Epidemiology and Outcome Following Post Surgical Admission to Critical Care

Electronic Supplementary Material

Epidemiology and Outcome Following Post Surgical Admission to Critical Care.

Rhodes A1, Moreno RP2, Metnitz B3, Hochrieser, H3. Bauer P3, Metnitz P4

1. Intensive Care Medicine

St Georges Healthcare NHS Trust and St Georges University of London

S Georges Hospital

London SW17 0QT

United Kingdom

E:

2. Unidade de Cuidados Intensivos Polivalente,

Hospital de St António dos Capuchos,

Centro Hospitalar de Lisboa Central, E.P.E.

1169-050 Lisbon, Portugal

E:

3.Section for Medical Statistics,

Center for Medical Statistics, Informatics, and Intelligent Systems

Medical University of Vienna,

Vienna, Austria

4.Dept. of Anesthesiology and General Intensive Care

Medical University of Vienna,

Vienna, Austria

E:

Corresponding Author:

Professor Philipp Metnitz

Dept. of Anesthesiology and General Intensive Care

Medical University of Vienna,

Vienna, Austria

T: +43 1 40400 6861

F: +43 1 95 22 843

E:

Methods

Database

This study used the database of the Austrian Center of Documentation and Quality Assurance in Intensive Care medicine (ASDI) ( The prospectively collected data of consecutive patients includes socio-demographic descriptors, existence of chronic conditions, reasons for admission to intensive care according to a predefined list of medical and surgical diagnoses, severity of illness as evaluated by the SAPS II model [1]computed using the worst physiologic values during the first 24 hours in the ICU (and corrected for inaccuracies in the evaluation of the Glasgow Coma score), therapeutic interventions as defined by the TISS 28 descriptors [2], ICU and hospital length of stay and vital status at ICU and hospital discharge.

Data quality

Database quality is prospectively assessed by evaluation for completeness of the documentation and for reliability. The SAPS II score is re-calculated in randomly selected patients for each ICU in order to assess the inter-rater quality control. Kappa coefficients and intra-class correlation coefficients are calculated for the re-scored data. Availability of the variables necessary to calculate the SAPS II was used as an indicator for the completeness of the data. The quality of the recorded data was categorized as excellent with respect to both completeness of records and inter-rater variability. The median number of missing parameters necessary for the calculation of the SAPS II score was 0 (0–2). Inter-rater quality control indicated an overall excellent grade of agreement. For all tested variables, there were minimal deviations between observers detected with the contribution to the variability being less than 1%

Statistical Analysis

Logistic multi-level regression analysis was used to determine whether or not hospital mortality was changing over time, as depicted by the calendar year for all surgical patients.The first step was to identify factors that were independently related to hospital mortality of surgical patients from multivariate analysis. The following factors were entered into the model based on their univariate relationship to outcome: SAPS II score, main diagnosis on admission and co-morbidities (as defined by a pre-defined list of co-morbidities which are not contained in the SAPS II-score). The reference classes used for each of these factors were elective surgery, ‘other’ types of surgery that are not described by any of the main diagnostic or surgical sites and the absence of the described co morbidities, respectively. The model was then repeated with the same variables but excluding both age and surgical admission type (elective / emergency) from the SAPS II score and entering them as separate and distinct variables. The goodness of fit of each of these models was then compared by means of the Brier score [3].

The second stage was to then evaluate the effects of time. This was performed by constructing a subsequent model using the same inference factors as above, but including calendar year as a co-variable to investigate a potential trend over time. To quantify the estimates in single calendar years the time variables were entered as eleven distinct yearly categories with the year 1988 as the reference. The individual ICU was entered into each analysis as a random factor in order to adjust for the clustering effect with the consequent breaking of the assumption that all observations are independent of one another applying the general estimation equation (GEE) approach. In order to finally understand whether a changing ICU length of stay of survivors over time was impacting on survival a further model was constructed against length of stay using the same factors as found to be relevant in the above analyses. Due to the multiplicity of tests performed and in order to avoid spurious associations and over-fitting only p values of less than 0.001 were considered as significant in order to allow for a more robust and consistent result [4].

References

1.Le Gall JR, Lemeshow S, Saulnier F, (1993) A new Simplified Acute Physiology Score (SAPS II) based on a European/North American multicenter study. JAMA 270: 2957-2963

2.Miranda DR, de Rijk A, Schaufeli W, (1996) Simplified Therapeutic Intervention Scoring System: the TISS-28 items--results from a multicenter study. Crit Care Med 24: 64-73

3.Brier G, (1950) Verification of forecasts expressed in terms of probability. Monthly weather review 78: 1-3

4.Bauer P, Potscher B, Hackl P, (1988) Model selection by multiple test procedures. Statistics 19: 39-44

.

ESM Table 1. Description of cohort.

ESM Table 2. Demographics and outcome data for surgical patients stratified by surgical category. The p-values refer to the comparison of all 7 surgical categories (not all categories shown) by Kruskal-Wallis or Chi-Square-test

ESM Table 3. Epidemiology and outcomes of surgery admitted to Intensive Care from 1998 to 2009; p: p-values from the Kruskal-Wallis or Chi-Squ.-test

Figure Legend

ESM Figure 1.

Trends in hospital mortality over time split by gender (black dots are females). Graphs on the left are for elective procedures and on the right are emergency.

ESM Figure 2

Graph of adjusted odds ratios as compared to the reference year 1998 with their corresponding 95% confidence intervals for non-surgical patients, scheduled and unscheduled surgery by calendar year.

ESM Figure 1

Figure 2.