Low-grade inflammation is negatively associated with physical Health-Related Quality of Life in healthy individuals: Results from The Danish Blood Donor Study (DBDS) (2024)

Khoa Manh Dinh, Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Project administration, Software, Visualization, Writing – original draft, Writing – review & editing,^1,* Kathrine Agergård Kaspersen, Investigation, Methodology, Validation, Visualization, Writing – review & editing,¹ Susan Mikkelsen, Investigation, Methodology, Validation, Writing – review & editing,¹ Ole Birger Pedersen, Conceptualization, Funding acquisition, Investigation, Methodology, Resources, Supervision, Writing – review & editing,² Mikkel Steen Petersen, Data curation, Funding acquisition, Investigation, Methodology, Software, Supervision, Writing – review & editing,¹ Lise Wegner Thørner, Funding acquisition, Investigation, Methodology, Resources, Supervision, Validation, Writing – review & editing,³ Henrik Hjalgrim, Conceptualization, Funding acquisition, Methodology, Resources, Supervision, Writing – review & editing,^4,5 Klaus Rostgaard, Data curation, Investigation, Methodology, Supervision, Writing – review & editing,⁴ Henrik Ullum, Conceptualization, Funding acquisition, Investigation, Methodology, Resources, Supervision, Writing – review & editing,³ and Christian Erikstrup, Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Resources, Supervision, Writing – original draft, Writing – review & editing¹

Study design and population

We studied participants from The Danish Blood Donor Study (DBDS), which has been described in detail previously [27–30]. Briefly, DBDS was initiated March 2010 as a multicentre, public-health study and biobank (www.dbds.dk). Currently, more than 110,000 blood donors aged 18–67 years participate in DBDS. At study enrolment, participants completed a four-page questionnaire on health-related items (HRQL), including current smoking status, alcohol consumption, physical activity, diet, anthropometric measurements, and (among women) use of combined oral contraception (OC), childbirth, and menopausal status. Plasma samples were stored, and donors gave permission to link their data to public registers. Overall, fewer than 5% of invited donors have declined to participate [30]. Implicitly, all participants in the present study were healthy, non-pregnant, and non-lactating adults testing negative for HIV, hepatitis B, and C.

According to the definition of LGI, 176 individuals with CRP levels above 10 mg/L were excluded. Likewise, 138 individuals were excluded on the basis of missing values on CRP. For the SF-12 up to two missing items can be accepted and replaced with the population mean [31]. Nevertheless, we decided to exclude participants with more than one missing value in the SF-12 questionnaire, and a mean value (adjusted for age and sex) was used as a substitute for those with one missing item. A total of 919 participants were thus excluded. In the primary analysis, we excluded those who failed to respond to certain questionnaire items, namely: current smoking (1,024 participants were excluded), height (1,017), weight (927), body mass index (BMI) (1,017), physical activity at work (942), leisure-time physical activity (946), meat intake (956), fish intake (1,008) and childbirth (575). Five postmenopausal women were excluded as they stated usage of combined oral contraception. For waist circumference, 1,985 responses were missing, and these participants were excluded from analyses in which waist circumference was used as predictor. In total, 1,413 women and 1,102 men were thus excluded.

Measurements: C-reactive protein

From 1 March to 31 December 2010, 25,877 participants were included in DBDS. High-sensitivity CRP was measured in plasma samples drawn at enrolment from 17,024 participants. CRP was measured by a commercially available, high-sensitivity assay on an automated system (Ortho Vitros 5600, Ortho Clinical Diagnostics, Rochester, NY, USA). Blood was collected in ethylenediamine tetraacetic acid (EDTA)-containing, gel-separated tubes, centrifuged within six hours and stored at -20°C. Samples were stored for less than 12 months before CRP measurement. Measurements on thawed EDTA plasma were validated against fresh serum, which is the test material recommended by the manufacturer. The two measures were highly correlated (R²: 0.98). The CRP levels were somewhat lower in plasma (slope: 0.935). The coefficient of variation for the assay was below 4.8%. The measuring range of the assay was 0.10 to 15.00 mg/L. A default value of 0.05 mg/L was assigned to samples below the lower limit of detection; no samples beyond the upper limit of detection were encountered.

Measurements: Health-Related Quality of Life (HRQL)

The 36-item short-form health survey (SF-36) is the most commonly used descriptive HRQL measure and was developed for population surveys [32,33]. The 12-item short-form (SF-12) is an abbreviated version of SF-36. The questionnaire completed by all participants at enrolment included the SF-12 version one [32,33], see S1 File. SF-12 consists of 12 items from the SF-36, and the result is reported as a physical (PCS) and mental (MCS) component summary score. PCS and MCS are computed using scores of twelve questions and range from 0 to 100, where 0 score indicates the lowest and 100 indicates the highest level of health. As recommended by Quality Metric Inc., the SF-12 was scored giving weights to the individual items and adding the sum to a constant (57.65693 and 60.58847) in PCS and MCS, respectively. In Denmark, the median PCS and MCS are 56 and 54, respectively [32]. The correlation between SF-12 and SF-36 in Denmark on PCS and MCS is 0.95 and 0.96, respectively [34,35]. SF-36 and SF-12 are valid for assessing the general health in populations, and both scores are used as screening tools for specific diseases.

Definition of low-grade inflammation

We used the common practice of defining LGI as CRP level above 3 mg/L and below or equal to 10 mg/L [2,27,36]. Recent study has suggested participant-specific thresholds may be necessary when using one time measurement of CRP to distinguish acute from chronic inflammation in obese women to avoid introducing bias into study findings [37]. Therefore, sensitivity analyses of participants with CRP level above 10 mg/L are performed.

Statistical analysis

The characterization of the study population was stratified by sex, and data were presented as numbers, means with ranges, or frequency statistics. Groups were compared by the two-sample t-test for normally distributed data, the X² test for categorical data, and the Mann-Whitney U-test for non-normally distributed data.

The relationship between primary outcomes (PCS or MCS) and predictors was explored by mediation analysis with sensitivity analyses to examine whether LGI is a mediator in the association between predictors and outcomes [38] adjusted for smoking, age, BMI, waist circumference, physical activity at work and in leisure stratified as detailed below (e.g., the association between BMI and PCS was explored with LGI as possible mediator adjusted for smoking, age, waist circumference, physical activity at work and in leisure). Furthermore, unadjusted and adjusted multivariable linear regression analysis were conducted to compare with mediation analysis model.

Interaction tests between LGI and the predictors were included in all models. Analyses were stratified by the use of OC among premenopausal women, as reported in the questionnaire and confirmed in the Danish National Prescription Register, because OC is a strong predictor of LGI [39]. Residuals were plotted against predicted values to assess model assumptions of normality and independence.

The explored predictors included LGI (yes/no), defined as CRP above 3 mg/L and below or equal to 10 mg/L [2,3]; BMI as a continuous variable; waist circumference as a continuous variable; current smoking (yes/no), cumulative tobacco consumption in pack years (one pack year was defined as 20 g of tobacco every day for one year [40], where a cigarette was estimated to contain 1 g of tobacco; and a cheroot, a cigar, or a pipe 3 g of tobacco [41]); age as a continuous variable; physical activity at work (high vs. low, see definition below); leisure-time physical activity (high vs. low, see definition below)[42,43], childbirth (yes/no); meat intake (yes/no), defined as meat consumption or vegetarian; fish intake (yes/no), defined as consumption of fish more than twice weekly.

Low physical activity at work was defined as mainly sitting or standing and occasionally walking. High physical activity at work was defined as walking and lifting occasionally; or heavy physical work. Low leisure-time physical activity was defined as light physical activity less than 2 hours per week or two to four times per week. High leisure-time physical activity was defined as light physical activity more than 4 hours a week, or fatiguing physical activity for at least 2 hours per week [44].

SF-12 MCS and PCS follow the Poisson distribution in Danish populations [32]. Transformation of data by a power of four and five could approximate the normal distribution for MCS and PCS, respectively [45]. Because the transformed data are more difficult to interpret, we performed and presented all calculations on the original scale. The correlations were later checked by performing the regression analysis after transformation. Cronbach’s α was estimated to evaluate internal consistency of the SF-12.

Multiple imputation was used to remedy missing data in predictor variables that were assumed to be missing at random. Only relevant predictor variables for outcomes were included in the imputation procedure (BMI, current smoking status, waist circumference, physical activity at work and leisure-time) based on complete observed data variables of age and stratified by sex. A total of 100 imputed datasets were created and used for imputation modelling. Sensitivity analyses were performed to assess robustness [46,47].

Results were presented as regression coefficients (RC) with 95% confidence intervals (CI). Statistical analyses were performed using Stata/MP 14.1 for Windows (StataCorp LP, College Station, Texas, USA). A p-value below 0.05 was considered significant.

Ethics statement

Oral and written informed consent was obtained from all participants. The study was approved by The Scientific Ethical Committee of Central Denmark (M-20090237). Additionally, the biobank and research database have been approved by the Danish Data Protection Agency (2007-58-0015).

Characteristics of the study population

The characteristics of the study population are presented in Table 1 and have previously been reported [39]. Briefly, the study comprised 17,024 individuals (47.3% women). The median age was 37.8 years for women and 40.1 years for men. The median PCS score was 55.9 and 56.0, and MCS score was 53.5 and 54.8, for women and men, respectively. Among women and men 17.1% and 15.8% were smokers, respectively. In the cohort, a total of 1,542 (10.3%) had LGI.

Characteristics of the group with missing responses are presented in Table A in S2 File. Compared with the study population, this group had a higher fraction of women, a higher prevalence of obesity and current smokers, lower PCS and MCS scores for both sexes, higher CRP levels, and fewer using OC among women.

Characterisation of participants with CRP levels above 10 mg/L are presented in Table B in S2 File. Analysis including these individuals did not change results, see Table C in S2 File.

Predictors of PCS

PCS and associations with LGI among women using OC, women not using OC, and men are presented in Tables ​Tables22–5. An interaction between LGI and OC (p = 0.009) was observed as described previously [27] and consequently analyses were stratified according to the use of OC. No interaction between LGI and age in any strata was observed. The mediation analyses showed a statistically significant, however only a small, effect of smoking on PCS is mediated through LGI only among women not using OC and men (the average mediated effect was 3.0% among women not using OC and 4.8% among men, see Table 2). In the age-only adjusted linear regression analysis, LGI was associated with lower PCS in all three donor strata (Table 3). Effects persisted, although attenuated, after further adjusting for current smoking, BMI, waist circumference, physical activity at work and leisure-time (women using OC: RC = -0.36 points lower PCS in participants with LGI vs no LGI, CI: -0.94 to -0.19; women not using OC: RC = -0.63, CI: -1.05 to -0.21 and men: RC = -0.76, CI: -1.10 to -0.42), see Table 4. Estimates by multiple imputation were similar to results of the case-complete analyses, see Table 5.

BMI, current smoking, and waist circumference were negatively associated with PCS in all three donor strata. Further, increasing age and high physical activity at work were also negatively associated with PCS, but only among women not using OC, and among men. In contrast, high physical activity in leisure-time increased PCS in all three donor strata (Table ​(Table22 and ​and4).4). Estimates by multiple imputation were also similar to the effects of case-complete analyses, see Table 5.

Predictors of MCS

LGI was not associated with MCS in any strata and the effect of predictors was not mediated by LGI (Table 6). Both increasing age and physical activity in leisure-time significantly increased MCS in all strata, whereas physical activity at work increased MCS among men and women not using OC. Current smoking status was the only strong and statistically significant predictor of a reduction in MCS in all strata (Tables ​(Tables44 and ​and6).6). Estimates by multiple imputation demonstrated the same trend as results from the case-complete analyses (Table 5).

Diet and childbirths had no impact on HRQL

The relation between PCS and MCS with childbirth, meat intake, and fish intake were statistically insignificant in univariate regression analysis, therefore not included in further analysis, see Table D in S2 File.

Transformation of MCS and PCS

The results for the transformed PCS and MCS were nearly identical to the results for untransformed PCS and MCS (Table E in S2 File). Cronbach’s α was 0.77 and 0.75 for PCS and MCS, respectively, which is acceptable.

LGI affects negatively on PCS but had no impact on MCS

To the best of our knowledge, the present study is the first to combine HRQL subjective health assessment with an inflammatory biomarker and objectively measured clinical indicators in a mediation analysis model, demonstrating that among relevant predictors of HRQL only a partial effect of smoking on self-reported physical health is mediated through LGI.

Our finding in this large study strengthens the growing evidence of an association between low self-rated health and higher levels of inflammatory markers, including IL-6 [17,18,21,22], TNF-α [19,20], CRP [21,23–25], erythrocyte sedimentation rate (ESR) [26], and fibrinogen [48]. Our study consists of a large population of healthy blood donors of a wide age range using the comprehensive HRQL as the tool for the assessment of subjective health and analysis stratified for sex, and use of OC. Several studies have shown similar association between CRP and lower score in HRQL, but only among women [23], others only among men [24], or in a design not stratified by sex [21]. Conversely, one study reported no relation between low HRQL and CRP, but only with higher levels of IL-6 [18]. Compared with the current study, previous studies differ in terms of the inflammation marker investigated, the assessment method of self-rated health, size of study population, and the adjustments included in the analysis of the association between HRQL and LGI. The hitherto largest study, which includes 43,110 men aged 18–21 years only, reports that elevated levels of ESR are associated with an increased risk of low HRQL score by 7.7% after adjustments [26]. Among studies including both sexes and a large sample size, low HRQL is related to higher levels of CRP only among women in an Asian population although a lower CRP threshold was defined as LGI (hsCRP above 1 mg/L) [23]. By contrast, another study including 13,236 participants found similar associations only among men; however, the age range was 24–34 [24] compared with 40–69 years [23]. Our reported negative association of LGI on physical health among both genders is in concordance with one previous study with 13,773 apparently healthy individuals from a routine health examination at a medical centre demonstrating higher levels of CRP among the self-rated health ‘average’ group compared to ‘excellent’, however, individuals reporting ‘very poor’ and ‘poor’ were not included [25].

Relation between lifestyle factors and HRQL

The current study was able to replicate the independent negative relation between PCS and lifestyle factors with respect to current smoking, physical activity, and obesity in a cohort of healthy individuals. These results are in agreement with previous studies that report a similar association with low HRQL among participants selected by sex, age, and region in the population [49,50]. Unlike previous studies, we find that the negative impact of smoking on PCS is partly mediated through LGI, however, the percentage of the effect mediated through LGI is marginal.

Further, we found no association between consumption of meat or fish and HRQL; a finding that agrees well with previous study which fails to demonstrate any association between HRQL and Mediterranean Diet Score (low-dietary-quality vs. high-dietary-quality)[50]. Waist circumference or rather abdominal obesity as a predictor of LGI has been described [27]; however, to our knowledge no other studies have shown abdominal obesity as a predictor of reduced PCS or low HRQL. Contrary to previous studies, we investigated the effects of physical activity at work and in leisure-time separately with findings in accordance with recent studies with regards to the effect of physical activity in leisure-time [26,49,50]. Physical activity at work reduced PCS among men and women not using OC; whereas the same trend among women using OC was statistically insignificant. The lack of significance could be explained by lack of statistical power in the latter stratum.

Recent results from a study including 1,546 adults aged 21–100 years suggest surprising observation that aging is associated with improved mental health among older adults, despite loss of physical and cognitive function [51]. In our study, we did indeed find that aging improved MCS and reduced PCS among women not using OC, and among men. The lack of a significant relation between aging and reduction of PCS among women using OC could possibly be explained by a lower and more narrow age group (median age is 26.5 years among women using OC and 38.4 years among women not using OC [27]) which is related to the age dependent difference of choice of contraception among women. Various explanations for an improved mental health in later life have been proposed, including increased wisdom and experience with aging, or a suggestion that older individuals tend to be more emotionally balanced, better at complex social decision-making, and tend to process information with a more positive attitude. Such ballast in life and a generally “positive outlook” could explain higher level of subjective well-being in later life [52].

Strengths and limitations of the study

Our study comprised a large and hom*ogeneous population of healthy individuals, in as much as because blood donors must comply with strict criteria to be allowed to donate and are permanently excluded from blood donation if diagnosed with certain chronic diseases, including diabetes, cancer, hypertension, or even hypercholesterolemia. Because blood donors are a selected group, being generally healthier than the background population, they are uniquely suitable for studies of health associations among individuals without clinical symptoms. In addition, the age range of our study population was broad (18–67 years), a more balanced sex distribution compared to previous studies [23,25], and our use of HRQL with the SF-12 questionnaire as descriptive measure of self-rated health rather than single-item measure [17–19,23,24,26] gives more information regarding perceived physical and mental health. Further, participants with more than one missing value in the SF-12 items were excluded, even though up to two missing values is generally acceptable in studies of this kind. Moreover, our study included not only a mediation analyses but also multivariable linear regression analyses that essentially report similar effects; however, the contribution of the mediation analyses gives further understanding of whether other predictors lead to LGI which then affects HRQL and to which extent the effect of predictors on physical and mental HRQL is mediated through LGI. Our finding indicates that LGI had an independent impact on HRQL scores, although a partial and small effect of smoking on PCS is mediated through LGI.

The mediation analyses and linear regression analysis was supplemented with multiple imputation modelling in the analysis strategy, which is an accepted and recommended procedure for handling missing data and to obtain more accurate and unbiased effect estimates [46,47]. In the current study, the group with missing values overall did not match completely with the study population with complete data in terms of distribution of sex and other predictors, as mentioned earlier. Multiple imputation did not alter direction of effects. In addition, our statistical analysis approach prompted inclusion of relevant predictors in adjusting for confounders. Our study is limited by relying on CRP as the only biomarker of low-grade inflammation, where additional, simultaneous biomarkers would be preferable. Finally, the cross-sectional design precludes any conclusion regarding causal relationships.

S1 File

Danish and English questionnaires for the Danish Blood Donor Study.

(PDF)

S2 File

Table A. Missing group (n = 2,515). Numbers (%) or medians (interquartile ranges) of participants in the particular predictor variable with values available. BMI: body mass index, PCS: physical component score, MCS: mental component score, LGI defined as 3 mg/L < CRP ≤ 10 mg/L, CRP: C-reactive protein. * Number of participants with missing values in the particular predictor variable. Table B. Excluded participants with C-reactive protein levels above 10 mg/L (n = 176). Numbers (%) or medians (interquartile ranges). Table C. Sensitivity analysis, predictors of PCS and MCS estimated by multivariable linear regression analysis of case-complete participants with C-reactive protein level > 10 mg/L included (n = 14,547). RC: regression coefficient, OC: oral contraception, PCS: physical component score, MCS: mental component score, CRP: C-reactive protein, BMI: body mass index, LGI: low-grade inflammation. Note: Analysis performed without missing responses in abdominal obesity variable. Table D. Diet and childbirth as predictors of PCS and MCS estimated by univariate linear regression analysis of case-complete participants (n = 15,380). RC: regression coefficient, OC: oral contraception, PCS: physical component score, MCS: mental component score, CI: confidence interval. * Meat consumption defined as meat intake or vegetarian. ** Fish consumption defined as fish intake more than twice weekly (yes) or not (no). Table E. Predictors of PCS and MCS estimated by multivariable linear regression analysis after transformation of data by power of four (MCS) and five (PCS). RC: regression coefficient, OC: oral contraception, PCS: physical component score, MCS: mental component score, CRP: C-reactive protein, BMI: body mass index, LGI defined as 3 mg/L < CRP ≤ 10 mg/L.

(PDF)

1. Pepys MB, Hirschfield GM. C-reactive protein: a critical update. J Clin Invest. 2003. June;111(12):1805–12. 10.1172/JCI18921 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

2. Rifai N, Ridker PM. Population Distributions of C-reactive Protein in Ap- parently Healthy Men and Women in the United States: Implication for Clinical Interpretation. Clin Chem. 2003;49(4):666–9. [PubMed] [Google Scholar]

3. Imhof A, Fröhlich M, Loewel H, Helbecque N, Woodward M, Amouyel P. Distributions of C-reactive Protein Measured by High-Sensitivity Assays in Apparently Healthy Men and Women from Different Populations in Europe. Clin Chem. 2003. April1;49(4):669–72. [PubMed] [Google Scholar]

4. Allin KH, Nordestgaard BG. Elevated C-reactive protein in the diagnosis, prognosis, and cause of cancer. Crit Rev Clin Lab Sci. 2011. August;48(4):155–70. 10.3109/10408363.2011.599831 [PubMed] [CrossRef] [Google Scholar]

5. Hage FG, Szalai AJ. C-Reactive Protein Gene Polymorphisms, C-Reactive Protein Blood Levels, and Cardiovascular Disease Risk. J Am Coll Cardiol. 2007. September18;50(12):1115–22. 10.1016/j.jacc.2007.06.012 [PubMed] [CrossRef] [Google Scholar]

6. Emerging Risk Factors Collaboration, Kaptoge S, Di Angelantonio E, Lowe G, Pepys MB, Thompson SG, et al. C-reactive protein concentration and risk of coronary heart disease, stroke, and mortality: an individual participant meta-analysis. Lancet. 2010. January9;375(9709):132–40. 10.1016/S0140-6736(09)61717-7 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

7. Nielen MMJ, Van Schaardenburg D, Reesink HW, Twisk JWR, Van De Stadt RJ, Van Der Horst-Bruinsma IE, et al. Increased levels of C-reactive protein in serum from blood donors before the onset of rheumatoid arthritis. Arthritis Rheum. 2004. August;50(8):2423–7. 10.1002/art.20431 [PubMed] [CrossRef] [Google Scholar]

8. Tsilidis KK, Branchini C, Guallar E, Helzlsouer KJ, Erlinger TP, Platz EA. C-reactive protein and colorectal cancer risk: a systematic review of prospective studies. Int J Cancer J Int Cancer. 2008. September1;123(5):1133–40. [PubMed] [Google Scholar]

9. Grad E, Danenberg HD. C-reactive protein and atherothrombosis: Cause or effect?Blood Rev. 2013. January;27(1):23–9. 10.1016/j.blre.2012.12.001 [PubMed] [CrossRef] [Google Scholar]

10. Protein C-Reactive, Fibrinogen, and Cardiovascular Disease Prediction. N Engl J Med. 2012;367(14):1310–20. 10.1056/NEJMoa1107477 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

11. Nielsen ABS, Siersma V, Kreiner S, Hiort LC, Drivsholm T, Eplov LF, et al. The impact of changes in self-rated general health on 28-year mortality among middle-aged Danes. Scand J Prim Health Care. 2009. Jan;27(3):160–6. 10.1080/02813430903020446 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

12. Saarni SI, Suvisaari J, Sintonen H, Koskinen S, Härkänen T, Lönnqvist J. The health-related quality-of-life impact of chronic conditions varied with age in general population. J Clin Epidemiol. 2007. December;60(12):1288.e1–1288.e11. [PubMed] [Google Scholar]

13. Gill SC, Butterworth P, Rodgers B, Mackinnon A. Validity of the mental health component scale of the 12-item Short-Form Health Survey (MCS-12) as measure of common mental disorders in the general population. Psychiatry Res. 2007. July;152(1):63–71. 10.1016/j.psychres.2006.11.005 [PubMed] [CrossRef] [Google Scholar]

14. Steenstrup T, Pedersen OB, Hjelmborg J, Skytthe A, Kyvik KO. Heritability of Health-Related Quality of Life: SF-12 Summary Scores in a Population-Based Nationwide Twin Cohort. Twin Res Hum Genet. 2013. June;16(03):670–8. [PMC free article] [PubMed] [Google Scholar]

15. Wang H-M, Beyer M, Gensichen J, Gerlach FM. Health-related quality of life among general practice patients with differing chronic diseases in Germany: Cross sectional survey. BMC Public Health [Internet]. 2008. December [cited 2016 Nov 10];8(1). Available from: http://bmcpublichealth.biomedcentral.com/articles/10.1186/1471-2458-8-246 [Google Scholar]

16. Møller L, Kristensen TS, Hollnagel H. Self rated health as a predictor of coronary heart disease in Copenhagen, Denmark: J Epidemiol Community Health; 1996;50(4):423–428. [PMC free article] [PubMed] [Google Scholar]

17. Andreasson AN, Szulkin R, Undén A-L, von Essen J, Nilsson L-G, Lekander M. Inflammation and positive affect are associated with subjective health in women of the general population. J Health Psychol. 2013;18(3):311–320. 10.1177/1359105311435428 [PubMed] [CrossRef] [Google Scholar]

18. Janszky I, Lekander M, Blom M, Georgiades A, Ahnve S. Self-rated health and vital exhaustion, but not depression, is related to inflammation in women with coronary heart disease. Brain Behav Immun. 2005. November;19(6):555–63. 10.1016/j.bbi.2005.01.001 [PubMed] [CrossRef] [Google Scholar]

19. Lekander M, Elofsson S, Neve I-M, Hansson L-O, Undén A-L. Self-rated Health Is Related to Levels of Circulating Cytokines: Psychosom Med. 2004. July;66(4):559–63. 10.1097/01.psy.0000130491.95823.94 [PubMed] [CrossRef] [Google Scholar]

20. Undén A-L, Andréasson A, Elofsson S, Brismar K, Mathsson L, Rönnelid J, et al. Inflammatory cytokines, behaviour and age as determinants of self-rated health in women. Clin Sci. 2007. March1;112(6):363–73. 10.1042/CS20060128 [PubMed] [CrossRef] [Google Scholar]

21. Christian LM, Glaser R, Porter K, Malarkey WB, Beversdorf D, Kiecolt-Glaser JK. Poorer self-rated health is associated with elevated inflammatory markers among older adults. Psychoneuroendocrinology. 2011. November;36(10):1495–504. 10.1016/j.psyneuen.2011.04.003 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

22. Cohen HJ, Pieper CF, Harris T, Rao KMK, Currie MS. The association of plasma IL-6 levels with functional disability in community-dwelling elderly. J Gerontol A Biol Sci Med Sci. 1997;52(4):M201–M208. [PubMed] [Google Scholar]

23. Tanno K, Ohsawa M, Onoda T, Itai K, Sakata K, Tanaka F, et al. Poor self-rated health is significantly associated with elevated C-reactive protein levels in women, but not in men, in the Japanese general population. J Psychosom Res. 2012. September;73(3):225–31. 10.1016/j.jpsychores.2012.05.013 [PubMed] [CrossRef] [Google Scholar]

24. Shanahan L, Bauldry S, Freeman J, Bondy CL. Self-rated health and C-reactive protein in young adults. Brain Behav Immun. 2014. February;36:139–46. 10.1016/j.bbi.2013.10.020 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

25. Leshem-Rubinow E, Shenhar-Tsarfaty S, Milwidsky A, Toker S, Shapira I, Berliner S, et al. Self-rated health is associated with elevated C-reactive protein even among apparently healthy individuals. Isr Med Assoc J IMAJ. 2015. April;17(4):213–8. [PubMed] [Google Scholar]

26. Warnoff C, Lekander M, Hemmingsson T, Sorjonen K, Melin B, Andreasson A. Is poor self-rated health associated with low-grade inflammation in 43 110 late adolescent men of the general population? A cross-sectional study. BMJ Open. 2016;6(4):e009440 10.1136/bmjopen-2015-009440 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

27. Sørensen CJ, Pedersen OB, Petersen MS, Sørensen E, Kotzé S, Thørner LW, et al. Combined Oral Contraception and Obesity Are Strong Predictors of Low-Grade Inflammation in Healthy Individuals: Results from the Danish Blood Donor Study (DBDS). PLoS ONE. 2014. February6;9(2):e88196 10.1371/journal.pone.0088196 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

28. Kaspersen KA, Pedersen OB, Petersen MS, Hjalgrim H, Rostgaard K, Møller BK, et al. Obesity and Risk of Infection: Results from the Danish Blood Donor Study. Epidemiology. 2015. July;26(4):580–9. 10.1097/EDE.0000000000000301 [PubMed] [CrossRef] [Google Scholar]

29. Rigas AS, Sørensen CJ, Pedersen OB, Petersen MS, Thørner LW, Kotzé S, et al. Predictors of iron levels in 14,737 Danish blood donors: results from the Danish Blood Donor Study. Transfusion (Paris). 2014. March;54(3 Pt 2):789–96. [PMC free article] [PubMed] [Google Scholar]

30. Pedersen OB, Erikstrup C, Kotzé SR, Sørensen E, Petersen MS, Grau K, et al. The Danish Blood Donor Study: a large, prospective cohort and biobank for medical research. Vox Sang. 2012. April;102(3):271–271. 10.1111/j.1423-0410.2011.01553.x [PubMed] [CrossRef] [Google Scholar]

31. Perneger TV, Burnand B. A simple imputation algorithm reduced missing data in SF-12 health surveys. J Clin Epidemiol. 2005. February;58(2):142–9. 10.1016/j.jclinepi.2004.06.005 [PubMed] [CrossRef] [Google Scholar]

32. Steenstrup T, Pedersen OB, Hjelmborg J, Skytthe A, Kyvik KO. Heritability of Health-Related Quality of Life: SF-12 Summary Scores in a Population-Based Nationwide Twin Cohort. Twin Res Hum Genet. 2013;16(03):670–8. [PMC free article] [PubMed] [Google Scholar]

33. Bjørner JB, Damsgaard M, Watt T, Bech P, Rasmussen N, Kristensen T. Danish SF-36 manual—a health status questionnaire. Copenhagen: Lægemiddelindustriforeningen; [Internet]. Available from: 2003 [Google Scholar]

34. Gandek B, Ware JE, Aaronson NK, Apolone G, Bjorner JB, Brazier JE, et al. Cross-validation of item selection and scoring for the SF-12 Health Survey in nine countries: results from the IQOLA Project. International Quality of Life Assessment. J Clin Epidemiol. 1998. November;51(11):1171–8. [PubMed] [Google Scholar]

36. Kaspersen KA ard, Dinh KM, Erikstrup LT, Burgdorf KS, Pedersen OB, Sørensen E, et al. Low-Grade Inflammation Is Associated with Susceptibility to Infection in Healthy Men: Results from the Danish Blood Donor Study (DBDS). PloS One. 2016;11(10):e0164220 10.1371/journal.pone.0164220 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

37. Ishii S, Karlamangla AS, Bote M, Irwin MR, Jacobs DR, Cho HJ, et al. Gender, Obesity and Repeated Elevation of C-Reactive Protein: Data from the CARDIA Cohort. Kiechl S, editor. PLoS ONE. 2012. April30;7(4):e36062 10.1371/journal.pone.0036062 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

38. Lange T, Vansteelandt S, Bekaert M. A Simple Unified Approach for Estimating Natural Direct and Indirect Effects. Am J Epidemiol. 2012. August1;176(3):190–5. 10.1093/aje/kwr525 [PubMed] [CrossRef] [Google Scholar]

39. Sørensen CJ, Pedersen OB, Petersen MS, Sørensen E, Kotzé S, Thørner LW, et al. Combined Oral Contraception and Obesity Are Strong Predictors of Low-Grade Inflammation in Healthy Individuals: Results from the Danish Blood Donor Study (DBDS). PLoS ONE. 2014. February6;9(2):e88196 10.1371/journal.pone.0088196 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

40. Afzal S, Bojesen SE, Nordestgaard BG. Low Plasma 25-Hydroxyvitamin D and Risk of Tobacco-Related Cancer. Clin Chem. 2013. January5;59(5):771–80. 10.1373/clinchem.2012.201939 [PubMed] [CrossRef] [Google Scholar]

43. Krauss RM, Eckel RH, Howard B, Appel LJ, Daniels SR, Deckelbaum RJ, et al. AHA Dietary Guidelines: revision 2000: A statement for healthcare professionals from the Nutrition Committee of the American Heart Association. Stroke J Cereb Circ. 2000. November;31(11):2751–66. [PubMed] [Google Scholar]

44. Lee J, Nordestgaard BG, Dahl M. Elevated ACE activity is not associated with asthma, COPD, and COPD co-morbidity. Respir Med. 2009. September;103(9):1286–92. 10.1016/j.rmed.2009.04.003 [PubMed] [CrossRef] [Google Scholar]

45. Rigas AS, Pedersen OB, Sørensen CJ, Sørensen E, Kotzé SR, Petersen MS, et al. No association between iron status and self-reported health-related quality of life in 16,375 Danish blood donors: results from the Danish Blood Donor Study. Transfusion (Paris). 2015. July1;55(7):1752–6. [PubMed] [Google Scholar]

46. Sterne JAC, White IR, Carlin JB, Spratt M, Royston P, Kenward MG, et al. Multiple imputation for missing data in epidemiological and clinical research: potential and pitfalls. BMJ. 2009. September1;338(jun29 1):b2393–b2393. 10.1136/bmj.b2393 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

47. Little RJ, D’Agostino R, Cohen ML, Dickersin K, Emerson SS, Farrar JT, et al. The prevention and treatment of missing data in clinical trials. N Engl J Med. 2012;367(14):1355–1360. 10.1056/NEJMsr1203730 [PMC free article] [PubMed] [CrossRef] [Google Scholar]

48. Fielding R, Lam TH, Ho SY, Janus ED. Subjective health and fibrinogen in a healthy Chinese cohort. Br J Health Psychol. 2004;9(4):523–532. [PubMed] [Google Scholar]

49. Molarius A, Berglund K, Eriksson C, Lambe M, Nordstrom E, Eriksson HG, et al. Socioeconomic conditions, lifestyle factors, and self-rated health among men and women in Sweden. Eur J Public Health. 2007. February13;17(2):125–33. 10.1093/eurpub/ckl070 [PubMed] [CrossRef] [Google Scholar]

50. Haveman-Nies A, De Groot LC, Van Staveren WA. Relation of dietary quality, physical activity, and smoking habits to 10-year changes in health status in older Europeans in the SENECA study. Am J Public Health. 2003;93(2):318–323. [PMC free article] [PubMed] [Google Scholar]

51. Thomas ML, Kaufmann CN, Palmer BW, Depp CA, Martin AS, Glorioso DK, et al. Paradoxical Trend for Improvement in Mental Health With Aging: A Community-Based Study of 1,546 Adults Aged 21–100 Years. 2016. August;77(8):1019–25. [PMC free article] [PubMed] [Google Scholar]

52. Grossmann I, Na J, Varnum ME, Park DC, Kitayama S, Nisbett RE. Reasoning about social conflicts improves into old age. Proc Natl Acad Sci. 2010;107(16):7246–7250. 10.1073/pnas.1001715107 [PMC free article] [PubMed] [CrossRef] [Google Scholar]