The Relative Importance of Family, School, and Leisure Activities for the Mental Wellbeing of Adolescents: The Young-HUNT Study in Norway

Round 1

Reviewer 1 Report

This original research article aims to explore the mental wellbeing of Norwegian adolescents in relation to the social contexts of family, school, and leisure activities, testing the relative importance of the three social contexts based on the strength of their associations with adolescent mental wellbeing. Differences between adolescents with high and low SEP are explored. The discussion focuses on the practical implications of developing public health interventions targeted at families and school.

The article is valuable in its focus on the importance of the social contexts/social determinants of adolescent health, but it may benefit from including more recent references on the topic (including of WHO, HBSC). Other weaknesses that need to be addressed include:

-        A description of the local sociocultural context in Norway is completely absent both in the introduction and discussion. Results should be discussed also in the context of Norwegian adolescents’ way of life.

-         The description of the Young-HUNT Study in the Methods section is not adequate as it gives too little information (not even mentioning “in Norway”) and refers to a previously published paper. However, the Limitations section focuses on aspects of the study such as its design (cross-sectional) and its population (county in mid-Norway, consisting of mainly small cities and rural areas). It is not specified in the text when data collection took place, but there are years (2017-19) in the caption of table 1.

-         Using parental education as a marker of SEP is discussed as a limitation, but there is no rationale provided why it has been chosen. According to the Norwegian Institute of Public Health, there are substantial social inequalities in health in Norway, especially between educational groups – maybe that is a reason to use educational level as a marker of SEP in a Norwegian health study?

-         A major weakness to be addressed is the absence of description and discussion of the results regarding the outcome variables – psychological distress, loneliness, and life dissatisfaction, which should be a main focus of the paper (and the title of the article starts with “mental wellbeing”). The means provided in Table 1 suggest that the levels of mental wellbeing variables are low in the sample (how does this relate to other studies conducted in Norway?) and it is unclear if the differences between low SEP and high SEP, presented in table 1, are statistically significant. The associations between the social contexts and the three outcome variables are presented and discussed only in terms of comparing the strength of the associations as a marker of the relative importance of the social contexts. The associations are not described and discussed per se. Family and school contexts reveal associations as expected and leisure activities are unrelated (how does this correspond to other studies)?

-         At present the discussion focuses disproportionally on implications for practice – development of interventions – at the expense of discussing the study results.

-         The leisure activities studied are very heterogeneous – some are performed alone, some with others, but unorganized, and some with others and organized. Not differentiating between types of activities hampers interpretation and is a limitation. How are leisure activities, performed with others, and especially organized ones, “private arenas within an adolescent’s life”? (Lines 236-237).

-         “These results may suggest that socioeconomic inequalities in certain populations may be more evident within private social environments compared to shared social systems such as the school environment”. (Lines 238-240). This may be related to the specific sociocultural context and economic inequality in Norway. Are there private or elite schools in Norway?

Author Response

Reviewer 1:

This original research article aims to explore the mental wellbeing of Norwegian adolescents in relation to the social contexts of family, school, and leisure activities, testing the relative importance of the three social contexts based on the strength of their associations with adolescent mental wellbeing. Differences between adolescents with high and low SEP are explored. The discussion focuses on the practical implications of developing public health interventions targeted at families and school.

R1,  C1

The article is valuable in its focus on the importance of the social contexts/social determinants of adolescent health, but it may benefit from including more recent references on the topic (including of WHO, HBSC). Other weaknesses that need to be addressed include:

Thank you for your comments!

We have now updated some of the literature. In particular, we have included some relevant recent international reports (World Health Organization., 2020) and national reports (Bakken, 2020, 2022; Institute of Health Equity., 2023).

R1, C2

-        A description of the local sociocultural context in Norway is completely absent both in the introduction and discussion. Results should be discussed also in the context of Norwegian adolescents’ way of life.

We have now included a description of the local sociocultural context in Norway in the description (lines 91 to 106) and have tried to frame the discussion in light of the local sociocultural context when appropriate.

R1, C3

-         The description of the Young-HUNT Study in the Methods section is not adequate as it gives too little information (not even mentioning “in Norway”) and refers to a previously published paper. However, the Limitations section focuses on aspects of the study such as its design (cross-sectional) and its population (county in mid-Norway, consisting of mainly small cities and rural areas). It is not specified in the text when data collection took place, but there are years (2017-19) in the caption of table 1.

We have now revised the method section to include more information about the Young-HUNT4 survey, see lines 129 to 135.

R1, C4

-         Using parental education as a marker of SEP is discussed as a limitation, but there is no rationale provided why it has been chosen. According to the Norwegian Institute of Public Health, there are substantial social inequalities in health in Norway, especially between educational groups – maybe that is a reason to use educational level as a marker of SEP in a Norwegian health study?

We have now included a reference showing that parental education seems to be a relevant marker for SEP when investigating socioeconomic inequalities in children and adolescents in general (lines 74 to 76) and refer to a recent report on social inequalities in Norway in which education was used as a marker for SEP (lines 202 to 203) .

R1, C5

-         A major weakness to be addressed is the absence of description and discussion of the results regarding the outcome variables – psychological distress, loneliness, and life dissatisfaction, which should be a main focus of the paper (and the title of the article starts with “mental wellbeing”). The means provided in Table 1 suggest that the levels of mental wellbeing variables are low in the sample (how does this relate to other studies conducted in Norway?) and it is unclear if the differences between low SEP and high SEP, presented in table 1, are statistically significant.

The main aim of the article was to investigate the relative importance of the social contexts for the mental wellbeing of adolescents, but we agree that this could be addressed to create a more comprehensive article. Thus, we have included an investigation on the association between parental education and the outcomes (lines 268 to 275). Furthermore, we highlight in the discussion that the levels of psychological distress, loneliness, and life dissatisfaction seems to be comparable to data from the recent national UngData-survey, though the questions used and how the questions were operationalized, which may limit comparability (lines 325 to 328). In general, it was difficult to make a clear comparison to other samples in Norway as how mental health and wellbeing is being measured varies. Lastly, we have revised the title of the article to emphasize that the relative importance of the included social contexts is the main interest.

R1, C6

The associations between the social contexts and the three outcome variables are presented and discussed only in terms of comparing the strength of the associations as a marker of the relative importance of the social contexts. The associations are not described and discussed per se. Family and school contexts reveal associations as expected and leisure activities are unrelated (how does this correspond to other studies)?

We agree, we now include a more thorough discussion of the separate associations with the outcomes, see lines 358 to 3368. However, as we have revised which leisure activities we include (see answer R1, C8) in the analyses, the results were slightly altered.

R1, C7

-         At present the discussion focuses disproportionally on implications for practice – development of interventions – at the expense of discussing the study results.

We agree and we have now shortened the discussion of practical implications  and have expanded the discussion of study results.

R1, C8

-         The leisure activities studied are very heterogeneous – some are performed alone, some with others, but unorganized, and some with others and organized. Not differentiating between types of activities hampers interpretation and is a limitation. How are leisure activities, performed with others, and especially organized ones, “private arenas within an adolescent’s life”? (Lines 236-237).

            In the present article, we wanted to have a broad perspective on leisure activities, both organized and unorganized, as adolescents may participate in different leisure activities during the week. However, we agree that including activities the adolescents partake in alone may hamper interpretation as the focus is on social arenas. Thus, we have removed these activities from the analyses, and include activities that are social in nature by either explicit stating that the activity is conducted with others or within a common social arena (lines 185 to 193). Further, we have also included literature in the perspective of broad leisure activity participation, which shows that participating in various leisure activities is related to mental wellbeing (lines 62 to 67). Regarding the sentence “private arenas within an adolescent’s life”, we have revised the sentence, see lines 341 to 343.

R1, C9

-         “These results may suggest that socioeconomic inequalities in certain populations may be more evident within private social environments compared to shared social systems such as the school environment”. (Lines 238-240). This may be related to the specific sociocultural context and economic inequality in Norway. Are there private or elite schools in Norway?

We agree that the results may be specific to the sociocultural context of the study sample. Thus, we have included a short discussion of how these results may differ depending on the sample. Furthermore, we highlight that other studies in Norway have found socioeconomic inequalities within the school system of Norway, and thus these results may contrast with other student populations within Norway as well, see lines 345 to 350.

References

Bakken, A. (2020). Ungdata. Ung i Distrikts-Norge, NOVA Rapport 3/20.

Bakken, A. (2022). Ungdata 2022. Nasjonale Resultater, NOVA Rapport 5/22 (NOVA Rapport 5/22).

Institute of Health Equity. (2023). Rapid review of inequalitites in health and wellbeing in Norway since 2014. https://www.instituteofhealthequity.org/resources-reports/rapid-review-of-inequalities-in-health-and-wellbeing-in-norway-since-2014/read-the-full-report.pdf

World Health Organization. (2020). Spotlight on adolescent health and well-being.  Findings from the 2017/2018 Health Behaviour in School-aged Children (HBSC) survey in Europe and Canada. International report. Volume 2. https://www.who.int/europe/initiatives/health-behaviour-in-school-aged-children-(hbsc)-study

Reviewer 2 Report

Peer review of “Mental wellbeing of adolescents and the relative importance of family, school, and leisure activities: The Young-HUNT Study in Norway”

 

I read the manuscript entitled “Mental wellbeing of adolescents and the relative importance of family, school, and leisure activities: The Young-HUNT Study in Norway” with great interest.

The overarching aim of this work was to explore how family cohesion, school climate and leisure activities were associated with psychological distress, loneliness and life dissatisfaction in adolescents aged 13 to 19 years old. The study had three objectives:

1.       To describe family cohesion, school climate and leisure activities according to parental education level

2.       To explore the association of family cohesion, school climate and leisure activities with psychological distress, loneliness and life dissatisfaction

3.       To see if a moderative role of parental education level on this association existed

Overall comments

The manuscript is well written and easy to understand. Using terms closer to what was actually explored could give a more precise glimpse of this study. As highlighted in the discussion section, parental education level is not sufficient to assess socioeconomic position (SEP). Unless a more comprehensive construct of SEP can be used, I suggest using “parental education” (or any other related term) instead of SEP.

Introduction

The first cited paper to show youth mental health is on the decline is from the HUNT study. Systematic review, papers on international datasets or from representative samples would be more appropriate. Otherwise the introduction is really nice.

Materials and methods

Study population

Although the HUNT study has been described elsewhere, basic information (i.e., cross-sectional design, dates, data collection mode) regarding the HUNT4 would give a clearer sight of the study context. These information are written elsewhere in the manuscript but having them all here really facilitate comprehension of the work.

Outcomes

Has the SCL-10 been validated in Norwegian adolescents or in adolescents? Is construction of an average score the usual way of using this scale?

Loneliness and life dissatisfaction were assessed using single items from two scales. This is often the case in surveys assessing various life dimension where full scales need to be shorten so that the questionnaire is not too long. If known, the rationale for choosing these two items should be added in the method section. Moreover, the limitation of using only single items to assess complex life dimensions need to be discussed in the corresponding part of the discussion.

Social context

How were chosen the ten question regarding leisure activities? Were they based on what was found in national statistics about what Norwegian adolescents usually do? If so, adding it in the manuscript would strengthen the methods. In the discussion, it is said that the survey was conducted in a more rural/small city. How would that affect the possibility of leisure activities for adolescents? Especially cultural ones? For example, if the adolescents need a drive to go to the theatre maybe this question is not the most relevant for assessing leisure activities in this area? Maybe a small discussion regarding access to leisure activities should be added in the discussion section.

Family SEP

As suggested above, the term socioeconomic position should be replaced by parental education level (or similar terms) to better reflect what was actually done.

Covariates

Regarding parental cohabitation, the sentence might need to be revised. How could the possible answers to “do you parents cohabit?” only be “No”; “Yes, they lived separately or were separated, but they later moved back together again” and “Yes, they were divorced or separated”?

Was parental mental health assessed?

Statistical analyses

With respect to the study aims, maybe the statistical plan could be revised. Below is a proposition:

1.       Descriptive statistics in total sample of each covariates, including parental education level, social context and mental health (can be included as supplementary material)

2.       Exploring association of parental education level with family cohesion, school climate, and leisure activities

a.       Bivariate linear regression between parental education level and family cohesion, school climate and leisure activities

b.       Adjusted linear regression between parental education level and family cohesion, school climate and leisure activities (adjusted for participants age, sex, perceived family financial situation and parental cohabitation)

3.       Exploring association of family cohesion, school climate, and leisure activities with psychological distress, loneliness, life dissatisfaction + moderative role of parental education level

a.       Bivariate linear regression between family cohesion, school climate and leisure activities and psychological distress, loneliness, life dissatisfaction

b.       Interaction test of parental education level on these associations

c.       Adjusted linear regression, stratified according to parental education level if the interaction tests are significant

d.       Maybe considering adding the three social context exposures in the same model to see how the coefficients vary?

                                                               i.      Linear regression adjusted for sex, age, parental cohabitation, and subjective family economy

                                                             ii.      Adding family cohesion to the previous model

                                                           iii.      Adding school climate to the previous model

                                                           iv.      Adding leisure activities to the previous model

Could the authors explain more thoroughly what they did to correct for potential clustering of siblings within families?

Results

Table 1: as for participants’ sex the value in each cell are not mean and standard deviation, maybe the table’s form can be revised. Also, please add the descriptive statistics of all covariates in the table.

Table 2: presenting exposures as column and outcomes as line is rather unusual. If readers go straight to the tables to have a quick glimpse at the results the table’s current form might be confusing. But otherwise it is really clear.

Discussion

The limitation section could be strengthen by discussing the use of single items to assess loneliness and life dissatisfaction, as well as characteristics not accounted for (parental mental health, adolescents health characteristics, bullying,…) Explaining more thoroughly how each limitation can impact the results (overestimation, underestimation, type of bias…) could also benefit this section.

Author Response

Reviewer 2:

Peer review of “Mental wellbeing of adolescents and the relative importance of family, school, and leisure activities: The Young-HUNT Study in Norway”

I read the manuscript entitled “Mental wellbeing of adolescents and the relative importance of family, school, and leisure activities: The Young-HUNT Study in Norway” with great interest.

The overarching aim of this work was to explore how family cohesion, school climate and leisure activities were associated with psychological distress, loneliness and life dissatisfaction in adolescents aged 13 to 19 years old. The study had three objectives:

1. To describe family cohesion, school climate and leisure activities according to parental education level
2. To explore the association of family cohesion, school climate and leisure activities with psychological distress, loneliness and life dissatisfaction
3. To see if a moderative role of parental education level on this association existed

Thank you for your comments!

R2, C1

Overall comments

The manuscript is well written and easy to understand. Using terms closer to what was actually explored could give a more precise glimpse of this study. As highlighted in the discussion section, parental education level is not sufficient to assess socioeconomic position (SEP). Unless a more comprehensive construct of SEP can be used, I suggest using “parental education” (or any other related term) instead of SEP.

Thank you for your comments! We have revised the MS and use “parental education” rather than SEP.

R2, C2

Introduction

The first cited paper to show youth mental health is on the decline is from the HUNT study. Systematic review, papers on international datasets or from representative samples would be more appropriate. Otherwise the introduction is really nice.

We have replaced the reference using HUNT data and now include a systematic review.

R2, C3

Materials and methods

Study population

Although the HUNT study has been described elsewhere, basic information (i.e., cross-sectional design, dates, data collection mode) regarding the HUNT4 would give a clearer sight of the study context. These information are written elsewhere in the manuscript but having them all here really facilitate comprehension of the work.

We have now revised the method section to include more information about the Young-HUNT4 survey, see lines 129 to 135.

R2, C4

Outcomes

Has the SCL-10 been validated in Norwegian adolescents or in adolescents? Is construction of an average score the usual way of using this scale?

The SCL-10 has been validated in Norwegian adolescents, see (Strand et al., 2003). Regarding a construction of an average score, this is one of two usual ways of managing this questionnaire, and have been conducted in a Norwegian adolescent sample (Larsen et al., 2021). The second way is to dichotomize the variable using cut-off point for substantial symptom burden, which was estimated to be 1.85 in Norwegian adolescents (Strand et al., 2003). A general recommendation within epidemiology is to refrain from categorizing variables (Dawson & Weiss, 2012).   

R2, C5

Loneliness and life dissatisfaction were assessed using single items from two scales. This is often the case in surveys assessing various life dimension where full scales need to be shorten so that the questionnaire is not too long. If known, the rationale for choosing these two items should be added in the method section. Moreover, the limitation of using only single items to assess complex life dimensions need to be discussed in the corresponding part of the discussion.

Yes, we agree that using single items hampers the investigation. we do not know the rationale of choosing these two items in the Young-HUNT questionnaire, but we have included a more thorough discussion of the limitations of using single items when investigating complex life dimensions in the study limitation (lines 502 to 515).

R2, C6

Social context

How were chosen the ten question regarding leisure activities? Were they based on what was found in national statistics about what Norwegian adolescents usually do? If so, adding it in the manuscript would strengthen the methods. In the discussion, it is said that the survey was conducted in a more rural/small city. How would that affect the possibility of leisure activities for adolescents? Especially cultural ones? For example, if the adolescents need a drive to go to the theatre maybe this question is not the most relevant for assessing leisure activities in this area? Maybe a small discussion regarding access to leisure activities should be added in the discussion section.

The leisure activities included were not chosen based on national statistics, which of course would have been more relevant. We wanted, however, to measure a broad spectrum of leisure activities, taking in to account that adolescents may have different types of leisure activities they wish to participate in. Thus, we included available questions about leisure activities from the Young-HUNT4 survey. We have revised the MS to include literature on the measure of breadth of leisure activities and mental health and wellbeing, which shows how participating in several different leisure activities is related to mental health and wellbeing. Furthermore, we have revised which leisure activities were included in the study as the aim is to compare social contexts it would be more appropriate to focus on leisure activities that are social by nature. 

Regarding accessibility of leisure activities, we agree that adolescents living in more rural areas may have limited access to some of the included activities. Thus, we have revised the MS to include this point in the discussion, see lines 442 to 452.

R2, C7

Family SEP

As suggested above, the term socioeconomic position should be replaced by parental education level (or similar terms) to better reflect what was actually done.

We have now replaced the term socioeconomic position with parental education level when referring to the measure of the present study. We do, however, state that we wish to investigate potential differences in socioeconomic position using parental education level but highlight in the limitation that other measures of socioeconomic position are necessary to get a full understanding of potential socioeconomic differences.

R2, C8

Covariates

Regarding parental cohabitation, the sentence might need to be revised. How could the possible answers to “do you parents cohabit?” only be “No”; “Yes, they lived separately or were separated, but they later moved back together again” and “Yes, they were divorced or separated”?

Thank you for highlighting this sentence. We have now revised the sentence and it should now be correct, lines 214 to 215.

R2, C9

Was parental mental health assessed?

Unfortunately, we did not have information regarding parental mental health in this project.

R2, C10

Statistical analyses

With respect to the study aims, maybe the statistical plan could be revised. Below is a proposition:

1. Descriptive statistics in total sample of each covariates, including parental education level, social context and mental health (can be included as supplementary material)
2. Exploring association of parental education level with family cohesion, school climate, and leisure activities
3. Bivariate linear regression between parental education level and family cohesion, school climate and leisure activities
4. Adjusted linear regression between parental education level and family cohesion, school climate and leisure activities (adjusted for participants age, sex, perceived family financial situation and parental cohabitation)
5. Exploring association of family cohesion, school climate, and leisure activities with psychological distress, loneliness, life dissatisfaction + moderative role of parental education level
6. Bivariate linear regression between family cohesion, school climate and leisure activities and psychological distress, loneliness, life dissatisfaction
7. Interaction test of parental education level on these associations
8. Adjusted linear regression, stratified according to parental education level if the interaction tests are significant
9. Maybe considering adding the three social context exposures in the same model to see how the coefficients vary?
10. Linear regression adjusted for sex, age, parental cohabitation, and subjective family economy
11. Adding family cohesion to the previous model

iii.      Adding school climate to the previous model

1. Adding leisure activities to the previous model

Thank you for a clear alternative plan for statistical analysis.  We agree that using regression analysis when investigating the association between parental education and family cohesion, school climate and leisure activity participation is more appropriate. Furthermore, we have included descriptive statistics of all covariates in table 1. Lastly, we do agree that including all social contexts within the same model is worth investigating and report these results in lines 297-299. We have created a table included with three models showing how the coefficients vary when including all social contexts. However, we propose that this could be included as a supplementary table, as the conclusion of the study does not change. The proposed supplementary table is included in the end of the MS. 

R2, C11

Could the authors explain more thoroughly what they did to correct for potential clustering of siblings within families?

Yes, we have included a more thorough explanation for the correction of the standard errors, see lines 241 to 244.

R2, C12

Results

Table 1: as for participants’ sex the value in each cell are not mean and standard deviation, maybe the table’s form can be revised. Also, please add the descriptive statistics of all covariates in the table.

We have revised the table, so it states that either No. or mean are provided, and SD or % are provided. Further, descriptive statistics for all covariates are now provided.

R2, C13

Table 2: presenting exposures as column and outcomes as line is rather unusual. If readers go straight to the tables to have a quick glimpse at the results the table’s current form might be confusing. But otherwise it is really clear.

We agree that it is somewhat unusual to present exposures as columns and outcomes as lines, but we felt that showcased the results clearly regarding the relative importance of the predictors family, school, and leisure activities. However, we have included an alternative table, presented with outcomes as columns and predictors as lines. We leave the decision to the editor; both will suffice from our point of view.

R2, C4

Discussion

The limitation section could be strengthen by discussing the use of single items to assess loneliness and life dissatisfaction, as well as characteristics not accounted for (parental mental health, adolescents health characteristics, bullying,…) Explaining more thoroughly how each limitation can impact the results (overestimation, underestimation, type of bias…) could also benefit this section.

We have included a discussion of these points in the study limitations, see 494 to 508. However, we have not stated in which way this may overestimate or underestimate the results as this is difficult to predict (Van Smeden et al., 2020), but we agree that the points mentioned may create sources of bias in the study.

References 

Dawson, N. V., & Weiss, R. (2012). Dichotomizing continuous variables in statistical analysis: A practice to avoid. Medical Decision Making, 2(32), 225–226. https://doi.org/10.1177/0272989X12437605

Larsen, T. B., Urke, H., Tobro, M., Årdal, E., Waldahl, R. H., Djupedal, I., & Holsen, I. (2021). Promoting Mental Health and Preventing Loneliness in Upper Secondary School in Norway: Effects of a Randomized Controlled Trial. Scandinavian Journal of Educational Research, 65(2), 181–194. https://doi.org/10.1080/00313831.2019.1659405

Strand, B. H., Dalgard, O. S., Tambs, K., & Rognerud, M. (2003). Measuring the mental health status of the Norwegian population: A comparison of the instruments SCL-25, SCL-10, SCL-5 and MHI-5 (SF-36). Nordic Journal of Psychiatry, 57(2), 113–118.

Van Smeden, M., Lash, T. L., & Groenwold, R. H. (2020). Reflection on modern methods: Five myths about measurement error in epidemiological research. International Journal of Epidemiology, 49(1), 338–347.