The RISKDIFF option in the TABLES statement provides estimates of risks (binomial proportions) and risk differences for tables. This analysis might be appropriate when comparing the proportion of some characteristic for two groups, where row 1 and row 2 correspond to the two groups, and the columns correspond to two possible characteristics or outcomes. For example, the row variable might be a treatment or dose, and the column variable might be the response. For more information, see Collett (1991); Fleiss, Levin, and Paik (2003); Stokes, Davis, and Koch (2012).
Let the frequencies of the table be represented as follows.
Column 1 
Column 2 
Total 

Row 1 



Row 2 



Total 


n 
By default when you specify the RISKDIFF option, PROC FREQ provides estimates of the row 1 risk (proportion), the row 2 risk, the overall risk, and the risk difference for column 1 and for column 2 of the table. The risk difference is defined as the row 1 risk minus the row 2 risk. The risks are binomial proportions of their rows (row 1, row 2, or overall), and the computation of their standard errors and Wald confidence limits follow the binomial proportion computations, which are described in the section Binomial Proportion.
The column 1 risk for row 1 is the proportion of row 1 observations classified in column 1,
which estimates the conditional probability of the column 1 response, given the first level of the row variable. The column 1 risk for row 2 is the proportion of row 2 observations classified in column 1,
The overall column 1 risk is the proportion of all observations classified in column 1,
The column 1 risk difference compares the risks for the two rows, and it is computed as the column 1 risk for row 1 minus the column 1 risk for row 2,
The standard error of the column 1 risk for row i is computed as
The standard error of the overall column 1 risk is computed as
Where the two rows represent independent binomial samples, the standard error of the column 1 risk difference is computed as
The computations are similar for the column 2 risks and risk difference.
By default, the RISKDIFF option provides Wald asymptotic confidence limits for the risks (row 1, row 2, and overall) and the risk difference. By default, the RISKDIFF option also provides exact (ClopperPearson) confidence limits for the risks. You can suppress the display of this information by specifying the NORISKS riskdiffoption. You can specify riskdiffoptions to request tests and other types of confidence limits for the risk difference. For more information, see the sections Confidence Limits for the Risk Difference and Risk Difference Tests.
The risks are equivalent to the binomial proportions of their corresponding rows. This section describes the Wald confidence limits that are provided by default when you specify the RISKDIFF option. The BINOMIAL option provides additional confidence limit types and tests for risks (binomial proportions). For more information, see the sections Binomial Confidence Limits and Binomial Tests.
The Wald confidence limits are based on the normal approximation to the binomial distribution. PROC FREQ computes the Wald confidence limits for the risks and risk differences as
where Est is the estimate, is the th percentile of the standard normal distribution, and is the standard error of the estimate. The confidence level is determined by the value of the ALPHA= option; by default, ALPHA=0.05, which produces 95% confidence limits.
If you specify the CORRECT riskdiffoption, PROC FREQ includes continuity corrections in the Wald confidence limits for the risks and risk differences. The purpose of a continuity correction is to adjust for the difference between the normal approximation and the binomial distribution, which is discrete. See Fleiss, Levin, and Paik (2003) for more information. The continuitycorrected Wald confidence limits are computed as
where cc is the continuity correction. For the row 1 risk, ; for the row 2 risk, ; for the overall risk, ; and for the risk difference, . The column 1 and column 2 risks use the same continuity corrections.
By default when you specify the RISKDIFF option, PROC FREQ also provides exact (ClopperPearson) confidence limits for the column 1, column 2, and overall risks. These confidence limits are constructed by inverting the equaltailed test that is based on the binomial distribution. For more information, see the section Exact (ClopperPearson) Confidence Limits.
PROC FREQ provides the following confidence limit types for the risk difference: AgrestiCaffo, exact unconditional, HauckAnderson, MiettinenNurminen (score), Newcombe (hybridscore), and Wald confidence limits. Continuitycorrected forms of Newcombe and Wald confidence limits are also available.
The confidence coefficient for the confidence limits produced by the CL= riskdiffoption is %, where the value of is determined by the ALPHA= option. By default, ALPHA=0.05, which produces 95% confidence limits. This differs from the testbased confidence limits that are provided with the equivalence, noninferiority, and superiority tests, which have a confidence coefficient of % (Schuirmann 1999). For more information, see the section Risk Difference Tests.
AgrestiCaffo Confidence Limits
AgrestiCaffo confidence limits for the risk difference are computed as
where , ,
and is the th percentile of the standard normal distribution.
The AgrestiCaffo interval adjusts the Wald interval for the risk difference by adding a pseudoobservation of each type (success and failure) to each sample. See Agresti and Caffo (2000) and Agresti and Coull (1998) for more information.
HauckAnderson Confidence Limits
HauckAnderson confidence limits for the risk difference are computed as
where and is the th percentile of the standard normal distribution. The standard error is computed from the sample proportions as
The HauckAnderson continuity correction cc is computed as
See Hauck and Anderson (1986) for more information. The subsection "HauckAnderson Test" in the section Noninferiority Tests describes the corresponding noninferiority test.
MiettinenNurminen (Score) Confidence Limits
MiettinenNurminen (score) confidence limits for the risk difference (Miettinen and Nurminen 1985) are computed by inverting score tests for the risk difference. A scorebased test statistic for the null hypothesis that
the risk difference equals can be expressed as
where is the observed value of the risk difference (),
and and are the maximum likelihood estimates of the row 1 and row 2 risks (proportions) under the restriction that the risk difference is . For more information, see Miettinen and Nurminen (1985, pp. 215–216) and Miettinen (1985, chapter 12).
The % confidence interval for the risk difference consists of all values of for which the score test statistic falls in the acceptance region,
where is the th percentile of the standard normal distribution. PROC FREQ finds the confidence limits by iterative computation, which stops when the iteration increment falls below the convergence criterion or when the maximum number of iterations is reached, whichever occurs first. By default, the convergence criterion is 0.00000001 and the maximum number of iterations is 100.
By default, the MiettinenNurminen confidence limits include the bias correction factor in the computation of (Miettinen and Nurminen 1985, p. 216). For more information, see Newcombe and Nurminen (2011). If you specify the CL=MN(CORRECT=NO) riskdiffoption, PROC FREQ does not include the bias correction factor in this computation (Mee 1984). See also Agresti (2002, p. 77). The uncorrected confidence limits are labeled as "MiettinenNurminenMee" confidence limits in the displayed output.
The maximum likelihood estimates of and , subject to the constraint that the risk difference is , are computed as
where
For more information, see Farrington and Manning (1990, p. 1453).
Newcombe Confidence Limits
Newcombe (hybridscore) confidence limits for the risk difference are constructed from the Wilson score confidence limits
for each of the two individual proportions. The confidence limits for the individual proportions are used in the standard
error terms of the Wald confidence limits for the proportion difference. See Newcombe (1998a) and Barker et al. (2001) for more information.
Wilson score confidence limits for and are the roots of
for . The confidence limits are computed as
For more information, see the section Wilson (Score) Confidence Limits.
Denote the lower and upper Wilson score confidence limits for as and , and denote the lower and upper confidence limits for as and . The Newcombe confidence limits for the proportion difference () are computed as
If you specify the CORRECT riskdiffoption, PROC FREQ provides continuitycorrected Newcombe confidence limits. By including a continuity correction of , the Wilson score confidence limits for the individual proportions are computed as the roots of
The continuitycorrected confidence limits for the individual proportions are then used to compute the proportion difference confidence limits and .
Wald Confidence Limits
Wald confidence limits for the risk difference are computed as
where , is the th percentile of the standard normal distribution. and the standard error is computed from the sample proportions as
If you specify the CORRECT riskdiffoption, the Wald confidence limits include a continuity correction cc,
where .
The subsection "Wald Test" in the section Noninferiority Tests describes the corresponding noninferiority test.
Exact Unconditional Confidence Limits
If you specify the RISKDIFF option in the EXACT statement, PROC FREQ provides exact unconditional confidence limits for the
risk difference. PROC FREQ computes the confidence limits by inverting two separate onesided tests (tail method), where the
size of each test is at most and the confidence coefficient is at least ). The conditional exact method, which is described in the section Exact Statistics, does not apply to the risk difference because of a nuisance parameter (Agresti 1992). The unconditional method (which fixes only the row margins) eliminates the nuisance parameter by maximizing the pvalue over all possible values of the parameter (Santner and Snell 1980).
By default, PROC FREQ uses the unstandardized risk difference as the test statistic to compute the confidence limits. If you specify the RISKDIFF(METHOD=SCORE) option, the procedure uses the score statistic to compute the confidence limits (Chan and Zhang 1999). The score statistic is a less discrete statistic than the unstandardized risk difference and produces less conservative confidence limits (Agresti and Min 2001). For more information, see Santner et al. (2007). The section Confidence Limits for the Risk Difference describes the computation of the risk difference score statistic. For more information, see Miettinen and Nurminen (1985) and Farrington and Manning (1990).
PROC FREQ computes the exact unconditional confidence limits as follows. The risk difference is defined as the difference between the row 1 and row 2 risks (proportions), , and and denote the row totals of the table. The joint probability function for the table can be expressed in terms of the table cell frequencies, the risk difference, and the nuisance parameter as
The % confidence limits for the risk difference are computed as
where
The set A includes all tables with row sums equal to and , and denotes the value of the test statistic for table a in A. To compute , the sum includes probabilities of those tables for which (), where is the value of the test statistic for the observed table. For a fixed value of , is taken to be the maximum sum over all possible values of .
PROC FREQ provides tests of equality, noninferiority, superiority, and equivalence for the risk (proportion) difference. The following analysis methods are available: Wald (with and without continuity correction), HauckAnderson, FarringtonManning (score), and Newcombe (with and without continuity correction). You can specify the method by using the METHOD= riskdiffoption; by default, PROC FREQ provides Wald tests.
The equality test for the risk difference tests the null hypothesis that the risk difference equals the null value. You can specify a null value by using the EQUAL(NULL=) riskdiffoption; by default, the null value is 0. This test can be expressed as versus the alternative , where denotes the risk difference (for column 1 or column 2) and denotes the null value.
The test statistic is computed as
where the standard error is computed by using the method that you specify. Available methods for the equality test include Wald (with and without continuity correction), HauckAnderson, and FarringtonManning (score). For a description of the standard error computation, see the subsections "Wald Test," "HauckAnderson Test," and "FarringtonManning (Score) Test," respectively, in the section Noninferiority Tests.
PROC FREQ computes onesided and twosided pvalues for equality tests. When the test statistic z is greater than 0, PROC FREQ displays the rightsided pvalue, which is the probability of a larger value occurring under the null hypothesis. The onesided pvalue can be expressed as
where Z has a standard normal distribution. The twosided pvalue is computed as .
If you specify the NONINF riskdiffoption, PROC FREQ provides a noninferiority test for the risk difference, or the difference between two proportions. The null hypothesis for the noninferiority test is
versus the alternative
where is the noninferiority margin. Rejection of the null hypothesis indicates that the row 1 risk is not inferior to the row 2 risk. See Chow, Shao, and Wang (2003) for more information.
You can specify the value of with the MARGIN= riskdiffoption. By default, . You can specify the test method with the METHOD= riskdiffoption. The following methods are available for the risk difference noninferiority analysis: Wald (with and without continuity correction), HauckAnderson, FarringtonManning (score), and Newcombe (with and without continuity correction). The Wald, HauckAnderson, and FarringtonManning methods provide tests and corresponding testbased confidence limits; the Newcombe method provides only confidence limits. If you do not specify METHOD=, PROC FREQ uses the Wald test by default.
The confidence coefficient for the testbased confidence limits is % (Schuirmann 1999). By default, if you do not specify the ALPHA= option, these are 90% confidence limits. You can compare the confidence limits to the noninferiority limit, –.
The following sections describe the noninferiority analysis methods for the risk difference.
Wald Test
If you specify the METHOD=WALD riskdiffoption, PROC FREQ provides an asymptotic Wald test of noninferiority for the risk difference. This is also the default method. The
Wald test statistic is computed as
where () estimates the risk difference and is the noninferiority margin.
By default, the standard error for the Wald test is computed from the sample proportions as
If you specify the VAR=NULL riskdiffoption, the standard error is based on the null hypothesis that the risk difference equals – (Dunnett and Gent 1977). The standard error is computed as
where
If you specify the CORRECT riskdiffoption, the test statistic includes a continuity correction. The continuity correction is subtracted from the numerator of the test statistic if the numerator is greater than 0; otherwise, the continuity correction is added to the numerator. The value of the continuity correction is .
The pvalue for the Wald noninferiority test is , where Z has a standard normal distribution.
HauckAnderson Test
If you specify the METHOD=HA riskdiffoption, PROC FREQ provides the HauckAnderson test for noninferiority. The HauckAnderson test statistic is computed as
where and the standard error is computed from the sample proportions as
The HauckAnderson continuity correction cc is computed as
The pvalue for the HauckAnderson noninferiority test is , where Z has a standard normal distribution. See Hauck and Anderson (1986) and Schuirmann (1999) for more information.
FarringtonManning (Score) Test
If you specify the METHOD=FM riskdiffoption, PROC FREQ provides the FarringtonManning (score) test of noninferiority for the risk difference. A score test statistic
for the null hypothesis that the risk difference equals – can be expressed as
where is the observed value of the risk difference (),
and and are the maximum likelihood estimates of the row 1 and row 2 risks (proportions) under the restriction that the risk difference is –. The pvalue for the noninferiority test is , where Z has a standard normal distribution. For more information, see Miettinen and Nurminen (1985); Miettinen (1985); Farrington and Manning (1990); Dann and Koch (2005).
The maximum likelihood estimates of and , subject to the constraint that the risk difference is –, are computed as
where
For more information, see Farrington and Manning (1990, p. 1453).
Newcombe Noninferiority Analysis
If you specify the METHOD=NEWCOMBE riskdiffoption, PROC FREQ provides a noninferiority analysis that is based on Newcombe hybridscore confidence limits for the risk difference.
The confidence coefficient for the confidence limits is % (Schuirmann 1999). By default, if you do not specify the ALPHA= option, these are 90% confidence limits. You can compare the confidence limits
with the noninferiority limit, –. If you specify the CORRECT riskdiffoption, the confidence limits includes a continuity correction. See the subsection "Newcombe Confidence Limits" in the section Confidence Limits for the Risk Difference for more information.
If you specify the SUP riskdiffoption, PROC FREQ provides a superiority test for the risk difference. The null hypothesis is
versus the alternative
where is the superiority margin. Rejection of the null hypothesis indicates that the row 1 proportion is superior to the row 2 proportion. You can specify the value of with the MARGIN= riskdiffoption. By default, .
The superiority analysis is identical to the noninferiority analysis but uses a positive value of the margin in the null hypothesis. The superiority computations follow those in the section Noninferiority Tests by replacing – by . See Chow, Shao, and Wang (2003) for more information.
If you specify the EQUIV riskdiffoption, PROC FREQ provides an equivalence test for the risk difference, or the difference between two proportions. The null hypothesis for the equivalence test is
versus the alternative
where is the lower margin and is the upper margin. Rejection of the null hypothesis indicates that the two binomial proportions are equivalent. See Chow, Shao, and Wang (2003) for more information.
You can specify the value of the margins and with the MARGIN= riskdiffoption. If you do not specify MARGIN=, PROC FREQ uses lower and upper margins of –0.2 and 0.2 by default. If you specify a single margin value , PROC FREQ uses lower and upper margins of – and . You can specify the test method with the METHOD= riskdiffoption. The following methods are available for the risk difference equivalence analysis: Wald (with and without continuity correction), HauckAnderson, FarringtonManning (score), and Newcombe (with and without continuity correction). The Wald, HauckAnderson, and FarringtonManning methods provide tests and corresponding testbased confidence limits; the Newcombe method provides only confidence limits. If you do not specify METHOD=, PROC FREQ uses the Wald test by default.
PROC FREQ computes two onesided tests (TOST) for equivalence analysis (Schuirmann 1987). The TOST approach includes a rightsided test for the lower margin and a leftsided test for the upper margin . The overall pvalue is taken to be the larger of the two pvalues from the lower and upper tests.
The section Noninferiority Tests gives details about the Wald, HauckAnderson, FarringtonManning (score), and Newcombe methods for the risk difference. The lower margin equivalence test statistic takes the same form as the noninferiority test statistic but uses the lower margin value in place of –. The upper margin equivalence test statistic take the same form as the noninferiority test statistic but uses the upper margin value in place of –.
The testbased confidence limits for the risk difference are computed according to the equivalence test method that you select. If you specify METHOD=WALD with VAR=NULL, or METHOD=FM, separate standard errors are computed for the lower and upper margin tests. In this case, the testbased confidence limits are computed by using the maximum of these two standard errors. These confidence limits have a confidence coefficient of % (Schuirmann 1999). By default, if you do not specify the ALPHA= option, these are 90% confidence limits. You can compare the testbased confidence limits to the equivalence limits, .
The BARNARD option in the EXACT statement provides an unconditional exact test for the risk (proportion) difference for tables. The reference set for the unconditional exact test consists of all tables that have the same row sums as the observed table (Barnard 1945, 1947, 1949). This differs from the reference set for exact conditional inference, which is restricted to the set of tables that have the same row sums and the same column sums as the observed table. See the sections Fisher’s Exact Test and Exact Statistics for more information.
The test statistic is the standardized risk difference, which is computed as
where the risk difference d is defined as the difference between the row 1 and row 2 risks (proportions), ; and are the row 1 and row 2 totals, respectively; and is the overall proportion in column 1, .
Under the null hypothesis that the risk difference is 0, the joint probability function for a table can be expressed in terms of the table cell frequencies, the row totals, and the unknown parameter as
where is the common value of the risk (proportion).
PROC FREQ sums the table probabilities over the reference set for those tables where the test statistic is greater than or equal to the observed value of the test statistic. This sum can be expressed as
where the set A contains all tables with row sums equal to and , and denotes the value of the test statistic for table a in A. The sum includes probabilities of those tables for which (), where is the value of the test statistic for the observed table.
The sum Prob() depends on the unknown value of . To compute the exact pvalue, PROC FREQ eliminates the nuisance parameter by taking the maximum value of Prob() over all possible values of ,
See Suissa and Shuster (1985) and Mehta and Senchaudhuri (2003).