Glucose Regulation in Acute Stroke Patients (GRASP) Study: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Glucose Regulation in Acute Stroke Patients (GRASP) Study (NCT00282867), a randomized phase 2 study. The primary outcome is Hypoglycemic Events.

2. Clinical Question

The registry record identifies the study as randomized and phase 2, with no masking. Its posted outcome data allow description of the observed counts and unadjusted comparisons. They do not, by themselves, establish whether any differences are attributable to the glucose-control strategy, nor do they resolve other clinical questions about outcomes after stroke. The record does not report detailed eligibility criteria, the method of randomization, or the rationale for the selected glucose targets.

3. Trial Design

GRASP was a randomized, parallel-group phase 2 study sponsored by the University of Virginia. The registry record gives enrollment as 74 and status as completed. It lists a start date of 2006-05 and primary completion date of 2008-02. The design had three groups: tight control, loose control and usual care. Their target glucose levels were respectively 70-110 mg/dL, 70 - 200 mg/dL and 70 - 300 mg/dL.

The record lists IV glucose insulin and potassium, GIK, as a drug intervention and standard care as another intervention. It does not report a complete mapping between these intervention labels and the three group descriptions, so the group names and target levels are the clearest available description of the comparison. The trial was unmasked: the masking field is NONE. The registry does not give further details about allocation concealment, the randomization procedure, or whether participants, clinicians or outcome assessors were aware of group assignment.

The posted primary outcome is hypoglycemic events, measured as participants with an event up to 5 days. The arm counts are 7 of 23 in tight control, 1 of 24 in loose control, and 1 of 25 in usual care. One secondary outcome is posted: favorable 3 Month Modified Rankin, measured at 3 months. The record lists 4 outcome measures overall, but the other outcome details are not reported in the registry record. It lists no reported analyses.

This is a feasibility, safety and preliminary-efficacy study according to its stated purpose. The registry record does not report the original sample-size justification, the intended primary analysis, or a protocol-level statistical analysis plan. The enrollment figure should not be treated as a description of the numbers contributing to every outcome; the outcome-specific arm denominators are reported separately.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Hypoglycemic Events, measured over up to 5 days.

The primary endpoint counts participants who experienced hypoglycemic events up to 5 days. It is a binary participant-level outcome: each participant either had an event during the stated period or did not. The posted results give the number with an event and the number of participants in each group. The endpoint is therefore different from a count of every hypoglycemic episode; the registry describes participants with events, not the total number of episodes.

In general, an endpoint measuring an adverse event can help describe the safety profile of a treatment strategy over a specified follow-up period. For this study, the registry's stated purpose included assessing the feasibility and safety of insulin infusions, and hypoglycemic events are directly relevant to that stated safety question. The endpoint is limited to events during the period up to 5 days, so it does not describe events outside that timeframe.

The registry record does not provide a detailed event definition, thresholds used to identify hypoglycemia, severity categories, or rules for counting recurrent events. It also does not report whether outcomes were assessed by masked reviewers. Those omissions limit how specifically the counts can be interpreted. The endpoint is useful for describing what was recorded in the groups, but it does not alone capture all dimensions of safety, feasibility or stroke recovery.

7. Results Posted in the Registry

• Loose Control Group: 4.2% (1/24) versus 30.4% (7/23) on control; risk difference -26.3 percentage points (95% CI -46.7 to -5.8), risk ratio 0.14 (0.02–1.03); unadjusted two-proportion z-test p = 0.017.
• Usual Care Group: 4.0% (1/25) versus 30.4% (7/23) on control; risk difference -26.4 percentage points (95% CI -46.7 to -6.1), risk ratio 0.13 (0.02–0.99); unadjusted two-proportion z-test p = 0.014.

8. Secondary Efficacy Endpoints

The registry also posts results for secondary outcomes. For each outcome below, the endpoint type, the test applied to the posted summary statistics, the result and its interpretation are given in turn. Secondary endpoints are usually not adjusted for multiple testing, so they support rather than confirm the primary finding.

Favorable 3 Month Modified Rankin

• Endpoint type: Binary: the percentage of participants with the event (counts rebuilt from the posted percentages and group sizes).
• Test: Unadjusted two-proportion z-test, with a Wald 95% confidence interval for the risk difference.
• Result:
  • Loose Control Group: 25.0% (6/24) versus 43.5% (10/23) on control; risk difference -18.5 percentage points (95% CI -45.1 to 8.2), risk ratio 0.58 (0.25–1.33); unadjusted two-proportion z-test p = 0.181.
  • Usual Care Group: 32.0% (8/25) versus 43.5% (10/23) on control; risk difference -11.5 percentage points (95% CI -38.8 to 15.8), risk ratio 0.74 (0.35–1.54); unadjusted two-proportion z-test p = 0.412.
• Interpretation:
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Loose Control Group and the control. The interval gives the range of effects that remain compatible with the data.
  • The 95% confidence interval includes no difference, so these data do not show a statistically significant difference between Usual Care Group and the control. The interval gives the range of effects that remain compatible with the data.

The posted secondary outcome is favorable 3 Month Modified Rankin, recorded at 3 months. Across its two comparisons with tight control, the estimates point below zero, but both confidence intervals include zero and the unadjusted p-values do not provide clear evidence of a difference. The intervals also allow for differences in either direction, so the secondary findings remain uncertain rather than demonstrating equivalence or no effect.

This outcome addresses a later functional measure, while the primary outcome records hypoglycemic events up to 5 days. The two endpoints therefore describe distinct aspects of the trial over different timeframes. The primary pattern cannot be assumed to predict the secondary pattern, and the secondary findings do not validate or overturn the primary comparisons. They should be considered complementary but separate pieces of evidence.

Unadjusted secondary results carry less weight when several comparisons or outcomes may have been examined. In general, repeated testing increases the opportunity for at least one apparently unusual result to arise by chance. The registry record lists 4 outcome measures, reports one secondary outcome, and lists no reported analyses. It does not report a multiplicity plan, a hierarchy of outcomes, or which analyses were prespecified. Consequently, the posted secondary comparisons should be treated as exploratory rather than as stand-alone confirmatory evidence. The registry's stated study purpose includes preliminary efficacy, and the secondary results fit that preliminary context; they do not support a definitive conclusion about functional outcomes.

Secondary outcomes registered for this trial, for reference (only those above were recalculated):

• Favorable 3 Month Modified Rankin

9. Statistical Interpretation

10. Statistical Methods Explained

The primary outcome is binary at the participant level: the record reports whether participants experienced hypoglycemic events up to 5 days. For this type of endpoint, a group proportion is the number of participants with an event divided by the number assessed in that group. The registry reports the event counts and group denominators, and the posted risk-difference comparisons summarize the contrast between each comparator and tight control. A risk difference expresses an absolute difference in event proportions. A negative value in these comparisons indicates a lower observed proportion in the comparator group than in tight control.

A confidence interval describes a range of values reasonably compatible with the data under the analysis used. It communicates uncertainty around the point estimate, rather than identifying a range that must contain the true effect. For a risk difference, zero represents no difference in proportions. The primary-outcome intervals are below zero, whereas the secondary-outcome intervals include zero. General interpretation of whether intervals include zero should be considered alongside their width, the study design, outcome definitions and analysis methods.

The posted p-values are unadjusted. A p-value describes how unusual data at least as incompatible with a specified null hypothesis would be if that null hypothesis and the analysis assumptions held. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of a difference. Since multiple group comparisons and more than one outcome are present, unadjusted p-values do not account for the possibility of chance findings across a set of tests. The registry record does not report a multiplicity adjustment or a prespecified testing hierarchy.

What does the risk difference convey?

The risk difference is an absolute contrast between the proportions in two groups. It can be useful because it directly expresses how far apart the observed event frequencies are on the probability scale. The estimate depends on which group is used as the reference and on the order of subtraction. Here the posted comparisons set tight control as the reference and report the loose-control and usual-care contrasts. The sign should be interpreted using that comparison direction, not as a label for benefit or harm independent of the endpoint.

The registry also posts relative measures, including risk ratios and odds ratios with confidence limits. A risk ratio compares proportions as a ratio; an odds ratio compares the odds of an event. These measures use different scales from the risk difference, and they can communicate the same observed contrast in different ways. The present explanations prioritize the risk difference because it is the metric used for the posted comparison estimates and planning illustrations. The record does not report a sponsor analysis for these comparisons; the analysis count is zero.

Why does small-group uncertainty matter?

General statistical principle: when an outcome is binary and the groups are small, the observed proportions can be sensitive to the outcome status of a small number of participants. Confidence intervals help show the uncertainty that accompanies such estimates. A single point estimate does not convey that uncertainty on its own. It is also important not to confuse the observed arm sizes with the total enrollment: each outcome has its own denominators, and the registry does not provide a participant-flow account explaining every difference.

General statistical principle: the choice of statistical method affects the interval and p-value, especially with sparse events. The record does not report the method used to calculate the posted intervals, whether any continuity correction or exact method was used, or whether covariates were included. The comparison results are labeled unadjusted, and no reported analyses are listed. It is therefore not possible to infer from the registry record that the comparisons account for prognostic factors or other design features beyond the randomized group labels.

What does the planning calculation represent?

The planning illustrations are based on the observed effect and use a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. They show illustrative participants per arm for the stated power targets and effect fractions. In general, power is the probability that a specified analysis will detect an effect of a chosen size under stated assumptions if that effect is present. A planning calculation depends on the assumed effect, allocation, significance level and statistical method. These illustrations do not establish the assumptions used to design the original trial; those assumptions are not reported in the registry record.

General statistical principle: over-dispersion means that observed variation is greater than a chosen model expects. It can make standard errors and confidence intervals based on that model too small if it is not accounted for. Over-dispersion is particularly relevant to some count outcomes, but the primary endpoint posted is a participant-level binary outcome, not a repeated-event count. The registry does not report recurrent-event analyses or an over-dispersion assessment. Because the endpoint counts participants with an event, it should not be interpreted as the total number of hypoglycemic episodes.

Randomization can support a fair comparison by balancing factors on average, but a randomized label does not guarantee balance in a small trial. It also does not remove uncertainty or make an unreported analysis plan knowable. The registry does not describe allocation concealment, baseline characteristics in the posted outcome details, missing-data handling, or the population included in each analysis. Such details matter when assessing whether observed comparisons are robust and how broadly they can be interpreted.

11. Confidence Intervals and Precision

The primary-outcome confidence intervals provide a range of uncertainty around the posted risk differences. For loose control compared with tight control, the estimated risk difference is -0.263, with limits of -0.467 and -0.058. For usual care compared with tight control, it is -0.264, with limits of -0.468 and -0.061. Both intervals are entirely below zero, but their width shows that the magnitude of the difference is not estimated precisely. The point estimates should not be read as exact effects.

For the secondary outcome, favorable 3 Month Modified Rankin at 3 months, the loose-control comparison has an estimated risk difference of -0.185, with limits of -0.451 and 0.082. The usual-care comparison has an estimate of -0.115, with limits of -0.388 and 0.158. Both intervals span zero and include values in either direction. Thus, the posted secondary estimates are compatible with a range of possible differences and do not provide a precise estimate of the direction or size of the contrast.

In general, a confidence interval represents uncertainty under the method and assumptions used to construct it. It is not a guarantee that the true value lies within its limits, and it does not address bias from unmeasured factors, missing outcomes, outcome misclassification or limitations in the analysis. The registry record does not report the calculation method for these intervals or details about missing data. The small outcome-specific group denominators and the limited event counts are important context when interpreting their precision.

12. Planning a Similar Trial: Sample Size

This is a planning illustration, not a re-evaluation of the original trial. It asks: if a new study wanted to detect a difference like the one observed, or a smaller one, how large would it need to be? It is deliberately not “observed power” of the finished trial, which is not a meaningful quantity.

The planning illustrations report participants per arm for each two-arm comparison; they do not report a total number of events. They use the observed effect as a starting point and show planning scenarios for detecting 100% of the observed effect, 75% of the observed effect, or 50% of the observed effect. The stated assumptions are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. These are planning illustrations, not the assumptions documented for the original trial; the trial's own design assumptions are not reported in the registry record.

The 80% power column means the illustrative number of participants per arm under which the calculation gives an 80% chance of detecting the chosen effect, if that effect is present and the assumptions hold. The 90% power column uses a higher target chance of detection and therefore generally requires more participants per arm for the same assumed effect. Neither column describes the probability that a result from the completed study is correct.

Smaller effects are harder to distinguish from random variation, so planning for a smaller fraction of the observed effect requires more participants per arm. These figures are not observed power. They are prospective illustrations that depend on the assumed effect and method, and they should not be treated as evidence about what the completed trial was designed to detect. The registry does not report the original sample-size justification.

13. How the Sample Size Is Derived

Power is the probability that a study of a given size rejects the null hypothesis when a specific true effect exists. It depends on four things: the effect to detect, the variability of the outcome, the significance level (here a two-sided α of 0.05, so z_1−α/2 = 1.960) and the desired power (z_1−β = 0.842 for 80% and 1.282 for 90%).

Formula

n is the number of participants in each arm, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.304; treatment proportion p₁ = 0.042; average p̄ = 0.173.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.5349 = 1.0484.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5016 = 0.4222.
4. n = (1.0484 + 0.4222)² / (-0.2627)² = 31.3, rounded up to 32 participants per arm for 80% power.

The smaller-effect rows in the table above repeat the same calculation with the effect multiplied by 0.75 and 0.5 (for ratios, applied to the logarithm of the ratio).

R code

alpha <- 0.05
z_a   <- qnorm(1 - alpha / 2)          # two-sided
p0 <- 0.304347826086957                          # control proportion
p1 <- 0.0416666666666667                          # treatment proportion
n_per_arm <- function(power, frac = 1) {
  q1 <- p0 + (p1 - p0) * frac               # smaller effect = fraction of the observed difference
  pbar <- (p0 + q1) / 2
  z_b <- qnorm(power)
  ceiling(round((z_a * sqrt(2 * pbar * (1 - pbar)) + z_b * sqrt(p0 * (1 - p0) + q1 * (1 - q1)))^2 / (q1 - p0)^2, 6))
}
for (f in c(1, 0.75, 0.5))
  cat(sprintf("%3.0f%% of observed effect: n per arm = %s (80%% power), %s (90%% power)\n", 100 * f,
              format(n_per_arm(0.80, f), big.mark = ","), format(n_per_arm(0.90, f), big.mark = ",")))

# Check against R's built-in function (unrounded n per arm)
cat("power.prop.test:", round(power.prop.test(p1 = p0, p2 = p1, power = 0.80)$n, 1), "\n")

# Simulation: how often does the z-test reject at the 80%-power sample size?
set.seed(282867)
n    <- n_per_arm(0.80)
reps <- 20000
x0 <- rbinom(reps, n, p0); x1 <- rbinom(reps, n, p1)
ph <- (x0 + x1) / (2 * n)
z  <- ((x1 - x0) / n) / sqrt(ph * (1 - ph) * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(z) > z_a, na.rm = TRUE), reps))

Output from running the code

100% of observed effect: n per arm = 32 (80% power), 42 (90% power)
 75% of observed effect: n per arm = 65 (80% power), 87 (90% power)
 50% of observed effect: n per arm = 165 (80% power), 220 (90% power)
power.prop.test: 31.3 
Simulated power at n = 32 per arm: 0.856 (20000 replicates)

The simulation draws many pairs of event counts at the planned sample size and counts how often the two-proportion z-test rejects the null. It should land close to the target power; at very small required sample sizes it can fall below it, because the normal approximation is rough there.

14. Multiplicity, Missing Data and Analysis Populations

The registry record lists 4 outcome measures and posts one secondary outcome in addition to the primary outcome. It lists two comparisons for each posted outcome, each using tight control as the reference. No analyses are reported in the registry. The record does not describe a prespecified testing hierarchy, a family of primary hypotheses, or a correction for multiple comparisons. The posted p-values are unadjusted, so they should not be interpreted as controlling the overall chance of a false-positive finding across outcomes and comparisons.

General statistical principle: multiplicity matters when several statistical tests are considered together. Even if each test uses a conventional threshold, the chance of at least one false-positive result across a collection of tests can be greater than for one test alone. A prespecified strategy can clarify which result is primary and how other findings should be interpreted. The registry record does not report such a strategy, so no particular adjustment or decision rule can be inferred.

Missing-data handling is not reported in the registry record. The outcome-specific denominators are given, but participant flow, reasons for unavailable outcome data, and any imputation or sensitivity analyses are not described. The analysis population is also not reported: the record does not say whether comparisons followed an intention-to-treat approach, a per-protocol approach, or another definition. It does not report how protocol deviations, withdrawals, or participants with incomplete follow-up were handled.

These omissions limit assessment of whether the posted estimates could be affected by missing outcomes or analysis choices. They do not establish that data were missing or that any particular method was used. The registry record also does not provide enough detail to assess the prespecification of the posted analyses or to determine whether all planned outcomes were reported.

15. What the Result Does, and Does Not, Mean

The posted primary results describe fewer participants with recorded hypoglycemic events in the loose-control and usual-care groups than in the tight-control group during follow-up up to 5 days. The unadjusted risk-difference estimates and their confidence intervals place both comparisons below zero. This is a description of observed group differences, with uncertainty around their magnitude; it is not, by itself, a definitive causal conclusion.

The record identifies the study as randomized, which is a design feature intended to support comparison between groups. But the registry does not provide details of the randomization process, allocation concealment, analysis population, missing-data handling, or adjusted analyses. The outcome definitions and methods for identifying events are also not fully reported. These limitations matter when moving from observed contrasts to broader claims about what caused them.

The primary endpoint covers a limited timeframe and one type of event. It does not summarize all safety outcomes or all aspects of feasibility. The secondary outcome is a separate measure at 3 months, with imprecise estimates that do not resolve the direction of the difference. Neither endpoint alone represents the entirety of stroke recovery or the complete benefit-risk profile of a glucose-control strategy.

The study's stated purpose was to assess feasibility, safety and preliminary efficacy of insulin infusions in hyperglycemic acute ischemic stroke patients. The posted results contribute preliminary information about hypoglycemic events and the listed functional outcome. The record does not provide a complete account of feasibility results, all outcome measures, or a detailed prespecified analysis plan. Therefore, the posted data should be understood within the limits of a phase 2 study and the information documented in the registry, without treating these comparisons as a final evaluation of the strategies.

16. Why This Trial Matters Statistically

This trial illustrates why outcome definition and timeframe are central to clinical-trial interpretation. The primary endpoint counts participants with hypoglycemic events up to 5 days, while the secondary endpoint records a favorable Modified Rankin outcome at 3 months. Those outcomes answer different questions. Clear separation of short-term safety outcomes from later functional outcomes prevents a result for one from being mistakenly generalized to the other.

It also illustrates the value of reporting both point estimates and confidence intervals. The primary comparisons have estimates below zero with intervals that remain below zero, while the secondary comparisons have wider intervals that include zero. Considering intervals alongside p-values makes the degree of uncertainty more visible and discourages a binary reading of results as simply positive or negative.

The trial's three-group design gives two comparisons against a shared tight-control reference. That structure raises practical interpretation questions about comparison direction and multiplicity. The unadjusted results are informative descriptions, but the registry does not report a multiplicity strategy or the original analysis plan. That absence makes it especially important to distinguish exploratory evidence from prespecified confirmatory conclusions.

Finally, the planning illustrations show how assumed effect size and power targets influence the number of participants required per arm. They are not evidence of the trial's original design rationale or of its observed power. Taken together, the registry supports careful statistical reading: identify the endpoint, comparison, uncertainty and analysis limitations before drawing broader conclusions.

17. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record does not report detailed eligibility criteria, baseline characteristics, allocation concealment, or the randomization procedure. It describes the study as unmasked, but does not specify who knew the assigned group or whether outcome assessment was independent. These details could affect assessment of the trial's conduct and the potential for measurement or performance differences between groups.

The intervention descriptions are not fully mapped to the three group labels. The record lists IV glucose insulin and potassium, GIK, and standard care, and it reports target glucose levels for tight control, loose control and usual care. It does not provide enough detail to establish how each listed intervention was implemented in each group. Treatment adherence, co-interventions and achieved glucose levels are not reported in the registry entries.

The endpoint description is limited. The record does not report a detailed definition or ascertainment method for hypoglycemic events, severity categories, or how recurrent events were treated. Although the outcome is reported as participants with an event, the registry does not provide further rules for determining event status. The primary outcome timeframe is up to 5 days, and the posted secondary outcome is at 3 months; outcomes beyond those stated timeframes are not described.

Outcome-specific denominators are reported, but participant flow and reasons for any differences between enrollment and outcome counts are not reported in the registry record. Missing-data methods, analysis populations, protocol deviations, and sensitivity analyses are also not reported. The registry lists 4 outcome measures but provides detail for the primary outcome and one secondary outcome. It lists no reported analyses and does not provide the original sample-size justification or a full statistical analysis plan. These limitations constrain evaluation of completeness, prespecification and robustness.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Glucose Regulation in Acute Stroke Patients (GRASP) Study — NCT00282867.
• Registry record last updated 2009-08-21; status last verified 2009-07.
• Results first posted 2009-08-21.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

21. Record Summary

GRASP, NCT00282867, was a completed randomized, unmasked, parallel phase 2 study sponsored by the University of Virginia. The registry gives enrollment as 74, a start date of 2006-05 and primary completion date of 2008-02. Its stated purpose was to assess feasibility, safety and preliminary efficacy of insulin infusions for hyperglycemic acute ischemic stroke patients.

The three groups were tight control, loose control and usual care, with target glucose levels of 70-110 mg/dL, 70 - 200 mg/dL and 70 - 300 mg/dL. The primary outcome was hypoglycemic events, measured as participants with an event up to 5 days. One secondary outcome, favorable 3 Month Modified Rankin at 3 months, is posted. The registry lists 4 outcome measures and no reported analyses.

The primary comparisons show lower observed event proportions in the loose-control and usual-care groups than in tight control, with unadjusted confidence intervals below zero. The secondary comparisons are less precise, with intervals that include zero. The registry record does not report the original sample-size justification, detailed analysis plan, missing-data handling, or analysis populations, so the posted findings are best read as preliminary comparisons with important uncertainty and reporting limitations.