A risk score for conversion from laparoscopic to open cholecystectomy

A risk score for conversion from laparoscopic to open cholecystectomy
of 6
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  A risk score for conversion from laparoscopic to open cholecystectomy Nuri Aydın Kama, M.D.*, Murat Kologlu, M.D., Mutlu Doganay, M.D., Erhan Reis, M.D.,Mesut Atli, M.D., Mete Dolapci, M.D.  Ankara Numune Hospital, 4th Department of Surgery, Ankara, Turkey Manuscript received July 26, 2000; revised manuscript January 8, 2001 AbstractBackground:  Laparoscopic cholecystectomy has become the standard operative procedure for cholelithiasis, but there are still some patientsrequiring conversion to open cholecystectomy mainly because of technical difficulty. Our aim was to develop a risk score for prediction of conversion from laparoscopic to open cholecystectomy. Methods:  Preoperative clinical, laboratory, and radiologic parameters of 1,000 patients who underwent laparoscopic cholecystectomy wereanalyzed for their effect on conversion rates. Six parameters (male sex, abdominal tenderness, previous upper abdominal operation,sonographically thickened gallbladder wall, age over 60 years, preoperative diagnosis of acute cholecystitis) were found to have significanteffect in multivariate analysis. A constant and coefficients for these variables were calculated and formed the risk score. Results:  Overall 48 patients required conversion to open cholecystectomy (4.8%). These patients had significantly higher scores (mean 6.9versus  7.2,  P  0.001). Increasing scores resulted with significant increases in conversion rates and probabilities ( P  0.001). Ideal cut-off point for this score was  3; conversion rate was 1.6% under  3, but 11.4% over this value ( P  0.001). Conclusions:  Conversion risk can be predicted easily by this score. Patients having high risk may be informed and scheduled appropriately.An experienced surgeon has to operate on these patients, and he or she has to make an early decision to convert in case of difficulty. © 2001Excerpta Medica, Inc. All rights reserved. Keywords:  Laparoscopic cholecystectomy; Conversion; Risk factors; Severity of Illness Index Laparoscopic cholecystectomy has become the standard op-erative procedure for treating cholelithiasis. Today morethan 80% of cholecystectomies are carried out laparoscopi-cally. Decreased postoperative pain and discomfort, de-creased postoperative ileus, earlier oral intake, decreasedpostoperative hospital stay, earlier return to normal activity,and improved cosmetic results are the known advantages of laparoscopic cholecystectomy over the open procedure. De-spite the increasing experience, approximately 2% to 15%of attempted laparoscopic cholecystectomies have to beconverted to open operation, usually because of dense ad-hesions and inflammation [1–9]. Some preoperative vari-ables may help to predict the risk of conversion in a partic-ular patient or difficulty of the operation. An appreciationfor these predictors of conversion may allow appropriateplanning by the patient, the surgeon, and the institution.The specific aim of this study was to develop a scoringsystem for predicting the risk of conversion from laparo-scopic to open cholecystectomy. Methods We examined our computer database of patients under-going laparoscopic cholecystectomy and retrospectivelyanalysed the data of our first 1,000 consecutive patientsoperated on between March 1992 and June 1999 in AnkaraNumune Hospital 4th Department of Surgery. Univariateanalysis (chi-square) was applied to the preoperative vari-ables of patients to determine their effect on rates of con-version to laparotomy. The variables included were, age,sex, medical history, previous operations, previous attacksof acute cholecystitis, present symptoms and physical ex-amination findings, labaratory data and findings of imagingstudies (especially ultrasonography and endoscopic retro-grade cholangiopancreatography [ERCP] findings). The di-agnosis of an attack of acute cholecystitis was based on * Corresponding author’s address: Uzmanlar Tıp Grubu, 3. Cadde, 41.Sokak, No: 3/1, Bahc¸elievler, 06500, Ankara, Turkey. Tel.:   90-312-3122829; fax:  90-312-3092829.  E-mail address:  nak4ceranur@superonline.comThe American Journal of Surgery 181 (2001) 520–5250002-9610/01/$ – see front matter © 2001 Excerpta Medica, Inc. All rights reserved.PII: S0002-9610(01)00633-X  acute abdominal pain with signs of peritoneal irritation inthe right upper quadrant and the need for emergency admis-sion to the hospital associated with fever (  38.5°C) andleukocytosis (  12,000 cells/mm 3 ) in the presence of cho-lelithiasis. We were not too strict with the “fever” parame-ter, but the other signs were present in patients with adiagnosis of acute cholecystitis. Only the patients clearlydocumented to have these signs before were taken as “hav-ing a history of previous attacks.”The variables with a statistically significant effect ac-cording to univariate analysis were selected and a multivar-iate analysis (stepwise logistic regression) was applied tothem. By using the variables having independent significanteffect on conversion rates, a logistic model was developed;a constant and coefficients for these significant variableswere calculated by another multivariate analysis (not step-wise) [10], which were than rounded and multiplied by 10for ease of use. These coefficients and constant formed therisk score for conversion from laparoscopic to open chole-cystectomy (RSCLO).All patients were scored and divided into 5 groups forcomparisons: (1) patients with a score between   20 and  11; (2) patients with a score between   10 and   1; (3)patients with a score between 0 and 9; (4) patients with ascore between 10 and 19; and (5) patients with a score  19.Observed conversion rates according to scores were calcu-lated and compared. Also probabilities of conversion ac-cording to scores were calculated as described by Hunt et al[10].To determine the discriminatory ability of the score,different cut-off points were chosen, sensitivity and speci-ficity of these cut-off scores to predict conversion rates werecalculated seperately and receiver-operator characteristiccurve (ROC  sensitivity versus specificity) was drawn. Inother words, every point on the ROC curve represents thesensitivity-specificity pair of a different cut-off point in thescore. The area under the curve was calculated. The idealcut-off score with the highest sensitivity-specificity pair wasfound [10].The proportion of patients having a calculated probabil-ity of conversion   10% or   90% according to score wasdetermined, which is called “sharpness.” This value showsthe rate of sharp and precise predictions made according tothe scoring system. Probabilities between 10% and 90%were called “nonsharp predictions.”Operation times of patients with different scores werecompared. These data were affected by conversion ratesbut also indirectly showed the difficulty of the operation;as converted patients have longer operation times and con-version rates increase with higher scores, mean operationtimes for high scores may be affected with the rate of converted patients. So we compared the operation timesboth in whole patient population and in patients who un-derwent successful laparoscopic cholecystectomy (con-verted patients excluded).The chi-square test was used for comparison of propor-tions. Student’s  t   test and one-way analysis of variance(ANOVA) were used for comparison of means. Statisti-cally,  P  0.05 was considered significant. SPSS version 8.0for Windows software was used in statistical analyses. Results A total of 1,000 consecutive patients were included in thestudy. There were 804 women and 196 men; mean age was49.6 (range 16 to 85). Major complications (bile duct orgastrointestinal injury, intraabdominal hemorrhage) oc-curred in 2.8% of patients and 0.8% of patients requiredreoperation because of complications. Total operative mor-tality rate was 0.4%. Conversion to laparotomy was re-quired in 48 patients (4.8%); 35 were because of difficultyin dissection and delineating the anatomy, 8 were because of injury to bile ducts or gastrointestinal system, 4 were be-cause of uncontrolled hemorrhage, and 1 was for the treat-ment of common bile duct stones.Among all preoperative data, 6 variables were found tohave significant association with conversion rates according Table 1Factors significantly affecting conversion ratesVariables Conversion rates(%) P  value(univariate) P  value(multivariate)Sex Male 11.2   0.0001 0.0008Female 3.2Abdominal tenderness Present 6.3 0.007 0.013Absent 2.5Previous upper abdominal operation Present 7.3 0.037 0.014Absent 4.0Gallbladder wall (ultrasonography) Thickened 30.8   0.0001 0.009Normal 4.1Age (years)   60 7.0 0.037 0.052  60 3.9Acute cholecystitis Present 16.7   0.0001   0.0001Absent 2.8521  N.A. Kama et al. / The American Journal of Surgery 181 (2001) 520–525  to univariate analysis (Table 1). These variables were eval-uated by multivariate analysis and all were found to bestatistically significant (age was borderline significant). Ac-cording to these analyses, male patients, patients with ab-dominal tenderness on preoperative physical examination,patients with a history of previous upper abdominal opera-tion, patients with a thickened gallbladder wall on ultra-sonography (  4 mm), patients aged 60 years or older, andpatients with a clinical diagnosis of acute cholecystitis orwith a history of previous attacks of acute cholecystitisrequired conversion significantly more than others. Amongthe patients who required conversion to laparotomy, 45.8%were male, 79.2% had abdominal tenderness on physicalexamination, 37.5% underwent an upper abdominal opera-tion previously, 16.7% had thickened gallbladder wall ul-trasonographically, 41.7% were aged 60 years or older, and50.0% had clinically acute cholecystitis or suffered previousattacks of acute cholecystitis.These 6 variables with their coefficients and a constantwere included in the scoring system (Table 2). The sum of coefficients and constant gives the final score of the patientand can take a value between  20 and 41. Increasing scoreswere associated with significantly increased conversionrates ( P   0.001) (Fig. 1). Mean    SEM score of patientswho required conversion was 6.9  1.8, but this value was  7.2    0.3 for patients who underwent successful laparo-scopic cholecystectomy ( P  0.001).As shown in Fig. 2, higher scores result with increasedprobability of conversion. The important point is that, thecalculated probability of conversion for a patient with themaximum score of 41 is nearly 80%, which means that anexperienced laparoscopic surgeon can still complete theoperation laparoscopically. Overall 95.3% of the patientscan be correctly predicted to undergo successful laparo-scopic cholecystectomy or require conversion to open cho-lecystectomy with the developed logistic model used todesign the score.The ROC curve is shown in Fig. 3. The area under thecurve is 0.83; this means that, a randomly selected patientfrom the group who required conversion has a higher scorethan a randomly selected patient from the group who un-derwent successful laparoscopic cholecystectomy at least83% of the time. Ideal cut-off point for the score wascalculated to be  3. The sensitivity was 78% and specificitywas 72% for this cut-off score for predicting conversion.The conversion rate below  3 was 1.6% and over  3 was11.4% ( P   0.001). Among patients who required conver-sion, 77.1% had a score over  3.As we have stated previously, the probability of conver-sion for the maximum score is nearly 80%, so when calcu-lating sharpness we took the proportion of patients with aprobability of conversion less than 10%. The proportion of sharp predictions was 90.3% and of nonsharp predictions,9.7% when this score was used.Operation time was taken as an indicator of difficulty.Either evaluated in all patient groups or in successful lapa- Table 2Risk score for conversion from laparoscopic to open cholecystectomyVariables CoefficientsSex Male 11Female 0Abdominal tenderness Present 9Absent 0Previous upper abdominal operation Present 8Absent 0Gallbladder wall (ultrasonography) Thickened 13Normal 0Age (years)   60 5  60 0Acute cholecystitis Present 15Absent 0Constant   20Fig. 1. Conversion rates according to score.  P  0.001 for the difference of conversion rates according to scores.Fig. 2. Probability of conversion to open cholecystectomy according toscore.522  N.A. Kama et al. / The American Journal of Surgery 181 (2001) 520–525  roscopic cholecystectomy groups (converted patients ex-cluded), increasing scores resulted with longer operationtimes (Table 3). The difference between the first two groups(score between   20 to   11 and   10 to   1) was notsignificant when compared one by one; but the majority of patients in these groups had a score under the cut-off point  3 and had a very low risk of conversion. This findingshows the reliability of the cut-off point. When convertedpatients were excluded, the difference between last twogroups (score between 10 to 19 and  19) did not reach tostatistical significance due to the limited number of patientswith a score   19, but still there was a difference. Whencalculated according to the cut-off point, mean    SEMoperation times were 51.3    0.9 minutes under   3 and68.2    2.1 minutes over   3 for all patients ( P   0.001).Mean    SEM operation times were 50.5    0.9 minutesunder   3 and 60.2    1.7 minutes over   3 for patientsunderwent successful laparoscopic cholecystectomy (con-verted patients excluded;  P  0.001).Among 28 patients having major complications, 12 pa-tients required conversion to laparotomy for the manage-ment of complications, in 8 patients complications wererealized postoperatively and they were reoperated, and inthe rest of the patients the complications were managedeither laparoscopically or postoperatively by nonsurgicalmethods. Five patients suffered intraabdominal organ orvascular injury due to puncture during trocar insertion at thebeginning of the operation; so these complications had noassociation with difficulty level of the operation. In 10patients, complications occurred mainly due to an inexpe-rienced surgeon operating and technical insufficiency; thesecomplications included avulsion of the cystic artery or thecystic duct. The score was over   3 in the remaining 13patients, and difficulties were encountered in dissection anddelineating the anatomy in these patients because of severeinflammation or fibrosis. Operating surgeons insisted oncompleting these operations laparoscopically, which re-sulted with complications like common bile duct or hepaticduct injury and duodenal injury. Perhaps the complicationsmight have been prevented if a decision of conversion tolaparotomy had been made in these patients, constituting56.5% of the complicated patients (when trocar injurieswere excluded). The mean    SEM operation time was99.4  10.7 minutes in these 28 patients. Comments The known advantages and safety of laparoscopic cho-lecystectomy made it the standard treatment of cholelithia-sis. Today the majority of cholecystectomies are carried outlaparoscopically and surgery residents learn laparoscopiccholecystectomy before open cholecystectomy. Conversionto open procedure is required in 2% to 15% of patients[1–9]. The need for conversion to laparotomy is neither afailure nor a complication, but an attempt to avoid compli-cations. In more than 50% of conversions, the main indica-tion is difficulty in dissection and unclear anatomy becauseof adhesions or inflammation; the remaining patients requireconversion because of complications or unexpected findings[1–4,6–8,11].Risk factors determining conversion from laparoscopicto open cholecystectomy have been investigated in numer-ous studies [1,3,4,6,7,9]. The most significant factor asso-ciated with conversion is the presence or previous attacks of acute cholecystitis, and some other proposed risk factors arerelated to this variable [1,3,4,6–9,12,13]. The conversionrate changes between 10% and 50% in patients with acutecholecystitis [1–3,11,12,14]. Risk depends on previous at-tacks and the grade of present inflammation where thehighest risk is expected in patients with gangrenous chole- Fig. 3. Receiver-operator characteristic curve (ROC) for the score. The areaunder the curve is 0.83.Table 3Mean operation times according to scoresScore Mean  SEM operation timesAll patients(n  1,000)*Converted patientsexcluded (n  952)**  20–  11 50.9  1.0 50.6  1.0  10–  1 54.3  1.8 51.5  1.60–9 64.9  2.7 59.1  2.410–19 78.3  4.7 71.9  4.4  19 104.8  10.5 75.4  8.0*  P   0.001 for the difference between score groups in all patients(one-way ANOVA). When compared one by one, except the differencebetween the first two groups, all the differences between score groups arestatistically significant ( P   0.05, Tukey-Kramer multiple comparisonstest).**  P   0.001 for the difference between score groups in patients whounderwent successful laparoscopic cholecystectomy (converted patientsexcluded; one-way ANOVA). When compared one by one, except thedifference between the first two groups and the difference between the lasttwo groups, all the differences between score groups are statisticallysignificant ( P   0.05, Tukey-Kramer multiple comparisons test).523  N.A. Kama et al. / The American Journal of Surgery 181 (2001) 520–525  cystitis or empyema [2,7,11]. In our patients we realizedthat presence of acute cholecystitis at the time of laparo-scopic cholecystectomy and documented previous attacks of acute cholecystitis had similar effects on conversion risk, sowe took these as a single variable.Another risk factor is the increased thickness of gallblad-der wall on preoperative ultrasonography that represents thepresent inflammation or fibrosis due to previous attacks of cholecystitis [1,3,4,6,13,15]. Abdominal tenderness or ri-gidity on preoperative physical examination is also an im-portant factor in defining the risk [1] that is associated withinflammation. The reason for older age being a risk factor issuggested to be a longer history of gallstones and increasednumber of cholecystitis attacks [3,4,6,7,9,11,12,14]. Be-sides, elderly patients have a higher likelihood of compli-cated biliary pathology [3,4,12]. Previous abdominal oper-ations (especially upper abdominal) generally cause thedissection to be more difficult because of dense adhesions[1,3,9] and may require conversion to laparotomy.The reason of higher conversion rates in male patientsremains unexplained in the literature, but male sex is ac-cepted as a significant risk factor in most series [3,7,16]. Wefound this variable to be an independent risk factor, whichmeant that it did not have any association with other vari-ables. In our series mean operation time was longer in malepatients (65.1 versus 54.7 minutes,  P  0.001). Our obser-vation was that male patients had more intense inflamma-tion or fibrosis resulting in more difficult dissection both inthe triangle of Calot and through the plane between gall-bladder and liver, when compared with female patients witha similar history.None of these risk factors are contraindication to lapa-roscopic cholecystectomy, but may help us predict the dif-ficulty of the procedure. Obesity or the experience of oper-ating surgeon were not found to be significantly associatedwith conversion rates in our study. Clinical or laboratorydata indicating common bile duct stone are presented as arisk factor in some series [1,6]. This depends on the treat-ment policy of the institution for common bile duct stonesand the experience in laparoscopic common bile duct ex-ploration [3,5]. Our policy is to perform a preoperativeERCP and extract common bile duct stones if present inpatients with clinical or laboratory findings of common bileduct stones; so these variables were not found to be asso-ciated with conversion rates in our study.Preoperative prediction of the risk of conversion is animportant aspect of planning laparoscopic surgery. By thehelp of accurate prediction, patients may be informed ap-propriately and they may have a chance to make arrange-ments regarding their work and family. On the other hand,surgeon may schedule the time and team for the operationappropriately [3,4,7]. Patients predicted to have a high risk of conversion have to be operated on by an experiencedteam; these patients are not suitable for routine residenttraining [7]. On the other hand, residents need some trainingwith difficult cases; in this situation senior residents may beallowed to operate on selected high-risk patients with anattending experienced laparoscopic surgeon. Surgeons inthe early phase of the learning curve have to select patientswith low risk of conversion [3,7], especially if they are notoperating under the supervision of an experienced lapa-roscoppic surgeon. When operating on a high-risk patient,the surgeon has to make an early decision to convert if he orshe encounters difficulty in dissection; early conversionshortens the operation time and decreases morbidity [2,4,7,14]. A policy of converting if there is no progress in dis-section of the Calot’s triangle within 15 to 30 minutes maybe adopted for high-risk patients [1]. Today laparoscopiccholecystectomy can be performed safely as an outpatientprocedure where patients are discharged within the sameday, and this procedure decreases average hospital costssignificantly. Patient selection is very important for outpa-tient facilities, and low-risk patients have to be selected[1,3,6–8,17,18]. Patients predicted to have high risk have tobe scheduled for longer hospitalization and more intensivefacilities [1,3,6].Scoring systems are frequently used to predict the prog-nosis of critically ill patients, and less frequently in otherareas of clinical practice. These systems help us accuratelygroup patients in homogeneous subsets and comparisonsbetween different patient populations can be made morereliably. Also scoring may be an inclusion or exclusioncriteria in clinical studies, and sometimes a guide to moni-toring or treatment in clinical practice. A score can reducenumerous clinical variables and knowledge to a small num-ber. Developing scoring systems may be possible for everyarea in clinical practice. There are numerous studies aboutthe risk of conversion in laparoscopic cholecystectomy andin some studies predictions are made using some variables[1,3,7], but patients were not scored in these studies. Anaccurate scoring system is needed on this issue to makemore reliable evaluations and provide better surgical prac-tice. The scoring system helps the surgeons make standardpredictions independent of their clinical experience. A well-defined cut-off score may clarify what “difficult laparo-scopic cholecystectomy” means, and surgeons from differ-ent parts of the world can communicate better on this issue.We developed a risk score to predict the probability of conversion in laparoscopic cholecystectomy. The score hasa good discriminatory ability with an area of 0.83 under theROC curve (the maximum value may be 1.0, which isactually impossible). For 90.3% of patients sharp predic-tions can be made. It is possible to group patients accordingto their scores or discriminate high-risk and low-risk pa-tients using the cut-off point with a sensitivity and speci-ficity over 70%. The probability of conversion can be pre-dicted according to score and this makes appropriate patientselection possible. Even the patients with maximum scoredo not have a probability of 100% conversion; this showsthat an experienced surgeon can complete laparoscopic cho-lecystectomy successfully in a high-risk patient. Conversionrates, difficulty, and operation time are related to each other. 524  N.A. Kama et al. / The American Journal of Surgery 181 (2001) 520–525
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks