Comparative Analysis of APACHE II and P-POSSUM: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Comparative Analysis of APACHE II and P-POSSUM (NCT02471612), a registered phase not stated study. The primary outcome is Area Under the Receiver Operating Curve (ROC) as a Measure of the Accuracy of the APACHE II and P-POSSUM Scoring Systems to Predict Mortality.

2. Clinical Question

The clinical question is comparative and prognostic: it concerns the ability of two risk scores to order or distinguish mortality outcomes, not whether either score changes outcomes. The brief summary states that the study compares APACHE II and P-POSSUM in emergency laparotomy. The registry also lists outcomes relating to length of stay, postoperative ventilator and inotropic support, cardiac morbidity, acute kidney injury, and re-exploration. Numerical results for these outcomes are not reported in the registry record. Because both scores were applied to the same patients, their assessment is not a randomized comparison of two treatment arms.

3. Trial Design

The registry describes a completed study at Tata Main Hospital involving patients undergoing emergency laparotomy. Enrollment is listed as 159. The study period is described as December 2013 to November 2014, and the registry lists the study as completed. The description says that both APACHE II and P-POSSUM scores were assigned on the day of surgery and that mortality was followed over a 30-day timeframe. The record does not report the numerical mortality results or the corresponding AUC estimates.

This is a comparison of two scoring systems applied within the same clinical cohort, not a randomized treatment trial. The registry explicitly indicates that allocation is not stated and that the groups are not a like-for-like control comparison. The scoring-system labels in the serious-event record should therefore not be interpreted as independent treatment groups. Both scores concern the same patients, and no randomized allocation between them is described.

The registry identifies the primary outcome as the AUC for each scoring system's prediction of mortality. It also lists secondary outcomes covering length of stay, postoperative ventilator support, postoperative inotropic support, cardiac morbidity defined as acute myocardial infarction or arrhythmias needing treatment, acute kidney injury, and re-exploration. The registry lists seven outcome measures overall, but numerical results for these listed outcomes are not reported in the registry record.

The registry does not report a detailed protocol, a sample-size justification, an analysis population, or a statistical analysis plan. It also does not report how participants with incomplete follow-up or missing score or outcome information were handled. The record's enrollment figure is 159, while the serious-event records list 157 participants at risk under each scoring-system label. The reasons for this difference are not reported in the registry record. This distinction matters when interpreting the event counts: the enrollment number and the event-record denominators describe different entries in the record and should not be treated as interchangeable.

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

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: All patients undergoing emergency laparotomy at Tata Main Hospital form December 2013 to November 2014 were scored with APACHE II \& P-POSSUM scoring systems on the day of surgery. The patients were followed up till at least 30 days after discharge or death (during admission or within 30 days after discharge).

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 Area Under the Receiver Operating Curve (ROC) as a Measure of the Accuracy of the APACHE II and P-POSSUM Scoring Systems to Predict Mortality, measured over 30 days.

The primary endpoint is the area under the receiver operating characteristic curve for APACHE II and P-POSSUM as predictors of mortality. The registry gives a 30-day timeframe and describes the AUC as a measure of the accuracy of prediction. In general, an ROC curve traces sensitivity against one minus specificity across possible thresholds. Its AUC summarizes how well a score separates participants with the outcome from those without it across those thresholds.

In general, an AUC is useful because it assesses discrimination across a range of score cut-offs rather than describing performance only at one chosen cut-off. An AUC near the chance-discrimination level indicates limited separation, while a larger value indicates stronger separation. The registry itself states that the AUC can range from 0.5 to 1.0, identifies 1.0 as perfect discriminatory ability, describes values above 0.8 as good, values from 0.60 to 0.80 as moderate, and values below 0.60 as poor. These are the registry's stated interpretive bands; they do not substitute for the study's actual estimates.

The record names cut-offs of >=24 for APACHE II and >=63 for P-POSSUM. A cut-off can support a classification at a particular score, but it does not by itself convey how well a system discriminates across all possible thresholds. The registry says the primary measure is AUC, yet does not report either AUC estimate or its 95% confidence interval. Consequently, the relative discrimination of the two scores and the precision of either estimate cannot be assessed from the registry record.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

What does the AUC measure?

In general, the area under the receiver operating characteristic curve is a measure of discrimination. An ROC curve considers how sensitivity and specificity change as a score threshold is moved. Sensitivity describes the proportion of participants with the outcome who are classified as positive at a threshold; specificity describes the proportion without the outcome who are classified as negative. The curve displays sensitivity against one minus specificity, as the registry description states. The AUC compresses this curve into a single summary of how well the score separates outcome and non-outcome observations.

The registry describes the AUC as ranging from 0.5 to 1.0, with 1.0 indicating perfect discriminatory ability. It also supplies interpretive bands: above 0.8 as good, from 0.60 to 0.80 as moderate, and below 0.60 as poor. Such bands are general interpretive aids, not a replacement for the actual estimate, confidence interval, or clinical context. An estimate alone also does not reveal whether predicted risks agree with observed risks; that separate property is calibration.

Why are the confidence intervals important?

The primary outcome is described as an AUC with a 95% confidence interval. In general, a confidence interval expresses the uncertainty associated with an estimate under the statistical method used. A narrower interval indicates greater precision than a wider interval, while an interval that spans a broad range of plausible values leaves greater uncertainty about performance. The registry record does not report the AUC estimates or their intervals, so neither the estimated discrimination nor its precision can be evaluated.

Confidence intervals also help prevent over-reading a point estimate. Two estimates may differ numerically while their uncertainty remains substantial. For a comparison between scores applied to the same participants, the uncertainty of the difference matters directly; separate intervals around each score do not, by themselves, provide a complete test of the difference. No paired comparison method or comparison interval is reported in the registry record.

What role do the stated cut-offs play?

The registry names a cut-off of >=24 for APACHE II and >=63 for P-POSSUM. In general, converting a continuous or ordered score into a positive-versus-negative classification allows performance to be described at a particular threshold. Sensitivity and specificity at that point can inform how often the threshold identifies outcome cases and how often it classifies non-cases correctly. However, a single threshold gives only one operating point on the ROC curve. It does not provide the full discrimination summary represented by the AUC.

The record does not report sensitivity, specificity, predictive values, the number of patients on either side of each threshold, or how the thresholds were chosen. The cut-offs therefore cannot be used to infer classification performance. The registry's AUC descriptions are not a substitute for those missing threshold-specific results.

Why does the paired design matter?

In general, when two prediction systems are evaluated in the same participants, each system's performance is measured against the same outcomes. The resulting performance estimates are related rather than independent. A method for comparing them should account for that pairing. The registry says both scores were assigned on the day of surgery, which supports interpreting this as a within-cohort comparison. It does not describe the statistical method used for an AUC comparison, nor does it report a test or interval for a difference between the scores.

The study is not described as randomized between scoring systems. A score is an assessment, not a treatment assignment, and the record explicitly notes that the listed groups are not a like-for-like control comparison. Thus, any comparison concerns predictive performance in the cohort, not a causal effect of one scoring system over another on mortality.

What can be inferred from the registry analysis information?

The registry lists one analysis, but the numerical output for the primary measure is not reported in the registry record. The record also does not state a statistical analysis plan, whether the AUC analysis was adjusted, or how score distributions and outcome status were handled. No model specification, covariate adjustment, or assessment of calibration is reported. These omissions limit reproducibility and make it impossible to assess whether the reported endpoint, if analyzed, was evaluated according to a prespecified method.

The record names a 30-day timeframe and describes following participants in relation to hospital stay, discharge, or death. It does not provide detailed follow-up completeness or a precise account of outcome ascertainment for every participant. In general, incomplete or inconsistent outcome ascertainment can affect estimates of discrimination, because the score-outcome comparisons depend on knowing which participants experienced the endpoint. The registry does not report its approach to missing outcome or score information.

How should planning calculations be understood?

In general, sample-size planning calculations describe how many participants might be needed under chosen assumptions to detect a specified effect with a chosen statistical approach. They are planning illustrations, not evidence that a study had a particular probability of finding a result after it was conducted. The registry's planning description refers to a two-sided significance level of 0.05, equal allocation, a normal approximation, and effects expressed as fractions of an observed effect. It also distinguishes illustrations expressed per arm from time-to-event illustrations expressed as total events.

Those assumptions do not make the study a randomized two-arm comparison. The registry describes both scoring systems being applied to the same participants, and it does not report the observed effect or numerical planning illustrations in the record available for interpretation. Planning figures should not be described as observed power or used to infer whether the study was adequately sized. The registry also does not report a sample-size justification for the original study design.

10. Multiplicity, Missing Data and Analysis Populations

The registry lists seven outcome measures overall. It identifies mortality prediction by AUC as the primary outcome and names secondary outcomes relating to length of stay, postoperative ventilator support, postoperative inotropic support, cardiac morbidity, acute kidney injury, and re-exploration. Numerical results for these outcomes are not reported in the registry record.

When a study evaluates several outcomes or makes several statistical comparisons, the chance of at least one apparently notable result can increase. In general, multiplicity methods or a prespecified hierarchy can help explain how several comparisons were interpreted. The registry does not report a multiplicity strategy, a hierarchy among secondary outcomes, or which comparisons were formally tested. It is not possible to determine from the record whether any adjustment was planned or used.

The registry does not report missing-data counts, missingness patterns, or methods for handling unavailable scores or outcomes. It also does not report a defined analysis population, such as whether all enrolled participants were included in each endpoint analysis or whether any exclusions occurred. These details matter because the people included in an analysis may affect estimates of discrimination and their uncertainty.

Enrollment is listed as 159, while the serious-event entries show 157 participants at risk under each scoring-system label. The reason for this difference, and whether it relates to the primary endpoint analysis, is not reported in the registry record. The record also does not provide a flow of participants through follow-up or explain whether all participants had complete mortality status within the stated timeframe.

11. 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 0% reporting threshold.

The serious-event record lists APACHE-2 Scoring and P-POSSUM, with 157 participants at risk under each label. It records zero participants affected in both the serious-event and other-event fields for each label. These are event-record figures, not evidence that one scoring system is safer than the other. Both scores were applied to the same patients, and the registry does not describe randomized treatment groups.

The deaths field is not numerically reported. A missing death count must not be read as zero deaths. The record also does not list specific serious adverse-event terms. The zero affected counts therefore describe only the event fields as entered; they do not establish that no adverse events or deaths occurred during the study.

The serious-event denominators of 157 differ from the enrollment figure of 159, and the reason is not reported in the registry record. The event figures should be interpreted within their stated denominators and should not be used to infer the primary mortality outcome or the comparative accuracy of APACHE II and P-POSSUM.

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

The registry record establishes that the study was designed to compare APACHE II and P-POSSUM for predicting mortality among patients undergoing emergency laparotomy. Both scores were recorded on the day of surgery, and the primary measure was the AUC over a 30-day timeframe. It also identifies thresholds for each score and lists several secondary outcomes. These details clarify the prediction question, but they do not reveal how well either score performed.

The record does not report the AUC estimates or 95% confidence intervals. It therefore does not support a conclusion about which system discriminated better, whether either estimate was precise, or whether any apparent difference was statistically compatible with no difference. The numerical result of a direct comparison is also not reported. The listed score thresholds cannot fill this gap because threshold-specific sensitivity and specificity are not reported.

Even if AUC results were available, they would address discrimination rather than every aspect of prediction. In general, discrimination concerns separation of outcome cases and non-cases. Calibration concerns agreement between predicted risks and observed outcomes. The registry does not report calibration measures, nor does it provide data to assess clinical consequences at the stated cut-offs. Predictive values and classification counts are also not reported.

This study is a comparison of prediction systems within a cohort, not a comparison of treatments. The registry does not describe randomized assignment to scoring systems, and both systems were applied to the same participants. The analysis cannot be interpreted as showing that using one score causes a change in mortality. Nor do the serious-event entries establish that no deaths occurred: the death count is not numerically reported there.

The appropriate conclusion is limited to the information documented: the study question, population, scoring systems, and primary endpoint are stated, but the outcome estimates needed to judge predictive performance are not reported in the registry record. Completion status and enrollment do not substitute for those missing statistical results.

13. Why This Trial Matters Statistically

Evaluating prognostic scores is important because clinical researchers need to know whether a score can distinguish participants with different outcome risks in the setting where it is used. A comparison of APACHE II and P-POSSUM in emergency laparotomy addresses this type of predictive-performance question. The registry specifies mortality as the primary outcome and AUC as the measure, giving readers a clear indication of the intended statistical target.

The design feature that both systems were applied to the same patients is statistically relevant. In general, evaluating scores within one cohort can make their relative performance directly comparable, while requiring the analysis to account for the paired outcome data. The registry does not report the method used for that comparison. Recognizing the paired structure helps prevent treating the scoring-system labels as separate randomized groups.

The registry also demonstrates why a stated endpoint is not enough to interpret a study's findings. AUC values and confidence intervals are needed to judge discrimination and precision; the record does not report them. Threshold-specific results and calibration measures would answer additional questions, but are also not reported. The stated cut-offs alone do not establish how accurately participants were classified.

The study's record lists enrollment of 159 and completed status, as well as secondary outcomes that capture other aspects of postoperative care and morbidity. These details describe scope, but no numerical results for the primary or secondary outcomes are reported. For statistical interpretation, transparency about estimates, uncertainty, analysis populations, and missing data is essential alongside the design description.

14. 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 main limitation for statistical interpretation is that the registry record does not report the primary AUC estimates or their 95% confidence intervals. Without these results, discrimination and precision cannot be evaluated, and the comparison between scoring systems cannot be quantified. The record also does not report a statistical test or interval for the difference between their AUCs.

The registry does not report sensitivity, specificity, predictive values, classification counts at the stated cut-offs, or calibration results. It gives cut-offs of >=24 for APACHE II and >=63 for P-POSSUM, but does not explain their selection. The reported endpoint and cut-offs therefore do not provide enough information to assess performance at a particular threshold or agreement between predicted and observed risks.

The record does not report a detailed analysis plan, sample-size justification, analysis population, handling of missing data, or completeness of follow-up. It describes a mortality timeframe of 30 days but does not provide participant-level follow-up information. These omissions make it difficult to assess how the endpoint was ascertained and which participants contributed to an analysis.

Enrollment is listed as 159, whereas the serious-event entries list 157 at risk under each score label; the reason for this difference is not reported. The record does not provide numerical mortality counts in those event entries. In addition, the scoring systems were applied to the same cohort rather than assigned as randomized arms, so the comparison is about prediction performance and not a causal treatment contrast.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Comparative Analysis of APACHE II and P-POSSUM — NCT02471612.
• Registry record last updated 2019-10-15; status last verified 2016-05.
• Results first posted 2016-06-27.
• 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.

18. Record Summary

The registry describes a completed study at Tata Main Hospital comparing APACHE II and P-POSSUM for prediction of mortality after emergency laparotomy. Enrollment is listed as 159, and the study period is described as December 2013 to November 2014. Both scoring systems were recorded on the day of surgery. The primary endpoint is AUC for mortality over a 30-day timeframe, with a 95% confidence interval specified as the dispersion measure.

The record lists cut-offs of >=24 for APACHE II and >=63 for P-POSSUM, and six secondary outcomes concerning length of stay, postoperative support, cardiac morbidity, acute kidney injury, and re-exploration. The numerical primary and secondary outcome results are not reported in the registry record. Serious-event entries list 157 at risk and zero affected in the recorded event fields for each score label; deaths are not numerically reported. The registry does not describe randomized allocation, a detailed analysis plan, missing-data handling, or the reason event denominators differ from enrollment.