Connected Healthcare for Individuals Living at Home With Chronic Conditions: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Connected Healthcare for Individuals Living at Home With Chronic Conditions (NCT04339296), a randomized na study. The primary outcome is Recorded Medication Adherence.

2. Clinical Question

3. Trial Design

This was a randomized, parallel-group trial with no masking. The registry lists the study as completed, with an enrollment of 91. The study description states a target sample size of 100, with 50 participants planned for medication management with Spencer and 50 for medication management without Spencer. The number enrolled was therefore not the same as the stated target; reasons for that difference are not reported in the registry record.

The recorded start date was 2019-03-01, and the primary completion date was 2020-07-31. The primary outcome timeframe was 26 weeks. The sponsor was Alberta Health Services, Calgary. The listed phase was NA, which is consistent with a device-focused study rather than a drug phase designation.

The intervention group used Spencer for medication management. The control group continued its existing approach, which could include blister packs, strip packs, pill organizers, or plastic prescription vials. The design was unmasked, so participants and study personnel were not described as blinded to group assignment. Whether outcome assessment involved any additional masking is not reported in the registry record.

The registry identifies recorded medication adherence as the primary outcome. It reports two outcome measures and three analyses overall, but detailed methods for outcomes other than the primary measure are not reported in the registry record. The published primary endpoint summaries include 20 intervention participants and 22 control participants. Those outcome counts are smaller than the enrollment figure, and the registry’s description identifies one intervention participant who withdrew after receiving Spencer without contributing adherence data. Further details about the flow of participants and the handling of missing observations are not reported in the registry record.

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.

Before assignment: A total of 91 participants were assessed for eligibility and 50 were randomized into the two groups; 24 allocated to intervention and 26 allocated to control. A total of 16 spencer participants and 17 control participants completed the study. The study dropout rate was 33% (n=8) and 35% (n=9) in the intervention and control group, respectively.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: Intention-to-treat analysis was utilized. From the 21 intervention participants that started using spencer, adherence data was only collected from 20 participants. One participants withdrew right after receiving spencer and no data was collected. From 24 control participants, no adherence data was collected from two lost to follow participants.

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.

Population: From the 50 participants randomized; 24 were allocated to intervention and other 26 allocated to control. However, one participant in each group refused to participate after allocation and didn't complete the baseline.

6. Primary Endpoint

The registry lists the primary outcome as Recorded Medication Adherence, measured over 26 weeks.

The primary endpoint was recorded medication adherence, expressed as a percentage, over 26 weeks. For the intervention group, the AdhereNet platform recorded whether a medication pouch was removed within the two-hour window after dispensing. Each medication time was scored as 100 if the pouch was removed and 0 if it was not. Weekly adherence was calculated from daily dispensing adherence values, and average weekly adherence was used as the primary outcome.

In the control group, adherence was recorded on medication calendars completed by participants. Calendars were customized to medication timing, and participants marked missed medication times. The endpoint was used to quantify adherence across the study period and compare the groups’ recorded averages.

The two groups did not use the same recording mechanism: the intervention measure came from system records, while the control measure came from participant calendars. This distinction matters when interpreting the comparison because the outcome reflects the operational definition and recording process used in each group. The registry does not report validation of the two recording approaches against a common independent measure.

7. Results Posted in the Registry

• Intervention Group: mean 98.35 versus 91.17 on control; difference 7.18 (95% CI 2.99 to 11.37), standardized effect d = 0.99; z-test p = <0.001.

8. Statistical Interpretation

9. Statistical Methods Explained

A mean is a measure of the center of a continuous outcome. In this trial, the outcome was a percentage of medication adherence recorded over the stated timeframe. A group mean condenses the observed values into one summary, while the standard deviation describes how dispersed the values are around that mean. The reported intervention mean was 98.35 with a standard deviation of 2.15; the control mean was 91.17 with a standard deviation of 9.76. The standard deviations describe variability among the recorded participant outcomes, not uncertainty in the means themselves.

The registry reports a two-sided t-test comparing the groups. In general, a two-sided test evaluates evidence for a difference in either direction rather than only for a prespecified direction. A t-test uses the observed difference and its estimated sampling variability to assess how compatible the data are with a null hypothesis of no mean difference, under the assumptions of the test. The registry gives a p-value of <0.01 and describes the analysis as an equivalence analysis. The equivalence margin and detailed test specification are not reported in the registry record, so the equivalence label cannot be fully evaluated from the posted information.

The separate recalculation is an unadjusted comparison derived from the posted means, standard deviations, and group sizes. It reports a mean difference of 7.18, a 95% confidence interval from 2.994 to 11.37, and p < 0.001. It uses a pooled standard deviation of 7.225 for its standardized effect-size calculation, with a reported standardized difference of 0.994. These recalculated quantities are not adjusted for other variables. The record does not report an adjusted primary analysis or enough detail to determine whether the sponsor used covariate adjustment, a different variance assumption, or another procedure. Because the discrepancy flag is true, the sponsor’s own analysis takes precedence over the recalculation.

In general, a standardized mean difference expresses a contrast relative to a measure of variability, allowing the contrast to be described on a scale without the original outcome units. It does not eliminate differences in how the endpoint was measured. Here, the intervention group’s adherence came from dispenser-system records and the control group’s adherence from participant calendars, so standardization does not make the measurement processes equivalent.

The reported standard errors for the groups are 0.481 and 2.081. In general, a standard error describes the sampling uncertainty of an estimate, whereas a standard deviation describes variation among individual observations. Standard errors depend on both the observed variability and the number of observations. The registry does not provide a detailed derivation of the sponsor’s standard errors or confidence interval, and the sponsor’s analysis does not include a posted confidence interval.

The recalculated confidence interval is an uncertainty range for the unadjusted mean difference under the recalculation’s assumptions. In general, a confidence interval is a range of values compatible with the data and the statistical model at the stated confidence level; it is not a guarantee that the true value lies in that particular interval. Its width reflects the amount of information and variability in the data as well as the method used to construct it.

Statistical significance and equivalence are different questions. A conventional test for a difference asks whether the data provide evidence against a null of no difference. An equivalence analysis instead asks whether the difference is sufficiently contained within prespecified bounds. In general, equivalence requires a clearly specified margin and a method designed to evaluate that margin. The record provides the sponsor’s equivalence label and p-value but does not report the margin, so the basis for that label is not available for independent interpretation. The recalculated unadjusted test is not an equivalence analysis.

In general, over-dispersion means observed variability is greater than a model assumes. When a method underestimates variability, standard errors may be too small and inferential results may be too optimistic. The posted summaries show different standard deviations between groups, but that alone does not establish over-dispersion or identify the sponsor’s variance assumptions. The registry does not report diagnostics for the distribution of adherence values, the handling of bounded percentage data, or a sensitivity analysis using alternative variance methods.

Multiplicity arises when several outcomes, comparisons, or analyses are tested, increasing the chance of at least one apparently positive finding if results are interpreted without accounting for the testing structure. The registry lists two outcome measures and three analyses, but the hierarchy of tests, multiplicity adjustment, and distinction between confirmatory and exploratory analyses are not reported in the registry record. The primary endpoint is identified, but that alone does not reveal how the other analyses affected interpretation.

10. Confidence Intervals and Precision

The recalculated unadjusted mean difference is 7.18, with a 95% confidence interval from 2.994 to 11.37. This interval expresses uncertainty around the estimated difference under the recalculation’s method and assumptions. It is not the confidence interval from the sponsor’s own analysis: the registry reports no confidence limits for that analysis.

In general, a confidence interval gives a range of estimates compatible with the observed data under a specified statistical model. A narrower interval indicates greater precision under that model, while a wider interval indicates more uncertainty. The interval does not capture every potential source of uncertainty. In particular, it does not account for possible bias from different adherence-recording methods, missing outcome data, or any unreported features of the analysis.

The recalculation uses the posted group sizes of 20 and 22 and their reported means and standard deviations. The modest number of outcome observations and the group-specific variability contribute to the uncertainty represented by the interval. The control group’s standard deviation is 9.76, compared with 2.15 in the intervention group; the recalculated pooled standard deviation is 7.225. These summaries do not establish why variability differed, and the registry does not report further distributional detail.

The sponsor’s analysis reports a mean difference of 7.18 and p <0.01, but no confidence interval. The recalculation reports p < 0.001 and a confidence interval. Since the registry flags a discrepancy, the sponsor’s analysis takes precedence. The difference in inferential reporting should therefore be kept visible rather than treating the recalculated interval as if it were the sponsor’s official uncertainty estimate.

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 planning figures are participants per arm, not total events. They are illustrations based on the observed effect, not the original trial’s design calculations. The stated assumptions are a two-sided significance level of 0.05, equal allocation between arms, and a normal approximation; each planning row uses the stated fraction of the observed effect as the effect to detect. The trial’s original sample-size assumptions and justification are not reported in the registry record.

The 80% power column describes the number of participants per arm that the planning calculation estimates would be needed to have an 80% chance of detecting the specified effect under those assumptions, if that effect were the true one. The 90% power column gives the corresponding planning number for a 90% chance. These are planning probabilities conditional on the assumptions, not guarantees about a future study’s result.

The rows consider planning for 100% of the observed effect, 75% of the observed effect, and 50% of the observed effect. At 100% of the observed effect, the illustration gives 16 participants per arm for 80% power and 22 per arm for 90% power. At 75%, it gives 29 per arm for 80% power and 38 per arm for 90% power. At 50%, it gives 64 per arm for 80% power and 86 per arm for 90% power. The pattern illustrates the general principle that smaller effects require more participants to detect with the same power.

These figures are not observed power for the completed trial. They use the observed effect as a planning reference and should not be interpreted as evidence about the probability that the trial’s result is true or reproducible. The original protocol’s assumptions are not reported in the registry record.

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, Δ is the difference in means to detect and σ is the common standard deviation.

Worked calculation for this trial

1. Observed difference Δ = 7.180; pooled standard deviation σ = 7.225.
2. n = 2 × (1.96 + 0.842)² × 7.225² / 7.180² = 15.9, rounded up to 16 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
delta <- 7.17999999999999                      # observed difference in means
sigma <- 7.22536694569902                     # pooled standard deviation
n_per_arm <- function(power, frac = 1) {
  ceiling(round(2 * (z_a + qnorm(power))^2 * sigma^2 / (delta * frac)^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 (t-based, so very slightly larger)
cat("power.t.test:", round(power.t.test(delta = delta, sd = sigma, power = 0.80)$n, 1), "\n")

# Simulation: exact two-sample t-test using simulated means and variances
set.seed(4339296)
n    <- n_per_arm(0.80)
reps <- 20000
m_diff <- rnorm(reps, delta, sigma * sqrt(2 / n))             # difference in sample means
s2     <- sigma^2 * rchisq(reps, 2 * (n - 1)) / (2 * (n - 1))   # pooled sample variance
tstat  <- m_diff / sqrt(s2 * 2 / n)
cat(sprintf("Simulated power at n = %s per arm: %.3f (%d replicates)\n", format(n, big.mark = ","), mean(abs(tstat) > qt(0.975, 2 * (n - 1))), reps))

Output from running the code

100% of observed effect: n per arm = 16 (80% power), 22 (90% power)
 75% of observed effect: n per arm = 29 (80% power), 38 (90% power)
 50% of observed effect: n per arm = 64 (80% power), 86 (90% power)
power.t.test: 16.9 
Simulated power at n = 16 per arm: 0.784 (20000 replicates)

The simulation draws many sets of group means and variances at the planned sample size and counts how often the two-sample t-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. Analyses Reported by the Sponsor

The registry reports the sponsor’s primary comparison as a two-sided t-test, with a mean difference of 7.18 and p <0.01. It labels the analysis as equivalence, but does not report an equivalence margin or confidence interval. Those omissions prevent a full assessment of how the equivalence conclusion was specified and evaluated.

An unadjusted recalculation from the posted group summaries also gives a mean difference of 7.18, but reports a 95% confidence interval from 2.994 to 11.37 and p < 0.001. The recalculation is not the sponsor’s analysis and does not reproduce the sponsor’s stated equivalence procedure. The registry flags a discrepancy between the posted analysis and the recalculated results; the sponsor’s own analysis therefore takes precedence.

The record does not explain the source of the discrepancy. It does not provide sufficient detail to attribute it to a particular test specification, variance assumption, adjustment, or other analytic choice. The two sets of inferential statements should not be blended: the sponsor’s p-value and equivalence label are the registry-reported analysis, while the interval and different p-value belong to an unadjusted recalculation.

Both report the same numeric mean difference, but a common point estimate does not make their inferential procedures interchangeable. The sponsor’s analysis should be cited as the trial’s reported result, with the absence of a posted confidence interval and equivalence margin noted. The recalculation is useful as a transparent comparison based on group summaries, not as a correction or substitute for the sponsor’s result.

14. Multiplicity, Missing Data and Analysis Populations

The registry describes the primary analysis as intention-to-treat. However, the posted outcome summaries include 20 intervention participants and 22 control participants, while enrollment was 91. The record says 21 intervention participants started using Spencer and adherence data were collected from 20; one participant withdrew after receiving Spencer and no data were collected. The registry does not explain why the reported control outcome count is 22 or how any participants without adherence data were handled in the primary analysis.

An intention-to-treat analysis generally seeks to analyze participants according to their randomized group, regardless of adherence to the assigned intervention. In general, missing outcomes create a challenge for that approach because a participant with no outcome cannot contribute an observed endpoint value without a stated method for handling the missingness. The registry does not report whether imputation, a model-based method, or another missing-data procedure was used. It also does not describe sensitivity analyses for missing data. The stated intention-to-treat label cannot by itself establish how missing observations were addressed.

The endpoint was recorded differently by group: platform records for Spencer users and participant calendars for controls. The record does not report a common measurement procedure, validation study, or sensitivity analysis addressing this difference. This creates a measurement issue distinct from missingness: even when a value is recorded, the processes used to generate it differ.

The registry lists two outcome measures and three analyses. It identifies recorded medication adherence as the primary outcome, but does not report the full analysis hierarchy, whether other tests were prespecified, or whether any multiplicity adjustment was used. The equivalence margin and the detailed basis for the sponsor’s equivalence analysis are also not reported. These omissions make it difficult to assess how the primary analysis relates to the other analyses or how the sponsor’s statistical conclusion was constructed.

Participant flow after randomization, reasons for missing observations beyond the noted intervention withdrawal, and analysis-population details are not fully reported in the registry record. These gaps limit assessment of potential attrition bias and the reproducibility of the reported analysis.

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

The posted results describe higher mean recorded adherence in the intervention group than in the control group for the participants included in the outcome summaries. The intervention mean was 98.35 and the control mean was 91.17. The sponsor reported a mean difference of 7.18 and p <0.01; the unadjusted recalculation also gives a mean difference of 7.18, with a 95% confidence interval from 2.994 to 11.37 and p < 0.001. Because a discrepancy is flagged, the sponsor’s own analysis takes precedence.

These statistics do not establish that every participant in one group had higher adherence than every participant in the other. Group means summarize observations and do not describe the full distribution or each participant’s result. The larger control-group standard deviation indicates more spread in its recorded values, but the registry does not report enough distributional information to characterize the shapes of the outcome distributions.

The outcome was recorded through different mechanisms in the two groups. Spencer users’ adherence was based on dispenser records, while control participants marked missed doses on calendars. The contrast therefore reflects the group-specific recording procedures as well as the underlying medication-taking behavior. The registry does not report validation of these methods against a shared measure.

The findings concern recorded adherence over 26 weeks in the participants contributing to the posted outcome summaries. They do not by themselves establish effects on health status, quality of care, or any other outcome. The registry’s primary result does not supply evidence about outcomes beyond those measured and reported. Missing outcome handling and the reason for the control outcome count are not fully described.

The sponsor’s equivalence label should not be interpreted without the missing equivalence margin and fuller method description. Likewise, the recalculated interval is not the sponsor’s interval. Statistical evidence, measurement quality, missing data, and the clinical meaning of the endpoint are separate considerations; the registry record leaves important details about several of them unresolved.

16. Why This Trial Matters Statistically

This trial raises several statistical issues that are common in pragmatic studies of digital health tools. First, the outcome is operationalized differently between groups: system-recorded dispensing and participant-reported calendar tracking. A comparison of group averages can be calculated, but interpretation depends on whether the recording processes capture the same underlying construct with comparable accuracy. The registry does not report a common validation measure.

Second, the endpoint is a percentage and is bounded by its definition, while the reported analysis is a t-test of means. In general, analysts consider whether a continuous-outcome method is suitable for the data’s distribution and variability. The record does not report distributional diagnostics or alternative analyses, so the appropriateness of specific modeling assumptions cannot be independently assessed from the registry information.

Third, the registry uses an equivalence label but does not report the margin that defines acceptable equivalence. General statistical practice distinguishes failure to detect a difference from affirmative evidence that two groups are sufficiently similar under a prespecified margin. The posted label and p-value alone do not provide the information needed to evaluate that distinction.

Fourth, the outcome counts are substantially smaller than overall enrollment, and the handling of missing primary-outcome data is not described in detail. An intention-to-treat label is informative about the intended analysis population, but it does not explain how unavailable outcomes were treated. Participant flow and missing-data methods are needed to assess the robustness of the comparison.

Finally, the discrepancy between the sponsor’s analysis and an unadjusted recalculation illustrates why point estimates, confidence intervals, p-values, and analysis labels should be reported together with clear methods. The same mean difference can be accompanied by different inferential reporting depending on the analysis procedure. The sponsor’s analysis takes precedence, but fuller reporting would make the result easier to evaluate.

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 provides limited detail about the population. The study description focuses on community-dwelling older adults with chronic conditions, but exact age criteria, eligibility rules, and participant characteristics are not reported in the registry record. The listed conditions cover several distinct health circumstances, and the record does not report how many participants had each condition or the extent of medication burden.

The primary endpoint used different measurement systems across groups. Spencer adherence was derived from platform dispensing records, whereas control adherence was recorded by participants on medication calendars. No common independent adherence measure or validation results are reported. Since the intervention itself generated its adherence data, comparability and possible differences in recording behavior are important considerations.

The outcome summaries include 20 intervention participants and 22 control participants, compared with enrollment of 91. The registry notes one intervention participant withdrew after receiving Spencer and had no adherence data. The reason for the control count and the handling of all missing endpoint data are not reported in the registry record. Although the analysis is described as intention-to-treat, the implementation of that approach is not detailed.

The sponsor reports an equivalence analysis but does not provide an equivalence margin or confidence interval. A discrepancy is flagged between the sponsor’s analysis and an unadjusted recalculation; the source is not explained. The sponsor analysis takes precedence, but the record does not include enough methodological detail to reproduce it or fully evaluate its inference.

The registry does not report detailed participant flow, baseline balance, adherence distribution diagnostics, multiplicity procedures, or sensitivity analyses. The posted primary result should therefore be interpreted as a comparison of recorded adherence summaries with unresolved questions about measurement, missing data, and analytic specification.

18. Related Tutorials

19. Related Statistical Calculators

20. Sources

• ClinicalTrials.gov: Connected Healthcare for Individuals Living at Home With Chronic Conditions — NCT04339296.
• Registry record last updated 2020-10-20; status last verified 2020-09.
• Results first posted 2020-10-20.
• 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

NCT04339296 is a completed randomized, parallel-group, unmasked trial of an in-home electronic medication dispensing system for people living at home with chronic conditions. The study description identifies Spencer as the intervention and continued usual medication management as the control approach. Enrollment was 91, and the stated target was 100.

The primary outcome was recorded medication adherence over 26 weeks. The intervention mean was 98.35 among 20 participants, and the control mean was 91.17 among 22 participants. The sponsor reported a mean difference of 7.18 with a two-sided p-value of <0.01 and labeled the analysis as equivalence. An unadjusted recalculation reports a 95% confidence interval from 2.994 to 11.37 and p < 0.001; because a discrepancy is flagged, the sponsor’s analysis takes precedence.

Important details are not reported in the registry record, including the equivalence margin, full missing-data methods, and the reason for the control outcome count. Adherence was recorded differently in the two groups, which should inform interpretation of the comparison.