What is the Relative Risk?

The relative risk is defined as the probability of disease in the exposed group divided by the probability of disease in the unexposed group(1). In this dataset instead of disease the outcome is infant death.

How is Relative Risk Helpful?

The relative risk statistic is helpful because we are better able to determine if there is a significant increase or decrease in risk associated with the exposure and outcome variables.

What Variables Can be Used?

Most variables can be used as the exposure variable, whether it be a row variable, region, and or cohort indicator variable. If multiple cohort indicator variables are used they should be within the same grouping, and total must be included in the variable selection box.

How is Significance Determined?

If the number 1.0 is between the lower and higher confidence interval then the result is not statistically significant, otherwise the result is statistically significant.

How do I gain access to the data?

Simply email Yang Wang and request a username and password.

Relative Risk and 95% Confidence Interval Formulas (2)

	Infant Deaths	Surviving Live Births	All Live Births
Risk Group	a	b	a + b
Non-Risk Group	c	d	c + d
Total	a + c	b + d	n

Relative Risk (RR)=

[a / (a + b)] / [c / (c + d)]

For the null hypothesis, RR = 1

95% Confidence Intervals =

Standard Error (SE) of the RR = square root( [b / {a*(a+b) } ] + [d / { c*(c+d) } ] )

95% Confidence Interval = RR * e^{(+1.96 * [SE RR] )}

References:
1. Pagano M, Gauvreau K, Principles of Biostatistics. Duxbury Press, Belmont, CA. 1993
2. Szklo M, Nieto J, Epidemiology: Beyond the Basics. Aspen Publishers, Gaithersburg, MD. 2000

How to Calculate the Relative Risk Using the Cohort Module?

Step One

In order to generate relative risks for the data the user must first select a row variable, all options are valid. In our example we chose Race of Child.

Step Two

Second the user must use the column variable default of Survival Status. If this is chosen as a row variable the generated table will need to be rotated in order to calculate the relative risk.

Step Three

Third users can make any changes needed in the drop down menus, regions, and/or cohort indicator variables. If multiple regions or cohort indicators are selected, it must also be selected as the row or column variable. In the example below, all options were left at their default levels.

Step Four

Generate the Table and click on the Calculate Relative Risk button

If the calculate relative risk button does not appear, try rotating the table or check to see if all of the steps were followed correctly.

Step Five

Once the selections have been made and the table has been generated, users must select which variables are the Risk and/or Non-Risk Groups then click the Submit button. In our example we selected Race of Child: White as the non-risk group and the Race of Child: Black and Race of Child: Other as the risk groups.

If multiple cohort indicator variables have been selected, to generate relative risk users must select risk and non-risk variables within the same grouping such as Education, or Ethnicity. If not a warning will appear and prevent users from calculating the statistic. The cohort indicator variable Total must also be included to calculate the relative risk.

The table below will automatically change, populating the risk and non-risk boxes with the respective data according to the users selections. If a single cohort indicator is selected then the module will calculate the risk or non-risk group by taking the difference between the number of observations of the indicator selected and the total number of deaths, surviving live births, and total live births from the selections made. The user should note that this calculation includes unknown or missing observations from the exposed variable.

Note: Some groupings are not mutually exclusive, meaning that in a single grouping an infant may be in two or more categories, these non-mutually exclusive groupings include: congenital anomalies, prenatal care, medical risk factors, obstetrical procedures, complications of labor and delivery, method of delivery, abnormal conditions of the newborn.

Since many of the variables are check boxes and one infant can have multiple checks in a single grouping. For example, an infant could have Down's Syndrome and also have a Congenital Heart Malformation, both of which are under the congenital malformations grouping. For most of these variables the group name is the total number of infants with a condition listed in that grouping.

Variables dealing with the cause of death are not able to generate relative risk statistics since there is no surviving birth data for these infants.

Step Six

Interpretation of Statistic

Once these selections have been made, the user will receive the relative risk statistic with the lower and upper 95% confidence intervals along with an interpretation of the statistic. The interpretation uses only simple logic to determine significance and reliability of the statistic is determined by the user, since users determine what variables are used in the calculation. Significance is determined whether the confidence intervals contain the null value of 1, not by a P-Value calculation.

Relative risks and confidence intervals that are less than one imply that the risk category selected is less likely to have an infant death than the non-risk category. To have the statistic be more meaningful switch the risk and non-risk groups in the previous page, then regenerate the statistic.