This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record. Numerical trial results on this page are restricted to the results posted in the ClinicalTrials.gov record; calculations added by Clinical Biostats are labelled as such.
1. Trial at a Glance
This page covers the registry results for Fracture Risk With Thiazolidinediones (NCT01055223), a registered phase not stated study. The primary outcome is Number of Low Impact Fractures in Males and Females After 6 Months of Exposure to TZD.
| Feature | Registry value |
|---|---|
| ClinicalTrials.gov identifier | NCT01055223 |
| Phase | Phase not stated |
| Condition | Fractures, Bone, Type II Diabetes |
| Allocation | Not Stated |
| Design model | Not Stated |
| Masking | Not Stated |
| Enrollment | 98483 |
| Arms in the primary outcome | n/a |
| Lead sponsor | GlaxoSmithKline |
| Results posted | Yes |
2. Clinical Question
Population
People with type 2 diabetes, aged 18-65 years, exposed to a thiazolidinedione (TZD). Eligibility required an ICD-9 code for type 2 diabetes and at least 6 months of exposure to rosiglitazone, pioglitazone, or troglitazone during follow-up available in the database.
Intervention
Exposure to TZD plus spironolactone, or exposure to TZD plus amiloride. The record describes spironolactone and amiloride as diuretics being assessed in combination with TZD.
Comparator
People exposed to TZD only. The study description identifies TZD-only exposure as the comparison for both combinations.
Primary question
Among people with type 2 diabetes exposed to TZD, is exposure to spironolactone or amiloride associated with the odds of low-impact fracture compared with TZD exposure alone? The registry labels the primary outcome as the number of low-impact fractures in males and females after 6 months of TZD exposure.
The research question concerns an association between medication exposure and recorded fracture diagnoses, not an assigned treatment effect. The registry describes the study as nested case-control and says controls were selected from TZD-exposed patients without a fracture diagnosis during follow-up. It also describes comparisons involving fractures at specific sites, including the hand, foot, upper arm, wrist, and hip. The primary outcome is broader: any low-impact fracture as defined through the listed ICD-9 diagnostic codes. The registry record does not report estimates of fracture odds or other comparative outcome results, so it does not establish the direction or precision of any association.
3. Trial Design
The registry describes a completed nested case-control study among people with type 2 diabetes exposed to TZD. Although the record uses clinical-trial registry conventions, the described design is observational: exposure groups are defined by medication use, and allocation is not stated. The record itself cautions that the arms are not a like-for-like control comparison. The enrollment listed in the record is 98483; the number included in each analysis is not reported in the registry record.
The study population was people aged 18-65 years with an ICD-9 code for type 2 diabetes and at least 6 months of TZD exposure during database follow-up. The database period named for the primary outcome runs from January 1, 1997 through December 31, 2008. The primary outcome description defines cases as incident low-impact fractures after the earliest date of a first TZD prescription. For each case, up to four controls were randomly selected from TZD-exposed patients without a fracture diagnosis during follow-up. Controls were matched on age (+ 5 years), gender, and year of fracture diagnosis.
The listed exposure groups are TZD only, TZD plus spironolactone, and TZD plus amiloride. The TZD-only group could include rosiglitazone, pioglitazone, or troglitazone. The record lists a primary measure for low-impact fractures and secondary measures for fractures at specified sites and follow-up descriptions. Start is recorded as 2009-05, primary completion as 2010-03, and status as COMPLETED. Details of the database construction, exposure classification rules, and analysis plan 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.
Overall Study
| Stage | TZD 6-month Cohort (Including TZD 12-month Cohort) |
|---|---|
| STARTED | 98483 |
| COMPLETED | 98483 |
| NOT COMPLETED | 0 |
Analysis population for the primary outcome, as defined in the registry: Type 2 diabetes patients 18-65 years old exposed to TZD. To be eligible for the study, a subject must have had at least one ICD-9 code for type 2 diabetes and have at least 6 months of exposure to TZD (RSG, PIO, or troglitazone) during their follow-up time available in the database.
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.
| Characteristic | Unit | TZD 6-month Cohort (Including TZD 12-month Cohort) |
|---|---|---|
| Participants analysed | 98483 | |
| Age, Continuous | Years | 52.2 (SD 8.33) |
| Sex: Female, Male: Female | Participants | 37738 |
| Sex: Female, Male: Male | Participants | 60745 |
6. Primary Endpoint
The registry lists the primary outcome as Number of Low Impact Fractures in Males and Females After 6 Months of Exposure to TZD, measured over From the earliest date of first TZD prescription to the fracture diagnosis date (Cases) or the end of follow-up in the database (Controls) between January 1, 1997 and December 31, 2008.
| Element | Registry entry |
|---|---|
| Result type | Number |
| Measure of spread | not applicable |
| Unit | fractures |
| Endpoint classification | not reduced to a single comparison |
The primary endpoint is titled “Number of Low Impact Fractures in Males and Females After 6 Months of Exposure to TZD.” It concerns fractures identified from ICD-9 diagnostic codes in Uniform Billing-92 records and Health Care Finance Administration records. The codes listed in the registry record are 805-807.4x, 808-810.xx, and 812-829.xx. A case is defined as an incident fracture with a listed ICD-9 code occurring after the study period begins, which is the earliest date of the first TZD prescription.
The endpoint is used to capture clinically recorded low-impact fractures within the population and to define cases for the nested case-control comparison. Controls were selected from TZD-exposed patients without a fracture diagnosis during follow-up. The timeframe description extends from the earliest date of the first TZD prescription to the fracture diagnosis date for cases, or the end of database follow-up for controls, between January 1, 1997 and December 31, 2008.
Although the endpoint title refers to a number of fractures, the registry describes a case-control design and states that the study aims to compare odds. The number of fractures, case counts, and any odds estimates are not reported in the registry record. Details about how repeat diagnoses or multiple fractures in one person were handled are also not reported in the registry record.
7. Why No Re-analysis Is Shown
8. Statistical Interpretation
The registry record defines a clinically focused association question: whether fracture odds differ among TZD-exposed people with additional exposure to spironolactone or amiloride compared with people exposed to TZD only. It describes how cases and matched controls were selected, but it does not report comparative estimates for the primary outcome or the listed secondary outcomes. In particular, the registry record does not report case counts, odds ratios, confidence intervals, or outcome-specific p-values. The direction, magnitude, and precision of any association therefore cannot be assessed from the record.
The listed enrollment of 98483 should not be read as the number of cases or the number analysed. It is the registry enrollment figure; the analysed sample size and group-specific counts are not reported in the registry record. Nor does the completed status supply evidence about the observed association. Completion describes the study status recorded in the registry, not the direction or statistical certainty of a result.
Matching is described on age (+ 5 years), gender, and year of fracture diagnosis, with up to four controls selected for each case. Matching can help make cases and controls more comparable on the named characteristics, but it does not by itself establish comparability on other features or remove all possible confounding. The record does not describe an adjusted analysis, how matching was accounted for in the statistical model, or how exposure to spironolactone and amiloride was handled over time. Those analysis details are not reported in the registry record.
The study is observational, and allocation is not stated. Thus medication exposure was not described as randomly assigned. Any contrast would be an association within the studied database population, not, by itself, proof that one exposure caused a change in fracture risk. The registry record also notes that the arms are not a like-for-like control comparison, a caution relevant to interpretation of any group contrast.
The record describes secondary outcomes for hand, foot, upper arm, wrist, and hip fractures, with sex-specific and combined descriptions and follow-up intervals. It lists 6 outcome measures and 36 analyses, but does not report the estimates or the analysis details corresponding to them. Without those results and methods, readers cannot determine whether findings were consistent across fracture sites, how precise any estimate was, or how the many analyses were interpreted. No effect or safety conclusion can be drawn from the registry record alone.
9. Statistical Methods Explained
A nested case-control design identifies people with an outcome of interest within a defined population and selects controls from that same population. Here, the outcome is a recorded low-impact fracture among people with type 2 diabetes exposed to TZD. The registry says that controls were selected from TZD-exposed patients without a fracture diagnosis during follow-up. This approach can be useful when the source population is large, because detailed comparisons can focus on cases and a selected set of controls rather than requiring every person to be analysed in the same way. The registry record does not report the database sampling procedures beyond its description of case and control selection.
What matching contributes
Matching selects controls with characteristics similar to those of cases on specified variables. The registry names age (+ 5 years), gender, and year of fracture diagnosis as matching characteristics and allows up to four randomly selected controls per case. Matching can improve comparability on those characteristics, but it does not ensure that cases and controls are similar on unmeasured or unmatched factors. A suitable analysis generally needs to account for the matched structure; the specific method used for this study is not reported in the registry record. It would therefore be inappropriate to attribute a particular regression model or adjustment strategy to the study.
What an odds comparison means
Odds describe the relative frequency of an outcome in relation to its absence. In a case-control design, the sampled case and control counts are determined by the selection process, so the raw fraction of sampled participants who are cases is not automatically an estimate of risk in the source population. A comparison of odds can describe the association between the exposure groups and being a case, subject to the design and analysis assumptions. The registry says that the study aims to compare odds, but the actual odds estimates are not reported in the registry record.
Precision and uncertainty
A confidence interval is a range of values that reflects uncertainty around an estimate under the assumptions of the statistical method. A narrower interval indicates greater precision than a wider interval, while an interval that includes the null value may be compatible with a range of possibilities that includes no association. These are general statistical interpretations, not findings from this study. The registry record does not report confidence intervals or any other uncertainty estimates for the primary or secondary outcomes, so precision cannot be evaluated.
Potential confounding
In observational comparisons, differences in exposure groups can reflect characteristics related to both medication use and the outcome. This is confounding. Matching addresses only the characteristics used in matching, and statistical adjustment can address additional measured characteristics if they are collected and appropriately modelled. The registry record does not report which potential confounders were measured, whether adjustment was performed, or how medication exposure was defined over follow-up. Those omissions limit what can be inferred about the comparability of the exposure groups.
Outcome classification
The registry defines fractures using specified ICD-9 codes captured from Uniform Billing-92 and Health Care Finance Administration records. Use of administrative diagnostic codes provides a stated operational definition, but the record does not report validation of those codes against clinical records or describe how coding errors were handled. Misclassification occurs when an outcome or exposure is assigned incorrectly. In general, such errors can affect estimates, and the direction of that effect depends on how classification errors relate to exposure and outcome. No misclassification assessment is reported in the registry record.
Multiplicity and model assumptions
When researchers assess several outcomes or conduct multiple analyses, the chance of observing an apparently unusual result somewhere among them can increase. Appropriate interpretation depends on the analysis plan, the relationships among outcomes, and whether findings were designated as primary or exploratory. The record lists 6 outcome measures and 36 analyses, but the statistical plan and any multiplicity approach are not reported in the registry record. It also does not report model diagnostics, handling of missing values, or sensitivity analyses. These are general methodological considerations; no particular procedure can be attributed to this study.
10. Multiplicity, Missing Data and Analysis Populations
The registry record lists 6 outcome measures and 36 analyses. It names a primary low-impact fracture outcome and secondary outcomes concerning fractures of the hand, foot, upper arm, wrist, and hip. The registry descriptions include sex-specific outcomes and follow-up intervals. Conducting multiple comparisons raises a general interpretive issue: the more outcomes or analyses examined, the greater the opportunity for apparently notable findings to arise by chance. The record does not report whether a multiplicity adjustment was planned or used, how the primary outcome was distinguished in the analysis, or whether secondary analyses were designated as exploratory.
Missing-data methods are not reported in the registry record. This includes how missing exposure, outcome, or matching information was handled, whether exclusions occurred because information was incomplete, and whether the amount or pattern of missingness was assessed. In general, missingness can affect estimates when the availability of information is related to exposure or outcome; its consequences depend on the data and the analysis.
The primary outcome description specifies the source records and diagnostic codes, but does not report a validation procedure or rules for resolving conflicting records. The analysis population, numbers of cases and controls, and group-specific participant counts are not reported in the registry record. The record also does not describe whether all eligible participants contributed to each outcome analysis or how follow-up ended for individuals. Without these details, readers cannot assess attrition, data completeness, or the extent to which the analysed groups represent the eligible population.
Allocation is not stated, and no statistical analysis plan is reported in the registry record. A planned approach to missing data, multiplicity, and analysis populations therefore cannot be inferred from the listed outcomes alone.
11. What the Result Does, and Does Not, Mean
The registry record describes a study designed to examine whether fracture diagnoses were associated with exposure to spironolactone or amiloride alongside TZD, compared with TZD exposure alone. It provides the population criteria, the broad case and control definitions, matching characteristics, and diagnostic-code description. These details clarify the research question and how the investigators intended to identify cases and controls.
It does not report the primary or secondary outcome estimates. The record therefore does not establish whether fracture odds were higher, lower, or similar across the described exposure groups, nor how precisely any difference was estimated. Enrollment of 98483 is not a substitute for the number of cases or the number included in each analysis. The registry record also does not report confidence intervals, p-values, or the analysis methods needed to interpret comparative estimates.
Even if an association estimate were available, an observational nested case-control comparison would not by itself establish that medication exposure caused a change in fracture occurrence. Matching addresses only the named characteristics, and the record does not report adjustment for other potential confounders. Exposure classification, missing data, and the handling of multiple outcome analyses are also not described in sufficient detail to assess their influence.
The study's completed status indicates its registry status, not a conclusion about the clinical question. The listed adverse-event fields do not supply comparative fracture results or an estimate of treatment effect. The appropriate interpretation of the posted record is consequently limited to the study design and intended comparisons; the outcome findings and their uncertainty are not reported in the registry record.
12. Why This Trial Matters Statistically
This study raises a useful statistical distinction between a registry's detailed description of a research question and the evidence needed to answer that question. The record specifies an observational nested case-control design, a population with defined TZD exposure, a case definition based on diagnostic codes, and matched control selection. Those details make the intended comparison understandable. However, without outcome estimates and uncertainty measures, the association itself cannot be evaluated.
The design also highlights why matched case-control data require careful analysis. The matching factors—age (+ 5 years), gender, and year of fracture diagnosis—shape how cases and controls were selected. Analysis that ignores the matched structure can fail to reflect the design, while matching alone does not control differences in other characteristics. The specific statistical treatment of matching and confounding is not reported in the registry record.
Multiplicity is another important issue. The record lists 6 outcome measures and 36 analyses, spanning fracture types, sex descriptions, and follow-up intervals. When many comparisons are made, readers need to know which outcomes were primary, whether analyses were prespecified, and how uncertainty across comparisons was handled. Those details are not reported in the registry record.
Finally, the record illustrates that completion, enrollment, and outcome evidence are distinct. A completed study status and a listed enrollment do not show what the outcome estimates were or how precise they were. For clinical researchers and biostatisticians, the useful lesson is to distinguish design information from results, and to avoid drawing conclusions about an association when estimates, intervals, and analysis details are unavailable.
13. 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 outcome information in the registry record is limited to definitions and planned comparisons; primary and secondary results are not reported. Consequently, the magnitude and precision of the association cannot be assessed. Although enrollment is listed as 98483, the record does not report the numbers of cases and controls, the number analysed, or counts by exposure group.
The study is observational, and allocation is not stated. The record warns that the arms are not a like-for-like control comparison. It describes matching on age (+ 5 years), gender, and year of fracture diagnosis, but does not report whether other potential confounders were measured or adjusted for. The statistical method used to account for matching is not reported in the registry record.
The primary outcome relies on ICD-9 diagnoses from named administrative record sources. The record does not report validation of the diagnostic codes, adjudication procedures, or how repeated fracture diagnoses were treated. Exposure definitions and changes in medication use during follow-up are not described in detail. The registry also does not report how missing data, incomplete follow-up, or exclusions were handled.
Finally, the listed 6 outcome measures and 36 analyses create a need for clear prespecification and multiplicity handling, neither of which is described. The statistical analysis plan, sensitivity analyses, and model assumptions are not reported in the registry record. These limitations concern the information available in the record and should not be read as proof that the study did or did not address them outside it.
14. Related Tutorials
15. Related Statistical Calculators
16. Sources
- ClinicalTrials.gov: Fracture Risk With Thiazolidinediones — NCT01055223.
- Registry record last updated 2017-05-30; status last verified 2017-05.
- Results first posted 2011-04-18.
- 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.
Continue through the Clinical Biostats statistical pathway
Use this trial's endpoint as a starting point for deeper study of effect measures, confidence intervals, and sample-size planning.
17. Record Summary
Fracture Risk With Thiazolidinediones (NCT01055223) is recorded as a completed nested case-control study sponsored by GlaxoSmithKline. The listed enrollment is 98483. The study concerns people with type 2 diabetes aged 18-65 years who had at least 6 months of exposure to TZD. Its exposure groups are TZD only, TZD plus spironolactone, and TZD plus amiloride.
The primary outcome is the number of low-impact fractures in males and females after 6 months of TZD exposure, identified using specified ICD-9 codes in administrative records. Controls were selected from TZD-exposed patients without a fracture diagnosis and matched to cases on age (+ 5 years), gender, and year of fracture diagnosis. The registry lists secondary fracture outcomes for specified sites and follow-up descriptions.
The record lists start as 2009-05 and primary completion as 2010-03. Comparative outcome estimates and the statistical analysis plan are not reported in the registry record.