Dataset: Menopause

Analyses By: Sean Zion

Date: 2018-01-03

Summary:

This is a preliminary analyses of data set collected by Geers & Colleagues. The data comes from a survey in which women were asked questions about their feelings about menopause. This analysis is primarily interested in the mindsets of women about menopause.

Target outcomes:

We want to look at the following: - Descriptive statistics for the individual mindsets across all women and stratified by menopause status. - Relationship between mindsets about menopause and willingness to pursue different treatments

m <- read.csv("menopause_111517.csv")

Descriptive Statistics Across Mindsets

Individual Mindsets

Body Capable

describe(body_capable_avg)
##    vars   n mean   sd median trimmed  mad min max range  skew kurtosis
## X1    1 365 3.67 0.94   3.67    3.73 0.99   1   5     4 -0.43    -0.07
##      se
## X1 0.05

Body Adversary

describe(body_adversary_avg)
##    vars   n mean   sd median trimmed  mad min  max range skew kurtosis
## X1    1 364 1.65 0.63   1.67    1.56 0.49   1 4.33  3.33 1.08     1.04
##      se
## X1 0.03

Menopause Manageable

describe(men_manage_avg)
##    vars   n mean   sd median trimmed  mad min max range  skew kurtosis
## X1    1 371  4.3 0.84   4.33    4.44 0.99   1   5     4 -1.77     4.14
##      se
## X1 0.04

Menopause Catastrophe

describe(men_cat_avg)
##    vars   n mean   sd median trimmed  mad min max range skew kurtosis   se
## X1    1 361 2.22 0.99      2    2.12 0.99   1   5     4 0.72     0.16 0.05

Menopause Opportunity

describe(men_opp_avg)
##    vars   n mean   sd median trimmed  mad min max range  skew kurtosis
## X1    1 359 3.06 0.93      3    3.07 0.99   1   5     4 -0.18     0.03
##      se
## X1 0.05

Overall Mindsets

Capability of the Body

describe(capability_avg)
##    vars   n mean   sd median trimmed  mad min max range  skew kurtosis
## X1    1 357 4.01 0.64      4    4.04 0.74   2   5     3 -0.33    -0.32
##      se
## X1 0.03

Manageability of Menopause

describe(manageability_avg)
##    vars   n mean   sd median trimmed  mad  min max range  skew kurtosis
## X1    1 360 4.04 0.71      4    4.08 0.74 1.33   5  3.67 -0.52     -0.1
##      se
## X1 0.04

Mindset Means by Menopause Status

Mean Agreement with Each Individual Mindset Across Menopause Status

ggplot(data=means, aes(x=Status, y=mean, fill=variable, width=0.75)) + geom_bar(stat="identity", width=1.0, position=position_dodge()) + labs(x="Status", y="Mean Mindset") + scale_fill_manual(values=c("#FF9999","#CC3333","#FF6666","#99CCFF","#0066CC"))

Mean Agreement with Overall Mindsets Across Menopause Status

ggplot(data=means2, aes(x=Status, y=mean, fill=variable, width=0.75)) + geom_bar(stat="identity", width=1.0, position=position_dodge()) + labs(x="Status", y="Mean Mindset") + scale_fill_manual(values=c("#CC3333","#0066CC"))

Number of Subjects Endorsing Each Category of Menopause Status in Total Sample

ggplot(data=m2) +
  geom_bar(mapping = aes(x = Menopause_status)) +
  labs(title="Distribution of Menopause Status in Sample", x="Status", y="Number of Subjects")

Correlation Matrices - All Women

Correlation Matrix of Mindsets and Outcomes in ALL Women

all_mindset_outcome_corr2 <- cor(all_mindset2, outcomes2)
corrplot(all_mindset_outcome_corr2, title="Correlation Matrix of Mindsets and Outcomes in All Women", method="color", type="full", addCoef.col="black", addgrid.col="black", tl.srt=45, tl.col="midnightblue", bg="gray94", col = colorRampPalette(c("blue4", "white", "red3"))(100))

Correlation Matrix of Mindsets and Treatment Willingness and Confidence in Menopausal Women

all_mindset_outcome_corr3 <- cor(all_mindset2, outcomes3)
corrplot(all_mindset_outcome_corr3, title="Correlation Matrix of Mindsets and Outcomes in All Women", method="color", type="full", addCoef.col="black", addgrid.col="black", tl.srt=45, tl.col="midnightblue", bg="gray94", col = colorRampPalette(c("blue4", "white", "red3"))(100))

Correlations in the total population are relatively weak. This fits with the pattern of results we have been seeing in looking at mixed health samples. Below, we look at the same correlations in women who are currently experiencing menopause symptoms

Correlation Matrices - Menopausal Women

Correlation Matrix of Mindsets and Outcomes in Menopausal Women

mindset_outcome_corr2 <- cor(mmindset, moutcomes2)
corrplot(mindset_outcome_corr2, title="Correlation Matrix of Mindsets and Outcomes in Menopausal Women", method="color", type="full", addCoef.col="black", addgrid.col="black", tl.srt=45, tl.col="midnightblue", bg="gray94", col = colorRampPalette(c("blue4", "white", "red3"))(100))

Correlation Matrix of Mindsets and Treatment Willingness and Confidence in Menopausal Women

mindset_outcome_corr3 <- cor(mmindset, moutcomes3)
corrplot(mindset_outcome_corr3, title="Correlation Matrix of Mindsets and Outcomes in Menopausal Women", method="color", type="full", addCoef.col="black", addgrid.col="black", tl.srt=45, tl.col="midnightblue", bg="gray94", col = colorRampPalette(c("blue4", "white", "red3"))(100))

Again we are seeing the same pattern of results - the mindsets are more strongly correlated with outcomes in the sample for which the mindset is most relevant. In this case, women who are currently experiencing menopause symptoms.

Some interesting findings:

  • Holding more adaptive mindsets about the body and menopause is generally positively correlated with confidence in non-pharmacological menopause interventions, like exercise, diet, and herbal treatments, though not for the willingness to use these treatments.
  • Similarly, these mindsets are negative correlated with willingness to use HRT, but are weakly positively correlated with the confidence in HRT.
  • Women with maladaptive mindsets (Catastrophic and Adversary) tend to be willing to try dietary changed but are not confident in the ability of any treatments to work.
  • Women with the mindset that menopause is catastrophic endorse fewer positive emotions (i.e., calm, peaceful, relaxed) and more negative emotions (i.e., nervousness, fear, sad, and annoyed) when thinking about menopause. -Women who believe in the capability of their bodies are more likely to endorse positive emotions (i.e., calm, peaceful, relaxed) and less likely to endorse negative emotions (i.e.m fear, embarrassed, dad, and annoyed).

Stepwise Regression Models

First, we need to create summary scores for some of the outcomes of interest. To do this we will calculate a summary score for health, consisting of three questions assessing general health, health compared to 1 year previously, and health compared to others. The alpha score for this composite is 0.78 suggesting a composite score is appropriate for these three questions.

We will do the same for the questions measuring positive affect and negative affect from the PANAS.

Next, we want to create composite scores for the three broad interventions: Hormone Replacement Therapy (HRT), Behavioral Interventions (Diet & Exercise), and Alternative Medicine Approaches (Herbal Treatments, Acupuncture, & Meditation). For each of these three interventions, we will create a ‘willingness to use’ and a ‘confidence’ composite.

Next, we can run a series of regression models. These models will first look at whether or not menopause status predicts the outcome of interest. They will then add in the key mindsets to see if the mindsets improve the predictive ability of the model.

Overall Health

sjt.lm(mod1, mod2, mod3, group.pred=FALSE, p.numeric = TRUE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    health_composite   health_composite   health_composite
    B p   B p   B p
(Intercept)   3.66 <.001   2.87 <.001   2.70 <.001
status_yn   -0.10 .440   -0.07 .559   -0.10 .433
capability     0.19 .007  
manageability       0.24 <.001
Observations   264   264   264
R2 / adj. R2   .002 / -.002   .030 / .023   .052 / .045

Looking across the total sample, we created a series of step wise regression models. Model 1, a limited model looking at whether menopause status (Yes = menopausal symptoms; No = pre or post menopausal) predicts a composite score measuring self-reported general health. We found that menopause status (whether or not a subject is currently experiencing menopause symptoms) is not predictive of general health.

Both the Mindset that the Body is Capable and that Menopause is Manageable are included, separately, in Models 2 & 3. Both mindsets are significantly related to overall health, above and beyond menopause status.

Overall Health (including Age as a covariate)

sjt.lm(moda, modb, modc, group.pred=FALSE, p.numeric = TRUE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    health_composite   health_composite   health_composite
    B p   B p   B p
(Intercept)   3.29 <.001   2.26 <.001   2.45 <.001
Age   0.01 .005   0.01 <.001   0.01 .013
status_yn   -0.14 .260   -0.12 .328   -0.13 .272
capability     0.24 <.001  
manageability       0.22 <.001
Observations   264   264   264
R2 / adj. R2   .032 / .024   .072 / .061   .074 / .064

Because both menopause status and overall health are related to age, we modified the step wise regression models above by adding in Age as a co-variate.

Age is significantly related to the health composite score, a measurement of overall health (Model 1)

When Body Capability and Menopause Manageability are included in separate step wise regression models (Models 2 and 3), both are significantly related to overall health, above and beyond either Age (which remains significant), and menopause status (which remains insignificant)

Positive Affect

sjt.lm(mod1b, mod2b, mod3b, mod4b, group.pred=FALSE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    pos_affect   pos_affect   pos_affect   pos_affect
    B p   B p   B p   B p
(Intercept)   2.59 <.001   2.31 <.001   1.77 <.001   1.51 <.001
status_yn   -0.12 .430   -0.15 .319   -0.10 .529   -0.12 .424
Age     0.01 .093    
capability       0.20 .023  
manageability         0.26 <.001
Observations   264   264   264   264
R2 / adj. R2   .002 / -.001   .013 / .006   .022 / .015   .044 / .036

Neither menopause status nor age have a significant relationship with positive affect (Mod 1 & 2).

However, Capability of the body is significantly positively related to positive affect (Mod 3). Manageability is significantly positively related to positive affect (Mod 4).

Negative Affect

sjt.lm(mod1c, mod2c, mod3c, mod4c, group.pred=FALSE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    neg_affect   neg_affect   neg_affect   neg_affect
    B p   B p   B p   B p
(Intercept)   1.92 <.001   2.24 <.001   4.00 <.001   4.17 <.001
status_yn   0.18 .216   0.21 .132   0.18 .188   0.20 .122
Age     -0.01 .031   -0.01 .002   -0.00 .109
capability       -0.40 <.001  
manageability         -0.50 <.001
Observations   264   264   264   264
R2 / adj. R2   .006 / .002   .024 / .016   .113 / .103   .196 / .186

Menopause status is unrelated to negative affect (Mod 1), however younger age is significantly related to negative affect (Mod 2). Keeping age as a co-variate, capability of the body (Mod 3) and manageability of menopause (Mod 4) are both significantly negatively predictive of negative affect.

Confidence in HRT

sjt.lm(mod1d, mod2d, mod3d, group.pred=FALSE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    confidence_hrt   confidence_hrt   confidence_hrt
    B p   B p   B p
(Intercept)   3.17 <.001   2.75 <.001   2.41 <.001
status_yn   0.27 .187   0.30 .154   0.26 .198
capability     0.10 .541  
manageability       0.19 .214
Observations   98   98   98
R2 / adj. R2   .018 / .008   .022 / .001   .034 / .014

Neither menopause status (Mod 1), Capability of the Body (Mod 2), nor Manageability of Menopause (Mod 3) were significantly related to confidence in hormone replacement therapy (HRT).

Willingness to use HRT

sjt.lm(mod1d2, mod2d2, mod3d2, group.pred=FALSE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    willing_hrt   willing_hrt   willing_hrt
    B p   B p   B p
(Intercept)   2.92 <.001   5.38 <.001   3.36 <.001
status_yn   -0.01 .971   -0.19 .487   -0.01 .983
capability     -0.60 .009  
manageability       -0.11 .605
Observations   98   98   98
R2 / adj. R2   .000 / -.010   .070 / .050   .003 / -.018

Menopause status is not significantly related to willingness to use hormone replacement therapy (HRT). However, Capability of the Body is a significant negative predictor of willingness to use HRT (Mod 2). Manageability of Menopause is not a significant predictor of willingness to use HRT (Mod 3).

Confidence in Behavioral Interventions

sjt.lm(mod1e, mod2e, mod3e, group.pred=FALSE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    confidence_behavior   confidence_behavior   confidence_behavior
    B p   B p   B p
(Intercept)   3.51 <.001   1.86 .004   2.89 <.001
status_yn   -0.01 .967   0.12 .546   -0.01 .943
capability     0.40 .010  
manageability       0.16 .285
Observations   98   98   98
R2 / adj. R2   .000 / -.010   .067 / .047   .012 / -.009

Menopause status is not significantly related to confidence in behavioral interventions for menopause symptoms (Exercise & Diet). Capability of the Body is a significant positive predictor of confidence in behavioral interventions (Mod 2). Manageability of Menopause, however, is not significantly related to confidence in behavioral interventions (Mod 3).

Willingness to use Behavioral Interventions

sjt.lm(mod1e2, mod2e2, mod3e2, group.pred=FALSE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    willing_behavior   willing_behavior   willing_behavior
    B p   B p   B p
(Intercept)   4.20 <.001   3.52 <.001   3.55 <.001
status_yn   -0.14 .441   -0.09 .636   -0.14 .420
capability     0.16 .267  
manageability       0.16 .229
Observations   98   98   98
R2 / adj. R2   .006 / -.004   .019 / -.002   .021 / .001

Neither menopause status (Mod 1), Capability of the Body (Mod 2), nor Manageability of Menopause (Mod 3) are significantly related to willingness to try behavioral interventions for menopause symptoms.

Confidence in Alternative Medicine

sjt.lm(mod1f, mod2f, mod3f, group.pred=FALSE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    confidence_altmed   confidence_altmed   confidence_altmed
    B p   B p   B p
(Intercept)   2.78 <.001   3.02 <.001   3.31 <.001
status_yn   0.26 .136   0.24 .180   0.26 .129
capability     -0.06 .681  
manageability       -0.13 .312
Observations   98   98   98
R2 / adj. R2   .023 / .013   .025 / .004   .034 / .013

Neither menopause status (Mod 1), Capability of the Body (Mod 2), nor Manageability of Menopause (Mod 3) are significantly related to confidence in the efficacy of alternative medicine (herbal treatments, acupuncture, and meditation) for menopause symptoms.

Willingness to try Alternative Medicine

sjt.lm(mod1f2, mod2f2, mod3f2, group.pred=FALSE, emph.p = TRUE, sep.column = TRUE, show.ci=FALSE)
    willing_altmed   willing_altmed   willing_altmed
    B p   B p   B p
(Intercept)   3.19 <.001   3.40 <.001   3.39 <.001
status_yn   0.15 .471   0.13 .536   0.15 .467
capability     -0.05 .754  
manageability       -0.05 .742
Observations   98   98   98
R2 / adj. R2   .005 / -.005   .006 / -.014   .007 / -.014

Neither menopause status (Mod 1), Capability of the Body (Mod 2), nor Manageability of Menopause (Mod 3) are significantly related to willingness to try alternative medicine (herbal treatments, acupuncture, and meditation) for menopause symptoms.

Results

Mindsets about menopause and mindsets about the body are related to important health outcomes in women.

Whether or not a woman is experiencing menopause symptoms seems to be unrelated to self-reported health (B=-0.14; p=0.026), whereas age is significantly related to self-reported health (B=0.01; p=0.005).

Controlling for age and menopause status, holding the mindset that menopause is manageable (B=0.22; p<0.001) and that the body is capable (B=0.24; p<0.001) significantly relate to greater self reported health, as measured by a composite of three health status questions.

These two mindsets are positively related to positive affect, and negatively related with negative affect. Holding the mindset that menopause is manageable (B=0.26; p<0.001) or the mindset that the body is capable (B=0.20; p<0.05) are both predictive of higher positive affect. Similarly, the mindset that menopause is manageable (B=-0.50; p<0.001) and that the body is capable (B=-0.40; p<0.001) are both negatively related to negative affect. These relationships existed when both age and menopause status were controlled for.

Mindsets were also related to certain treatment preferences. Capability of the Body was found to be a significant negative predictor of willingness to use hormone replacement therapy (HRT) (B=-0.60; p<0.01), and a significant positive predictor of confidence in behavioral interventions (i.e., exercise and diet), for menopause symptoms (B=0.40; p<0.05).

Discussion

Mindsets may be important targets for interventions for women experiencing menopause symptoms, and may be related to certain psychological characteristics and treatment preferences. Women who hold the mindset that menopause is manageable and the body is capable have better self reported health. While this is the first time these mindsets have been studied in menopausal women, similar results have been found in patients with chronic illnesses like cancer, cardiovascular disease, diabetes, and chronic pain (Zion, Dweck, & Crum, 2017). It also appears that adaptive mindsets about menopause and the body are related to higher levels of positive affect and lower levels of negative affect, however the causality of this relationship remains uncertain. Finally, mindsets about the body may be an important psychological variable for understanding treatment preferences for managing symptoms of menopause. Women who endorsed the mindset that their body’s are capable show less willingness to use hormone replacement therapy and have higher confidence in behavioral interventions like exercise and diet for managing menopause symptoms.