Exploring gendered practices by social position in epidemiology: the gender score applied to Germany

Background

Gender as a relational concept is rarely taken into account in epidemiology, yet an in-depth reflection on gender conceptualisation and operationalisation can advance gender analysis in quantitative health research, allowing for more valid evidence to support public health interventions. We constructed a context-specific gender score to assess how its discriminatory power differed in sub-groups defined by social positions investigated in intersectional analyses, i.e. sex/gender, race, class, age and sexual orientation.

Methods

We created a gender score based on gendered social practices on a masculinity-femininity continuum (ranging from 0-1) using data of the German Socioeconomic Panel. With density plots, we exploratively compared distributions of gendered social practices and their variation across social groups.

Results

We included 13 gender-related variables to define a gender score in our sample (n = 20,767). Variables on family and household structures presented with the highest weight for the gender score. Calculating tertiles, the score ranged between 0.01 and 0.68, 0.69 and 0.93 and 0.94 to 1 in women and 0 and 0.18, 0.19 and 0.40 and 0.41 to 1 in men, showing that the distribution for women is more skewed than for men. Comparing social groups, we saw that young individuals, those without children, not living with a partner or currently living in a same-sex/gender partnership, showed more overlap between feminine/masculine social practices among men and women.

Conclusions

Our explorative findings showed that the distribution of gendered social practices differed among social groups, which empirically backed up the theoretical notion of gender being a context-specific construct. Economic participation and household structures remain essential drivers of heterogeneity in practices among women and men in most social positions. Concerted efforts must continue to overcome these gender (in)equalities - which are important determinants of health inequalities.

Key messages

Family and household structures are crucial to constructing the gender score, indicating that interpersonal relationships are key determinants of gendered social practices. Including gendered practices in representative health monitoring data could allow for differentiation of biological sex and socially constructed gendered practices that drive health inequalities.

Extremely elevated lipoprotein (a) as an independent risk factor for peripheral arterial disease in large patient cohort

Background

Elevated lipoprotein (a) [LP (a)] levels are an independent, genetic, and causal factor for cardiovascular disease and associated with myocardial infarction (MI). Although the association between circulating levels of lipoprotein(a) [Lp(a)] and risk of coronary artery disease (CAD) is well established, its role in risk of peripheral arterial disease (PAD) remains unclear. PAD affects over 236 million individuals and follows ischaemic heart disease (IHD) and cerebrovascular disease (CVD) as the third leading cause of atherosclerotic cardiovascular morbidity worldwide. LP (a) is genetically determined, stable throughout life and yet refractory to drug therapy. While 30 mg/dl is considered the upper normal value for LP (a) in central Europe, extremely high LP (a) levels (>150mg/dl) are rare in the general population.

The aim of our study was to analyse the correlation between lipoprotein (a) [LP (a)] levels and an incidence of PAD in high-risk patients.

Patients and methods

We reviewed the LP (a) concentrations of 52.898 consecutive patients admitted to our cardiovascular center between January 2004 and December 2014. Of these, 579 patients had LP (a) levels above 150 mg/dl (mean 181.45±33.1mg/dl). In the control collective LP (a) was <30mg/dl (n=350). Other atherogenic risk factors in this group were HbA1c 6.58±1.65%, low density lipoprotein (LDL) 141.99±43.76 mg/dl, and body mass index 27.81±5.61. 54.40% were male, 26.07% were smokers, 93.2% had hypertension, and 24% had a family history of cardiovascular diseases. More than 82.6% were under statins. The mean glomerular filtration rate (GFR) was 69.13±24.8 ml/min [MDRD (Modification of Diet in Renal Disease)].

Results

45.00% (n=261) of the patients with LP (a) >150mg/dl had PAD. The prevalence of PAD in patients with LP (a) <30mg/dl in our control collective was 15.8%. (P- Value 0.001). Patients with LP (a) >150mg/dl had a significantly increased risk for PAD (Odds ratio 4.36, 95% CI 2.94–6.72, p: 0.001).

19.1% of patients were re-vascularized by percutaneous angioplasty (PTA) and 7.09% of patients had to undergo peripheral vascular bypass (PVB). Mean LP (a) level in patients with PAD was 182.6±31.61.

Conclusion

Elevated LP (a) levels above 150 mg/dl are associated with a significantly increased risk of PAD in our collective and it confirms our hypothesis. Over one fourth of these patients had severe PAD and requiring revascularization therapy. We need more prospective studies to confirm our findings.

Funding Acknowledgement

Type of funding source: None

Social cohesion as a mechanism behind neighbourhood effects on mental or physical health

Background

The neighbourhood where an individual resides have become a relevant element in the study of health inequalities. While there is substantial evidence that environmental factors affect health, far less evidence of the hypothesised role of social mechanisms in the causal chain between neighbourhood characteristics and health is available.

Methods

Using data from the longitudinal German SOEP panel, we built measures of perceived neighbourhood characteristics and a measure of social cohesion for participant living in urban areas. Participants were included in a longitudinal cohort of 4 year observation time starting in 2004. Using linear regression models, we tested the role of social cohesion as a mediator between the neighbourhood characteristics at baseline and a health outcome (mental and physical component of the SF-12) at the end of follow-up adjusted for socio-economic variables.

Results

For the 4 year follow-up 10 665 participants were included. There was no evidence of a mediation effect of our measure of social cohesion between geographical or institutional characteristics and physical health. The effect of social cohesion on mental health was 2.03 ([1.44, 2.59]) and on physical health 1.78 ([1.27, 2.29]). Social cohesion mediated associations between biological (17%, 11%) or built (11%, 14%) characteristics and both physical and mental health and association between institutional characteristics and mental health (7%).

Discussion

We found evidence for a possible role of social cohesion in the pathway between neighbourhood characteristics and health. Only perceived characteristics have been used in this work and the respective role of objective and subjective measures of neighbourhood characteristics should be further investigated. Moreover, validated measures of social cohesion should be developed.

Key messages

The perception of social cohesion mediates the effect of specific perceived characteristics of neighbourhood on mental and physical health. Interventions seeking to improve social cohesion in neighbourhoods might have positive consequences on the health and wellbeing of inhabitants.

Adequate utilisation of emergency services in Germany: a differential by migration background?

Background

The role of emergency services (ES) is to provide a round-the-clock acute care. In recent years, inadequate use of emergency services has been internationally thematised because of overcrowding and the cost associated. There is only limited evidence regarding the pattern of utilisation of ES by populations with a migration background.

Methods

Consecutive patients visiting three ES in Berlin from July 2017 to July 2018, and who fulfilled inclusion criteria such as being conscious, were recruited into the study. A questionnaire covered medical conditions, reasons to visit the ES, and socio-economic factors. A criterion for the adequacy of utilisation was developed using a sub-sample of patients with information provided by physicians. Differences between migrants (1st generation), their offspring (2nd generation), and non-migrants were evaluated using logistic regression.

Results

2 327 patients were included, 901 had a migration background. The utilisation was adequate if the patient was admitted to hospital and/or if all three following criterion were fulfilled: reported to have been advised by a physician to visit the ES; reported strong pain; and reported a high perceived urgency (both ≥7 on a scale from 0 to 10). Adjusting for gender, age, condition and number of recent visits to ES, 1st generation migrants had a significantly higher chance than non-migrants to have an inadequate utilisation of services (OR 1.30; 95% CI [1.01; 1.68]). For 2nd generation persons, this was not statistically significant.

Conclusions

First generation migrants have a higher chance of inadequate ES use compared to non-migrants. A similar study 20 years ago had similar findings. This implies that existing information leaflets on ES addressing specific population groups are not sufficient to improve adequacy of use. Structural changes in the healthcare system as well as improved methods of communication respecting the needs of diverse subgroups of patients need to be considered.

Key messages

Existing information leaflets on emergency services addressing specific population groups are not sufficient to improve adequacy of use. Structural changes in the healthcare system as well as improved methods of communication respecting the needs of diverse subgroups of patients need to be considered.

An ego-centred approach for the evaluation of health inequalities in urban areas

Background

Neighbourhood possesses attributes, structural, physical and social, for which pathways to health inequalities could be hypothesized. Hence, neighbourhood is a complex mixture of factors which cannot be simply defined by a delineation on a map, making common definitions of neighbourhood (e.g. administrative borders) problematic. We present a new concept for the evaluation of contextual health inequalities in an urban setting.

Methods

An ego-centred approach to neighbourhood effects on health allows to establish to what degree the health outcomes of a person are on average correlated to the health outcomes of his/her neighbours. This approach does not necessitate the definition of what a neighbourhood is, or of its boundaries. Using data from the BaBi birth cohort following up 958 mother-child pairs in Bielefeld/Germany we illustrate how the method provides information about the spatial structure of a possible association between unmeasured neighbourhood factors and birthweight. Spatially correlated birthweight indicates a neighbourhood effect on maternal health.

Results

A parametric model of the correlation structure gives two indicators: a distance after which health outcomes are no longer correlated (practical range), and the strength of correlation (RSV). We modelled birthweight directly and residuals after controlling for (spatially correlated) covariates. After adjusting for the mother’s demographics and neighbourhood characteristics, birthweights remained spatially correlated with RSV of 11% and a practical range of 128 m.

Conclusions

Modelling the spatial correlation of a health outcome provides a measure of the degree of health correlation, thus offering new evidence on the production of health inequalities while incorporating current modelling approaches. Moreover, it measures heterogeneity in a city. This could be used as an indicator for policy makers or town planners to identify areas in need of socioeconomic investment.

Key messages

Modelling the spatial correlation of health outcomes is an approach which enable to assess unmeasured neighbourhood effects. The health correlation neighbourhood approach helps to investigate the production of health inequalities and to identify urban areas in need of socioeconomic investment.

P5528Extremely elevated lipoprotein (a) as an independent risk factor for coronary heart disease in large patient cohort

Background

Elevated lipoprotein (a) [LP (a)] levels are an independent risk factor for coronary heart disease (CHD) and associated with myocardial infarction (MI). CHD took a devastating toll in Europe and in the United States in the 20th century, killing more people than any other disease. It remains the leading cause of death in most countries worldwide. CHD shares risk factors with atherosclerosis. It has been shown that elevated LP (a) levels are associated with an increased risk for CHD across various ethnic groups. LP (a) is genetically determined, stable throughout life and yet refractory to drug therapy. While 30 mg/dl is considered the upper normal value for LP (a) in central Europe, extremely high LP (a) levels (>150mg/dl) are rare in the general population.

The aim of our study was to analyse the correlation between lipoprotein (a) [LP (a)] levels and an incidence of coronary heart disease (CHD) in high-risk patients.

Patients and methods

We reviewed the LP (a) concentrations of 52.898 consecutive patients admitted to our cardiovascular center between January 2004 and December 2014. Of these, 579 patients had LP (a) levels above 150 mg/dl (mean 181.45±33.1mg/dl). In the control collective LP (a) was <30mg/dl (n=350). Other atherogenic risk factors in this group were HbA1c 6.58±1.65%, low density lipoprotein (LDL) 141.99±43.76 mg/dl, and body mass index 27.81±5.61. 54.40% were male, 26.07% were smokers, 93.2% had hypertension, and 24% had a family history of cardiovascular diseases. More than 82.6% were under statins. The mean glomerular filtration rate (GFR) was 69.13±24.8 ml/min [MDRD (Modification of Diet in Renal Disease)].

Results

64.98% (n=373) of the patients with LP (a) >150mg/dl had CHD. The prevalence of CHD in patients with LP (a) <30mg/dl in our control collective was 37.14%. (P- Value 0.0001). Patients with LP (a) >150mg/dl had a significantly increased risk for CHD (Odds ratio 5.98). 12.72% (n=73) of these patients suffered from CHD with single-vessel disease (VD), 14.63% (n=84) from CHD with 2VD and 37.63% (n=216) from CHD with 3VD. 47.92% of patients were re-vascularized by percutaneous coronary angioplasty (PTCA) and 37.06% of patients had to undergo coronary artery bypass grafting (CABG). 19.13% of patients had both, PTCA and CABG. Mean LP (a) level in patients with 1-vessel CHD was 181.5±29.98, in patients with 2-vessel CHD 178.94±34.26 and in patients with 3-vessel CHD 180.97±32.38 mg/dl.

Conclusion

Elevated LP (a) levels above 150 mg/dl are associated with a significantly increased risk of CHD in our collective and it confirms our hypothesis. Most of these patients had severe CHD with 3-vessel disease (VD) requiring coronary revascularization therapy. We need more prospective studies to confirm our findings.

The influence of migration on women's satisfaction during pregnancy and birth: results of a comparative prospective study with the Migrant Friendly Maternity Care Questionnaire (MFMCQ)

INTRODUCTION

Approximately 21% of Germany's inhabitants have been born abroad or are of direct descent of immigrants. A positive birth experience has an effect on a woman's mental health and her future family planning choices. While international studies showed that immigrant women are less satisfied with their birth experience, no such study has been conducted in Germany until now.

METHODS

At our center of tertiary care in Berlin, with approximately 50% immigrants among patients, pregnant women of at least 18 years of age were offered participation in this study. A modified version of the Migrant Friendly Maternity Care Questionnaire (MFMCQ) designed by Gagnon et al. in German, English, French, Spanish, Arabic and Turkish was used. We compared non-immigrant women to immigrant women and women with direct descent of immigrants. For certain analysis, the latter two groups were included together under the category "migration background".

RESULTS

During the study period, 184 non-immigrant, 214 immigrant women and 62 direct descendants of immigrants were included. The most frequent countries of origin were Syria (19%), Turkey (17%), and Lebanon (9%). We found a slight difference between groups regarding age (non-immigrants: mean 33 years versus women with any migration background: mean 31) as well as parity with more non-immigrants delivering their first child. No difference in the satisfaction with care was observed between immigrant and any migration background groups (p ≥ 0.093 in the two-sided Fisher's exact test). At least 75.8% of all participating women reported complete satisfaction with care during labor, birth and after birth. Interestingly, the level of German language proficiency did not influence the immigrant patient's satisfaction with care.

CONCLUSION

The study results show no difference regarding overall satisfaction with care during labor and birth despite a relevant language barrier. We are for the first time providing the MFMCQ in German and Turkish. Further future analyses on the impact of patient expectations on satisfaction with care will be conducted.

Evaluation of tubomanometry as a routine diagnostic tool for chronic obstructive Eustachian tube dysfunction

OBJECTIVE

The objective of this study was to demonstrate the reliability of tubomanometry (TMM) described by Estéve in the diagnosis of chronic obstructive Eustachian tube (ET) dysfunction.

STUDY DESIGN

Combined prospective and retrospective clinical study.

SETTING

Tertiary referral centre, affiliated to university.

METHODS

Two hundred and fifteen healthy subjects were examined once, 25 healthy subjects underwent TMM weekly for 6 weeks, and six healthy subjects were tested three times a day on at least three different days. The results of tubomanometry in healthy subjects were compared to data obtained from 171 patients with chronic obstructive ET dysfunction.

RESULTS

In healthy subjects, there was an immediate opening of the ET at 30-50 mbar with an R-value ≤ 1 in at least 94% of the cases. In patients with chronic ET dysfunction, an opening of the ET could be registered in only 42% of patients at 30 mbar and in 58% at 50 mbar. The average of the R-value in these subjects always indicated towards a delayed opening (R > 1). When measurements are repeated in the same subject with a weekly interval, the intraclass correlation (ICC) was 0.49 for the TMM with 30 mbar, 0.51 for the TMM with 40 mbar and 0.52 for the TMM with 50 mbar in healthy people. For the patients with symptoms of ET dysfunction, the ICC for up to four repeated measures was 0.50 for the TMM with 30 mbar, 0.53 for the TMM with 40 mbar and 0.54 for the TMM with 50 mbar. A complete agreement of the results in repeated measurements within seconds was present in 86% for 30 and 40 mbar and in 79% for 50 mbar. The ICC was 0.61 for the TMM with 50 mbar, 0.62 for the TMM with 40 mbar and 0.68 for the TMM with 30 mbar.

CONCLUSIONS

Tubomanometry can support the diagnosis of ET dysfunction. An R-value ≤ 1 indicates a regular function of the ET, an R-value >1 indicates a delayed opening of the ET, and no definable R-value means no detectable opening of the ET. TMM is a reliable and valid instrument to support the diagnosis of chronic obstructive ET dysfunction.

Self-rated treatment outcomes in medical rehabilitation among German and non-German nationals residing in Germany: an exploratory cross-sectional study

Background

In many European countries, foreign nationals experience, on average, less favorable treatment outcomes in rehabilitative care than the respective majority population. In Germany, this for example is reflected in a lower occupational performance and a higher risk of disability retirement after rehabilitation as analyses of routine data show. However, little is known about the perspective of health care users. The aim of the present study was to compare self-rated treatment outcomes between German and non-German nationals undergoing in-patient medical rehabilitation in Germany.

Methods

We analyzed data from a cross-sectional representative rehabilitation patient survey of 239,811 patients from 642 clinics in Germany who completed about 3 weeks of in-patient rehabilitative treatment. The self-rating of the treatment outcome was based on a dichotomized Likert scale consisting of three items. A multilevel logistic regression analysis adjusted for various demographic, socio-economic, health and other covariates was conducted to examine differences in the self-rated treatment outcome between German and non-German nationals.

Results

Of the 239,811 respondents 0.9 % were nationals from Turkey, 0.8 % had a nationality from a former Yugoslavian country, 0.9 % held a nationality from the South European countries Portugal, Spain, Italy or Greece and 1.9 % were nationals from other countries. Non-German nationals reported a less favorable self-rated outcome than Germans. Adjusted odds ratios [OR] for reporting a less favorable treatment outcome were 1.24 (95 %-confidence interval [95 %-CI]: 1.12–1.37) for nationals from the South European countries Portugal/Spain/Italy/Greece, 1.62 (95 %-CI: 1.45–1.80) for Turkish nationals and 1.68 (95 %-CI: 1.52–1.85) for nationals from Former Yugoslavia.

Conclusions

Knowledge on health outcomes from the patients’ point of view is important for the provision of patient-centered health care. Our study showed that non-German nationals report less favorable outcomes of rehabilitative care than Germans. This may be due to cultural and religious needs not sufficiently addressed by health care providers. In order to improve rehabilitative care for non-German nationals, rehabilitative services must become sensitive to the needs of this population group. Diversity management can contribute to this process.

Routine screening for pain combined with a pain treatment protocol in head and neck cancer: a randomised controlled trial

BACKGROUND

We compared the effectiveness and cost of a pain screening and treatment program, with usual care in head and neck cancer patients with significant pain.

METHODS

Patients were screened for the presence of pain and then randomly assigned to either an intervention group, consisting of a pain treatment protocol and an education program, or to usual care. Primary outcome was change in the Pain Severity Index (PSI) over three months.

RESULTS

We screened 1074 patients of whom 156 were randomized to either intervention or usual care. Mean PSI was reduced over three months in both groups, with no significant difference between the two groups. The Pain Management Index (PMI) at three months, was significantly improved in the intervention group compared with usual care (P<0.001), as was Patient Satisfaction (mean difference in scores was statistically significant: -0.30 [-0.60 to -0.15]). All subjects reported clinically significant levels of anxiety and depression throughout the study. Treatment costs were significantly higher for intervention (mean=£400) compared with usual care (£200), with a low likelihood of being cost-effective.

CONCLUSIONS

There was no difference in the Pain Severity Index between the two groups. However there were significant improvements in the intervention group in patient satisfaction and PMI. The pain screening process itself was effective. Sufficient benefit was demonstrated as a result of the intervention to allow continued development of pain treatment pathways, rather than allowing pain treatment to be left to nonformalised ad hoc arrangements.

Estimating dichotomised outcomes in two groups with unequal variances: a distributional approach

Dichotomisation in medical research is sometimes necessary for decision-making or communication purposes. This practice has been criticised in the case of continuous data, and it has been said that means should be compared instead. However when the two groups have unequal variances, comparing means might not show the whole picture as a particular group with a risk defined by a threshold in an outcome may have been affected differently by an intervention than when there is a simple shift of distribution. A statistically sound method using a distributional approach for the dichotomisation of normally distributed outcomes has been described under the assumption of equal variances. This assumption is not sustainable in some situations, and in this work, we develop the method further to cover the case of unequal variances. Through examples from the literature and our own data, we illustrate the effect of unequal variance on dichotomised estimates and present a validation of the method through simulations.

Dichotomising continuous data while retaining statistical power using a distributional approach

Dichotomisation of continuous data is known to be hugely problematic because information is lost, power is reduced and relationships may be obscured or changed. However, not only are differences in means difficult for clinicians to interpret, but thresholds also occur in many areas of medical practice and cannot be ignored. In recognition of both the problems of dichotomisation and the ways in which it may be useful clinically, we have used a distributional approach to derive a difference in proportions with a 95% CI that retains the precision and the power of the CI for the equivalent difference in means. In this way, we propose a dual approach that analyses continuous data using both means and proportions to replace dichotomisation alone and that may be useful in certain situations. We illustrate this work with examples and simulations that show good performance of the parametric approach under standard distributional assumptions from our own research and from the literature.