Secondary Logo

Journal Logo

Original Article

Serum creatinine and perioperative outcome - a matched-pairs approach using computerised anaesthesia records

Dehne, M. G.; Junger, A.; Hartmann, B.; Quinzio, L.; Röhrig, R.; Benson, M.; Hempelmann, G.

Author Information
European Journal of Anaesthesiology: February 2005 - Volume 22 - Issue 2 - p 89-95
doi: 10.1017/S0265021505000177


Little is known about the prevalence of cardiovascular events and mortality in individuals with renal failure. Serum creatinine concentration is widely used as a marker for renal failure despite the existence of more sensitive markers for renal damage such as proteinuria and creatinine clearance. Depending on the laboratory methods used, a serum creatinine level higher than 1.3-1.5 mg dL−1 has been arbitrarily chosen as a measure of renal failure in a normal patient population. The HOPE study [1,2] has shown a significantly increased incidence of cardiovascular events and mortality in patients with incipient renal failure. In that study a direct association between the incidence of cardiovascular events and the level of serum creatinine was also found. Similar associations between renal and cardiovascular morbidity have been reported in previous studies after vascular [3,4] or cardiac [5,6] surgery.

One of the major requirements of an Anaesthesia Information Management System is a comprehensive documentation of the anaesthetic procedure as well as the incorporation of additional information from other hospital information systems. It should provide sufficient data for scientific purposes as well as for medical and administrative purposes in order to justify its high cost. Such a system may be especially useful to analyse events with a low incidence because the population to be studied has to be very large. Using the resources of our department's computerised anaesthesia record keeping system, we designed a retrospective matched case-control study in order to assess the possible effects of increased preoperative creatinine (>1.3 mg dL−1) on the incidence of intraoperative cardiovascular events and hospital mortality in patients undergoing non-cardiac surgery.

Materials and methods

In 1995 an Anaesthesia Information Management System (NarkoData, IMESO GmbH, Hüttenberg, Germany) [7,8] was installed in our tertiary care teaching hospital. It was upgraded in 1997 to version 4. With this client-server system running under Microsoft™ Windows NT™ (Microsoft Corporation, Redmond, Washington, USA), anaesthesia procedures have been routinely recorded at more than 125 anaesthesia work stations including operating rooms, pre-anaesthesia care units and administration areas.

The attending anaesthesiologist records all patient-related data during the preoperative ward round (biometrical data, history, pre-existing diseases, examination results and additional investigations) and obtains informed consent from the patient. These data are entered into the electronic anaesthesia record on the preceding day for an elective procedure and immediately before the operation in the case of an emergency procedure. Each patient's characteristics data and preoperative laboratory results are imported directly from the central hospital information system. Vital signs, monitor readings and ventilatory data are imported via RS232 interfaces. Systolic, diastolic and mean arterial pressure (SAP, DAP and MAP, respectively), as well as heart rate (HR) are recorded online at a minimum of 5 min intervals if using a non-invasive technique or every 3 min if invasive monitoring is in use. The record keeping system collects all data relevant to anaesthesia during the procedure including administered drugs, laboratory results, vital signs and the data set for quality assurance according to the German Society of Anesthesiology and Intensive Care Medicine (DGAI) [9]. Any drugs administered are entered manually at the moment of administration. One important aspect of the system is that data is recorded in the form of preconfigured menu fields rather than as free text.

The programme runs as a local application on workstation computers. Data backups are made on the local hard disk and on a network server. At the end of anaesthesia, a completed protocol file is subsequently imported into an Oracle™ database (version 7, Oracle Corporation, Redwood Shores, California, USA). The database is designed according to the principles of a relational data model and consists of 163 tables: 90 tables with 450 attributes for time dependent data and 73 tables with 230 attributes for configuration and data description. For example, a drug entry is confined to three dependent tables of which the major one contains the name, drug characteristic flags (e.g. antibiotic, opioid, etc.) and links to the application form and default value tables. With this relational construction, similar configurations can be easily defined.

All mandatory fields (e.g. age, gender, height, weight and ASA status) must be entered before the anaesthetic record can be closed and stored and this is undertaken automatically. Data are also checked automatically for plausibility, allowing events to be documented only in a logical manner. Only mandatory data have been used for this study.

Data extraction

Fifty-eight-thousand-four-hundred-and-fifty-eight patients having undergone non-cardiac surgery recorded with the computerised anaesthesia record keeping system over a period of 4 yr since 1999 were studied. Patients under the age of 18 yr were excluded.

Preoperative creatinine values measured by the central hospital laboratory within 30 days before surgery were entered into the record keeping system database. Hospital mortality was determined using data imported from the central hospital information system. Structured query language (SQL) was used to retrospectively select intraoperative cardiovascular events including hypotension, hypertension, bradycardia and tachycardia, from the database according to the definition of the German Society of Anesthesiology and Intensive Care Medicine (DGAI) [9]. Relevant cardiovascular events were defined as follows:

  • Hypotension: A decrease in MAP >30% within a 10 min interval and administration of a vasoconstrictor or a positive inotropic drug (epinephrine, norepinephrine, dopamine, dobutamine, dopexamine, amezinium metilsulfate (Supratonin®; Grünenthal GmbH, Aachen, Germany), cafedrine/theodrenaline (Akrinor®; AWD.pharma GmbH & Co. KG, Dresden, Germany), enoximone, milrinone) within 20 min of the beginning of the decrease. Additional volume administration was not considered.
  • Hypertension: An increase in MAP >30% within a 10 min interval and administration of an anti-hypertensive drug (nifedipine, urapidile, clonidine, hydralazine, droperidol, glyceryl trinitrate and sodium nitroprusside) within 20 min of the beginning of the increase.
  • Bradycardia: HR <50 beats min−1 for at least 5 min and intravenous (i.v.) drug administration to increase HR (atropine, orciprenaline, ipratropium bromide and epinephrine) or pacing within 15 min of the beginning of the bradycardia.
  • Tachycardia: HR >100 beats min−1 for at least 5 min and i.v. drug administration to decrease HR (beta blocker, calcium channel blocker, cardiac glycoside, sodium channel blocker and potassium channel blocker), cardioversion or defibrillation within 15 min of the beginning of the tachycardia.

A randomly selected sample of 20% of the anaesthetic records containing at least one cardiovascular event was reviewed independently by two investigators to verify the automatically detected cardiovascular events and to eliminate possible artefacts. Neither the number, duration nor type of events were analysed separately.

The method of matched pairs was used for an evaluation of the impact of increased preoperative creatinine on hospital mortality and incidence of intraoperative cardiovascular events. Cases were defined as patients with a preoperative creatinine >1.3 mg dL−1. In our laboratory this is considered abnormal based on the laboratory's population distribution to include 95% of normal subjects. Matched controls were automatically selected from among all patients over the study period according to matching variables. A patient in the matched control group was required to have a creatinine level of ≤1.3 mg dL−1. Only patients (cases and controls) who had at least one preoperative creatinine value documented within 30 days before surgery were included. According to our normal hospital surgical practice, a battery of routine investigations is undertaken on every in-patient and emergency case and this includes serum creatinine.

The matching criteria included ASA physical status [10], high-risk surgery (intracranial, thoracic, abdominal and major vascular surgery; following the pattern of the revised cardiac risk index of Lee and colleagues [11] for classification of high-risk surgery), urgency of surgery (elective; urgent - surgery within 6 h after admission; emergency - surgery within 2 h after admission), age and gender. The selection of the matched controls was performed in a stepwise manner, first attempting to match on ASA physical status, then type of surgery, then urgency of surgery, then age and finally gender. Only one control was matched to each case.

The aim was to examine whether an increased preoperative creatinine level is associated with hospital mortality and incidences of intraoperative cardiovascular events in non-cardiac surgery. Crude mortality ratio was calculated by the ratio of the hospital mortality rate of patients in the case group divided by the mortality rate in matched controls.

Statistical analysis

For statistical evaluation, data were exported from the database into the SPSS® statistics program (SPSS Software GmbH, Munich, Germany). Either Χ2-test or the Fisher's exact test for independent samples were used to detect statistically significant differences between the overall groups of cases and matched control patients in outcome variables. Power was determined using the software GPOWER Version 2.0 [12].

Since the patients in the case group and the matched control group were not randomly assigned with respect to risk of an increased preoperative creatinine >1.3 mg dL−1, we developed logistic regression models using the entire method to predict the impact of an increased creatinine on hospital mortality and incidences of intraoperative cardiovascular events. All matching criteria as well as serum creatinine >1/3 mg dL−1 were included in both models. These variables were analysed as categories using dummy variables for ASA physical status, high-risk surgery, urgency of surgery, gender and increased creatinine. Age was handled as a continuous variable. Level of significance was set at P < 0.01.


Three-thousand-and-twenty-eight patients (5.2%) were found to have a creatinine >1.3 mg dL−1. Matching was successful for 54.5% of the cases, leading to 1649 cases and 1649 controls. The case patients had a mean creatinine (mean ± standard deviation) of 3.3 ± 2.2 mg dL−1 whereas the matched control patients had a mean creatinine of 1.0 ± 0.2 mg dL−1. The mean age of the case patients was 61.8 ± 15.3 yr, and of the matched control patients 61.6 ± 15.4 yr. The case patients had a mean body mass index (BMI) of 27.1 ± 18.5 kg m−2 whereas the matched control patients had a mean BMI of 27.0 ± 20.8 kg m−2. The distribution of the matching variables in cases and controls is shown in Table 1.

Table 1
Table 1:
Matching variables in the cases and matched controls.

The matched control patients had a crude mortality rate of 0.9% (n = 15) vs. 2.2% (n = 36) for the case patients with an increased creatinine (P = 0.003). The crude mortality ratio of cases to controls was 2.75. Intraoperative cardiovascular events were detected from the database in 30.1% of the case patients (n = 496) and in 28.3% of the matched control patients (n = 466) (P = 0.25, power = 0.46). In the samples of records checked manually to determine the accuracy of automatic detection, none of the cardiovascular events were artefacts.

We detected an association between ASA physical status, high-risk surgery and older age and the occurrence of cardiovascular events. However, preoperative increased creatinine was not associated with a higher incidence of intraoperative cardiovascular events (Table 2).

Table 2
Table 2:
Results of the logistic regression models using the outcome measures as dependent variables.


There are currently few publications reporting clinical studies using data collected using an anaesthesia information management system. Such a system should make it possible to rapidly evaluate data to address medical, administrative and scientific queries and this should distinguish such systems from manual recording systems [13,14]. This has already been demonstrated by other authors using automatic systems for data evaluation [14-18]. In our study we have assessed the suitability of automatically generated data for analysing the attributable effects of increased preoperative serum creatinine on perioperative outcome (hospital mortality and intraoperative cardiovascular events) in patients undergoing non-cardiac surgery.

Our retrospective study is based on extensive data gathered during routine clinical work. The low incidence (5.2% of the studied patients) of a raised serum creatinine (>1.3 mg dL−1) is comparable to the distribution of serum creatinine levels in the US population [19]. This confirms that the patients in our study represent a non-selected group. Matching was successful in over 50% of a large patient population (more than 58 000) enabling us to determine significant differences. This is an important benefit of the approach. The clinical relevance of possible significant findings in the statistical analysis of an extremely large number of patients should always be kept in mind. It may also be possible to use these data for internal quality assurance programmes in the future.

All data was subjected to integrity checks resulting in a higher and relatively uniform quality of data in comparison to a traditional manual method of record keeping [16,17]. In addition, vital signs are automatically transferred to the patient record, reducing the impact of smoothing which is known to take place in manual systems [18-20]. Using our system we were able to detect intraoperative cardiovascular events automatically from the database [20]. The method by which data is collected from existing databases is becoming increasingly important [13,21-23]. Data from other information systems was also incorporated into our study (laboratory data, hospital information system) and additional manual data entry was avoided.

One of the major requirements for the routine use of computer-aided documentation systems for scientific purposes is the integration of electronic data collection into the daily clinical work process and the selection of a standardised, clearly structured database. In our opinion, an SQL-based standard database system is ideal for this purpose while the database structure should be consistent with the principles of the relational data model and a clearly defined description of the database structure be available.

Increased serum creatinine and perioperative outcome. The level of creatinine as a definitive indicator of renal failure has been arbitrarily chosen and differs in previous studies [1] between 1.4 and 1.5 mg dL−1. In our institution a serum creatinine level of more than 1.3 mg dL−1 is the defining line for renal failure as agreed between the laboratory physicians and clinicians. As serum creatinine is easy and cheap to measure, it is a widely accepted routine parameter for renal failure, despite its acknowledged limitations [21,22] and the availability of other more precise diagnostic methods.

The results of our study show that increased levels of serum creatinine are associated with an elevated risk of adverse outcomes in non-cardiac surgery, confirming results of previous studies mainly carried out in cardiac [5,6,23] and vascular [3,4,24] surgery. Several other studies have focused on the association of renal failure with morbidity and mortality [1,25]. Patients with renal failure were found to have a substantially increased risk of cardiovascular death and total mortality [1]. The association of renal failure with risk for adverse outcomes is strongly related to coexisting cardiovascular diseases and associated risk factors [25]. Vascular deficiency is both a result and a cause of chronic renal failure [11,26] and renal function is known to correlate with several cardiovascular risk factors. The results from our study only showed an association between renal failure and mortality whereas the incidence of intraoperative cardiovascular events was not different.

None of the reported studies on mortality in surgical patients have included a risk stratification of patients according to ASA physical status or type of surgery and urgency of surgery. The aim of this study was to establish whether increased serum creatinine level is a risk factor for morbidity and mortality. For this reason the study was designed as a matched-pairs case-control study investigating well-established risk factors for perioperative adverse events. The ASA classification is widely acknowledged and may, under certain circumstances, be a predictor of morbidity [27] and mortality [28]. Undisputedly the type of surgery also has a major impact on the surgical risk [11]. Similar to the Revised Cardiac Risk Index by Lee and colleagues [11] we classified surgical risk as high (intracranial, thoracic, abdominal and major vascular) or low (all other surgical procedures). Other possible risk factors included urgency of surgery, age and gender.

We are well aware of the existence and impact of other possible risk factors, especially of a cardiovascular nature, that may potentially compromise perioperative outcome. However, these comorbidities are reflected in the ASA classification included in our model. The statistical analysis of the matched-pairs model was restricted to the investigation of a small number of variables in order to retain a high number of cases and controls. A more sophisticated statistical analysis of such a model, including more detailed parameters of comorbidities would result in a much reduced number of cases and controls.


It is necessary to consider some limitations to our analysis. Due to the observational nature of the study, we were unable to identify and analyse all potential confounding factors. However, we consider the observational nature of our study a strength rather than a limitation. The matched case-control model based on a large number of data sets of patients undergoing non-cardiac surgery makes the results of our study more representative than those of other studies designed as randomised trials which tend to include a highly selected patients cohort. ‘Data mining’ may be considered a meaningful supplement or in some cases an alternative to prospective studies when the quantity of data is sufficient and the aim of the study as well as the criteria for inclusion and exclusion are well-defined. Other authors have also suggested the need for a possible re-evaluation of retrospective analysis in medicine using modern technology [21-23]. This method opens up the possibility of carrying out studies that would otherwise be impossible either because of ethical reasons or because they include risks for the patients involved.

Serum creatinine was used as an indicator of renal failure. Limitations of its use are well known [21,29]. Indeed, serum creatinine is a poor screening test for renal failure in elderly patients [22] because it is affected by muscle bulk (and breakdown), volume of distribution and rate of excretion. We are aware of methods used to estimate the creatinine clearance such as the Cockcroft-Gault equation [30] but clinicians do not routinely use this method. Furthermore, tubular secretion (and therefore creatinine excretion) may also be affected by the presence of drugs such as cimetidine, trimethoprim and some cephalosporins. We did not study the impact of such drugs on serum creatinine. Moreover no data was available on the cause, duration and severity of renal failure in our study. We did not distinguish between patients with potentially reversible and those with chronic renal failure. Our database did not provide consistent information on creatinine clearance and more detailed analysis of renal function such as the presence and degree of proteinuria. This lack of information potentially limits our ability to more accurately assess the impact of serum creatinine levels over 1.3 mg dL−1 on hospital mortality and the incidence of intraoperative cardiovascular events.

ASA grading was used as one of the bases of the case matching. It must also be remembered that a raised serum creatinine may affect the ASA grading making this not a wholly independent factor.


In summary, this analysis of computerised documented data using an anaesthesia recording system found that an elevated serum creatinine concentration of more than 1.3 mg dL−1 as a routine marker for renal function was associated with poorer outcome in patients undergoing non-cardiac surgery. These results may have some clinical implication but would benefit from further prospective studies.


We would like to thank Moredata GmbH, Giessen, for their help in data management and statistical evaluation.

Source of financial support

Financial support for this study was provided in part by a grant from IMESO GmbH, Hüttenberg, Germany. The founding agreement ensured the authors' independence in designing the study, interpreting the data, writing and publishing the report.

PD Dr. med. M. Benson is partner of the IMESO GmbH (Hüttenberg, Germany) and employee of the University Hospital Giessen. None of the other authors or participants have any financial interest in the subject matter, materials, or equipment discussed or in competing materials.


1. Mann JF, Gerstein HC, Pogue J, Bosch J, Yusuf S. Renal insufficiency as a predictor of cardiovascular outcomes and the impact of ramipril: the HOPE randomized trial. Ann Intern Med 2001; 134: 629-636.
2. Gerstein HC, Bosch J, Pogue J, Taylor DW, Zinman B, Yusuf S. Rationale and design of a large study to evaluate the renal and cardiovascular effects of an ACE inhibitor and vitamin E in high-risk patients with diabetes. The MICRO-HOPE Study. Microalbuminuria, cardiovascular, and renal outcomes. Heart Outcomes Prevention Evaluation. Diabetes Care 1996; 19: 1225-1228.
3. Szczech LA, Best PJ, Crowley E, et al. Outcomes of patients with chronic renal insufficiency in the bypass angioplasty revascularization investigation. Circulation 2002; 105: 2253-2258.
4. Brady AR, Fowkes FG, Greenhalgh RM, Powell JT, Ruckley CV, Thompson SG. Risk factors for postoperative death following elective surgical repair of abdominal aortic aneurysm: results from the UK Small Aneurysm Trial. On behalf of the UK Small Aneurysm Trial participants. Br J Surg 2000; 87: 742-749.
5. Anderson RJ, O'Brien M, MaWhinney S, et al. Mild renal failure is associated with adverse outcome after cardiac valve surgery. Am J Kidney Dis 2000; 35: 1127-1134.
6. Mangano CM, Diamondstone LS, Ramsay JG, Aggarwal A, Herskowitz A, Mangano DT. Renal dysfunction after myocardial revascularization: risk factors, adverse outcomes, and hospital resource utilization. The Multicenter Study of Perioperative Ischemia Research Group. Ann Intern Med 1998; 128: 194-203.
7. Benson M, Junger A, Quinzio L, et al. Clinical and practical requirements of online software for anesthesia documentation: an experience report. Int J Med Inf 2000; 57: 155-164.
8. Benson M, Junger A, Quinzio L, et al. Data processing at the anesthesia workstation: from data entry to data presentation. Meth Inf Med 2000; 39: 319-324.
9. DGAI-Kommission ‘Qualitätssicherung und Datenverarbeitung in der Anästhesie'. Kerndatensatz Qualitätssicherung in der Anästhesie. Anästh Intensivmed 1993; 34: 331-335.
10. American Society of Anesthesiologists (ASA). New classification of physical status. Anesthesiology 1963; 24: 111.
11. Lee TH, Marcantonio ER, Mangione CM, et al. Derivation and prospective validation of a simple index for prediction of cardiac risk of major noncardiac surgery. Circulation 1999; 100: 1043-1049.
12. Erdfelder E, Faul F, Buchner A. GPOWER: a general power analysis program. Behavior Res Meth, Instr Comput 1996; 28: 1-11.
13. Eichhorn JH. Anesthesia record keeping. Int J Clin Monit Comput 1993; 10: 109-115.
14. Edsall DW, Deshane P, Giles C, Dick D, Sloan B, Farrow J. Computerized patient anesthesia records: less time and better quality than manually produced anesthesia records. J Clin Anesth 1993; 5: 275-283.
15. Benson M, Junger A, Fuchs C, et al. Using an anesthesia information management system to prove a deficit in voluntary reporting of adverse events in a quality assurance program. J Clin Monit 2000; 16: 211-217.
16. Benson M, Junger A, Quinzio L, et al. Influence of the method of data collection on the documentation of blood pressure readings with an Anesthesia Information Management System (AIMS). Meth Inf Med 2001; 40: 190-195.
17. Reich DL, Wood RK, Mattar R, et al. Arterial blood pressure and heart rate discrepancies between handwritten and computerized anesthesia records. Anesth Analg 2000; 91: 612-616.
18. Reich DL, Timcenko A, Bodian CA, et al. Predictors of pulse oximetry data failure. Anesthesiology 1996; 84: 859-864.
19. Jones CA, McQuillan GM, Kusek JW, et al. Serum creatinine levels in the US population: Third National Health and Nutrition Examination Survey. Am J Kidney Dis 1998; 32: 992-999.
20. Röhrig R, Junger A, Hartmann B, et al. The incidence and prediction of automatically detected intraoperative cardiovascular events in noncardiac surgery. Anesth Analg 2004; 98: 569-577.
21. Perrone RD, Madias NE, Levey AS. Serum creatinine as an index of renal function: new insights into old concepts. Clin Chem 1992; 38: 1933-1953.
22. Swedko PJ, Clark HD, Paramsothy K, Akbari A. Serum creatinine is an inadequate screening test for renal failure in elderly patients. Arch Int Med 2003; 163: 356-360.
23. Anderson RJ, O'Brien M, MaWhinney S, et al. Renal failure predisposes patients to adverse outcome after coronary artery bypass surgery. VA Cooperative Study #5. Kidney Int 1999; 55: 1057-1062.
24. Szczech LA, Reddan DN, Owen WF, et al. Differential survival after coronary revascularization procedures among patients with renal insufficiency. Kidney Int 2001; 60: 292-299.
25. Culleton BF, Larson MG, Wilson PW, Evans JC, Parfrey PS, Levy D. Cardiovascular disease and mortality in a community-based cohort with mild renal insufficiency. Kidney Int 1999; 56: 2214-2219.
26. Detsky AS, Abrams HB, Forbath N, Scott JG, Hilliard JR. Cardiac assessment for patients undergoing noncardiac surgery. A multifactorial clinical risk index. Arch Intern Med 1986; 146: 2131-2134.
27. Wolters U, Wolf T, Stutzer H, Schröder T. ASA classification and perioperative variables as predictors of postoperative outcome. Br J Anaesth 1996; 77: 217-222.
28. Lagasse RS. Anesthesia safety: model or myth? A review of the published literature and analysis of current original data. Anesthesiology 2002; 97: 1609-1617.
29. Levey AS, Perrone RD, Madias NE. Serum creatinine and renal function. Annu Rev Med 1988; 39: 465-490.
30. Cockcroft DW, Gault MH. Prediction of creatinine clearance from serum creatinine. Nephron 1976; 16: 31-41.


© 2005 European Society of Anaesthesiology