This was a population-based study of the relationship between educational level and self-reported health in Tehran, Iran. There was a distinct pattern of self-reported health among those with different educational levels, showing a dose-response relationship between education and the risk of 'less than good' self-rated health status. However, one might argue that a sample from the urban capital (Tehran) is not necessarily representative of the entire country. In general this is true, but since Tehran has became a multicultural metropolitan area with a mixture of different socioeconomic and ethnic backgrounds, a sample from the general population in Tehran could at least be regarded as representative of the urban population of Iran . It has been suggested that studies of self-reported health including quality of life assessments provide information on the health of a population that are usually invisible in traditional analyses of population health . Responses to questions of this type (self-reported health or self-reported morbidity) may vary according to mode of administration, but studies have shown that such responses are nevertheless valid .
The findings from the present study indicate that in general people with a higher educational level rated their health status more highly than people with a lower educational level. There are several explanations for education-related health inequalities within and between countries. The most straightforward is that the effect of education varies from one place to another for unknown reasons, which would make it difficult to account for differences in the magnitude of the association between education and various health indicators in different countries . In addition, international studies have shown that educational health inequalities vary in magnitude between countries . A recent publication from the Eurothine Project , which compares inequalities in health in more than 20 European countries, suggests that although income and education are 'upstream' determinants of health inequalities, this is the feature of European welfare regimes that could provide evidence for the magnitudes of educational health inequalities between countries. They found that South European welfare regimes had the largest health inequalities while countries with Bismarckian welfare regimes tended to have the smallest. Although the other welfare regimes ranked relatively close to each other, the Scandinavian regimes were placed less favorably than the Anglo-Saxon and East European .
Another possible explanation is that in societies where everyone has the same access to education, educational level is not a good indicator of socioeconomic position. Thus, even if there is an inverse relationship between education and self-reported health, this does not demonstrate inequality in health in such societies . On the other hand, it can be argued educational attainment could reflect dissimilarities among individuals in terms of work conditions, economic status, lifestyle and the use of health care services, so education has a significant impact on the observed inequalities in health among different socioeconomic subgroups of populations . Others have suggested that educational inequalities in health might be attributable to the fact that education reflects the different life course accumulations of material and psychosocial hazards to which people have been exposed . Since not everyone has the same access to education in Iran, we suspect that the latter explanation applies to the variation in self-reported health by educational level. Interestingly, a study using data from the 2003 US Current Population Health Survey indicated that people in the very high income bracket tend to report slightly worse health, which may be explained by their lower education .
We found that women rated their health more poorly than men. This was definitely not due to age since women were younger than men on average. Thus one might argue that education contributes to the observed differences between men and women in self-reported health. Studies have shown that education is one of the most important contributors to gender inequalities in health . There are also several other explanations for such observed differences: economic dependence, employment, marital status, family position and family demands are among the factors found to contribute to gender differences in self-rated health [25, 26]. However, it is argued that diversity in life style is not the most important reason for gender differences in social inequalities in health .
We found that women reported having significantly more chronic medical conditions. Studies from some other developing and transitional countries have yielded similar results, with some exceptions [28, 29]. For instance, a Syrian study identified gender-specific determinants of poor self-rated health including being married, low socioeconomic status, and not having social support for women; and smoking, and low physical activity for men . However, the present study showed that married men and particularly married women were worse off (in terms of health) than their single counterparts. One might argue that the married were merely older than the single, and being older implies worse health.
In addition to gender, the results from the present study clearly indicate that age has an independent effect on self-reported health. It is therefore argued that age, sex and social class make distinct contributions to specific morbidities and should be recognized as a transparent and robust approach to the assessment of morbidity-based inequality .
Finally, the limitations of this study should be considered in interpreting the results. The design was cross sectional and therefore could not indicate whether education really causes inequality in self-rated health or whether existing inequality due to other factors causes poorer health, and this in turn leads to lack of success in appropriate education. Furthermore, we used binary logistic regression analysis and dichotomized a 5-level ordinal scale to yield 'good or better than good' and 'less than good' self-rated health. Thus, excellent, very good and good self-ratings of health (for example) are assumed to be the same, but in fact they are not. There are other ways of analyzing such data and tackling this problem, for example by multinomial regression analysis, where ordinal data can be used without collapsing categories. However, each of these approaches to analysis has its own limitations. The study did not collect data on other measures such as health behaviors (diet, exercise, smoking, etc.) or measures such as blood pressure or body mass index. The contributions of these variables to self-rated health remain unknown. Collecting such data is recommended for future studies. We used years of formal education as a measure of socioeconomic position; it would be useful if reliable data on income could be collected for future investigations on the topic.