• No results found

Estimating the available populations and identifying the sampling procedure

CHAPTER SIX: METHODOLOGY

6.3 Estimating the available populations and identifying the sampling procedure

This study was initiated prior to the publication o f the 1991 OPCS Census (1993). The ethnic composition and number o f first births per ethnic group was therefore estimated. Discussions with local health professional identified the largest West African communities to be of Nigerian and Ghanaian origin. The Nigerian, Ghanaian and Bangladeshi populations living in East London were thought to be recent migrants so satisfying part of the inclusion criteria, but there was no data available to indicate the numbers of mothers from these groups delivering their first babies.

In order to estimate the number of births per year by ethnicity, a review o f the available Birth Notification records for Hackney held at Department of Community Child Health, St. Leonard's Hospital was undertaken by the author. These records contain the name and address of the mother, matemal age, parity of the infant and birth weight, and it was hoped that they would provide a viable method for eventually recmiting the sample groups. The records did not include information regarding the duration o f matemal residence.

6.3.1 Assessing the Numbers of Births over 2500grams per Ethnic Group North East Thames Health Authority, of which Hackney is a part, defined ethnicity using the global categories o f Asian, African or Caucasian and had only started to record ethnicity in Febmary 1990. Although these categories were inadequate for determining specific numbers bom to the Bangladeshi, Nigerian and Ghanaian

groups, they provided the best possible estimate of the numbers o f births to broad ethnic groupings in the area. The estimates are based on births from 1 February to 31 December 1990 (11 months) when approximately 3400 births were registered.

6.3.2 Estimating the Asian Population

During the 11-month period, 310 births were classified as o f Asian parentage. O f these, 89 were first live births (22%), and of these, 14 (16%) were ineligible as their birth weight was less than 2500 grams. The country o f origin o f the remaining 75 first-live-birth infants was not known but it was estimated by local health workers that approximately half were from Bangladesh. This suggested that the total number of births that would have been eligible for inclusion during this 11-month period was 37. Health workers warned that changing policies regarding re-housing of

Bangladeshi families meant many were now being relocated to neighbouring Tower Hamlets.

6.3.3 Estimating the African Population

The total number of births where the ethnicity of the parent was classified as Afiican was 332. O f those, 191(62%) were identified as first live births; 20 (10%) were recorded with a birth weight being under 2500 grams. It was not clear what proportion o f the remaining 171 were Nigerian.

6.3.4 Verifying ethnicity data coded unknown or missing

In addition, 617 births (34% of the total number recorded) were classified as 'ethnic origin unknown' or 'missing'. The use o f first and second names has been found to be a reliable method for identifying Muslim Asians in this country (Nicholl, Bassett

and Ulijasjek,1986); but its reliability at differentiating between Bangladeshi and Pakistani names is unclear. Some Nigerians names are easily recognisable, but the accuracy o f using this method has not been verified with the Nigerian population in this country. Such a large proportion of unknown and missing codes made estimates o f the eligible population difficult. Discussions with Hackney Health Authority identified a possible solution that would still allow the birth records to be used as the sampling fi-ame. Local health visitors would be asked to verify individually the ethnicity o f the mothers whose infants birth records indicated were in her/his case load. This would also provide an opportunity for verifying whether mothers were Pakistani or Bangladeshi, Nigerian or Ghanaian.

All the Health Visitors in Hackney were informed by letter about the nature of the study (see Appendix A), and that they might be contacted again in the near future to help with specific enquiries. Each clinic was visited and a personal contact

established. Although the response was generally co-operative, it became apparent during piloting, that obtaining information on the ethnicity of the 1990 births would not be sufficiently comprehensive to enhance the estimates already calculated using the broad ethnic categories.

The review process had indicated some potential differences between the African and Asian groups. There were fewer first births to Asian mothers; Asian birth

weights were lower than the norm; and the mothers were younger at the time o f their first birth.

allow pair-wise matching on birth weight across the three groups. Matching for birth weight across the groups would have potentially biased the sample by excluding the smaller Bangladeshi and larger Nigerians infants because they would be impossible to pair. Instead concurrent eligible births were chosen.

It also became apparent that if the records were to be used as the basis of the

sampling frame, it would not be possible to secure the information on ethnicity from the health visitors before the infant was 6 weeks old, the intended time of the first visit. Thus, the only method that would allow access to the whole population would be through direct contact with the antenatal clinics.

6.3.5 Sampling procedure

Applications were made to Hackney and subsequently. Tower Hamlets Ethical Committees for permission to access the antenatal records and attend the antenatal clinics at the Homerton Hospital, Hackney and Royal London Hospital, Tower Hamlets. Although it had initially been hoped that all participants could be recruited from the Homerton Hospital, there were insufficient numbers of Bangladeshi

mothers that met the inclusion criteria. Additional mothers from all three ethnic groups were recruited from the antenatal clinics at the Royal London Hospital. An identical procedure for identifying mothers was used at both sites.

6.3.6 Recruitment procedure

Mothers that met the inclusion criteria were identified from the antenatal records held at the clinics at the Homerton and Royal London Hospitals. Details o f the mother's ethnicity, age, previous birth history and occasionally details of when she

had arrived in the UK were included. The notes were reviewed during the daily clinic.

Once identified from the hospital records, the mother was approached while waiting in the clinic for her antenatal appointment. The author introduced herself, explained the purpose o f the study and gave the mother an explanatory leaflet (Appendix B). It was explained that participation in the study would involve three visits to her home; that during each visit an anthropometric assessment o f the infant would be

undertaken; and that she would be interviewed about her infant's feeding and

behaviour. An Infant Diary (Appendix C) was shown to the mother as an illustration o f the kind of information the study was interested in collecting. It was explained that if she agreed to be included at this stage, she might be contacted postnatally and asked again if she wished to participate. Mothers were told that inclusion in the study would be random according to the gender of her baby. In fact, infants were not selected on the basis of gender, but this was thought to be a less stigmatising reason for exclusion than explaining the infant eligibility criteria.

Details obtained from the antenatal records were verified, including ethnicity and her length o f residence in the UK. Where necessary, for example when a move was anticipated, a second contact address or telephone number was obtained. The Hospital translators or an accompanying family member provided translation for non-English speaking mothers. Verbal consent to be contacted after the delivery was obtained from those willing to be included in the study.