Table of Contents

APPENDIX D


Technical Notes and Formulas

NATALITY REPORT

Significance Tests for Proportions

State Versus Subgroup Comparisons

The following test was used to determine whether or not a subgroup (county, city or race) percent (p1) is different from the state percent (P). The Null and Alternate hypotheses tested are:

          Ho : p1 equal P vs. H1 : p1 not equal P

The formula for the Z-score is:   Z = (p1 - P)/SQRT[(P * Q)/n1]

          where:   p1 is the subgroup percent,
                       P is the state percent,
                       Q is 1-P,
                       n1 is the number of births in the subgroup (denominator of p1), and
                       SQRT denotes the square root of the expression in brackets.

In a single comparison, if the absolute value of the resulting Z-score is greater than or equal to 1.96, the null hypothesis is rejected and it can be concluded that the subgroup percent is not equal to the state percent, but rather is larger (or smaller) than the state percent. This critical value (1.96) represents a two-tailed test at alpha equal to .05 (.025 in each tail of the normal distribution).

When a large number of comparisons are made, as is the case when comparing percents for all the counties or all the cities against the state percent, the probability that a difference is significant increases substantially due to chance alone. Bonferonni's theory of inequalities is used to account for the increased chance of a significant result when making multiple comparisons.7 The significance level for each individual comparison is adjusted by the number of comparisons made. For the county comparisons, the alpha level was set to .05/92, or .0005. This corresponds to a critical Z-score of 3.48 for a two-tailed test. For the city comparisons, the alpha level was set to .05/26, or .002. This corresponds to a critical Z-score of 3.10 for a two-tailed test.

Subgroup Versus Subgroup Comparisons

The following test was used to compare two subgroup percents. It is used to compare one county versus another county or one city versus another city. It is also used, for example, to determine if the percent in two racial subgroups of the state are significantly different from each other. The Null and Alternate hypotheses are:

          Ho : p1 = p2 vs. H1 : p1 not equal p2

The formula for the Z-score is:   Z = (p1 - p2) / SQRT[(p*q/n1) + (p*q/n2)]

          where:    p1 is the percent for one group,
                        p2 is the percent for the second group,
                        p is the pooled percent,
                        q is 1-p,
                        n1 is the number of births in group 1,
                        n2 is the number of births in group 2,
                        SQRT denotes the square root of the expression in brackets,

and the pooled percent is:   p = (f1 + f2 )/( n1 +n2)

          where:   f1 = the number with the characteristic in group 1,
                       f2 = the number with the characteristic in group 2.

For single comparisons, if the resulting Z-score is greater than or equal to 1.96, the null hypothesis is rejected and it can be concluded that the two subgroup percents are not equal. This critical value (1.96) represents a two tailed test at alpha equal to .05 (.025 in each tail of the normal distribution).

When making multiple comparisons, the critical Z values must be adjusted according to the number of comparisons being made. Multiple subgroup comparisons were not made within the context of this report.

Significance Tests for Rates

State Versus Subgroup Comparisons

A comparison between rates was considered statistically significant if the difference between the rates would have occurred by chance less than five times out of 100 (i.e., p < 0.05). All statistical tests were conducted using the Z-score method, described in basic statistics texts. While the form of the Z test is similar, the calculation of the standard error for a rate is not straightforward. The following formula can be used to approximate the standard error of a rate:8

          SE(rate) = rate / [events]1/2

          where, in this case, "events" is the number of births used to calculate the rate.

Subsequently, the Z test for the difference between the state and county rates takes the form:

          Z = (State rate - County rate) / SE(diff)

where the SE(diff) is defined as:

          SE(diff) = [(SE(s))2 + (SE(c))2 ]1/2

where SE(s) is the standard error of the state rate and SE(c) is the standard error of the county rate.

When testing the differences between the Indiana rate and the rate for each of the 92 counties, the large number of possible comparisons increases the chance that a rate would be statistically significant due to chance alone. To account for multiple comparisons, based on Bonferroni's theory of inequalities, the significance level for each individual comparison was set equal to 0.05/92, where 92 is the number of comparisons made.7 Thus, any individual comparison with an associated "p" value less than 0.0005 (Z=3.48) was considered to be statistically significant.

When testing the difference between the state rate and the 26 city rates, the significance level was set to 0.05/26. The associated "p" value less than .002 is considered significant. This corresponds to a critical Z score of 3.10 for a two-tailed test.

Subgroup Versus Subgroup Comparisons

The above test can be used to compare two subgroup rates. It is used to compare one county to another county or one city to another city. It is also used, for example, to determine if the rate in two racial subgroups of the state are significantly different from each other.

Significance Tests for Risk Ratios

The ratio of the incidence rate in one group to that in another is referred to as a rate ratio. Likewise, the ratio of the cumulative incidence or risk in two groups is termed the risk ratio or relative risk. One advantage of such ratio measures is the ease of intuitive understanding (i.e., disease occurrence is increased nearly 15-fold among those who smoke). Also, this ratio is independent of the absolute incidence rates in the two groups and therefore is directly interpretable.9

The risk ratios in this report were calculated using the Statcalc function in the Centers for Disease Control's Epi Info, version 6.02.10 A two-by-two table is created which contains the number of individuals affected by the object of choice (e.g., low birth weight) out of the entire population compared to the number of individuals affected by the object in the comparison group out of the entire population of the comparison group. The top left quadrant would contain the number of individuals affected in the study population, the top right quadrant would contain the number of individuals affected in the control or comparison population. The bottom left quadrant would contain the number of individuals not affected in the total population (i.e., total population - affected individuals). The bottom right quadrant would contain the number of individuals in the comparison population not affected by the condition. Subsequently, the Chi square and relative risk are calculated.

Method of Computing Cesarean Delivery Rates

Overall Cesarean Rate

To calculate this rate, divide the total number of births by cesarean delivery by the total number of births minus the not stated method of delivery, and multiply by 100 to express the rate as a percent.

Overall cesarean rate =                          Number of births by cesarean                        x 100
                             Total number of births - "not stated" method of delivery

Primary Cesarean Rate

To calculate this rate, divide the number of first cesarean births by the total number of births to women who have not had a previous cesarean delivery, and multiply by 100 to express as a percent.

Primary cesarean rate =                                Number of primary cesarean births                            x 100
                             Total # births - VBACs - repeated cesareans - "not stated" method

VBAC (Vaginal Birth After Cesarean) Rate

To calculate this rate, divide the number of vaginal births to women who had a previous birth by cesarean delivery by the total number of births (vaginal and cesarean) to women who had a previous birth by cesarean delivery, and multiply by 100 to express as a percent.

VBAC rate =                       VBACS                       x 100
                VBACS + repeated cesareans


INDUCED TERMINATION OF PREGNANCY REPORT

Estimates for Unknown Gestation

Table of Contents