Elisabeth Loder, MD, MPH BMJ Editorial Team Dear Prof. Loder, · 2016. 5. 5. · 1 Elisabeth Loder, MD, MPH BMJ Editorial Team [email protected] Dear Prof. Loder, We appreciate the comments

1

Elisabeth Loder, MD, MPH

BMJ Editorial Team

[email protected]

Dear Prof. Loder,

We appreciate the comments provided by the reviewers of the BMJ for our manuscript,

“Trajectory of body shape in early and middle life and all-cause and cause-specific mortality:

results from two prospective US cohort studies”. The comments were quite useful and we thank

you for the opportunity to submit the paper for your continued consideration after revisions.

The revised manuscript with “track change” of the edits made from the previous article and the

specific responses to referees’ comments have been uploaded in accordance with your

specifications.

Thank you in advance for further consideration of our manuscript. Please feel free to contact me

with any questions. On behalf of my colleagues, I hope that this revised manuscript will now be

acceptable for publication in the BMJ.

Sincerely yours,

Mingyang Song, M.D., Sc.D.

Clinical and Translational Epidemiology Unit and Division of Gastroenterology,

Massachusetts General Hospital and Harvard Medical School,

55 Fruit St, Boston, MA 02114

[email protected]

2

Comments from the manuscript meeting committee

* Our statistician has a number of concerns that are detailed in his separate statistical report.

He will be looking at any revision, so please address his inquiries carefully.

* Despite the fact that results are not especially surprising, we thought the way the information

is presented might resonate with the public, who probably are more likely to remember body

shape than what they ate or weighed at a particular point in the past.

First, please revise your paper to respond to all of the comments by the reviewers. Their reports

are available at the end of this letter, below.

In your response please provide, point by point, your replies to the comments made by the

reviewers and the editors, explaining how you have dealt with them in the paper.

We appreciate the committee’s comments and the opportunity to re-submit this manuscript. We

have addressed the reviewers’ comments as detailed below in our point-by-point response.

Specifically, as per the statistician’s thoughtful comments, we have updated our methods of

rescaling BMI at age 50 by taking into account the BMI-somatotype relationship at younger age.

The data provided in the revised version and in this letter are from the analysis using the new

method. We believe that these changes have substantially improved the manuscript. We hope that

the revised manuscript will be suitable for publication and we welcome any additional feedback.

Response to Referee #1

Comments:

The manuscript describes a novel approach to assess the impact of adiposity on mortality.

Overall, I would give high priority for publication.

Point #1: The higher RRs in women compared to men are not restricted to the category of people

who started heavy, where women an average have higher BMI than men; for example in the

group lean-marked increase women have (roughly) twice excess risk compared to men for all

causes of death. One explanation is that the somatotypes seem to perform better in women than

in men (see ref. 20). Alternatively, the difference may reflect some heterogeneity in the biological

of adiposity early in life between sexes. In any case, more discussion on this issue would be

welcome.

We appreciate the reviewer’s comments. We agree that the better performance of somatotypes in

assessing body size in women than in men may have contributed to the gender difference

3

observed in the trajectory-mortality associations. Additionally, because trajectories were

constructed within each cohort separately and sex-specific reference groups were used in the

analysis, it is difficult to directly compare the results between women and men. Similar notions

apply to most previous studies that have used sex-specific percentiles to assess early-life BMI in

relation to adult mortality. Among studies that have used common cutoffs in both genders, there

is no strong evidence suggesting that obesity, whether in early life1 2

or in middle or late

adulthood,3-5

is differentially associated with mortality risk in women versus men. We have

added more discussion on the sex difference in the revised manuscript.

Discussion

Of note, the relative risk estimates were stronger in women than in men,

although strict comparison between the two sexes is difficult because the

trajectories were created within each cohort separately and sex-specific

reference groups were used in the analysis. Better performance of the

pictograms in assessing body shape in women (see Methods) may have

contributed to the stronger results than in men. Another explanation may be

related to the much higher BMI in the heavy-stable/increase group in women

than in men (e.g., mean BMI at age 50: 32.9 vs. 29.0 kg/m2, Table 1).

(DISCUSSION, Page 23)

Point #2: Similarly, the negative interaction with smoking deserves more discussion. Although it

is obvious that smoking may confound the association between BMI/somatotype and mortality

across categories, it is less clear whether confounding alone would explain the much smaller

effect compared to never smokers within BMI/somatotype categories, and why the confounding

effect would be stronger in women than in men.

We agree that the trajectory-mortality associations were strikingly stronger in never smokers than

in all subjects or in ever smokers. Similar difference has been reported in many previous

studies.6-9

For example, in a pooled study of 19 cohorts (n=1.46 million), compared to men with

a BMI of 22.5-24.9 kg/m2, the hazard ratios of all-cause mortality among all subjects versus

never smokers only were 1.03 and 1.17 for BMI of 27.5-29.9 kg/m2, 1.16 and 1.43 for BMI of

30.0-34.9 kg/m2, 1.44 and 1.95 for BMI of 35-39.9 kg/m

2, and 1.93 and 2.86 for BMI of 40-49.9

kg/m2, respectively.

8 Similar results were reported in women. Therefore, the strong effect

modification by smoking we observed in the current study is consistent with previous reports.

Moreover, compared to studies of BMI measured once at baseline or at a certain age, our study

may be more able to capture the confounding effect of smoking by assessing the body shape over

a long period of life, because smoking status can change at any time point in life and any

accompanied effect on body weight would have been incorporated into our analysis of all

subjects, which may have exaggerated the difference in the results from the restricted analysis

among never smokers.

4

Regarding the sex difference, we agree that the difference in trajectory-mortality associations

between never and ever smokers appeared somewhat more evident in women than in men. One

explanation may be related to the higher proportion of current smokers among women than

among men (14% vs. 7%) within our cohorts. Furthermore, there is some, albeit inconsistent,

evidence suggesting that a greater metabolic influence of smoking in women than men and

women tend to gain more weight after smoking cessation than men.10-12

Therefore, excluding

smokers from the analysis may have a larger impact on the trajectory-mortality associations in

women than men. We have added these discussions to the revised manuscript, as follows:

Discussion

Indeed, consistent with previous studies,5 13

we found that body-shape

trajectory was more strongly associated with mortality among never smokers

than among ever smokers. This difference appeared more striking in women

than in men, which may be due to the higher proportion of current smokers

among women than among men (14% vs. 7%) within our cohorts. Moreover,

some, albeit inconsistent, evidence suggests a greater metabolic influence of

smoking in women than men, and women tend to gain more weight after

smoking cessation than men.10-12

Therefore, excluding smokers from the

analysis may have a larger impact on the trajectory-mortality associations in

women than men.

(DISCUSSION, Page 24)

Point #3: Although the paper represents an elegant epidemiologic exercise, and might inform on

underlying biologic mechanisms, its contribution to clinical practice and public health is less

clear. How much does the analysis of lifespan trajectories add to a straightforward assessment of

current (i.e., adulthood) adiposity? RRs in the category lean-marked increase do not appear (at

least in men) to differ much from RRs among those who started heavy. Should a general

practitioner try to assess the trajectory of their patient (presumably with a large amount of

misclassification) rather than just measuring current BMI?

We appreciate the reviewer’s positive comments that our study “represents an elegant

epidemiologic exercise and might inform underlying biological mechanisms”. We agree that in

men the RR of all-cause mortality for the lean-marked increase group did not substantially differ

from that for the heavy-stable/increase group in men (among never smokers: 1.19 vs. 1.28).

However, considering that the lean-marked increase group includes over twice more participants

than the heavy-stable/increase group, dissecting the two groups is not trivial from a public health

point of view, and any moderate difference in the RRs can have a tremendous impact on

estimating the population risk of death attributable to each of the two trajectories. Regarding

whether assessing patients’ trajectory would produce significantly more information than just

measuring current BMI, we believe that it is definitely an interesting question that deserves a

5

formal assessment in future studies, by comparing clinically relevant measures (e.g., predictive

capability of the two instruments for mortality) and also taking into account the cost associated

with each tool. Nevertheless, given the advancement of technology, especially the growing use of

electronic medical records,14

assessment of individuals’ trajectory will conceivably have an

improved accuracy and may not be an overwhelming task for clinical practitioners to perform in

the future.


Comments:

This study took advantage of the multiple body-shape data before mid-life obtained

retrospectively in conjunction with a BMI transformed body-shape data to investigate the

association between body-shape trajectories and mortality risk. A model-based modeling

approach was used to identify 5 distinct trajectories of body shape, i.e., lean-stable, lean-

moderate increase, lean-marked increase, medium-stable/increase, and heavy-stable/increase.

Authors found that the lean-stable had the lowest mortality risk and the consistently heavy body-

shape was associated with the highest mortality risk. In addition, this phenomenon was more

apparent in non-smokers than in smokers and for CVD than for other diseases. Significant

trajectory-smoking interactions were clearly demonstrated.

Major comments

This study has made use of a very unique set of body shape data collected retrospectively for

several time points before mid-life, which allowed for generating body-shape trajectories and

studying how these different trajectories were associated with mortality. Personally, I feel that

this type of study is what is needed to challenge the so called U-shape relation for BMI-mortality

which relies on BMI data on one time point.

We appreciate the reviewer’s comments, and agree that our trajectory analysis can better address

some of the methodological challenges currently encountered in assessing the influence of body

weight on long-term mortality. We have detailed as below our response to each of the reviewers’

comments.

Point #1: Title: The paper is about the trajectory of body shape “before mid-life” not “across

the lifespan”. Some modification should be made for the title and across the text.

We have modified the title and the text as per the reviewer’s suggestion.

Title: Trajectory of body shape in early and middle life and all-cause and

6

cause-specific mortality: results from two prospective US cohort studies

Abstract

Objective: To assess body-shape trajectories in early and middle life in

relation to risk of mortality

Conclusions: Using the trajectory approach, we found that heavy body shape

from age 5 up to 50, especially the increase in middle life, was associated with

higher mortality.

Introduction

Therefore, to extend our knowledge, we used a different, trajectory-based

approach to assess the relationship between body shape in early and middle

life and risk of all-cause and cause-specific mortality in two large US cohort

studies.

Discussion

To our knowledge, this is the first study to investigate adiposity throughout

early and middle life in relation to mortality.

In conclusion, we found that heavy body shape throughout early and middle

life, especially the increase in middle life, was associated with higher

mortality.

Point #2: Abstract and Discussion: This paper should stress not only the increased mortality risk

of heavy body shape, but also the lowest risk of the lean stable in the conclusion statement to

refute the previous viewpoint on increased risk at low BMI.

We have modified the abstract and discussion as per the reviewer’s suggestion.

Abstract

Conclusion:

Using the trajectory approach, we found that heavy body shape from age 5 up

to 50, especially the increase in middle life, was associated with higher

mortality. In contrast, individuals who maintained a stably lean body shape

had the lowest mortality. These results indicate the importance of weight

management across the lifespan.

Discussion

By comparing mortality risk between trajectory subgroups, we found that

participants who remained heavy from age 5 to 50 had the highest risk of

7

death, whereas those who maintained a stably lean body shape had the lowest

mortality. Compared with the latter group of individuals, even those who were

lean in childhood or adolescence but gained weight in middle life were at

higher risk of mortality.

In conclusion, we found that heavy body shape throughout early and middle

life, especially the increase in middle life, was associated with higher

mortality. In contrast, individuals who maintained a stably lean body shape

had the lowest mortality. These results indicate the health benefit of weight

management across the lifespan.

Point #3: Please make clear how the BMI data at age 50 is converted to body shape information

in the method section. Although authors pointed out the more details about the conversion are

provided in Supplementary Table 1, it is not clear by reading the Supplementary Table 1. Where

is the data coming from in Supplementary Table 1? Would you please provide the reference?

As per the comments by Reviewer 3, we have revised our method of rescaling BMI at age 50 by

taking into account the BMI-somatotype relationship at younger age in a linear mixed effects

model. Detailed descriptions about the methods have been provided in the response to the Point

#1 of Reviewer 3 on pages 9-12.

Point #4: Line 31: Reference 24 was submitted in 2015. Please provide the journal information

was provided.

We have provided detailed citation for Reference 24 (the current Reference 21).

References

21. Song M, Willett WC, Hu FB, et al. Trajectory of body shape across the

lifespan and cancer risk. Int J Cancer 2015.

(REFERENCES, Page 31)

Point #5: Line 33 to line 42: It is not clear how authors calculated the mean BMI at different

ages in each trajectory group in Table 1, since body shape information was recalled, and BMI

was not measured for earlier life time points.

In the two cohorts, we have collected body weight data every two years since baseline by our

follow-up questionnaire. We used these data to calculate the BMI for ages 40 and 50 as shown in

Table 1. In addition, we asked participants to recall their body weight at adolescence (age 18 for

women and 21 for men) in 1980 in the Nurses’ Health Study and in 1986 in the Health

Professionals Follow-up Study, and included these data in Table 1. Because some participants

8

had already been older than 40 or 50 at baseline, or did not provide their weight at age 18 or 21,

the BMI data were not available for all the participants included in the analysis. We have noted

this in the footnote of Table 1. We have also provided more details in the methods to clarify this

point.

Table 1 Footnote b Data were not available in all participants because some participants had

already been older than 40 or 50 at baseline, or did not provide their body

weight at age 18 or 21.

(TABLE 1 FOOTNOTE, Page 13)

Methods

Body shape assessment

Height and body weight were queried on biennial follow-up questionnaires.

We used these data to calculate the BMI at age 50 and then converted it to the

same scale as somatotypes in younger ages. More details about this

conversion are provided in the Supplementary Methods. In addition, recalled

body weight at age 18 was inquired in 1980 in the NHS, and weight at age 21

was inquired in 1986 in the HPFS, as previously described.15

We used these

data to calculate the BMI at adolescence.

(METHODS, Page 8)

Point #6: Line 43-48: It is stated that “those in the lean-stable group were more physically

active, tended to use multivitamin, and consumed a healthier diet than those in the other

groups.” Statistical testing should be carried out.

Because of the large sample size, all the statistical tests yielded a P value of <0.001. We have

included these results in the table footnote.

Table 1 Footnote a All variables are standardized by age at baseline (1976 for women in the

Nurses’ Health Study and 1986 for men in the Health Professionals Follow-up

Study). Means are presented for continuous variables. Due to the large sample

size, P values for testing the difference across the trajectory groups were all

<0.001 for the variables listed in the table.

(TABLE 1 FOOTNOTE, Page 13)


9

Comments:

The authors use two large cohorts to construct five shape trajectory groups from age 5 to 50, and

then compare mortality in the diffferent groups. I have some comments on the study design,

analysis and presentation.

Point #1: The key exposure here is the recalled somatotype at ages 5, 10, 20, 30 and 40, and the

corresponding rating at age 50 is inferred from the age 40 rating. The way this is done strikes me

as clunky, based on just the previous rating and ignoring the earlier ones. It also requires the age

40 rating to be present to estimate the age 50 rating. Since the purpose is to represent shape over

the life course it would surely be better to use all the available shape ratings and BMI at 40 to

impute the age 50 rating.

The description on page 39 implies, though does not state, that BMI was measured at ages 40

and 50: “…we assessed the average BMI from age 47 to 53 as the BMI for age 50. We then

divided BMI at these two ages into 9 categories”. Which two ages?

“The cutoff points for each category were calculated as the median BMI of this category at age

40 plus a constant to account for weight gain from age 40 to 50”. But why use the median as the

upper cut-off, which will misclassify half those in the group? Surely one needs cutoffs midway

between the group medians?

Also, using a single value of 1.5 kg/m2 for 10-year BMI gain ignores the fact that BMI is

increasing over time in some groups but not in others. The calculation needs to take into account

all the available information on individual trajectories.

We appreciate the reviewer’s thoughtful comments. First, we want to clarify that the somatotype

rating at age 50 is not inferred from the age 40 rating, but rather is based on the real BMI at age

50. What we did was just to find the cutoffs to rescale the BMI at age 50 into 9 groups, to be

consistent with the somatotype categorization in earlier ages.

Second, regarding how we assessed the BMI at age 40 and 50, we obtained the BMI data from

our biennial follow-up questionnaires, in which we queried participants about their current

weight every two years. To reduce random variation, we used the average BMI from age 37 to 43

to represent the BMI for age 40, and the BMI from age 47 to 53 as the BMI for age 50.

We agree that using the body shape ratings in earlier years will improve the rescaling accuracy

for BMI at age 50, that using medians as the cutoffs may have resulted in misclassification, and

that using a single value for the 10-year BMI gain ignores the fact that BMI is increasing over

time in some individuals but not in others. As per the reviewer’s comments, we have changed our

rescaling method. Instead of using the BMI data at age 40 only, we used all the time points at

which we had both BMI and somatotype data to run a linear mixed effects model, in which we

assumed there is a linear relationship between BMI and somatotype. To allow such relationship

to vary among individuals, we included a random intercept and a random slope in the model.

10

Therefore the model is specified as follows:

�� = �� + �� + �� + �� + ��

where �� denotes the BMI for individual � at age and �� denotes the corresponding

somatotype, �� and �� specify the fixed intercept and slope, �� and �� specify the random,

subject-specific intercept and slope, and �� represents the within-subject measurement error.

Furthermore, it is assumed that �~��0, ��, ��~��0, �� , and that � and �� are mutually

independent. We used an unstructured variance-covariance matrix (i.e., without making any

particular assumption about the covariance structure) for estimation.

We modeled BMI instead of somatotype as the dependent variable because BMI is

approximately normally distributed, whereas somatotype is a discrete variable ranging from 1 to

9. As such, the conditional mean BMI for individual � at age can be written as:

�[��|��, ��] = �� + �� + �� + ��

Based on the model output, we then calculated the somatotype rating at age 50 �� as

��[�� |�!�,�"�]#$!#�!�$"%�"�

&. To keep the estimated somatotypes within the range of 1 to 9, for �� that

was larger than 9, we rounded it as 9, and for �� that was smaller than 1, we rounded it as 1.

There were 4 and 3 time points available for the linear mixed effects modeling in women and

men, respectively, as summarized below.

Time point Women

(Nurses’ Health Study)

Men

(Health Professionals Follow-up Study)

1 BMI: age 18

Somatotype: age 20

BMI: age 21

Somatotype: age 20

2 BMI: age 30

Somatotype: age 30

BMI: age 40

Somatotype: age 40

3 BMI: age 40

Somatotype: age 40

BMI: in 1988

Somatotype: in 1988

4 BMI: in 1988

Somatotype: in 1988

The details about anthropometric assessments at each time point are described below. It should

be noted that there may be missing data for each of the time points, but the missing data can be

accommodated by the linear mixed effects model.

In women (Nurses’ Health Study):

• Time point 1: in 1980 participants were asked to recall their body weight at age 18. We

then calculated their BMI, and linked it to their somatotype ratings at age 20.

• Time points 2 and 3: participants were queried about their current body weight in our

biennial follow-up questionnaires. We used these data to calculate their BMI at age 30

(the mean BMI from age 27 to 33) and 40 (the mean BMI from age 37 to 43), and then

linked it to their somatotype ratings at age 30 and 40.

11

• Time point 4: we calculated participants’ BMI in 1988 based on the follow-up

questionnaires, and then linked it to their somatotype ratings in 1988 when we also

queried about their current body shape.

In men (Health Professionals Follow-up Study):

• Time point 1: in 1986 participants were asked to recall their body weight at age 21. We

then calculated their BMI, and linked it to their somatotype ratings at age 20.

• Time point 2: participants were queried about their current body weight in our biennial

follow-up questionnaires. We used these data to calculate their BMI at age 40 (the mean

BMI from age 37 to 43), and then linked it to their somatotype ratings at age 40. We did

not include the BMI at age 30, because almost all participants were already at their 40s

when enrolled into the cohort and their BMI data at age 30 were not available.

• Time point 3: we calculated participants’ BMI in 1988 based on the follow-up

questionnaires, and then linked it to their somatotype ratings in 1988 when we also

queried about their current body shape.

Using these re-calculated somatotype ratings for age 50, we re-constructed the trajectory within

each cohort. As shown below, the plots look very similar to what we had before, except for the

slight increase of body shape levels in the “lean-stable” group. Such small increase may reflect

the natural, body fatness-independent change of body shape over time (e.g., although a person

may maintain the same BMI from childhood to middle adulthood, his/her body shape may

appear to increase during this time period).

Using these newly created trajectories, we assessed the relationship to mortality. Reassuringly

the results were quite similar to what we obtained before. For example, the HR (95% CI) for all-

cause mortality in the “heavy-stable/increase” group compared to the “lean-stable” group

changed from 1.48 (1.40-1.57) to 1.45 (1.38-1.54) in women, and from 1.20 (1.13-1.28) to 1.22

(1.14-1.29). Such similarity provides further support for the robustness of our findings. For the

reviewer’s convenience to compare the results, we have summarized the old and new main

findings in the table below.

Table. Hazard ratio of all-cause and cause-specific mortality according to trajectories of

body shape from age 5 to 50 among women in women and men using the old and new

rescaling method for BMI at age 50

Lean-stable

Lean-moderate

increase

Lean-marked

increase

Medium-

stable/increase

Heavy-

stable/increase

Women

All-cause

HR (95% CI), old 1 (reference) 1.04 (1.00-1.08) 1.12 (1.08-1.17) 1.03 (0.99-1.08) 1.48 (1.40-1.57)

HR (95% CI), new 1 (reference) 1.05 (1.01-1.09) 1.29 (1.23-1.35) 1.06 (1.02-1.10) 1.45 (1.38-1.54)

Cardiovascular disease



Coronary heart disease



12

Stroke



Cancer



Other causes



Men

All-cause



Cardiovascular disease



Coronary heart disease



Stroke



Cancer



Other causes



We have updated the manuscript and all the tables and figures in the revised version using the

new method. All the numbers referred in this response letter are from the updated results. We

have also updated the method description for rescaling of the BMI at age 50 in the

Supplementary Materials, as shown below.

Supplementary Materials

Rescaling of BMI at age 50 years

To minimize random variation, we assessed the average BMI from age 37 to

43 to represent the BMI for age 40, and the average BMI from age 47 to 53 as

the BMI for age 50. We then rescaled the BMI at age 50 into 9 categories,

consistent with the grouping of somatotypes (ranging from 1 to 9) at younger

ages. The rescaling was conducted by using a linear mixed effects model, in

which we assumed there is a linear relationship between BMI and somatotype

at each age. We then used such relationship at younger ages to derive the

somatotype at age 50 from the corresponding BMI at age 50. To allow this

relationship to vary among individuals, we included a random intercept and a

random slope in the model, as specified below:

�� = �� + �� + �� + �� + ��

13

where �� denotes the BMI for individual � at age and �� denotes the

corresponding somatotype, �� and �� specify the fixed intercept and slope, �� and �� specify the random, subject-specific intercept and slope, and �� represents the within-subject measurement error. Furthermore, it is assumed

that �~��0, ��, ��~��0, ��, and that � and �� are mutually independent.

We used an unstructured variance-covariance matrix (i.e., without making any

particular assumption about the covariance structure) for estimation.

We modeled BMI instead of somatotype as the dependent variable because

BMI is approximately normally distributed, whereas somatotype is a discrete

variable ranging from 1 to 9. Thus, the conditional mean BMI for individual � at age can be written as:

�[��|��, ��] = �� + �� + �� + �� Based on the model output, we then calculated the somatotype rating at age 50

�� as ��[�� |�!�,�"�]#$!#�!�$"%�"�

&. To keep the estimated somatotypes within the

range of 1 to 9, for �� that was larger than 9, we rounded it as 9, and for ��

that was smaller than 1, we rounded it as 1.

There were 4 and 3 time points available for the linear mixed effects modeling

in women and men, respectively, as summarized below.

Time point Women

(Nurses’ Health Study)

Men

(Health Professionals Follow-up Study)

1 BMI: age 18

Somatotype: age 20

BMI: age 21

Somatotype: age 20

2 BMI: age 30

Somatotype: age 30

BMI: age 40

Somatotype: age 40

3 BMI: age 40

Somatotype: age 40

BMI: in 1988

Somatotype: in 1988

4 BMI: in 1988

Somatotype: in 1988

The details about anthropometric assessments at each time point are described

below. It should be noted that there may be missing data for each of the time

points, but the missing data can be accommodated by the linear mixed effects

model.

In women (Nurses’ Health Study):

• Time point 1: in 1980 participants were asked to recall their body weight

at age 18. We then calculated their BMI, and linked it to their somatotype

ratings at age 20.

• Time points 2 and 3: participants were queried about their current body

weight in our biennial follow-up questionnaires. We used these data to

calculate their BMI at age 30 (the mean BMI from age 27 to 33) and 40

14

(the mean BMI from age 37 to 43), and then linked it to their somatotype

ratings at age 30 and 40.

• Time point 4: we calculated participants’ BMI in 1988 based on the

follow-up questionnaires, and then linked it to their somatotype ratings in

1988 when we also queried about their current body shape.

In men (Health Professionals Follow-up Study):

• Time point 1: in 1986 participants were asked to recall their body weight

at age 21. We then calculated their BMI, and linked it to their somatotype

ratings at age 20.

• Time point 2: participants were queried about their current body weight

in our biennial follow-up questionnaires. We used these data to calculate

their BMI at age 40 (the mean BMI from age 37 to 43), and then linked it

to their somatotype ratings at age 40. We did not include the BMI at age

30, because almost all participants were already at their 40s when

enrolled into the cohort and their BMI data at age 30 were not available.

• Time point 3: we calculated participants’ BMI in 1988 based on the

follow-up questionnaires, and then linked it to their somatotype ratings in

1988 when we also queried about their current body shape.

Point #2: Related to this, Figure 1 shows that 28% of individuals were excluded as they had

“missing somatotype data for more than two different age points”, or put more simply, fewer

than 4 somatotypes. This serious data loss strikes me as unnecessary – Figure 1 shows that the

group trajectories are essentially linear (despite the cubic fit), which means that anyone with 2

or more somatotypes could reasonably be analysed. Even requiring a minimum of 3 ought to

reduce the dropout appreciably.

We appreciate the reviewer’s thoughtful comments. However, lowering the required number of

complete data points would not reduce the dropout appreciably, because most of the excluded

participants had missing somatotype data for 5 or 6 age points. This is understandable because all

the somatotype data from age 5 to 40 were collected at the same time, and the participant who

provided somatotype data for one age was likely to answer all the somatotype questions. The

following table shows the number of the excluded participants due to missing somatotype data

according to the number of missing data points.

Number of the age points that

have missing somatotype data Women Men

3 208 305

4 86 149

5 26,315 5,186

6 7,323 8,061

Total 33,932 13,701

Therefore, the gain in sample size will be minimal (208 in women, 305 in men) if we change the

15

exclusion criterion to missing somatotype data for more than three (instead of two) age points.

Point #3: A second data exclusion is for BMI < 18.5 kg/m2. I can see that this is meant to reduce

reverse causation, but since the whole purpose is to model BMI, it seems illogical to omit

individuals whose BMI is arbitrarily low. In any case the numbers involved are tiny and will

make no difference to the results.

Yes, we excluded participants with BMI of <18.5 kg/m2 to reduce reverse causation, because

these participants likely have some underlying chronic disease that may predispose them to early

death. As the reviewer pointed out, because of the small numbers involved, our results were

robust to this exclusion. For example, the HRs (95% CI) of all-cause mortality for the heavy-

stable/increase group in women were 1.45 (1.38-1.54) in the analysis including participants with

BMI of <18.5 kg/m2, and 1.48 (1.40-1.57) after excluding these participants. Since BMI <18.5

kg/m2 technically falls outside the range of normal BMI, we think that it is preferable to exclude

them.

Point #4: Related to point 6 above, the essential linearity of the group trajectories indicates that

a simple random-slope-random-intercept model would lead to broadly the same conclusions, and

the relative size of the slope and intercept random effects in predicting mortality would quantify

the importance of mean BMI versus BMI gain.

We agree that the essential linearity of body-shape measurements with age makes the random-

slope-random-intercept model a potential choice to assess subject-specific trajectory. Also, as the

reviewer pointed out, comparing the relative size of the slope and intercept random effects in

predicting mortality would give us some sense about the importance of mean BMI versus BMI

gain. However, we believe that the approach we employed has some unique advantages. First, it

avoids the difficulty in interpretation of the results obtained from statistical modeling of different

dimensions of adiposity (e.g., mean BMI versus BMI gain in the approach suggested by the

reviewer), because these measures are not biologically independent. For example, the mean BMI

at age 50 depends largely on how BMI changes over early years, and therefore it is difficult, if

not biologically meaningful, to separate their independent effects on mortality. In contrast,

instead of comparing different adiposity measures, our approach focuses on comparison of

different groups of individuals who have distinct body-shape trajectory profiles, and addresses a

more tangible and clinically relevant question: how does mortality differ among individuals with

different trajectories of body shape.

Another advantage of classifying individuals into distinct groups is that it provides an intuitive

way to probe into the population heterogeneity in the susceptibility of body shape change across

the lifespan, which does not only yield an easily-digested message for the public but also has

great implications for future research. For example, further studies examining the relative

16

contributions of individuals’ genetics and behaviors to their trajectory profiles will provide

critical insights into tailored prevention strategies. If it is found that, for instance, the heavy-

stable/increase group has a larger genetic component, whereas the lean-marked increase group is

more behaviorally oriented, public health strategies of behavioral change should then be targeted

towards the latter group. In addition, these approaches, for example identifying risk genes, are

much more feasible for mutually exclusive phenotypes than by using multivariable analysis,

where the meaning of a variable is conditional on other variables.

A potential limitation of our approach is information loss due to discrete grouping. However, the

good discrimination of our trajectory-building model (mean posterior probability of trajectory

assignment: 0.9) and well-tracked change in BMI across trajectories indicate that the trajectories

we identified can parsimoniously summarize, without a significant loss of information, the

predominant features of lifetime body shape in the study population. Additionally, it was

reassuring that we obtained similar results after excluding participants with suboptimal trajectory

assignment.

Point #5: Is there a way to superimpose on the group trajectories of Figure 1 the mean BMI

values appearing in Table 1? It would provide some validation of the group allocation.

We tried to add to Figure 1 all the mean BMI values in Table 1. But because the Y axis

encompasses the somatotype ratings for 6 different ages and across 5 different trajectory groups,

superimposing all the BMI data does not seem to work in Excel. Instead, we have added the

mean BMI values for age 50 only to the figures, as shown on the next page.

17

18

Point #6: The hazard ratios in Tables 2 and subsequently are adjusted for a string of covariates

including lifestyle factors such as physical activity, alcohol consumption and dietary score. Are

these factors not on the causal pathway, and hence should not be adjusted for? The research

question involves the shape trajectories versus mortality, which must be due at least in part to

lifestyle differences – so why adjust for them if the interest is in the trajectories themselves?

It is a challenge to control for confounding in such studies as our current one of the exposure

across the lifespan, because some covariates can be both mediators and confounders for the

exposure-outcome relationship. Here we adjusted for several lifestyle factors because they may

influence individuals’ body-shape trajectory and are also important predictors for mortality. Not

adjusting for these factors may lead to confounding bias (i.e., the observed relationship between

body shape and mortality may be due to other lifestyle factors that are related to body shape).

However, as the reviewer pointed out, there can also be over-adjustment because individuals

with certain body shapes may be more likely to change their lifestyle that can in turn influence

their mortality risk. We tried to minimize such over-adjustment by using the cumulative average

lifestyle information collected throughout early and mid-life to age 50, because such long-term

habitual exposure is less susceptible to temporary lifestyle changes that were caused by body-

shape change at some time points during follow-up.

We also compared the crude and adjusted hazard ratios among never smokers. As shown in the

table below, the results were quite similar; indicating that neither confounding nor over-

adjustment had a substantial influence on our results, after we account for the confounding effect

by smoking.

19

Crude and adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) of all-cause and cause-

specific mortality according to trajectories of body shape from age 5 to 50 among never smokers of

women in the Nurses’ Health Study and men in the Health Professionals Follow-up Study

Lean-stable

Lean-moderate

increase

Lean-marked

increase

Medium-

stable/increase

Heavy-

stable/increase

Women

All-cause

Crude HR (95% CI) 1 (reference) 1.09 (1.03-1.16) 1.53 (1.43-1.65) 1.05 (0.98-1.13) 1.75 (1.60-1.92)

Adjusted HR (95% CI) 1 (reference) 1.08 (1.02-1.15) 1.43 (1.33-1.54) 1.04 (0.97-1.12) 1.63 (1.49-1.79)

Cardiovascular disease (CVD)



Coronary heart disease (CHD)



Stroke



Cancer



Other causes



Men

All-cause



Cardiovascular disease (CVD)



Coronary heart disease (CHD)



Stroke



Cancer



Other causes



In the multivariable model, we adjusted for age (continuous), height (continuous), race (non-white or white),

pack-years of smoking (0, 1-<6, 6-≤20, or >20), family history of cancer (yes or no), history of lower

gastrointestinal endoscopy (yes or no; for analysis of total cancer), multivitamin use (yes or no), regular

aspirin/NSAID use (yes or no), history of physical exam (yes and for screening, yes and for symptoms, or no),

mammography (women only, yes and for screening, yes and for symptoms, or no; for analysis of total cancer),

menopausal hormone therapy (women only, past use, current use, or no), prostate-specific antigen test (men

only, yes or no; for analysis of total cancer), physical activity (in quintiles), alcohol consumption (0-<0.5, 0.5-

<2, 2-<8, or ≥8 g/d), and AHEI dietary score (in quintiles).

20

Point #7: Table 2 shows that for deaths due to stroke in women smokers (though not in men),

groups 3 and 4 are significantly protected relative to the lean-stable group. This is surely worth a

mention.

These protective associations did not show up any more in our updated results. Now the HRs

(95% CI) for stroke mortality in female smokers were 0.94 (0.72-1.23) for the “lean-marked

increase” group and 0.98 (0.80-1.22) for the “medium-stable/increase” group.

Point #8: The interaction in Figure 2B is said to be insignificant (P = 0.46), yet the two trends

look strikingly different, and much more so than in Figures 2A and 2C. Is it correct? It’s not

entirely clear what sort of interaction has been fitted – I assume it's comparing the linear trends

across shape groups in the two diabetes groups.

We did not assume a linear trend across trajectory groups. Instead, we used the likelihood ratio

test with 3 degrees of freedom to calculate the P value for interaction by treating the trajectory

group as a categorical variable (4 categories, with the medium-stable/increase group combined

with the lean-moderate increase group), and comparing the model with the product terms

between trajectory groups (3 indicator variables) and diabetic history (binary variable) to the

models without these terms. Although the hazard ratios (HRs) for CVD mortality were quite

different between diabetics and non-diabetics, they demonstrated more of an additive

relationship. In contrast, for all-cause and cancer mortality, there is more evident deviation from

a multiplicative relationship, which is what we are testing for (i.e., multiplicative interaction).

For example, for CVD mortality, the product of the HRs for the independent effects of history of

diabetes (i.e., the HR for the lean-stable group with diabetes) and heavy-stable/increase

trajectory (i.e., the HR for the heavy-stable/increase group without diabetes) was 1.29*1.97 =

2.54, which accounts for about 87% of the HR for the joint effect of the two factors (2.92). In

contrast, for cancer mortality, the product of the two HRs was 0.93*1.18=1.10, which accounts

for only half of the HR for the joint effect (2.08). Given that the interaction test in this case is to

assess deviation from multiplicative relationship (rather than additive interaction), it is not

surprising that the P value is not statistically significant for CVD mortality compared to that for

all-cause or cancer mortality.

As a sensitivity analysis, we also calculated the P value for interaction by comparing the linear

trends across trajectory groups in the two diabetes strata (the four trajectory groups from left to

right were ordered as 1, 2, 3, and 4). The results were similar to what we had: P=0.004 for all-

cause mortality, 0.46 for CVD mortality, and 0.001 for cancer mortality.

In addition, to see which trajectory group contributes most to the significant interaction, we also

examine the P value for interaction for each individual trajectory group (i.e., Wald test for each

of the product terms between trajectory group and diabetic history). As shown in the following

21

table, only the heavy-stable/increase group showed significant interaction for all-cause and

cancer mortality, indicating that this group dominates the overall interaction; whereas no

significant interaction was detected for CVD mortality.

Table. P value for the individual interaction test for each trajectory group with diabetic history

Lean/medium-

moderate increase

Lean-marked

increase

Heavy-

stable/increase

All-cause mortality 0.60 0.12 0.03

CVD mortality 0.40 0.22 0.69

Cancer mortality 0.97 0.28 0.003

We have updated the text and added these results to the figure legend, as follows.

Figure 2. Joint association of trajectories of body shape and history of

type 2 diabetes with risk of all-cause (A), cardiovascular (B), and cancer

(C) mortality among never smokers

Lean-moderate increase and medium-stable/increase groups were combined as

the “lean/medium-moderate increase” group due to small number of cases.

Multivariable Cox-proportional hazards model was used to calculate the

hazard ratio within each cohort after adjusting for the same set of covariates as

in Tables 2 and 3. Participants were categorized into 8 groups according to

trajectories and history of type 2 diabetes, with those in the lean-stable group

and without history of type 2 diabetes as the reference. Pooled results from the

two cohorts are shown here. The P values for interaction shown in the figures

were calculated from likelihood ratio test with three degrees of freedom by

comparing the model with the product terms between diabetic history (binary)

and the trajectory groups (indicator variables for the three non-reference

groups) to the model without these terms. We also calculated the P value for

interaction for each individual trajectory group by Wald test. For all-cause

mortality, the individual P value for interaction was 0. 60 for the

lean/medium-moderate increase group, 0.12 for the lean-marked increase

group, and 0.03 for the heavy-stable/increase group; the corresponding P

values for cardiovascular mortality were 0.40, 0.22, and 0.69; and for cancer

mortality were 0.97, 0.28, and 0.003.

(FIGURE LEGEND, Page 36)

The figures are shown on the next page.

22

23

References

1. Hirko KA, Kantor ED, Cohen SS, Blot WJ, Stampfer MJ, Signorello LB. Body mass

index in young adulthood, obesity trajectory, and premature mortality. Am J Epidemiol

2015;182:441-50.

2. Ma J, Flanders WD, Ward EM, Jemal A. Body mass index in young adulthood and

premature death: analyses of the US National Health Interview Survey linked mortality files. Am

J Epidemiol 2011;174:934-44.

3. Adams KF, Schatzkin A, Harris TB, et al. Overweight, obesity, and mortality in a large

prospective cohort of persons 50 to 71 years old. N Engl J Med 2006;355:763-78.

4. Prospective Studies C, Whitlock G, Lewington S, et al. Body-mass index and cause-

specific mortality in 900 000 adults: collaborative analyses of 57 prospective studies. Lancet

2009;373:1083-96.

5. Berrington de Gonzalez A, Hartge P, Cerhan JR, et al. Body-mass index and mortality

among 1.46 million white adults. N Engl J Med 2010;363:2211-9.

6. Manson JE, Willett WC, Stampfer MJ, et al. Body weight and mortality among women.

24

The New England journal of medicine 1995;333:677-85.

7. Adams KF, Schatzkin A, Harris TB, et al. Overweight, obesity, and mortality in a large

prospective cohort of persons 50 to 71 years old. The New England journal of medicine

2006;355:763-78.

8. Berrington de Gonzalez A, Hartge P, Cerhan JR, et al. Body-mass index and mortality

among 1.46 million white adults. The New England journal of medicine 2010;363:2211-9.

9. Hjellvik V, Selmer R, Gjessing HK, Tverdal A, Vollset SE. Body mass index, smoking,

and risk of death between 40 and 70 years of age in a Norwegian cohort of 32,727 women and

33,475 men. European journal of epidemiology 2013;28:35-43.

10. Tian J, Venn A, Otahal P, Gall S. The association between quitting smoking and weight

gain: a systemic review and meta-analysis of prospective cohort studies. Obes Rev 2015;16:883-

901.

11. Chiolero A, Faeh D, Paccaud F, Cornuz J. Consequences of smoking for body weight,

body fat distribution, and insulin resistance. Am J Clin Nutr 2008;87:801-9.

12. Williamson DF, Madans J, Anda RF, Kleinman JC, Giovino GA, Byers T. Smoking

cessation and severity of weight gain in a national cohort. N Engl J Med 1991;324:739-45.

13. Pischon T, Boeing H, Hoffmann K, et al. General and abdominal adiposity and risk of

death in Europe. N Engl J Med 2008;359:2105-20.

14. Hsiao CJ, Hing E. Use and characteristics of electronic health record systems among

office-based physician practices: United States, 2001-2013. NCHS Data Brief 2014:1-8.

15. Song M, Hu FB, Spiegelman D, et al. Adulthood Weight Change and Risk of Colorectal

Cancer in the Nurses' Health Study and Health Professionals Follow-up Study. Cancer Prev Res

(Phila) 2015;8:620-7.

Elisabeth Loder, MD, MPH BMJ Editorial Team Dear Prof. Loder, · 2016. 5. 5. · 1 Elisabeth Loder, MD, MPH BMJ Editorial Team [email protected] Dear Prof. Loder, We appreciate the comments

Documents