Hot Flashes: The Role of Age and Physical Health in Menopausal Symptoms

Author

Lydia B

Statistics Course Project

The Introduction

Dataset Research

A hot flash is a sudden and intense feeling of heat, normally felt in the upper body, especially the face, neck, and chest. The sensation is often paired with sweating, chills, and sometimes anxiety. It is a common symptom of menopause, the transition periods in a woman’s life when she moves from her reproductive years to the post-reproductive stage. Hot flashes typically last from 2 to 5 minutes and could happen several times a day or, for some, less frequently. They are often followed by cold or shivering after the flash.

Defining Technical Terms

BMI - stands for body mass index, if someones BMI is over 30 they are classified as obese. In this data set 0: No, the participant’s BMI is less than 30. and 1: Yes, the participant’s BMI is 30 or greater.

Estra or (Baseline Estradiol)- The level of estradiol, a form of estrogen, measured at baseline. Estradiol is the primary female sex hormone responsible for regulating the menstrual cycle and other reproductive processes.

FSH or (Baseline Follicle-Stimulating Hormone) - Measures follicle-stimulating hormone (FSH), which is involved in regulating the menstrual cycle by stimulating the ovaries to produce eggs.

LH or (Baseline Luteinizing Hormone) - A hormone that plays a role in the menstrual cycle. It triggers ovulation (the release of an egg from the ovary).

Testo or (Baseline Testosterone) - Represents the level of testosterone in the blood. Testosterone, while typically associated with male health, also plays an important role in women, influencing libido, bone health, and muscle mass.

DHEAS or (Baseline Dehydroepiandrosterone Sulfate) - A hormone produced by the adrenal glands and is a precursor to both estrogen and testosterone.

PCS12 - A self-reported outcome measure assessing the impact of health on an individual’s everyday life. It is often used as a quality of life measure.

History

Hot flashes were noted in ancient texts, but not scientifically studied until the 20th century. Then, in 1940s-1950s, estrogen therapy was introduced as a treatment for menopause-related symptoms, including hot flashes. It wasn’t until the 2000s when non-hormonal treatments were introduced such as, SSRIs, gabapentin, and clonidine. Today, research focuses on personalized medicine, genetic factors, and non-hormonal treatments, with new approaches like neurokinin receptor antagonists actively being tested.

Details about Stats, how it was collected, bias?

The data collected was an observational cohort study where the researchers followed a group of people and observed them over time slowly collecting data, researchers did not manipulate or intervene. Recruitment was specifically stratified by race (African American vs. Caucasian) to ensure that both groups were equally, participants were randomly recruited through random digit dialing from the general population of Philadelphia County, PA. This was to ensure a diverse pool of participants.

Only women aged 35-47 who were in the late reproductive years, experiencing regular menstrual cycles, had intact ovaries and uterus, and were English-speaking, were used in the study. And were excluded if they had conditions that might interfere with ovarian function like diabetes or liver disease, were pregnant or breastfeeding, or had used hormonal medications. This was to ensure that hormonal treatments did not alter the results about the occurrence of hot flashes.

Bias

The random digit dialing, while ensure diversity, does not guarantee a completely random sample because it could be biased by factors like access to telephones or willingness to participate in the study

Because subjects were recruited from Philadelphia County, PA only, it may not be representative of other geographic areas or demographic groups like, rural populations or people from different cultural backgrounds.

This study was specific to African American and Caucasian women so it may limit the ability to generalize the results to other racial or ethnic groups like Hispanics or Asians.

The study excludes women with conditions such as diabetes or cancer. This introduces selection bias because the study does not include women with these conditions, even though they may also experience hot flashes.

Since the hot flash data was based on self-reports, there is a potential for recall bias, where participants may underreport or overreport the occurrence of hot flashes.

Overarching question

Is there an association between physical health (PCS12 score) and the occurrence of hot flashes in the late reproductive years?

Your work with the data

load the libraries

library(tidyverse)

── Attaching core tidyverse packages ──────────────────────── tidyverse 2.0.0 ──
✔ dplyr     1.1.4     ✔ readr     2.1.5
✔ forcats   1.0.0     ✔ stringr   1.5.1
✔ ggplot2   3.5.1     ✔ tibble    3.2.1
✔ lubridate 1.9.3     ✔ tidyr     1.3.1
✔ purrr     1.0.2     
── Conflicts ────────────────────────────────────────── tidyverse_conflicts() ──
✖ dplyr::filter() masks stats::filter()
✖ dplyr::lag()    masks stats::lag()
ℹ Use the conflicted package (<http://conflicted.r-lib.org/>) to force all conflicts to become errors

library(tidymodels)

── Attaching packages ────────────────────────────────────── tidymodels 1.2.0 ──
✔ broom        1.0.6     ✔ rsample      1.2.1
✔ dials        1.3.0     ✔ tune         1.2.1
✔ infer        1.0.7     ✔ workflows    1.1.4
✔ modeldata    1.4.0     ✔ workflowsets 1.1.0
✔ parsnip      1.2.1     ✔ yardstick    1.3.1
✔ recipes      1.1.0     
── Conflicts ───────────────────────────────────────── tidymodels_conflicts() ──
✖ scales::discard() masks purrr::discard()
✖ dplyr::filter()   masks stats::filter()
✖ recipes::fixed()  masks stringr::fixed()
✖ dplyr::lag()      masks stats::lag()
✖ yardstick::spec() masks readr::spec()
✖ recipes::step()   masks stats::step()
• Use suppressPackageStartupMessages() to eliminate package startup messages

Load the data

getwd()

[1] "C:/Users/lydia/Downloads/Stats 217/stats final project"

setwd("C:/Users/lydia/Downloads/Stats 217/stats final project")
hot_flash <- read_csv("hflash.csv")

Rows: 375 Columns: 14
── Column specification ────────────────────────────────────────────────────────
Delimiter: ","
dbl (14): pt, ageg, aagrp, edu, d1, f1a, pcs12, hotflash, bmi30, estra, fsh,...

ℹ Use `spec()` to retrieve the full column specification for this data.
ℹ Specify the column types or set `show_col_types = FALSE` to quiet this message.

head(hot_flash)

# A tibble: 6 × 14
     pt  ageg aagrp   edu    d1   f1a pcs12 hotflash bmi30 estra   fsh    lh
  <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>    <dbl> <dbl> <dbl> <dbl> <dbl>
1     3     2     0     1     0     0  56.8        0     0 107.   3.00  2.98
2     6     3     0     1     0     0  59.2        0     0  31.2 11.2   5.76
3     7     1     0     1     0     1  57.7        0     0  13.4 14.5   5.60
4     8     1     0     1     1     0  55.8        1     0  10.6  5.53  2.26
5     9     2     0     1     0     1  55.9        0     0  24.1  9.78  2.6 
6    10     3     0     1     0     0  NA          1     0  37.3 10.3   3.40
# ℹ 2 more variables: testo <dbl>, dheas <dbl>

Boxplot for distribution of pcs12 scores across age groups

hot_flash2 <- hot_flash |>
  mutate(age_type = case_when(ageg == 1 ~ "35-39 years",
                              ageg == 2 ~ "40-44 years",
                               ageg == 3 ~ "45-48 years"))

ggplot(hot_flash2, aes(age_type, pcs12, fill = age_type)) +
         geom_boxplot()

Warning: Removed 13 rows containing non-finite outside the scale range
(`stat_boxplot()`).

These box plots illustrate the distribution of pcs12 scores across three age groups: 35-39 years, 40-44 years, and 45-48 years. I am exploring pcs12 scores and age to find out if at a certain age pcs scores go up or down. This graph is showing the median pcs12 scores for the 35-39 age group is the highest among the three groups while the median level is lowest in the 45-48 age group. The plots are suggesting a trend of decreasing pcs12 scores with age.

PCS12 (Physical Component Scale of the SF12)

ggplot(hot_flash2, aes(pcs12)) +
geom_density()

Warning: Removed 13 rows containing non-finite outside the scale range
(`stat_density()`).

The density plot of PCS-12 shows a wide range of physical health scores, with a peak around 50 implying the average health is around there. The left skew implies that more people have a higher physical health score rather than a lower score.

ba for proportion of BMI

hot_flash1 <- hot_flash |>
  mutate(bmi30 = ifelse(bmi30 == 0, "<30", ">30"))

hot_flash3 <- hot_flash1 |>
  count(bmi30)|>
  mutate(prop=n/sum(n))
head(hot_flash3)

# A tibble: 2 × 3
  bmi30     n  prop
  <chr> <int> <dbl>
1 <30     236 0.629
2 >30     139 0.371

ggplot(hot_flash3, aes(x = bmi30, y = prop, fill = bmi30)) +
  geom_col()

description of bmi30 (body mass index) bargraph

This bar graph shows the proportion of people who have a bmi over and under 30. The plot shows that a larger proportion of individuals in the data set have a BMI of 30 or higher.

summary statistics

for anova test

summarytable <- hot_flash2 |>
 group_by(hotflash) |>
 summarise(mean_delay = mean(pcs12), sd_delay = sd(hotflash), median_delay = median
(hotflash), count = n())
summarytable

# A tibble: 2 × 5
  hotflash mean_delay sd_delay median_delay count
     <dbl>      <dbl>    <dbl>        <dbl> <int>
1        0         NA        0            0   257
2        1         NA        0            1   118

for t-test

hot_flash2 |>
 group_by(hotflash) |>
 count()

# A tibble: 2 × 2
# Groups:   hotflash [2]
  hotflash     n
     <dbl> <int>
1        0   257
2        1   118

3 DIFFERENT statistical techniques

ANOVA

hot_flash2$ageg<- as.factor(hot_flash2$ageg)

aov_pcs12_ageg <- aov(pcs12 ~ ageg, data = hot_flash2)
summary(aov_pcs12_ageg)

             Df Sum Sq Mean Sq F value Pr(>F)
ageg          2     14    6.97   0.098  0.907
Residuals   359  25482   70.98               
13 observations deleted due to missingness

# Tidy the model
tidy(aov_pcs12_ageg)

# A tibble: 2 × 6
  term         df   sumsq meansq statistic p.value
  <chr>     <dbl>   <dbl>  <dbl>     <dbl>   <dbl>
1 ageg          2    13.9   6.97    0.0982   0.907
2 Residuals   359 25482.   71.0    NA       NA

t-test

unique(hot_flash2$hotflash)

[1] 0 1

hot_flash2$hotflash <- as.factor(hot_flash2$hotflash)

Generate 1000 reps.

null_dist <- hot_flash2 |>
 specify(pcs12 ~ hotflash) |>
 hypothesize(null = "independence") |>
 generate(reps = 1000, type = "permute") |>
 calculate(stat = "diff in means", order = c("0", "1"))

Warning: Removed 13 rows containing missing values.

head(hot_flash2)

# A tibble: 6 × 15
     pt ageg  aagrp   edu    d1   f1a pcs12 hotflash bmi30 estra   fsh    lh
  <dbl> <fct> <dbl> <dbl> <dbl> <dbl> <dbl> <fct>    <dbl> <dbl> <dbl> <dbl>
1     3 2         0     1     0     0  56.8 0            0 107.   3.00  2.98
2     6 3         0     1     0     0  59.2 0            0  31.2 11.2   5.76
3     7 1         0     1     0     1  57.7 0            0  13.4 14.5   5.60
4     8 1         0     1     1     0  55.8 1            0  10.6  5.53  2.26
5     9 2         0     1     0     1  55.9 0            0  24.1  9.78  2.6 
6    10 3         0     1     0     0  NA   1            0  37.3 10.3   3.40
# ℹ 3 more variables: testo <dbl>, dheas <dbl>, age_type <chr>

observed difference in means

obs_dist <- hot_flash2 |>
 # Specify the response and explanatory variables
 specify(pcs12 ~ hotflash) |> # syntax is y ~ x
 calculate(stat = "diff in means", order = c("0", "1"))

Warning: Removed 13 rows containing missing values.

obs_dist

Response: pcs12 (numeric)
Explanatory: hotflash (factor)
# A tibble: 1 × 1
   stat
  <dbl>
1  3.20

calculate p value

pvalue <- get_p_value(null_dist, obs_dist, direction = "two-sided")
pvalue

# A tibble: 1 × 1
  p_value
    <dbl>
1   0.002

Chi-squared test for age and hotflash

chisq.test(hot_flash2$ageg, hot_flash2$hotflash)


    Pearson's Chi-squared test

data:  hot_flash2$ageg and hot_flash2$hotflash
X-squared = 0.73582, df = 2, p-value = 0.6922

Conclusion

ANOVA

The ANOVA revealed that the P-value of .907 is not significant so therefore there is no significant evidence that there is a difference in the physical health scores between the age groups, 35-39 years, 40-44 years, and 45-48 years.

t test conclusion

The permutation test displayed a p-value of 0.002, which implies a statistically significant difference in PCS12 scores between women who experience hot flashes and those who do not.

Chi squared test age and hotflash conclusion

The Chi-squared test for yielded a p-value of .6922 , which displays that there is no significant association between the two variables. We fail to reject the null. There is no significant evidence to suggest that there is an association between the age groups and the occurrence of hot flashes in the late reproductive years.

Useful to the general public

ANOVA

The results show that age may not significantly affect physical health (PCS12 scores) during menopause, which helps redirect focus to other potential factors, such as BMI, hormone levels, or lifestyle. This information can guide healthcare providers in allocating resources effectively and encourage women to seek support for menopause symptoms. It also highlights the need for future research into other influences on physical health during menopause, ultimately improving public understanding and care.

T-test

The t-test and permutation test results suggest a significant difference in PCS12 scores between women who experience hot flashes and those who do not, further reinforcing the idea that hot flashes may be linked to poorer physical health.

Chi Squared Test Age and Hotflash

If you’re between 35 and 48 years old and experiencing hot flashes, don’t assume it’s because of your age. This analysis shows that age alone is not a significant factor, so it might be helpful to consider other possible causes or speak to a healthcare professional for further investigation.

In conclusion, while some of my data points suggest a potential relationship between physical health (PCS12 score) and the occurrence of hot flashes in the late reproductive years, most of the analysis indicates no significant association. Although the boxplot suggests that physical health tends to decrease with age, the ANOVA test indicates that age may not significantly affect physical health. Furthermore, despite the t-test showing significant differences in PCS scores between women who experience hot flashes and those who do not, the chi-squared tests on the association between hot flashes and PCS scores, as well as hot flashes and age, both showed no significant results. While these findings could be seen as inconclusive, overall, the data leans toward no significant association between physical health (as measured by the PCS12 score) and the occurrence of hot flashes in the late reproductive years.

My Opinion

This data confused me a lot going into this project I was almost sure that there would be a difference in between physical health (PCS12 score) and the occurrence of hot flashes in the late reproductive years but it is looking like there isnt. Especially at the end with all of the tests when I started getting all different outputs pointing to different things I got really confused. Interesting outcome I definitely wonder if this data is accurate though.

Bibliography

“Hot Flashes – Teaching of Statistics in the Health Sciences.” Causeweb.org, 2023, causeweb.org/tshs/hot-flashes/. Accessed 26 Nov. 2024.

Mayo Clinic. “Hot Flashes - Symptoms and Causes.” Mayo Clinic, 12 Dec. 2023, www.mayoclinic.org/diseases-conditions/hot-flashes/symptoms-causes/syc-20352790.

National Institute on Aging. “What Is Menopause?” National Institute on Aging, 30 Sept. 2021, www.nia.nih.gov/health/menopause/what-menopause.

“Hot Flashes: What Can I Do?” National Institute on Aging, 30 Sept. 2021, www.nia.nih.gov/health/menopause/hot-flashes-what-can-i-do.