Chronic Low Back Pain Research Project: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Chronic Low Back Pain Research Project (NCT00108550), a randomized phase 2 study. The primary outcome is Transformed Descriptor Differential Scale-Pain Intensity Scores Adjusted for Time.

2. Clinical Question

The registry record describes a Phase 2 study of gabapentin for chronic low back pain. Its central question was whether pain-intensity scores over the treatment period differed between participants assigned to gabapentin and those assigned to placebo. The primary endpoint was a transformed measure of self-reported current pain intensity, adjusted for time across the scheduled assessments.

This is a comparison of average scores over repeated measurements, not simply a comparison of pain ratings at one final visit. The registry record says the single reported values represent adjusted means over all time points. Its primary result is therefore interpreted as an average treatment-group contrast over the measurement period. That framing matters: it describes the fitted comparison across the recorded assessments, while the posted result does not give a separate treatment contrast for each assessment time.

The outcome description identifies higher scores as indicating higher pain intensity. Its underlying self-report measure used a 20 point scale, with adjectival anchors ranging from faint to strong. Before analysis, scores were transformed to a continuous 0-1.5 scale. The clinical question consequently concerns differences in this transformed measure, rather than a direct difference on the original 20 point scale.

3. Trial Design

The registry record classifies the study as a randomized, parallel-group Phase 2 trial with triple masking. It was sponsored by the US Department of Veterans Affairs and is recorded as completed. The recorded start was 2004-10, and primary completion was 2009-07. Enrollment is listed as 108. The arm-level outcome records show 53 participants for placebo and 55 for gabapentin; the record describes the primary analysis population as all randomized participants receiving study drug.

One group received gabapentin orally three times daily for 12 weeks. The listed dose started at 300 mg per dose and could be increased to a maximum of 1200 mg per dose. The comparator group received inert placebo capsules identical in size and shape to the experimental capsules, also taken orally three times daily for 12 weeks. The registry record does not report further details about how the dose was increased or how adherence was assessed.

For the primary outcome, pain intensity was assessed at baseline and through Week 12, with interim measurements at Weeks 1, 2, 3, 4, 5, 7 and 9. The endpoint description says the reported values are adjusted means across all time points. The record also lists a disability measure, the Roland and Morris Disability Index, adjusted for time. The primary outcome and the disability outcome are continuous measures, but they assess different aspects of the condition: reported pain intensity and reported disability, respectively.

The registry record does not report the allocation sequence method, the parties included in triple masking, the exact eligibility criteria, or the procedures used to maintain masking. It also does not report the planned sample-size justification or a detailed protocol for dose modification. Those omissions limit how fully the trial procedures can be reconstructed from the record alone. The randomized, parallel design establishes the broad structure of the comparison; it does not, by itself, supply all the details needed to assess implementation or adherence.

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

Analysis population for the primary outcome, as defined in the registry: Analysis of all randomized participants receiving study drug

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 Transformed Descriptor Differential Scale-Pain Intensity Scores Adjusted for Time, measured over Baseline to Week 12 with Interim Measurement at Weeks 1, 2, 3, 4, 5, 7 and 9.

The primary endpoint is the Transformed Descriptor Differential Scale-Pain Intensity score adjusted for time. It captures participants’ self-reports of current chronic back pain intensity. The original rating task used a 20 point scale and 12 adjectival descriptor anchors, asking participants to rate pain as more or less intense relative to words such as faint, moderate and strong. Higher scores indicated higher pain intensity.

Before analysis, the measure underwent an order-preserving mean-matching variance-stabilizing transformation, placing the score on a continuous 0-1.5 scale. This transformation changes the scale on which the reported averages are expressed; it does not make the transformed values directly interchangeable with ratings on the original 20 point scale. The registry record does not provide a simple clinical interpretation for a particular difference on the transformed scale.

The assessment schedule included baseline and repeated measurements through Week 12, with interim measurements at Weeks 1, 2, 3, 4, 5, 7 and 9. The posted endpoint values are described as adjusted means of transformed pain intensity over all time points. Using repeated observations can characterize the average pattern across follow-up rather than relying only on a single final measurement. It can also make the interpretation depend on the statistical model used to combine observations over time.

The endpoint is relevant to the trial’s stated analgesic question because it measures reported pain intensity directly. It remains a patient-reported scale, however, and the transformed scale’s units are not themselves a direct description of how participants felt. The primary result should therefore be read as a between-group statistical contrast in transformed scores, with the confidence interval indicating the uncertainty around that contrast.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry reports a mixed models analysis for the primary continuous outcome. In general, a mixed model is used when observations are related within a participant, as repeated measurements over time usually are. The model can represent both the average pattern across participants and the correlation among measurements from the same participant. Treating every visit as if it came from an unrelated person would ignore that within-participant relationship and could misstate uncertainty.

In this trial, the primary endpoint description says the reported values are adjusted means over all time points. The analysis therefore concerns an average over repeated measurements rather than a single visit alone. The registry record does not state the precise fixed effects, random effects, covariance structure, time parameterization, or adjustment variables. Those details are important for reproducing a repeated-measures analysis and are not reported in the registry record.

A mean difference expresses the estimated contrast between the two groups on the outcome scale used for analysis. Here, that scale is the transformed 0-1.5 pain-intensity scale. A positive estimate, given the group order in the registry comparison, places the gabapentin mean above the placebo mean; because higher scores indicate higher pain intensity, the direction is toward higher reported intensity in the gabapentin group. The numerical estimate is small relative to the scale range, but no clinical-importance threshold for this transformed scale is reported in the registry record.

A confidence interval describes the range of effect estimates reasonably compatible with the data and the model, under the procedure used to calculate it. Its width conveys precision: a narrower interval indicates more precise estimation than a wider interval. The primary interval runs from -0.044 to 0.105 and crosses zero. That means the posted estimate does not isolate a direction of effect with confidence at the conventional zero-difference reference. It should not be read as a probability that the true effect falls within the interval, nor as proof that the treatment effects are equal.

The p-value addresses compatibility with a null hypothesis of no difference under the model and its assumptions. It is not the probability that the null hypothesis is true, and it does not measure the size or clinical importance of an effect. The registry reports p = 0.424 for the primary analysis. Interpreting that value alongside the estimate and confidence interval gives a fuller picture than treating it as a stand-alone pass-or-fail decision. The record does not report a prespecified decision threshold or an adjustment for testing multiple outcomes.

The endpoint was transformed before analysis. In general, transformations can help make statistical modeling more suitable when the original data have features such as skewness or unequal variance. An order-preserving transformation retains the ranking of observations while changing the distances between values. A variance-stabilizing transformation is intended to make variability more consistent across the measurement range. The registry describes the transformation as order-preserving, mean-matching and variance-stabilizing, but does not report its full formula or diagnostic assessments. As a result, the transformed-scale values are not directly interpreted as raw-scale points.

Repeated-measure models can make use of observations across follow-up, but their interpretation depends on how the model represents time and within-person correlation. General statistical practice also considers how missing observations are handled. For example, likelihood-based analyses can use available observations under assumptions about the missingness process; they do not automatically resolve bias if missingness depends on unobserved outcomes. The registry record does not report the amount or pattern of missing primary-outcome data, the method for handling missing observations, or the assumptions used. Those matters therefore cannot be evaluated from the posted analysis description.

The analysis population is stated as all randomized participants receiving study drug. This is a specific description, not enough detail to determine whether every randomized participant was included in every endpoint model or how participants without usable outcome measurements were treated. The record does not report a separate intention-to-treat analysis or a per-protocol analysis. It also does not report whether any protocol deviations led to exclusions.

The posted sample-size illustrations are planning calculations based on the observed effect for the secondary disability comparison, not the trial’s original design justification. They apply a two-sided significance level of 0.05, equal allocation and a normal approximation, and describe participants per arm for 80% or 90% power at stated fractions of that observed effect. For planning at 100% of the observed effect, the illustrations are 784 participants per arm for 80% power and 1,049 for 90% power. At 75% of the observed effect, they are 1,393 and 1,865 participants per arm, respectively. At 50% of the observed effect, they are 3,134 and 4,195 participants per arm, respectively. These are planning illustrations, not observed power for the completed trial. The registry record does not report the trial’s own sample-size justification, so the illustrations should not be interpreted as evidence about the original design assumptions.

In general, smaller effects require more information to estimate precisely than larger effects, all else being equal. The planning figures illustrate how required sample size can increase when planning for a smaller fraction of an observed effect. They do not establish that the observed effect is reproducible or that the assumed effect is clinically important. Sample-size calculations depend on the target effect and analytic assumptions; they do not compensate for uncertainty in an effect estimate taken from a single study.

The registry also reports an adjusted comparison for the secondary disability outcome. Its sponsor-reported analysis and the simple recalculation are close but not identical. The registry’s analysis takes precedence for describing the sponsor’s result; the recalculation is a simple, unadjusted check using posted summary statistics and does not replace the fitted model. More generally, recalculations from group summaries may not reproduce a model that uses repeated observations, covariate adjustment, or a particular missing-data approach. A discrepancy, if present, would require explicit attention; the record indicates no discrepancy for the primary comparison.

10. Analyses Reported by the Sponsor

The sponsor’s primary analysis is the mixed models comparison reported in the registry: a mean difference of 0.030, a 95% confidence interval from -0.044 to 0.105, and p = 0.424. The posted primary results do not include a separate recalculated comparison. The registry indicates no discrepancy for this endpoint, so the sponsor’s reported analysis is the relevant result to interpret.

The secondary disability outcome has both a sponsor-reported mixed models result and a simple recalculation from posted summary statistics. The sponsor reported a mean difference of -.5343, with a 95% confidence interval from -1.940 to 0.872 and p = 0.458. The recalculation is an unadjusted comparison based on the posted group summaries; it is not the same procedure as the reported mixed model. Such a recalculation can be useful as a check on direction and approximate scale, but it cannot reproduce every feature of an analysis based on repeated measurements or other model specifications.

For interpretation, the sponsor’s own analysis takes precedence over a summary-statistic recalculation. The registry does not indicate a discrepancy for the primary result, and the secondary estimates are broadly similar in direction and uncertainty. Neither analysis should be detached from its confidence interval, outcome scale, analysis population, or the information not reported in the registry record about model specification and missing observations.

11. Multiplicity, Missing Data and Analysis Populations

Multiplicity arises when several outcomes, time points, contrasts, or analyses are examined. In general, each additional statistical test creates another opportunity for a result to appear notable by chance when no underlying effect is present. Common approaches include prespecifying a primary outcome, controlling an overall error rate, or treating secondary analyses as exploratory. The registry identifies a primary endpoint and one posted secondary endpoint, but does not report a multiplicity adjustment, testing hierarchy, or detailed prespecified statistical analysis plan. It is therefore not possible to determine from the registry record whether the reported p-values were intended as confirmatory tests or how any family of tests was handled.

The primary endpoint itself combines repeated measurements over time. The record identifies the assessment schedule and says the reported values are adjusted means over all time points, but does not describe whether individual visit contrasts were also tested or how they would have been handled. It does not report interim analyses or stopping rules. Those design details are not reported in the registry record and should not be inferred from the completed-trial status.

Missing data are another important consideration for repeated-measure outcomes. The registry record does not report the number of missing observations, reasons for missingness, or their distribution between groups. It also does not explain the method used to address missing outcome data or the assumptions underlying that method. In general, a mixed model may use available repeated measurements, but the validity of an analysis depends on assumptions about why observations are missing. The record does not contain enough detail to assess whether those assumptions were plausible or whether sensitivity analyses were conducted.

The analysis population is described as all randomized participants receiving study drug. The record does not clarify how that population was implemented for each model, whether participants who stopped treatment continued outcome assessments, or whether any participants were excluded from an analysis. It does not report separate results for intention-to-treat or per-protocol populations. These omissions limit assessment of how the analysis population and missing observations might affect the findings.

12. Safety Results

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

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 registry reports zero serious adverse events among 53 placebo participants at risk and zero among 55 gabapentin participants at risk. It also reports other, non-serious adverse events for 21 placebo participants and 42 gabapentin participants. These are participant counts as recorded in the registry; they should not be converted into percentages or interpreted as rates without additional information about the event definitions, observation period, and analysis method.

The record’s serious adverse event fields show no serious events in either group, and no top serious-event terms are listed. Death counts are not reported in the registry record. The entry for other adverse events indicates that such events were recorded in both groups, with more participants listed in the gabapentin group. These figures alone do not describe the severity, duration, recurrence, relationship to study treatment, or clinical importance of each event.

Safety counts from a trial should be interpreted alongside the way events were collected and classified, the duration of follow-up, and the completeness of reporting. Those details are not reported in the registry record. The posted counts provide a limited description of recorded events; they do not establish comparative safety or prove that serious harms could not occur.

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

The primary result is an estimate of the average difference in transformed pain-intensity scores across the scheduled measurements. Its confidence interval includes zero and spans values in both directions. The posted data therefore do not provide a precise directional estimate of the average contrast. The p-value likewise does not establish that the treatment groups are identical, and the result should not be translated into a claim of equivalence.

The endpoint was transformed before analysis. It is measured on a continuous 0-1.5 scale, while the underlying self-report task began on a 20 point scale. The transformation and time adjustment mean that the reported comparison is not a simple difference in raw ratings at a single visit. The record does not provide a threshold for a clinically important difference on the transformed scale. Consequently, statistical uncertainty cannot be converted into a confident judgment about clinical importance from the posted information alone.

The result applies to the trial’s recorded comparison: gabapentin versus inert placebo in a randomized, parallel-group study of chronic low back pain, with the primary analysis population described as randomized participants receiving study drug. The record does not give enough detail to assess how missing outcomes, adherence, or specific eligibility criteria affected the estimate. It also does not establish how broadly the findings apply beyond the participants and conditions represented in this study.

The primary finding should not be described as proof that gabapentin does or does not work. It is a statistical result with uncertainty, based on a particular transformed endpoint and model. The secondary disability outcome provides additional context but does not resolve the primary uncertainty, and the registry does not report enough information to assess multiplicity fully. The serious and other adverse-event counts are similarly limited descriptions, not a complete assessment of safety.

A careful conclusion is that the registry reports no clear, precise between-group difference in the primary transformed pain-intensity outcome, while the uncertainty interval permits differences in either direction. More detailed protocol, analysis, missing-data and clinical-interpretation information would be needed to make stronger claims about the magnitude, importance or generalizability of the result.

14. Why This Trial Matters Statistically

This trial illustrates why an estimated effect, its confidence interval, the outcome scale, and the analysis method should be read together. A single p-value cannot show the magnitude or precision of a treatment contrast. Here, the primary interval crosses zero, so the posted analysis does not pin down a direction of average difference. At the same time, that uncertainty is not evidence that all potentially relevant differences have been ruled out.

The endpoint also demonstrates how measurement choices shape interpretation. The original pain ratings were transformed to a continuous scale and summarized across repeated observations. That approach may be suitable for the intended analysis, but it means the treatment contrast needs to be understood on the transformed scale and in the context of the repeated-measures model. A numeric difference in that scale is not automatically a direct statement about patients’ experience.

The registry’s secondary outcome illustrates a further statistical issue: separate outcomes can add context without providing independent confirmation. The primary and secondary comparisons both retain uncertainty, and the record does not report whether multiplicity was addressed. The distinction between a sponsor’s model-based analysis and an unadjusted calculation from group summaries also underscores that analyses answering apparently similar questions can differ in what information they use.

Finally, the record contains important limits on interpretation. It does not report the detailed analysis plan, missing-data handling, sample-size justification, or a clinical-importance threshold for the transformed endpoint. These omissions do not invalidate the posted results, but they constrain the strength of conclusions that can be drawn from them. Recognizing those limits is part of interpreting a clinical trial statistically rather than reducing it to a binary significance judgment.

15. 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 provide a full protocol or statistical analysis plan. It does not report the precise mixed-model specification, covariance structure, adjustment variables, or diagnostic checks. Those details matter for understanding how repeated pain scores were combined and how the estimated uncertainty was calculated. The record also does not report the transformation formula, making it difficult to translate differences on the transformed scale back to the original rating task.

The handling of missing outcomes is not described. The record does not give the number or pattern of missing measurements, reasons for missingness, or sensitivity analyses. Since the primary outcome was measured repeatedly, missing observations and the assumptions used to analyze them can affect the interpretation of the fitted average contrast. The stated analysis population does not by itself resolve these questions.

The registry does not report a clinical-importance threshold for the transformed pain scale or a rationale for the planned enrollment. It is therefore not possible to assess the observed precision against a prespecified clinically meaningful difference or to reconstruct the trial’s original power assumptions. The sample-size illustrations based on the secondary observed effect are planning examples, not a substitute for the original design justification.

The record also does not report a multiplicity strategy, interim analyses, stopping rules, adherence information, or detailed eligibility criteria. These omissions limit assessment of the trial’s broader conduct and inferential context. Although the study is recorded as triple-masked, the record does not specify who was masked or how masking was maintained. The adverse-event listing provides counts but not detailed event descriptions, severity, timing, attribution or deaths. These limitations support cautious interpretation of both efficacy and safety findings.

16. Related Tutorials

17. Related Statistical Calculators

18. Sources

• ClinicalTrials.gov: Chronic Low Back Pain Research Project — NCT00108550.
• Registry record last updated 2014-02-10; status last verified 2013-12.
• Results first posted 2014-02-10.
• 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.

19. Record Summary

The registry record describes a completed, randomized, triple-masked, parallel-group Phase 2 trial sponsored by the US Department of Veterans Affairs. It compared oral gabapentin with matching inert placebo over 12 weeks in chronic low back pain. Enrollment is listed as 108, and the primary analysis population is described as randomized participants receiving study drug.

The primary endpoint was transformed descriptor-based pain intensity, adjusted for time across baseline and repeated measurements through Week 12. The registry reports a mixed models comparison with a confidence interval that crosses zero. One secondary outcome, the Roland and Morris Disability Index adjusted for time, is also posted. The record does not report a detailed analysis plan, missing-data methods, multiplicity strategy, or clinical-importance threshold for the transformed primary measure.

Reported adverse-event counts include no serious events in either arm, while other adverse events were recorded in both groups. These counts are not a complete account of event characteristics. Overall, the posted results describe uncertain average treatment-group contrasts and should be interpreted within the limits of the registry’s methodological detail.