OUP user menu

Health and religiosity among Israeli Jews

Amir Shmueli
DOI: http://dx.doi.org/10.1093/eurpub/ckl105 104-111 First published online: 28 July 2006


Background: The objective of this paper is to explore the connection between self-reported health and religiosity among Israeli Jews, using several self-reported health measures. Methods: Data were collected by two health surveys covering 1999 individuals in 1993 and 2505 individuals in 2000, representing the population of Jewish Israelis aged 45–75 years residing in urban communities in those years. Self-reported health was measured by (i) reported chronic conditions, (ii) the SF-36 instrument, and (iii) a visual analogue scale of health-related quality of life. Religiosity was measured by a self-reported five-category scale. Results: Controlling for a large array of socio-demographic characteristics, while no religiosity gradient was found in reported chronic morbidity, religious persons generally report worse health than secular persons on the other measures. The gap is larger in the SF-36's role-performance scales, and among women and Israelis from Asian-African origin. Discussion: The mixed results are consistent with the ambiguity of the religiosity effect on health reported in recent surveys. However, trying to reconcile between longer life expectancy of religious persons found in earlier Israeli and other research and poorer reported health found above, the paper emphasizes the possible differences in the perception of ‘normal’ life and roles, and argues that the SF-36 health measures might suffer from a religiosity-related reporting heterogeneity, distorting their association with mortality in the population.

  • Israel
  • Judaism
  • religiosity
  • self-reported health
  • SF-36
  • VAS

Several reviews summarize the sociological, epidemiological, medical, and psychological research on the health–religiosity connection. Ellison and Levin1 conclude: ‘In these studies, salutary effects of religious involvement persist despite an impressive array of statistical controls for social ties, health behaviours, and socio-demographic variables’. This applies to both physical and mental health (MH). Easterbrook2 provides a popular review of recent studies showing the protective effect of religiosity on health, and concludes, following Koening et al.,3 that: ‘Lack of religious involvement has an effect on mortality that is equivalent to 40 years of smoking one pack of cigarettes per day’. Sloan et al.4 present a medical and different view, claiming that: ‘Even in the best studies, the evidence of an association between religion, spirituality, and health is weak and inconsistent’. Finally, Powell et al.5 conclude that the only confirmed finding is that: ‘In healthy participants, there is a strong, consistent, prospective, and often graded reduction in risk of mortality in church/service attenders’.

Several studies examined the health–religiosity connection among Israeli Jews. Friedlander et al.6 found that secular persons had higher prevalence of smoking, higher plasma levels of cholesterol and triglycerides, and lower levels of LDL cholesterol, controlling for age, sex, ethnic origin, body mass, and social class. No account was taken of possible differences in medication use or compliance to medicines prescribed to reduce cholesterol or triglycerides levels. These differences were attributed, at least partially, to different dietary regimes and nutrient intakes commanded by observing Kosher meat and avoiding dairy products concurrently or right after meat consumption. The lower levels of risk for coronary heart disease among religious persons might explain the lower incidence of acute myocardial infarction in these groups compared with secular groups found earlier.7 In a case–control study of survivors of a first myocardial infarction,8 secular persons had substantially higher risk than religious persons in both sexes, controlling for age, ethnic origin, education, smoking behaviour, physical exercise, and body mass. Another study9 examined the survival–religiosity connection by comparing religious and secular kibbutzim (collective settlements). It used 11 secular and 11 religious kibbutzim, matched on geographical location, using the same hospital, on members older than 40 years and year of establishment. Since all kibbutzim share similar social and economic structure, social support—which is believed to be a major channel through which religiosity exercises its effect on health—is held constant. Using 16 years of all-cause mortality data, it was found that mortality was considerably higher in secular kibbutzim. Most of the research reviewed above used mortality or morbidity as health measures. An exception is the study by Anson et al.10 who studied the relationship between religiosity and self-reported health in Israel. Their findings showed a protective effect of membership in a religious community on mental and physical health. Specific measures of individual religiosity either overlapped community membership or were not related to health outcomes.

Finally, a recent report of the Central Bureau of Statistics and the Ministry of Health ranked Israeli localities according to mortality rates and related them to socio-economic and demographic indices.11 The results indicate that two localities, characterized by exceptionally high proportions of religious persons—Jerusalem and Bnei-Brak—have the lowest age-adjusted death rate, and in particular in the 75+ age group. These differences cannot be explained by income differences, as these two cities have relatively low mean income. They have, however, relatively high levels of (religious) education which might contribute to higher survival. These results are consistent with the conclusions of Powell et al.5

Several explanatory mechanisms via which religious involvement may lead to positive health outcomes have been proposed.1,9 They include safer behaviour and personal lifestyles (religious Jews observe, in addition, dietary regimes and several other commandments, which might be beneficial for one's health, such as avoiding pork consumption and travelling on Saturdays and Jewish holidays, and consuming less red meat and dairy products); social integration and support; and a sense of meaning and positive emotions.

There are, however, several possible explanations for an adverse effect of religiosity on health.1,12,13 These include erosion of self-esteem, feeling of competence and coping styles, and foster feelings of guilt and shame; belief in Divine determinism and justice or other directives and norms may induce passivity and abdication of responsibility with regard to prevention, early detection, and compliance activities; and religious congregations can be sources of stress and of negative support.

The purpose of the present analysis is to explore the health–religiosity connection among urban Israeli Jews aged 45–75 years, using several common self-reported health and health-related quality of life (HRQL) measures, and controlling for a large array of socio-demographic characteristics.

In addition to estimating the religiosity gradient in the population, we focused on two interactions as follows: the religiosity gradient among men and women, and the religiosity gradient across ethnic origin groups. The religious involvement of men and women differ in the Jewish religion in Israel. Among the religious and the orthodox groups, men are much more involved in religious study, prayers and the synagogue life than women.14 Jewish religion practice and customs are also quite different between Israeli Jews whose ethnic origin is Europe or North America, those from Asian or North African countries, and post-1990 immigrants from the former USSR.

Data and variables

The data

The data used for the present analysis comes from two similar health surveys conducted by the Gertner Institute in 1993 and 2000. Stratified samples of 1999 individuals in 1993 and 2505 individuals in 2000 represented the population of Jewish Israelis aged 45–75 years residing in urban communities in the respective years (see15 for further details). Through full sit-down face-to-face interviews, information on many areas related to health was collected.

The measurement of health

Self-reported health was measured using three well-known measures.

Number of chronic conditions (denoted by CHRON)

The respondents were asked to report on the presence (diagnosed by a physician) of eight chronic conditions as follows: ulcers, kidney disease, arthritis, cancer, hypertension, asthma, heart disease, and diabetes. The number of chronic conditions ranges from zero to eight.

The visual analogue rating scale of HRQL

The survey included a visual analogue scale with anchors of ‘death’ (at 0) and ‘full health’ (at 100). The scale (horizontal, 20 cm length) was presented to the respondents, and they were asked to report the score on this scale that best represents their HRQL during the previous month.

The SF-36 instrument

The questionnaire included a Hebrew version of the Medical Outcomes Study Short-Form 36 (SF-36) instrument.16,17 The instrument includes 36 questions measuring eight health domains as follows: physical functioning (PF), physical limitations in roles' performance such as work (RP), pain (BP), general health (GH), emotional and mental well-being (MH), emotional limitations (such as pain, anxiety, or depression) in roles' performance (RE), social functioning (SF), and vitality (VT). Each domain is measured on a 0–100 scale, where a score of 100 signifies perfect health.

The measurement of religiosity

Self-reported religiosity was measured on a five-category scale as follows: secular, partially secular, observant (or traditional), religious, and orthodox. This scale has been found valid and reliable in eliciting religious beliefs and practice among Israeli Jews.18 Secular persons typically do not perform any religious rites. They might, however, perform the least-demanding commandments such as having a Mezuza on their front door. Partially secular people perform the ‘light’ commandments only, such as attending synagogue on the High Holidays, fasting on Yom Kippur. Observant persons keep ‘heavier’ commandments, such as keeping a Kosher kitchen, attending synagogue and avoiding any work (including using a car) on Saturdays. Religious persons keep more demanding commandments such as attending the synagogue daily, covering hair (women), wearing a hat (men). The orthodox persons try to observe all commandments more strictly and in more details, they live in separate communities and dress differently.

It should be noted that measuring religiosity in that way avoids possible endogeneity problems, which arise when religiosity is measured, as is done in several studies, in terms of the frequency in which people visit the church. The intensity of visits to church (or to synagogue) might be affected by the person's mobility and health state. On the other hand, it might introduce a bias if sick and elderly persons tend to turn to religion practice more than healthy and young individuals.10

Other personal characteristics

A wide range of socio-economic and demographic characteristics were included in the analysis. These included age, sex, years of schooling, subjective economic status, marital status, the size of the location of residency, sickness fund (health plan) membership (both in 1993 and in 2000, four sick funds were operating in Israel), and ethnic origin. The empirical definitions of the variables are given in table 1.

View this table:
Table 1

Variables definitions and descriptive statistics (n = 4504)

Self-reported health
    CHRONNumber of chronic conditions0.9570–8
    HRQLVisual analogue rating scale of HRQOL69.4060–100
    GHGeneral health scale (SF-36)63.3210–100
    PFPhysical functioning scale (SF-36)78.0110–100
    RERole emotional scale (SF-36)78.9290–100
    RPRole physical scale (SF-36)69.4990–100
    SFSocial functioning scale (SF-36)80.7580–100
    VTVitality scale (SF-36)56.7310–100
    BPBodily pains scale (SF-36)70.2420–100
    MHMental health scale (SF-36)67.2210–100
    SECULARSecular0.377Base category
    PARTSECPartially secular0.2110–1
    OBSERVObservants (traditional)0.2880–1
    PRIMAR0–8 years of schooling0.214Base category
    HIGHSCH9–12 years of schooling0.5210–1
    UNIVERS13+ years of schooling0.2650–1
Sick fund (health plan) membership
    CLALClalit sick fund0.707Base category
    MACCMaccabi sick fund0.1610–1
    MEUHMeuhedet sick fund0.0600–1
    LEUMLeumit sick fund0.0720–1
Ethnic origin
    IS-ISSecond generation Israeli born0.117Base category
    USSRPost-1990 immigrants from former USSR0.1130–1
    EUR-AMContinent of birth of father is Europe-America0.3860–1
    AS-AFContinent of birth of father is Asia or Africa0.3850–1
Other socio-demographic and economic characteristics
    AGEIn years57.81045–75
    ECONVery good or good economic status0.4900–1
    BIGCITYResidency location’s size >200 K inhabitants0.2640–1

The statistical strategy

Preliminary analysis showed that the focal relationships remained stable over the 2 years. Consequently, in order to identify the effect of religiosity on health with greater power, multivariate regression analysis was performed on the pooled (2 years) sample.

For HRQL and the eight SF-36 scales, classical regression (ordinary least squares, OLS) was used. For CHRON, being a count scale, we used the negative binomial regression.

For the HRQL and the eight SF-36 scales several alternative specifications were tried as well. In the first, the dependent variables were measured in logarithms of the original scales, to approach a symmetric distribution. In the second, the nine equations were estimated simultaneously as a system of seemingly unrelated equations (SURE), to account for possible unobserved correlations among the error terms. In the third, in order to deal with possible ‘ceiling effects’ (a relatively large concentration at the upper score, 100), the equations were estimated by Tobit models with upper limit at 100. All specifications provided essentially similar conclusions.

The potential sex differences in the relationships were introduced into the equations by including interaction terms between sex and three religiosity groups (secular and partially secular serve as the base category, observant persons, and religious and orthodox persons). Similarly, the exploration of the health–religiosity connection in the different ethnic origin groups was done by adding interaction terms between the religiosity groups (recoded for this analysis into three groups: secular/partially secular, observant, and religious/orthodox) and ethnic origin affiliation to the regressions.



Table 1 provides the variables' definitions, means, and ranges. Thirty-eight per cent report on being secular, 21% define themselves as partially secular, 29% classify themselves as observant persons, 10% report on being religious, and 2% define themselves as being orthodox.

The mean age of the population under study is 58, with 48% being men. About half of the population report ‘very good’ or ‘good’ economic status, and the mean educational achievement is 11 years of schooling. Seventy-eight per cent are married. New (post-1990) immigrants from the former USSR constitute 11% of the population.

The religiosity, age, ethnic origin, and education distributions match the ones found in other studies and in national statistics.15

The religiosity effect on self-reported chronic morbidity

No differences in CHRON were found across the religiosity groups, controlling for all socio-economic and demographic characteristics. Similar results were obtained with regard to the prevalence (using Logistic models) of the three most common conditions in the population studied: hypertension, heart disease, and arthritis.

The religiosity effect on HRQL and SF-36 eight scales

Table 2 presents the religiosity gradient on self-reported health, adjusting for other personal characteristics. The base category is secular. The bottom row in table 2 provides the P-value for F-testing the existence of such a gradient. For HRQL and GH, no differences exist across the religiosity groups. For the other seven health scales, religiosity has a negative effect on self-reported health. Apart from SF, partially secular persons report similar health levels as secular persons. In role emotional (RE), observant persons do not differ from secular persons, while for MH only the observant persons report lower scores than secular persons. In general, however, the religiosity effect is monotone across the religiosity groups, increasing with increased religious involvement.

View this table:
Table 2

The health–religiosity connection

Other covariates
    F(4,n−19) P*0.4480.3810.00010.0160.0120.0100.0460.0050.072
  • Bold represents parameter significantly (0.05) different than zero

  • *F-test for the existence of a religiosity effect

The largest differences are found in the ‘role-performance’ scales as follows: PF (mobility, activities of daily living), SF (social interactions, mutual visits), RE (lower performance of daily roles because of emotional problems), and role physical (lower performance of daily roles because of physical problems).

Controlling for CHRON as well did not change the conclusions.

The health–religiosity connection in men and women

The results show (data not shown) that differences in gender-specific religiosity effects exist in HRQL and SF only. Religious men tend to report higher scores of HRQL than secular men (3.127 points, t-value = 2.3), while religious women tend to report lower scores than secular women (−2.796 points, t-value = 2.1). In SF, no differences exist among men, but religious and orthodox women report 7–8 points, respectively, less than secular women (t-value = 2.3–3.6, respectively).

The health–religiosity connection in different ethnic groups

Table 3 presents the religiosity gradient across the four ethnic groups. The religiosity groups in this analysis are secular/partially secular (base category), observant, and religious/orthodox. No differences in the religiosity gradient werefound in HRQL, RE, and MH, and they were omitted from the table.

View this table:
Table 3

The health–religiosity connection in ethnic origin groups (t-values in parenthesis)a,b,c

ObservReligs and orthdxObservReligs and orthdxObservReligs and orthdxObservReligs and orthdxObservReligs and orthdxObservOrthdxObservOrthdx
F(6,n−21) P*0.0060.0220.0050.00950.0270.0250.050
  • Bold represents parameter significantly (0.05) different than zero

  • *F-test for the existence of a religiosity effect (main and interactions effects)

  • a: Base religiosity category: secular and partially secular

  • b: No differences in the religiosity gradient across ethnicity groups were found in HRQL, RE, and MH

  • c: Covariates: all socio-demographic variables in table 1

Among second generation Israelis, no differences in health exist between secular/partially secular persons and observant persons, and religious/orthodox persons report lower health status than the secular/partially secular persons in CHRON (more chronic conditions on average) and in PF.

Among Jews from European-American origin, observant persons report worse health than secular/partially secular persons on all the measures, but religious/orthodox persons tend to report the best health status, although the difference is significant only in GH. We note that this is the only evidence on a protective effect of religiosity.

Among Jews from Asian-African origin, the health of observant persons is similar to that of secular/partially secular persons, but religious/orthodox persons report the worst health on the SF-36 domains.

Among post-1990 immigrants from the former USSR, the only health–religiosity differential is the higher prevalence of chronic conditions (CHRON) among observant persons compared with secular/partially secular persons. Combing observant persons with religious/orthodox persons to raise the power did not change the results. Similar effect is found among European-American Israelis.


In general, no religiosity effect on reported chronic morbidity was found. However, religious persons from European-American and Russian origin tend to report more chronic conditions than seculars. Earlier research documented a lower cardiovascular morbidity among religious persons, yet the evidence is generally mixed. Some of the differences in the findings might be related to the use of incidence (flow) versus prevalence (stock) of chronic conditions, and the lack of significant differences might result from the small numbers of sick persons.

The results indicate that in the general population, unlike the findings of Anson et al.10 from 1991, religiosity has an adverse effect on self-reported health as measured by the VAS and the SF-36 tools, in particular among women and among Jews from Asian-African origin. The adverse effect of religiosity on health among women might be explained by exhaustion, resulting from the higher demands on religious women's time and energy, leading also to a less responsible behaviour with regard to weight watching, physical exercising, and the use of preventive medicine (e.g. mammography). Jewish women are also less involved in the synagogue life and spiritual work. For religious persons from Asian-African ethnic origin, it seems that their belief in Divine determinism and justice might have induced a general feeling of dependency, passivity, and frailty expressed in their health reporting.

If religious persons report worse health, how does this reconcile with the earlier (Israeli and American) evidence on the protective effect of religiosity in terms of mortality in the general population? First, it must be noted that a similar contradiction exists in the gender–health connection: women live longer than men but consistently report worse health status and greater use of medical care, even after controlling for reproductive problems. A vast literature tried to explain the gender–differential in reported-health1921 focusing on reporting heterogeneity and on the different social roles occupied by men and women. Similarly, Sen,22 observing higher life expectancy as well as higher reported morbidity in the US than in India, argues for the examination of self perception of illness in a social context, and in particular taking note of what is considered ‘normal’ or ‘natural’ states of being versus clinically preventable or treatable conditions.

The reconciliation of the evidence that religious persons live longer and their worse scores on the SF-36 (in particular among men) might follow similar lines. It is based on the finding that the largest differences were found in the SF-36 role-performance scales, and on possible different perception of the nature of ‘normal life’ and ‘human roles’. Data from the 2002 Social Survey of the Israeli Central Bureau of Statistics (www.cbs.gov.il/socialsurvey) showed that while 62% of the adult secular population did some paid work during the week preceding the survey, that rate was 48% among observant and religious persons, and only 32% among orthodox persons. In other words, secular persons are engaged in more fixed and non-flexible roles in everyday life. If religious and orthodox men view their main roles as continuous study, prayer and approaching the Divine, naturally, earthly emotions, pain, and physical fatigue might be seen as disturbances and distraction, while similar emotional and physical states are not considered such a hindrance for secular ‘normal life’, jobs, and other everyday roles. A support for that argument might be the finding that religious men actually report higher HRQL than secular men, but report lower levels on the SF-36 scales, and in particular, the role-performance scales. Among religious men of Asian-African ethnic origin, lower reported levels on the role-performance scales might be also related to the traditional gender-related division of instrumental roles found in the households of that ethnic origin.

The HRQL-protective effect of religiosity among men (and its adverse effect among women) might be related to their higher practice of religious rites and studies, and their greater involvement in the synagogue life than religious women.

A practical conclusion is that, as in the case of the use of subjective evaluations of health across men and women or across cultures, the SF-36 scales might suffer from religiosity-related reporting heterogeneity, which affect their survival predicting power.


The results show that religiosity has a consistent adverse effect on reported health among Israeli urban Jews aged 45–75 years. This is particularly clear among women and Jews of Asian-African origin. The apparent contradiction between higher life expectancy enjoyed by religious Jews and worse (reported) health cannot be fully understood until a longitudinal study—collecting data on reported health, health behaviour, chronic morbidity, and mortality—is conducted. However, reporting heterogeneity, differences in what is considered ‘normal life’ or ‘human roles’ and what might disturb it between religious and secular persons, might have led to the observed health–religiosity gradient in the data. From a public health point of view, however, the findings indicate that religious Jewish women and religious Jews of Asian-African origin might be particularly vulnerable and at higher relative health risk.

Key points

  • The study question is to explore the connection between self-reported health and religiosity among Israeli Jews, using several self-reported health measures.

  • While no religiosity gradient was found in reported chronic morbidity, religious persons generally report worse health than secular persons on the other measures.

  • The gap is larger in the SF-36's role-performance scales, and among women and Israelis from Asian-African origin.

  • The mixed results are consistent with the ambiguity of the religiosity effect on health reported in recent surveys.

  • The SF-36 health measures might suffer from a religiosity-related reporting heterogeneity, distorting their association with mortality in the population.


I have benefited from comments made by Victor Fuchs, participants of the School of Public Health seminar, and from very useful discussions with Jeremy Kark and Judith Shuval. Any shortcoming is, however, mine. This study was partially funded by the Israeli National Institute for Health Policy Research. Two reviewers provided very useful comments.


View Abstract