Social disadvantages in childhood and risk of all-cause death and cardiovascular disease in later life: a comparison of historical and retrospective childhood information

Abstract

Background Childhood socioeconomic circumstances have been shown to contribute to adult mortality. The purpose of this study was to compare the association between objective historical records and recalled questionnaire-based information on childhood socioeconomic position (SEP) with regard to cardiovascular and all-cause mortality.

Methods We examined the association between a socially disadvantaged childhood and all-cause mortality, cardiovascular disease (CVD) mortality, coronary heart disease (CHD) mortality, and acute coronary events among male participants in the Kuopio Ischemic Heart Disease (KIHD) Risk Factor Study, a population-based cohort study in eastern Finland with follow-up until 2002. The historical data on childhood factors were collected from school health records (n = 698), mainly from the 1930s to the 1950s. Recall data on socioeconomic conditions in childhood were obtained from the baseline examinations of the KIHD cohort (n = 2682) in 1984–89.

Results According to original school health records the men who were socially disadvantaged in childhood had a 1.41-fold (95% confidence interval 1.01–1.97) age-adjusted and examination-year-adjusted risk of all-cause death, a 1.32-fold (0.83–2.11) risk of CVD death, a 1.48-fold (0.85–2.57) risk of CHD death, and a 1.50-fold (1.02–2.20) risk of acute coronary events. After adjustment for biological and behavioural risk factors and for the SEP in adulthood the association was attenuated in all-cause death but did not change in CVD death, CHD death, and acute coronary events. On the contrary, the questionnaire-based recalled childhood data on childhood SEP showed no associations with mortality or acute coronary events.

Conclusions With regard to adult mortality, the use of historical records concerning hygiene and living conditions collected in childhood may either provide more accurate measures of early-life socioeconomic conditions or capture more relevant aspects of childhood socioeconomic disadvantage than retrospective recall data.

Adverse socioeconomic conditions in childhood have been associated with mortality in later life.^1–4 Most life course studies have used retrospective cohorts or a case–control design, relying on participants' recall of early life socioeconomic position (SEP). A socially disadvantaged childhood has been identified through the subjects' own recollection of the father's occupational social class.^2–9 There has not been much systematic evaluation of the validity of recalled early life circumstances or of the possibility for recall errors to bias associations.¹⁰ Using recalled information may underestimate the true impact of childhood socioeconomic situation.

Few studies have used data on childhood social status from actual historical records.^11–13 The review by Galobardes et al.¹ showed that studies using objective data on childhood SEP tended to show stronger associations with mortality than studies using recalled information, and studies using more expansive measures of childhood SEP showed an effect that compared with studies using simple measures of recalled father's occupation.

The purpose of this study was to compare objective historical records and recalled questionnaire-based information on childhood SEP as a predictor of all-cause mortality, cardiovascular disease (CVD) death, coronary heart disease (CHD) death, and incidence of acute coronary events in later life among a subset of participants of the Kuopio Ischemic Heart Disease (KIHD) Risk Factor Study for whom objective early life data were available. We have previously published negative findings on associations between childhood SEP and adult mortality using recalled data on childhood SEP that included information on parental occupation and education.⁸ Historical data gathered at the time our participants were boys were collected by school health nurses who were familiar with the family background and the home conditions of the boys. The nurses regularly followed-up on health and behaviour of the students, made home visits, and were able to make fairly objective comparisons of the socioeconomic differences within the perspective of that time. They collected information on housing conditions, social situation, nutrition, and the hygiene of the child during home visits. Additionally in this study, we examined the possible effect of biochemical, behavioural, and socioeconomic factors on the relation between socially disadvantaged childhood and cardiovascular morbidity and mortality.

Materials and methods

Study population

The subjects were participants in the KIHD Risk Factor Study, which is a prospective population-based study designed to investigate risk factors for CVDs, including psychosocial and socioeconomic factors, in middle-aged and ageing men from Eastern Finland. The original study population consisted of a random age-stratified sample of 2682 men who were 42, 48, 54, or 60 years of age at baseline. The Research Ethics Committee of the University of Kuopio approved the study.¹⁴ School health records were available for 698 (22.3%) men. There were missing data because some of the archives, where school health records were stored, were destroyed during fires and the Second World War. The questionnaire-based childhood SEP variable was counted to 2682 (100.0%) men. A comparison of the historical study sample with the rest of the KIHD cohort revealed that participants and non-participants did not differ in systolic blood pressure (SBP), diastolic blood pressure, low-density lipoprotein (LDL) cholesterol, body mass index (BMI), alcohol consumption, former smoking, leisure time physical activity, and income. However, the study participants were on average somewhat younger than the rest of the KIHD cohort, they had lower incidence of acute coronary events after baseline, lower high-density lipoprotein (HDL), and were less likely to smoke currently (data not shown).

Socially disadvantaged childhood

Socially disadvantaged childhood was measured in two ways: (i) Historical data from school health records and (ii) Questionnaire-based recall information.

(i) Historical childhood information was obtained from elementary school health records filled by the school health nurses in the 1930s to 1950s and maintained by the schools or the municipality/city archives. The school health records contained data on family SEP, general hygiene of the student, and sanitary/socioeconomic circumstances at home based on the personal observations of school health nurses in school and during home visits. The school health records included certain places to fill in the nurse's comments, like circumstances at home (information about how many rooms there were at home and how many persons, and who were the persons living at home), general cleanliness (the assessment was either good, satisfactory, or poor), notes, and other comments (poor social conditions/circumstances at home, poverty, misery, neglect, etc.). Social welfare support was provided for children of poor homes, in the form of summer camps and school meals. A man was defined as socially disadvantaged in childhood if school health nurses had reported one or more of the following:

1. poor social conditions at home (deprivation, neglect etc.),

2. poor hygiene,

3. attending a special summer camp for children of poor background,

4. participation in school meal programme intended for children in real need.

If there was no mention of the items 1–4, the man was defined as a man socially not disadvantaged in childhood. Men socially disadvantaged in childhood formed the index group (32%) and men socially not disadvantaged were a reference group (68%) in the analysis. Distribution of the items in the historical childhood SEP index and proportions of socially disadvantaged and not disadvantaged men in childhood are shown in Tables 1 and 2.

(ii) An index used for measuring recalled childhood information was used in the earlier study.⁸ It was created to represent retrospectively recalled childhood SEP based on seven questions: father's and mother's occupations (unskilled manual/skilled manual/non-manual), father's and mother's educations (part of public school/primary or primary plus vocational/middle school or higher), whether or not a family lived on a farm and the size of the farm, and the degree to which their family was perceived as wealthy. These items were scored dichotomously and the scores summed. The subjects were assigned to low, medium, or high childhood SEP by the index tertiles.⁸

Baseline covariates

Age and examination year

Age was categorized into four groups: 1 = 42 years, 2 = 48 years, 3 = 54 years, and 4 = 60 years. Examination years were from 1984 to 1989.

Biological factors

The gathering of blood specimens¹⁵ and the measurement of serum lipids¹⁶ have been explained elsewhere. Two trained nurses measured resting blood pressure between 8 and 10 a.m. with a random-zero mercury sphygmomanometer (Hawksley, Lancing, England). The measurements were made after 5 min of supine rest at 5 min intervals with three measurements in supine, one in standing, and two in sitting position. The mean of the six measurements was used as SBP. The ratio of LDL to HDL cholesterol, and SBP were included in the analysis.

Behavioural factors

The assessment of alcohol consumption in grams per week¹⁷ and a leisure-time physical activity in hours per year from a 12 month history¹⁸ have been described previously. Cigarette smoking was estimated by self-reporting of never, former, or current smoking, and by pack-years. BMI was calculated as ratio of weight to the square of height in metres (kg/m²).¹⁹

Adulthood socioeconomic and psychosocial variables

Current socioeconomic status was assessed by the self-report of annual personal income and educational level. Education was categorized into four groups: less than elementary, elementary, full or some secondary, and high school or above.

Prevalent chronic disease

Participants were considered as having a prevalent ischaemic heart disease at baseline if they had a history of angina pectoris or a prior myocardial infarction, the use of nitro-glycerine or other antiangina medication for chest pain once a week or more frequently/currently, or angina pectoris on effort according to the Rose Questionnaire.²⁰ Self-reported history of stroke was also recorded.

Outcomes

Mortality

Deaths were ascertained by computer linkage to the National Death Registry utilizing the Finnish social security number that is used by all registries. All deaths occurring between study entry (March 1984–December 1989) and December 31, 2002 were included. Deaths coded with the Ninth International Classification of Diseases (ICD-9) codes 390-459 and the 10th revision (ICD-10) by codes I00-I99 were considered CVD deaths. Deaths coded with ICD-9 codes 410-414 and ICD-10 codes I20-I25 were included in the analysis of CHD deaths. The average follow-up time was 12.7 years (range 0.8–16.8 years). There were 59 CVD deaths and 44 CHD deaths during the follow-up period. Death codes were all validated according to the international criteria adopted by the WHO MONICA (MONItoring of Trends and Determinants of Cardiovascular Disease) Project.²¹

Acute coronary events

Data on fatal or non-fatal acute coronary events between the study entry and 1992 were collected prospectively and diagnostic classification was made by the FINMONICA coronary registry group.²² Since January 1, 1993, the events were obtained by computer linkage to the national computerized hospital discharge registry. Diagnostic information was collected from hospitals and events were classified by one internist using the same diagnostic criteria as in the FINMONICA project. The average follow-up time to the first coronary event was 11.9 years (range 0.1–16.8 years). If the subject had multiple non-fatal events during the follow-up, the first one was considered as the endpoint. Data were available through December 31, 2002 during which time 111 acute coronary events occurred in the cohort.

Statistical analysis

The association between socially disadvantaged childhood and the risk of all-cause death, CVD death, CHD death, and the risk of acute coronary events in later life were analysed with Cox proportional hazards models. Two separate analyses were performed: (i) with historical and (ii) with retrospective childhood SEP data. There were 698 cases in the historical childhood SEP analysis. Men socially disadvantaged in childhood formed the index group (32%) and men socially not disadvantaged were a reference group (68%) in the analysis. In the retrospective childhood SEP analysis there were 2682 cases. The bottom tertile of the recalled childhood SEP were compared with the two highest tertiles of the childhood SEP in the analysis. Because the distributions of the two childhood SEPs were so different, this was the only possible valid comparison (32%/68%). Other versions were also tried, but they did not change the results.

All analyses were performed using SPSS for Windows 11.0. Covariates were entered uncategorized as continuous variables into the Cox models with the exception of age, examination year, and educational level.

Relative hazards (RHs), adjusted for risk factors, were estimated as antilogarithms of coefficients from multivariate models. Their confidence intervals (CIs) were estimated under the assumption of asymptotic normality of the estimates.

Subsequent age-adjusted and examination-year-adjusted models separately examined the influence of biological risk factors (SBP, LDL/HDL), prevalent chronic diseases, psychosocial/personality characteristics (alcohol, smoking, BMI, physical activity), and SEP (income, education) on the relation between childhood disadvantage and cardiovascular events to analyse the impact of these potential mediating mechanisms. A full model that simultaneously adjusted for all of the risk factors was also analysed.

Results

Historical data

The mean ± standard deviation or prevalence for the baseline covariates: age, examination year, the biological and behavioural cardiovascular risk factors (resting SBP, HDL and LDL cholesterol, leisure time physical activity, BMI, alcohol consumption, and smoking), for men socially disadvantaged and socially not disadvantaged in childhood are shown in Table 3.

Mortality rates of all-cause death, CVD and CHD death, and incidence density of acute coronary events in men socially disadvantaged (n = 221) and socially not disadvantaged (n = 477) in childhood are also shown in Table 3.

According to original school health records the men who were socially disadvantaged in childhood had a 1.41-fold (95% CI 1.01–1.97) age-adjusted and examination-year-adjusted risk of all-cause death, a 1.32-fold (0.83–2.11) risk of CVD death, a 1.48-fold (0.85–2.57) risk of CHD death and a 1.50-fold (1.02–2.20) risk of acute coronary events. After adjustment for biological and behavioural risk factors, IHD at baseline, educational level, and income in adulthood, the association attenuated in all-cause death but remained unchanged in CVD death, CHD death, and in acute coronary events (Table 4).

Men with no prevalent ischaemic heart disease at baseline (n = 556) were also examined separately. The men who were socially disadvantaged in childhood had a 1.15-fold (0.75–1.76) age-adjusted and examination-year-adjusted risk of all-cause death, 1.01-fold (0.53–1.93) risk of CVD death, 1.09-fold (0.51–2.35) risk of CHD death, and 1.73-fold (1.05–2.85) risk of acute coronary events. The number of subjects in each group was reduced, and consequently the CIs became wider and the associations were not statistically significant any longer, with the exception of acute coronary events (Table 5).

Retrospective data

As in earlier analyses using a shorter follow-up⁸ men who were socially disadvantaged in childhood had no increased CVD risk. Socially disadvantaged men had a 1.05-fold (95% CI 0.89–1.23) age-adjusted and examination-year-adjusted risk of all-cause death, 0.95-fold (0.75–1.21) risk of CVD death, 1.01-fold (0.77–1.33) risk of CHD death, and 1.08-fold (0.90–1.30) risk of acute coronary events (Table 6).

In men with no prevalent ischaemic heart disease at baseline (n = 2005), the men who were socially disadvantaged in childhood had a 1.14-fold (0.93–1.41) age-adjusted and examination-year-adjusted risk of all-cause death, 1.02-fold (0.74–1.42) risk of CVD death, 0.93-fold (0.62–1.41) risk of CHD death, and 1.14-fold (0.88–1.47) risk of acute coronary events compared with the men not socially disadvantaged (data not shown).

Discussion

Previous research has suggested that adverse childhood experiences are associated with CVD morbidity and mortality^2,4,7 in later life. The review by Galobardes et al.¹ argued that studies using objective data on childhood SEP tend to show stronger associations with later health problems than studies using recalled information from childhood. Our findings support this contention to some extent. Historical childhood data showed that adverse social circumstances in childhood are independently associated with increased risk of all-cause mortality, CVD and CHD death, and acute coronary events even after adjustment for biological and behavioural risk factors, and for the SEP in adulthood, while use of recalled information on childhood factors showed no associations with CVD risk.

We showed that the association between childhood socioeconomic disadvantage and adult CVD might vary depending on the source of information used. The use of authentic childhood health data gives weight to these new findings. For the first time we were able to derive the measure of childhood social disadvantage directly from the original health records dating back to 1930s and 1940s, instead of relying only on the questionnaire-based recall data. The recall errors may bias the associations and underestimate the true impact of childhood socioeconomic situation may capture somewhat different aspects of childhood socioeconomic disadvantage.

By the standards of modern epidemiological research, the school health records were a very old source of information. In any case, they represent the original observations that were collected by health professionals at that time. The records were stored by either individual schools or by municipalities. Many of the old schools have been closed since that time, and at least one municipal archive was known to be destroyed in a fire during the past decades. About 9% of the original sample were Karelian refugees, who had to leave behind their schools, and in most cases, their health records, in the course of the Second World War. It is important that the final sample for the historical childhood SEP analysis (n = 698) was sufficient for our analyses, and there is no indication that the sample would be in any way gravely misrepresentative of the total KIHD study population (n = 2682). Another strength of the study was that confounding factors and potential mediators were comprehensively measured.

It is also possible that the effects of adverse social circumstances in childhood may have been manifested before this study was carried out. In other words, disadvantaged men may have died, disproportionally earlier, even before the follow-up time started.

The results of this study support the concept that using recalled information of childhood circumstances may underestimate the true impact of childhood socioeconomic situation. If there is no possibility of using historical data, it should be kept in mind that the retrospective approach may be somewhat misleading. Further studies are needed to clear this discrepancy between the analysis of historical and retrospective information.

The Kuopio Ischaemic Heart Disease Risk Factor Study was supported by grants by 41471, 1041086, and 2041022 from the Academy of Finland; 167/722/96, 157/722/98 from the Ministry of Education of Finland; HL44199 from the National Heart, Lung and Blood Institute of the United States and the City of Kuopio. Part of the cost of data collection of the historical records was funded by a grant from the Institute for Social Research at the University of Michigan, USA. L.K. was supported by grants from the Juho Vainio Foundation. The authors would like to thank Professor Jukka T Salonen for the use of the original survey and Professor Pertti Happonen, Pirjo Halonen, MSc, and Kimmo Ronkainen, MSc, for statistical assistance.

Guarantor: Professor Jussi Kauhanen

References