30
Between- and Within-Religion Differences in Hypertension Risks for Indonesian Women Natalie Jansen* University of Kansas Hypertensive status, or high blood pressure, is an indicator of future cardiovascular disease (CVD), heart attack, stroke, kidney disease, and premature mortality and disability. Using data from Wave 5 of the Indonesian Family Life Survey, I explore religious differences in women’s hypertensive risks. I find that Muslim women experience higher hypertension rates than Christian and Hindu women. There are no within-religion differences in hypertension rates for Muslim women when accounting for participation in daily salat prayer or when accounting for adherence to different traditions. I also find that Hindu women have lower hypertension risks than Christian and Muslim women, regardless of whether or not they eat beef. Finally, I find that Hindu women who do yoga and/or meditate daily have the lowest risks of hypertension compared to Muslims, Christians, and Hindu women who infrequently do yoga/meditate. * Corresponding author: 1415 Jayhawk Boulevard, Fraser 716, Lawrence, KS, 66045; [email protected]

Between- and Within-Religion Differences in Hypertension

  • Upload
    others

  • View
    3

  • Download
    0

Embed Size (px)

Citation preview

Between- and Within-Religion Differences in Hypertension Risks for Indonesian Women

Natalie Jansen*

University of Kansas

Hypertensive status, or high blood pressure, is an indicator of future cardiovascular disease

(CVD), heart attack, stroke, kidney disease, and premature mortality and disability. Using data

from Wave 5 of the Indonesian Family Life Survey, I explore religious differences in women’s

hypertensive risks. I find that Muslim women experience higher hypertension rates than

Christian and Hindu women. There are no within-religion differences in hypertension rates for

Muslim women when accounting for participation in daily salat prayer or when accounting for

adherence to different traditions. I also find that Hindu women have lower hypertension risks

than Christian and Muslim women, regardless of whether or not they eat beef. Finally, I find that

Hindu women who do yoga and/or meditate daily have the lowest risks of hypertension

compared to Muslims, Christians, and Hindu women who infrequently do yoga/meditate.

* Corresponding author: 1415 Jayhawk Boulevard, Fraser 716, Lawrence, KS, 66045;

[email protected]

Introduction

Hypertensive status, or high blood pressure, is an indicator of future cardiovascular

disease (CVD), heart attack, stroke, kidney disease, and premature mortality and disability (He &

Whelton, 1999; WHO, 2016). Furthermore, hypertension is responsible for at least 45 percent of

deaths as a result of heart disease and 51 percent of deaths caused by stroke globally (WHO,

2013). Past research demonstrates that hypertension and hypertension-related complications are

largely preventable, and making healthy lifestyle adjustments can lower blood pressure (Joffres

et al., 2013).

Concerted public health efforts are being made to alter individual behaviors largely by

striving to lower rates of alcohol consumption and tobacco intake– both previously-demonstrated

risk factors for hypertension (e.g., He & Whelton, 1999; Kayima et al., 2015; Khanam et al.,

2015; Abdul-Razak et al., 2016; Ibrahim, Hijazi, & Al-Bar, 2008). However, in Indonesia – a

low-middle income, Muslim-majority nation – fewer than three percent of women report tobacco

use (DHS, 2012), and fewer than five percent of women report drinking alcohol in the last 12

months (WHO, 2014). Despite the fact that only a small proportion of women in Indonesia

consume alcohol and/or tobacco products, rates of hypertension remain higher than the average

for women in Southeast Asia at 29.3 percent (WHO, 2016).

Low rates of tobacco use are attributed in part to cultural values which stigmatize women

as morally flawed (Barraclough, 1999) along with Islamic fatwas that encourage healthy living,

while low rates of alcohol consumption are attributed both to cultural stigmatization as well as

Islamic doctrine that prohibits alcohol intake. In addition to encouraging healthy living and

discouraging alcohol consumption, religion, and particularly Islam, plays a critical role in

women’s lives in Indonesia; Indonesian citizens must legally identify with a monotheistic

religion, although the overwhelming majority of Indonesians identify as Muslim.

Given the potential role religion has to provide structure to women’s daily lives, offer

social and spiritual support, and encourage healthy behaviors, understanding the role religion

plays on hypertension rates could offer additional insight into disparities in hypertension rates.

Religious involvement as measured by prayer participation, adherence to tradition, and selection

into different faiths, particularly in non-Christian-majority countries, is largely overlooked in

hypertension literature. In this study, I use Wave 5 of the Indonesian Family Life Survey to

examine differing risks in hypertensive status across and within religions.

The Role of Religion on Hypertensive Status

Many studies have found that religious involvement is associated with lower risks of

hypertension. The link between religion and hypertension has been attributed to a variety of

factors including encouragement toward healthy behaviors such as avoiding tobacco, exercising

regularly, and adhering to a particular diet (Koenig, 2012; Powell, Shahabi, & Thoresen, 2003),

strengthening social ties and fostering supportive environments (Krause, 2008; Bradley, 1995;

Holt et al., 2015) and providing a sense of security and self-control (Rounding, Lee, Jacobson, &

Ji, 2012). Koenig (2012) examined 63 studies and found that 36 reported lower blood pressures

for more religious individuals while seven reported significantly higher BP for more religious

individuals. Several studies have also found that religious attendance specifically is associated

with lower blood pressure (Banerjee, Boyle, Anand, Strachan, & Oremus, 2014; Bell, Bowie, &

Thorpe, 2012; Gillum & Ingram, 2006). Conversely, Buck, Williams, Musick, and Sternthal

(2009) found that attendance and public participation were not significantly related to

hypertensive status in a sample of adults from Chicago. Regarding the role of prayer, Buck et al.

also found that regular prayer was associated with a higher likelihood of hypertension. Similarly

Fitchett and Powell (2009) found in a study of middle-aged Protestant and Catholic American

women that daily spiritual experiences were not protective against hypertension.

Most research exploring hypertensive status relies on data from Christian and Catholic

respondents and does not explore between-religion variations in blood pressure. However,

research from rural India suggests that religious affiliation may differentially affect hypertensive

status; individuals belonging to minority religions have higher risks of hypertension compared to

Hindu individuals (Banerjee, Mukherjee, & Basu, 2016). In a country such as Indonesia, where

about 87 percent of the population practices Islam, understanding both the role that Islamic

practices play, as well as how Islam – as a majority religion – measures up to Christianity and

Hinduism, is an important next step in understanding the mechanisms through which religion

may affect hypertensive status.

Religions and Coexistence in Indonesia

In Indonesia, all citizens are declared religious through the Indonesian Constitution

(Gudorf, 2012). However, only six religions are legally recognized: Islam, Buddhism, Hinduism,

Catholicism, Protestantism, and Confucianism (An-Na-Im, 2008). Indonesia largely manages

religions through state administration of religious practices such as mandates to marry within an

individual’s religion, regulations on constructing houses of worship and performing mission

work, and requirements to claim a religion in order to acquire legal documents, obtain state-

administered identification, and travel outside of the country (Seo, 2012). Some argue that these

policies are discriminatory measures against minority religions in the country, particularly

Christianity (Jones, 2005) – due in part to hostility stemming from attempted evangelizing and

conversion (Hoon, 2013). Furthermore, Christianity is a product of Dutch colonialism, and the

legacy of Dutch origins has left Christians better-educated on average than Muslims and with

disproportionate control of the Indonesian economy (Jones, 2005), despite only comprising about

10 percent of the population (Hoon, 2016).

On the other hand, Hinduism and Islam coexist in several domains in Indonesia including

art, festivals, and religious gatherings and do not exhibit the same tensions as Christianity

(Suprato, 2016). In adhering to the legal recognition of monotheistic religions, many modern

Indonesian Hindus have adapted their practices to reflect a single Almighty God surrounded by

ancient gods who have been demoted to angels (McDaniel, 2010). Furthermore, the rich,

longstanding history of Hinduism in Indonesia dates back to the sixth century and has heavily

influenced Indonesian culture (Pringle, 2010).

As a result of the complex and hostile relationship between Christianity and Islam, as

well as the unique and coexistent relationship between Hinduism and Islam, and contrary to past

research demonstrating that minority religions experience higher rates of hypertension, I

hypothesize that:

1A. Christians will experience lower risks of hypertension than Muslim respondents.

1B. Hindu women will not experience significantly different risks of hypertension

compared to Muslim women.

Muslim Traditions in Indonesia

Indonesian Muslims primarily belong to one of two organizations: the Nahdlatul Ulama

(NU) and Muhammadiyah. NU is the largest traditionalist organization with 40 million members,

and Muhammadiyah is the leading modernist organization with 29 million members. Both

organizations are national and provide thousands of universities, schools, and medical facilities

(Pringle, 2010). Muhammadiyah is more stringently organized and is influenced heavily by

puritanical Middle Eastern trends (Pringle, 2010). Conversely, while NU is more accommodating

of local cultural norms within its definition of Islam (Pringle, 2010).

Given that members of NU experience may fewer internal conflicts between their

adherence to Islam and their cultural Indonesian traditions, I hypothesize that:

2. Women from the Nahdlatul Ulama tradition will experience lower risks of

hypertension than Muhammidiyah women, Muslim women without a tradition,

Christian women, and Hindu women.

Salat Prayer and Islam

Among the many religious practices performed by Muslims in Indonesia, one of the most

important is salat – the call to pray five times per day (Henkel, 2005). Salat includes recitations

and fixed sequences of physical movements (standing, prostrating, kneeling, sitting) in the

direction of Mecca. Most Muslims agree that regardless of one’s one level of conformity to

Islam, salat participation is required (Bowen, 1989). Past research has demonstrated that salat

movements can improve physical fitness (Ibrahim, Sian, Shanggar, & Razack, 2013; Reza,

Urakami, & Mano, 2001; Alabdulwahab, Kachanathu, & Oluseye, 2013), psychological well-

being due to its meditative nature (Doufesh et al., 2012), cognitive functioning (Bai, Ye, Zhu,

Zhao, & Zhang, 2012), and cardiac blood flow (Doufesh et al., 2013). Prayer enhances cardiac

autonomic control and increases bararoflex sensitivity (Bernardi, Sleight, Bandinelli, & Cencetti,

2001; Berntson, Norman, Hawkley, & Cacioppo, 2008). However, on the other hand, prayer

participation has been hypothesized to reflect mobilization of spiritual resources as a reaction to

poor health (Salmoirago-Blotcher et al., 2013). Given the meditative and physical nature of salat

prayer as well as its daily importance to Islam itself, I hypothesize that:

3. Muslim women who adhere to salat (by praying at least five times per day) will

have lower risks of hypertension than Muslim women who pray 0 times or 1-4

times per day and Christian women and Hindu women.

Setting

Indonesia is an archipelago nation located in Southeast Asia and is comprised of more

than 17,500 islands, 300 different ethnicities, and 580 different languages and dialects. Indonesia

is the fourth most populated country in the world following China, India, and the United States

with a population of approximately 255 million, although approximately 130 million of its

inhabitants reside on the Javanese cluster of islands (Indonesian Embassy, 2016). Despite ethnic

heterogeneity, 87 percent of Indonesia’s population identifies as Muslim.

Indonesia has long played an important role in global trade with a rich history dating

back to the 1500s when it was colonized by the Portuguese for its spices. Although the

Portuguese were largely unsuccessful, the Dutch established the Dutch East India Company in

1602 and maintained rule in Indonesia until the 1940s. Both the Dutch and Portuguese presence

heavily influenced the evangelization of Christianity across the nation, and the urge for Christian

movements to evangelize persists today. Following a brief interlude in which the Japanese ruled

during World War II, Indonesia formally established independence in 1949.

Since achieving independence, Indonesia has been ruled by two dictators, Sukarno

followed by Suharto. Suharto came to power after Sukarno in 1968 following an army-led anti-

communist purge in 1965 in which over 500,000 people were killed. However, Indonesia was

among the hardest hit by the Asian Financial Crisis in the late 1990s, and amidst the crisis

protests led to Suharto’s resignation in 1998. Following Suharto’s reign Indonesia has

increasingly democratized and has continued to operate under secular governance. For this

reason, Indonesia is seen as progressive with women’s rights relative to most Muslim-majority

countries (Pringle, 2010).

In recent years, Indonesia has been a site for six terrorist attacks by militant Islamic

organizations between 2002 and 2016. Indonesia was also devastated by an earthquake and

tsunami in 2004 that killed approximately 230 thousand people. A series of forest and plantation

fires due to slash and burn techniques has also led to dense hazes that have killed dozens of

people due to low visibility and respiratory illnesses.

Data

The data come from Wave 5 of the Indonesian Family Life Survey (IFLS5), a

longitudinal survey collected in 13 of Indonesia’s 27 provinces (Strauss, Witoelar, & Sikoki,

2016). The IFLS has been collected since 1993 and represents 83 percent of Indonesia’s

population with a sample size of more than 30,000. Individual surveys as well as data from

community leaders and social service centers such as schools, health clinics, and hospitals are

included.

The data for my analyses come from both the household- and community-level data.

Initially starting with a sample of 30,872, limiting the sample to women reduced the total to

16,706. I excluded women who moved outside of the surveyed communities, which further

restricted the sample size to 10,882. Some women had missing health data, and restricting these

women led to a sample size of 10,818 women. Finally, I restricted the sample to women who are

the household head or the household head’s spouse. After doing this, I excluded two households

for having two women who reported being spouses of the household head, presumably in

polygynous unions. Restricting the sample to household heads and household head spouses led to

a final sample size of 7,355 women nested within 310 communities.

Dependent Variable. In the IFLS5, regular interviewers were trained in collecting physical

measurements. Blood pressure was taken as the first anthropometric measure and was taken three

times on alternating arms. An Omron meter, HEM-7203, was used. (Strauss, Witoelar, & Sikoki,

2016). The Omron meter is fully automatic and operates on the oscillometric principle (Omron,

2016). That is, blood flowing through an artery between systolic and diastolic pressures produces

vibrations in the arterial wall that are translated to electric signals and provided as a digital

readout (Berger, 2001). Normal-sized cuffs were used in most circumstances, although large

cuffs were available if needed (Strauss, Witoelar, & Sikoki, 2016).

Importantly, past research has demonstrated the value of home blood pressure

measurements as opposed to measurements taken in a doctor’s office, as in-office measurements

are often higher than those taken at home (Bliziotis, Destounis, & Stergiou, 2012). At-home

measurements allow for more accurate readings due to the absence of the “white coat effect” –

having higher blood pressure during a doctor’s visit (Fuchs, de Mello, & Fuchs, 2013).

Furthermore, home blood pressure monitoring better predicts organ damage (Chobanian et al.,

2003) and hypertension (Pickering et al., 2008; Ward, Takahasi, Stevens, & Heneghan, 2012).

To account for differences in at-home blood pressure readings compared to a doctor’s office,

hypertension thresholds are adjusted accordingly and explained below.

Hypertension – Using the mean systolic and diastolic measurements for blood pressure, I

constructed a dichotomous measure of hypertension. In accordance with the Joint National

Committee on Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (2003),

individuals with a mean home-measured BP of more than 135/85 mm Hg are classified as

hypertensive. As supported by (Parati et al., 2008), this threshold is also the accepted indicator

for hypertension in older individuals and pregnant women. Women who do not fall above this

threshold but report taking hypertension medication are also classified as hypertensive.

Systolic Blood Pressure (SBP) – Systolic blood pressure refers to the amount of pressure

in the arteries during heart muscle contraction. I calculated SBP as a mean of three systolic

measurements taken. For those respondents missing measurements, the available measurements

were used to construct a systolic blood pressure value. For example, for women with two

readings, the mean of the two readings was used to create their SBP value.

Diastolic Blood Pressure (DBP) – Diastolic blood pressure refers to the amount of

pressure in the arteries when the heart is in-between beats. DBP was calculated as the mean of

three diastolic measurements. As with SBP, those women with fewer than three readings had

DBPs calculated using the available number of measurements.

Independent Variables. Religious affiliation was self-reported. Women reported affiliations

with Islam, Catholicism, Christianity, Hinduism, and Buddhism. Due to the low number of

Buddhist women in the sample (n=5), I exclude Buddhist respondents from analyses. Catholic

and Christian women are combined to form a “Christian” comparison group. To address

hypothesis 1, I disaggregate Muslim, Christian, and Hindu women with Muslim being the

reference. In order to address hypothesis 2, I disaggregate Muslim women by which Islamic

tradition they feel closest to. The tradition Nahdlatul Ulama (traditionalist) serves as the

reference category, and comparison groups are Muhamadiyah (modernist), not close to any

tradition, Christian, and Hindu. Finally, I disaggregate Muslim women by prayer frequency – 0

prayers per day, 1-4 prayers per day, 5 prayers per day, and 6 or more prayers per day – and

compare those groups to Christian and Hindu women in order to explore hypothesis 3.

Following initial findings, I also include additional variables to disaggregate Hindu

women by whether or not they eat beef as well as whether or not they participate in daily yoga or

meditation. I separate Hindu women into these categories because red meat consumption is

associated with higher risks of hypertension (Pan et al., 2012), while regular meditation and/or

yoga are associated with lower risks of hypertension (Wachholtz & Pargament, 2005). Using the

same reference groups and categories for Muslim and Christian women detailed above, I replace

the single aggregated Hindu group with Hindu women with and without dietary restrictions as

the independent variable for each model, followed by an independent variable with Hindu

women disaggregated by those who do and not participate in daily yoga or meditation.

Community-Level Controls. First, I separate each community into the major grouping of

islands it belongs to. The reference category is Sumatra, and the comparison groups are Java and

“other major islands.” This categorization of island groupings reflects the documentation

provided by the IFLS. Second, I include a dichotomous measure of urban/rural with urban

serving as the reference group. Proportion of community households with electricity is centered

at 35 percent – the minimum percent electricity in the sample. In order to control for health

center accessibility, I include a continuous variable for the number of health posts in the

community ranging from 1 to 38. Finally, I include a categorical variable for the level of

education of the community leader. The reference group is “junior high or less,” and comparison

groups are “high school” and “university/post-secondary.”

In order to accurately partition the variance accounted for as between-person variance

(individual-level) and between-community variance (community-level) to test for contextual-

level effects (Hox, Moerbeek, & van de Schoot, 2010), I also include community-level variables

corresponding with all individual-level characteristics. For continuous variables, I use the

community mean of the variable (e.g. the community’s mean number of household assets for a

given community). For categorical variables, I include a measure for the proportion of the

community reporting each response.

Individual-Level Controls. I included a continuous asset scale with scores ranging from 0-29.

The scale is weighted with more expensive items such as home ownership being worth more.

Possible assets are: house lived in, other house/building, land outside of farm land, livestock or

fish pond, poultry, hard crop trees (perennials), vehicle, household appliances such as a radio or

TV, savings/certificates of deposit/stocks, receivables, jewelry, household furniture and utensils,

and other assets.

Additionally, I include a continuous variable for age centered at the minimum age in the

sample of 15. Ages range from 15 to 101. Marital status is dichotomously coded with married

being the reference and single/divorced/separated/widowed being the comparison group. I also

include a binary measure of highest level of education attained. The reference group is

elementary school or less compared to junior high or more. Employment status is coded as

unemployed (reference) versus employed. A continuous measure for the number of community

programs the respondent has been involved with in the last 12 months ranges from 0 to 14. I also

dichotomously code for whether or not there are children under 15 years old residing in the

household with those women who answered “yes” being the reference group.

Self-reported childhood health status is a continuous Likert-type item ranging from

excellent (0) to poor (4). As many of the women in the sample are of childbearing age, I include

a dichotomous measure to account for whether a woman is pregnant and/or breastfeeding with

the reference being “neither pregnant nor breastfeeding.” As anemia, diabetes, and high

cholesterol are strong indicators of hypertension, I include a binary variable for those taking

medication for anemia, diabetes, and/or high cholesterol, with “no medication” being the

reference group. Smoking status is a dichotomous measure for whether the respondent has a

history of smoking (including those who quit and those who still smoke).

In order to account for healthcare quality, I include a categorical measure of healthcare

satisfaction. The reference category is women who reported less-than-adequate healthcare, and

the comparison groups are “adequate healthcare” and “more-than-adequate healthcare.” Women

were also asked about their food security; women with less-then-adequate food serve as the

reference, and comparison groups are women with “adequate food intake” and “more-than-

adequate food intake.” I also control for whether a woman was fasting at the time of the survey

with those who answered “no” being the reference group.

As regular physical activity has been shown to be a major predictor of hypertension, I

construct a binary variable for whether or not the respondent meets guidelines for adequate

physical activity, defined as participating in at least 25 minutes of vigorous activity 3 days a

week or 30 or more minutes of moderate activity 5 days a week (AHA, 2016). Women who fell

into one or both of these categories when reflecting on their weekly activity are categorized as

“physically active” (reference), while those who did not are categorized as “physically inactive.”

Finally, I construct a binary BMI variable calculated as: weight(in kg)/(height(in m)2). Per CDC

guidelines, those whose body mass index falls above the cut-off point of 25.0 are classified as

overweight or obese, with “not overweight” as the reference category. Because less than 5

percent of the sample are categorized “underweight,” they are included as “not overweight or

obese.”

Data Analysis

I estimate the relationship between the independent variables and their relationship with

hypertensive status, defined as having a home-based blood pressure greater than 135/85mm Hg,

using multilevel logistic regression models in Stata Statistical Software 14 (StataCorp, 2015).

Respondents are nested within communities, and all independent variables and controls are

included as fixed effects. Initially, I use three models to test hypotheses 1-3. However,

unanticipated findings regarding risks of hypertension in Hindu women resulted in 2 additional

models that disaggregate Hindu women by beef consumption and yoga/meditation practices.

In constructing the models, likelihood ratio tests for significant improvement in model fit

upon the inclusion of each independent variable are used. Variables that do not provide

significant improvement in model fit are trimmed and variables that significantly improve model

fit are retained. I perform pairwise comparisons to examine mean differences across comparison

groups for the independent variable in each model. Models are estimated using maximum

likelihood (ML) with unstructured covariances. Covariances are unstructured in order to allow

for unique variances and covariances without constraints. Because each variance and covariance

is estimated uniquely from the data, this structure results in the best possible model fit despite the

loss of degrees of freedom.

It is important to mention limitations. First, there is a possible issue of reverse causality

regarding religion and hypertensive status. However, most hypertensive individuals do not

experience symptoms (WHO, 2016). Furthermore, there is no reason to suspect that hypertensive

status would result in women differentially self-selecting into certain religions or Islamic

traditions. Even for Muslim women who may experience symptoms due to hypertension,

symptoms should not affect the number of daily prayers women report, as Muslim individuals

are expected to adapt prayer customs to physical abilities. Second, blood pressure measurements

are a mean of three measurements taken on a single day. A more accurate accounting of BP

would happen over a larger period of time. However, again there is no reason to believe that

more accurate BP readings would affect one religion or tradition over another.

Results

Descriptive statistics for the independent variables and individual-level controls by

hypertensive status are provided in Table 1. For continuous controls, I provide the percent

hypertensive for the mean response. Twenty seven percent of Muslim respondents, 27 percent of

Christian respondents, and 18 percent of Hindu respondents are categorized as hypertensive.

Within Muslim respondents, 26 percent of NU Muslim women, 26 percent of Muhammadiyah,

and 28 percent of Muslim women who do not identify with a tradition are hypertensive.

Nineteen percent of those who do not pray daily, 19 percent of those who pray 1-4 times per day,

27 percent of women who pray 5 times per day, and 32 percent of women who prayer six or

more times per day are categorized as hypertensive.

------------------------------------------------TABLE 1 ABOUT HERE ----------------------------------

Table 2 provides multivariable results. These results are reported as relative risk ratios,

calculated by converting odds ratios as: RR=OR/[(1-P0)+(P0 x OR)] with P0 representing the

incidence for unexposed risk group (Zhang & Kai, 1998). In Model 1, I test hypotheses 1A and

1B by comparing relative risks of hypertension for Muslim, Christian, and Hindu women.

Inclusion of religion as fixed effects significantly improves model fit via a likelihood ratio test

relative to a model including all other fixed effects (χ2(2)=9.79, p=.008). Christian women have

marginally significantly lower risks of hypertension at 0.71 times the rate of Muslim women,

therefore lending some support to hypothesis 1A. Hindu women have significantly lower rates of

hypertension at less than one third the rate of Muslim women. As such, Model 1 does not

provide support for hypothesis 1B; that is, Hindu women and Muslim women do not experience

comparable risks of hypertension. Hindu women experience significantly lower rates of

hypertension than Muslim women. There are no statistically significant differences between

Christian and Hindu women.

-------------------------------------------TABLE 2 ABOUT HERE -----------------------------------------

In estimating Model 2, the disaggregation of Muslim women by tradition results in

significant improvement in model fit relative to a model including all other fixed effects

(χ2(4)=10.68, p=.03). However, relative to Model 1, there is no significant improvement in

model fit (χ2(2)=0.76, p=.684). Model 2 does not offer support to hypothesis 2 – there are no

significant differences in hypertension rates among Muslim women from different traditions.

Although Muhammidiyah women experience lower rates of hypertension than NU women, the

difference is nonsignificant. Compared to the other major religions, Christian women have

marginally significantly lower risks of hypertension when compared to NU Muslim women

(RR=0.71, p=.065), and Christian women have significantly lower risks compared to Muslim

women who do not adhere to a tradition (RR=0.60, p=.053) (not shown). Hindu women

experience significantly lower risks of hypertension compared to all Muslim women regardless

of tradition, although the largest difference in risk is between Hindu women and Muslim women

who do not adhere to a tradition at slightly above one quarter the rate (RR=0.269, p=.006).

Model 3 provides hypertension risks for Muslim women with differences in number of

reported daily prayers. Relative to Model 1, Model 3 does not provide significant improvement

in Model fit (χ2(3)=1.84, p=.606). Model 3 provides no evidence in support of hypothesis 3;

there are no significant differences in hypertension rates for Muslim women who pray different

amounts each day. However, pairwise comparisons show that Hindu women experienced

statistically significantly lower rates of hypertension when compared to Muslim women from

every category of prayer (p<.01). Additionally, Christian women experienced significantly lower

rates of prayer when compared to Muslim women who prayed 5 times per day (RR=.595,

p=.046).

Given the unanticipated findings in each of the models showing that Hindu women have

lower relative risks of hypertension than Muslim women of every tradition as well as Muslim

women of every category of prayer, I include additional models disaggregating Hindu women by

beef consumption followed by disaggregating Hindu women by daily yoga and meditation

(Table 3) and compare these models to Model 1 via likelihood ratio tests. Model 4 does not

significantly improve model fit relative to the religion model (χ2(1)=1.05, p=.306). In Model 4, I

disaggregate Hindu women by women who do and do not consume beef. Hindu women who eat

beef experience hypertension at one third the rate of Muslim women (p=.02), and Hindu women

who do not eat beef experience hypertension at approximately one fifth the rate of Muslim

women (p=.005). There are no significant differences in hypertension rates between Hindu

women who do and do not eat beef.

------------------------------------------TABLE 3 ABOUT HERE ------------------------------------------

In Model 5, I disaggregate Hindu women by women who do yoga/meditate infrequently

and those who do yoga/meditate every day. Model 5 provides significant improvement in model

fit relative to the aggregated religion model (Model 1) and therefore is the model of best fit

(χ2(1)=3.92, p=.048). Hindu women regardless of yoga/meditation experience statistically

significantly lower rates of hypertension than Muslim women. Those who do yoga/meditate

infrequently experience risks at 0.38 times the rate of Muslim women (p=.045), while those who

do yoga/meditate daily experience hypertension risks at one fifth the rate of Muslim women

(p=0.002). There was also a significant difference in rates of hypertension when comparing

Hindu women who do daily yoga/meditation versus those who do not; women who do

yoga/meditate experience risks of hypertension at 40 percent the rate of Hindu women who

infrequently do yoga/meditate.

Discussion

Although past findings from India suggest that minority religions experience lower rates

of hypertension than majority religions (Banerjee, Mukherjee, & Basu, 2016), my findings from

Indonesia show that Hindu women as well as Christian women fare better than Muslim women

with regard to hypertensive risks after controlling for key individual-level, community-level, and

contextual-level effects. The findings provide some support for hypothesis 1A – that is, Christian

women experience marginally significantly lower rates of hypertension than Muslim women in

four of the five models. This could be attributed in part to connections to the Western world that

provide benefits to Christians that Muslim women may not have access to (Jones, 2005).

Furthermore, there are not significant differences between Muslim women of different

traditions. Although I hypothesized that women who belong to the Nahdlatul Ulama tradition

would experience lower rates of hypertension than women who do not adhere to a tradition or

Muhammadiyah women, tradition does not result in significant differences in hypertension rates.

This finding could provide evidence for cohesion between the traditions; Muslim women in

Indonesia regardless of tradition adhere to the same core beliefs and participation in the salat as a

profession of faith (Pringle, 2010).

Additionally, despite past findings showing that women who participate in salat

experience better cognitive functioning attributed largely to the physical and meditative nature of

Islamic prayer (Doufesh et al., 2013), I estimate no significant differences in hypertensive rates

for Muslim women with varying degrees of daily prayer. Although salat prayer involves a series

of physical movements, the survey question used to gauge prayer participation for Muslim

women does not specify whether women are engaging with the physical aspects of prayer. Given

that Muslims are expected to adapt daily prayer to their physical limitations, salat prayer may not

provide meaningful physical activity for women who already have limitations. Furthermore,

women engaging in more prayer may be mobilizing spiritual resources as a reaction to poor

health (Salmoirago-Blotcher et al., 2013), and I cannot explore differences in possible reasons

women pray at higher rates due to limitations to the data.

Surprisingly, Hindu women have significantly lower risks of hypertension than Muslim

women in all five models presented, contrary to my hypothesis that Hindu and Muslim women

would have comparable hypertension risks (hypothesis 1B). Even after disaggregating Hindu

women by those who eat beef and those who do not, both Hindu groups experience lower

hypertension risks than Muslim women, although Hindu women with restricted diets experience

the lowest rates of hypertension relative to Muslim women. The nonsignificant difference in

women who eat and do not eat beef could reflect recent findings that unprocessed red meat is not

a significant predictor of hypertension (Lajous, 2015), although the survey questionnaire makes

no distinction between the type of meat consumed.

Disaggregating Hindu women by yoga and meditation practices may provide the most

insight into mechanisms through which religion may influence hypertensive risks; Hindu women

who meditate or practice yoga daily experience lower risks of hypertension than all other

religious groups including Hindu women who meditate or practice yoga infrequently.

Participation in daily yoga and meditation through Hinduism, which provides physical health

benefits, reflects past literature examining the role religion plays on encouraging healthy

behaviors and exercise (Koenig, 2012). Given past findings demonstrating the value of yoga for

cardiac health (Satin, Linden, & Millman, 2014; Rosenberg, 2015) as well as findings showing

that meditative practices infused with spirituality provide additional health benefits when

compared to secular meditation (Wachholtz & Pargament, 2005), intentional yoga and

meditation infused with Hindu beliefs may provide additional cardiac health benefits. Future

research may compare women who practice yoga for religious purposes versus women who

practice yoga secularly to examine possible differences in health benefits.

Next Steps

This study provides unique insight into the between-religion differences in hypertensive

rates in Muslim-majority setting. As I prepare the paper for presentation at PAA, I will take

advantage of the longitudinal nature of the data to explore possible differences in beef

consumption and yoga/meditation practices for Hindu women across the earlier waves to see if

changes in food consumption and physical activity behaviors over time reflect changes in

hypertensive risks. I will also expand my analyses to include models examining differences in

severity of isolated systolic hypertension (ISH) in Wave 5 as well as examining the relationship

between religions and hypertensive status in earlier waves of the IFLS as a predictor of mortality

in Wave 5.

References

Abdul-Razak, S., Daher, A. M., Ramli, A. S., Ariffin, F., Mazapuspavina, M. Y., Ambigga, K. S., ... &

Ng, K. K. (2016). Prevalence, awareness, treatment, control and socio demographic determinants

of hypertension in Malaysian adults. BMC public health, 16(1), 1.

Abdullahi Ahmed An-Na'im,. (2008). Islam and the Secular State.

AlAbdulwahab, S. S., Kachanathu, S. J., & Oluseye, K. (2013). Physical activity associated with prayer

regimes improves standing dynamic balance of healthy people. Journal of physical therapy

science, 25(12), 1565.

Bai, R., Ye, P., Zhu, C., Zhao, W., & Zhang, J. (2012). Effect of salat prayer and exercise on cognitive

functioning of Hui Muslims aged sixty and over.Social Behavior and Personality: an

international journal, 40(10), 1739-1747.

Banerjee, A. T., Boyle, M. H., Anand, S. S., Strachan, P. H., & Oremus, M. (2014). The relationship

between religious service attendance and coronary heart disease and related risk factors in

Saskatchewan, Canada.Journal of religion and health, 53(1), 141-156.

Banerjee, S., Mukherjee, T. K., & Basu, S. (2016). Prevalence, awareness, and control of hypertension

in the slums of Kolkata. Indian Heart Journal.

Barraclough, S. (1999). Women and tobacco in Indonesia. Tobacco Control,8(3), 327-332.

Bell, C. N., Bowie, J. V., & Thorpe Jr, R. J. (2012). The interrelationship between hypertension and

blood pressure, attendance at religious services, and race/ethnicity. Journal of religion and

health, 51(2), 310-322.

Bernardi, L., Sleight, P., Bandinelli, G., & Cencetti, S. (2001). Effect of rosary prayer and yoga mantras

on autonomic cardiovascular rhythms: comparative study. British Medical Journal, 323(7327),

1446.

Berntson, G. G., Norman, G. J., Hawkley, L. C., & Cacioppo, J. T. (2008). Spirituality and autonomic

cardiac control. Annals of Behavioral Medicine,35(2), 198-208.

Bliziotis, I. A., Destounis, A., & Stergiou, G. S. (2012). Home versus ambulatory and office blood

pressure in predicting target organ damage in hypertension: a systematic review and meta-

analysis. Journal of hypertension, 30(7), 1289-1299.

Bowen, J. (1989). Salat in Indonesia: The Social Meanings of an Islamic Ritual.Man, 24(4), 600-619.

Bradley, E. (1995). Religious Involvement and Social Resources: Evidence from the Data Set"

Americans' Changing Lives". Journal for the scientific study of Religion, 259-267.

Buck, A. C., Williams, D. R., Musick, M. A., & Sternthal, M. A. (2009). An Examination of Mediators

in the Relationship between Religiosity, Blood Pressure, and Hypertension. Social Science and

Medicine, 68(2).

Chobanian, A. V., Bakris, G. L., Black, H. R., Cushman, W. C., Green, L. A., Izzo Jr, J. L., ... &

Roccella, E. J. (2003). The seventh report of the joint national committee on prevention,

detection, evaluation, and treatment of high blood pressure: the JNC 7 report. Jama, 289(19),

2560-2571.

DHS (2012). Indonesia DHS, 2012 – Final report. http://dhsprogram.com/publications/publication-

fr275-dhs-final-reports.cfm.

Doufesh, H., Faisal, T., Lim, K. S., & Ibrahim, F. (2012). EEG spectral analysis on Muslim

prayers. Applied psychophysiology and biofeedback,37(1), 11-18.Fitchett, G., & Powell, L. H.

(2009). Daily spiritual experiences, systolic blood pressure, and hypertension among midlife

women in SWAN. Annals of Behavioral Medicine, 37(3), 257-267.

Doufesh, H., Ibrahim, F., Ismail, N. A., & Ahmad, W. A. W. (2013). Assessment of heart rates and

blood pressure in different salat positions.Journal of Physical Therapy Science, 25(2), 211-214.

Gillum, R. F., & Ingram, D. D. (2006). Frequency of attendance at religious services, hypertension, and

blood pressure: the Third National Health and Nutrition Examination Survey. Psychosomatic

Medicine, 68(3), 382-385.

Gudorf, C. E. (2012). Religion, Law, and Pentecostalism in Indonesia.Pneuma, 34(1), 57-74.

He, J., & Whelton, P. K. (1999). Elevated systolic blood pressure and risk of cardiovascular and renal

disease: overview of evidence from observational epidemiologic studies and randomized

controlled trials. American heart journal, 138(3), S211-S219.

Henkel, H. (2005). ‘BETWEEN BELIEF AND UNBELIEF LIES THE PERFORMANCE OF SALĀT

’: MEANING AND EFFICACY OF A MUSLIM RITUAL. Journal of the Royal

Anthropological Institute, 11(3), 487-507.

Holt, C. L., Clark, E. M., Wang, M. Q., Williams, B. R., & Schulz, E. (2015). The Religion–Health

Connection Among African Americans: What Is the Role of Social Capital?. Journal of

Community & Applied Social Psychology,25(1), 1-18.

Hoon, C. Y. (2013). Between evangelism and multiculturalism: The dynamics of Protestant Christianity

in Indonesia. Social Compass, 60(4), 457-470.

Hoon, C. Y. (2016). Religious aspirations among urban Christians in contemporary

Indonesia. International Sociology, 0268580916643853.

Hox, J. J., Moerbeek, M., & van de Schoot, R. (2010). Multilevel analysis: Techniques and applications.

Routledge.

Ibrahim, N. K., Hijazi, N. A., & Al-Bar, A. A. (2008). Prevalence and determinants of prehypertension

and hypertension among preparatory and secondary school teachers in Jeddah. J Egypt Public

Health Assoc, 83(14), 183-203.

Ibrahim, F., Sian, T. C., Shanggar, K., & Razack, A. H. (2013). Muslim Prayer Movements as an

Alternative Therapy in the Treatment of Erectile Dysfunction: A Preliminary Study. Journal of

physical therapy science,25(9), 1087.

Joffres, M., Falaschetti, E., Gillespie, C., Robitaille, C., Loustalot, F., Poulter, N., ... & Campbell, N.

(2013). Hypertension prevalence, awareness, treatment and control in national surveys from

England, the USA and Canada, and correlation with stroke and ischaemic heart disease

mortality: a cross-sectional study. BMJ open, 3(8), e003423.

Jones, N. (2005). REDISCOVERING PANCASILA: RELIGION IN INDONESIA'S PUBLIC

SQUARE. The Brandywine Review of Faith & International Affairs, 3(1), 23-30.

Kayima, J., Nankabirwa, J., Sinabulya, I., Nakibuuka, J., Zhu, X., Rahman, M., ... & Kamya, M. R.

(2015). Determinants of hypertension in a young adult Ugandan population in epidemiological

transition—the MEPI-CVD survey. BMC public health, 15(1), 830.

Khanam, M. A., Lindeboom, W., Razzaque, A., Niessen, L., & Milton, A. H. (2015). Prevalence and

determinants of pre-hypertension and hypertension among the adults in rural Bangladesh:

findings from a community-based study. BMC public health, 15(1), 1.

Koenig, H. G. (2012). Religion, spirituality, and health: The research and clinical implications. ISRN

psychiatry, 2012.

Krause, N. (2008). The social foundation of religious meaning in life.Research on Aging, 30(4), 395-

427.

Lajous, M., Bijon, A., Fagherazzi, G., Rossignol, E., Boutron-Ruault, M. C., & Clavel-Chapelon, F.

(2014). Processed and unprocessed red meat consumption and hypertension in women. The

American journal of clinical nutrition, 100(3), 948-952.

McDaniel, J. (2010). Agama Hindu dharma Indonesia as a new religious movement: Hinduism recreated

in the image of Islam. Nova Religio: The Journal of Alternative and Emergent Religions, 14(1),

93-111.

Pan, A., Sun, Q., Bernstein, A. M., Schulze, M. B., Manson, J. E., Stampfer, M. J., ... & Hu, F. B.

(2012). Red meat consumption and mortality: results from 2 prospective cohort studies. Archives

of internal medicine,172(7), 555-563.

Parati, G., Stergiou, G. S., Asmar, R., Bilo, G., de Leeuw, P., Imai, Y., ... & O'Brien, E. (2008).

European Society of Hypertension guidelines for blood pressure monitoring at home: a summary

report of the Second International Consensus Conference on Home Blood Pressure

Monitoring. Journal of hypertension, 26(8), 1505-1526.

Pickering, T. G., Miller, N. H., Ogedegbe, G., Krakoff, L. R., Artinian, N. T., & Goff, D. (2008). Call to

action on use and reimbursement for home blood pressure monitoring A joint scientific statement

from the American Heart Association, American Society of Hypertension, and Preventive

Cardiovascular Nurses Association. Hypertension, 52(1), 10-29. Powell, L. H., Shahabi, L., &

Thoresen, C. E. (2003). Religion and spirituality: Linkages to physical health. American

psychologist, 58(1), 36.

Pringle, R. (2010). Understanding Islam in Indonesia: Politics and diversity. Honolulu: University of

Hawaiʻi Press.

Reza, M. F., Urakami, Y., & Mano, Y. (2001). Evaluation of a new physical exercise taken from salat

(prayer) as a short-duration and frequent physical activity in the rehabilitation of geriatric and

disabled patients. Annals of Saudi medicine, 22(3-4), 177-180.Rosenberg, K. (2015). Yoga may

Improve Cardiac and Metabolic Risk Factors. AJN The American Journal of Nursing, 115(4), 60.

Salmoirago-Blotcher, E., Hunsinger, M., Morgan, L., Fischer, D., & Carmody, J. (2013). Mindfulness-

based stress reduction and change in health-related behaviors. Journal of Evidence-Based

Complementary & Alternative Medicine, 2156587213488600.

Satin, J. R., Linden, W., & Millman, R. D. (2014). Yoga and psychophysiological determinants of

cardiovascular health: comparing yoga practitioners, runners, and sedentary individuals. Annals

of Behavioral Medicine, 47(2), 231-241.

Seo, M. (2012). Defining ‘religious’ in Indonesia: toward neither an Islamic nor a secular

state. Citizenship Studies, 16(8), 1045-1058.

Strauss, J., Witoelar, F., & Sikoki, B. (2016). The fifth wave of the Indonesia Family Life Survey

(IFLS5): Overview and field report.

Suprapto, S. (2016). THE THEOLOGY OF TOLERANCE IN HINDU AND ISLAM: MAINTAINING

SOCIAL INTEGRATION IN LOMBOK-INDONESIA.ULUMUNA, 19(2), 329-352.

Wachholtz, A. B., & Pargament, K. I. (2005). Is spirituality a critical ingredient of meditation?

Comparing the effects of spiritual meditation, secular meditation, and relaxation on spiritual,

psychological, cardiac, and pain outcomes. Journal of behavioral medicine, 28(4), 369-384.

Ward, A. M., Takahashi, O., Stevens, R., & Heneghan, C. (2012). Home measurement of blood pressure

and cardiovascular disease: systematic review and meta-analysis of prospective studies. Journal

of hypertension,30(3), 449-456.

WHO (2013). Global health observatory data.

http://www.who.int/gho/ncd/risk_factors/blood_pressure_prevalence/en/.

WHO (2014). Alcohol consumption: Levels and patterns. Retrieved 22 September 2016 from http://ww

w.who.int/substance_abuse/publications/global_alcohol_report/profiles/idn.pdf.

WHO (2016). A global brief on hypertension: silent killer, global public health crisis. World. Retrieved

from http://tinyurl.com/cywh2eh.

Table 1: Percent Hypertensive for Independent Variables and Individual-Level Controls

% Hypertensive (SD)

Independent Variables

Muslim Tradition

NU 26.33(44.04)

Muhammadiyah 26.08(43.94)

No Tradition 28.31(45.07)

Muslim Prayer

0 Prayers Per Day 18.94(39.27)

1-4 Prayers Per Day 19.13(39.36)

5 Prayers Per Day (salat) 26.88(44.34)

6 or More Prayers Per Day 31.97(46.65)

Christian 27.35(44.64)

Hindu Dietary Restrictions

No Dietary Restrictions 18.98(39.28)

Does Not Eat Beef 14.63(35.56)

Hindu Yoga/Meditation Participation

Daily Yoga/Meditation 12.43(33.09)

Infrequent Yoga/Meditation 23.46(42.50)

Individual-Level Controls

Age (mean age=43) 34.83(47.91)

Community Participation (mean=2) 28.17(44.99)

Education (ref=elementary or less) 33.72(47.28)

Junior High or More 18.12(38.53)

Employment (ref=unemployed) 27.81(44.81)

Employed 25.20(43.42)

Household Assets (mean=11) 24.63(43.12)

Kids Under 15 in Household (ref=yes) 38.53(48.68)

No 16.75(37.34)

Marital status (ref=married) 24.11(42.78)

Single/separated/widowed 39.05(48.81)

BMI (ref=not overweight or obese) 20.34(40.26)

Overweight or obese 33.13(47.07)

Childhood Health (mean=2) 26.99(44.40)

Tobacco use (ref=never tried) 25.92(43.82)

Formerly smoked, currently smokes 34.15(47.51)

Medication Use (ref=no medications) 25.82(43.77)

Takes medication 32.80(47.00)

Pregnant or Breastfeeding (ref=no) 28.60(45.19)

Yes 9.40(29.20)

Healthcare Quality (ref=less-than-adequate) 31.46(46.45)

Average 26.37(44.07)

More-than-adequate 20.63(40.48)

Food security (ref=less than adequate) 30.66(46.13)

Average 27.10(44.45)

More-than-adequate 21.34(40.98)

Fasting (ref=no) 26.31(44.03)

Yes 24.40(43.04)

Physically active (ref=no) 26.69(44.24)

Yes 25.75(43.73)

Table 2: Hypertensive Relative Risk Ratios by Religion, Muslim Tradition, and Daily Prayers

Model 1 Model 2 Model 3

Religion (ref=Muslim)

Christian 0.710†

Hindu 0.311**

Religion (ref=Muslim NU)

Muslim Muhammadiyah

0.948

Muslim No Tradition

1.022

Christian

0.707†

Hindu

0.278**

Religion (ref=Muslim & 0 daily prayers)

Muslim & 1-4 daily prayers

1.059

Muslim & 5 daily prayers

1.087

Muslim & 6 or more daily prayers

1.012

Christian

0.757

Hindu

0.311*

N=7,355; †p<.10, *p<.05, **p<.01, ***p<.001

Table 3: Hypertensive Relative Risk Ratios by Religion, Hindu Diet, and Hindu Yoga/Meditation

Model 4 Model 5

Religion (ref=Muslim)

Christian 0.712†

Hindu & No Dietary Restrictions 0.325*

Hindu & Does Not Eat Beef 0.219**

Religion (ref=Muslim)

Christian

0.709†

Hindu & Meditates/Does Yoga

Infrequently

0.395*

Hindu & Meditates/Does Yoga Daily

0.199**

N=7,355; †p<.10, *p<.05, **p<.01, ***p<.001