Central Neck Dissection in Patients With Clinical Node Negative Thyroid Cancer: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Central Neck Dissection in Patients With Clinical Node Negative Thyroid Cancer (NCT02138214), a randomized phase 2 study. The primary outcome is Percentage of Participants With Transient Hypoparathyroidism, as Defined by a Day 1 Serum Parathyroid Hormone (PTH) Level of < 10 pg/ml.

2. Clinical Question

3. Trial Design

This was a phase II, randomized, double-masked, parallel-group trial of surgery for node-negative thyroid cancer. The registry records enrollment of 117 and a completed status. The trial start date was 2014-06-06, and primary completion was 2019-10-30. The record lists Stage I, Stage II and Stage III papillary thyroid cancer among the conditions.

Randomization applied to Arms I and II. Arm I received total thyroidectomy alone. Arm II received total thyroidectomy with ipsilateral prophylactic central lymph node dissection. Arm III was a standard-of-care group for patients not eligible for randomization into either randomized arm; those patients were treated according to patient and physician preference. The registry describes Arm III as added later as a comparison for qualitative aims. The meaning of double masking—specifically who was masked—is not reported in the registry record.

The stated purpose was to compare surgical approaches and assess quality of life, as well as outcomes including hypocalcemia, voice and swallowing problems, and clinical recurrence. The primary outcome addressed a postoperative complication measure: the percentage with transient hypoparathyroidism, defined using day 1 serum PTH. The registry lists 13 outcome measures and 6 analyses. It does not report enough detail to reconstruct how all those outcomes and analyses were handled statistically.

The primary endpoint comparison was between Arm I and Arm II only. Arm III was not included because the registry says its data were not collected for this outcome: the measure was intended to compare complications between the total-thyroidectomy and CND arms and did not apply to SOC arm participants. One participant in Arm II was a screen failure, according to the endpoint population description. The number enrolled in the whole trial should therefore not be treated as the denominator for this particular comparison.

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: 117 participants consented. 27 participants were excluded from the study before assignment to arms (15 participants withdrew before surgery, 12 participants did not meet intraoperative inclusion criteria). 90 participants were allocated in 3 arms

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: 1 participant in the Arm II(CND) was a screen failure. Arm III(SOC) data was not collected for this outcome measure as outcome is a comparison of complication rates between TT and CND arms and does not apply to SOC arm participants. The SOC arm was added later as comparison for the qualitative aims only.

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 Percentage of Participants With Transient Hypoparathyroidism, as Defined by a Day 1 Serum Parathyroid Hormone (PTH) Level of < 10 pg/ml, measured over Post-operative day 1.

The primary endpoint is the percentage of participants with transient hypoparathyroidism, defined in the record as a serum parathyroid hormone (PTH) level below 10 pg/ml on postoperative day 1. The measure is binary at the participant level: each participant either meets the stated day 1 threshold or does not. The reported arm values are percentages, and the comparison is between total thyroidectomy alone and total thyroidectomy with ipsilateral prophylactic central lymph node dissection.

This endpoint is used to capture an early postoperative biochemical complication. In general, a defined laboratory threshold can make an outcome more reproducible than an unstructured clinical impression, because the same stated criterion can be applied to each participant. That general advantage depends on consistent timing and measurement; details of laboratory procedures, assessment completeness, and how missing PTH measurements were handled are not reported in the registry record.

The endpoint is specifically transient hypoparathyroidism as operationalized by the day 1 PTH criterion. It is not a direct measure of permanent hypoparathyroidism, overall quality of life, voice or swallowing problems, recurrence, or the full balance of surgical outcomes. Those topics appear among other outcomes in the record, but the result discussed for the primary endpoint addresses only the stated day 1 measure. Arm III is not part of this endpoint comparison.

7. Results Posted in the Registry

• Arm I (no CND): 33.3% (10/30) versus 24.1% (7/29) on control; risk difference 9.2 percentage points (95% CI -13.8 to 32.2), risk ratio 1.38 (0.61–3.14); unadjusted two-proportion z-test p = 0.436.

8. Statistical Interpretation

9. Statistical Methods Explained

The primary outcome is a binary event measure: the registry defines an event as a day 1 serum PTH level below 10 pg/ml. The arm-level summaries are percentages of participants meeting that definition. For a binary outcome, the central descriptive question is how often the event occurred in each group. Here, the posted percentages are 24.1% for Arm II and 33.3% for Arm I. The accompanying event counts are 7 and 10, and the endpoint denominators are 29 and 30.

What does the risk difference express?

The recalculated comparison uses a risk difference, with Arm I minus Arm II as its direction. A risk difference subtracts one group’s event proportion from the other group’s event proportion. Positive values indicate a larger observed event proportion in Arm I under this direction; negative values indicate a larger observed proportion in Arm II. The recalculated estimate is 0.092. This is an unadjusted comparison based on the posted summary statistics, not an adjusted model estimate.

A risk difference is an absolute contrast on the proportion scale. It is distinct from a risk ratio or odds ratio, which compare groups using a relative scale. The record includes relative measures in the recalculated comparison, but the main interpretation focuses on the risk difference because it states the difference in event proportion directly. No adjustment for baseline characteristics is reported for the recalculation.

What does the interval add?

A confidence interval expresses the uncertainty around an estimate under the assumptions of the method used to construct it. In general, a wider interval indicates less precision, while an interval spanning the null value for a risk difference—zero—does not distinguish the observed data from no difference at the corresponding interval level. The recalculated interval spans negative and positive values. The confidence level and interval construction details are not reported in the registry record, so the exact nominal coverage should not be inferred from the posted result.

For this comparison, the interval indicates that the point estimate is not a precise account of the range of differences compatible with the data. It is important not to read the interval as a probability statement about the true difference, or as proof that the groups are equal. It describes sampling uncertainty under the assumptions of the calculation. Other sources of uncertainty, such as outcome measurement, missing data, and the analysis population, are not captured by the interval alone.

How should the p-values be read?

The registry reports a Fisher Exact test with p = 0.567. Fisher’s Exact test is a method for comparing proportions in a two-group contingency table, and it can be useful when event counts are small. The record does not specify whether the test was one-sided or two-sided, nor does it report a detailed statistical analysis plan. The recalculation gives an unadjusted p-value of 0.436 using the posted summaries. It is a separate, simple recalculation and should not be conflated with the sponsor’s reported test.

In general, a p-value measures how unusual data at least as discrepant as those observed would be under a specified null hypothesis and test procedure. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. A p-value above a conventional threshold does not establish equivalence or absence of an effect. The record does not state the threshold used for the sponsor’s analysis, so no threshold-based sponsor conclusion should be inferred.

Why does the small event count matter?

With a binary outcome and modest group sizes, a small change in the event count can noticeably change the estimated percentage. Statistical methods account for uncertainty in different ways, and exact methods such as Fisher’s Exact test avoid relying on some large-sample approximations. Even an exact test cannot make a small dataset precise: the amount of information still depends on the observed counts and denominators.

In general, over-dispersion is a concern when observed variation is larger than a model assumes. It is more commonly relevant to count or clustered-data models than to this simple binary comparison. The registry does not report a model-based analysis of over-dispersion for the primary endpoint. For the posted comparison, the main interpretive issue is the limited precision associated with the event counts and the uncertainty interval, rather than a reported over-dispersion adjustment.

The primary endpoint comparison excludes Arm III by design of the outcome collection described in the record. The recalculation compares only Arms I and II and uses the posted group summaries. Details of randomization procedures, allocation concealment, analysis-set definitions, covariate adjustment, and missing-data methods are not reported in the registry record. Those omissions limit the ability to independently assess whether the analysis fully followed a prespecified plan.

10. Confidence Intervals and Precision

The main precision feature is the uncertainty interval around the recalculated risk difference. The estimate, Arm I minus Arm II, is 0.092, while the interval extends into both negative and positive values. This means the data are consistent with effects in either direction under the interval calculation. Because the interval crosses zero, it does not establish a difference at the interval’s confidence level; the confidence level itself is not reported in the registry record.

The point estimate should not be read alone. The posted percentages are 33.3% in Arm I and 24.1% in Arm II, but the comparison is based on 10 and 7 endpoint events in groups with denominators of 30 and 29. With counts of this size, the observed percentages may shift appreciably if event classifications differ or additional observations are available. The interval reflects sampling uncertainty under its calculation, but not every possible source of uncertainty.

The registry-reported Fisher Exact p-value is 0.567, while the unadjusted recalculation reports p = 0.436. Neither value is a measure of precision or effect magnitude. The interval is more informative for seeing the range of differences compatible with the recalculated estimate, while the p-values summarize evidence against a specified null under different testing procedures. The reason for the numerical difference between the tests is not reported in the registry record.

The group comparison is limited to Arms I and II, and Arm III did not contribute primary endpoint data. The record also notes a screen failure in Arm II. It does not report endpoint missingness beyond that statement, confidence-interval construction details, or the analysis population definition. These gaps matter because uncertainty from incomplete outcome measurement or analysis-set choices is not necessarily reflected in a calculation based only on the posted summaries.

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 illustrative calculations based on the observed effect, rather than the trial’s original design assumptions. For this binary endpoint, the figures are participants per arm, not total events. The calculation assumptions stated in the record are a two-sided significance level of 0.05, equal allocation between the two arms, and a normal approximation. The effect used for each planning scenario is the stated fraction of the observed effect.

The 80% power column describes the number of participants per arm in a hypothetical study that would have an 80% chance of detecting the specified effect under those assumptions, if that effect were the true one. The 90% power column describes a higher target chance of detecting that same specified effect. These are design-planning probabilities, not guarantees for a future study and not evidence that the completed trial had a particular probability of finding a result.

For planning for 100% of the observed effect, the illustrations give 379 participants per arm for 80% power and 507 participants per arm for 90% power. For planning for 75% of the observed effect, they give 659 participants per arm for 80% power and 881 participants per arm for 90% power. For planning for 50% of the observed effect, they give 1,444 participants per arm for 80% power and 1,932 participants per arm for 90% power. Smaller effects require more participants under these assumptions because they are harder to distinguish from random variation.

These values are not observed power. They are planning illustrations derived from the observed effect, and they should not be interpreted as the trial’s original sample-size target or justification. The original design assumptions and sample-size justification 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, p₀ and p₁ are the control and treatment proportions, and p̄ is their average.

Worked calculation for this trial

1. Control proportion p₀ = 0.241; treatment proportion p₁ = 0.333; average p̄ = 0.287.
2. First term: z_1−α/2 × √[2p̄(1−p̄)] = 1.96 × 0.6400 = 1.2543.
3. Second term: z_1−β × √[p₀(1−p₀) + p₁(1−p₁)] = 0.842 × 0.6367 = 0.5358.
4. n = (1.2543 + 0.5358)² / (0.0920)² = 379.0, rounded up to 379 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 <- 0.241379310344828                          # control proportion
p1 <- 0.333333333333333                          # 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(2138214)
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 = 379 (80% power), 507 (90% power)
 75% of observed effect: n per arm = 659 (80% power), 881 (90% power)
 50% of observed effect: n per arm = 1,444 (80% power), 1,932 (90% power)
power.prop.test: 379 
Simulated power at n = 379 per arm: 0.801 (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. Analyses Reported by the Sponsor

The registry reports the sponsor’s analysis as a Fisher Exact test with p = 0.567. The posted descriptive values are 24.1% in Arm II and 33.3% in Arm I. The recalculation is a simple unadjusted comparison from those summary statistics; it gives a risk difference of 0.092 for Arm I minus Arm II and p = 0.436. The recalculation is supplementary and is not the sponsor’s reported test.

The registry does not flag a discrepancy between the posted summary and the sponsor’s reported analysis. The statistical details needed to reproduce the Fisher Exact result—such as the exact test specification and any analysis conventions—are not reported in the registry record. The p-values should therefore be presented as results from their respective analyses, not treated as interchangeable estimates of effect.

The sponsor’s p-value is evidence under the assumptions of its reported test, not a measure of effect size or probability that the groups are the same. The recalculated risk difference and interval provide a separate description of magnitude and uncertainty based on the posted counts and denominators. Neither the test result nor the recalculation alone establishes equivalence or resolves the range of underlying effects compatible with the data.

14. Multiplicity, Missing Data and Analysis Populations

The registry lists 13 outcome measures and 6 analyses, and it names several secondary outcomes, including transient and permanent hypocalcemia, voice and swallowing problems, quality of life, clinical recurrence, and use of natural language processing to extract quality-of-life measures from interview narratives. The record does not report whether or how multiplicity was addressed across outcomes, analyses, time points, or comparisons. It therefore cannot establish whether the reported p-values were adjusted, whether one outcome was designated as the sole confirmatory test, or how any family of hypotheses was defined.

Multiplicity matters in general because testing many questions can increase the chance of obtaining at least one result that appears unusual by chance. That concern does not imply that a particular result is invalid; it means interpretation should account for the number and organization of analyses. The registry’s endpoint-level test result should not be treated as if it were necessarily the only statistical test in the trial.

The endpoint population description says one participant in Arm II was a screen failure and that Arm III data were not collected for the primary outcome. It does not state whether other endpoint measurements were missing, how any missing values were handled, or whether the analysis used an intention-to-treat, per-protocol, or other population. The denominators reported for the comparison are 29 in Arm II and 30 in Arm I, but the registry record does not explain the full relationship between those denominators and the randomized population.

Details of exclusions after randomization, protocol deviations, imputation, sensitivity analyses, and the treatment of participants without a day 1 PTH measurement are not reported in the registry record. The record also does not provide a statistical analysis plan or say whether analyses were prespecified. These limitations prevent a full audit of selection and missing-data assumptions. They do not alter the posted descriptive values, but they constrain how confidently the analysis can be generalized to all randomized participants.

15. 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 5% reporting threshold.

Most frequent non-serious adverse event terms

The posted serious adverse event figures are zero affected participants in Arm I, zero in Arm II, and zero in Arm III. The registry lists 30 participants at risk in Arm I, 29 in Arm II, and 30 in Arm III for this safety summary. These are the reported figures; no additional rate calculation is needed to interpret the listed counts.

For other adverse events, the record reports 23 affected participants out of 30 at risk in Arm I, 25 out of 29 in Arm II, and 9 out of 30 in Arm III. These figures are separate from the serious adverse event counts. The registry does not provide event-term details in the serious-event list, which is empty, and it does not describe the severity criteria or the observation window for these adverse-event summaries.

Zero reported serious adverse events in these groups should be read as a description of the recorded safety data, not proof that serious events cannot occur or that the procedures have identical safety profiles. The groups are limited in size, and details about safety follow-up, ascertainment, and adverse-event definitions are not reported in the registry record. Arm III was not randomized with Arms I and II, so its safety figures should not be interpreted as a randomized comparison with either surgical arm.

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

The primary endpoint result describes the percentage of participants meeting a day 1 PTH threshold used to define transient hypoparathyroidism. The posted value was 24.1% for total thyroidectomy with CND and 33.3% for total thyroidectomy alone. The unadjusted recalculated risk difference, oriented as Arm I minus Arm II, was 0.092. The uncertainty interval spans values on both sides of zero, and the registry reports a Fisher Exact p-value of 0.567.

These findings do not precisely determine the direction or size of an underlying difference in this endpoint. The estimate is based on a small number of observed events, and the interval indicates substantial uncertainty. The sponsor-reported test does not provide evidence that can be translated into proof of no difference, just as the point estimate alone does not prove that the procedures differ.

The endpoint is one early postoperative biochemical measure. It is not a substitute for permanent hypoparathyroidism, other complications, patient-reported quality of life, recurrence, or a broad assessment of harms and benefits. Although those areas appear among the trial’s outcomes, the primary result does not answer them. The standard-of-care group is also outside the primary endpoint comparison because the registry says its data were not collected for this measure.

The result should be interpreted in the context of the trial design and record limitations. Arm I and Arm II were randomized, but details of the randomization process, who was masked, analysis populations, missing data, and the multiplicity plan are not reported in the registry record. The result therefore supports a cautious description of the observed endpoint data, rather than a definitive claim about comparative effectiveness or safety.

17. Why This Trial Matters Statistically

This trial raises a useful statistical question: how should a binary complication outcome be compared when the observed event counts are limited and the point estimate is uncertain? Reporting arm percentages alone can make the observed contrast appear more definitive than it is. The risk difference and its interval add essential context by expressing the contrast and showing that plausible values extend in both directions.

The record also illustrates the distinction between a sponsor-reported test and a simple recalculation from posted summary data. The Fisher Exact result is p = 0.567, while the unadjusted recalculation gives p = 0.436. Presenting these as separate analyses, rather than choosing one without explanation, helps readers distinguish the official reported test from a supplemental summary-statistic calculation. The recalculation is not a replacement for the trial’s own analysis.

Another statistical issue is the endpoint population. Randomization involved Arms I and II, while Arm III was a nonrandomized standard-of-care group and did not contribute data for this primary outcome. This makes it important to keep the target comparison clear and not combine groups that had different eligibility and treatment assignment pathways. The screen-failure note and the unreported analysis-set details further show why denominators need to be understood in relation to the endpoint.

Finally, the registry includes multiple outcomes and analyses without reporting how multiplicity was handled. A statistically careful interpretation distinguishes the primary endpoint from secondary questions, treats uncertainty as central, and avoids turning a nonsignificant test into a claim of equivalence. These considerations are broadly relevant when interpreting surgical trials with multiple outcomes and limited precision.

18. 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.
• Participant counts were reconstructed from posted percentages and arm sizes, so small rounding differences are possible.

The primary endpoint comparison includes Arms I and II only. Arm III data were not collected for this outcome, and that group consisted of patients who were not eligible for randomization into the two surgical arms. The endpoint population description also notes a screen failure in Arm II. The registry does not report a complete participant flow for this measure or define the analysis population in enough detail to determine how all randomized participants were handled.

The event counts were 7 in Arm II and 10 in Arm I, with denominators of 29 and 30. The resulting estimate is imprecise: its recalculated interval crosses zero. The registry-reported Fisher Exact p-value and the unadjusted recalculation are different results from different analyses. Details needed to reproduce the sponsor’s test are not reported in the registry record.

The record does not report the trial’s sample-size justification, original design assumptions, allocation concealment procedures, masking details, or a statistical analysis plan. It also does not describe missing-data methods, whether covariate adjustment was planned, or how multiplicity across the outcomes and analyses was handled. Confidence-interval construction details are not reported.

The endpoint is limited to a day 1 laboratory threshold. Measurement procedures, completeness of assessment, and the relationship between this measure and longer-term outcomes are not detailed in the registry record. Arm III’s care followed patient and physician preference, so it cannot be treated as a randomized comparator. The posted primary result therefore applies to the specified endpoint comparison and should not be generalized to all trial outcomes or interpreted as a complete assessment of the surgical approaches.

19. Related Tutorials

20. Related Statistical Calculators

21. Sources

• ClinicalTrials.gov: Central Neck Dissection in Patients With Clinical Node Negative Thyroid Cancer — NCT02138214.
• Registry record last updated 2020-12-21; status last verified 2020-02.
• Results first posted 2020-12-21.
• 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.

22. Record Summary

NCT02138214 is a completed phase II randomized, double-masked, parallel-group trial of total thyroidectomy alone versus total thyroidectomy with ipsilateral prophylactic central lymph node dissection for node-negative thyroid cancer. The record lists enrollment of 117 and conditions including Stage I, Stage II and Stage III papillary thyroid cancer.

The primary endpoint was the percentage with transient hypoparathyroidism, defined as day 1 serum PTH below 10 pg/ml. The posted values were 24.1% in Arm II and 33.3% in Arm I. The registry reports a Fisher Exact p-value of 0.567; an unadjusted recalculation gives a risk difference of 0.092 for Arm I minus Arm II, with an interval spanning both negative and positive values. The record does not report the original sample-size rationale, analysis populations, missing-data methods, or multiplicity approach.