The European Journal of Public Health Advance Access published online on July 22, 2008
The European Journal of Public Health, doi:10.1093/eurpub/ckn069
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Letter to the Editor |
Comment on "Induced abortion in Denmark: effect of socio-economic situation and country of birth"
Received May 2, 2008, accepted July 2, 2008A recent study by Rasch et al. in this journal concluded that ‘Immigrant women comprise a vulnerable group, with a poor socio-economic status. This situation exposes immigrant women to increased risk of induced abortion’.1 This statement seems to suggest that the excess risk of induced arbotion experienced (particularly by immigants from non-Western countries) are due to their poor socio-economic status (SES). I think that the support for this conclusion as based on the methods used and data presented by the auhors is weak. In the following, I will explain why I think this is the case. Based on the data reported by the authors, I provide a reanalysis to support my argument.
The strategy used by Rasch et al. (see table 4) is to compare the odds ratios (ORs) for ethinicity before and after control for the putative mediators occupation and income. I want to point out that the approach of comparing ORs before and after adjustment for mediators have been repeatedly shown to be error prone.2,3
An important, but implicit asumption made by Rasch et al. is that the effect of occupation and income on the risk of abortion is the same across ethnic groups, i.e. no interaction between ethnicity occupation/income. I do not think that this is a reasonable a priori assumption to make.4 In a Danish context, it has recently been shown that education and income have different effects on several reproductive outcomes in different ethnic groups.5 However, due to the careful reporting of data by Rasch et al., it is easy to produce the estimates for occupaction and income statified by ethicnity and visa versa. I read the data from Table 2 into a speadsheet application and renalysed it. Confidence intervals were omitted for the sake of brevity.
If we first consider the effects of SES within each ethnic group it appears that the socio-ecnomic gradient in induced abortion is less among non-Western immigrants than among Danes. If we then look at the ethnic disparity within SES groups, the ORs do not appear to be constant. For example, within the two most disadvantaged occupational groups the ethnic difference between non-Western immigrants and Danes is abscent (among the unemployed) or even inverse (among the unskilled). However, in all other strata of occupation the OR between Danes and non-Western immigrants are higher than the marginal OR of 2.09 between these two groups. I note that the interaction between occupation and ethnicity is statistically significant at conventional levels of significance.
So what happens when the authors assume that effects of SES are constant within the ethnic groups (and visa versa) when in fact they are not? Because the group of Danish women constitutes the majority of the sample (82% of cases, 89% of controls), this group has the biggest ‘say’ in what the effect of SES on the risk of induced abortion is. This can be seen by comparing the column labelled ‘All’ with the ones for each of the three ethnic groups in table 1: the ‘Danish’ estimates are quite close to that of ‘All’. In essence this means that in the SES-adjusted analyses ethnic differences are calculated as if SES has the effect on abortion that it has among Danish women. The consequence is that the value of SES in explaining the ethnic disparities in abortion is not correctly assessed.
|
In conclusion, I do think that the paper by Rasch et al. is a valid and important contribution, but I think that one should exert causion with regards to the paper's conclusions on the role of SES in explaining the ethnic disparities in induced abotion: the role of ethnicity and SES in relation to induced abortion might be slightly more complicated than what is indicated by Rasch et al.
Laust H Mortensen
Correspondence:Laust H. Mortensen, National Institute of Public Health, Øster Farimagsgade 5, 1399 Copenhagen K, Denmark, tel: +45 3920 7777, fax: +45 3920 8010, e-mail: lmo{at}niph.dk
Acknowledgement
The author would like to thank his colleagues at the National Institute of Public Health for providing valuable comments.
References
1 Rasch V, Gammeltoft T, Knudsen LB, et al. Induced abortion in Denmark: effect of socio-economic situation and country of birth. Eur J Public Health (2008) 18:144–9.
2 Mackinnon DP, Lockwood CM, Brown CH, et al. The intermediate endpoint effect in logistic and probit regression. Clin Trials (2007) 4:499–513.
3 Kaufman JS, MacLehose RF, Kaufman S. A further critique of the analytic strategy of adjusting for covariates to identify biologic mediation. Epidemiol Perspect Innov (2004) 1:4.[CrossRef][Medline]
4 Braveman PA, Cubbin C, Egerter S, et al. Socioeconomic status in health research: one size does not fit all. JAMA (2005) 294:2879–88.
5 Villadsen SF. Etnisk ulighed i dødfødsel og spædbarnsdød i Danmark 1981-2003. In: Masters Thesis (2007) University of Copenhagen.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||