1 Background

In the 2017 hurricane season, there were 16 extreme weather and climate disaster events, with each individual event resulting in economic costs of over $1 billion and cumulative costs of over $300 billion, a new U.S. annual record.1 Beyond the financial strain created by these storms, the most significant impact lies on loss of life and the short and long-term health consequences of the individuals and communities that are directly affected. Discussions on strains caused by these events often focus on mitigating short-term effects, such as minimizing loss of life, ensuring access to basic needs, and the reestablishment of communications and infrastructure. However, longer-term, disasters have also been known to provoke large-scale migration, growth of pre-existing inequities, psychological trauma, and the exacerbation of chronic conditions.2 These effects are more notable in historically vulnerable populations such as racial and ethnic minorities, persons with disabilities, older adults, and those experiencing poverty.3 Meaning that, even though devastation is widespread following disasters, communities with deeply rooted inequities and pre-existing social concerns are disproportionately affected by catastrophes.

Puerto Rico was devastated in September of 2017 by two powerful hurricanes, resulting in approximately 3,000 deaths attributed to Hurricane Maria, the collapse of the power grid, decimated potable water systems, and long-term public health concerns.4-6 High mortality following Maria has been explained by politicians, researchers, and public figures as the result of a highly vulnerable population, inadequate planning and response from the government, Puerto Rico’s weakened infrastructure, and the funding challenges of the health system prior to the hurricane.7-9 In terms of self-reported health, statistics show strong disparities when comparing Puerto Rico to the rest of the United States. In Puerto Rico, adults report poor or fair general health at a 34 percent rate in 2016, compared to the nationwide average of 18 percent.10 The prevalence of chronic health conditions also demonstrated inequalities, with the prevalence of diseases such as diabetes (15% vs 11%), heart disease (11% vs 7%), and HIV diagnoses per 100,000 (17.1 vs 14.7) all being higher in Puerto Rico even before the hurricane.10

Interruption in healthcare services has been identified in disaster literature as a crucial contributor to mortality and morbidity following catastrophic events.11 Scarcity of healthcare assistance is notably impactful in communities with higher proportion of older adults, given that many of them are living with chronic health conditions and rely on electricity and life-sustaining equipment, as is the case in Puerto Rico.12 Consistent with this, a study published in the New England Journal of Medicine concluded that disruption of medical services and delayed care was the primary cause of sustained mortality in the months following Hurricane Maria.7

Limitations in access to healthcare and the vulnerability of the population are the result of Puerto Rico facing its worst financial crisis in its history, plaguing the population with social conditions that put its population at risk of adverse health outcomes. Puerto Rico’s recession is the result of a $74 billion dollar debt that has grown since the early 2000’s and led government leaders, such as previous governor Alejandro Garcia-Padilla to state that the debt is unpayable.13 In the years since that statement, the United States appointed a Fiscal Control Board in 2016, which was tasked with overseeing Puerto Rico’s finances to ensure that the island’s debt is paid. Policies enacted since the beginning of the economic crisis and the appointment of the Fiscal Control Board have resulted in disinvestment in healthcare, social services, and public education.14-15 Additionally, Puerto Rico’s residents have been severely affected by the ongoing economic recession, with the U.S. Census Bureau reporting a median household income of $19,775 and a poverty rate of over 44 percent.16 This can result in major public health issues, with disaster literature suggesting that person’s experiencing poverty are particularly vulnerable when facing catastrophes as they lack the resources and mobility to effectively prepare and recover from events.17 The effect of poverty on Puerto Rico’s recovery can be significant, considering that a substantial amount of literature has linked poverty to poor health outcomes, adverse social issues, and creates significant barriers for individuals and communities to overcome.18

Furthermore, in Puerto Rico, one of the main results of the economic challenges has been a mass exodus of the population to the United States in search of employment opportunities and economic growth. Most recent statistics from Puerto Rico’s Institute of Statistics show that the population had decreased to 3.3 million in 2017 before the hurricane, 10% lower than the 3.7 million residents when the U.S. Census was conducted in 2010, compared to 5.5 million Puerto Ricans living in the United States.19-20 Depopulation is only expected to intensify, with projections from the U.S. Census Bureau estimating that the population will decrease to less than 3 million by 2025 and about 2 million in 2050.21

Puerto Rico’s unique political, social, and economic issues are prone to create long-term challenges in public health moving forward. This study examines the impact of Hurricane Maria on Public Health in Puerto Rico by using data from the Center for Disease Control and Prevention’s Behavioral Risk Factor and Surveillance System (BRFSS). It builds on existing disaster literature by examining changes in social, behavioral, and health trends before and after a natural disaster event on a state-wide population with great vulnerability and complex social issues. This study also considers how the event affected different subsets of the population, including those that have been found by previous research to be vulnerable when facing these events.

2 Data Source

Data from the CDC’s Behavioral Risk Factor Surveillance system will be analyzed in R and SPSS. Datasets will be downloaded from the 2014, 2015, 2016, 2017, and 2018 BRFSS and will be combined into a single file using R and Excel. The Behavioral Risk Factor Surveillance System is a nation-wide system of health-related telephone surveys that collect state data about U.S. residents regarding their health-related risk behaviors, chronic health conditions, and use of preventative services. The BRFSS was established in 1984 and collects data in all 50 states as well as the District of Columbia and three U.S. territories. This surveillance system completes more than 400,000 adult interviews each year, making it the largest continuously conducted health survey system in the world. By collecting behavioral health risk data at the state and local level, the BRFSS has become a powerful tool for targeting and building health promotion activities.

3 Research Questions

This project aims to understand the effect of Hurricane Maria on health and wellness in Puerto Rico. This will be done by examining longitudinal population-level data from the BRFSS and examining variables that measure physical and mental health, socio-demographic characteristics, health-related behaviors, and potential risk factors. Specifically, this study aims to answer the following research questions:

To what extent has the prevalence of physical and mental health conditions changed in the time following the hurricane as compared to before the hurricane?
To what extent have there been behavioral changes in the time following the hurricane as compared to before the hurricane?
How have behavioral and health consequences affected demographic groups and vulnerable populations in the time following the hurricane as compared to before the hurricane?
What, if any, protective factors have facilitated community resilience among certain subsets of the population

4 Data Management

4.1 Organizing myself

https://docs.google.com/spreadsheets/d/1DIWtF7s-HivligMGUapZUgHjHqinGc1286XG9cBaM6U/edit?usp=sharing

4.2 Dataset creation

BRFSS2014NEW <- BRFSS2014 %>% 
  select('_STATE', 'IMONTH', 'IDAY', 'IYEAR', 'GENHLTH', 'PHYSHLTH', 
         'MENTHLTH', 'POORHLTH', '_AGE_G', '_INCOMG', 'SEX', 'EDUCA', 
         'MARITAL', 'EMPLOY1', '_BMI5CAT', '_BMI5',
         'ADDEPEV2', 'CVDINFR4', 
         'HAVARTH3', 'DIABETE3', 'EXERANY2', 'SMOKDAY2', 'DRNK3GE5',
         'AVEDRNK2', 'DRNKANY5') %>%
  rename(AGE_G ='_AGE_G', INCOMG = '_INCOMG', 
           BMI5CAT = '_BMI5CAT', BMI5 = '_BMI5')
BRFSS2015NEW <- BRFSS2015 %>% 
  select('_STATE',  'INTERVIEW MONTH','INTERVIEW DAY', 'INTERVIEW YEAR',
         'GENERAL HEALTH', 
         'NUMBER OF DAYS PHYSICAL HEALTH NOT GOOD', 
         'NUMBER OF DAYS MENTAL HEALTH NOT GOOD',
         'POOR PHYSICAL OR MENTAL HEALTH', '_AGE_G', '_INCOMG',
         'RESPONDENTS SEX', 'EDUCATION LEVEL',
         'MARITAL STATUS', 'EMPLOYMENT STATUS', '_BMI5CAT', '_BMI5', 
         'ACTIVITY LIMITATION DUE TO HEALTH PROBLE',
         'HEALTH PROBLEMS REQUIRING SPECIAL EQUIPM',
         'EVER TOLD YOU HAD A DEPRESSIVE DISORDER',
         'EVER DIAGNOSED WITH HEART ATTACK', 'TOLD HAVE ARTHRITIS',
         '(EVER TOLD) YOU HAVE DIABETES', 
         'EXERCISE IN PAST 30 DAYS', 'FREQUENCY OF DAYS NOW SMOKING', 
         'BINGE DRINKING', 
         'AVG ALCOHOLIC DRINKS PER DAY IN PAST 30',
         'DRINK ANY ALCOHOLIC BEVERAGES IN PAST 30') %>% 
  rename(IMONTH = 'INTERVIEW MONTH', IDAY = 'INTERVIEW DAY', 
         PHYSHLTH = 'NUMBER OF DAYS PHYSICAL HEALTH NOT GOOD',
         IYEAR = 'INTERVIEW YEAR', GENHLTH = 'GENERAL HEALTH', 
         MENTHLTH = 'NUMBER OF DAYS MENTAL HEALTH NOT GOOD', 
         POORHLTH = 'POOR PHYSICAL OR MENTAL HEALTH',
         SEX = 'RESPONDENTS SEX', 
         EDUCA = 'EDUCATION LEVEL',
         MARITAL = 'MARITAL STATUS', 
         EMPLOY1 = 'EMPLOYMENT STATUS',
         QLACTLM2 = 'ACTIVITY LIMITATION DUE TO HEALTH PROBLE',
         USEEQUIP = 'HEALTH PROBLEMS REQUIRING SPECIAL EQUIPM',
         ADDEPEV2 = 'EVER TOLD YOU HAD A DEPRESSIVE DISORDER',
         CVDINFR4 = 'EVER DIAGNOSED WITH HEART ATTACK',
         HAVARTH3 = 'TOLD HAVE ARTHRITIS',
         DIABETE3 = '(EVER TOLD) YOU HAVE DIABETES',
         EXERANY2 = 'EXERCISE IN PAST 30 DAYS',
         SMOKDAY2 = 'FREQUENCY OF DAYS NOW SMOKING',
         DRNK3GE5 = 'BINGE DRINKING', 
         AVEDRNK2 = 'AVG ALCOHOLIC DRINKS PER DAY IN PAST 30',
         DRNKANY = 'DRINK ANY ALCOHOLIC BEVERAGES IN PAST 30',
         AGE_G ='_AGE_G', INCOMG = '_INCOMG', BMI5CAT = '_BMI5CAT', 
         BMI5 = '_BMI5')
BRFSS2016NEW <- BRFSS2016 %>% 
  select('_STATE', 'IMONTH', 'IDAY', 'IYEAR', 'GENHLTH', 'PHYSHLTH', 
         'MENTHLTH', 'POORHLTH', '_AGE_G', '_INCOMG', 'SEX', 'EDUCA', 
         'MARITAL', 'EMPLOY1', '_BMI5CAT', '_BMI5',
         'ADDEPEV2', 'CVDINFR4', 
         'HAVARTH3', 'DIABETE3', 'EXERANY2', 'SMOKDAY2', 'DRNK3GE5',
         'AVEDRNK2', 'DRNKANY5') %>%
  rename(  AGE_G ='_AGE_G', INCOMG = '_INCOMG', 
           BMI5CAT = '_BMI5CAT', BMI5 = '_BMI5')
BRFSS2017NEW <- BRFSS2017 %>% 
  select('_STATE', 'IMONTH', 'IDAY', 'IYEAR', 'GENHLTH', 'PHYSHLTH', 
         'MENTHLTH', 'POORHLTH', '_AGE_G', '_INCOMG', 'SEX', 'EDUCA', 
         'MARITAL', 'EMPLOY1', '_BMI5CAT', '_BMI5',
         'ADDEPEV2', 'CVDINFR4', 
         'HAVARTH3', 'DIABETE3', 'EXERANY2', 'SMOKDAY2', 'DRNK3GE5',
         'AVEDRNK2', 'DRNKANY5') %>%
  rename(AGE_G ='_AGE_G', INCOMG = '_INCOMG', 
           BMI5CAT = '_BMI5CAT', BMI5 = '_BMI5')

ANALYSISDATA <- bind_rows(BRFSS2014NEW, BRFSS2015NEW, BRFSS2016NEW,BRFSS2017NEW)

4.3 Variable Creation

ANALYSISDATA$IDATE <- paste0(ANALYSISDATA$IMONTH, "/", ANALYSISDATA$IDAY, "/", ANALYSISDATA$IYEAR)
ANALYSISDATA$IDATE <- as.Date(ANALYSISDATA$IDATE, 
                              format = "%m/%d/%Y")
ANALYSISDATA$MARIA <- ymd("2017-09-20")
ANALYSISDATA$POSTHURRICANE <- 0 
ANALYSISDATA$POSTHURRICANE[ANALYSISDATA$IDATE > ANALYSISDATA$MARIA] <- 1

4.4 Deleting Missing Data

ANALYSISDATA$GENHLTH [ANALYSISDATA$GENHLTH ==7]<-NA
ANALYSISDATA$GENHLTH [ANALYSISDATA$GENHLTH ==9]<-NA

ANALYSISDATA$MENTHLTH [ANALYSISDATA$MENTHLTH ==77]<-NA 
ANALYSISDATA$MENTHLTH [ANALYSISDATA$MENTHLTH ==88]<-0 
ANALYSISDATA$MENTHLTH [ANALYSISDATA$MENTHLTH ==99]<-NA 

ANALYSISDATA$PHYSHLTH [ANALYSISDATA$PHYSHLTH ==77]<-NA 
ANALYSISDATA$PHYSHLTH [ANALYSISDATA$PHYSHLTH ==88]<-0 
ANALYSISDATA$PHYSHLTH [ANALYSISDATA$PHYSHLTH ==99]<-NA 

ANALYSISDATA$POORHLTH [ANALYSISDATA$POORHLTH ==77]<-NA 
ANALYSISDATA$POORHLTH [ANALYSISDATA$POORHLTH ==88]<-0 
ANALYSISDATA$POORHLTH [ANALYSISDATA$POORHLTH ==99]<-NA 

ANALYSISDATA$INCOMG [ANALYSISDATA$INCOMG ==9]<-NA

ANALYSISDATA$SEX [ANALYSISDATA$SEX ==9]<- NA

ANALYSISDATA$EDUCA [ANALYSISDATA$EDUCA ==9]<- NA

ANALYSISDATA$MARITAL [ANALYSISDATA$MARITAL ==9]<- NA

ANALYSISDATA$EMPLOY1 [ANALYSISDATA$EMPLOY1 ==9]<- NA

ANALYSISDATA$ADDEPEV2 [ANALYSISDATA$ADDEPEV2 ==7]<- NA
ANALYSISDATA$ADDEPEV2 [ANALYSISDATA$ADDEPEV2 ==9]<- NA

ANALYSISDATA$CVDINFR4 [ANALYSISDATA$CVDINFR4 ==7]<- NA
ANALYSISDATA$CVDINFR4 [ANALYSISDATA$CVDINFR4 ==9]<- NA

ANALYSISDATA$HAVARTH3 [ANALYSISDATA$HAVARTH3 ==7]<- NA
ANALYSISDATA$HAVARTH3 [ANALYSISDATA$HAVARTH3 ==9]<- NA

ANALYSISDATA$DIABETE3 [ANALYSISDATA$DIABETE3 ==7]<- NA
ANALYSISDATA$DIABETE3 [ANALYSISDATA$DIABETE3 ==9]<- NA

ANALYSISDATA$EXERANY2 [ANALYSISDATA$EXERANY2 ==7]<- NA
ANALYSISDATA$EXERANY2 [ANALYSISDATA$EXERANY2 ==9]<- NA

ANALYSISDATA$SMOKDAY2 [ANALYSISDATA$SMOKDAY2 ==7]<- NA
ANALYSISDATA$SMOKDAY2 [ANALYSISDATA$SMOKDAY2 ==9]<- NA

ANALYSISDATA$DRNK3GE5 [ANALYSISDATA$DRNK3GE5 ==77]<- NA
ANALYSISDATA$DRNK3GE5 [ANALYSISDATA$DRNK3GE5 ==88]<- NA
ANALYSISDATA$DRNK3GE5 [ANALYSISDATA$DRNK3GE5 ==99]<- NA

ANALYSISDATA$AVEDRNK2 [ANALYSISDATA$AVEDRNK2 == 77]<- NA
ANALYSISDATA$AVEDRNK2 [ANALYSISDATA$AVEDRNK2 == 99]<- NA

ANALYSISDATA$DRNKANY5 [ANALYSISDATA$DRNKANY5 ==7]<- NA
ANALYSISDATA$DRNKANY5 [ANALYSISDATA$DRNKANY5 ==9]<- NA

5 Initial Data Testing

5.1 Average Self-reported general health

t.test(ANALYSISDATA$GENHLTH~ANALYSISDATA$POSTHURRICANE)

## 
##  Welch Two Sample t-test
## 
## data:  ANALYSISDATA$GENHLTH by ANALYSISDATA$POSTHURRICANE
## t = 3.154, df = 911.94, p-value = 0.001663
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.04783733 0.20544007
## sample estimates:
## mean in group 0 mean in group 1 
##        3.045078        2.918440

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= GENHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average View of Personal Health in General in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Description of Personal General Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))

5.2 Average days of poor mental health

t.test(ANALYSISDATA$MENTHLTH~ANALYSISDATA$POSTHURRICANE)

## 
##  Welch Two Sample t-test
## 
## data:  ANALYSISDATA$MENTHLTH by ANALYSISDATA$POSTHURRICANE
## t = -5.7729, df = 881.81, p-value = 1.079e-08
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -3.071774 -1.513040
## sample estimates:
## mean in group 0 mean in group 1 
##        3.839893        6.132300

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= MENTHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Poor Mental Health Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 with Poor Mental Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))

5.3 Average days of poor physical health

t.test(ANALYSISDATA$PHYSHLTH~ANALYSISDATA$POSTHURRICANE)

## 
##  Welch Two Sample t-test
## 
## data:  ANALYSISDATA$PHYSHLTH by ANALYSISDATA$POSTHURRICANE
## t = -0.4486, df = 905.83, p-value = 0.6538
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.8427703  0.5291785
## sample estimates:
## mean in group 0 mean in group 1 
##        5.321724        5.478520

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= PHYSHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Poor Physical Health Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 with Poor Physical Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))

5.4 Average days that health impeded usual activities

t.test(ANALYSISDATA$POORHLTH~ANALYSISDATA$POSTHURRICANE)

## 
##  Welch Two Sample t-test
## 
## data:  ANALYSISDATA$POORHLTH by ANALYSISDATA$POSTHURRICANE
## t = 1.3541, df = 461.11, p-value = 0.1764
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -0.311451  1.691758
## sample estimates:
## mean in group 0 mean in group 1 
##        6.609394        5.919240

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= POORHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Impeded Usual Activities Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 Impeded Usual Activities")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))

6 Focusing on Mental Health

6.1 Puerto Rico days with poor mental health

ggplot(ANALYSISDATA, 
       aes(x=IDATE, 
           y=MENTHLTH))+ 
 geom_jitter(size = 1/2)+ 
  geom_smooth(color="red")+ 
  geom_vline(xintercept = as.numeric(as.Date("2017-09-20")),  
             linetype = 4,
             color = "red") +
  annotate(geom = "text", 
           x = as.Date("2017-09-20") + 10, 
           y = 15, 
           label = "Hurricane Maria", 
           color = "red",
           angle = 90,
           hjust = 0) +
  labs(title = "Self-reported Days of Poor Mental Health in Puerto Rico Over Time",
       subtitle = "Days in last 30 days with poor mental health",
       x = "Date",
       y = "Number of Days with Poor Mental Health") +
  scale_x_date(date_breaks = "3 months") + 
  theme_minimal() +
  theme(axis.text.x = element_text(size = rel(.75),
                                   angle = 45))

6.2 Puerto Rico days with poor mental health without individual cases

ggplot(ANALYSISDATA, 
       aes(x=IDATE, 
           y=MENTHLTH))+ 
 # geom_jitter(size = 1/2)+ 
  geom_smooth(color="red")+ 
  geom_vline(xintercept = as.numeric(as.Date("2017-09-20")),  
             linetype = 4,
             color = "red") +
  annotate(geom = "text", 
           x = as.Date("2017-09-20") + 10, 
           y = 0, 
           label = "Hurricane Maria", 
           color = "red",
           angle = 90,
           hjust = 0) +
  labs(title = "Self-reported Days of Poor Mental Health in Puerto Rico Over Time",
       subtitle = "Days in last 30 days with poor mental health",
       x = "Date",
       y = "Number of Days with Poor Mental Health") +
  scale_x_date(date_breaks = "3 months") + 
  theme_minimal() +
  theme(axis.text.x = element_text(size = rel(.75),
                                   angle = 45))

6.3 Hurricane mental health impact by sex

ANALYSISDATA <- ANALYSISDATA %>% 
  filter(!is.na(SEX))

ANALYSISDATA$SEX2 <- factor(ANALYSISDATA$SEX, levels = c(1, 2), labels = c("Male", "Female"))

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= MENTHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Poor Mental Health Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 with Poor Mental Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))+facet_wrap(~SEX2)

6.4 Hurricane mental health impact by age

ANALYSISDATA$AGE_G2 <- factor(ANALYSISDATA$AGE_G, 
                              levels = c(1, 2, 3, 4, 5, 6), 
                              labels = c("Age 18 to 24", "Age 25 to 34", 
                                         "Age 35 to 44", "Age 45 to 54",
                                         "Age 55 to 64", "Age 65 or older"))

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= MENTHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Poor Mental Health Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 with Poor Mental Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))+facet_wrap(~AGE_G2)

6.5 Hurricane mental health impact by body mass index

ANALYSISDATA <- ANALYSISDATA %>%
  filter(!is.na(BMI5CAT))

ANALYSISDATA$BMI5CAT2 <- factor(ANALYSISDATA$BMI5CAT,
                                levels = c(1, 2, 3, 4),
                                labels = c("Underweight", "Normal Weight",
                                           "Overweight", "Obese"))

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= MENTHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Poor Mental Health Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 with Poor Mental Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))+facet_wrap(~BMI5CAT2)

6.6 Hurricane Mental Health Impact by Income Level

ANALYSISDATA <- ANALYSISDATA %>%
  filter(!is.na(INCOMG)) 

ANALYSISDATA$INCOMG2 <- factor(ANALYSISDATA$INCOMG,
                                levels = c(1,2,3,4,5),
                                labels = c("Less than $15,000", 
                                           "$15,000 to less than $25,000",
                                           "$25,000 to less than $35,000",
                                           "$35,000 to less than $50,000",
                                           "$50,000 or more"))

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= MENTHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Poor Mental Health Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 with Poor Mental Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))+facet_wrap(~INCOMG2)

prop.table(table(ANALYSISDATA$INCOMG2))

## 
##            Less than $15,000 $15,000 to less than $25,000 
##                   0.45083639                   0.29061025 
## $25,000 to less than $35,000 $35,000 to less than $50,000 
##                   0.10246547                   0.07454683 
##              $50,000 or more 
##                   0.08154106

6.7 Hurricane Mental Health Impact by Relationship Status

ANALYSISDATA <- ANALYSISDATA %>%
  filter(!is.na(MARITAL)) 

ANALYSISDATA$MARITAL2 <- factor(ANALYSISDATA$MARITAL,
                                levels = c(1,2,3,4,5,6),
                                labels = c("Married", "Divorced", "Widowed",
                                           "Separated", "Never Married", 
                                           "Unmarried couple"))

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= MENTHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Poor Mental Health Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 with Poor Mental Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))+facet_wrap(~MARITAL2)

6.8 Hurricane Mental Health Impact by Employment Status

ANALYSISDATA <- ANALYSISDATA %>%
  filter(!is.na(EMPLOY1)) 

ANALYSISDATA$EMPLOY2 <- factor(ANALYSISDATA$EMPLOY1,
                                levels = c(1,2,3,4,5,6,7,8),
                                labels = c("Employed", "Self-Employed",
                                           "Long-term unemployed",
                                           "Short-term unemployed",
                                           "Homemaker", "Student",
                                           "Retired", "Unable to work"))

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= MENTHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Poor Mental Health Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 with Poor Mental Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))+facet_wrap(~EMPLOY2)

6.9 Hurricane Mental Health Impact by Educational Attainment

ANALYSISDATA <- ANALYSISDATA %>%
   filter(!is.na(EDUCA))

ANALYSISDATA$EDUCA2 <- factor(ANALYSISDATA$EDUCA,
                               levels = c(1,2,3,4,5,6),
                               labels = c("Never attended school", 
                                          "Some schooling", 
                                          "Some high school", 
                                          "High school", 
                                          "Some college", 
                                          "College graduate"))

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= MENTHLTH, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average Days of Poor Mental Health Per 30 Days in Puerto Rico ", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average Days in Last 30 with Poor Mental Health")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))+facet_wrap(~EDUCA2)

prop.table(table(ANALYSISDATA$EDUCA2))

## 
## Never attended school        Some schooling      Some high school 
##           0.004331031           0.104412970           0.076788014 
##           High school          Some college      College graduate 
##           0.249268407           0.250380428           0.314819150

7 Behavioral Changes following the Hurricane

7.1 Exercise Patterns

ANALYSISDATA <- ANALYSISDATA %>%
   filter(!is.na(EXERANY2))

ANALYSISDATA$EXERANY3 <- factor(ANALYSISDATA$EXERANY2,
                                levels = c(1,2),
                                labels = c("Yes", "No"))


ggplot(data=ANALYSISDATA)+geom_bar(mapping = aes(x=EXERANY3, y= ..prop.., group=1, color=EXERANY3))+facet_wrap(~POSTHURRICANE)+theme_minimal()+labs(title = "Exercise Behaviors in Puerto Rico ", subtitle = "Any physical activity in last 30 days", x= "Before Maria vs After Maria", y= "Percentage of the population")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))

7.2 Smoking Behaviors

ANALYSISDATA <- ANALYSISDATA %>%
   filter(!is.na(SMOKDAY2))

ANALYSISDATA$SMOKDAY3 <- factor(ANALYSISDATA$SMOKDAY2,
                                levels = c(1,2,3),
                                labels = c("Every day", "Some days",
                                           "Not at all"))

ggplot(data=ANALYSISDATA)+geom_bar(mapping = aes(x=SMOKDAY3, y= ..prop.., group=1, color=SMOKDAY3))+facet_wrap(~POSTHURRICANE)+theme_minimal()+labs(title = "Current smoking behaviors ", x= "Before Maria vs After Maria", y= "Percentage of the population")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))

7.3 Alcohol Consumption

ANALYSISDATA <- ANALYSISDATA %>%
   filter(!is.na(DRNKANY5))

ANALYSISDATA$DRNKANY6 <- factor(ANALYSISDATA$DRNKANY5,
                                levels = c(1,2),
                                labels = c("Yes", "No"))

ggplot(data=ANALYSISDATA)+geom_bar(mapping = aes(x=DRNKANY6, y= ..prop.., group=1, color=DRNKANY6))+facet_wrap(~POSTHURRICANE)+theme_minimal()+labs(title = "Current alcohol consumption", subtitle = "At least one drink in past 30 days", x= "Before Maria vs After Maria", y= "Percentage of the population")+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))

t.test(ANALYSISDATA$AVEDRNK2~ANALYSISDATA$POSTHURRICANE)

## 
##  Welch Two Sample t-test
## 
## data:  ANALYSISDATA$AVEDRNK2 by ANALYSISDATA$POSTHURRICANE
## t = 0.072783, df = 94.53, p-value = 0.9421
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -1.794233  1.930791
## sample estimates:
## mean in group 0 mean in group 1 
##        6.231069        6.162791

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= AVEDRNK2, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Average drinks per drinking occassion", subtitle = "Before Maria vs After Maria", x= "Before Maria vs After Maria", y= "Average amount of drinks per drinking occasion in last 30 days")+ theme(text = element_text(size = 10))+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))

ggplot(ANALYSISDATA, aes(x=POSTHURRICANE, y= DRNK3GE5, fill=factor(POSTHURRICANE)))+stat_summary(fun.y = mean, geom = "bar")+theme_minimal()+theme(axis.text.x = element_blank())+labs(title = "Binge Drinking Events", x= "Before Maria vs After Maria", y= "Average amount of binge drinking ocassions in past 30 days")+ theme(text = element_text(size = 10))+ scale_fill_manual(name = "Time Period", values = c("grey93", "red3"), labels = c("before the hurricane", "after the hurricane"))

t.test(ANALYSISDATA$DRNK3GE5~ANALYSISDATA$POSTHURRICANE)

## 
##  Welch Two Sample t-test
## 
## data:  ANALYSISDATA$DRNK3GE5 by ANALYSISDATA$POSTHURRICANE
## t = 0.61152, df = 51.58, p-value = 0.5435
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -1.161098  2.178686
## sample estimates:
## mean in group 0 mean in group 1 
##        4.819905        4.311111

testing

Summer 2019

Mislael A. Valentin Cortes

2019-07-30