gen.POSSUM: A function to compute POSSUM scores

In dannyjnwong/HSRC: Useful functions for perioperative research

Description

This function will parse a dataframe and produce POSSUM scores to predict perioperative mortality and morbidity. To use the function, you will need to manipulate your dataframe to have columns with the structure detailed below.

Usage

gen.POSSUM(x)

Arguments

x

A dataframe or tbl where each row is a patient observation, and the columns are POSSUM predictor variables. x must contain the following column names (not necessarily in order): Age continuous variable, numeric or integer JVP binary variable, whether the patient has raised JVP or not Cardiomegaly binary variable, whether the patient has cardiomegaly on CXR or not Oedema binary variable, whether the patient has peripheral oedema or not Warfarin binary variable, whether the patient normally takes warfarin or not Diuretics binary variable, whether the patient normally takes a diuretic medication or not AntiAnginals binary variable, whether the patient normally takes anti-anginal medication or not Digoxin binary variable, whether the patient normally takes digoxin or not AntiHypertensives binary variable, whether the patient normally takes blood pressure meds or not Dyspnoea categorical variable, can be: "Non" = None; "OME" = On exertion; "L" = Limiting activities; "AR" = At rest Consolidation binary variable, whether the patient has consolidation on CXR PulmonaryFibrosis binary variable, whether the patient has a history of pulmonary fibrosis or imaging findings of fibrosis COPD binary variable, whether the patient has COPD or not SysBP continuous variable, pre-op systolic blood pressure (in mmHg) HR continuous variable, pre-op pulse/heart rate (in beats per min) GCS continuous variable, pre-op Glasgow Coma Scale (3-15) Hb continuous variable, pre-op Haemoglobin (in g/L), please note the units! WCC continuous variable, pre-op White Cell Count (in * 10^9cells/L) Ur continuous variable, pre-op Urea (in mmol/L) Na continuous variable, pre-op Sodium concentration (in mmol/L) K continuous variable, pre-op Potassium concentration (in mmol/L) ECG categorical variable, can be "ND" = Not done; "NOR" = Normal ECG; "AF6090" = AF 60-90; "AF>90" = AF>90; "QW" = Q-waves; "4E" = >4 ectopics; "ST" = ST or T wave changes; "O" = Any other abnormal rhythm OpSeverity categorical variable, the surgical severity, can be Min = Minor; Int = Intermediate; Maj = Major; Xma = Xmajor; Com = Complex ProcedureCount categorical variable, number of procedures patient underwent in the last 30 days including this one, can be "1" = 1; "2" = 2; "GT2" = >2 EBL categorical variable, the estimated blood loss, can be "0" = 0-100ml; "101" = 101-500ml; "501" = 501-999ml; "1000" = >=1000 PeritonealContamination categorical variable, whether there was peritoneal soiling, can be "NA" = Not applicable; "NS" = No soiling; "MS" = Minor soiling; "LP" = Local pus; "FBC" = Free bowel content pus or blood Malignancy categorical variable, whether the patient has malignant disease, can be "NM" = Not malignant; "PM" = Primary malignancy only; "MNM" = Malignancy + nodal metastases; "MDM" = Malignancy + distal metastases OpUrgency categorical variable, NCEPOD classifications of urgency, can be "Ele" = Elective; "Exp" = Expedited; "U" = Urgent; "I" = Immediate

Value

A dataframe (or tbl), which you can assign to an object, with the following variables:

PhysScore

The physiological score for POSSUM

OpScore

The operative score for POSSUM

POSSUMLogit

The log-odds for morbidity as calculated by POSSUM

pPOSSUMLogit

The log-odds for mortatlity as calculated by pPOSSUM

Converting to probability scale

The function will produce POSSUMLogit and pPOSSUMLogit values which are on the log-odds scale To convert to probabilities (0 to 1 scale), use arm::invlogit(). See: invlogit.

References

• Copeland GP, Jones D, Walters M. POSSUM: A scoring system for surgical audit. Br J Surg. 1991 Mar 1;78(3):355–60. http://onlinelibrary.wiley.com/doi/10.1002/bjs.1800780327/abstract.

• Prytherch DR, Whiteley MS, Higgins B, Weaver PC, Prout WG, Powell SJ. POSSUM and Portsmouth POSSUM for predicting mortality. Br J Surg. 1998 Sep 1;85(9):1217–20. http://onlinelibrary.wiley.com/doi/10.1046/j.1365-2168.1998.00840.x/abstract

Examples

## Not run: 
#Example of pre-processing to rename data variables to match expected column names
library(tidyverse)

test_data <- raw_data %>%
select(Age = S01Age,
        JVP = S03ElevatedJugularVenousPressureJvp,
        Cardiomegaly = S03RadiologicalFindingsCardiomegaly,
        Oedema = S03PeripheralOedema,
        Warfarin = S03DrugTreatmentWarfarin,
        Diuretics = S03DrugTreatmentDiureticTreatment,
        AntiAnginals = S03DrugTreatmentAntiAnginal,
        Digoxin = S03DrugTreatmentDigoxinTherapy,
        AntiHypertensives = S03DrugTreatmentAntiHypertensive,
        Dyspnoea = S03Dyspnoea,
        Consolidation = S03RadiologicalFindingsConsolidation,
        PulmonaryFibrosis = S03PastMedicalHistoryPulmonaryFibrosis,
        COPD = S03PastMedicalHistoryCOPD,
        SysBP = S03SystolicBloodPressureBpAtPreAssessment,
        HR = S03PulseRateAtPreoperativeAssessment,
        GCS = S03GlasgowComaScaleGcsPreInductionOfAnaesthesia,
        Hb = S03Hb,
        WCC = S03WhiteCellCountWcc,
        Ur = S03Urea,
        Na = S03Na,
        K = S03K,
        ECG = S03EcgFindings,
        OpSeverity = S02PlannedProcSeverity,
        ProcedureCount = S04ProcedureCount,
        EBL = S04BloodLoss,
        PeritonealContamination = S04PeritonealContamination,
        Malignancy = S04Malignancy,
        OpUrgency = S02OperativeUrgency)

## End(Not run)

test_data <- patients
test_output <- gen.POSSUM(test_data)
head(test_output)

dannyjnwong/HSRC documentation built on May 18, 2019, 2:37 a.m.