IJE Advance Access originally published online on December 22, 2005
International Journal of Epidemiology 2006 35(2):294-298; doi:10.1093/ije/dyi287
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Article |
Mother's consent to linkage of survey data with her child's birth records in a multi-ethnic national cohort study
1 Centre for Paediatric Epidemiology and Biostatistics, Institute of Child Health, University College London, London, UK
2 Centre for Longitudinal Studies, Institute of Education, University of London, London, UK
* Corresponding author. Department of Biostatistics and Computing, Institute of Psychiatry, Kings College London, De Crespigny Park, Denmark Hill, London SE5 8AF, UK. E-mail: rosemary.tate{at}iop.kcl.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
Background The increased use of computer-based records has facilitated linkage of routine data with that obtained for research. When children are involved, parental consent for linkage is usually required. The Millennium Cohort Study, of 18 819 UK babies born in 2000–02, over-sampled families from disadvantaged and ethnic wards, providing the opportunity to investigate factors associated with mother's consent to access her child's birth records.
Methods Factors considered included ward type and mother's socioeconomic status, ethnicity, education, age, and language. Logistic regression was used to investigate the relationship of these factors with consent.
Results Consent for linkage to birth register and/or hospital maternity data was obtained from 92% of the cohort mothers. The proportions consenting differed according to the mother's country of residence, age, and education, with consent being less likely among minority ethnic group mothers, lone parents, and those with higher degrees or no qualifications. Where interviews had been translated, consent was significantly less likely if the interpreter was a male.
Conclusion A large proportion of mothers who were interviewed gave permission for linkage. However, there were some groups who were less likely to do so, particularly those from minority ethnic groups. These sources of non-consent bias should be taken into account when analysing linked data from socially and ethnically mixed populations. Efforts should be made to understand the reasons for non-consent, which in turn will help determine the best ways to encourage more mothers to consent in future.
Keywords Ethnicity, hospital records, maternal consent, Millennium Cohort Study, registries
Accepted 11 November 2005
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The increasing use of computerized records has greatly facilitated the linkage of data obtained for research purposes with routinely collected information, which can be used to augment and also to validate the data acquired for research. However, ethical guidelines in the UK and many other countries require written consent in order to use a person's health records and other confidential/personal information. Since consent will not be universal, it is important to determine whether those who consent differ from those who do not, and if so in what way, in order to allow for possible bias in analyses using linked data.1 However, few data have yet been published on this subject.2
In the Millennium Cohort Study (MCS) parents of 
19 000 babies born between 2000 and 2001 were interviewed when the infants were 9 months old. At the end of the interview mothers were asked to sign a consent form agreeing to access to records relating to her pregnancy and the baby's birth. The answers provided by the mothers, to a large range of questions about family and social circumstances, helped us to identify the factors related to this consent.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Study population
The MCS is a longitudinal survey of social, economic, and health-related factors, of babies born between September 2000 and January 2002 who were living in the UK and eligible to receive child benefit at 9 months of age. Child Benefit is a universal benefit for all children whose parents are UK residents, and it is almost universally claimed. The population was stratified by UK country and electoral ward type, defined as minority ethnic (at least 30% of the ward population ‘Black’ or ‘Asian’ at the 1991 census), disadvantaged (upper quartile of the ward-based Child Poverty Index, excluding minority ethnic wards), or advantaged (not in the top quartile of this index or ethnic). Disadvantaged wards and ethnic wards (which were only in England) were over-represented in the sample as were families living in Wales, Scotland, and Northern Ireland. Full information on the MCS and its sampling design is described elsewhere.3
Interviews were carried out with the parents of 18 553 families (for 18 819 MCS children). Since we are concerned with mother's consent, in this report we focus on the 18 505 families where the natural mother was one of the interviewees.
Mothers were asked questions about family and individual circumstances. Interviews were conducted in English, but in 4% (n = 773) of cases her partner (n = 387), a friend (n = 155), another member of the household (n = 130), or the interviewer (n = 101) translated some or all of the questions.
The factors considered in this consent study included child's sex, whether the child was the first born, mother's age, socioeconomic status, lone parenthood status (at time of interview), ethnic group, and academic qualifications. We also investigated the effect of the language spoken at home and used in the interview, and if this was not English, Welsh, or Gaelic, the gender of the translator. The mother's socioeconomic status was based on mother's occupation and classified in accordance with the National Statistics Socioeconomic Classification (NS-SEC).4 Ethnicity was categorized in accordance with guidelines from the Office for National Statistics5: British/Irish White, Other White, Mixed, Indian, Pakistani, Bangladeshi, Black Caribbean, Black African and ‘Other’ ethnic group. The white mothers were divided in order to distinguish those from the British Isles from their (mainly European) counterparts, who are likely to have different cultural characteristics.6
Consent procedures
Informed consent to participate in the MCS First survey was obtained orally at a preliminary visit by the interviewer. The families first learned about the study in a letter with accompanying information leaflet that came from the Department of Social Security (now Department for Work and Pensions) asking them to indicate if they did not want their address passed on to the survey. Those who had not opted out at that stage were given further information and another information sheet by the interviewer, who proceeded to arrange an interview if they opted in at that stage. This procedure was agreed with the Multi-centre Research Ethics Committees in 2001 to avoid the automatic exclusion of people who could not read English, which would have been entailed by a postal opt-in.
At the end of their visit, interviewers asked the mother to carefully read a copy of a form requesting: (i) to access information about her ‘pregnancy and the baby's birth from birth registration, clinical, central, or hospital records’ and (ii) to ‘follow her baby's National Health Service registration where necessary’. Interviewers were asked to read this consent form to any respondents who were not able to read it for themselves and the form was provided in several languages. If mothers were willing to give one or both permissions they were asked to sign and date the form and print their name. Respondents who did not give consent were asked to give their reasons. This was the only written consent form involved in the first survey. It contained an explanation of the linkages proposed, an assurance of confidentiality, and the uses to which the information would be put. The informants kept a copy of the form they had signed.
Receipt and completion of consent forms
A total of 17 221 consent forms were returned by the interviewers to the study organizers. After checking the signatures, 16 983 were accepted as a valid consent. Of the remainder, nine mothers subsequently withdrew consent, 19 forms received had no permissions, and 212 forms were disqualified, either because the signature could not be verified (n = 40) or because the form had been signed by the partner (n = 172). Over half (97) of the forms signed by the partner were from Indian, Pakistani, or Bangladeshi families.
Reasons for not giving consent were recorded by 1016 of the 1282 mothers for whom no form had been received. Approximately half gave no specific reasons, or just said they did not want to sign, 85 mothers said they felt they had already given enough information, and 67 said that they could not understand the form. This appeared to be mainly due to language difficulties, even though 43 had a translator, as 64 of these mothers were from a minority ethnic group, 61 of whom said that they spoke a language other than English at home.
As the aim of this paper is to examine the characteristics of mothers who gave consent, all mothers who had returned a form with permission and who had not subsequently withdrawn consent were labelled as consenters (n = 17 195). The remaining 1310 mothers were labelled as non-consenters. We included amongst consenters, anyone who provided a signed form, even if the signature could not be verified or was signed by the partner.
Analysis
Significant differences (P 
0.05) between the (weighted) proportions of consenters and non-consenters were identified using Pearson's chi-square statistic. Multiple logistic regression, using a combination of forward and backward feature selection, was used to relate the adjusted odds of consent with family characteristics and social circumstances. Adjusted Wald tests were used to test for differences between regression coefficients. All analysis was conducted using STATA 8.2 (Stata Corporation, TX, USA), using SVY commands to allow for the design effect.7
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Factors that were significantly related to consent are shown in Table 1. Non-consent was highest in Northern Ireland, and among those with no reported occupation and those who either had a higher degree or no other qualifications. Ethnicity was a significant factor with a much higher proportion of non-consent in ethnic wards, and amongst Asian and Black mothers. Both interview and home language were significantly related to non-consent and the proportion of non-consent among those speaking a language other than English at home was highest among mothers who had a male translator (this was true for mothers from all ethnic groups except Black and ‘Other’). Mother's age was significant, with the percentage of non-consent being lowest among mothers aged 20 or younger. In England non-consent was highest in London (10%, or 8% after excluding ethnic wards). The gender of the interviewer for the sample as a whole (78% female) was not significantly related to non-consent and neither was the sex or birth order of the baby.
|
Table 2 shows the adjusted odds ratios for the factors that significantly predicted non-consent after adjusting for potential confounders. Residence in Scotland and Northern Ireland was still significantly related to non-consent, after adjusting for the other factors, as was academic qualification level and lone parenthood. Most minority ethnic groups were significantly more likely (with odds ratio of 1.8 or more) not to consent than British/Irish White mothers. The addition of interview language (or language spoken at home) to the model had little effect on the odds ratios for the individual ethnic groups, and was not a significant factor (P = 0.2) for non-consent, when education was taken into account, since a much larger proportion (45% versus 12%) of ethnic minority mothers with no academic qualifications had their interviews wholly or partly translated. However, the odds were significantly higher if the interview was translated and the interpreter was male. Mother's socioeconomic status was not significant (P = 0.8) after adjusting for qualifications.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
The literature on consent to record linkage in epidemiological research is sparse2,8 and few papers investigate parental consent, probably because little information has been collected on the characteristics of non-consenters. The unique nature of the MCS provided the opportunity to investigate these characteristics in a large group of mothers who had agreed to participate in a survey about the health and development of their children. To our knowledge, this is the first report on maternal agreement to record linkage with birth record data.
Over 92% of the mothers who had taken part in the survey agreed to linkage with their baby's health records. This is consistent with rates reported by Dunn et al.2 and the same as was reported in a recent article that investigated consent to linking primary care records among older patients in London in a postal questionnaire.9 However, a lower consent rate (72%) was recently reported by Klassen et al.10 for linkage of the child's provincially collected health records in a postal survey of caregivers of young children in Vancouver.
The proportion consenting in our study was significantly lower in Northern Ireland and among minority ethnic groups. There was no evidence that the higher proportion of ethnic minority mothers not giving consent was due to language problems, although the likelihood of non-consent increased if the partner or another male had acted as interpreter. This suggests that consent may be related to cultural differences, possibly increased suspicion of health research11 particularly among women and/or their partners from ethnic minority groups. While the gender relations could be owing directly to cultural differences it could be that the presence of a male at the point of decision negatively influences a mother's decision even if she is British/White. However, this seems unlikely since Dunn et al.2 who analysed data from seven general population surveys conducted in the UK, found that, in the under 50 age group, males were only slightly less likely to consent to linkage than females, and Klassen et al.10 found no significance difference in consent between male and female caregivers. In addition, we found that lone parents (who were less likely to have a male partner present) were less likely to consent than those with a partner.
Maternal academic qualification was also significant, with mothers with either a degree or no academic qualifications being less likely to consent. This suggests that those with the most information about official records and those with the least were both more likely to be wary of allowing access to the information. Once qualifications are controlled for, there is a tendency for mothers in their 20s and 30s to be less likely to give consent than those giving birth at younger or older extremes. Most of the factors that related to non-consent to linkage, for example education and ethnicity, have also been identified as factors that affect response rates to social surveys.12 In the MCS, the overall survey response rates were lower amongst those living in an ethnic ward (66%) than in advantaged (76%) or disadvantaged wards (
72%) in England,3 indicating that fewer families from ethnic minority groups took part in the survey. Response was also lower in Northern Ireland (66%), which also had the lowest consent rate, than in other countries (72%+). There is no evidence as yet as to the possibility of response bias by education.
Since the effects of non-response and non-consent are multiplicative, the differences in the proportions with linked data may become wider when the data are linked. For example, if data from all those who consented are successfully linked, the percentages of the original sample with linked data will be 56, 68, and 72% for the ethnic, disadvantaged, and advantaged wards, respectively, more than doubling the original gap between the ethnic and other wards. Similarly the percentage with linked data from Northern Ireland will drop to 59% as compared with 68%+ for other UK countries.
It is important to take these sources of bias into account when analysing survey data, from the MCS or other similar surveys, linked with child health records, particularly in studies focussing on ethnic minority groups or children from Northern Ireland. A better solution, however, would be to increase response and consent rate among the unrepresented groups, as for example was done13 during the recruitment of women with infants for a controlled trial of two different strategies for providing them with support. Here special efforts were made to recruit a multi-ethnic multi-language sample, by, for example, increasing the number of preliminary visits to ascertain language needs and employing interpreters where needed. Although the efforts were costly, they succeeded in considerably increasing the proportion of women from socially disadvantaged and minority ethnic groups in their study. They also helped in understanding some of the reasons why the mothers did not consent. Further qualitative work on reasons for non-consent, including an investigation of what aspects of request for linkage to record data lead people from identified groups to say no, could be of great benefit in helping elucidate strategies for more inclusive recruitment for both survey participation and consent.
KEY MESSAGES
|
|
Acknowledgments |
|---|
We would like to thank all the Millennium Cohort families who provided the data for this study. We also thank members of the Millennium Cohort Study Management Team at the Centre for Longitudinal Studies, Institute of Education, University of London. The MCS is funded by the ESRC and a consortium of government departments led by ONS. The Record Linkage exercise is one of the enhancements to the Millennium Cohort Study funded by the Department of Health through a consortium led by the Office of National Statistics. Research at the Institute of Child Health and Great Ormond Street Hospital for Children NHS Trust benefits from R&D funding received from the NHS Executive. The other members of the Millennium Cohort Study Child Health Group who contributed to this work are: Suzanne Bartington, Helen Bedford, Neville Butler, Tim Cole, Lucy Griffiths, Summer Hawkins, Catherine Law, Catherine Peckham, Lamiya Samad, and Suzanne Walton.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
1 Jenkins SP, Cappellari L, Lynn P, Jackle A, Sala E. Patterns of consent: evidence from a general household survey. ISER Working Paper 2004-27, Colchester, University of Essex, December, 2004.
2 Dunn KM, Jordan K, Lacey RJ, Shapley M, Jinks C. Patterns of consent in epidemiologic research: Evidence from over 25,000 responders. Am J Epidemiol 2004;159:1087–94.
3 Plewis I. Millennium cohort study first survey: Technical report on sampling. 3rd edition. London: Institute of Education, University of London, 2004.
4 Rose D, Pevalin DA. A Researcher's Guide to the National Statistics Socio-economic Classification. London: Sage Publications, 2003.
5 ONS. Ethnic group statistics. a guide for the collection and classification of ethnicity data. Technical report. Available at: http://www.statistics.gov.uk/about/ethnic_group_statistics/ (Accessed 2003).
6 Tate AR, Dezateux C, Cole TJ, Davidson L, The Millennium Cohort Study Child Health Group. Factors affecting a mother's recall of her baby's birth weight. Int J Epidemiol 2005;34:688–95.
7 StataCorp. STATA Survey Data Reference Manual Release 8. Texas: Stata Corporation, 2003.
8 Woolf SH, Rothemich SF, Johnson RE, Marsland DW. Selection bias from requiring patients to give consent to examine data for health services research. Arch Family Med 2000;9:1111–18.
9 Harris T, Cook D, Victor C, Beighton C, Dewilde S, Carey I. Linking questionnaires to primary care records: factors affecting consent in older people. J Epidemiol Community Health 2005;59:336–38.
10 Klassen A, Lee S, Barer M, Raina P. Linking survey data with administrative health information: characteristics associated with consent from a neonatal intensive care unit follow-up study. Can J Public Health 2005;96:151–54.[Medline]
11 Corbie-Smith G, Thomas SB, St George DMM. Distrust, race, and research. Arch Intern Med 2002;162:2458–63.
12 Foster K. Evaluating non-response on household surveys. Report of a study linked to the 1991 census. Technical report. London: Governmental Statistical Service, 1998.
13 Oakley A, Wiggins M, Turner H, Rajan L, Barker M. Including culturally diverse samples in health research: a case study of an urban trial of social support. Ethn Health 2003;8:29–39.[CrossRef][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||