Utility of serum albumin levels to predict postoperative complications in pancreatic surgeries in a tertiary cancer institute

Title****– Utility of  serumalbumin levels to predict postoperative complications in pancreatic surgeriesin a tertiary cancer institute

Backgroundand rationale– Albumin is commonly known as  an acute phase reactant which is  down regulated by inflammatory signals. Trauma,burns and surgical stress are known to lead to immediate drop in serum albuminlevels. The most important  cause of  postoperative serum  albumin losses  are due to sequestration into the third space.Capillary leak is a well known phenomenon which  accounts for >75% of albumin decrease in the early postoperative phase and appearsto be related to the magnitude of systemic inflammatory response.  Recent studies have suggested that it may beused to predict immediate post-surgical  outcomes. Since postoperative complicationsadd further stress to post-surgical stress, a continued drop in serum albuminlevels may be useful to predict post-surgical outcomes. We, therefore, want toassess whether serial measurement of decrease in serum albumin levels in theimmediate postoperative period after pancreatic resections  are useful in predicting short termpostoperative outcomes.

Aims ofthe study–To evaluate the utility of postoperative serum albumin decrease afterpancreatic resections  on post operativeday  1 ,3 and 5  for predicting immediate postoperativeoutcomes i.e morbidity and mortality within 30 days.

 Outcomes – Primary outcomesmeasured will be short term postoperative complications as per Clavien Dindoclassification

 Secondaryoutcomes  - Need for vasopressors, need forpostoperative ventilation, ICU and Hospital Length of Stay (LOS), status athospital discharge or at 30 days, whichever is earlier (alive or dead)

STUDY DESIGN–Prospective Observationalcohort study

STUDY PERIOD**:**Duration of studywill be for 12 months.

Methods

**Ethics-**The study will be undertaken after obtainingapproval from Institutional Ethics Committee(IEC).

Patients- The study will includeall  patients at our institute posted for pancreatic resections under general anaesthesia,within 12 months period, after approval from IEC. The study will be conductedin the Department of Anaesthesia, Critical Care and Pain at Tata MemorialCancer Centre

InclusionCriteria- All patients above 18 years undergoing elective major pancreaticresections under General Anaesthesia

ExclusionCriteria-

Other abdominal surgeries (liver resections, gallbladder, gastrectomies, oesophagectomies and colorectal surgeries)

SAMPLESIZE- All pancreatic surgeries in a periodof one year approximately 169  patients

Detailsof Study Procedures Involved-

Data will be collected from the electronic medicalrecord system (EMR) of the hospital which contains patients’ clinical,laboratory and radiological data. Data will also be recorded from the patient’scase file, the anaesthesia and PACU (intraoperative and postoperative)management summaries, monitoring charts and the surgical notes and will beentered in a prospectively maintained database. There will be no change inroutine anaesthesia practice and care. No additional interventions will beperformed. No additional hospital visits will be required.

Thefollowing data will be collected

Demographics,Comorbidities,Charlson comorbidity index, preoperative Hemogram, Renal andLiver function tests, coagulation studies

Patientundergoing pancreatic resections  will beanalysed  for the length of theirhospital stay maximum upto 30 days.The  postoperative morbidity, length of hospital stay, ICU stay and readmissions , durationof ventilation  and status at hospitaldischarge or at 30 days, whichever is earlier (alive or dead) will be noted.

Postoperative complications noted will be graded byseverity by use of the validated Clavien-Dindo system; grades I-II will beconsidered as minor and III-IV will be considered as major complications, respectively. Mortality will be  noted as Clavien Dindo 5 complication

The hospitalstay will be counted from the day of surgery till discharge or death

The  most recentpreoperative serum albumin level (POD -1)  and the serum albumin levels on the first, thirdand fifth postoperative day will be noted . The maximal difference between thepreoperative and postoperative values  onday 1,3 and 5 will be considered. Δ Max: maximal difference between thepreoperative and postoperative values. Δ POD 1: difference ofconcentration on POD −1 and POD 1 ; Δ POD3: difference of concentration onPOD−1 and POD 3 and Δ POD 5- difference of concentration on POD −1 and POD 5.

Samplesize

Total sample size 128 (64 in each group) will achieve80% power with size effect of 0.5 and with a significance level (alpha) of 5%using a two-sided two-sample t-test. As per complication rate of 30%, the finalsample size for this study was n=152. In order to adjust for 10% dropout ormissing data, final sample size resulted in n=169 (51 patients complication vs118 without complication).

 Statisticalanalysis plan:

Continuous variables will be  presented as mean with SD or median value withIQR depending on the normality of the distribution and compared using Student’st-test and Mann–Whitney U test, whereas categorical variables will be  given as frequencies with percentages andcompared with χ2 test. Receiver operating characteristic (ROC) curves will be  applied to obtain the area under the curve(AUC) and to determine ideal cut-offs.  Univariate analysis of categoricalvariables will be compared by Pearson’s Χ2 test or Fisher’s exact test.Logistic regression will be  applied toidentify independent predictors; variables with significance < 0.2 inunivariate analyses will be further included in multivariable analyses.Multivariate logistic regression (adjusted for Albumin infusion) and major riskfactors as independent variables. Predictive accuracy of a logistic regression modelwill be assessed using c-index. All analysis will two sided, and significancewill set at a p-value of 0.05.Statistical analyses will be performed using SPSS(the statistical package for social sciences) IBM Corp. Released 2017. IBM SPSSStatistics for Windows, Version 21.0. Armonk, NY: IBM Corp and R studio(version 1.2).

     References

Hübner M,Mantziari S, Demartines N, Pralong F, Coti-Bertrand P, Schäfer M. Postoperativealbumin drop is a marker for surgical stress and a predictor for clinicaloutcome: a pilot study. Gastroenterology research and practice. 2016 Oct;2016.

 Labgaa I, JoliatG, Kefleyesus A, et alIs postoperative decrease of serum albumin an earlypredictor of complications after major abdominal surgery? A prospective cohortstudy in a European centreBMJ Open 2017;7.