• No se han encontrado resultados

PLAN DE IMPLANTACIÓN

In document SEGURIDAD Y SALUD OCUPACIONAL (página 65-68)

SISTEMA INTEGRADO.

6.1 PLAN DE IMPLANTACIÓN

r (0.6) is the indicator rate;

f (1.75) is the design effect factor;

c (1.12) is the non-response constant;

e (0.05) is the margin of error to be tolerated;

p (0.02) is the proportion of the target population group (one-year-old children);

nh (3.40) is the average household size.

Below there is an explanation of the data used to estimate each part of this equation

4 = Normal distribution factor

The distance of the variances on either sides of the mean, so that approximately 95% of the area under the normal distribution curve is enclosed.

r = Key indicator rate

The DPT coverage rate was selected as the key indicator, because DPT immunisation is the least prevalent, and it relates to the smallest target group (children aged up to 12 months). The DPT coverage rate was officially estimated by nationally reported data to the FBiH Public Health Institute to be over 90%. The empirical data, however, indicated that the actual coverage rate may have been much lower, as explained next. Sarajevo Canton immunisation results showed that 4,590 out of 6,662 (68.9%) children received DPT vaccine in 1999.a As Sarajevo Canton – compared to the rest of the country – was well served with public health facilities, the immunisation rate for the country was likely to be even lower. A figure of 60% was taken to be representative

a

165

of the country as a whole for DPT vaccination by age 12 months. It is possible that the real figure might have been even lower.

An alternative method would have been to select BCG coverage. The official FBiH summary data indicated that the coverage rate for Sarajevo Canton was 74.2% in 1999. From the birth rate, it can be calculated that the actual percentage of newborn babies who were immunised may have been as low as 40.4%. The BCG coverage figures, therefore, are less reliable than those for DPT.

As it was important for this survey to take a conservative but realistic estimate for the indicator rates, a figure for r of 60% was used.

f = Sample design effect factor

This is the variance increase factor required to compensate for sample design in comparison with a simple random sample.

c = Non-response constant

The non-response constant was reviewed at the pre-testing stage of this survey and in a UNICEF supported surveys on breastfeeding and iodine deficiency in BiH based on MICS methodology. Three percent of households surveyed in the FBiH were non- responders, and 14% in the RS. In the 1999 RS Breastfeeding survey, (sample size 2,800 households), the non-response was found to be 8% of households surveyed. As the MICS sample plan and questionnaires are much more complex, with countrywide coverage, a conservative non-response factor of 12 % was used.

e = Margin of error

A standard statistical margin of error of 5% (or 0.05) was used, as is also recommended by the global MICS methodology. It gives an acceptable confidence interval around estimates from the sample.

n = Average household size

The precise size of the average household in BiH was not known. Available estimates of average household size included:

166

3.37 in RS (RS 1999 UNICEF supported Breastfeeding survey); (148) 3.50 in the Federation (Federation MICS pre-testing).

The official demographic trends were quite reliable with respect to this factor, and indicated that the RS 1999 Breastfeeding survey results were likely to be accurate for RS household size. The average household had traditionally been slightly bigger in the geographical area of the FBiH. Therefore, the average household size for the country was estimated to be 3.40.

p = Proportion of the target population

There were no official data on the proportion of one-year-old children in the population. The available estimates included:

1.12% (derived from vital statistics indicators);

1.50% (unofficial estimate of the RS Statistics Institute); 1.70% (results of the RS Breastfeeding survey);

1.42% and 2.82% (BiH MICS 2000 survey pre-testing in the FBiH and RS respectively).

Therefore, a conservative figure of 2% (i.e. p =0.02) was used.

3.21.6 Sampling frame

In developing the sampling frame, perhaps the most challenging technical aspect of the survey, several options were explored. The previous census of the Republic of Bosnia and Herzegovina in 1991a was available and had detailed household listings, but they were no longer useful due to population shifts. Only the census enumeration areas and maps had some value. A population census was carried out in RS in 1996, however there was no equivalent in FBiH and basic demographic data for the FBiH was unavailable. A household database, developed by the Organisation for Security and Cooperation in Europe (OSCE) during the national post-war elections in BiH, was

a As explained in the introduction chapter, this was the pre-war political structure with the Socialist

Federal Republic of Yugoslavia and the geographical boundaries matched the current boundaries of BiH.

167

examined but found to be substantially incomplete. It was decided to create a new sampling frame using the best available pooled information from a number of sources. The sources were:

• Results of the 1996 population census in the RS

• Population estimates for 1999 at the municipality level, provided by the statistical institutes of both entities

• Census enumeration areas and maps from the 1991 population census • Lists of voters, provided by the OSCE

• Results of previous UNICEF supported surveys (1999 breastfeeding surveys in both entities, MICS pre-tests)

• Results of the 1991 BiH population census

• List of inhabited settlements produced by the Republic of BiH Statistical Institute 1998

Stage 1: The geographical area of BiH (with the exception of Brcko district) was

In document SEGURIDAD Y SALUD OCUPACIONAL (página 65-68)