge on effects of meat production and consumption on environment, health and farmed animals' welfare has increased, but how it influences consumers' meat consumption behavior and attitudes may distribute unevenly across sociodemographic and -economic groups. Even though quite many previous studies have examined distributions of meat consumption across certain sociodemographic and-economic groups, only a few previous studies have examined associations of meat attachment or barriers to consuming plant-based diets, and no studies exist on associations between subjective importance of meat in diet (SIM) and sociodemographic and -economic factors. Further, no studies have examined whether RPM consumption varies according to SIM and simultaneously according to sociodemographic and -economic factors. Discovering which subgroups, in particular, find meat in their diet important is essential when striving to move towards healthier and climate-wiser diets and to understand the challenges of such dietary transition. Thus, aims of this study were to determine which sociodemographic and -economic factors associate with SIM. Further, this study investigated whether consumption of RPM differed between either groups according to SIM or according to sociodemographic and -economic factors, and when considering these factors simultaneously.

Method
This study was based on data from the FinHealth 2017 Study, conducted in Finland in 2017 (Borodulin & Sääksjärvi, 2019). Aims of the survey were to produce information on health, wellbeing, health behavior and functional capacity of adult Finns with health examinations (including measurements and drawing blood samples), questionnaires and interviews. The nationally representative sample comprised 10 247 individuals aged 18 and older. A total of 5952 (58%) individuals participated in the health examination. Of these, 5125 filled in the food frequency questionnaire (FFQ). Sample used in this study included participants aged 18-74 years with information available on all the essential variables (n = 4671). The FinHealth 2017 Study was performed in accordance with the Declaration of Helsinki and approved by the ethics committee of Helsinki and Uusimaa hospital district. All participants gave their written informed consent.
SIM was asked on a questionnaire with a question "Which of the following describe your food choices? That my diet contains a large proportion of meat products" with five response options (not important at all/not very important/can't say/fairly important/very important). The variable was utilized as such and as further categorized into two categories (not important at all, not very important or can't say/fairly important or very important). Also an alternative variable including two categories (not important at all, not very important, can't say or fairly important/very important) was tested but due to small number of individuals finding meat very important, this variable was omitted from main analyses in order to enable further interaction analyses.
Dietary intake was determined with a validated (Kaartinen et al., 2012;Männistö, Virtanen, Mikkonen, & Pietinen, 1996;Paalanen et al., 2006) semi-quantitative FFQ containing 134 foods, food groups and drinks and measuring habitual food intake over the last 12 months (Borodulin & Sääksjärvi, 2019;Valsta, Kaartinen, Tapanainen, Männistö, & Sääksjärvi, 2018). Intake of foods, ingredients, nutrients and energy was calculated with an in-house dietary software (Finessi) based on the national food composition database (Fineli®). RPM consumption was determined as intake of red meat and processed meat (beef, pork, lamb, game, offals, sausages, sausage cuts and cold cuts) as grams (g) per day and additionally as g per megajoule (MJ) per day, which demonstrates relative amount of RPM consumption in diet regardless of energy intake. For domain analyses considering those within the highest RPM consumption quintile, RPM variables were divided into sex-specific quintiles.
Sex, age, area, residential area, marital status, household structure, education, main activity, household income, and sufficiency of money to meet needs were chosen to represent sociodemographic and -economic status. Information for sex, age, area (Uusimaa (Southern Finland)/ North Savo (Eastern Finland)/Northern Finland/Pirkanmaa (Central Finland)/Southwest Finland) and residential area (urban areas/areas near urban areas, rural centres/remote rural areas) were obtained from the Population Register Centre. Age was further categorized into three categories (18-34/35-54/55-74 years).
For the other variables, information was collected with questionnaires. Marital status was categorized as follows: married, cohabiting or in a registered partnership/single/separated, divorced or widow. Household structure was categorized into: household with only one adult living alone/household with at least one adult and at least one underage child/household with at least two adults and no underage children. Education was based on self-reported number of years of fulltime studying including primary school and categorized into three groups on the basis of one's relative position with regard to his/her birth cohort and sex: low (the lowest tertile)/intermediate (middle tertile)/ high (the highest tertile). Main activity was categorized into: employed (including entrepreneurs and those working for a family business without salary)/other (including students, those retired, unemployed, on family-leave and others).
The income quintile variable was based on questions on total household income during the last year before tax deductions, and on number of adult and underage household members. The original household income question comprised ten pre-defined categories from 'less than 15 000 €', and '15 001-25 000 €' to 'more than 90 000 €'. In this study, upper limits of the categories (and in the highest category, lower limit multiplied by two) were divided by weighted sum of household members, given a value of 1.0 to the first adult, value of 0.7 to additional adults, and value of 0.5 to the underage household members (OECD Project on Income Distribution and Poverty). The remainder was further categorized into sex-specific quintiles.
Question on sufficiency of money to meet needs was categorized as follows: not at all or a little/moderately or mostly/completely.

Statistical methods
The linear and the logistic regression models were used, and the strength of association was estimated with model-adjusted means and odds ratios (OR), respectively. In the linear model, the dependent variable was continuous RPM consumption either in g/day or g/MJ/day, and the independent variable was a sociodemographic or -economic factor in question (Tables 3 and 4). In the logistic model, the dependent variable was SIM in two categories, and the independent variable was a sociodemographic or -economic factor in question (Tables 1, 2, 5a and 5b). Two adjustment models were used. In adjustment model 1 age in 10-year categories was adjusted for. In adjustment model 2 additionally area, residential area and education were adjusted for. Analysis weights were used in order to mitigate non-participation bias (Härkänen et al., 2016). The effect modification of SIM in the association between RPM consumption and sociodemographic and -economic factors was studied by including in the model an interaction term between the SIM variable and a sociodemographic or -economic factor at issue.
Even though many of the sex-interactions appeared non-significant, we opted to present results separately for men and for women due to significant level difference in RPM consumption and in SIM between sexes. Moreover, traditionally meat has been associated with masculinity (Kildal & Syse, 2017;Rothgerber, 2013) and thus, it was hypothesized that the associations would differ between sexes.
All of the analyses were conducted with SAS Enterprise Guide, version 7.15 HF7 (SAS Institute Inc., Cary, NC, USA).

Subjective importance of meat in diet
Overall, men found meat in their diet more important than women did. Of those who considered meat very important, 75% were men and 25% were women, while of those who did not consider meat important at all, 25% were men and 75% were women (Table 1). Of men, 53% found meat fairly important or very important and 8% did not consider it important at all, while the corresponding percentages for women were 29% and 22%, respectively (Fig. 1).
Even though generally, meat in diet was more important to men than to women (Table 1), associations between SIM and sociodemographic and -economic factors were mostly parallel in both sexes. Among both sexes, those finding meat important in diet were younger and more commonly living in Northern Finland or in remote rural areas compared to Uusimaa or urban areas, respectively (Table 2). Moreover, individuals with the lowest educational level found meat more important than individuals with high educational level.
Some exceptions, however, emerged, and among men only, those living in Pirkanmaa and Southwest Finland or in areas near urban areas and rural centres appeared to find meat more important than individuals in respective reference groups, although in these analyses, sex-interaction remained non-significant (Table 2). Moreover, men living in households with at least one child found meat important over two times more commonly than men living alone, while in women, no such difference existed (P-value for sex-interaction = 0.02). In addition, in men only, those who were employed considered meat more important than men with other main activities.

Red meat and processed meat consumption
RPM consumption (g and g/MJ) grew in parallel with categories of SIM in men and in women (Table 3). Those finding meat very important

Table 1
Subjective importance of meat in diet in men and in women. consumed over twofold as much RPM as those not finding meat important at all. Overall, men consumed more RPM than women did, both considered as in absolute intake (g) and in relation to energy intake (g/MJ) in each category. In addition to male-sex, RPM consumption was higher in both sexes in age group of 35-54 year-olds, those living in rural areas and in those with low education (Table 4). Moreover, of the areas, RPM consumption was highest in men living in Southwest Finland and in women living in Northern Finland, although in men, the differences between areas were less notable than in women. In men, marital status did not associate with RPM consumption, but in women, those who were single consumed less RPM than those with other marital statuses. In women, also those living alone consumed less RPM than women living in different households. When adjusting for age only, household income associated inversely with RPM consumption in women. Further Table 2 Weighted odds of finding meat in diet fairly or very important in categories of selected sociodemographic and -economic factors in men and in women (n = 4671).

Sociodemographic and -economic factors n Finding meat fairly or very important in diet
Men adjustment, however, attenuated the association into non-significant. In both sexes, RPM consumption was lowest in individuals who evaluated they have sufficiently money to meet needs. Generally, results of absolute intake (g) and intake in relation to energy intake (g/MJ) were mostly aligned.

Interactions between subjective importance of meat in diet and RPM consumption
Generally, those who found meat important in their diet also consumed more RPM compared to those not finding meat important irrespective of the sociodemographic or -economic group (Table S1). One statistically significant first-degree interaction and one second-degree sex-interaction emerged between SIM and sociodemographic and -economic factors when comparing RPM consumption (g and g/MJ) in such interaction groups. Accordingly, in men, while in those finding meat important RPM consumption was highest in those living alone, in those not finding meat important RPM consumption was highest in those with households with children. Further, while in women, 35-54 yearolds consumed the most RPM (g) within those finding meat important and in those not finding it important, in men finding meat important, highest RPM consumption could be seen in 55-74 year-olds.

Domain analysis of the highest RPM consumption (g and g/MJ)
In a domain analysis considering only those within the highest RPM consumption quintile (g and g/MJ), it appeared that men found meat in diet two to three times as important as women did (Table 1). Among those in the highest RPM quintile (g and g/MJ), odds of finding meat important seemed to decrease in parallel with age in both sexes (Tables 5A and 5B). While men in the highest quintile (g) and living in households with adults only found meat less important than corresponding men living alone or in households with children, in women, those in the highest quintile (g) and living in households with children found meat less important than women living alone (P for sex-interaction = 0.02) (Table 5A). Within the highest quintile (g) non-employed men found meat less important compared to employed men (Table 5A). In women, those in the highest RPM quintile (g) and having medium household income found meat more important than corresponding women with the lowest household income. Moreover, men, in particular in the highest quintile (g) and having moderately money to meet needs found meat twice as important compared to corresponding men not having enough money to meet needs (Table 5A). When considering the highest RPM consumption quintile in grams/MJ, men living in remote rural areas, men belonging in the highest household income quintile and men evaluating they have completely enough money to meet needs found meat approximately twice as important as men in the corresponding reference groups (Table 5B).

Interpretation of the findings
In this study, men found meat more important than women did. Men also consumed more RPM overall and in each category of SIM variable in grams and in grams/MJ. These findings on sex differences in RMP consumption are in line with previous literature (Guenther et al., 2005;Heuer et al., 2015;Meier & Christen, 2012;Pfeiler & Egloff, 2018;Prättälä et al., 2007;Sych et al., 2019;Valsta et al., 2018;Wang, Beydoun, Caballero, Gary, & Lawrence, 2010;Wozniak et al., 2020). Various reasons may explain the differences between sexes. Traditionally, meat consumption has been linked to masculinity (Bogueva & Marinova, 2020;Kildal & Syse, 2017;Leroy & Praet, 2015;Marinova & Bogueva, 2019;Rothgerber, 2013;Rozin et al., 2012). Furthermore, a Finnish study suggested that men have more barriers to consume plant-based foods (Pohjolainen et al., 2015). It may be that men generally prefer taste of other foods over plant-based foods, but behind this may also be deep-seated beliefs of men needing meat to be men and traditions of men preparing the meat (e.g. barbecuing) while women prepare other foods (Leroy & Praet, 2015). Previous studies have also indicated that women are more health conscious (Ek, 2015;Kearney, Kearney, Dunne, & Gibney, 2000) and take more often environmental and animal welfare-related issues into account (Clonan et al., 2015). In this study, domain analyses on individuals consuming the most RPM (individuals within the highest consumption quintile) revealed that men with high RPM consumption also found meat more important than women with high RPM consumption did. This new finding suggests that for men with high RPM consumption decreasing RMP consumption may be more challenging than for women with high RPM consumption to whom high consumption may be rather based on habits.
Age was inversely associated with SIM both in men and in women. RPM consumption (g and g/MJ), however, was highest in the middle age group of 35-54 year-olds. In the oldest age group, physical difficulties in eating and chewing may decrease RPM consumption (Hildebrandt, Dominguez, Schork, & Loesche, 1997 Table 3 Weighted red meat and processed meat consumption (g and g/MJ) in categories of subjective importance of meat in diet in men and in women (n = 4511).  Weighted red meat and processed meat consumption (g and g/MJ) in categories of selected sociodemographic and -economic factors in men and in women (n = 4511).  Egloff, 2018;Prättälä et al., 2007;Valsta et al., 2018), greater barriers to consume plant-based foods (Pohjolainen et al., 2015), but also with higher frequency of vegetarianism (Wozniak et al., 2020). Thus, it appears that younger adults have stronger, yet more segmented attitudes towards meat consumption, while for older individuals meat consumption may be less based on principles. Indeed, in the analyses considering only individuals in the highest RPM consumption quintile, it appeared that in older age groups SIM was less prevailing than in the youngest age group in both sexes. It seems that in older age-groups, high RPM consumption could be rather a habit than something based on conscious values. In the youngest age group, however, conscious values appear to determine the consumption habits, and thus, it may be more challenging to impact habits of this group. Men and women living in the area of Northern Finland (northern parts of Finland, including Lapland) found meat more important than men and women living in Uusimaa (southern part of Finland, including the Helsinki Metropolitan area) or in North Savo (mainly eastern parts of Finland). In men, meat was also found more important in those living in Pirkanmaa (covering roughly the middle areas of the country) and in Southwest Finland. In women, RPM consumption ran in parallel with SIM and was the highest in Northern Finland and the lowest in Uusimaa and North Savo suggesting that higher actual RPM consumption in certain regions is not based on habits and traditions only but also higher SIM in these regions. In men, however, a statistically significant difference in RPM consumption occurred only in RPM consumption proportional to energy intake (g/MJ), and was the highest in Southwest Finland and the lowest in North Savo. It may be that as men generally consume more RPM, regional differences do not show that distinct differences. Study on residential areas showed that men and women living in more rural areas both found meat more important and consumed more RPM than women and men living in more urban areas. Previous studies have shown mostly aligned regional differences in meat consumption in Finland between regions (Montonen et al., 2008) and in areas according to urbanization level (residential areas) (Vainio, Niva, Jallinoja, & Latvala, 2016). While in many low-and middle-income countries meat consumption is directly associated with urbanization level (Clonan et al., 2016), in high-income countries the association has opposite direction. An American study indicated that in rural areas red meat consumption is higher than in metropolitan areas (Guenther et al., 2005). Previous studies have suggested that individuals living in urban areas are more willing to change their diet towards more ecological (Pohjolainen, Tapio, Vinnari, Jokinen, & Rasanen, 2016). Moreover, Pohjolainen et al. (2015) indicated that rural residence is associated with greater barriers to consume plant-based foods. These results, combined with findings of this study, indicate that rural residence is one of the most meaningful determinants of challenges in transition towards more plant-based diet. Regional customs and food culture differences as such, but also uneven distribution of education across both regions and residential areas may contribute to such regional differences in SIM and RPM consumption, although education was adjusted for in the analyses.
In this study, marital status showed no associations with SIM. In women, however, differences in RPM consumption emerged, and the consumption was the highest in women who were married or cohabiting and the lowest in women who were single, which is in line with results from the Finnish Health 2000 Survey (Montonen et al., 2008), and is presumably at least partly, an implication of women preparing and eating meals that are appealing to their spouses. Associations between household structure subgroups and SIM and RPM consumption appeared divergent in men and in women. Noteworthy was that while men living in households with children found meat more important but did not consume more meat than men living alone, women living in households with children did not find meat in their diet any more important, but still consumed more meat than women living alone. In domain analyses on individuals within the highest RPM consumption quintile, SIM was the most prominent in men living in households with children but the least prominent in the corresponding women. These findings suggest that high RPM consumption is important to men but not to women living in households with children and for such women, high RPM consumption may be more based on habits. Interaction analysis between SIM and RPM consumption revealed that while in men finding meat important RPM consumption was highest in those living alone, in men not finding meat important RPM consumption was highest in those living in households with at least one child. It may be that in households with children, preparing dishes containing meat is more customary than in other households and leads to higher consumption regardless of not personally finding meat important. An Australian study found that family traditions are an important reason for meat consumption (Marinova & Bogueva, 2019). Guenther and colleagues (2005) found that 'higher than average' processed pork product consumption was common in households with small children. Sych et al. (2019), however, failed to find any differences between household types. Pohjolainen et al. (2015) concluded that eating meat is more often perceived as a norm in socio-culturally traditional families of 'parents with children'. Explanatory analyses of this study suggested that individuals in households with children also ate more often lunch and dinner but less often snacks than individuals in other households (data not shown). Thus, such more frequent eating of principal meals could be one reason accounted for higher consumption of RPM. In all, if especially women living in households with children consume a lot of meat as a consequence of preparing it to their children and spouses, but not due to finding meat important themselves, this group could be a significant target to impact to and strive to move their dietary choices towards more plant-based.
Educational level was inversely associated with both SIM and RPM consumption in men and in women. Also previous studies have indicated an inverse association between educational level (Guenther et al., 2005;Linseisen et al., 2002;Maguire & Monsivais, 2015;Pfeiler & Egloff, 2018;Sych et al., 2019) or SES (Heuer et al., 2015;Hulshof et al., 2003) and meat consumption. Lower education has also been linked to greater barriers to consume a plant-based diet (Pohjolainen et al., 2015). Pampel and colleagues (2010) concluded that poorer health behaviors among individuals with lower socioeconomic status (e.g. lower education) derive from several reasons, for instance, using unhealthy habits as "self-medication" in stressful everyday life situations, not perceiving one would benefit from adopting healthy lifestyle habits, having limited knowledge on healthy lifestyle, lack of self-efficacy, and conscious "class distinction". Thus, higher perceived SIM and higher RPM consumption among individuals with lower education may derive from these factors and poorer knowledge on health and environmental implications of meat. Indeed, higher education has been shown to associate with more favorable health behavior and greater efforts to follow a healthy diet (Kearney et al., 2000;Li & Powdthavee, 2015).
Whilst employed men found meat more important than men in other main activity categories, in women, no differences existed between such groups. The same associations held true in domain analyses including only individuals within the highest RPM consumption quintile. The group including other main activities besides employed individuals includes heterogeneous main activity situations (e.g. students, unemployed, retired) and, thus, reasons for not finding meat that important may derive from various motives. It is possible, however, that employed men have more traditional values (i.e. being supporter of the family), including also finding meat important part of diet, compared to students, unemployed or retired men. In women, employment possibly does not relate that strongly to such conventional values.
Household income did not associate with SIM but was inversely associated with RPM consumption in women. Moreover, both men and women evaluating they have enough money to meet needs consumed less RPM than men and women not having enough money to meet needs. It may be that in women, high income and both in men and in women sufficiency of money to meet needs correlates with higher health and environmental consciousness, and thus, with lower RPM consumption. Moreover, it is possible that such individuals better afford to select healthier and environmentally friendlier protein sources, such as fish or plant protein products. An American study indicated an inverse association between income and processed pork product consumption in data including men and women (Guenther et al., 2005). Similarly, in an UK study RPM consumption fell in parallel with growing income categories (Maguire & Monsivais, 2015). In a German study (Pfeiler & Egloff, 2018) and in a Swiss study (Sych et al., 2019), however, household income was unrelated to meat consumption. In domain analyses of this study including only individuals within the highest RPM consumption (g/MJ) quintile, men having the highest household income found meat more important than corresponding men with the lowest household income, while the association in women remained absent. Correspondingly, men in the highest RPM consumption (g/MJ) quintile and evaluating they

Table 5A
Weighted odds of finding meat in diet fairly or very important between categories of selected sociodemographic and -economic factors in men and in women within the highest meat consumption (g) quintile (n = 922).

Sociodemographic and -economic factors n
Finding meat in diet fairly or very important among those belonging to the highest meat consumption (g) quintile Men have completely enough money to meet needs found meat more important than corresponding men without enough money to meet needs. Direct association between wealth and SIM in men with high RPM consumption may derive from traditional masculine values associated with meat (Kildal & Syse, 2017).

Methodological considerations
Strengths of this study include a large nationally representative sample, population-based health examination study data with standardized measurement methods and a wide range of information available on the subjects (Borodulin & Sääksjärvi, 2019), and RPM

Table 5B
Weighted odds of finding meat in diet fairly or very important between categories of selected sociodemographic and -economic factors in men and in women within the highest meat consumption (g/MJ) quintile (n = 922).

Sociodemographic and -economic factors n
Finding meat in diet fairly or very important among those belonging to the highest meat consumption (g/MJ) quintile Men consumption measured with a validated and updated FFQ (Kaartinen et al., 2012;Männistö et al., 1996;Paalanen et al., 2006). Some methodological considerations, however, should be taken into account. First, even though participation rate of the FinHealth 2017 Study was relatively good (58% participated in the health examination), non-participationa common issue causing bias in population surveysmay have affected the results. This, however, was mitigated by using analysis weights in the analyses (Härkänen et al., 2016). Second, as the data was collected in 2017, the results represent the situation prior to Covid-19 pandemic. The pandemic may have affected consumption habits and sociodemographic and -economic position and, thus, it would be of importance to conduct a corresponding study utilizing more recent study data. Third, even though FFQ method may produce misestimations of food and energy intake (Kipnis et al., 2002), the aim of this study was not to estimate absolute consumption levels but to compare consumption between groups, for which FFQ is a suitable tool (Willett, 2013). Fourth, possibility of dietary misreporting exists in all surveys using self-reported information on dietary intake. Livingstone and Black (2003) concluded in their review that low-energy reporters reported more "good foods" such as meat. The original studies the authors made their summarization from were published on the 80'-90's, and concept of meat as being a "good food" seems to have changed since then. Accordingly, some studies based on a little more recent data have demonstrated underreporting of meat consumption due to being "undesirable" (Scagliusi, Polacow, Artioli, Benatti, & Lancha, 2003). Fifth, in this study and in line with several previous studies (Clonan et al., 2016;Maguire & Monsivais, 2015) red meat and processed meat were combined in the RPM variable in spite of possible different health effects, and different distributions of consumption in sociodemographic and -economic groups (Clonan et al., 2016). This was opted due to rather similar environmental effects of red and processed meats and in order to avoid expansion of the article too much. Sixth, SIM was inquired only with one question, which has not been validated against any previous measurement scales (i.e. the Meat-Attachment Questionnaire (Graca et al., 2015)), and thus, could produce biased results. As the variable, however, produced mostly parallel results compared to RPM consumption, it can be presumed that it is a relatively valid tool to measure the issue. Seventh, in a study including numerous analyses, the possibility of false positive findings exists. Eighth, potential confounding variables for two adjustment models were chosen based on their associations with exposure and outcome variables, based on existing literature on the associations, and based on variables available. Despite the adjustments, the possibility of residual confounding cannot be fully ruled out. Finally, in this study, we used two different RPM consumption variables: one considering absolute RPM intake in grams, and another considering RPM consumption in grams relative to energy intake in megajoules. The RPM consumption in grams tells about actual amount of RPM consumption, which can be higher in individuals with generally higher energy intake, while the RPM consumption in grams/MJ represents relative amount of RPM consumption in diet regardless of energy intake and can be high in spite of low energy intake and relatively low absolute RPM consumption if greater proportion of diet consists of RPM. Even though statistical significance of the findings differed in some of the results according to RPM consumption variable used, the results of the two measures were mostly parallel in direction.

Conclusions
As RPM consumption is associated with poorer health outcomes (Ekmekcioglu et al., 2018;Yip et al., 2018) and greater environmental impact (Hallström et al., 2015;Hyland et al., 2017;McMichael, Powles, Butler, & Uauy, 2007), reduction of the consumption has been suggested (European Public Health Association., 2017;Hemler & Hu, 2019;Willett et al., 2019). In order to be able to understand challenges of such dietary transition and to affect people's meat consumption habits, subgroups that consume more RPM but also subgroups that find meat more important in their diet should be identified. No previous studies, however, have been conducted on sociodemographic or -economic differences of SIM.
This study showed that both RPM consumption and SIM distribute unevenly across sociodemographic and -economic groups. In this study, RPM consumption distributions across subgroups were mostly in line with those of previous studies, and RPM consumption appeared higher in those finding meat important, in men, in 35-54 year-olds, in those living in rural areas and in those with low education than in corresponding reference groups. Conversely RPM consumption was lower in those who experienced they have enough money to meet needs. In addition, women who were single or living alone consumed less RPM than other marital or household groups.
Novelty of this study, which should be verified in other populations, was the demonstration of subgroups that find meat important in their diet. Such groups were men, young adults, those living in rural areas or in Northern Finland and those with low education. Moreover, men who lived in households with children or were employed, found meat more important than men in corresponding reference groups. Special attention should be put on these subgroups when outlining strategies to reduce meat consumption.
Another novelty of this study was the simultaneous examination of SIM and actual RPM consumption. The results indicated that among those who belonged to the highest RPM consumption quintile, meat was found less important in older individuals. Additionally, in men, it was found less important in those living in urban areas, those living in households with adults only, those who were not employed and those with the lowest household income or not enough money to meet needs. The preceding suggests that in these groups high RPM consumption could be more of a habit than something based on strong opinions, and could be impacted easier while in complementary groups, more effort may be needed to make a change in RPM consumption habits. Individualized measures should be tailored to impact meat consumption habits of different population groups, whose reasons for high consumption may derive from differing premises. Information provided by this study could be used when striving to change food consumption habits towards climate friendlier at population level.

Author contributions
LS-J, LV, PH and TM designed the research. LS-J conducted the research, analyzed the data and wrote the first draft of the manuscript. LS-J, LV, PH and TM participated in the interpretation of the results and commented the manuscript. All authors read the final manuscript and approved its submission.

Funding
This work (part of the JUST FOOD project) was supported by the Strategic Research Council (SRC) of the Academy of Finland [grant number: 327 370]. The funding source had no role in the design, analysis or writing of the article.

Ethics statement
The FinHealth 2017 Study was performed in accordance with the Declaration of Helsinki and approved by the ethics committee of Helsinki and Uusimaa hospital district. All participants gave their written informed consent.

Data availability
No data are publicly available due to being sensitive individual-level data.

Declaration of competing interest
None.