Does This Patient Have Bacterial Peritonitis or Portal Hypertension? How Do I Perform a Paracentesis and Analyze the Results?

A 44-year-old man with cirrhosis is admitted with fever but has no obvious source of infection. His physical examination reveals ascites and minimal abdominal tenderness.

In a previous Rational Clinical Examination article, Williams and Simel¹ discussed the previous scenario and suggested performing a diagnostic paracentesis to seek for the source of the patient's fever. Ascites is the most common major complication of cirrhosis.² The syndrome of infected ascites was first recognized and described in the European literature in the early 20th century.^{2 - 4} Infection occurs in as many as 27% of cirrhosis patients admitted to the hospital for evaluation and management of symptoms associated with ascites. The presentation of spontaneous bacterial peritonitis can range from silent to overt⁵ ; however, patients typically have nonspecific symptoms such as nausea, abdominal pain, malaise, fever, or mild confusion. In reports⁶ from the 1970s, the mortality rate from spontaneous bacterial peritonitis exceeded 90% but recent data⁷ show a lower mortality rate of 30%. Diagnosis and subsequent treatment of spontaneous bacterial peritonitis depend on the results from a diagnostic paracentesis. Paracentesis is also used in the diagnostic workup for other causes of ascites such as malignancy. Therapeutic paracentesis may be used to manage refractory ascites.

In general, paracentesis is quite safe. Several prospective and retrospective studies of diagnostic and therapeutic paracentesis report on significant bleeding and death (Table 1). The studies are too heterogeneous to allow meaningful meta-analysis but the overall rate of significant bleeding^{8 - 15} ranges from 0% to 2.7% and of death^{9 - 10 ,12 - 13 ,16} from 0% to 17%. Variability in bleeding rates may be influenced by preexisting renal dysfunction.¹² The heterogeneity in the risk of death is primarily due to a retrospective study of therapeutic paracentesis in hospitalized patients at a tertiary care hospital.¹⁶ The risk of death, renal failure, hyponatremia, and hepatic encephalopathy with therapeutic paracentesis are further discussed in “Results” (see “Plasma Expanders in Therapeutic Paracentesis”). A retrospective analysis of 242 consecutive diagnostic paracenteses reported complicated bowel perforation in 0.83% (95% confidence interval [CI], 0.10%-3.0%) and residual catheter tip fragment in 0.41% (95% CI, 0.01%-2.3%).¹⁰ The risk of a paracentesis leak is variable from 0% to 23%.^{8 ,14 ,16}

Quiz Ref IDRelative contraindications to paracentesis include an uncooperative patient, skin infection at the proposed puncture site, pregnancy, and severe bowel distension.¹⁷ One expert suggests that coagulopathy precludes paracentesis only if there is clinically evident fibrinolysis or disseminated intravascular coagulation.¹³

Searches of MEDLINE (using Ovid) from 1966 to April 2007 and EMBASE from 1980 to April 2007 identified English-language studies completed in a clinical setting. The search strategy for studies evaluating interventions to optimize success of the procedure used the terms paracentesis, ascites, ascites fluid, needle, thrombocyte count, platelet count, prothrombin time, partial thromboplastin time, bleeding, blood clotting test, blood coagulation tests, hemorrhage, bacterium culture, culture techniques, ultrasonography, and albumin. For studies of test accuracy, the terms paracentesis, ascites fluid, ascites, ascites fluid analysis, peritonitis, portal hypertension, cell count, leukocyte count, neutrophil count, hydrogen-ion concentration, and albumin were used. Additional articles were found by searching the bibliographies of retrieved articles.

Randomized trials of a predominantly adult population undergoing interventions to reduce adverse events from paracentesis and optimize success were included. If no randomized studies were identified, studies of lower quality including cohort, case-control, and case series were retrieved. Primary data or appropriate summary statistics had to be available for inclusion. If necessary, further data were obtained by contacting study authors. Interventions of interest included those that could be used at the time of paracentesis such as measuring coagulation status, use of ultrasound guidance, location of needle insertion, needle design, bedside vs delayed laboratory inoculation in culture bottles, and use of plasma expanders in therapeutic paracentesis. The outcomes of interest included obtaining a sufficient amount of ascitic fluid, number of attempts, and adverse events.

To examine the accuracy of ascitic fluid analysis in patients suspected of having spontaneous bacterial peritonitis or portal hypertension, we included studies with a predominantly adult population that were prospective in design, described the use of an appropriate reference standard, and applied the same index and reference test to all patients. Furthermore, primary data or appropriate summary statistics had to be available. Where necessary, additional data were obtained by contacting study authors. Diagnostic tests of interest included ascitic fluid white blood cell (WBC) count, ascitic polymorphonuclear (PMN) count, ascitic pH, blood–ascitic fluid pH gradient, and serum-ascites albumin gradient. The gradients were calculated by measuring the serum and ascitic fluid specimens obtained on the same day and subtracting the ascitic value from the serum value.

Two reviewers independently reviewed the abstracts and selected relevant publications that met inclusion criteria. Disagreements were resolved by discussion with the third reviewer. In cases of doubt, full-text articles were retrieved for review and discussion. Two investigators then independently reviewed all full-text articles to confirm that inclusion criteria were met. For intervention studies, data were collected on the quality of the study and method used for outcome assessment. For studies of test accuracy, a data collection form was used to extract information on study quality, sampling method, reference standard, presence of blinding, index test, and population demographics. Positive ascitic fluid cultures and invasive portal pressure measurements were used as reference standards for studies on spontaneous bacterial peritonitis and portal hypertension, respectively.

For studies of interventions, a 2-tailed Fisher exact test was used to examine the significance of the association between 2 categorical variables. For studies on the use of plasma expanders, statistical heterogeneity was assessed by the Q test. Odds ratios and absolute risk reductions were determined by using a random-effects model via the Mantel-Haenszel and inverse variance methods for combined and single studies, respectively. For studies of test accuracy, summary likelihood ratios (LRs) were calculated using a random-effects model (DerSimonian and Laird). The summary measures were calculated on the natural logarithm of the LR and then exponentiated to arrive at LRs. If 1 or more studies contained zeros in their 2 × 2 table, resulting in LR estimates of zero or infinity, 0.5 was added to the counts for all studies in that analysis. Statistical analysis was conducted using R (R Foundation for Statistical Computing, Vienna, Austria) version 2.5.0 and the rmeta contributed package version 0.8-2.

Seventy-three citations of potential studies on preprocedure coagulation investigations were found (Figure 1). Five were obtained for full-text review, but 3 were discarded because of retrospective design^{10 - 11} and because less than half of the patients had complete coagulation profiles.¹³ Only 2 prospective studies systematically obtained prothrombin times and platelet counts prior to paracentesis in all patients.^{8 - 9}

The first study included both diagnostic and therapeutic procedures in the emergency department under ultrasound guidance. Patients presenting with concomitant problems requiring transfusion were excluded. No bleeding complications occurred in any of the 96 diagnostic procedures (95% CI, 0%-3.8%), even though 6 paracenteses were carried out with a platelet count of less than 50 000/μL and 27 paracenteses were performed with an international normalized ratio (INR) of more than 1.5. Only 2 episodes of minor bleeding occurred among the 314 therapeutic procedures (0.0064%; 95% CI, 0.0008%-0.023%). Both occurred with an INR of 2.5 to 2.9 and platelet count of 50 000 to 99 000/μL. Neither resulted in hospitalization or transfusion.⁹

The second study included 1100 large-volume paracenteses performed by trained endoscopy assistants in an outpatient setting. There was no significant bleeding in any patient (95% CI, 0%-0.33%) despite 598 procedures that were performed with a platelet count of less than 50 000/μL and 292 with an INR of greater than 2.0.⁸

Therefore, preprocedure coagulation studies are likely unnecessary prior to paracentesis.

One hundred forty-eight citations of potential studies on the location of paracentesis or the role of ultrasound guidance were found. Three articles met the inclusion criteria.^{18 - 20} In the past, the midline was usually chosen as the site for paracentesis because of its avascularity; however, the ideal site for the procedure remains controversial. This research identified 2 studies that evaluated different locations for possible paracentesis using ultrasonography, but neither study objectively confirmed the site selection with actual paracentesis.^{18 - 19} In a study of 52 patients, the investigators compared the abdominal wall thickness and depth of ascites between the subumbilical midline and the left lower quadrant using ultrasonography. The abdominal wall was thicker (P < .0001) and the depth of the ascites shallower (P < .02) at the midline than at the left lower quadrant in the supine position.¹⁸ In another study of 27 patients, ascites detected by ultrasonography found the location of the distribution too variable to identify the ideal site for blind puncture, particularly in small or moderate ascites.¹⁹

Because the physical examination often fails at detecting smaller amounts of ascites,¹ clinicians recognize that the reference standard for detecting ascites is either ultrasonography, which can reveal as little as 100 mL of ascitic fluid,²¹ or paracentesis. One randomized trial of 100 emergency department patients compared bedside ultrasound-guided paracentesis vs the traditional technique. However, the allocation of randomization was not concealed and the physician performing the procedure completed the randomization by coin toss. The authors did not distinguish between diagnostic vs therapeutic procedures. Of the 56 patients assigned to the ultrasound group, 42 showed identifiable ascites on diagnostic imaging, of whom 40 underwent successful aspiration. In the 44 patients randomized to paracentesis without ultrasound evaluation, 27 underwent successful aspiration. The ultrasound-guided approach yielded a successful aspiration rate of 71% vs 61% without ultrasound (P = .39, using an intention-to-treat analysis). For 15 of the 17 patients in whom the traditional technique failed, the ultrasound technique was later used. In 13 patients, fluid was visualized on ultrasound and all underwent successful aspiration. Therefore, a role may exist for ultrasound guidance when the traditional technique fails. In the same study, the number of passes was recorded in 80 patients. Of these patients, there was no difference in the proportion of patients who required more than 1 attempt between the ultrasound group vs the traditional group (P = 1.00).²⁰

One hundred sixteen citations of potential studies on needle design were identified. Three full-text articles were retrieved.^{13 - 14 ,22} There were no randomized studies comparing needle gauge or length. We found 1 prospective study of 229 paracenteses in which the procedure was first attempted with a 1.5-inch, 22-gauge metal needle, and if unsuccessful, it was reattempted with a 3.5-inch, 22-gauge replacement. Seven procedures required ultrasound for localization. Successful aspiration was achieved in 94% of attempts with the shorter needle, and the remaining 6% of attempts with the 3.5-inch needle.¹³

Two full-text articles describing use of specialized needles were retrieved, but 1 was subsequently discarded because it was not a randomized study.²² In a randomized study of 40 patients with symptomatic ascites, large-volume paracentesis was performed with a 14-gauge, 3.25-inch Angiocath needle with the plastic sheath discarded and compared with use of a 15-gauge, 3.25-inch Caldwell needle/cannula in the other half. Details of the randomization were not clearly stated. Fewer patients in the Caldwell needle/cannula group required multiple peritoneal punctures (1 vs 6; P < .05) and fewer procedures were terminated due to poor fluid return (1 vs 8; P = .02). There was no significant difference in complications or volume of ascites removed.¹⁴

One study that evaluated inoculation of 10 mL of ascitic fluid into blood culture bottles immediately at the bedside vs a 4-hour delayed inoculation in the laboratory was identified.²³ Of the 29 sets that grew a pathogenic organism, 7 grew bacteria in the bedside samples only. There were no sets in which the delayed sample grew bacteria and the bedside sample did not. Thus, the yield on delayed inoculation among patients with bacterial infection was 77% vs 100%, favoring immediate inoculation (95% CI for the difference, 5.3%-40.0%). Quiz Ref IDFor every 5 patients who have their ascitic fluid immediately inoculated into a bedside culture bottle vs delayed inoculation by the laboratory, 1 additional patient with bacterial infection will be detected.

The study strategy identified 1264 citations of potential studies on the use of plasma expanders for therapeutic paracentesis. There was 1 citation in the Cochrane Library; however, it was only at the protocol stage.²⁴ Thirty-six full-text articles were identified for further evaluation and 27 were later excluded for a variety of reasons: comparison was with diuretics,^{25 - 27} intervention of interest was reinfusion of concentrated ascitic fluid,^{28 - 31} intervention of interest was not a plasma expander,^{32 - 33} the authors could not be contacted for primary data,³⁴ and study design was not a randomized trial.^{35 - 51} Of the 9 remaining prospective randomized studies,^{52 - 60} 6 provided details on the methods of randomization^{52 ,54 - 57 ,59} and none described the use of blinded outcomes assessment.

All 9 studies with data from 806 procedures, reported hyponatremia and renal impairment as outcomes of interest. The overall incidence of hyponatremia was 0% to 20% and for renal impairment 0% to 8.8%, depending on the definition of outcome and time to outcome used. There was no significant difference between therapeutic paracentesis with and without volume expansion with albumin, nor with nonalbumin plasma expanders compared with albumin for either hyponatremia or renal impairment (Table 2 and Table 3).^{52 - 60}

Seven studies, with data from 666 procedures, reported encephalopathy as an outcome.^{52 - 58} Encephalopathy occurred in 1.7% to 8.6% of procedures, with no significant advantage with and without volume expansion with albumin, nor with nonalbumin plasma expanders compared with albumin (Table 2 and Table 3).

The incidence of death following therapeutic paracentesis was 0% to 13% in 7 studies, which included 678 procedures.^{53 - 59} The heterogeneity in the risk of death among the studies was likely due to the different time frames used to define the outcome. In one study, for example, several patients died during hospitalization, several weeks after the procedure.⁵⁸ Regardless of the outcome definition used, albumin compared with no plasma expansion or other plasma expanders did not confer any statistically significant difference in the risk of death (Table 2 and Table 3).

Therefore, no existing study comparing albumin with other plasma expanders found a difference in morbidity or mortality following paracentesis. However, there is insufficient evidence comparing plasma expanders to either placebo or saline to be certain that there is no benefit.

Ascitic fluid and serum samples can be subjected to various analyses, but this discussion will concentrate on diagnosing spontaneous bacterial peritonitis or portal hypertension—the most common indications for the procedure. The diagnostic accuracy of the tests are summarized as LRs, the likelihood that a given test result would occur in a patient with the target disorder as opposed to a patient without. The reference standard for diagnosing spontaneous bacterial peritonitis remains controversial. For this analysis, a culture with positive results was used as the reference standard, recognizing that there is no consensus on how to interpret bacterascites or what PMN count level should be used to define a culture with negative results for neutrocytic ascites. The diagnosis of a culture with negative results for neutrocytic ascites is given to patients with an elevated PMN count and cultures with negative results. Conversely, a diagnosis of bacterascites is given to patients who have cultures with positive results without an elevation in the PMN count. The literature has identified, not uncommonly, ascitic patients with “mixed” etiology as the cause of ascites with coincidental portal hypertension. Invasive measures of portal pressure was the reference standard for diagnosing portal hypertension.

Seven hundred sixty-four citations of potential studies on the test accuracy of ascitic fluid WBC and PMN count were found. Twenty-eight full-text articles were identified for further evaluation, and 14 were later excluded for a variety of reasons: retrospective design,^{61 - 65} the study examined serum and not ascitic fluid WBC count,⁶⁶ the study examined lymphocyte count only,⁶⁷ the authors were contacted but no longer had primary data available,^{68 - 70} the authors could not be contacted to clarify if the study was prospective⁷¹ or to obtain primary data,^{5 ,72} or the control group in the study was defined by a PMN count of less than 250 cells/μL.⁷³ Of the remaining 14 prospective studies, all occurred in a clinical setting,^{74 - 87} 5 clearly described the use of consecutive sampling,^{75 - 76 ,80 ,84 ,87} 4 missed a minor proportion of individual test data,^{75 ,79 ,84 - 85} and none mentioned the use of blinding for either the index or reference test.

The studies used various cut-off points for both parameters (Table 4).^{74 - 87} The most accurate parameters to diagnose spontaneous bacterial peritonitis were an ascitic WBC count of greater than 1000 cells/μL (summary LR, 9.1; 95% CI, 5.5-15.1) and an ascitic PMN count of greater than 500 cells/μL (summary LR, 10.6; 95% CI, 6.1-18.3). While the literature supports a threshold of greater than 500 cells/μL as most accurate, most experts initiate antibiotic therapy when the PMN count is greater than 250 cells/μL, given still the greatly increased likelihood of spontaneous bacterial peritonitis at this lower PMN count threshold (summary LR, 6.4; 95% CI, 4.6-8.8) along with the potential fatality of spontaneous bacterial peritonitis.^{88 - 89} Quiz Ref IDConversely, an ascitic PMN count of 250 cells/μL or less lowers the likelihood of spontaneous bacterial peritonitis (summary LR, 0.20; 95% CI, 0.11-0.37).

The search strategy yielded 192 citations and review of these led to inclusion of 9 studies appropriate for data extraction (Table 5).^{68 ,74 - 75 ,78 - 79 ,81 ,83 ,87 ,90} Of the 20 full-text articles reviewed, 11 articles were excluded because the outcome of interest was malignancy and not spontaneous bacterial peritonitis,⁹¹ the data were preliminary results of another study,⁹² the authors were contacted but no longer had primary data available,^{69 ,72} the reference was an editorial of another study,^{93 - 94} the index test was not performed on many patients,⁸² or the authors could not be contacted to clarify the reference standard used,⁹⁵ to verify if the study was prospective,⁷¹ or for primary data.^{72 ,77} Of the remaining 9 prospective studies, all occurred in a clinical setting,^{68 ,74 - 75 ,78 - 79 ,81 ,83 ,87 ,90} 3 explicitly described use of consecutive sampling,^{68 ,75 ,87} all but 1 had complete data for index and reference tests,⁷⁵ and none mentioned the use of blinding for either the index or reference test. The sample sizes were small, ranging from 13 to 285. Of the included studies, the most recent was published in 1995; most other studies meeting inclusion criteria were published more than 20 years ago. We are uncertain about the routine clinical laboratory availability of ascitic fluid pH measurements.

An ascitic fluid pH of lower than 7.35 (summary LR, 9.0; 95% CI, 2.0-40.6) and a blood–ascitic fluid pH gradient of 0.10 or greater (LR, 11.3; 95% CI, 4.3-29.9) were the most accurate thresholds for diagnosing spontaneous bacterial peritonitis as evidenced by having the highest diagnostic odds ratios (calculated by positive LR/negative LR). Conversely, a blood–ascitic fluid pH gradient of lower than 0.10 lowers the likelihood of spontaneous bacterial peritonitis (LR 0.12; 95% CI, 0.02-0.77).

One hundred sixteen citations of potential studies on the test accuracy of the serum-ascites albumin gradient were identified. Review of these led to retrieval of 20 full-text articles for assessment and 4 prospective studies were subsequently identified for inclusion.^{96 - 99} Reasons for excluding trials were inappropriate choice of reference standard or failure to apply the same reference standard to all patients,^{95 ,100 - 105} unclear reference standard,¹⁰⁶ commentary publication,¹⁰⁷ the serum-ascites albumin gradient was used to study malignancy and not portal hypertension,^{73 - 74 ,76 ,91 ,108 - 109} or retrospective design.¹¹⁰

In the remaining 4 prospective studies,^{96 - 99} all were conducted in a clinical environment, 1 clearly described use of consecutive sampling,⁹⁶ all had complete individual data for both the index and reference test, and none described use of blinded assessment for the index or reference test. Quiz Ref IDPooling the 1029 samples from the 4 studies, an abnormal serum-ascites albumin gradient of 1.1 g/dL or greater (≥11 g/L) increases the likelihood of portal hypertension (summary LR, 4.6; 95% CI, 1.6-12.9). A normal serum-ascites albumin gradient (<1.1 g/dL [<11 g/L]) is very useful in decreasing the likelihood of portal hypertension (summary LR, 0.06; 95% CI, 0.02-0.20) (Table 6).^{96 - 99}

Ideally, a successful paracentesis should meet the following criteria: (1) sufficient fluid is obtained on the first attempt; (2) occurs without adverse events to the patient such as bleeding or bowel perforation; and (3) occurs with minimal discomfort to the patient during and after the procedure. The following description of the method to perform a paracentesis considers the best available evidence and expert opinion to facilitate successful completion.

The procedure, including its risks, should be explained to the patient and informed consent obtained if relevant in the practice setting. The explanation should include how the procedure will be performed, why it is being done, level of experience of the clinician, what complications may occur, and how these can be treated. For example, patients can be told that on average, out of 100 paracenteses, 3 will result in a bleeding complication and 1 in a bowel perforation. The risk of death from a diagnostic paracentesis is less than 1.2%. Patients should be asked if they are allergic to any medications including local anesthesia. The patient should urinate prior to the procedure.

The procedure is usually completed with the patient in the supine position. The head of the bed should be elevated 30° to 45° to allow fluid to accumulate in the patient's lower abdomen.¹³

The needle insertion site should be localized to the level of percussible dullness on the abdominal wall. To minimize damage to vascular structures, 2 possible sites may be selected: either a site in the midline 2 cm below the umbilicus or the area of flank dullness in the left lower quadrant lateral to the rectus abdominus muscle (Figure 2). The needle should never be inserted through superficial veins or through surgical scars, since scars may have collateral vessels or underlying adherent bowel.¹³

Wearing sterile gloves, the clinician should cleanse the puncture site with an antiseptic solution. Sterile drapes can be applied, leaving the puncture site exposed. Using a 25-gauge needle, 1 to 3 mL of local anesthetic should be injected (eg, lidocaine) to infiltrate the skin and then into the subcutaneous tissue. Two minutes should be allowed for the anesthetic to take effect. If performing a diagnostic procedure, a 22-gauge, 1.5-inch needle should be attached to a 20- to 50-mL syringe; and if performing a therapeutic tap, a 15-gauge, 3.25-inch Caldwell needle/cannula should be used. Using the Z-tract method, the abdominal skin should be stretched 1 to 2 cm caudad prior to insertion and tension maintained using the nondominant hand. It is thought that this technique may seal the needle tract and prevent persistent leakage of ascitic fluid, although no evidence supports this. The needle should be slowly advanced at a 45°- or 90°-angle while gently aspirating to allow free return of the ascitic fluid, once the peritoneal space is entered. If using the Caldwell needle/cannula, it should be withdrawn, leaving the outer cannula in place. Some clinicians use the angular technique, in which the needle is inserted at a 45°-angle; however no evidence supports this method.

For a diagnostic sample, an appropriate amount should be aspirated into the large syringe. For a therapeutic tap, an evacuation container should be connected using high-pressure connection tubing. After collecting sufficient fluid for analysis, the apparatus should be rapidly removed allowing the skin to resume its natural position. An adhesive bandage should be placed over the insertion site. If there is leakage of ascitic fluid, a pressure dressing should be applied.

If the first attempt fails, a reattempt should be done with landmarking of another site. If the patient has an excessively thick panniculus, the clinician should change to a 3.5-inch needle of equal gauge. There are also commercially available prepackaged paracentesis kits with catheter-over-needle devices. Alternatively, the procedure may be attempted by another individual or with ultrasound guidance.

Normal ascitic fluid should be clear. Ascitic fluid (10 mL) should be immediately inoculated into 1 aerobic and 1 anaerobic blood culture bottle at the bedside.¹³ Depending on the purpose of the paracentesis, the remainder of the ascitic fluid can be sent to the laboratory for obtainment of a cell count, Gram stain, pH, and albumin. Serum samples such as arterial pH and albumin that are used to calculate serum-ascitic fluid gradients should be drawn within 30 minutes of the paracentesis.

The American Board of Internal Medicine no longer recommends 3 directly supervised, successfully performed abdominal paracenteses as a minimum standard of competency.¹¹¹ Instead, the organization recommends that trainees be able to explain indications, contraindications, patient preparation, sterile techniques, pain management, handling of specimens and fluids obtained, interpretation of test results, and management of complications. These newer standards (revised in 2006 and updated in November 2007) were developed using expert consensus. In a survey, 71% of program directors agreed that all of their residents master the procedure.¹¹² A survey of 242 internal medicine residents found that they needed to complete 3 to 5 procedures to feel confident in their performance.¹¹³ In another survey of 128 graduating internal medicine residents, a median of 15 paracenteses were completed at the time of residency completion and 87% of residents felt confident in performing the procedure.¹¹⁴ While performing the paracentesis is not an explicit requirement of the American Board of Internal Medicine, we anticipate that many program directors will encourage their trainees to develop competency in this skill.

Only descriptive studies, with little evaluation of the various methods, provide evidence to guide the teaching of paracentesis. A comprehensive, cadaver-based workshop for first-year internal medicine residents was reported to improve residents' knowledge and comfort level in performing invasive bedside procedures including paracenteses; however, this study evaluated student self-reported data rather than objective measurement of technical skill proficiency.¹¹⁵ Other reports have cited the use of unembalmed cadavers as a useful educational approach to teaching paracentesis; but there has not been any rigorous evaluation of these methods.^{116 - 117} One study described the implementation of a pilot medical procedure service. Select faculty ensured familiarity with relevant anatomy, taught residents the psychomotor skills required for a specific procedure, and identified and managed complications. A complementary Web-based, multimedia curriculum outlining the indications, contraindications, risks and benefits, and anatomical landmarks of each procedure was also developed. Experienced hospitalists supervised 37 paracenteses performed by trainees, of which only 1 procedure resulted in serious complication (hemorrhage), for a complication rate (2.7%) consistent with other studies. Eighty-seven percent of the housestaff believed this procedure service helped in their education of medical procedures.¹¹⁸ However, there was no control group with housestaff who did not complete this program available for comparison.

This review highlights several gaps in the literature pertaining to the performance and interpretation of paracentesis. Future areas of study include how to teach and learn the procedure to optimize performance and determination of optimal needle gauge and length, amount of ascites removal at which plasma expanders should be considered, and best procedure location.

The patient in this study had a fever and ascites. Blood work included arterial pH and albumin measurement. A diagnostic paracentesis was completed after obtaining informed consent. Cloudy ascitic fluid was sent to the laboratory to obtain its cell count and differential, pH, albumin, Gram stain and culture. The patient's ascitic fluid PMN count was 623 cells/μL. Broad spectrum antibiotic therapy was initiated for presumed spontaneous bacterial peritonitis while awaiting culture results.

Quiz Ref IDA diagnostic paracentesis should be performed at the time of admission on any patient with ascites, when there is a clinical suspicion of spontaneous bacterial peritonitis. Ascitic fluid should be directly inoculated into blood culture bottles at the bedside.

In patients suspected of having spontaneous bacterial peritonitis, this study describes the appropriate laboratory results that represent greater likelihood of spontaneous bacterial peritonitis (Table 7). In patients with ascites, in whom portal hypertension is considered, a serum-ascites albumin gradient of 1.1 g/dL or greater (≥11 g/L) makes portal hypertension more likely (summary LR, 4.6; 95% CI, 1.6-12.9), while a gradient of less than 1.1 g/dL (<11 g/L) may rule it out when the prior probability estimate of portal hypertension is low (summary LR, 0.06; 95% CI, 0.02-0.20).