TPN-Induced Hyperglycemia: Impact on Clinical Outcome in Intensive Care Unit (ICU) and Non-ICU Patients: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for TPN-Induced Hyperglycemia: Impact on Clinical Outcome in Intensive Care Unit (ICU) and Non-ICU Patients (NCT00604669), a registered phase not stated study. The primary outcome is Mortality.

2. Clinical Question

The study asked whether hyperglycemia during total parenteral nutrition (TPN) was associated with worse clinical outcomes among hospitalized patients receiving TPN. The registry summary describes a retrospective chart review at Grady Memorial Hospital, covering patients treated with TPN from 1/01/06 to 12/31/06. The stated hypothesis was that patients who developed hyperglycemia would have greater morbidity and mortality than TPN patients with euglycemia.

The question is associational: it concerns whether recorded hyperglycemia and outcomes occurred together in this cohort. A retrospective comparison cannot by itself determine whether hyperglycemia caused an outcome, or whether other patient factors contributed to both. The registry record does not report a formal causal analysis, subgroup results for ICU and non-ICU patients, or numerical results for the stated comparisons.

3. Trial Design

Despite the word “trial” in the study context, the registry describes an observational, retrospective chart review rather than a randomized clinical trial. The investigators planned to review records for patients treated with TPN at Grady Memorial Hospital from 1/01/06 to 12/31/06. The registry lists an enrollment of 170 and a status of completed. The study dates are recorded as 2007-06 for the start and 2009-08 for primary completion.

The exposure was hyperglycemia during TPN, and the summary describes a comparison with TPN patients who remained euglycemic. The exposure was not assigned by investigators, and allocation is not reported in the registry record. The groups therefore should not be treated as randomized, like-for-like arms: their characteristics and clinical circumstances may differ before or during TPN. The record does not describe how hyperglycemia or euglycemia was defined, how patients were selected, or how the review handled incomplete records.

The stated outcomes of interest included infections, systemic sepsis, cardiac complications, acute renal failure, length of stay, and mortality. Mortality is the listed primary outcome. The registry description says it is assessed at the end of the chart review of all patients, but does not provide a more precise outcome window or state how mortality was ascertained. The record does not report whether outcomes were evaluated separately for ICU and non-ICU patients, although those groups appear in the study title.

The study was intended to inform the formulation of a prospective randomized trial on management of TPN-associated hyperglycemia. That stated future aim is distinct from the design of this chart review. No protocol for a subsequent trial is 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: Nothing to disclose.

Overall Study

Reasons for not completing this period:

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 Mortality, measured over at the end of the chart review of all patients.

The primary endpoint is listed as mortality, described as the mortality rate among patients with hyperglycemia while receiving TPN during the period 1/01/06 to 12/31/06. The timeframe is described as the end of the chart review of all patients. This endpoint captures whether death was recorded in the defined patient group, but the registry record does not specify the precise follow-up interval, the source used to confirm deaths, or the denominator used to calculate a rate.

Mortality is used as a clinical outcome because it represents death during the period of interest and can be assessed from patient records. In general, mortality comparisons between exposure groups can describe whether recorded deaths differ between those groups. They do not, on their own, establish that the exposure caused any difference. For an observational study, differences may reflect underlying illness, care circumstances, or other factors that also relate to hyperglycemia and mortality.

The registry lists the endpoint parameter as a number, while its unit is described as a mortality rate. It does not post a count, rate, denominator, comparison estimate, confidence interval, or p-value. The registry record also does not state whether mortality was analyzed as an unadjusted comparison or with adjustment for other characteristics. The other outcomes mentioned in the study summary are not presented as the primary endpoint.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry identifies a retrospective chart review and a mortality endpoint, but provides little detail about the statistical analysis. The concepts below explain how such analyses are generally interpreted; they do not describe methods confirmed for this study unless explicitly stated in the registry record.

What does an observational comparison estimate?

In general, an observational comparison describes how outcomes differ between groups formed by exposure status rather than by random assignment. Here, the stated exposure groups are TPN patients with hyperglycemia and TPN patients with euglycemia. The resulting comparison can show an association in the reviewed records. Because exposure was not assigned at random, group differences may also reflect factors related to patient condition or care. A causal interpretation requires suitable design and analysis assumptions, and those assumptions and analyses are not reported in the registry record.

How should mortality be summarized?

Mortality can be presented as a count of deaths, a proportion over a defined interval, or a rate using a specified time-at-risk denominator. These measures answer related but distinct questions. A count gives the number of recorded events; a proportion relates events to the number of people followed; a rate incorporates the time people contributed under observation. The registry labels the primary measure a mortality rate and its parameter as a number, but the calculation, denominator, and follow-up interval are not reported in the registry record. No mortality value is posted, so the measure cannot be independently interpreted.

What does a confidence interval convey?

As a general statistical concept, a confidence interval expresses uncertainty around an estimate under a specified analysis procedure. It helps readers assess how precisely an association has been estimated and which effect sizes remain compatible with the data. It does not ensure that the study is free of bias, and it does not resolve confounding in a nonrandomized comparison. No confidence interval is reported in the registry record.

What does a p-value convey?

In general, a p-value measures how unusual data at least as incompatible with a null hypothesis would be if that null model and its assumptions were correct. It does not measure the probability that the null hypothesis is true, nor does it quantify the importance of an effect. A p-value is most informative alongside an effect estimate and its uncertainty. Neither a p-value nor a numerical effect estimate is reported in the registry record.

Why can adjustment matter?

In general, regression or other adjustment methods can account for measured characteristics that differ between exposure groups and are relevant to the outcome. Adjustment does not automatically remove bias: it depends on the variables measured, the model, and its assumptions. The registry record does not describe adjusted analyses, covariates, model choice, or diagnostics. It is therefore not possible to tell whether the stated comparison was adjusted or how any adjustment would affect interpretation.

What does over-dispersion mean for count data?

As a general concept, count outcomes can vary more across observations than a simple statistical model assumes. This extra variability, called over-dispersion, can make standard errors too small if ignored, yielding intervals that are too narrow and tests that appear more conclusive than warranted. Modeling choices should reflect the outcome’s distribution and the study design. The registry record does not report a count-data model or any assessment of over-dispersion; no such analysis should be inferred from the endpoint description.

How should planning calculations be read?

Any sample-size figures accompanying the posted material are planning illustrations rather than observed results or observed power. The stated assumptions describe a two-sided significance level of 0.05, equal allocation between two arms, a normal approximation, and an effect defined as a stated fraction of the observed effect. Such calculations are conditional on their assumptions and are not evidence that the study had a particular probability of finding an effect. The registry record does not report a sample-size justification for the chart review or a prospective randomized study, and no specific planning count is reported in the registry record.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists one primary outcome measure, mortality, and mentions other outcomes of interest: infections, systemic sepsis, cardiac complications, acute renal failure, and length of stay. The record does not report whether these additional outcomes were formally analyzed, whether analyses were prespecified, or whether the investigators adjusted for multiple comparisons. When many outcomes or subgroup comparisons are examined, the chance of finding at least one apparently unusual result can increase. In general, multiplicity procedures can address that issue, but their choice depends on the analysis plan and goals. No such procedure is reported in the registry record.

Missing-data handling is also not described. For a chart review, records or specific clinical measures may be incomplete; the registry does not report whether any data were missing, how missingness was assessed, or whether cases were excluded or otherwise handled. Without these details, readers cannot determine whether the available records represent the full enrolled cohort or whether missing information could affect the comparisons.

The registry record does not identify the analysis population, such as whether all enrolled patients were included in each outcome analysis or whether eligibility for a particular analysis depended on available measurements. It also does not describe exclusion criteria, exposure classification procedures, or the handling of patients whose glycemic status was unclear. These details matter because decisions about inclusion and classification can change which observations contribute to an estimate.

The record reports no statistical analysis plan, model, covariates, subgroup plan, or interim analysis. Their use or absence is not reported in the registry record. These gaps, together with the absence of posted outcome estimates, limit assessment of multiplicity, missingness, and the population on which any result would be based.

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

The record documents a completed retrospective chart review of patients treated with TPN and states a research hypothesis about hyperglycemia, morbidity, and mortality. It identifies mortality as the primary outcome and lists other clinical outcomes of interest. It does not provide numerical findings for these outcomes. Consequently, the registry record cannot establish the direction, magnitude, or precision of any association between hyperglycemia and mortality or other outcomes.

The study question is observational. In general, a difference between patients with hyperglycemia and those with euglycemia would describe a difference between groups as observed in the medical records. It would not, without additional design and analysis support, demonstrate that hyperglycemia produced the outcome difference. Differences in patient characteristics, illness, or care could be related to both exposure and outcome. The registry record does not describe how such potential confounding was addressed.

The stated hypothesis is not a result. The completed status and enrollment of 170 do not reveal whether the hypothesis was supported, and the intended use of findings to formulate a prospective randomized study is not evidence about either the chart-review results or a later trial. No separate ICU and non-ICU estimates are reported in the registry record.

In general, an estimate paired with an uncertainty interval can help show both the observed association and the precision with which it was measured. Those results, as well as group sizes and outcome counts, are not reported in the registry record. The available information therefore supports a description of the study’s purpose, setting, and broad design, but not a conclusion about the observed mortality rate or the comparative outcomes.

12. Why This Trial Matters Statistically

This study is statistically interesting because it asks about outcomes in a setting where exposure status is observed during care rather than assigned by investigators. That distinction changes what a comparison can support. Random allocation can, under appropriate conditions, help balance measured and unmeasured characteristics between groups; retrospective exposure groups do not have that protection. Statistical adjustment may address measured differences, but it cannot guarantee removal of confounding, and the registry does not report whether adjustment was performed.

The question also highlights the importance of defining an endpoint precisely. “Mortality rate” requires a clear account of the outcome, denominator, and observation period to interpret a value. The record names mortality and gives a broad timeframe but does not report the denominator or detailed ascertainment. Such definitions are necessary for readers to understand what a reported estimate represents and whether it is comparable across groups.

The study summary lists several additional outcomes alongside mortality. If multiple outcomes are analyzed, interpretation depends on knowing which were primary, how the analyses were planned, and how uncertainty from multiple comparisons was handled. The registry lists one primary outcome but does not describe analysis planning or multiplicity methods. This makes transparent reporting especially important.

Finally, the registry describes the chart review as groundwork for formulating a prospective randomized study. That aim illustrates how observational work can motivate future research questions, while remaining distinct from evidence produced by a randomized comparison. The record’s lack of numerical results means it cannot show what the chart review found, but its design and reporting gaps clarify which details are needed to interpret an eventual estimate responsibly.

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 registry description leaves several trial-specific limitations unresolved. It does not state how hyperglycemia or euglycemia was defined, what glucose measurements were used, or how exposure was classified over the TPN period. Without those definitions, it is difficult to assess whether patients were assigned consistently to the stated observational groups or whether classification could vary across records.

The chart-review setting and dates are identified, but the record does not describe eligibility criteria, how potentially eligible patients were located, or whether all relevant records were available. It lists enrollment as 170 but does not provide the number of patients in each exposure group, baseline characteristics, or the number contributing to the mortality analysis. The possibility and treatment of incomplete records are not reported in the registry record.

The registry title names ICU and non-ICU patients, yet it does not report how those settings were defined, whether setting was captured at a particular point in care, or whether outcomes were analyzed separately by setting. The record also does not provide details on the duration of follow-up, mortality ascertainment, or how length of stay and the other listed outcomes were defined.

There is no posted statistical analysis plan, effect estimate, confidence interval, p-value, or description of adjustment for potential confounding. This prevents assessment of the strength and precision of the comparison, as well as of the analysis choices. The registry also does not report whether the chart review included all patients with TPN or a subset selected through another procedure. These limitations concern the information available in the registry record; they do not establish how the underlying investigators conducted work that is not described there.

14. Related Tutorials

15. Related Statistical Calculators

16. Sources

• ClinicalTrials.gov: TPN-Induced Hyperglycemia: Impact on Clinical Outcome in Intensive Care Unit (ICU) and Non-ICU Patients — NCT00604669.
• Registry record last updated 2013-12-13; status last verified 2013-11.
• Results first posted 2010-04-07.
• 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.

17. Record Summary

The registry identifies NCT00604669 as a completed study sponsored by Emory University. Its title concerns TPN-associated hyperglycemia and clinical outcomes in ICU and non-ICU patients. The description specifies a retrospective chart review at Grady Memorial Hospital of patients treated with TPN from 1/01/06 to 12/31/06. Enrollment is listed as 170; the study start is 2007-06 and primary completion is 2009-08.

Mortality is the sole listed primary outcome measure, described as the mortality rate among patients with hyperglycemia while on TPN. The summary also names infections, systemic sepsis, cardiac complications, acute renal failure, and length of stay as outcomes of interest, and states a hypothesis comparing hyperglycemic and euglycemic TPN patients. Numerical results, participant characteristics, outcome estimates, uncertainty measures, and analysis details are not reported in the registry record. The stated longer-term purpose was to inform formulation of a prospective randomized study.