The SEQDESIGN Procedure

Example 83.12 Creating a Two-Sided Asymmetric Error Spending Design with Early Stopping to Reject or Accept $H_0$

This example requests a four-stage two-sided asymmetric group sequential design for normally distributed statistics. The O’Brien-Fleming boundary can be approximated by a gamma family error spending function with parameter $\gamma =-4$ or –5, and the Pocock boundary can be approximated with parameter $\gamma =1$ (Hwang, Shih, and DeCani, 1990, p. 1440). The following statements use the gamma error spending function with early stopping to reject or accept the null hypothesis $H_0$:

ods graphics on;
proc seqdesign altref=2
               pss(cref=0 0.5 1)
               stopprob(cref=0 1)
               errspend
               plots=(asn power errspend)
               ;
TwoSidedAsymmetric: design nstages=4
                    method=errfuncgamma(gamma=1)
                    method(beta)=errfuncgamma(gamma=-2)
                    method(upperalpha)=errfuncgamma(gamma=-5)
                    alt=twosided
                    stop=both
                    beta=0.1
                    ;
run;
ods graphics off;

The design uses gamma family error spending functions with $\gamma =-5$ for the upper $\alpha $ boundary, $\gamma =1$ for the lower $\alpha $ boundary, and $\gamma =-2$ for the lower and upper $\beta $ boundaries.

The Design Information table in Output 83.12.1 displays design specifications and the derived maximum information. Note that in order to attain the same information level for the asymmetric lower and upper boundaries, the derived power at the upper alternative 0.93655 is larger than the specified $1-\beta = 0.90$.

Output 83.12.1: Design Information

The SEQDESIGN Procedure
Design: TwoSidedAsymmetric

Design Information
Statistic Distribution Normal
Boundary Scale Standardized Z
Alternative Hypothesis Two-Sided
Early Stop Accept/Reject Null
Method Error Spending
Boundary Key Both
Alternative Reference 2
Number of Stages 4
Alpha 0.05
Beta (Lower) 0.1
Beta (Upper) 0.06345
Power (Lower) 0.9
Power (Upper) 0.93655
Max Information (Percent of Fixed Sample) 104.0688
Max Information 3.162386
Null Ref ASN (Percent of Fixed Sample) 74.16654
Lower Alt Ref ASN (Percent of Fixed Sample) 59.10271
Upper Alt Ref ASN (Percent of Fixed Sample) 73.78797


The Method Information table in Output 83.11.2 displays the specified $\alpha $ and $\beta $ error levels and the derived drift parameter. With the same information level used for the asymmetric lower and upper boundaries, only one of the $\beta $ levels is maintained and the other is derived to have the level less than or equal to the specified level.

Output 83.12.2: Method Information

Method Information
Boundary Method Alpha Beta Error Spending Alternative
Reference
Drift
Function
Upper Alpha Error Spending 0.02500 . Gamma (Gamma=-5) 2 3.55662
Upper Beta Error Spending . 0.06345 Gamma (Gamma=-2) 2 3.55662
Lower Beta Error Spending . 0.10000 Gamma (Gamma=-2) -2 -3.55662
Lower Alpha Error Spending 0.02500 . Gamma (Gamma=1) -2 -3.55662


With the STOPPROB(CREF=0 1) option, the Expected Cumulative Stopping Probabilities table in Output 83.12.3 displays the expected stopping stage and cumulative stopping probabilities at each stage under the null reference $\theta = 0$ and under the alternative reference $\theta = \theta _{1}$.

Output 83.12.3: Stopping Probabilities

Expected Cumulative Stopping Probabilities
Reference = CRef * (Alt Reference)
CRef Ref Expected
Stopping Stage
Source Stopping Probabilities
Stage_1 Stage_2 Stage_3 Stage_4
0.0000 Lower Alt 2.851 Rej Null (Lower Alt) 0.00875 0.01556 0.02087 0.02500
0.0000 Lower Alt 2.851 Rej Null (Upper Alt) 0.00042 0.00190 0.00704 0.02500
0.0000 Lower Alt 2.851 Reject Null 0.00917 0.01746 0.02791 0.05000
0.0000 Lower Alt 2.851 Accept Null 0.00000 0.30125 0.79354 0.95000
0.0000 Lower Alt 2.851 Total 0.00917 0.31870 0.82145 1.00000
1.0000 Lower Alt 2.272 Rej Null (Lower Alt) 0.27499 0.58934 0.79601 0.90000
1.0000 Lower Alt 2.272 Rej Null (Upper Alt) 0.00000 0.00000 0.00000 0.00000
1.0000 Lower Alt 2.272 Reject Null 0.27499 0.58934 0.79601 0.90000
1.0000 Lower Alt 2.272 Accept Null 0.00000 0.01863 0.04935 0.10000
1.0000 Lower Alt 2.272 Total 0.27499 0.60797 0.84536 1.00000
0.0000 Upper Alt 2.851 Rej Null (Lower Alt) 0.00875 0.01556 0.02087 0.02500
0.0000 Upper Alt 2.851 Rej Null (Upper Alt) 0.00042 0.00190 0.00704 0.02500
0.0000 Upper Alt 2.851 Reject Null 0.00917 0.01746 0.02791 0.05000
0.0000 Upper Alt 2.851 Accept Null 0.00000 0.30125 0.79354 0.95000
0.0000 Upper Alt 2.851 Total 0.00917 0.31870 0.82145 1.00000
1.0000 Upper Alt 2.836 Rej Null (Lower Alt) 0.00002 0.00002 0.00002 0.00002
1.0000 Upper Alt 2.836 Rej Null (Upper Alt) 0.05945 0.33802 0.72323 0.93655
1.0000 Upper Alt 2.836 Reject Null 0.05947 0.33804 0.72325 0.93657
1.0000 Upper Alt 2.836 Accept Null 0.00000 0.01182 0.03131 0.06343
1.0000 Upper Alt 2.836 Total 0.05947 0.34986 0.75456 1.00000


Rej Null (Lower Alt) and Rej Null (Upper Alt) under the heading Source indicate the probabilities of rejecting the null hypothesis for the lower alternative and for the upper alternative, respectively. Reject Null indicates the probability of rejecting the null hypothesis for either the lower or upper alternative, Accept Null indicates the probability of accepting the null hypothesis, and Total indicates the total probability of stopping the trial.

With the PSS(CREF=0 0.5 1.0) option, the Power and Expected Sample Sizes table in Output 83.12.4 displays powers and expected sample sizes under hypothetical references $\theta = 0$ (null hypothesis $H_{0}$), $\theta = 0.5 \,  \theta _{1}$, and $\theta = \theta _{1}$ (alternative hypothesis $H_{1}$), where $\theta _{1}$ is the alternative reference. The expected sample sizes are displayed in a scale that indicates a percentage of its corresponding fixed-sample size design.

Output 83.12.4: Power and Expected Sample Size Information

Powers and Expected Sample Sizes
Reference = CRef * (Alt Reference)
CRef Ref Power Sample Size
Percent
Fixed-Sample
0.0000 Lower Alt 0.02500 74.1665
0.5000 Lower Alt 0.34601 75.8425
1.0000 Lower Alt 0.90000 59.1027
0.0000 Upper Alt 0.02500 74.1665
0.5000 Upper Alt 0.41647 85.3976
1.0000 Upper Alt 0.93655 73.7880


Note that at $c_{i}=0$, the null reference $\theta =0$, the power with the lower alternative is the lower $\alpha $ error 0.025, and the power with the upper alternative is the upper $\alpha $ error 0.025. At $c_{i}=1$, the alternative reference $\theta =\theta _{1}$, the power with the lower alternative is the specified power 0.90, and the power with the upper alternative 0.93655 is greater than the specified power 0.90 because the same information level is used for these two asymmetric boundaries.

With the PLOTS=POWER option, the procedure displays a plot of the power curves under various hypothetical references, as shown in Output 83.12.5. By default, powers under the lower hypotheses $\theta = c_{i} \,  \theta _{1l}$ and under the upper hypotheses $\theta = c_{i} \,  \theta _{1u}$, are displayed for a two-sided asymmetric design, where $c_{i}= 0, 0.01, 0.02, \ldots , 1.50$ and $\theta _{1l}=-1$ and $\theta _{1u}=1$ are the lower and upper alternative references, respectively.

Output 83.12.5: Power Plot

Power Plot


The horizontal axis displays the multiplier of the reference difference. A positive multiplier corresponds to $c_{i}$ for the upper alternative hypothesis, and a negative multiplier corresponds to $-c_{i}$ for the lower alternative hypothesis. For lower reference hypotheses, the power is the lower $\alpha $ error 0.025 under the null hypothesis ($c_{i}=0$) and is 0.90 under the alternative hypothesis ($c_{i}=1$). For upper reference hypotheses, the power is the upper $\alpha $ error 0.025 under the null hypothesis ($c_{i}=0$) and is 0.93655 under the alternative hypothesis ($c_{i}=1$).

With the PLOTS=ASN option, the procedure displays a plot of expected sample sizes under various hypothetical references, as shown in Output 83.12.6. By default, expected sample sizes under the lower hypotheses $\theta = c_{i} \,  \theta _{1l}$ and under the upper hypotheses $\theta = c_{i} \,  \theta _{1u}$ are displayed for a two-sided asymmetric design, where $c_{i}= 0, 0.01, 0.02, \ldots , 1.50$ and $\theta _{1l}=-1$ and $\theta _{1u}=1$ are the lower and upper alternative references, respectively.

Output 83.12.6: ASN Plot

ASN Plot


The horizontal axis displays the multiplier of the reference difference. A positive multiplier corresponds to $c_{i}$ for the upper alternative hypothesis, and a negative multiplier corresponds to $-c_{i}$ for the lower alternative hypothesis.

By default (or equivalently if you specify BETAOVERLAP=ADJUST), the SEQDESIGN procedure first derives boundary values without adjusting for the possible overlapping of the two one-sided $\beta $ boundaries based on two corresponding one-sided tests. Then the procedure checks for overlapping of the $\beta $ boundaries at the interim stages. Since the two $\beta $ boundaries overlap at stage 1, the $\beta $ boundary values for stage 1 are set to missing, the $\beta $ spending values at stage 1 are set to zero, and the $\beta $ spending values at subsequent stages are adjusted proportionally.

The Boundary Information table in Output 83.12.7 displays the information levels, alternative references, and boundary values. By default (or equivalently if you specify BOUNDARYSCALE=STDZ), the standardized Z scale is used to display the alternative references and boundary values. The resulting standardized alternative references at stage k is given by $\pm \theta _1 \sqrt {I_ k}$, where $\theta _1$ is the specified alternative reference and $I_ k$ is the information level at stage k, $k= 1, 2, 3, 4$.

Output 83.12.7: Boundary Information

Boundary Information (Standardized Z Scale)
Null Reference = 0
_Stage_   Alternative Boundary Values
Information Level Reference Lower Upper
Proportion Actual Lower Upper Alpha Beta Beta Alpha
1 0.2500 0.790597 -1.77831 1.77831 -2.37610 . . 3.33772
2 0.5000 1.581193 -2.51491 2.51491 -2.35714 -0.48408 0.29400 2.94871
3 0.7500 2.37179 -3.08012 3.08012 -2.34861 -1.36183 1.13898 2.50473
4 1.0000 3.162386 -3.55662 3.55662 -2.32105 -2.32105 1.95675 1.95675


With ODS Graphics enabled, a detailed boundary plot with the rejection and acceptance regions is displayed, as shown in Output 83.12.8.

Output 83.12.8: Boundary Plot

Boundary Plot


The Error Spending Information in Output 83.12.9 displays the cumulative error spending at each stage for each boundary.

Output 83.12.9: Error Spending Information

Error Spending Information
_Stage_ Information
Level
Cumulative Error Spending
Lower Upper
Proportion Alpha Beta Beta Alpha
1 0.2500 0.00875 0.00000 0.00002 0.00042
2 0.5000 0.01556 0.01863 0.01184 0.00190
3 0.7500 0.02087 0.04935 0.03132 0.00704
4 1.0000 0.02500 0.10000 0.06345 0.02500


With the $\beta $ boundary values missing at stage 1, there is no early stopping to accept $H_0$ at stage 1, and the corresponding $\beta $ spending at stage 1 is computed from the rejection region. For example, the upper $\beta $ spending at stage 1 (0.00002) is the probability of rejecting $H_0$ for the lower alternative under the upper alternative reference.

With the PLOTS=ERRSPEND option, the procedure displays a plot of the cumulative error spending on each boundary at each stage, as shown in Output 83.12.10.

Output 83.12.10: Error Spending Plot

Error Spending Plot