Familial Factors Associating with Youth Physical Activity Using a National Sample

Round 1

Reviewer 1 Report

The paper entitled “Familial factors associating with youth physical activity using a national sample” is consistent with the profile of the Journal Children.

1. The information presented in the abstract is adequate and describes thoroughly what the paper is about. I would recommend to change the title into more precise one: Familial factors associating with youth physical activity using an American national sample.

2. The introduction explains thoroughly the scientific background. Therefore I am convinced why this context is important. The authors use up-to-date literature to present the discussed problem in the paper.

3. Material and methods section is well prepared. The information provided in sections 2.1.-2.4 is clearly presented. The paper deserves recognition for its large sample (N=37,392).

4. The outcomes are clearly described. Key results are summarized with reference to study objectives. The discussion section provides the reference to other contributions in the studied area. The conclusions are supported by the results.

Author Response

Reviewer #1:

The paper entitled “Familial factors associating with youth physical activity using a national sample” is consistent with the profile of the Journal Children.

The information presented in the abstract is adequate and describes thoroughly what the paper is about. I would recommend to change the title into more precise one: Familial factors associating with youth physical activity using an American national sample.

-Thank you for this comment.

The introduction explains thoroughly the scientific background. Therefore I am convinced why this context is important. The authors use up-to-date literature to present the discussed problem in the paper.

-Thank you for this comment.

Material and methods section is well prepared. The information provided in sections 2.1.-2.4 is clearly presented. The paper deserves recognition for its large sample (N=37,392).

-Thank you for this comment.

The outcomes are clearly described. Key results are summarized with reference to study objectives. The discussion section provides the reference to other contributions in the studied area. The conclusions are supported by the results.

-Thank you for this comment.

Reviewer 2 Report

Thank you for providing the opportunity to review this paper. This is a comprehensive manuscript and a huge and representative sample is used which are really strong aspects of the study. Yet, a couple of questions raised while reading the manuscript especially related to the measurement of the main constructs, which therefore strongly influences the validity of some outcomes of the study.

Below, I’ll address all the issues point-by-point

Abstract L28-34: This information is now doubled. I would suggest to include the directions of the associations directly when the ORs are introduced (which is actually already the direction) to prevent this doubling.

L43: The authors are introducing child, home and environmental domains, which is conceptually very debatable. Please clearly distinguish types of environments and settings and/or apply a consistent use of ‘domains’. Child characteristic and the home environment are clearly on different levels and the use of the term ‘environmental domain’ is unclear. Do the authors mean social environment or physical environment and/or is the environmental type further specified for PA domains, such exercise etc.?

L51-53: Interactions are assumed and introduced based on the SEM, but these relevant assumptions are neglected within the further manuscript, which is disappointing. In the last decades, studies clearly stressed that it is not simply one or two characteristics in isolation that affect PA. I strongly suggest to includes interactions in the analyses taking the SEM assumptions into account. This well-powered study sample allows testing such interactions in an appropriate way.

L71-73: Why solely focusing on the rational for support for events? What is defined as event?

L87: … using data from… This adding seems irrelevant for a research question and is addressed in methods.

L140-143: Please provide some validation for the use of the question of frequency of 60 minutes p/d. Why not using more commonly used measured like frequency X duration in specific behaviors, such as sports, school PA, playing outside etc? The current question seems very much sensitive to social desirability and overestimation, which strongly affects the outcomes of the study. The current outcome does not allow to account for the recommended 60 minutes PA per day. In the question they ask for exercise and PA (without specification of at least moderate PA). This is a serious concern.

L145: The authors claim to study parental support, but the construct parental support comprises many more aspects than used in the current manuscript, e.g. see studies by Davison et al. In this study, a small component of parental support is taken into account. Moreover, it is unclear what exactly is meant by ‘events’ and whether a child and/or parents attends such events once or frequently, e.g. a training session at a sports club. Moreover, additional information is needed on what is meant with support. Is it about logistic support (bringing child), attending a training session/game etc and what is they impact of the parent’s behaviors in such occasion. Is simply being there and perhaps not only watching supportive?

L164: Was checked for a potential nested structure in the data, for instance because of state policies or regulations?

Table 1: The most relevant independent variable is highly skewed. How was this taken into account in the analyses?

L238-240: The claim that interventions should be multi-level, which may absolutely be true. However, the positioning of this statement at the start of the discussion section does not really make sense and further it is unclear based on what results in this study, this claim is made.

L253: ‘…may want to convey messages…’ How? What suggestions do the authors have for such messages taken into account that in this study no correction for own behavior was taken into account, ie. the message to attend events may not be sufficient?

L296: Addressing the whole range of health promotion behaviors is not a conclusion that can be drawn based on this study. This is actually a concern which is present in the whole conclusion section. Please rewrite based on the key outcomes of the current studies.

Author Response

Reviewer #2:

Abstract L28-34: This information is now doubled. I would suggest to include the directions of the associations directly when the ORs are introduced (which is actually already the direction) to prevent this doubling.                                                                                                                                      -Thank you for this helpful comment. We have revised the abstract to communicate the direction in the associations and to provide more useful concluding sentences (lines 20-34).

L43: The authors are introducing child, home and environmental domains, which is conceptually very debatable. Please clearly distinguish types of environments and settings and/or apply a consistent use of ‘domains’. Child characteristic and the home environment are clearly on different levels and the use of the term ‘environmental domain’ is unclear. Do the authors mean social environment or physical environment and/or is the environmental type further specified for PA domains, such exercise etc.?                                                                                                      -Thank you. We slightly revised this sentence to make it conceptually aligned with the SEM: “During childhood, numerous factors are thought to contribute to meeting physical activity guidelines, including individual determinants of behavior and determinants related to the social environment” (lines 41-44).

L51-53: Interactions are assumed and introduced based on the SEM, but these relevant assumptions are neglected within the further manuscript, which is disappointing. In the last decades, studies clearly stressed that it is not simply one or two characteristics in isolation that affect PA. I strongly suggest to includes interactions in the analyses taking the SEM assumptions into account. This well-powered study sample allows testing such interactions in an appropriate way.                                                                                                                                                                -Thank you for this insightful comment. We now conducted secondary analyses testing for joint effects (interactions) between each family-level and child-level variable (lines 198-204). Statistically significant joint effects are now presented within the text and in graphical format (lines 265-274; Figure 2).

L71-73: Why solely focusing on the rational for support for events? What is defined as event?                                                                                                                -Thank you for this comment. This NSCH item asks about attendance for child events and activities with no further elaboration on how “events” or “activities” are defined. We mention this as a limitation within the Discussion section (lines 344-348).

L87: … using data from… This adding seems irrelevant for a research question and is addressed in methods.                                                                                                                                                      -Thank you. This purpose statement has been revised (lines 96-98).

L140-143: Please provide some validation for the use of the question of frequency of 60 minutes p/d. Why not using more commonly used measured like frequency X duration in specific behaviors, such as sports, school PA, playing outside etc? The current question seems very much sensitive to social desirability and overestimation, which strongly affects the outcomes of the study. The current outcome does not allow to account for the recommended 60 minutes PA per day. In the question they ask for exercise and PA (without specification of at least moderate PA). This is a serious concern.                                                                                                                        -Thank you for this insightful comment. We have re-run all analysis using a new binary outcome variable that stratified children who met and who did not meet the 60 minute of physical activity per day guideline (lines 155-160). Because of this change, we now employed weighted logistic regression instead of weighted ordinal regression for all analyses (lines 189-190). Physical activity intensity was not assessed, we acknowledge this as a major limitation (lines 376-378).

L145: The authors claim to study parental support, but the construct parental support comprises many more aspects than used in the current manuscript, e.g. see studies by Davison et al. In this study, a small component of parental support is taken into account. Moreover, it is unclear what exactly is meant by ‘events’ and whether a child and/or parents attends such events once or frequently, e.g. a training session at a sports club. Moreover, additional information is needed on what is meant with support. Is it about logistic support (bringing child), attending a training session/game etc and what is they impact of the parent’s behaviors in such occasion. Is simply being there and perhaps not only watching supportive?                                                                     -Thank you for this helpful comment. We agree that the NSCH item only asks about one aspect of parental social support. After further reviewing the literature, we characterize the type of support as “conditional – watching/supervision” as discussed in Beets et al. We further describe this type of social support earlier within the Introduction section (lines 81-88). Although therm “logistic support” could also be used for this (as in Davison et al.). We compare/contrast the current study’s results with those found by Davison et al. within the Discussion section (lines 324-349).

L164: Was checked for a potential nested structure in the data, for instance because of state policies or regulations?                                                                                                                       -Thank you for this comment. The complex survey design was accounted for using STATA’s “svy” function. This included using “State of Residence” as the strata sampling unit. There was no other lower level nesting within the data structure; therefore, multilevel modeling with the sampling weights was not utilized.

Table 1: The most relevant independent variable is highly skewed. How was this taken into account in the analyses?                                                                                                                               -Thank you. Responses for the parental attendance variable ranged from rarely, sometimes, usually, and always. Because of the small prevalence of “rarely” and “sometimes” responses, the event attendance variable was dichotomized into rarely/sometimes and usually/always for analysis. Unfortunately, the categorizes were still skewed after this dichotomization; however, we feel that the absolute counts in each category are high enough to perform valid analysis without further manipulation of the variable. Nevertheless, we now include this as a limitation within the Discussion section (lines 376-384).

L238-240: The claim that interventions should be multi-level, which may absolutely be true. However, the positioning of this statement at the start of the discussion section does not really make sense and further it is unclear based on what results in this study, this claim is made.                                                         -Thank you. We have deleted this “multi-level” comment at the beginning of the Discussion section.

L253: ‘…may want to convey messages…’ How? What suggestions do the authors have for such messages taken into account that in this study no correction for own behavior was taken into account, ie. the message to attend events may not be sufficient?                                                                                                              -Thank you for this comment. We now provide our suggestion, given the revisions to this manuscript, within the discussion section (lines 342-349).

L296: Addressing the whole range of health promotion behaviors is not a conclusion that can be drawn based on this study. This is actually a concern which is present in the whole conclusion section. Please rewrite based on the key outcomes of the current studies.                                                                           -Thank you for this comment. We have revised the concluding paragraph and now comment only on parental correlates of physical activity (lines 387-396).                                                            

Reviewer 3 Report

The aim of the study is important and could be useful for physical activity promotion among families with kids. The authors state that "Although many of these previous studies highlight the importance of parental support for the promotion of child physical activity, they do not fully account for other familial factors that may confound the relationship..." Than cited studies which proved some of those relationships. This is very important basis for the study, unfortunately I do not see in the text the information which would be new and useful for stakeholders/ PA promoters etc. Generally you proved what was found in studies which you cited. Even more - most of them were made with better questionairs which measured really what you want (familiar support, level of PA) etc. The results which you get are probably overestimated and/or the results are probably very often parents wish, as I suppose. In my opinion you should mention this in limitation part, too. But there is a lot of potential in material, but you should try make more analyses and mayby with another statistics solve this. E.g Is it possible to answer the question if there are "agregation" effect among analized variables or if the place of residence get matter (maybe in some cases parental support could protect from not meeting PA recomendations). It would give more inside information about subject, esspecialy that you get representative sample.

I would like to mention some comments:

1.Abstract: there are no conclusion in it, but in results part are provided duplicated information (line 28-34)
2.Participants: It is no clear who was examined, from this part of text the information is that children, but in fact parents, I think it should be clarified;
3.Procedure - how do you know, if family has or has not access to web version of questionnaire? It is not clear if there was any agreement for participation signt by respondents.
4.Data processing - how was calculated BMI status?
5.Results:
For logistic regression results should be given more details (beta, SE, p-value, and result for model 1 and 2)

In line 196-201 you analyzed information on Figure 1. I was a little confused. As I understand from graph, there is no so big differences in prediction of 0 or 1-3 days meeting PA recommendation between high and low family support. I think that this is not so clear in your comments in this sentence. Secondly, I don't understand the clue... You predict meeting the PA recommendation from the level of meeting the recommendation (0,1-3,4-6,7 days/week)? This is not clear also what does it mean the 1-4 "probability" on line Y?

6. Discussion: line 259 - there is 31 in the end - wrong citation, I guess. 

Author Response

Reviewer #3:

The aim of the study is important and could be useful for physical activity promotion among families with kids. The authors state that "Although many of these previous studies highlight the importance of parental support for the promotion of child physical activity, they do not fully account for other familial factors that may confound the relationship..." Than cited studies which proved some of those relationships. This is very important basis for the study, unfortunately I do not see in the text the information which would be new and useful for stakeholders/ PA promoters etc. Generally you proved what was found in studies which you cited. Even more - most of them were made with better questionairs which measured really what you want (familiar support, level of PA) etc. The results which you get are probably overestimated and/or the results are probably very often parents wish, as I suppose. In my opinion you should mention this in limitation part, too. But there is a lot of potential in material, but you should try make more analyses and mayby with another statistics solve this. E.g Is it possible to answer the question if there are "agregation" effect among analized variables or if the place of residence get matter (maybe in some cases parental support could protect from not meeting PA recomendations). It would give more inside information about subject, esspecialy that you get representative sample.        -Thank you for this insightful comment. We added into the analysis a composite measure of family resilience (composite of 4 items). We now conducted secondary analyses testing for joint effects (interactions) between each family-level and child-level variable (lines 198-204). Statistically significant joint effects are now presented within the text and in graphical format (lines 265-274; Figure 2).

I would like to mention some comments:

1.Abstract: there are no conclusion in it, but in results part are provided duplicated information (line 28-34)                                                                                                                                                        -Thank you for this helpful comment. We have revised the abstract to communicate the direction in the associations and to provide more useful concluding sentences (lines 20-34).

2.Participants: It is no clear who was examined, from this part of text the information is that children, but in fact parents, I think it should be clarified;                                                                            -Thank you. We now clarify early in this section that it was the parents of the children who provided all information that was utilized in the current study (line 103).

3.Procedure - how do you know, if family has or has not access to web version of questionnaire?                                                                                                      -Thank you. An indicator variable on the NSCH data release provided whether or not a respondent completed an online or paper questionnaire (lines 123-124).

It is not clear if there was any agreement for participation signt by respondents.                               -Thank you. The NSCH screener questionnaire indicated to the respondents that completion of the survey was voluntary and there would be no penalties for refusing to answer questions (lines 115-117). Because this was a national survey administered by a government agency, there were no signed assent or consent forms.

4.Data processing - how was calculated BMI status?                                                                           -Thank you for this comment. The BMI class variable was sex-specific BMI-for-age categories calculated using the child’s age in months and the sex-specific 2000 Centers for Disease and Prevention Growth charts (lines 181-183).

5.Results:

For logistic regression results should be given more details (beta, SE, p-value, and result for model 1 and 2)                                                                                                                                     -Thank you for this helpful comment. The P-values for statistically significant associations are now reported within the text (lines 222-240). In STATA, for logistic regression, either the untransformed b-coefficient or Odds Ratio (OR) is in the output. The untransformed b-coefficient is not very interpretable for logistic regression, so just the OR is reported. Also, the SE is not in the STATA output for logistic regression, STATA just outputs the 95% CI for the OR.

In line 196-201 you analyzed information on Figure 1. I was a little confused. As I understand from graph, there is no so big differences in prediction of 0 or 1-3 days meeting PA recommendation between high and low family support. I think that this is not so clear in your comments in this sentence. Secondly, I don't understand the clue... You predict meeting the PA recommendation from the level of meeting the recommendation (0,1-3,4-6,7 days/week)? This is not clear also what does it mean the 1-4 "probability" on line Y?                                                                      -Thank you for this comment. Because we changed the outcome variable, per another Reviewer’s comments, to a binary physical activity variable (not ordered), we have deleted the old Figure 1. The new Figure 1A and 1B are hopefully easier to interpret and show how the probability of a child meeting the 60 minutes of physical activity per day guideline changes according to levels of the predictor variable.

Discussion: line 259 - there is 31 in the end - wrong citation, I guess.                             -Thank you. This typo has been corrected/deleted.

Round 2

Reviewer 2 Report

The authors conducted a thorough revision of their manuscript in which all comments were clearly addressed.

Due to substantial changes in the analysis, a small additional comment raised while reading. A composite measure was used for 'family resilience', however no details on the reliability of this measure were stated. Please provide some information on this, e.g. cronbach alpha. 

Author Response

Reviewer #2:

Due to substantial changes in the analysis, a small additional comment raised while reading. A composite measure was used for 'family resilience', however no details on the reliability of this measure were stated. Please provide some information on this, e.g. cronbach alpha

-Thank you for this comment. Cronbach’s alpha scale reliability coefficient has now been included in the Methods section (line 165).  

Reviewer 3 Report

Thank you for your respond, I'm satisfy with the revised version. Only one: line 245 - please add P value.

Author Response

Reviewer #3:

Thank you for your respond, I'm satisfy with the revised version. Only one: line 245 - please add P value.

-Thank you, the p-value has been added (lines 245-246).