ANALISIS DE CENTROS CULTURALES PARA INVIDENTES EN PUEBLA

The proportion o f women from the antenatal population who are eligible for screening depends on a universal or selective approach. It is one o f the functions o f

the model to predetermine the assumed true proportions o f eligible women for any

antenatal population broken down into ethnic groups. This is achieved by taking into consideration information about maternal ethnic group and Hb-pathy carrier state as well as information about the prevalence o f iron deficiency and clinically non

significant trait/homozygous state (see below) both o f which can lead to a low

MCH. The parameter must be distinguished from the later one ‘failure to screen

eligible w om an‘ which reflects screening practice and takes account o f the fact that in

Prevalence o f iron deficiency in pregnant women

The parameter o f relevance for the model is prevalence o f iron deficiency which causes a low MCH <27 pg and occurs at the time o f booking, by maternal ethnic group. It is a parameter likely to be influenced strongly by local demographic factors and appropriate figures could be expected to stem from antenatal Hb-pathy screening programmes. However, there are no such published data available for UK antenatal populations. Only one survey from a deprived inner London area with a high proportion o f Asian ethnic groups was found to be informative: amongst ethnic minority groups, 38% o f pregnant women were found to be iron deficient on grounds o f a low serum ferritin concentration (10 pg/1) with no differences between ethnic groups. In 24% o f pregnant women MCH was <27 pg which, after exclusion o f a and P thalassaemia traits, left a minimum o f 15% with an MCH <27 pg, most likely due to iron deficiency^^^.

Published data could thus only be used for a crude estimate o f the likely magnitude o f the effect. As baseline values we have chosen a probability o f 0.1 for all ethnic groups as a conservative estimate to demonstrate the possible impact o f iron deficiency on screening outcomes. This was varied in a sensitivty analysis to 0.05 and 0.3.

Apart from Hb-pathy carrier state, iron deficiency is only one parameter contributing to the prevalence o f a low MCH amongst pregnant women, the other being trait/homozygous state. Uncertainty about the prevalence o f iron deficiency has been

explored together with the prevalence for trait (see below) by varying a newly

created summary parameter called Tow MCH due to iron deficiency or

trait/homozygous state’ which combines both variables. T able 4.7 shows baseline values and ranges o f the summary parameter.

T able 4.7 Prevalence of ‘low M CH due to iron deficiency o r ‘a trait/hom ozygous state’ by ethnic group

+thal

Ethnic group Prevalence o f trait (%)

Prevalence o f low MCH due to trait/hom ozygous state^ (%)

Prevalence o f low M CH due to iron deficien cy or trait/hom ozygous state (%) at different lev els o f iron d eficien cy

L evel o f iron d eficien cy (%) B aseline value B aseline value B aseline value L ow H igh

10% 5% 30% B lack Caribbean 30 24 32 28 47 B lack African 30 24 32 28 47 B lack Other 30 24 32 28 47 Indian 50 59 64 62 72 Pakistani 50 59 64 62 72 Bangladeshi 25 19 27 23 44 Chinese 5 4 13 8 32 Other A sian 3 2 12 7 31 Other 3 2 12 7 31 Cypriot 25 19 27 23 44 Italian 3 2 12 7 31 North European 3 2 12 7 31

Figures are percentages

^ Calculation o f prevalence o f low MCH due to trait/hom ozygous state has been explained in box 3.1, page 57.

Prevalence o f tra it by ethnic group

Figures for the prevalence o f trait by ethnic group were based on estimates by

Petrou & ModelP^^ and WHO^'^. trait and the homozygous state are clinically

not significant and cannot be definitely diagnosed but only suspected by routine haematological tests showing an otherwise unexplained decrease in MCH and, in cord blood samples, a band suggestive of Hb Barts^^’^^^. Definite diagnosis requires genotyping and there are a variety o f deletional and non-deletional mutations described^. There are no published epidemiological data available from UK studies or screening programmes. Estimates for ethnic minority groups residing in the U K were therefore mainly based on studies in their countries o f origin using random cord blood samples or more recently, with available DNA technology, population samples. Figures for the North European population were informed by case series reporting

families o f British descent with trait^^\ Estimates for Asian and Cypriot ethnic

minority groups and North European have been checked against the observed

frequency o f otherwise unexplained low MCH measurements in haematological s a m p l e s ^ O v e r a l l however, there remains considerable uncertainty about the accuracy o f the estimates which together with estimates about the prevalence o f iron deficiency (see above) have been varied in a sensitivity analysis (table 4.7).

4.3.1.4 Wo ma n accepts screening

There was no specific information available regarding the percentage o f women accepting antenatal screening after informed consent, but it is estimated to approach 100% (personal communication EN Anionwu). Nationally and internationally there is broad consensus about the voluntary character o f antenatal screening and the need for consent^’^^’^^’^"^. Historically however testing for Hb-pathies and for sickle carrier status was seen as part o f routine antenatal care for the optimal management o f pregnant women^^. Uptake might thus vary according to the way in which consent is sought, ranging from assumed acceptance o f comprehensive antenatal care including diagnostic and screening procedures (implied consent) to screening specific consent, either in the form of generic consent for genetic screening for a variety o f conditions^^^ or as explicit disease specific opt-out and opt-in policies^^.

A baseline value o f 100% acceptance was assumed and not varied in a sensitivity analysis.

4.3.1.5 F a ilu re to screen eligible women

There is concern about an inherent risk o f failure to screen eligible women in a selective programme, either due to failure to offer screening or due to failure to carry

out the necessary tests'^’ However, there are no reliable quantitative published UK

studies directly addressing these issues. An unpublished audit from U niversity College London Hospital over a period o f three months found that 3.4% (9/268) o f non-North European mothers from the antenatal clinic had no laboratory test results available^^^. Adjaye et al'^ reported that over a three year period six out o f 10 mothers o f babies diagnosed with sickle cell disease postnatally had not been identified in a selective antenatal programme; tliree were due to non-attendance, one (10%) due to failure to offer screening despite eligibility. Indirect evidence o f the potential problem is given by an unpublished survey o f antenatal laboratory screening practice in the UK in which only four out o f 23 (17.4%) laboratories with selective screening strategies reported that they had information about ethnic origin o f the mother for more than 80% o f patients’^ As directly applicable data were lacking it was necessary to use best estimates and explore the influence o f this parameter value in a sensitivity analysis. The range o f arbitrary values employed were 0.5%, 1.5%, 3%, 5.5%.

As baseline values we assumed a 0.5% failure rate to screen eligible wom en in a universal programme in contrast to 5.5% in a selective programme, representing a 5% difference. The failure rate in a selective programme is only applicable to women selected solely by ethnic group but not those selected on grounds o f a low MCH. The latter were subjected to the same 0.5% failure to screen rate as women in a universal programme, because MCH measurement is a universal test performed on all antenatal women.

We inferred that without definite evidence to the contrary districts could not assume a selective failure to screen rate better than 5.5%. This is not to say that some units may be failing to achieve this but it was the highest rate that we felt could be countenanced.