Relationship between serum ferritin, alcohol intake, and social status in 2235 Danish men and women. Received: 24 July 1995 / Accepted: 13 December 1995.
Ann Hematol (1996) 72 : 145–151
Q Springer-Verlag 1996
ORIGINAL ARTICLE
N. Milman 7 M. Kirchhoff
Relationship between serum ferritin, alcohol intake, and social status in 2235 Danish men and women
Received: 24 July 1995 / Accepted: 13 December 1995
Abstract The objective was to examine the relationships between serum ferritin, alcohol intake, and socioeconomic factors (school education, occupational education, occupation, income, marital status, cohabitation status, housing, social class) in a population survey performed in Copenhagen County during 1982–1984. The participants were selected at random from the census register and comprised 2235 healthy Danish individuals, non-blood donors (1044 men, 1191 women) in cohorts being 30, 40, 50, and 60 years old. The participants gave a detailed social and medical history and had a clinical examination including blood samples. In all age-groups, men had significantly higher serum ferritin and alcohol intake than women. In men, there was no relationship between serum ferritin and social class. Significant relationships were observed between ferritin and occupation (unemployed and self-employed men had higher ferritin than those with other occupations) and ferritin and income (in younger men, ferritin displayed a steady increase with income). None of the social variables were related to the prevalence of iron deficiency or iron overload. Alcohol intake was related to occupation and income, but not to social class. In women, none of the social variables showed any significant relationship to ferritin levels or iron overload. The prevalence of small iron stores (serum ferritin ^30 mg/ l) was lower and the intake of alcohol was higher in women from high social classes. In both men and women, serum ferritin displayed highly significant positive correlations with alcohol intake. Likewise, the prevalence of iron overload (serum ferritin `90th percentile) was closely correlated to alcohol intake. In conclusion, socioeconomic factors per se had a minor influence on serum ferritin levels and iron status in Danes. The distinct association between alcohol intake and serum ferritin levels should be considered in future iron status surveys. N. Milman (Y) 7 M. Kirchhoff Department of Medicine, Næstved Hospital, DK-4700 Næstved, Denmark
Key words Alcohol 7 Ferritin 7 Iron metabolism 7 Men 7 Social class 7 Women
Introduction In normal subjects there exists a close relationship between serum ferritin levels and mobilizable body iron stores [1]. Measurement of serum ferritin is at the present the most simple and reliable parameter for evaluation of iron status in healthy individuals and in subjects with uncomplicated iron deficiency or iron overload [2]. Social factors have a considerable impact on the general health condition, on the prevalence of disease, and on life expectancy [3]. They may also influence nutritional habits, such as dietary and supplemental iron intake, consumption of meat and vegetables rich in vitamin C, and alcohol intake. The purpose of the present study was to evaluate associations between social factors, serum ferritin, and alcohol intake in a randomly selected population of 2235 Danish men and women in good health.
Subjects and methods Subjects The investigation was performed during 1982–1984, and the selection of the participants has been described in detail previously [4, 5]. A World Health Organization initiated study for the “Monitoring of Trends and Determinants in Cardiovascular Disease” (DAN-MONICA) was conducted at the Center for Population Studies in Glostrup in Copenhagen County. A randomly selected sample of 1433 Danish men and 1359 non-pregnant Danish women, in age-cohorts of 30, 40, 50, and 60 years, was derived from the census register. From this population, blood donors (389 men, 168 women) were excluded. The final series comprised 1044 men and 1191 women (Table 1). The majority of the participants were fit and healthy, and, according to the records, at most 3% might suffer from diseases (inflammatory disease, liver disease, malignancy) which might cause an inappropriate elevation of serum ferritin.
146 Table 1 Serum ferritin and alcohol intake (median and 5– 95 percentile) in 2235 Danish non-blood donors participating in the DAN-MONICA Survey
Age (years)
Men (n)
Serum ferritin (mg/l)
Alcohol (drinks/week)
Women (n)
Serum ferritin (mg/l)
Alcohol (drinks/week)
30
271 252
50
264
7 1–40 10 0–42 10 0–44 8 0–13 0.007 8 0–42 10 0–40 0.63
310
40
130 50–315 140 49–386 144 46–390 133 35–481 0.49 134 49–334 141 42–465 0.15
38 7–132 42 6–137 56 10–179 86 27–242 0.0001 39 7–135 67 14–212 0.0001
3 0–16 4 0–20 4 0–16 2 0–15 0.0001 3 0–18 3 0–15 0.01
60 p value 30c40
257 a
50c60 p value b a b
523 521
299 298 284 609 582
Kruskal-Wallis test Mann-Whitney test
Methods The participants gave a detailed social and medical history. The following variables concerning social and economic status were registered in a questionnaire. School education: I. 9 years of schooling; II. 10 years of schooling; III. 11–13 years of schooling. Occupational education: I. no education, unskilled, and semiskilled ~1year of education; II. 1–2 years of education; III. skilled education; IV. further education `2 years. Occupation: I. unemployed; II. unskilled, semiskilled occupation; III. skilled occupation, blue-collar worker; IV. whitecollar worker; V. self-employed. Annual income in 1981 figures: I. Low ~100!10 3 DKK; II. medium 100–149!10 3 DKK; III. intermediate 150–199!10 3 DKK; IV. high `199!10 3 DKK. Marital status: I. never married; II. married; III. divorced; IV. widowed. Cohabitation status: I. single; II. cohabitant. Housing: I. rental dwelling; II. owned dwelling. Social class of the individual: I. Class I–II; II. Class III; III. Class IV; IV. Class V. Social class of the family: Cohabitant women were classified according to the person with the highest social class: I. Class I–III; II. Class IV–V. Alcohol intake was registered in the questionnaire as number of drinks/week. In Denmark, a drink contains approximately 12 g ethanol. Serum ferritin was measured with an immunoradiometric assay (Phadebas Ferritin PRIST, Pharmacia Diagnostics AB, S75182 Uppsala). Serum ferritin values of ~15 mg/l indicated depleted body iron stores (iron deficiency), values of 15–30 mg/l indicated small iron stores. Thus, values of ^30 mg/l indicated small or depleted iron stores. Values of `300 mg/l were consistent with ample or increased iron stores (iron overload) [2].
Statistics Due to the skewed distribution of both serum ferritin and alcohol intake, nonparametric statistics were used in the analysis of the data. The Mann-Whitney and Kruskal-Wallis rank sum tests and the X 2-test were used to evaluate the significance of differences. Correlations were assessed by Spearman’s rank correlation coefficient (rs). The level of significance was chosen at the 5% level.
Results Serum ferritin In men, serum ferritin values were not significantly different among the four age groups (Table 1). At all ages,
men had significantly higher serum ferritin than women (Mann-Whitney test: p~0.0001). Premenopausal 30and 40-year-old women had similar serum ferritin values. In 50-year-old women, of whom approximately 50% were menopausal or postmenopausal, and in 60year-old women, who were all postmenopausal, serum ferritin increased significantly (Table 1). The male series was divided into two groups of 30- c 40-year-old and 50- c 60-year-old individuals. The female series was divided into three groups of 30- c 40-year-old, 50year-old, and 60-year-old individuals. Alcohol intake Alcohol intake in the cohorts is shown in Table 1. At all ages, men had significantly higher intake than women (Mann-Whitney test: p~0.0001). Men had a median intake of 9 drinks/week, and women a median intake of 3 drinks/week. The arithmetic mean intake was 13.4B0.45 (SEM) drinks/week in men and 5.0B0.19 drinks/week in women. In both sexes, 40- and 50-yearold subjects had significantly higher intake than 30- and 60-year-old individuals (p~0.001). In the entire series, the first, second, and third quartile for alcohol intake in men was 4, 9, and 18 drinks/week, and in women 1, 3, and 7 drinks/week. Alcohol intake and serum ferritin A significant positive relationship was found between alcohol intake in quartiles and serum ferritin in both men and women (Table 2). Also, significant positive correlations were observed between alcohol intake and serum ferritin in both sexes (Table 3). The relationship between alcohol intake and serum ferritin is further elucidated in Tables 4 and 5. An increase in alcohol intake was associated with a significant increase in serum ferritin. Men with serum ferritin `300 mg/l (Table 4)
147 Table 2 Relationship between serum ferritin (median and 5–95 percentile) and alcohol intake in 2235 Danes Age (years)
n
p value a
Alcohol intake, quartiles (drinks/week)
Table 5 Relationship between serum ferritin and alcohol intake (median and 5–95 percentile) in 1091 Danish women. The serum ferritin classes correspond approximately to the 25th, 26th–75th, 76th–89th and 90th percentile for all women Women-age (years)
I
IIcIII
IV
Men
Women
30c40
523
50c60
521
30c40
609
50
298
60
284
30c40 609
n
Serum ferritin (mg/l) 118 39–246 109 33–364
126 49–321 133 41–434
164 52–398 183 45–718
0.00001
28 6– 99 52 5–174 72 23–229
37 7–123 53 10–138 90 28–246
56 12–176 67 10–199 96 22–382
0.00001
0.00001
Serum ferritin (mg/l) ~30
30–90 n 91–140
0.32
n a
Kruskal-Wallis test
`140 n
Table 3 Correlation between alcohol intake (drinks/week) and serum ferritin (mg/l) in 2235 Danes
p value a a
Age (years) Men
Women
30c40 523
50c60 521
30c40 609
Spearman rs
0.26
0.24
0.22
0.16
0.10
p value
0.0001
0.0001
0.0001
0.004
0.08
n
50 298
60 284
60 284
Alcohol (drinks/week) 3 3 0–15 0–14 235 75
n 0.006
50 298
1 0–13 27
4 0–18 286
3 0–15 156
2 0–11 120
4 0–38 62
4 0–16 44
3 0–11 74
7 0–30 26
7 0–37 23
2 0–18 63
0.06
0.14
0.0001
Kruskal-Wallis test
Table 6 Relationship between serum ferritin and social variables in 2235 Danes. Figures indicate p values for differences in serum ferritin levels between groups (see text) Age (years)
Table 4 Relationship between serum ferritin and alcohol intake (median and 5–95 percentile) in 1044 Danish men. The serum ferritin classes correspond approximately to the 25th, 26th–75th, 76th–89th and 90th percentile for all men Men – age (years) 30c40 523
n Serum ferritin (mg/l) ~100 n 100–199 n 200–299 n `299 n p value a a
Kruskal-Wallis test
50c60 521
Alcohol (drinks/week) 6 7 0–31 0–40 159 166 8 1–36 238
10 0–31 195
12 0.5–48 88
11 0–40 88
15 3–80 38
16 0.7–53 72
0.00001
Men
Women
n
30c40 50c60 523 521
30c40 50 609 298
Social variable
Serum ferritin p value a
Serum ferritin p value a
School education Occupational education Occupation Income Cohabitation status Marital status Housing Social class, individual Social class, family a
60 284
0.84
0.07
0.22
0.54
0.17
0.42 0.007 0.04
0.74 0.35 0.76
0.98 0.09 0.22
0.87 0.56 0.92
0.003 0.48 0.34
0.61 0.81 0.45
0.55 0.84 0.05
0.75 0.23 0.54
0.31 0.06 0.43
0.68 0.08 0.94
0.07
0.58
0.81
0.23
0.12
0.77
0.14
0.94
Kruskal-Wallis or Mann-Whitney test
and women with relatively high serum ferritin levels `140 mg/l (Table 5) also had the highest alcohol intake.
0.00001
Socioeconomic variables The general results are depicted in Table 6. In women, none of the social variables displayed any significant re-
148 Table 7 Median serum ferritin and median alcohol intake in 1044 Danish men according to occupation Men – age (years) 30c40
50c60
30c40
50c60
Occupation
Serum ferritin (mg/l)
Alcohol (drinks/week)
Unemployed np156
166
140
10
7
Unskilled, semiskilled 126 np122
162
10
10
Skilled np143
119
133
10
9
White collar np498
129
133
7
9
Self-employed np125
164
165
13
16
p value a a
0.007
0.35
0.003
0.002
Kruskal-Wallis test
lationship to serum ferritin levels. The relationship between occupational education and serum ferritin in 60year-old women was probably coincidental. In men, significant relations were observed between serum ferritin and occupation, income, and housing. Occupation. It appears from Table 7 that 30- c 40year-old unemployed and self-employed men had similar serum ferritin levels, being higher than in men belonging to the other occupational groups. Income. In 30- c 40-year-old men, serum ferritin and alcohol intake displayed a steady increase with income (Fig. 1). Housing. In 30- and 40-year-old men there was no relationship between serum ferritin levels and housing (TaFig. 1 Relationship between income, median alcohol intake, and median serum ferritin in 1044 Danish men
ble 6). Men 50- c 60-year-old living in rental dwellings had significantly higher serum ferritin (median 159 mg/ l) compared with those living in their own dwellings (median 131 mg/l) (p~0.05). Men living in rental dwellings had a median alcohol intake of 10 drinks/week vs. 9 drinks/week in those living in their own dwellings (p~0.20). Social class. Both in men and in women, there was no significant difference between serum ferritin levels in the various social classes (Table 8; Figs. 2, 3). The low median serum ferritin level in 30- c 40-year-old men in social class IV was probably coincidental. In men, there was no significant difference between alcohol intake in the various social classes (Fig. 2). Women, 30- c 40 years old from social class III had significantly higher alcohol intake than those from the other social classes (p~ 0.01) (Fig. 3). Among 60-year-old women, those from social class I–II had significantly higher alcohol intake than those from the other social classes (p~0.05) (Fig. 3). Women from family social class I–III had a significantly higher alcohol intake than those from family social class IV–V (30- c 40-yearold, p~0.07; 50-year-old, p~0.01; 60-year-old, p~0.01). Small iron stores and iron overload. Among men, 30- c 40-year-old women, and 60-year-old women, there was no relationship between the prevalence of small iron stores or iron overload and social class (Table 9) or any of the other socioeconomic variables. Among 50-yearold women, those from social class I–III had a significantly lower prevalence of small iron stores compared with those from social class IV–V (Table 9).
149 Table 8 Relationship between serum ferritin (median and 5– 95 percentile) and social class of the individual in 2235 Danes
Age (years) Men 30c40 523
n
Serum ferritin (mg/l) 140 148 48–334 50–480
III
np320
139 57–379
IV
np287
V
np169
a
Fig. 3 Relationship between social class, median alcohol intake, and median serum ferritin in 1191 Danish women
50c60 521
Social class I–II np268
p value a
Fig. 2 Relationship between social class, median alcohol intake, and median serum ferritin in 1044 Danish men
Women
Kruskal-Wallis test.
30c40 609
50 298
60 284
np128
40 4–103
52 10–184
112 38–200
144 45–457
np276
39 7–155
67 17–146
75 27–237
119 48–332
133 24–547
np414
37 6–136
53 8–193
80 23–280
136 51–332
133 34–398
np373
45 10–131
55 8–192
91 24–271
0.07
0.58
0.81
0.23
0.12
150 Table 9 Prevalence of small iron stores (serum ferritin ^30 mg/l) and iron overload (serum ferritin `300 mg/l) in 2235 Danes according to social class of the individual Age (years)
n Social class I–II
S-ferritin (mg/l) ^ 30 `300
III
Men
Women
30c40 50c60 523 521
30c40 50 609 298
60 284
Percentage of subjects 0.7 5.7
1.6 11.8
35.6 0
18.1 0
0 0
^ 30 `300
0.7 9.7
2.3 14.8
35.1 0.6
11.9 0
7.2 1.8
IV
^ 30 `300
0.6 7.1
6.1 15.3
39.7 0
30.4 0.9
7.5 1.3
V
^ 30 `300
0 6.2
2.3 11.4
36.4 1.4
24.5 2.0
10.7 3.1
p value a
0.86
0.29
0.48
0.10
0.67
I–III
^ 30 `300
1.0 7.7
2.0 13.9
35.4 0.4
14.8 0
6.8 1.4
IV–V
^ 30 `300
0.4 6.8
4.6 13.7
37.5 0.6
29.1 1.4
10.5 2.4
p value a a
0.83
0.24
0.60
0.02
0.57
x 2-test
Discussion The relationship between social status and disease or death is a central theme in social medicine, and previous surveys have shown that a low social status is associated with increased morbidity and mortality [3]. Studies in Denmark have demonstrated an increased prevalence of cardiovascular risk factors in persons from lower social classes [6]. The recorded alcohol intake in the present series was based on a questionnaire. No attempt was made to verify the stated intake by measurement of biochemical markers such as mean red cell volume or glycosylated serum transferrin. Men stated a median alcohol intake in drinks/week which corresponded to approximately 108 g ethanol/week, with a 95th percentile of 492 g/ week; the mean stated alcohol intake in drinks/week corresponded to 161 g ethanol/week. Women stated a median alcohol intake in drinks/week which corresponded to approximately 36 g ethanol/week, with a 95th percentile of 204 g/week; the mean alcohol intake in drinks/week corresponded to 60 g ethanol/week. In 1980–1984, the registered intake of pure ethanol in adult Danes was on the average 12 l/year (9472 g ethanol/ year), corresponding to a mean intake of 182 g/week. Consequently, the stated alcohol intake may well be considerably lower than the real intake. In the present study there was no relationship between alcohol intake and social class in men, thus con-
firming the results by Møller et al. [6]. The relationship between alcohol intake and occupation was in accordance with the results of an earlier Danish survey done in 1981 [7], reporting a high intake among self-employed persons and manual workers, and a smaller intake among salaried employees. In addition, in the present series, 30- c 40-year-old men displayed a positive relationship between income, alcohol consumption, and serum ferritin. In the present study there was a positive relationship between alcohol intake and social class in women, in accordance with the findings of Sælan et al. [7]. In healthy individuals, the serum ferritin concentration is an indicator of mobilizable body iron stores [1, 2], which in the long-term are influenced by dietary and supplemental iron intake and iron absorption on the one hand, and by iron losses on the other hand. However, so far no study has been able to show a direct correlation between dietary iron intake and serum ferritin [4, 5]. In Danish men, serum ferritin levels increase from adolescence to 30 years of age and then remain fairly constant throughout adulthood into old age, with a low prevalence of iron deficiency [4]. However, there is an increase in the prevalence of high ferritin values above 300 mg/l with age [4], and in men the problems associated with iron overload have attracted much attention during recent years. In Sweden, Olsson et al. [8] reported a prevalence of genetic hemochromatosis of 0.4%, which means that 13% of the population are heterozygotic carriers of the hemochromatosis allele. In women, serum ferritin levels remain low from adolescence (menarche) to the menopause [9]. After menstruation has ceased, serum ferritin levels increase gradually during the subsequent 10 postmenopausal years to approach male levels [9]. Blood donation is the single man-made factor having the greatest impact on iron status and serum ferritin [10, 11], and in order to avoid any confounding effect, blood donors were a priori excluded from this study. The most interesting observation was the consistently positive relationship between alcohol intake and serum ferritin values in both men and women. An increase in alcohol intake was associated with an increase in serum ferritin levels. Men and women with serum ferritin values above the 90th percentile had the highest alcohol intake in all age groups. The mechanisms by which alcohol affects serum ferritin are unclarified. Alcohol may influence the serum ferritin in several ways. First, alcohol has some enhancing effect on intestinal iron absorption, especially in wine drinkers, and may in this way contribute to increased body iron stores and serum ferritin [12]. Second, alcohol may exert a toxic effect on hepatocytes and cause an elevation of serum ferritin through a release of intracellular ferritin [13]. However, the vast majority of the participants in the present series were in good health and undoubtedly had no biochemical evidence of liver cell damage. Third, alcohol may have an enhancing effect on the intracellular ferritin synthesis.
151
Leggett et al. [14] demonstrated a similar relationship between alcohol intake and serum ferritin in Australian men and women. An increase in alcohol intake was accompanied by an increase in serum ferritin levels. The daily intake of alcohol, regardless of type (beer, wine, spirits) was still positively related to the concentration of ferritin among men with normal serum gammaglutamyl transferase [14]. These findings show that alcohol intake is a “confounder” in epidemiological surveys of iron status and suggest that the amount of alcohol consumed daily should be registered in future surveys. It appeared from the present series that in menstruating women, serum ferritin levels and the prevalence of iron deficiency and iron overload were independent of social class and other socioeconomic factors. In the NHANES II study in the USA, there was a trend towards a higher prevalence of iron deficiency in women with a low level of education and in women belonging to low social classes and fulfilling the criteria for poverty [15]. A similar trend was observed in 50year-old Danish women. However, the breach between high and low social classes is far more pronounced in the USA than in Denmark, where social class distinctions have been leveled due to the social wellfare policy implemented during the 1960s. Furthermore, the impact of menstruation (including method of contraception) and pregnancy on iron balance is of such a magnitude [9] that it will overshadow any minor influence that socieconomic factors might have. In men, serum ferritin levels displayed no relationship to social class, but significant relations existed with occupation and income. In the NHANES II study, there was no relationship between the level of education or poverty status and the prevalence of iron deficiency [15]. In conclusion, the serum ferritin concentration was markedly influenced by alcohol intake, whereas socioeconomic factors per se had only a minor influence on serum ferritin in Danes.
References 1. Walters GO, Miller FM, Worwood M (1973) Serum ferritin concentration and iron stores in normal subjects. J Clin Pathol 26 : 770–772 2. Milman N, Pedersen NS, Visfeldt J (1983) Serum ferritin in healthy Danes: relation to marrow haemosiderin iron stores. Dan Med Bull 30 : 115–120 3. Syme SL, Berkman LF (1976) Social class, susceptibility, and sickness. Am J Epidemiol 104 : 1–8 4. Milman N, Kirchhoff M (1991) Iron stores in 1433, 30- to 60year-old Danish males. Evaluation by serum ferritin and haemoglobin. Scand J Clin Lab Invest 51 : 635–641 5. Milman N, Kirchhoff M (1992) Iron stores in 1359, 30- to 60year-old Danish women. Evaluation by serum ferritin and hemoglobin. Ann Hematol 64 : 22–27 6. Møller L, Kristensen TS, Hollnagel H (1992) Social status and cardiovascular risk factors in Danish men. Ugeskr Laeger 154 : 8–13 7. Sælan H, Møller L, Køster A (1992) Alcohol intake in a Danish cohort during 11 years. Scand J Soc Med 20 : 87–93 8. Olsson KS, Marsell R, Ritter B, Olander B, Åkerblom E, Ôstergård H, Larsson O (1995) Iron deficiency and iron overload in Swedish male adolescents. J Intern Med 237 : 187– 194 9. Milman N, Kirchhoff M, Jørgensen T (1992) Iron status markers, serum ferritin and hemoglobin in 1359 Danish women in relation to menstruation, hormonal contraception, parity, and postmenopausal hormone treatment. Ann Hematol 65 : 96–102 10. Milman N, Kirchhoff M (1991) Influence of blood donation on iron stores assessed by serum ferritin and haemoglobin in a population survey of 1433 Danish males. Eur J Haematol 47 : 134–139 11. Milman N, Kirchhoff M (1991) The influence of blood donation on iron stores assessed by serum ferritin and hemoglobin in a population survey of 1359 Danish women. Ann Hematol 63 : 27–32 12. Turnbull A (1974) Iron absorption. In: Jacobs A, Worwood M (eds) Iron in biochemistry and medicine. Academic Press, London, pp 369–403 13. Milman N, Graudal N, Hegnhøj J, Christoffersen P, Pedersen NS (1994) Relationships among serum iron status markers, chemical and histochemical liver iron content in 117 patients with alcoholic and non-alcoholic liver disease. Hepato-gastroenterol 41 : 20–24 14. Leggett BA, Brown NN, Bryant SJ, Duplock L, Powell LW, Halliday JW (1990) Factors affecting the concentration of ferritin in serum in a healthy Australian population. Clin Chem 36 : 1350–1355 15. Pilch SM, Senti FR (eds) (1984) Assessment of the iron nutritional status of the US population based on data collected in the second National Health and Nutrition Examination Survey, 1976–1980
