Source: Health Research and Information Division, ESRI
Creator: Information Unit, Department of Health

Information on every birth in the Republic of Ireland is submitted to the NPRS by hospital administrative staff and all practicing independent midwives. The information collected includes data on pregnancy outcomes (with particular reference to perinatal mortality and important aspects of Perinatal care), as well as descriptive social and biological characteristics of mother’s giving birth. The time frame to which the information relates is from 22 weeks gestation to the first week of life. In accordance with WHO guidelines, live births weighing less than 500 grams are not included in the national statistics presented. ICD-10 came into effect for births from 1 January 2005 onwards. The CSO commenced using ICD-10 from 1 January 2007. The data cover the years from 1999 to 2011. 'Not Stated' or 'Not Known' values are not included in the calculation of percentages.

Interpret these data cautiously. As a general rule, this website does not include confidence limits in its charts and maps. It aims to provide visual tools that allow you to explore an underling table or dataset. If you find something that you think is important, we strongly urge you to explore it more rigorously – consulting an experienced data analyst if appropriate – before taking any action based on that finding.

1. Statistical precision
Indicator values are prone to statistical error (the difference between an estimated value and the true value). The statistical error associated with an indicator depends on the population subgroup (e.g. the population of a county or LGD) that it refers to. Such differences in levels of statistical error can distort what we see in maps and charts. They can make some relationships involving indicators and attributes appear “real” (practically meaningful or statistically significant) when they are in fact spurious; other relationships that are “real” can be masked. These differences in statistical error can even distort the shape of plots or the colour patterns we see in maps.

For example,

• Many indicator values estimates are derived from sample surveys, and different sample sizes from different population subgroups will lead to different levels of precision in the indicator values for these subgroups.
• Different population subgroups have different population sizes which means that rate estimates for these subgroups will also have different confidence limits.
• The true value of a percentage or a rate can influence the level of statistical error of any estimate.

2. Scales and legends
The scales used on chart axes and in can also distort our perceptions:

• The range of values allowed on chart axes can accentuate relationships making them appear more “real” than they actually are.
• The radial arms of spider plots of scaled data show the position of the value (in a population subgroup) relative to the minimum and maximum values of that indicator. Because these minimum and maximum depend on the indicator, relative positions of different population subgroups on different radial arms are not directly comparable.
• The cut-off values used to determine the colours to shade areas of a map are default selections and do not necessarily represent meaningful values of the indicator. Areas with very similar data values can be shown with different shades. You should always note the actual values.