Article Text

Download PDFPDF

Intergenerational social mobility and the risk of hypertension
  1. Lovisa Högberg1,
  2. Sven Cnattingius1,2,
  3. Cecilia Lundholm1,
  4. Pär Sparén1,
  5. Anastasia N Iliadou1
  1. 1Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
  2. 2Department of Medicine, Clinical Epidemiology Unit, Karolinska Institutet, Stockholm, Sweden
  1. Correspondence to Lovisa Högberg, Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, PO Box 281, SE-171 77 Stockholm, Sweden; lovisa.hogberg{at}


Background Low socioeconomic status (SES) has been linked to increased risk of hypertension, a known risk factor for cardiovascular disease. How the risk is altered by intergenerational social mobility is not well known. The aim of this study is to investigate parental SES, adult SES and the intergenerational social mobility in relation to hypertension risk.

Methods By using data from the Swedish Twin Registry, the authors obtained information about both parental and adult SES and hypertension in 12 030 individuals born from 1926 to 1958. Generalised estimating equations were used to estimate ORs with 95% CIs.

Results Low parental SES was associated with increased odds of hypertension (OR 1.42, 95% CI 1.14 to 1.76). Low SES in adulthood was associated with increased odds for women but not for men (OR 1.40, 95% CI 1.15 to 1.70 and OR 1.01, 95% CI 0.83 to 1.24, respectively). Compared with the stable low social status group, the upward mobile group had lower odds of hypertension (OR 0.82, 95% CI 0.70 to 0.97). Compared with the stable high social status group, the results for the downward mobile group indicated an increased risk. A co-twin case-control analysis indicated that the results were independent of familial factors.

Conclusions These findings suggest that the risk of hypertension associated with low parental social status can be modified by social status later in life. Possibly, this could be targeted by public health or political interventions. As parental social status has an impact on later health, such interventions should be introduced early.

  • Social mobility
  • social class
  • occupations
  • hypertension
  • hypertension DI
  • social epidemiology
  • social inequalities

Statistics from


Hypertension or high blood pressure is a major risk factor for cardiovascular disease.1 An increasing trend is seen in both industrialised and developing countries,2 and by 2025, the number of adults with hypertension has been predicted to increase by 60%.3 Low socioeconomic status (SES) has been associated with an increased risk of hypertension,4 as well as other poor health outcomes.4–6 Such differences represent social inequity in health and have increased in Sweden since the 1990s.7

SES often ‘runs in the family’, that is, people are more likely to have an occupation within the same SES group as their parents.8 9 The opposite is social mobility, which is defined as the movement or opportunities for movement between social classes or occupational groups.10 How social mobility is related to hypertension risk is not thoroughly studied. It has been hypothesised that socially mobile individuals would have an increased risk of hypertension as the change leads to psychosocial stress and disrupts the personal support system of the individual.11 It has also been suggested that social mobility minimises health differences between SES groups, as individuals possess health characteristics from both social groups they have belonged to.12 However, more commonly, upward social mobility have been hypothesised to improve health and downward to deprive people of good health.13

In this study, we aim to investigate the effects of parental SES, SES in adulthood and intergenerational social mobility on risk of hypertension in a population-based sample of Swedish twins. We also aim to investigate the importance of common genes and shared environment to the associations studied by performing a nested co-twin case-control analysis.


Study design and study population

Eligible participants were like-sexed twins, born in Sweden from 1926 to 1958 and included in the Swedish Twin Registry (n=37 392) (figure 1). Of these subjects, 30 305 (81%) responded to a postal questionnaire about health and lifestyle in 1973. From 1998 to 2002, all twins born in 1958 or earlier were, irrespective of their participation in the questionnaire in 1973, asked to participate in a telephone interview called the Screening Across the Lifespan Twin (SALT) Study. The response rate was 74%. In the present study, 21 493 individuals (88% of the study population) responded to both the questionnaire and the telephone interview. Information on hypertension was obtained in 21 214 twins. Among these, 20 560 (97%) had information on SES in adulthood and hypertension; 12 324 (58%) had information on both parental SES and hypertension and 12 030 (57%) had information on both parental and adult SES as well as hypertension.

Figure 1

Flow chart of the study population. SES, socioeconomic status.


Information about hypertension was provided in the SALT interview (1998–2002), where questions about medical history and current use of prescribed medications were posed. A subject was classified as having hypertension if answering ‘yes’ to the questions: ‘Do you have or have you had high blood pressure?’ and ‘Do you take any medication daily?’ and then naming an antihypertensive drug. Antihypertensive drugs were identified as having an Anatomic Therapeutic Chemical Classification System code of C02, C03, C07, C08 or C09 and being listed in the Swedish Drug Compendium during 1997–2002.14 A subject was defined as not having hypertension if answering ‘no’ to both questions or naming a drug not listed as an antihypertensive medication. Subjects not answering the two questions, answering yes to the question about having hypertension but not naming an antihypertensive drug or answering no to the question but naming an antihypertensive drug were defined as unclassifiable (n=2934; 12.1% of the defined study population).


Information about birth and maternal characteristics, including both parents' occupations, is routinely documented at birth in delivery records by the attending midwife. The recording and preservation of these records in local delivery archives are required by law. Information about birth year and birth characteristics, including sex, birth weight, birth length and gestational age (calculated by using information on the mother's last menstrual period), was collected from the delivery records, as well as information about maternal characteristics, including maternal age at delivery, number of previous childbirths (parity) and occupation (father's and/or mother's occupation). Parental SES at the time of birth was defined as the highest SES of either parent based on their occupation. The classification from Statistics Sweden15 was used, but the groups were merged into three categories: low SES (unskilled and skilled blue-collar workers and low-level white-collar workers), high SES (intermediate- and high-level white-collar workers) and self-employed (entrepreneurs and farmers). The classification of occupations into SES groups relies to a large extent on education and type of union organisation. Blue-collar workers and low-level white-collar workers often have the same amount of education.15

Information about adult weight, height, smoking habits and alcohol consumption was collected from the 1973 postal questionnaire. Body mass index (BMI) was calculated as the ratio of weight and height squared (in kilograms per square metre). Smoking status was categorised as those who had ever smoked (current and former smokers) and those who had never smoked. Alcohol consumption was classified according to the recommendations by the WHO as category 1 (women: 0–19 g alcohol/day; men: 0–39 g alcohol/day), category 2 (women: 20–39 g alcohol/day; men: 40–59 g alcohol/day) and category 3 (women: >40 g alcohol/day; men: >60 g alcohol/day).16

Information about the individuals' occupation was obtained in the SALT interview. SES in adulthood was classified and merged in the same categories as parental SES.15 Those currently employed were asked about their occupation in the last 12 months, those unemployed were asked about their last occupation and those retired were asked about their primary occupation in adulthood.

Social mobility was defined as change between parental SES and SES in adulthood. This is the intergenerational mobility between the parental and offspring generations. Because of the diversity in background within the self-employed workers, we were unable to determine whether their mobility was upward or downward, and this group was therefore excluded from analyses of social mobility.

Statistical methods

To evaluate associations between parental SES, SES in adulthood, social mobility and risk of hypertension, generalised estimating equations were performed, taking the dependency within twin pairs into account (SAS V.9.2). The results are presented as adjusted ORs with 95% CIs. No crude ORs are presented because of the known strong association between age and hypertension. To investigate possible mechanisms by which parental SES, SES in adulthood and social mobility may influence odds, we present different adjusted models.

In order to investigate whether any shared familial factors confound the associations, we also conducted a co-twin case-control analyses using conditional logistic regression. In this analysis, we investigate the association in twins discordant for hypertension. This design resembles a matched case-control analysis (hypertensive twin as case and non-hypertensive co-twin as control), and the matching refers to the common genes and environment that the twins share. For example, if the twin with high SES in adulthood has lower odds of hypertension than his/her co-twin, this would indicate that the association between low social status and hypertension is not attributed to familial factors.

The study was approved by the regional ethical review board at Karolinska Institutet, Stockholm, Sweden.


Table 1 describes rates of hypertension by birth year, sex, zygosity, parental SES, adult SES and social mobility. The strong association between birth year and rates of hypertension was expected since birth year mirror age. Hypertension was more common in the low SES group, both for parental and adult SES. Rates of hypertension were also increased among twins with lower birth weight, higher BMI, shorter stature and increasing alcohol consumption in adulthood (data not shown). Rates of hypertension were similar among dizygotic and monozygotic twins.

Table 1

Rates of hypertension by birth year, sex, zygosity, parental and adult SES and social mobility among 21 214 Swedish like-sexed twins born from 1926 to 1958

Among those who had missing values for parental SES (N=8890), the rate of hypertension was higher compared with the overall cohort (19.5% vs 15.8%). However, the higher rate was entirely explained by differences in age; those with missing values for parental SES were older compared with the overall cohort (data not shown). The 654 individuals with missing values for SES in adulthood also had a higher rate of hypertension (24.5%) compared with the overall cohort (15.8%). Twins with missing information on SES in adulthood were generally older, had a higher BMI and less years of schooling compared with twins with information on SES in adulthood. However, these variables only partly explained the excess risk of hypertension among individuals with missing values for SES in adulthood (data not shown).

Table 2 shows adjusted ORs of hypertension in relation to parental SES and SES in adulthood. Those who had a low parental SES had 55% increased odds of hypertension compared with those with high parental SES, after adjusting for birth year and sex. Among those with parental self-employed SES, there was a 23% increase in odds. These estimates were essentially unchanged after adjusting for both birth and adult characteristics, but further adjustment for SES in adulthood slightly attenuated the risks (table 2, fourth model). Analyses using only the individuals with complete information on all variables (complete case analyses) showed similar results (data not shown). There was no interaction between parental SES and sex (p=0.95).

Table 2

Hypertension in relation to parental SES at time of birth and SES in adulthood in a cohort of Swedish like-sexed twins born from 1926 to 1958

There was a significant interaction between adult SES and sex (p=0.03). Therefore, we present sex-stratified analyses for SES in adulthood and hypertension. Among men, there was no association between low adult SES and hypertension, but among women, there were a 31% increase in odds in the low SES group, adjusted for birth year (table 2). Complete case analyses showed similar results (data not shown).

Compared with those with stable high SES in both generations, those with stable low SES in both generations had increased odds of hypertension (OR adjusted for age and sex: 1.73, 95% CI 1.37 to 2.18) (not shown in table). Compared with the stable low SES group, the upward mobile group had a decreased risk of hypertension (OR adjusted for age and sex: 0.85, 95% CI 0.74 to 0.97) (table 3). Compared with the stable high SES group, the results for downward movers indicated increased odds of hypertension (OR adjusted for age and sex: 1.24, 95% CI 0.88 to 1.76). These results were essentially unchanged after adjustment for birth and adult characteristics. Complete case analyses showed strengthened results to the same directions as above (data not shown).

Table 3

Intergenerational social mobility among Swedish like-sexed twins born from 1926 to 1958 and risk of hypertension

To investigate the effect of familial confounding, we performed the co-twin control analyses stratified for parental SES (presented in table 4). Only data for those with low parental SES are presented because there was too little variation between twin pairs to yield meaningful estimates in the other groups. Although not statistically significant, the point estimate indicated decreased odds of hypertension for the upward mobile twin compared with their stable low SES co-twin.

Table 4

Co-twin case–control analysis of twins discordant for hypertension in relation to their SES in adulthood


Our results show an association between hypertension and low parental SES. Between hypertension and low adult SES, we found an association for women but not for men. Upward social mobility decreased the odds of hypertension, and in the downward mobile group, there was an indication of increased odds. The results also indicate that the reduced risk associated with moving upwards may be independent of familial factors.

Both low childhood SES and adult SES have previously been associated with hypertension risk, although results have been more consistent for adult SES.17 There are few previous studies on social mobility and hypertension. Two studies found no effect of intergenerational socioeconomic mobility on hypertension risk.11 18 One of these studies was limited due to small sample size and low prevalence of hypertension, limitations well acknowledged by the authors.18 Similar to our results, two prior studies have found that SES both in childhood and in adulthood independently influenced the risk of hypertension.19 20 James et al19 found, with stable high SES as reference, that the upward mobile group had a four times greater risk of hypertension, the downward mobile group had a sixfold increased risk and the stable low SES group had a sevenfold increased risk. Waitzman and Smith20 also reported an increased risk of hypertension in the downward mobile group. Krieger et al21 reported increased blood pressure in working class twins compared with their professional twin sisters. Furthermore, an association between income decline and hypertension has been shown.22 Both parental and adult social class have separately been associated with blood pressure levels as well as risk of cardiovascular disease.23 24

Our results are consistent with findings from some of the previous studies on social mobility and hypertension and also with studies focussing on childhood SES, adult SES and hypertension.19 20 23 25 We found an association between low adult SES and hypertension only for women. Previously, an inverse SES gradient in hypertension risk has also been found more consistently in women than in men.4 This has been explained by a stronger SES gradient in high BMI for women.4 26 We see these results also after adjusting for BMI, but measurements of BMI were collected many years before the outcome measure and a potential gradient in BMI can have evolved during this time.

This study benefits from a large well-characterised study population including both men and women. In our analyses, we have corrected for the dependency between twin pairs by using generalised estimating equations. We also had information about several other covariates. As far as we know, this is the first study to present a co-twin case-control analysis addressing the possibility of familial confounding. Furthermore, it has been shown that twins do not differ in cardiovascular mortality from the general population.27 28

We investigated two possible mechanisms by which parental SES and social mobility may influence risk of hypertension. First, through foetal growth restriction, measured as low birth weight for gestational age, which is associated with both social factors29 30 and with subsequent risk of hypertension.31 However, when we adjusted for birth weight and gestational age, risks remained essentially unchanged, suggesting independent effects of foetal growth restriction and early social factors on subsequent risk of hypertension. Second, although parental SES also influences lifestyle factors in adulthood,32 adjustment for risk factors in adulthood (including BMI, smoking and alcohol consumption) did not essentially change estimates of odds of hypertension related to parental SES or social mobility. In the relation between social status and hypertension risk, all covariates could be regarded as mediators rather than confounding factors, and adjusting for such factors is an attempt to investigate how much of the association between SES and hypertension that can be attributed to the specific covariates. It has to be noted that SES is a theoretical construct, in this study operationalised through occupation; this captures only one part of SES but is usually considered to be an adequate proxy. In table 2, we separate the effects of parental and adult SES, but measures of social status at different time points in life are correlated and intertwined. In real life, people are influenced by both childhood and adult experiences.

One limitation of the study is that information on parental SES was missing for a substantial part of the study population, and the group with missing values had a higher rate of hypertension than the overall cohort. If these subjects are more likely to have low SES, the risk of hypertension related to low SES may have been underestimated. SES in adulthood was self-reported and might be open to recall bias for the part of the study population that was not working at the time of the interview (33% in total: 27% retired and 6% unemployed or not working due to other causes). The measure of parental SES was obtained from original birth records and is therefore not open to recall bias. Subjects were defined as hypertensive if reporting both high blood pressure and a prescribed antihypertensive medication, which is a rather strict definition. Sixteen per cent in our cohort were classified as having hypertension (they were between 40 and 76 years of age), which probably is an underestimation. In a Swedish sample, including subjects between 35 and 64 years, the prevalence of hypertension was 26%.33 Another report estimated that 27% of the whole Swedish adult population (age older than 20 years) had hypertension.34 The most likely primary cause of the low prevalence of hypertension in our cohort is misclassification, where hypertensive subjects have incorrectly been classified as not having hypertension. Self-reported hypertension has been shown to have high specificity (from 80% to 95%) but low sensitivity (43% to 82%), and women are up to twice as likely as men to correctly report hypertension.35 This might explain the slightly higher rate of hypertension among women in our study population. Given the high specificity and probably low sensitivity of the diagnosis, and assuming non-differential misclassification of hypertension by SES, we may have underestimated the effect of SES on hypertension.35 If the diagnosis and treatment of hypertension are socially patterned, those with low SES would probably be less likely to report hypertension than those with high SES. Again, this would, if anything, lead to an underestimation of the association between SES and hypertension.

The measures of adult characteristics, such as BMI, smoking and alcohol consumption, were collected in 1973, well before the outcome measure in 1998–2002. We acknowledge that these characteristics may have changed over time. However, we considered it as an advantage for the study to have measures of risk factors before the occurrence of the outcome especially since advice concerning lifestyle changes has an important role in the treatment of hypertension and since compliance to these advices might differ between socioeconomic groups.

In conclusion, this study shows a positive association between low parental SES and risk of hypertension, indicating that effects of SES on blood pressure start early in life. However, upward social mobility was associated with a decreased risk of hypertension, and the results indicated that those who experienced downward social mobility had an increased risk of hypertension. Furthermore, the results also indicated that the association, at least for the upward mobile group, might not be confounded by familial factors. These findings suggest that the risk of hypertension associated with low parental social status could be modified by social status later in life. This could possibly be targeted by early introduced public health or political interventions.

What is already known on this subject

  • Social inequities in health is a public health concern in most countries, and for hypertension, there is a socioeconomic gradient in risk with the socially disadvantaged at higher risk.

  • Low socioeconomic status both early and later in life have been shown to be positively associated with hypertension risk in most industrialised countries.

  • How intergenerational social mobility alters the risk of hypertension has rarely been studied.

What this study adds

  • Low parental socioeconomic status was associated with an increased risk of hypertension. Low adult socioeconomic status was associated with an increased risk for women but not for men.

  • Stable low socioeconomic status in two generations was associated with an increased risk of hypertension. Upward social mobility was associated with a decreased risk of hypertension, and in the downward socially mobile group, there was an indication of an increased risk.

  • Social inequities in hypertension are likely to start early in life but can be modified by social status later in life.


We are thankful to Dr Brian Lee, Drexel University School of Public Health, for helpful comments and proof reading of the manuscript.



  • Previous presentations An abstract of the paper has been presented as a poster at the 6th World Congress on Developmental Origins of Health and Disease (DOHaD) in Santiago, Chile, in November 2009.

  • Funding This study was supported by grants from the Swedish Cancer Society (grants No. 4594-B01-01XAC and 4594-B04-04XAB), the Swedish Council For Working Life and Social Research (grants No. 2004-1645, 2004-0174 and 2007-0231), the Swedish Research Council (grants No. K2006-71X-14676-04-2 and K2008-54X-20638-01-3), and the European Union-funded Network of Excellence Lifespan (FP6 036894).

  • Competing interests None.

  • Ethics approval The study was approved by the Regional Ethical Review Board in Stockholm, Sweden.

  • Provenance and peer review Not commissioned; externally peer reviewed.

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.