Provision of HIV Self-Test Kit to Emergency Department Patients: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Provision of HIV Self-Test Kit to Emergency Department Patients (NCT03021005), a randomized na study. The primary outcome is Number of Participants Who Did HIV Testing.

2. Clinical Question

3. Trial Design

The registry describes a completed randomized, parallel-group pilot study of providing HIV self-testing kits to emergency department patients. The study took place at the Johns Hopkins Hospital emergency department, according to the brief summary. The registry lists an enrollment of 200, a start date of 2016-03, a primary completion date of 2020-01, and a completed status. The registry record does not report the trial’s original sample-size justification or the assumptions used to plan enrollment.

Participants were randomized to a self-testing-kit group or a no-kit reference group. The registry lists no masking. In the intervention description, participants could take home a free FDA-approved self-testing kit, and the group description also mentions five referral cards for partners and peers. The comparator group was not offered the kit or referral cards. Both groups were described as completing a short survey about sociodemographic information and HIV-testing experiences and perceptions.

The primary outcome was the number of participants who did HIV testing, assessed by telephone follow-up at one month after enrollment. The outcome record lists 63 participants in the self-testing-kit group and 72 in the no-kit group, with 36 and 9 participants, respectively, recorded as having done testing. The outcome description separately reports that 39 participants in the kit group and 26 in the no-kit group were lost to follow-up. The registry record does not explain how these outcome counts relate to all randomized participants or specify how missing follow-up information was handled in the analysis.

The registry lists one outcome measure and no posted secondary outcomes or analyses. It does not report a detailed statistical analysis plan, allocation procedure, concealment process, or whether any additional analyses were conducted.

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:

Analysis population for the primary outcome, as defined in the registry: 39 participants in Self Testing Kit group and 26 participants in No Self Testing Kit group were lost to follow-up.

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 Number of Participants Who Did HIV Testing, measured over At 1 month post enrollment.

The primary endpoint is the number of participants who did HIV testing by one month after enrollment. The registry says that HIV testing rate was determined by telephone follow-up after the index emergency department visit. The posted outcome counts are 36 participants in the self-testing-kit group and 9 in the no-kit group; the corresponding group counts recorded for the outcome are 63 and 72.

This endpoint addresses whether participants went on to test for HIV after the emergency department encounter, rather than measuring a laboratory or clinical outcome. It is aligned with the study’s stated aim of examining uptake and engagement with HIV testing after offering a self-testing kit. A count of participants who tested is a binary-outcome summary: each participant is classified as having or not having done testing during the outcome period. The registry record does not provide further detail about what testing methods qualified, how testing was verified, or how the telephone follow-up classified uncertain or unreported testing.

The endpoint is useful for describing the observed testing behavior in the two randomized groups, but its interpretation depends on follow-up and outcome ascertainment. The record reports losses to follow-up in both groups and does not report a missing-data procedure. Consequently, the posted counts should be read as the recorded outcome results, not assumed to represent every enrolled participant. The one-month assessment also speaks only to testing by that follow-up point; the registry record does not report a longer-term primary outcome.

7. Results Posted in the Registry

• No Self Testing Kit: 12.5% (9/72) versus 57.1% (36/63) on control; risk difference -44.6 percentage points (95% CI -59.1 to -30.2), risk ratio 0.22 (0.11–0.42); unadjusted two-proportion z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a binary event: a participant either did HIV testing by the one-month follow-up or did not. For this kind of endpoint, group results can be described as counts and risks. The registry reports 36 participants with testing among 63 in the self-testing-kit group, and 9 among 72 in the no-kit group. A risk is the proportion of a group with the event. The comparison results give the group risks as 0.571 and 0.125, respectively. The registry record does not give further detail about how incomplete or uncertain reports of testing were classified.

What does the risk difference describe?

The risk difference subtracts one group’s risk from the other group’s risk. In the posted comparison, the no-kit group is compared with the self-testing-kit group, and the reported difference is -0.446. A negative estimate means that the recorded risk was lower in the no-kit group. The risk difference is an absolute measure: it describes the separation between the group risks on the probability scale. It is not a relative change and should not be read as the percentage of participants who would change their behavior if offered a kit.

How do relative measures differ?

The registry also reports a risk ratio and an odds ratio for the no-kit group relative to the kit group. A risk ratio divides one group’s risk by the other group’s risk. An odds ratio compares the odds, where odds are the chance of the event relative to the chance of no event. Both are relative measures, but they answer different mathematical questions. When outcomes are common, an odds ratio can be farther from the neutral value than the risk ratio, so an odds ratio should not be described as though it were a risk ratio. The posted comparison gives the risk ratio as 0.219 and the odds ratio as 0.107.

What does the p-value tell the reader?

The registry reports p < 0.001 for the unadjusted comparison. In general, a p-value describes how compatible the observed data are with a specified null model, under the assumptions of the statistical test. A small p-value is not the probability that the null hypothesis is true, nor does it measure the size or practical importance of an effect. The registry record does not identify the test procedure or provide a detailed analysis plan, so the test’s implementation cannot be assessed from the record. The p-value should be read alongside the effect estimate, its confidence interval, and the study’s limitations.

Why does randomization matter?

In general, random assignment aims to make the groups comparable at the start of a study, on average, so that differences in outcomes can be interpreted as evidence about the assigned interventions. The record identifies the study as randomized and parallel-group, but does not report the allocation sequence, allocation concealment, or implementation details. Randomization does not itself resolve incomplete outcome ascertainment, deviations from assigned treatment, or limitations in applying the results beyond the trial setting. The record does not describe an analysis by assigned group or another analysis population.

Why is over-dispersion not the central issue?

Over-dispersion is a concern when the variability in count data is greater than a model assumes, often because observations are more variable or dependent than expected. In general, it matters for some count-based models, where underestimating variability can make intervals too narrow and tests too optimistic. The primary outcome is reported as participants with an event, rather than as a rate of repeated events or a count process with exposure time. The registry does not report use of a count model or provide a dispersion estimate, so over-dispersion is not a feature that can be evaluated from the posted analysis.

The registry record lists one primary outcome, no secondary outcomes, and no posted analyses. It does not report whether there were interim analyses, adjustment for covariates, a prespecified multiplicity strategy, or a sensitivity analysis for missing outcomes. These omissions limit how fully the statistical procedure can be reconstructed. The reported estimates are best described as the registry’s unadjusted comparison of the recorded one-month outcome counts.

10. Confidence Intervals and Precision

The registry’s risk-difference estimate for the no-kit group relative to the self-testing-kit group is -0.446, with a confidence interval from -0.591 to -0.302. A confidence interval describes a range of effect estimates that are reasonably compatible with the data under the interval’s statistical assumptions. In general, it reflects sampling uncertainty; it does not account automatically for every possible source of bias, such as missing outcomes or errors in outcome classification.

The reported interval for the risk difference lies below zero, consistent with a lower recorded risk in the no-kit group than in the kit group. The estimate is an absolute contrast, and the interval indicates uncertainty about the size of that contrast. The interval is not a range containing a fixed proportion of individual participant outcomes, and it is not a guarantee that the true effect lies within its endpoints.

The registry also reports intervals for the risk ratio and odds ratio. For the risk ratio, the estimate is 0.219, with a confidence interval from 0.115 to 0.418. For the odds ratio, the estimate is 0.107, with a confidence interval from 0.045 to 0.253. These measures use different scales, so their numerical values and intervals should not be compared as though they were the same effect measure. The estimates and intervals all describe the no-kit group relative to the self-testing-kit group.

Precision depends on the amount of outcome information and the variability in the data, among other factors. The registry reports losses to follow-up in both groups, but does not state how those missing outcomes were handled in the interval calculations. The intervals therefore express statistical uncertainty under the analysis used, not uncertainty about all possible consequences of incomplete follow-up. The registry record does not report the method used to construct the confidence intervals, so that aspect of their calculation cannot be evaluated.

11. 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 sample-size table and curve are planning illustrations based on fractions of the observed effect, not the trial’s original enrollment plan. The registry record does not report the assumptions used for the trial’s own design or sample-size justification. The posted planning assumptions are a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. For each illustration, the effect to detect is the stated fraction of the observed effect.

The figures are participants per arm, not total participants across both arms. The table gives separate planning figures for 80% power and 90% power. In plain terms, power is the probability that a planned study would detect an effect of the specified size at the chosen significance level, if that effect were present and the model assumptions held. The 80% and 90% columns show the estimated participants per arm required under those respective planning targets. They are not estimates of the number enrolled in this pilot.

The rows describe planning for 100%, 75%, and 50% of the observed effect. Detecting a smaller effect generally requires more participants, because a smaller difference is harder to distinguish from sampling variation. The posted figures increase as the assumed effect fraction decreases: for 100% of the observed effect, the illustrations are 17 participants per arm for 80% power and 22 for 90% power; for 75%, they are 33 and 44; for 50%, they are 78 and 103.

These calculations are conditional illustrations, not a forecast of what a future study will achieve. They use the observed effect as their starting point, which may not match the effect in a different study. This is not observed power.

12. 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 = 0.571; treatment proportion p₁ = 0.125; average p̄ = 0.348.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6737 = 1.3205.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.5952 = 0.5009.
4. n = (1.3205 + 0.5009)² / (-0.4464)² = 16.6, rounded up to 17 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 <- 0.571428571428571                          # control proportion
p1 <- 0.125                          # 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(3021005)
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 = 17 (80% power), 22 (90% power)
 75% of observed effect: n per arm = 33 (80% power), 44 (90% power)
 50% of observed effect: n per arm = 78 (80% power), 103 (90% power)
power.prop.test: 16.6 
Simulated power at n = 17 per arm: 0.819 (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.

13. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure: the number of participants who did HIV testing at one month after enrollment. It lists no posted secondary outcomes and no posted analyses. The record does not report a multiplicity adjustment, a hierarchy of outcomes, or a prespecified strategy for interpreting multiple tests. Since no secondary outcome results are posted, the record does not allow assessment of whether additional outcomes were measured or analyzed but not reported.

The outcome description reports that 39 participants in the self-testing-kit group and 26 in the no-kit group were lost to follow-up. This is important because the primary outcome was determined by telephone follow-up. The outcome counts nevertheless appear with group counts of 63 and 72, and the record does not explain how these counts relate to all randomized participants. It also does not say whether participants lost to follow-up were excluded, treated as not having tested, handled through imputation, or addressed in another way. Those details are not reported in the registry record.

The record does not describe an intention-to-treat analysis, a per-protocol analysis, or another analysis population. It does not state whether every randomized participant was analyzed according to assigned group, or whether the reported counts reflect only participants with available follow-up. Because the missing-data approach and analysis population are not specified, the possible influence of missing outcomes on the observed comparison cannot be quantified from the registry record.

The registry also does not report interim analyses, stopping rules, prespecified subgroup analyses, or sensitivity analyses. These items should not be assumed to have occurred or not occurred; they are simply not reported in the registry record. The posted p-value and confidence intervals should therefore be interpreted as results of the reported comparison, without assuming unreported adjustments or procedures. A fuller assessment of multiplicity and missing-data handling would require a more detailed analysis description.

14. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0.01% reporting threshold.

The registry’s adverse-event record lists zero serious adverse events in the self-testing-kit group and zero in the no-kit group. It also lists zero deaths in each group. The serious-event and other-event fields record 102 participants at risk in the kit group and 98 in the no-kit group; the affected counts for other adverse events are zero in both groups. No serious adverse-event terms are listed.

These are the safety figures recorded in the registry, and they should be read as reported counts rather than evidence that no adverse events could have occurred. The record does not provide definitions of the events being monitored, the method or duration of safety surveillance, or further detail about how the at-risk groups relate to the outcome groups. It also does not describe how adverse events were solicited or classified. The listed zero counts do not establish that the intervention is free of risk, and they do not provide a comparative safety analysis beyond these reported figures.

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

The primary outcome shows that more participants in the self-testing-kit group than in the no-kit group were recorded as having done HIV testing by one month. The posted counts are 36 of 63 in the kit group and 9 of 72 in the no-kit group. The registry’s unadjusted comparison reports a risk difference of -0.446 for the no-kit group relative to the kit group, with p < 0.001. The confidence interval for that difference does not include zero, indicating that the reported estimates are inconsistent with an equal group risk under the interval’s assumptions.

That result describes the recorded outcome in this randomized pilot study. It does not establish that offering a kit caused all of the observed difference, or that the same result would occur in another emergency department, a different population, or with a different approach to offering self-testing. The registry record does not report how missing follow-up outcomes were handled, and it reports losses to follow-up in both groups. This limits assessment of how incomplete follow-up might affect the comparison.

The study’s intervention involved more than simply a kit offer: the kit group was also described as receiving referral cards for partners and peers. The posted primary result cannot separate the contribution of the kit from the contribution of the referral cards, or from other aspects of the intervention. The record also does not provide detailed information about testing verification, allocation procedures, or the statistical analysis plan.

The result does not establish longer-term testing behavior, partner or peer testing uptake, or other health outcomes. Those are not included as posted secondary outcome results in the registry record. The zero adverse-event counts should not be interpreted as proof of absence of risk, given the limited safety detail reported. The appropriate reading is narrow: the registry reports a difference in one-month testing counts between the randomized groups, with uncertainty and limitations that leave several explanatory and generalizability questions unanswered.

16. Why This Trial Matters Statistically

This trial illustrates a statistical distinction between an intervention’s intended mechanism and the outcome used to evaluate it. The study’s stated aim concerned uptake and engagement with HIV testing after offering self-testing, while the primary endpoint was the number of participants who did testing by one month. That endpoint is a direct behavioral outcome, but it does not by itself explain what led participants to test or whether testing continued later.

The binary outcome allows the group results to be expressed as counts, risks, and contrasts. Reporting both the group counts and an absolute risk difference makes the scale of the comparison more interpretable than a p-value alone. Relative measures are also reported, but the risk ratio and odds ratio answer different questions and can suggest different magnitudes. Reading the measures together helps keep statistical significance separate from effect size and scale.

The reported losses to follow-up are also statistically important. A randomized design can support a fair comparison, but missing outcome information can weaken that interpretation if follow-up differs or if missingness relates to testing behavior. Since the registry does not describe the missing-data procedure or analysis population, readers cannot determine from the record how those losses entered the primary analysis.

Finally, this is a pilot study with a single posted primary outcome and no posted secondary analyses. Its result can motivate questions about feasibility, engagement, outcome ascertainment, and the design of later studies, but the registry record does not report the original planning assumptions or enough analysis detail to reconstruct the full inferential process. The planning illustrations based on fractions of the observed effect are useful for understanding how assumed effect size influences participant requirements, but they are not evidence about the power of the completed trial.

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.

Several limitations specific to this trial affect interpretation of the posted outcome. The registry describes a pilot conducted among emergency department patients who declined routine HIV screening and patients at increased risk for HIV. The record does not report detailed eligibility criteria or the characteristics of those represented in the outcome counts, so the extent to which the result applies to other patients or settings cannot be determined from the registry alone.

The primary outcome was assessed by telephone at one month. The registry record does not describe whether reported testing was independently verified, what counted as testing, or how uncertain responses were classified. It also does not report outcomes beyond the one-month assessment. Consequently, the result cannot establish longer-term testing behavior.

Losses to follow-up are reported for both groups, but the registry does not explain how they were handled. The outcome group counts are not reconciled with the registry’s enrollment figure, and the analysis population is not specified. This makes it difficult to assess the extent to which missing data could have influenced the comparison.

The intervention group was offered both a home test kit and referral cards for partners and peers, while the comparator received neither. The primary outcome cannot identify the separate contribution of each component. The record also lacks details about randomization implementation, allocation concealment, and the statistical analysis plan. No secondary outcomes or analyses are posted, and the registry record does not report the original sample-size justification, interim analyses, or sensitivity analyses. These details should be treated as not reported, rather than inferred from the observed results.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Provision of HIV Self-Test Kit to Emergency Department Patients — NCT03021005.
• Registry record last updated 2021-10-08; status last verified 2021-10.
• Results first posted 2020-03-20.
• 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

The registry describes NCT03021005, “Provision of HIV Self-Test Kit to Emergency Department Patients,” as a completed randomized, parallel-group pilot study with no masking. The listed enrollment is 200. The study began in 2016-03, and primary completion was recorded as 2020-01. The sponsor is Johns Hopkins University.

The primary outcome was the number of participants who did HIV testing at one month after enrollment, assessed by telephone follow-up. The posted counts were 36 of 63 in the self-testing-kit group and 9 of 72 in the no-kit group. The registry reports an unadjusted comparison with p < 0.001 and a risk-difference confidence interval that excludes zero. It also reports losses to follow-up in both groups, without describing the missing-data method or analysis population.

The registry lists no secondary outcomes or analyses. It records zero serious adverse events and deaths in both groups. Further details about the trial’s original sample-size rationale, analysis plan, and follow-up procedures are not reported in the registry record.