Article Text

Download PDFPDF

Using self-rated health for analysing social inequalities in health: a risk for underestimating the gap between socioeconomic groups?
  1. C Delpierre1,2,
  2. V Lauwers-Cances1,
  3. G D Datta3,
  4. T Lang1,
  5. L Berkman2
  1. 1
    Inserm U558, Toulouse, France
  2. 2
    Harvard School of Public Health, Department of Society, Human Development and Health, Cambridge, Massachusetts, USA
  3. 3
    Harvard School of Public Health, Department of Epidemiology, Cambridge, Massachusetts, USA
  1. Dr C Delpierre, INSERM U558, 37 allées Jules Guesde 31073 Toulouse, France; delpier{at}cict.fr; cdelpier{at}hsph.harvard.edu

Abstract

Background: The use of self-rated health (SRH) for measuring health inequalities could present some limits. The impact of the same disease on SRH could be different according to health expectations people have which are associated with social characteristics. The aim of this study was to analyse the link between physical health status and SRH, according to level of education.

Method: Data from the National Health and Nutrition Examination Survey for the years 2001–4 were used. Multivariate logistic regression analyses were performed for assessing the relation between health status and SRH according to educational level.

Results: The sample consisted of 4661 men and 4593 women. Reporting functional limitation was associated more strongly with poor SRH in higher educated women than in lower educated women (OR, 8.73, 95% CI 5.87 to 12.98 vs OR, 3.97, 95% CI 2.93 to 5.38 respectively), as was reporting respiratory disease (OR, 5.17, 95% CI 3.67 to 7.30 vs OR, 2.60, 95% CI 1.72 to 3.95 respectively), cardiovascular disease (OR, 9.79, 95% CI 6.22 to 15.40 vs OR, 3.34, 95% CI 2.29 to 4.87 respectively) and dental problems (OR, 4.37, 95% CI 3.22 to 5.92 vs OR, 2.58, 95% CI 1.97 to 3.39 respectively). Reporting functional limitation was associated more strongly with poor SRH in higher educated men than in lower educated men (OR, 7.71, 95% CI 5.04 to 11.79 vs OR, 4.87, 95% CI 3.30 to 7.18 respectively), as reporting oral problems (OR, 2.62, 95% CI 1.84 to 3.74 vs OR, 3.63, 95% CI 2.81 to 4.68 respectively).

Conclusions: The impact of health problems on SRH is stronger among better educated individuals. This phenomenon could lead to an underestimate of the health inequalities across socioeconomic groups.

View Full Text

Statistics from Altmetric.com

Since the original study by Mossey and Shapiro in 1982,1 many studies have shown that self-rated health (SRH) is associated with mortality, in various populations, even after adjusting for a variety of factors.26

This measure, based on a single-item, is attractive because it is easy to use and refers to a wide, multidimensional definition of health.3 However, SRH is strongly associated with physical and mental health status.79 Singh-Manoux et al7 showed in two large studies that physical and mental health status explained 34.7% and 41.4% of the variance in SRH respectively, suggesting that SRH could be a reasonable measure of objective health status.

However, as reported by Adams and White10 SRH may also result from a balance between one’s actual health status and the best health that one could expect for oneself. And expectations that people have about their health can vary according to several factors, such as socioeconomic status (SES), cultural or social issues, which can lead to differences in reporting health status for the same “objective” health status, and constitute a limitation of using SRH as an objective health measure. According to comparison processes theory, people compare themselves with others with whom they are socially alike.11 12 As shown by Ross and Van Willigen,13 socially advantaged groups, such as the well educated, have the highest expectations about their quality of life and health. Therefore, in the presence of the same illness, more highly educated people may experience a greater negative impact on their perceived health than lower educated people in whom expectations are lower. This phenomenon could lead to an underestimation of the health inequalities existing between socioeconomic groups when using SRH as an indicator of health.

Three recent papers have shown that the predictive value of SRH on subsequent mortality differed according to level of education: two studies showed that the predictive value was higher for high educated people14 15 and one showed the opposite result.16 But to our knowledge, little work has investigated how SES might modify the association between physical health status and SRH. The aim of this study was thus to assess the link between physical health status and SRH, according to socioeconomic position.

METHODS

Study population and sample design

We used data from the 2001–2 and 2003–4 National Health and Nutrition Examination Survey (NHANES), described in detail elsewhere.17 In brief, NHANES is a cross-sectional, nationally representative survey of the civilian non-institutionalised population of the USA. The NHANES design is a stratified, multistage, probability sample. This multistage sample is based on a selection of counties, blocks, households and persons within households. Data are collected through an in-person home interview and a physical examination, conducted by physicians, at mobile examination centres.

Of the 25 917 people selected in the study, 21 161 (81.6%) people completed the in-person home interview, and 20 120 (77.6%) subsequently completed a physical interview/examination. We restricted our analyses to respondents who completed the home interview and the physical interview/examination. We excluded participants younger than 20 years (n = 10 351) and pregnant women (N = 515). The final sample consisted of 4661 men and 4593 women.

Self-rated health

Individuals rated their health as excellent, very good, good, fair or poor. The responses were dichotomised in our analyses: individuals reporting excellent, very good or good health were classified as having good SRH and those reporting fair or poor health as having poor SRH.

Socioeconomic position

Socioeconomic position was assessed by using level of educational attainment, categorised as: less than 12 years, 12 years and more than 12 years.

Physical health status

Though we acknowledge that mental health conditions are likely to impact SRH, we have limited this analysis to physical conditions. Because SRH is a measure of general health, we limited our analysis to chronic conditions that might impact one’s perception of one’s overall health status.

Chronic pain was assessed using two questions: (1) experience of pain that lasted more than 24 hours during the previous month, and (2) duration of pain. Participants who reported experiencing pain for at least 1 month were considered to have a chronic pain.

Chronic disease prevalence was assessed by the self-reported presence of respiratory disease (asthma, chronic bronchitis, emphysema), or a history of cancer or cardiovascular disease (congestive heart failure, coronary diseases, angina pectoris, heart attack, stroke). Functional limitations were self-reported and assessed with scales of activities of daily living (ADLs), instrumental activities of daily living (IADLs), mobility and upper/lower body strength, and social participation. Participants who reported some difficulty, much difficulty or who were unable to do one of the activities were considered to have functional limitations.

Data about vision, hearing and oral health were considered for evaluating the presence of other chronic conditions. Participants who reported trouble hearing or who were deaf were considered to have a hearing deficiency. Participants in whom distance vision was assessed to be worse than 20/25, upon medical examination, in either eye with their current correction (no correction, distance glasses and/or contact lenses) were classified to have visual deficiency. Participants who reported their mouth/teeth to be in fair or poor condition were considered to have oral health problems.

Covariates

The sociodemographic variables used in the analyses were age, gender, ethnicity (non-Hispanic white, non-Hispanic black, Mexican–American, or other), marital status (married/living with partner, single, divorced/separated, widowed) and health insurance (no insurance, public, private).

We were interested in assessing intermediates between education and SRH. Thus we also constructed models including smoking status, alcohol consumption, physical activity and body mass index (BMI).

Smoking status was coded as never, former or current from self-reports. A current smoker was defined as someone who had smoked at least 100 cigarettes during his or her lifetime and still smoked. A former smoker was someone who had smoked at least 100 cigarettes during his or her lifetime but no longer smoked. A participant who had not smoked at least 100 cigarettes was considered to have never smoked.

Abstainers were defined as individuals who reported, during the medical examination, not drinking in the previous year or those who reported drinking on fewer than 12 occasions in the past year. Drinkers were defined as those who reported drinking on at least 12 occasions in the previous year and were classified according to the average daily volume of alcohol consumption: 0, <1, ⩽2, >2 drinks/day.

Participants reported their moderate (ie, walking, bicycling for pleasure), vigorous (ie, running, lap swimming) or muscle-strengthening activities over past 30 days. Individuals who practised one of these activities were classified as active and those who reported neither vigorous nor moderate nor muscle-strengthening activities as non-active.

BMI was calculated as weight (kg)/height (m2), which were measured during the physical examination. Overweight was defined as a BMI of 25.0–29.9 and obesity as a BMI of 30.0 or higher.

Analysis

Socioeconomic position may operate differentially on SRH in men and women, as observed in several studies which showed different determinants of SRH according to gender.1820 Thus all our analyses were run separately for men and women.

To test if educational level modified the relation between individual physical health conditions and SRH, we constructed a logistic regression model with SRH as the outcome and included terms for education, the health condition, and the interaction between education and the health condition. As a statistical interaction was detected for most conditions, we performed multivariate logistic regression analyses stratified by educational level. Potential confounders and mediators of the relation between physical health status and SRH were introduced in sequence. First, we adjusted for variables likely to be confounders, such as age and ethnicity. Additionally we adjusted for variables that may be mediators through which education may influence the association between physical health status and SRH, for example marital status, health insurance, BMI, smoking, drinking and physical activity.

We used the medical examination clinic sampling weights to produce our weighted estimates and sampling errors (SEs). Sampling weights were used to adjust for non-response bias and the oversampling of black people, Mexican–Americans, low income individuals, adolescents and older people in NHANES. Statistical analyses were performed using SAS (version 9.1; SAS Institute; Cary, North Carolina).

RESULTS

Characteristics of the study sample are listed in tables 1 and 2 separately for men and women.

Table 1 Social characteristics and health behaviours of the study sample; National Health and Nutrition Examination Survey, 2001–4, adjusted for sampling design
Table 2 Health by educational level among men and women; National Health and Nutrition Examination Survey, 2001–4

Lower educational level was associated with poorer SRH, in men and women (table 2), regardless of physical health status (tables 3 and 4). Men with lower level of education were more likely to have functional limitations, cardiovascular disease, eye and oral health problems than men with more than 12 years of education (table 2). There were no differences in the proportion of people with chronic pain, cancer, respiratory disease and hearing problems according to educational level. Among women, lower educational level was associated with all the assessed health conditions except for cancer and respiratory disease (table 2).

Table 3 Proportion of people reporting poor self-reported health (SRH) according to the presence or absence of health conditions, by level of education
Table 4 Proportion of people reporting poor self-reported health (SRH), according to the presence of absence of health conditions, by level of education

Tables 3 and 4 show the relation between health status and poor SRH stratified on educational level in men and women.

Among men (table 3), all physical health conditions were associated with SRH at every educational level, with the exception of the relation between cancer, eye problems and SRH. Whatever educational level, reporting cancer was not associated with poor SRH. The proportion of men reporting poor SRH in the presence of functional limitations increased more for those with more than 12 years of education than for those with less than 12 years of education, because of reporting oral problems, though tests for interaction were not significant (interaction test p = 0.06 and p = 0.08 respectively for functional limitations and oral health). After adjusting for age and ethnicity, functional limitations were associated more strongly with poor SRH in more highly educated men than in those with lower educational attainment (OR, 7.71, 95% CI 5.04 to 11.79 for the most educated vs OR, 4.87, 95% CI 3.30 to 7.18 for the least educated), as were oral health (OR, 2.62, 95% CI 1.84 to 3.74 for the most educated vs OR, 3.63, 95% CI 2.81 to 4.68 for the least educated).

Among women (table 4), all physical health conditions, except visual deficiency and cancer, were associated with poor SRH at every educational level. Visual problems were associated with poor SRH only among women with the highest level of education and cancer was not associated with poor SRH in the least educated women. The proportion of women with poor SRH in the presence of functional limitations increased more for those with more than 12 years of education than for those with less than 12 years of education (interaction test p = 0.02). This was also the case for respiratory disease (interaction test p = 0.04), cardiovascular disease (interaction test p = 0.0005) and oral health problems (interaction test p = 0.002). After adjusting for age and ethnicity, functional limitations were associated more strongly with poor SRH in more highly educated women than in those with lower educational attainment (OR, 8.73, 95% CI 5.87 to 12.98 for the most educated vs OR, 3.97, 95% CI 2.93 to 5.38 for the least educated). This was also the case for respiratory disease (OR, 5.17, 95% CI 3.67 to 7.30 for the most educated vs OR, 2.60, 95% CI 1.72 to 3.95 for the least educated), cardiovascular disease (OR, 9.79, 95% CI 6.22 to 15.40 for the most educated vs OR, 3.34, 95% CI 2.29 to 4.87 for the least educated), and oral health problems (OR, 4.37, 95% CI 3.22 to 5.92 for the most educated vs OR, 2.58, 95% CI 1.97 to 3.39 for the least educated). The association between chronic pain and poor SRH was stronger for the most highly educated women than for the lowest educated women, because of the association between eye problem and SRH‘, though the test for interaction was not significant.

All these models were also constructed adjusting for marital status, health insurance, BMI, smoking, drinking and physical activity simultaneously. Adjustment for the full set of covariates led to no significant changes in the difference in odds ratios according to level of education (data not shown).

DISCUSSION

To our knowledge, this study is the first in USA to assess social conditions that could modify the relation between physical health status and SRH. The present study suggests that educational level influences the weight given to certain physical health conditions when reporting their perceived health, and that the impact of physical health conditions on poor SRH is greater among more highly educated people, particularly among women.

Physical health status, except for vision deficiency, was assessed from self-report, with all the limits that this method entails. However, we mainly studied chronic disabling diseases that may be less susceptible to over-reporting. Indeed, self-reports could be reasonably accurate for certain chronic conditions, like cardiac disease, stroke, malignancies, non-specific lung disease.21 22 As reported by Haapanen et al23 the agreement between questionnaire data and medical records could be good for well-known chronic diseases that have clear diagnosis criteria. Idler et al24 showed that the knowledge of a chronic illness strengthened the association between SRH and mortality. Thus, use of patient-reported health conditions may be a valid method to identify some chronic illnesses and may constitute a pertinent indicator to analyse the association between physical health status and SRH. However, future studies should assess the relation between objective health, SRH and social factors. Another limitation is that level of education could impact the self-reports of health conditions. It is possible that the health service access differs according to education and that this access is lower for low educated people. Thus the probability to be diagnosed and to be aware of the presence of a disease could be lower for poorly educated people, which would increase the underestimation of the magnitude of the health inequalities between socioeconomic groups. However results of literature on the association between education level and self-reports are inconsistent.2123 An additional limitation is that only non-institutionalised adults were sampled. Those whose health status resulted in institutionalisation were not included in this study. If institutionalisation because of health problems impacts SRH differently according to educational level, these exclusions may modify the influence of education on the relation between health status and SRH.

Some might consider the use of educational level as the measure of socioeconomic position as an additional limitation. Educational level, in contrast to occupation or income, is a resource that is a personal characteristic,25 is acquired first over the life course, and contributes to occupational class position and income.26 Educational level, the root of individual well-being according to Ross and Van Willigen,13 influences health behaviours and is typically attained before health problems such as functional disability or cardiovascular diseases. Thus reverse causality is less likely with educational attainment than with occupational class or income.

As hypothesised, SRH was positively associated with educational level. For every health condition those with lower educational attainment were more likely than those with higher educational attainment to report poor SRH. This is probably because they are also more likely to have health problems.

However, our results showed that when health problems occur, the repercussions of these problems on SRH are worse in those who are more highly educated. One hypothesis to explain this finding is that the expectations about health increase with level of education. Brouwer and van Exel27 reported that the expectations regarding health and quality of life were higher for highly educated people than for those with a moderate level of education. Ross and Van Willigen13 showed that those with higher expectations about their quality of life were more frequently dissatisfied. Thus, the repercussions of health problems on subjective health may be worse for those with higher health expectations. More highly educated people are also more likely to be aware about the consequences of health problem, in terms of morbidity or mortality risks. Hofman et al28 showed that higher educated people with cancer expected more treatment side-effects than lower educated people. Therefore, in the case of illness, they may expect more problems and so report poor SRH more frequently. We observed the influence of the level of education after adjustment on different covariables. As shown by Huisman et al,15 it seemed that these covariates (marital status, health insurance, BMI, health behaviours) did not mediate the association between physical health status and SRH differentially by education.

Our results also show that the influence of educational level on the relation between health status and SRH may differ according to disease. There was no link between poor SRH and educational level among people with cancer and no differences according to the level of education on the relation between SRH and cancer in either sex. Taylor and Lobel29 showed that health expectations are different among cancer patients. They found in threatening situations, such as having cancer, social comparison occurred differently than in non-threatening situations. They found that cancer patients evaluated their situation in comparison with those who were less fortunate. This phenomenon may have limited the effect of cancer on poor SRH, independently of educational level.

The effect of educational level on the relation between health status and SRH was stronger in women than men. Previous studies have reported that women are more aware of their health,27 less likely to overestimate their future health quality of life, expect to live longer30 and more often report poor SRH6 than men. These gender differences in health expectations may also play a role in the impact of disease on perceived health. Educational level has also been shown to be more important for well-being in women than men. Other than education, women have fewer socioeconomic resources such as power, authority or income. Thus, women could depend more heavily on education than men for quality of life, well-being and health.25 Future studies should assess the relation between SRH and health status according to factors other than education, such as income or job activity, particularly for men.

Our results are consistent with results of two recent articles that have shown that lower health ratings were more strongly associated with mortality for adults with higher education.14 15 Indeed, according to our results, the propensity to report poor health status when health problems were reported to exist was higher among more highly educated people. Thus, by using SRH, the distinction between individuals with and without a health problem is more precise among those who are in the more highly educated group. If reported diseases or disabilities are associated with death, poor SRH may constitute a more accurate indicator of disease and mortality risk among the higher educated group than among the lower educated group. Conversely, if the higher educated group is more likely to report less serious ailments, with a weak impact on mortality, this phenomenon could lead to a weaker association between SRH and mortality in the higher educated group, as proposed by Sing-Manoux et al.16

Our results show the relation between health status and SRH is modified by level of education and that the association between health conditions and SRH is greater among more highly educated individuals. This phenomenon could lead to an underestimation of the magnitude of health inequalities existing between socioeconomic groups when using SRH as a general measure of health. As requested by Sen,31 “there is a strong need for scrutinising the statistics on self perception of illness in a social context by taking note of levels of education, availability of health facilities, and public information on illness and remedy”.

What is already known on this subject

  • SRH results from a balance between one’s actual health status and the best health that one could expect for oneself.

  • Expectations that people have about their health can vary according to several factors, like SES, cultural or social issues, which can lead to differences in reporting health status for the same “objective” health status.

What this study adds

  • Our results show the relation between health status and SRH is modified by level of education and that the impact of health conditions on SRH is greater among more highly educated individuals, particularly among women.

  • This phenomenon could lead to an underestimation of the magnitude of health inequalities existing between socioeconomic groups when using SRH as a general measure of health.

Acknowledgments

We acknowledge Centers for Disease Control and Prevention (CDC), the National Center for Health Statistics (NCHS) for giving us the possibility to use National Health and Nutrition Examination Survey Data. Hyattsville, Maryland: US Department of Health and Human Services, Centers for Disease Control and Prevention.

REFERENCES

View Abstract

Footnotes

  • Competing interests: None.

  • Funding: This research was supported by the MiRe (“Mission de la Recherche”) and the Direction of Research, studies, evaluation and statistics, France.

  • Ethics approval: The NHANES 1999–2004 study was approved by the NCHS Research Ethics Review Board (protocol #98-12).

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.