IJE Advance Access originally published online on October 30, 2007
International Journal of Epidemiology 2007 36(6):1214-1221; doi:10.1093/ije/dym214
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Does the predictive power of self-rated health for subsequent mortality risk vary by socioeconomic status in the US?
1Center for Social Epidemiology and Population Health, University of Michigan, Ann Arbor, MI 48104-2548, USA.
2Present address: Population Studies Center, University of Michigan, Ann Arbor, MI 48104-2548, USA.
*Corresponding author. Center for Social Epidemiology and Population Health, 1214 S. University Ave, 2nd Floor, University of Michigan, Ann Arbor, MI 48104-2548, USA. E-mail: jenndowd{at}umich.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Background The purpose of this study is to test whether the predictive power of an individual's self-rated health (SRH) on subsequent mortality risk differs by socioeconomic status (SES) in the United States.
Methods We use the National Health Interview Survey 1986–94 linked to Multiple Cause of Death Files 1986–97 (NHIS–MCD). Analyses are based on non-Hispanic Black and White adults 25 and older (n = 358 388). Cox proportional hazard models are used to estimate the effect of SRH on mortality risk during follow-up. Interactions of SRH and level of education and SRH and level of income are used to assess differences in the predictive power of SRH for subsequent mortality risk.
Results The effect of SRH on subsequent mortality risk differs by level of education and level of income. Lower health ratings are more strongly associated with mortality for adults with higher education and/or higher income relative to their lower SES counterparts.
Conclusions Our findings suggest that individuals with different education or income levels may evaluate their health differently with respect to the traditional five-point SRH scale, and hence their subjective health ratings may not be directly comparable. These results have important implications for research that tries to quantify and explain socioeconomic inequalities in health based on self-rated health.
Keywords Self-rated health, socioeconomic status, mortality, predictive ability
Accepted 26 September 2007
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Self-rated health (SRH) is a commonly used outcome measure in research on socioeconomic inequalities in health. Most SRH assessments ask respondents to rate their current health on a five-point scale ranging from poor to excellent. Its popularity lies in its simplicity and strong predictive power of future morbidity and mortality, even controlling for current measures of health such as chronic conditions and functional limitations.1–3 Though the evaluation process that individuals use to assess their health is still poorly understood, this global health rating appears to be more inclusive than current physical status, encompassing assessments of health behaviours, psychological well-being, trajectories in health over time and social well-being.4,5
With the widespread use of SRH to measure the magnitude of socioeconomic inequalities in health and test potential pathways responsible for these inequalities, it is crucial to know whether SRH corresponds to ‘true’ health status in the same way for different socioeconomic groups. Many studies employing SRH as an outcome in the health inequalities literature justify its use based on SRH's ability to predict future mortality in the population as a whole.6–8 If the association between SRH and objective health measures such as mortality differs significantly across comparison groups, whether due to heterogeneous reporting standards across groups or differences in the content of the health evaluation, it would call into question the validity of SRH as an outcome for analysis of health inequalities. To our knowledge, no study has tested whether the relationship between the traditional five-point SRH scale and mortality varies by education or income among US adults.
It is important to emphasize that differences across socioeconomic status (SES) in how individuals evaluate their health may affect not only the measurement of health inequalities, but also attempts to understand the causes of those inequalities. If subgroups report different levels of health for a given level of ‘true’ health, empirical tests of pathways such as health behaviours will always fail to explain the portion of health inequalities due to these reporting differences, no matter how well the mechanism may explain group differences in ‘true’ health. This has the potential to lead researchers to the wrong conclusions about what mechanisms are or are not important in explaining health inequalities.
Theoretically, there is reason to think that SRH evaluations may differ by social and cultural context. Individuals may understand and use the ordinal response scales in systematically different ways such as the propensity to use extreme categories, or general optimism and pessimism. If one group has consistently stricter standards for what is considered ‘excellent’ health for instance, they will report systematically worse health than other groups. This type of uniform shift in the thresholds used to define response categories is referred to as an ‘index shift’, while any change in the relative positions of reporting thresholds such as distances between categories, is referred to as a ‘cut-point shift’.9 This problem has been referred to as ‘state-dependent reporting bias’,10‘response category differential item functioning’11,12 and ‘reporting heterogeneity’.13 Figure 1 illustrates a hypothetical example of reporting heterogeneity with respect to the five-point self-reported health scale by SES. Thresholds for reporting each health category for the high SES group are shifted to the left, and the relative positions of the thresholds also differ by SES group, an example of a ‘cut-point shift’.
|
Empirically, problems in the comparability of SRH scales have been found across countries, with growing evidence that identical health status translates into different self-assessed thresholds in different countries.14–17 Evidence of reporting heterogeneity for subjective health measures across SES groups within countries has been found in the UK, France, Canada, Israel and the US using a variety of benchmarks for ‘objective’ health.9,13,18–21
With respect to mortality, a number of recent studies have found significant race/ethnic and gender differences in the predictive effect of SRH in the US.22–25 For instance, SRH is a weaker predictor of mortality for less-acculturated Hispanic adults, which some have suggested is due to a tendency to put a greater weight on non-life-threatening issues such as emotional and mental well-being in their health assessment,22 though language and cultural differences in response styles are also plausible. SRH has also been shown to be a stronger predictor of mortality for men than for women, although some studies found no significant difference or even a stronger effect for women compared with men.2,24,26–30
Fewer studies have examined socioeconomic differentials in the health evaluation process by comparing the effect of SRH on mortality across different levels of SES, and all of these studies involve European countries. Using the Swedish Survey of Living Studies, Burstrom and Fredlund found that the relative relationship between SRH and mortality is stronger for those in higher occupational classes. However, they report that the absolute mortality risk differences for people reporting poor versus good health were similar across occupational class.31 Doorslaer and Gerdtham, using the same Swedish data, found that the effect of SRH on mortality risk does not vary by income or education.32 Using the German Socio-Economic Panel (SOEP), Jürges found a significant interaction effect of SRH and income on mortality for women but not for men, with poor health being a weaker predictor of mortality for higher income women.33 A recent study by Huisman et al. from the Netherlands found that the predictive ability of SRH for mortality was greater for men in the highest educational groups compared to the lowest, with no differences for women.34
The current international evidence on this issue is decidedly mixed. The presence, absence and magnitude of reporting heterogeneity by SES is likely to vary across countries. To our knowledge, only one study has examined differential effects of SRH on mortality by SES in the US, using a different measure of self-rated health. Using the 1987 National Medical Expenditure Survey, Franks, Gold and Fiscella found that subscales ranging from 0 to 100 measuring health perceptions, physical function, role function and mental health are more predictive of mortality for individuals with a higher educational attainment.30 The current article tests whether SRH measured on the traditional five-point scale differs in predicting future mortality risk by education and income groups in the U.S.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Study population
The analyses are based on the National Health Interview Survey linked to Multiple Cause of Death (NHIS–MCD) dataset. The NHIS is an annual cross-sectional survey of households that produces a sample representative of the US civilian non-institutionalized population ages 0–99 for the purpose of tracking long-term trends in disease and disability prevalence in the US. Respondents 18 years and older who were interviewed from 1986 to 1994 were linked to data in the National Death Index (NDI) for years 1986–97. The NHIS and NDI are linked using a probabilistic matching algorithm based on 12 criteria to determine the vital status of all eligible NHIS participants. It is estimated that the matching methods correctly identify 99% of all living NHIS respondents and over 97% of those who died.35 We limit our analysis to adults 25 and older because a high proportion of younger adults have not completed their education. We also limit our analysis to non-Hispanic Whites and non-Hispanic Blacks and non-proxy interviews. The sample used in the analyses thus comprises 358 388 individuals with 36 466 deaths during the follow-up.
Measures
SRH is measured on a five-point scale (1 = excellent, 2 = very good, 3 = good, 4 = fair and 5 = poor), based on the question ‘Would you say that your health in general is excellent, very good, good, fair or poor?’ We use both the entire five-point scale, with dummy variables for each category, and also test the commonly used dichotomized indicator of 1 = poor/fair health and 0 = excellent/very good/good. We examine all-cause and cause-specific mortality, grouped into five categories based on the ninth revision of the International Classification of Diseases (ICD): cardiovascular disease and diabetes (ICD 390–448, 250), respiratory diseases (ICD 480–496), cancers (ICD 140–239), external (ICD 800–969) and other causes.
Education is measured as the highest completed year of schooling. We recode this measure into three categories: less than high school, high school completion and more than high school. Household income is reported in 26 categories in the NHIS. We code income as the midpoint of each category (using $65 000 for the incomes above $50 000), adjust for inflation using the consumer price index to 1986 dollars, and then divide into quartiles. Models using completed years of education and household income as continuous variables yielded results similar to those presented here (available from the authors).
Controls for existing health conditions include major activity limitations (0 = no limitation, 1 = limited), number of bed days in the past year (0 = no bed days, 1 
0 bed days), restricted activity in the past 14 days (1 = restricted or 0 = not restricted), and BMI from self-reported height and weight (<18, 18–24.9, 25–30, >30). Other covariates include age, sex, race (non-Hispanic Black and non-Hispanic White), census region of residence, marital status and household size.
Statistical analysis
We use Cox proportional hazards models to estimate the effect of SRH, education, income and their interactions on the hazard of dying during the follow-up period. Time to death is measured in years. Proportionality for all predictors was examined using log-minus-log survival plots. The plots showed no severe departures from proportionality, with a slight tendency toward converging hazards for education, income, SRH and sex and a slight divergence with time for race. All analyses employ weights provided by the NCHS and correct for complex survey design using SAS-callable SUDAAN Version 9.0 (Research Triangle Institute, USA).
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
15.7% of respondents reported fair or poor health at the interview, and 10.2% of the sample died during follow-up (Table 1). For all-cause and cause-specific mortality, poor/fair health is a stronger predictor of subsequent mortality for adults with more education (Table 2). The relative risk of all-cause mortality for those reporting poor/fair health versus excellent/very good/good health is 2.82 (95% CI: 2.67, 2.98) for those in the highest educational category, 2.26 (95% CI: 2.16, 2.37) for those in the middle education category, and 1.79 (95% CI: 1.73, 1.86) for those in the lowest education category. While controlling for activity limitations and BMI lowers the overall effect of SRH on mortality somewhat (Table 2, Model 2), the pattern of effects by education is similar. Analysis by cause of death confirms that these results are not driven by the differential distribution of cause of death by SES.
|
|
A similar pattern is evident by income (Table 3). Individuals in the top income quartile who report fair or poor health have 3.65 (95% CI: 3.33, 3.99) times the risk of dying relative to those who report excellent/very good/good health, while individuals in the third, second and bottom quartile who report fair or poor health have 2.69 (95% CI: 2.51, 2.88), 2.14 (95% CI: 2.02, 2.27) and 1.80 (95% CI: 1.73, 1.87) times the risk of dying, respectively, relative to those who report excellent/very good/good health (Model 1). The results are again robust to controls for objective health and hold across all causes of death.
|
Due to the large number of interactions, Figures 2 and 3 summarize models estimated with dummy variables for each category of the five-point SRH scale (excellent as reference), showing the relative effect of SRH on all-cause mortality by education and income, respectively. The effect of reporting good, fair or poor health compared with excellent health on subsequent mortality risk is lower for individuals with less education and lower income. Results by cause of death and controlling for activity limitations were similar to the results from the dichotomized SRH models.
|
|
We conducted additional analyses to ascertain the robustness of our findings. The length of follow-up does not appear to play a large role in the findings. We estimated models that restricted the time to follow-up to 0–4 years and found little difference from the follow-up of up to 11 years. Models stratified by race, sex and age all show the same substantive and significant differences whereby poor health for high-SES individuals is more strongly predictive of mortality than for lower-SES groups. The differences across SES appear somewhat stronger for men than for women, and somewhat stronger for Whites, compared with Black adults. The patterns across age are inconsistent: income appears to have a stronger effect on the SRH coefficients at higher ages, while education has a stronger effect at younger ages (results available upon request).
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
Because of the widespread use of SRH in public health and social science research, especially in discussions of socioeconomic disparities in health, it is important to fully understand how much of the SES differences in SRH are attributable to ‘true’ health differences and how much are due to differences in how individuals report their health according to their own norms and expectations. This article takes one approach to testing whether SRH corresponds differently to objective health by SES by examining SES differentials in the way SRH predicts mortality. We find large differences in the predictive power of SRH by education and income, with lower health ratings showing a stronger association with mortality risk for individuals with more education and/or higher income. The results are consistent across different causes of death and with controls for some objective health measures. These findings are in contrast to results from Sweden which found no significant interactions with education, income and trichotomized self-reported health with respect to follow-up mortality,32 but consistent with recent work from The Netherlands showing a significant interaction in men for the highest versus lowest educational group.34 The extent to which SES groups differ in their use of categorical health scales may vary from country to country.
As in the Dutch study,34 our results hold when looking at absolute differences in mortality rates by SRH across SES groups. Comparing absolute differences in age- and sex-adjusted mortality rates between those with poor/fair health versus those with good, very good or excellent health across income and education groups, we found larger absolute differences in mortality for those in the higher education and higher income groups (17% and 25% differences, respectively). Our results are consistent with previous work in the US using the 1987 National Medical Expenditure Survey which found that subscales ranging from 0 to 100 measuring health perceptions, physical function, role function and mental health were more predictive of mortality for individuals with a higher educational attainment.30 Our findings are also consistent with results from NHANES III in the US indicating that higher income people may have more lenient standards for their health, reporting physical difficulties or inabilities only when they reach a more severe level of tested limitation.21,36 In our analysis, this type of ‘index shift’ would correspond to the fact that a person with more education or higher income would have to be in worse health, as measured by mortality risk, in order to rate themselves in a lower SRH category compared with someone with less education or less income.
An important limitation of this study is the fact that mortality is not the only way to measure objective health status, and our analysis cannot distinguish whether SRH is less predictive of mortality for lower SES individuals because of heterogeneous reporting standards or because the health evaluation process of lower SES individuals incorporates more non-fatal ailments such as chronic pain or poor mental health. Nonetheless, mortality is arguably the most objectively measured physical health outcome available, and the large literature on the predictive power of SRH for mortality1 is often cited as the primary justification for its use as a summary health measure in a broad range of research. Thus the finding that SRH does not predict mortality as well at lower levels of SES has important implications for research on health inequalities that has assumed the magnitude of SES inequalities in SRH corresponds well to SES inequalities in mortality.
One interpretation of our findings is that individuals with higher education and higher income may have more health knowledge and/or contact with health services, making them more accurate predictors of their own mortality risk via SRH ratings. If this is the case, it does not necessarily reflect reporting bias in the use of the scale, but rather differential measurement error across groups. This would support the notion that a single measure of SRH does not correspond to ‘objective’ health as well for lower SES individuals.
Another possible interpretation of our findings is that for the same ‘objective’ level of morbidity, lower SES individuals are more pessimistic in their health ratings because they have fewer material and social resources with which to deal with their conditions, leading to a higher level of suffering at the same mortality risk. If this is true, it would imply that even if poor health ratings do not correspond to the same mortality risk for low SES individuals, they may correspond to the same level of subjective ‘suffering’, and it remains an open question by what metric we wish to measure health inequalities. This scenario is difficult to exclude under any test of ‘true’ versus reporting differences, but our analysis implicitly assumes that there is a ‘true’ and measurable objective health status, for which mortality risk is a good proxy.
While we did not examine specific chronic conditions in this study because of the design of the NHIS prior to 1997, we hope to shed light on some of these questions in future work with new data by testing whether different distributions of chronic conditions and comorbidities can explain SES differentials in SRH by testing whether individuals with the same profile of physical and psychological conditions but different SES report similar health status. This technique has recently been used to try to differentiate ‘true’ versus reporting differences in SRH by gender and across different European countries.16,37
The differences by SES found here are sizeable enough to be of concern in the estimation of health inequalities using SRH in the US and for efforts to test mechanisms responsible for inequalities using this measure. Future work should build on these findings to estimate the size of potential biases in studies of health inequalities that employ SRH as an outcome, helping to inform researchers as to whether suitable corrections can be made, or whether this commonly used measure is ultimately too misleading to be used in the study of health inequalities. Murray et al. outline several strategies for fixing levels of unobserved latent health variables in order to isolate the source of variation in assessments across populations.38 One approach is to use anchoring vignettes, where respondents rate the health of fictitious individuals with given health conditions or disabilities and the evaluations of these fictitious individuals are used to estimate reporting thresholds (or cut-points) which may depend on individual characteristics, and then used to adjust the rating of the respondent's own health.11,39 While these techniques were developed with cross-country comparisons in mind, they could be equally important in ensuring comparability of subjective measures across groups within countries.
In conclusion, we find evidence that the predictive power of SRH for mortality differs by education and income in the US, suggesting either differential reporting standards by SES, or health evaluations that include more non-fatal risk factors or mental health components for low SES groups. These results suggest caution in using SRH to quantify and explain health inequalities by education and income in the US, and also suggest urgency in implementing new techniques to ensure comparability of subjective health measures so often utilized in social science and public health research.
|
Acknowledgements |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
J.B.D. acknowledges financial support from the Robert Wood Johnson Health and Society Scholars Program at the University of Michigan and A.Z. acknowledges support provided by NICHD grant No. 1 R01 HD053696. The authors thank Bob Hummer and two anonymous reviewers for helpful comments.
Conflicts of interest: None Declared.
KEY MESSAGES
|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAcknowledgementsReferences |
|---|
1 Idler EL, Benyamini Y. Self-rated health and mortality: a review of twenty-seven community studies. J Health Soc Behav (1997) 38:21–37.[CrossRef][Web of Science][Medline]
2 DeSalvo KB, Bloser N, Reynolds K, et al. Mortality prediction with a single general self-rated health question. A meta-analysis. J Gen Intern Med (2006) 21:267–75.[CrossRef][Web of Science][Medline]
3 Idler EL, Kasl SV. Self-ratings of health: do they also predict change in functional ability? J Gerontol B Psychol Sci Soc Sci (1995) 50:S344–53.[Abstract]
4 Ferraro KF, Farmer MM, Wybraniec JA. Health trajectories: long-term dynamics among black and white adults. J Health Soc Behav (1997) 38:38–54.[CrossRef][Web of Science][Medline]
5 Schnittker J. When mental health becomes health: age and the shifting meaning of self-evaluations of general health. Milbank Q (2005) 83:397–423.[CrossRef][Web of Science][Medline]
6 Deaton AS, Paxson CH. Aging and inequality in income and health. Am Econ Rev (1998) 88:248–53.[Web of Science]
7 Kunst AE, Bos V, Lahelma E, et al. Trends in socioeconomic inequalities in self-assessed health in 10 European countries. Int J Epidemiol (2005) 34:295–305.
8 Power C, Matthews S, Manor O. Inequalities in self-rated health: explanations from different stages of life. Lancet (1998) 351:1009–13.[CrossRef][Web of Science][Medline]
9 Lindeboom M, van Doorslaer E. Cut-point shift and index shift in self-reported health. J Health Econ (2004) 23:1083–99.[Web of Science][Medline]
10 Kerkhofs M, Lindeboom M. Subjective health measures and state dependent reporting errors. Health Econ (1995) 4:221–35.[CrossRef][Web of Science][Medline]
11 King G, Murray CJL, Salomon J, et al. Enhancing the validity and cross-cultural comparability of measurement in survey research. Am Polit Sci Rev (2004) 98:191–207.[CrossRef][Web of Science]
12 King G, Wand J. Comparing incomparable survey responses: evaluating and selecting anchoring vignettes. Polit Anal (2007) 15:46–66.
13 Shmueli A. Socio-economic and demographic variation in health and in its measures: the issue of reporting heterogeneity. Soc Sci Med (2003) 57:125–34.[CrossRef][Web of Science][Medline]
14 Banks J, Marmot M, Oldfield Z, et al. The SES health gradient on both sides of the Atlantic. In: NBER Working Paper 12674 (2006) Cambridge, MA.
15 Kapteyn A, Smith JP, van Soest A. Self-reported work disability in the US and The Netherlands. In: Labor and Demography 0504006 (2005) Santa Monica, CA: EconWPA.
16 Jürges H. True health vs response styles: exploring cross-country differences in self-reported health. Health Econ (2007) 16:163–78.[CrossRef][Web of Science][Medline]
17 Zimmer Z, Natividad J, Lin H-S, et al. A cross-national examination of the determinants of self-assessed health. J Health Soc Behav (2000) 41:465–81.[CrossRef][Web of Science][Medline]
18 Adams J, White M. Is the disease risk associated with good self-reported health constant across the socio-economic spectrum? Public Health (2006) 120:70–75.[CrossRef][Web of Science][Medline]
19 Etile F, Milcent C. Income-related reporting heterogeneity in self-assessed health: evidence from France. Health Econ (2006) 15:965–81.[CrossRef][Web of Science][Medline]
20 Humphries KH, van Doorslaer E. Income-related health inequality in Canada. Soc Sci Med (2000) 50:663–71.[CrossRef][Web of Science][Medline]
21 Melzer D, Lan T-Y, Tom BDM, et al. Variation in thresholds for reporting mobility disability between national population subgroups and studies. J Gerontol A Biol Sci Med Sci (2004) 59:1295–303.
22 Finch BK, Hummer RA, Reindl M, et al. Validity of self-rated health among Latino(a)s. Am J Epidemiol (2002) 155:755–59.
23 Ferraro KF, Kelley-Moore JA. Self-rated health and mortality among Black and White adults: examining the dynamic evaluation thesis. J Gerontol B Psychol Sci Soc Sci (2001) 56:S195–205.
24 Benyamini Y, Blumstein T, Lusky A, et al. Gender differences in the self-rated health-mortality association: is it poor self-rated health that predicts mortality or excellent self-rated health that predicts survival? Gerontologist (2003) 43:396–405.
25 McGee DL, Liao Y, Cao G, et al. Self-reported health status and mortality in a multiethnic US cohort. Am J Epidemiol (1999) 149:41–46.
26 Benjamins MR, Hummer RA, Eberstein IW, et al. Self-reported health and adult mortality risk: an analysis of cause-specific mortality. Soc Sci Med (2004) 59:1297–306.[CrossRef][Web of Science][Medline]
27 Bath PA. Differences between older men and women in the self-rated health-mortality relationship. Gerontologist (2003) 43:387–95.
28 Deeg DJH, Kriegsman DMW. Concepts of self-rated health: specifying the gender difference in mortality risk. Gerontologist (2003) 43:376–86.
29 Strawbridge WJ, Wallhagen MI. Self-rated health and mortality over three decades. Res Aging (1999) 21:402.
30 Franks P, Gold MR, Fiscella K. Sociodemographics, self-rated health, and mortality in the U.S. Soc Sci Med (2003) 56:2505–14.[CrossRef][Web of Science][Medline]
31 Burstrom B, Fredlund P. Self rated health: is it as good a predictor of subsequent mortality among adults in lower as well as in higher social classes? J Epidemiol Community Health (2001) 55:836–40.
32 van Doorslaer E, Gerdtham U-G. Does inequality in self-assessed health predict inequality in survival by income? evidence from Swedish data. Soc Sci Med (2003) 57:1621–29.[CrossRef][Web of Science][Medline]
33 Jurges H. Self-assessed health, reference levels and mortality. In: Paper No. 0457, MEA Discussion Paper Series (2005) Mannheim Research Institute for the Economics of Aging, University of Mannheim.
34 Huisman M, van Lenthe F, Mackenbach J. The predictive ability of self-assessed health for mortality in different educational groups. Int J Epidemiol (2007) 36:1207–13.
35 Horm J. Multiple causes of death for the national health interview survey. In: Record linkage techniques – Proceedings of an International Workshop and Exposition (1997) Washington, D.C: 1997: National Research Council.
36 Iburg K, Salomon JA, Tandon A, et al. Cross-population comparability of physician-assessed and self-reported measures of health. In: Global Programme on Evidence for Health Policy, Discussion Paper No. 14 (2001) Geneva: World Health Organization.
37 Case A, Paxson CH. Sex differences in morbidity and mortality. Demography (2005) 42:189–214.[CrossRef][Web of Science][Medline]
38 Murray CJL, Tandon A, Salomon J, et al. Enhancing cross-population comparability of survey results. In: GPE Discussion Paper No 35 (2001) Geneva: WHO/EIP.
39 Tandon A, Murray CJL, Salomon JA, et al. Statistical models for enhancing cross-population comparability. In: Global Programme for Evidence on Health Policy Discussion Paper No. 42 (2002) Geneva: World Health Organization.
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  T. A. Eikemo, M. Huisman, F. Perlman, and K. Ringdal Educational health inequalities in former Yugoslavia: evidence from the South-East European Social Survey Project Eur J Public Health, December 15, 2009; (2009) ckp200v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. E. Alley, B. J. Soldo, J. A. Pagan, J. McCabe, M. deBlois, S. H. Field, D. A. Asch, and C. Cannuscio Material Resources and Population Health: Disadvantages in Health Care, Housing, and Food Among Adults Over 50 Years of Age Am J Public Health, November 1, 2009; 99(S3): S693 - S701. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Delpierre, V. Lauwers-Cances, G. D. Datta, L. Berkman, and T. Lang Impact of Social Position on the Effect of Cardiovascular Risk Factors on Self-Rated Health Am J Public Health, July 1, 2009; 99(7): 1278 - 1284. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C Delpierre, V Lauwers-Cances, G D Datta, T Lang, and L Berkman Using self-rated health for analysing social inequalities in health: a risk for underestimating the gap between socioeconomic groups? J Epidemiol Community Health, June 1, 2009; 63(6): 426 - 432. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S V Subramanian and K. Ertel Author's Response: Self-rated health may be adequate for broad assessments of social inequalities in health Int. J. Epidemiol., February 1, 2009; 38(1): 319 - 320. [Full Text] [PDF]
|
|
|
|
|
|
T. Bago d'Uva, O. O'Donnell, and E. van Doorslaer Differential health reporting by education level and its impact on the measurement of health inequalities among older Europeans Int. J. Epidemiol., December 1, 2008; 37(6): 1375 - 1383. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Subramanian and K. Ertel Is the use of self-rated health measures to assess health inequalities misleading? Int. J. Epidemiol., December 1, 2008; 37(6): 1436 - 1440. [Full Text] [PDF]
|
|
|
|
|
|
J. B. Dowd and A. Zajacova Author's response Int. J. Epidemiol., December 1, 2008; 37(6): 1438 - 1439. [Full Text] [PDF]
|
|
|
|
|
|
A. Singh-Manoux, M. J Shipley, M. Zins, and J. E Ferrie Author's Response Int. J. Epidemiol., December 1, 2008; 37(6): 1439 - 1440. [Full Text] [PDF]
|
|
|
|
|
|
S. Ebrahim Uses of epidemiology, ways of living and dying. Int. J. Epidemiol., December 1, 2007; 36(6): 1159 - 1160. [Full Text] [PDF]
|
|
|
|
|
|
A. Quesnel Vallee Self-rated health: caught in the crossfire of the quest for 'true' health? Int. J. Epidemiol., December 1, 2007; 36(6): 1161 - 1164. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||