창고지기

INTRODUCTION/DEFINITIONS — Diverticulitis represents micro- or macroscopic perforation of a diverticulum. It was previously believed that obstruction of diverticula, eg, by fecaliths, increased diverticular pressure and caused perforation; such obstruction is now thought to be rare. The primary process is thought to be erosion of the diverticular wall by increased intraluminal pressure or inspissated food particles. Inflammation and focal necrosis ensue, resulting in perforation. (See "Epidemiology and pathophysiology of colonic diverticular disease").

From a therapeutic viewpoint, diverticulitis can be divided into complicated and uncomplicated presentations (show figure 1). (See "Clinical manifestations and diagnosis of colonic diverticular disease").

• Complicated diverticulitis refers to the presence of a perforation, obstruction, an abscess, or a fistula. Approximately 25 percent of patients diagnosed with diverticulitis for the first time present with complicated diverticulitis. Nearly all of these patients require surgery.

• Uncomplicated diverticulitis, accounting for 75 percent of cases, refers to diverticulitis without the complications noted above. The majority of these patients respond to medical therapy, although up to 30 percent require surgery.

This topic review will focus on the management of patients with acute diverticulitis. Several major medical organizations have also proposed recommendations, which are available online [1-5]:

• The American Society of Colon and Rectal Surgeons (www.fascrs.org)

• The Society for Surgery of the Alimentary Tract (www.ssat.com)

• The American College of Gastroenterology (www.acg.gi.org/)

• European Association of Endoscopic Surgeons (www.eaes-eur.org/).

The guidelines from the American Society of Colon and Rectal Surgeons and the American College of Gastroenterology are particularly useful in providing the supporting documentation used by the panels.

DIAGNOSIS — The diagnosis of acute diverticulitis is often suspected based upon the history and the physical examination. In the acute stage, further studies are performed to confirm the diagnosis and to rule out other sources of acute abdominal signs. Computer tomographic scanning has become the optimal method of investigation in patients suspected of having acute diverticulitis. (See "Clinical manifestations and diagnosis of colonic diverticular disease").

UNCOMPLICATED DIVERTICULITIS — Conservative treatment (with bowel rest and antibiotics) is successful in approximately 70 to 100 percent of patients with acute uncomplicated diverticulitis [6-13]. A CT scan can be helpful for identifying patients who are likely to respond to conservative therapy [8,9]. This was illustrated in a study that included 56 patients with acute diverticulitis, 37 of whom were identified as having uncomplicated disease by CT [9]. All of these 37 were treated successfully with nonsurgical therapy and discharged in an average of 6.8 days. Nineteen patients had complicated disease (abscess, fistula, peritonitis, obstruction) and were treated appropriately. Three of the initial cohort of 59 patients (5 percent) had a mistaken clinical diagnosis of acute diverticulitis. Similar findings were noted in another series in which medical therapy was successful in 97 percent of patients with uncomplicated diverticulitis by CT [8]. (See "Clinical manifestations and diagnosis of colonic diverticular disease").

Several issues must be addressed when caring for patients with uncomplicated diverticulitis, including whether treatment can be achieved safely in an outpatient, the selection of specific antibiotics, dietary recommendations, the need for follow-up investigation, the likelihood of recurrence, and the choice and timing of operative intervention in those who do not respond adequately.

Selection for outpatient management — The decision regarding whether to hospitalize a patient with diverticulitis depends upon several factors, including the severity of presentation, the ability to tolerate oral intake, the presence of comorbid diseases, and the available support system. Patients selected for outpatient management should be reliable and understand the indications for seeking immediate medical attention. These include an increase in fever or abdominal pain or the inability to consume adequate fluids. As a general rule, the elderly, immunosuppressed, those with significant comorbidities, and those with high fever or significant leukocytosis should be hospitalized.

Choice of antibiotics — Diverticulitis represents a localized infection that requires treatment with antibiotics. The choice of antibiotics should be based upon the usual bacteria, which are principally Gram negative rods and anaerobes (particularly E. coli and B. fragilis). Reasonable choices include a quinolone with metronidazole, amoxicillin-clavulanate, or sulfamethoxazole-trimethoprim with metronidazole. The selection may be influenced by the patient's history of allergies or previous use of antibiotics.

Our usual outpatient regimen includes:

• Ciprofloxacin, 500 mg PO twice daily plus metronidazole, 500 mg PO three times daily, although amoxicillin-clavulanate (875/125 mg twice daily) is an acceptable alternative. Treatment should be continued for 7 to 10 days. Oral ciprofloxacin achieves levels similar to those with intravenous administration, has broad coverage of enteric Gram negative pathogens, and (similar to amoxicillin-clavulanate) requires only twice daily dosing, thus improving compliance. For patients intolerant to metronidazole, clindamycin may be an acceptable alternative. For patients intolerant to metronidazole as well as beta lactam agents, moxifloxacin has reasonable gram-negative and anaerobic coverage.

Patients requiring hospitalization should receive empiric broad-spectrum intravenous antibiotics directed principally at colonic anaerobic and gram-negative flora until culture results of any abscess obtained by percutaneous aspirate or surgical drainage are available. Metronidazole is the antibiotic of choice for anaerobic coverage, while gram-negative coverage can be achieved with a third-generation cephalosporin (ceftriaxone 1 to 2 g daily or cefotaxime 1 to 2 g every six hours) or a fluoroquinolone (ciprofloxacin 400 mg IV every 12 hours or levofloxacin 500 mg IV daily). Short courses (five to seven days) of aminoglycoside-containing regimens are reasonably safe in patients with normal renal function; however longer courses of therapy should be avoided when possible (when alternative drugs are available).

Single agent coverage is also reasonable with the following alternative agents:

• Beta lactamase inhibitor combination such as ampicillin-sulbactam (3 g every six hours), piperacillin-tazobactam (3.375 g or 4.5 g every six hours), or ticarcillin-clavulanate (3.1 g every four hours)

• Carbapenem such as imipenem (500 mg every six hours) or meropenem (1 g every eight hours)

Coverage with a single antibiotic with activity against colonic flora is as effective as combination therapy [10,14]. For patients intolerant to the above agents, tigecycline may be a reasonable alternative agent.

Regardless of which empiric regimen is chosen initially, therapy should be altered to reflect the organisms recovered if aspiration or drainage is performed.

The second-generation cephalosporin cefoxitin and the cephamycin cefotetan cover anaerobes but are not sufficiently active for Gram negative aerobic bacteria to use empirically for serious infection.

Dietary recommendations — Outpatients should be instructed to consume clear liquids only. Clinical improvement should be evident after two to three days, after which the diet can be advanced slowly. Patients requiring hospitalization can be treated with clear liquids or NPO with intravenous hydration.

The role of dietary therapy in the prevention of recurrence has not been established in well-designed randomized controlled trials with long-term follow-up. Nevertheless, patients should generally be advised to consume a high fiber diet once the acute phase has resolved. This recommendation is based mostly upon uncontrolled studies, which have suggested that long-term fiber supplementation may reduce the incidence of recurrence [11,15]. Indirect support for fiber supplementation can also be inferred from studies that have demonstrated a protective effect from diets high in fiber on the development of diverticular disease [16]. Two small controlled trials on fiber supplementation for prevention of recurrence had conflicting results [17,18].

Seeds and nuts — Patients with diverticular disease have historically been advised to avoid whole pieces of fiber (such as seeds, corn, and nuts) because of concern that the undigested fragments could become lodged within a diverticulum, thereby inciting an episode of diverticulitis. The commonly heard advice to avoid small undigestible foods (such as seeds) because they may theoretically become lodged in a diverticulum is completely unproven and is probably little more than an old wives' tale. However, there continues to be considerable practice variation with regard to this recommendation. In a survey of colorectal surgeons, one-half believed that the avoidance of seeds and nuts was of no value [19].

The authors have seen tens of thousands of diverticula and never seen a single seed. Furthermore, in a large observational study (The Health Professionals Follow-up Study) there was an inverse association between nut and popcorn consumption and the risk of diverticulitis and diverticular bleeding [20]. In addition, no association was found between consumption of corn and diverticulitis or between nut, corn or popcorn consumption and diverticular bleeding or uncomplicated diverticulosis. Thus, we do not suggest that patients with diverticulosis avoid seeds, corn and nuts.

Monitoring the clinical course — Outpatients should be advised to call for increasing pain, fever, or the inability to tolerate fluids, all of which may be an indication for hospitalization. Complications should be sought in all patients with clinical deterioration or those who fail to improve within two to three days. Abscesses amenable to percutaneous drainage should be treated. Laparotomy is necessary if there is no reversible condition, if abscesses cannot be drained, or if drainage does not result in improvement (show table 1). Surgery should proceed without delay if a patient's condition deteriorates (increased pain, more localized peritonitis or diffuse tenderness, increased white count), and there is no reversible condition.

Alternative diagnoses should also be considered. These include acute appendicitis, Crohn's disease, colon cancer, ischemic colitis, pseudomembranous colitis, complicated ulcer disease, ovarian cyst or abscesses or torsion, and ectopic pregnancy.

Recommendations following resolution — Two to six weeks after recovery, patients should undergo an evaluation of the colon to exclude other diagnostic considerations (such as colonic neoplasia) and to evaluate the extent of the diverticulosis (information that may be important to direct therapy should future complications of diverticulosis occur). This is usually accomplished with a colonoscopy, although a flexible sigmoidoscopy plus barium enema is a reasonable alternative. As noted above, patients should be advised to consume a diet high in fiber. (See "Patient information: Diverticular disease").

Prognosis after resolution — Following successful conservative therapy for a first attack of diverticulitis, 30 to 40 percent of patients will remain asymptomatic, 30 to 40 percent will have episodic abdominal cramps without frank diverticulitis, and one-third will proceed to a second attack of diverticulitis [11,21-23]. It is generally believed that the prognosis is worse with a second attack, since some studies have reported that the rate of complicated diverticulitis in such patients approaches 60 percent and the mortality rate is doubled [21,24,25]. These data have been the foundation for recommending an operation after a second attack.

However, more recent data suggest that the prognosis after a second attack may be better than generally recognized [26,27]. This was illustrated in a series of 110 patients who were treated conservatively for a first episode of diverticulitis [26]. Follow-up was complete in 83 patients, 15 of whom (18 percent) recurred once while six (7 percent) recurred twice or more during a median of 10.5 years. Only nine of these patients required inpatient therapy; two underwent operation and none required an emergency procedure. The authors concluded that recurrent diverticulitis can be treated in a similar fashion to a first episode and that medical therapy remains a valuable alternative to surgery. This study was useful in having long median follow-up. However, 28 of 83 patients (34 percent) died during this period. The cause of death was unrelated to diverticulitis in 25 and unknown in 3. Although this high death rate was perhaps not unexpected because the median age at first recurrence was 70 years, it limits the conclusions that can be drawn regarding the natural history of the disease. In another report, morbidity and mortality were not significantly different between patients with more than two episodes of diverticulitis compared with those with one or two prior attacks [27].

Thus, elective surgery is not necessary for all patients who respond to medical therapy. However, the possible increased risk of complications with a second episode emphasizes the importance of identifying patients who are more likely to have a recurrence after a first attack of diverticulitis and those who are at increased risk for complications. Patients in whom elective surgery has been recommended following a single attack of diverticulitis include young patients (variously defined in the literature as less than 40 or 50 years of age) and those who are immunosuppressed. However, the issue of resection for young patients is controversial (see "young patients" below).

Those who are treated operatively are generally felt to be cured, with progression of diverticulosis in the remaining colon occurring in only 15 percent [28], and a need for further surgery in only 2 to 11 percent [28-30]. However, up to 27 percent of these patients may describe abdominal pain postoperatively in the same location; these persistent symptoms may be explained by coexistent irritable bowel syndrome rather than by recurrent diverticulitis.

There may be a small number of patients who do not follow the classic pattern of either resolving with antibiotics or requiring operation. Such patients have been described as having "smoldering" diverticulitis. One of the largest series included 47 patients who had ultimately undergone sigmoid resection [31]. Patients were selected from a group of 930 patients who had undergone surgical resection for diverticular disease during the study period; thus they represented about 5 percent of patients overall.

Acute or chronic inflammatory changes were present in 76 percent of resected specimens. Complete resolution of symptoms occurred in 77 percent, with 88 percent becoming pain free. The authors concluded that smoldering diverticular disease is uncommon and poorly defined but that resection can be beneficial. Despite these findings, a diagnosis of smoldering diverticulitis must be made carefully, as less stringently selected patients may have irritable bowel syndrome or other causes of symptoms. Furthermore, since all patients had undergone resection, this study does not provide insight into the natural history of this disorder.

COMPLICATED — Additional factors must be considered in patients with one of the complications of diverticulitis, such as obstruction, perforation, abscess formation, or fistula formation.

Peritonitis — Diffuse peritonitis mandates resuscitation, broad-spectrum antibiotics, and emergency exploration; diagnostic studies are rarely necessary. Examples of antibiotic regimens we have used are:

• Ampicillin (2 g IV every six hours), gentamicin (1.5 to 2 mg/kg IV every 8 hours), and metronidazole (500 mg IV every 8 hours)

• Imipenem/cilastin (500 mg IV every six hours)

• Piperacillin-tazobactam (3.375 g IV every six hours)

Perforated diverticulitis has a mortality rate of 6 percent for purulent peritonitis and 35 percent for fecal peritonitis (peritonitis with fecal soilage) [32].

The importance of administering an appropriate initial empiric antibiotic regimen was illustrated in a review of 425 patients who required surgery for community-acquired secondary peritonitis, including patients with diverticulitis [33]. In 13 percent of patients, the initial antibiotic regimen was inappropriate, defined as not covering all bacteria subsequently isolated or not covering both aerobic and anaerobic organisms in the absence of culture results. Resolution of the infection with initial or step-down therapy after primary surgery was significantly less likely with an inappropriate regimen (53 versus 79 percent); failure to achieve such clinical success was associated with a six day prolongation in hospitalization (20 versus 14 days).

Obstruction — Diverticular obstruction is rarely complete, usually allowing for bowel preparation. A primary concern is differentiation from carcinoma; even with negative biopsies, resection is mandatory in lesions in which there is suspicion for malignancy based upon their appearance alone (show radiograph 1). Resection with primary anastomosis is usually possible; a colostomy is required if the preparation is inadequate or on table colonic lavage can be considered to permit primary anastomosis.

Perforation — Free intraperitoneal rupture of diverticulitis is unusual; however, such patients have high mortality rates of 20 to 30 percent [32,34]. Treatment usually involves a two-stage procedure as described below.

Abscess — Abscesses occur in 16 percent of patients with acute diverticulitis without peritonitis [35] and in 31 to 56 percent of those requiring surgery for diverticulitis [23]. Prior to interventional radiologic techniques, abscesses mandated operative intervention, often with two-stage procedures. Percutaneous drainage now permits elective single stage surgery in 60 to 80 percent of patients [36-38]; furthermore, in selected patients with contraindications to surgery, catheter drainage may be sufficient to relieve symptoms [37]. Drainage is usually performed through the anterior abdominal wall; abscesses deep in the pelvis or obscured by other organs may be accessed transgluteally, or through the rectum or vagina [39,40].

The catheter is left in place until drainage is less than 10 mL in 24 hours; this may take as long as 30 days [37]. Catheter sinograms during this period can show persistent communication between fistula and bowel, and permit assessment of resolution of the abscess cavity. Surgical intervention is mandatory if improvement does not occur.

Until recently, all abscesses have been considered to represent complicated diverticulitis, and surgery has typically been recommended once resolution has occurred. However, with continued improvements in CT scan technology, smaller abscesses can be identified. Some of these are so small (ie, less than 3 to 4 cm) that they are not amenable to CT guided drainage. Physicians have by necessity moved to treating such patients with antibiotics without drainage. There is some evidence to suggest that these patients have a similar course as those with uncomplicated diverticulitis and can thus avoid an operation after a first attack. Further research is necessary and is actively being encouraged by one of the national colorectal surgical societies.

Fistulas — Diverticulitis involves the sigmoid colon in approximately 90 percent of cases. Thus, fistulas arising from diverticulitis usually involve the sigmoid colon. The approach to this problem is discussed separately. (See "Acute diverticulitis complicated by fistula formation").

PRINCIPLES OF SURGERY — As noted above, up to 30 percent of patients with uncomplicated diverticulitis require surgical intervention during the initial attack. The role of surgery in patients who respond to conservative therapy depends upon the clinical setting. Surgery has generally been advised after a first attack of complicated diverticulitis or after two or more episodes of uncomplicated diverticulitis. However, the risk and benefits of surgery should be tailored to the individual patient [41].

Growing experience with laparoscopic techniques for performing colonic resections suggests that it is associated with a shorter recovery time than traditional open approaches [5,42-49]. Laparoscopic resection appears to be best suited for patients in whom the episode of acute diverticulitis has resolved and in patients with stage I or II disease according to the Hinchey classification (see "Intraoperative assessment" below) provided that a surgeon experienced with laparoscopic surgery of the colon is available [5].

The goals of surgery are to remove the septic focus by resection of the colon, to treat obstruction or fistula, and to restore bowel continuity while minimizing morbidity and mortality. Surgical mortality rate is approximately 1.3 to 5 percent depending upon the severity of illness and the presence of comorbidities [24,50].

Preoperative preparation — All patients receive antibiotics, a second or third generation cephalosporin or more broad-spectrum antibiotics, depending upon the degree of contamination. Examples of regimens we have used are: cefazolin (1 g IV every eight hours) plus metronidazole (500 mg IV every 8 hours); ampicillin-sulbactam (1.5 g IV every six hours); or ticarcillin-clavulanate (3.1 g IV every six hours). For patients with more extensive contamination, we use the regimens described above for peritonitis.

Bowel preparation is often possible in nonemergent situations. Patients should be advised before surgery of the possibility of a stoma and the abdominal wall should be marked. The modified lithotomy position allows the use of a circular stapler if an anastomosis is possible. Intraoperative proctoscopy permits assessment and emptying of the rectum.

Intraoperative assessment — In the emergency setting, the abdomen is explored through a midline incision; in the elective setting (usually at least six weeks after the episode of diverticulitis has resolved) a laparoscopic approach may be possible. Assessment of peritoneal contamination using Hinchey's classification determines the advisability of a primary anastomosis [51]:

• Stage I — pericolic or mesenteric abscess
• Stage II — walled-off pelvic abscess
• Stage III — generalized purulent peritonitis
• Stage IV — generalized fecal peritonitis

In elective or semielective situations, resection and primary anastomosis are often possible, since the disease is well-localized or has significantly resolved. Patients with Hinchey stages I and II disease can usually have a bowel preparation preoperatively. The bowel must be well-vascularized, nonedematous, tension-free, and well-prepared. The distal resection margin is in the upper third of the rectum, where the taenia coalesce; the proximal margin is where the colon becomes soft and nonedematous. It is not necessary to remove all diverticula-bearing colon proximal to the intended anastomosis to prevent recurrence since diverticula proximal to the descending/sigmoid colon are unlikely to result in further symptoms [30]. Contraindications to primary anastomosis are fecal or purulent peritonitis, associated medical conditions, poor nutrition, and immunosuppression [52].

Two-stage procedures are typically used in emergency situations with peritoneal contamination (show figure 2). A common approach is the Hartmann procedure, which involves resection of the diseased colon, an end-colostomy, and creation of a rectal stump; this is followed by colostomy closure three months later [53]. Another approach entails resection, primary anastomosis, and proximal diverting stoma (colostomy or ileostomy); the second stage is stoma closure. The latter approach is used when there are relative contraindications to primary anastomosis, but no purulent or fecal peritonitis and the bowel is nonedematous. It is preferred to the first approach in these settings because of the difficulty of reversing a Hartmann procedure after generalized peritonitis [22,23]. Although still very controversial, there is increasing experience with primary anastomosis in the acute setting in the presence of an unprepared colon [54]. As an alternative, on-table colonic lavage may be considered [55].

Resection, diverting colostomy and a Hartmann rectal stump is the preferred approach in patients with fecal peritonitis and in most cases with purulent peritonitis (show figure 2). A mucous fistula, in which the distal end of the transected bowel is brought through the abdominal wall, is often not possible after resection of the entire sigmoid. Many surgeons mark the rectal stump with a long nonabsorbable suture and tack it to the anterior abdominal wall or sacral promontory.

The classic three-stage procedure is now rarely indicated [56] because it is associated with a higher mortality rate (26 versus 7 percent after resection) [32]. The first stage is drainage of the diseased segment, plus a proximal diverting stoma, without resection. This is followed by resection with primary anastomosis, and then by closure of the stoma. In exceptional cases, when inflammation precludes safe dissection of iliac vessels and ureters or the patient is unstable, it may be necessary to drain the area and create a proximal stoma. Drainage and diversion may also be performed to stabilize the patient prior to transfer to a large center.

If a mucous fistula is created, a portion of the rectosigmoid and possibly the sigmoid must, by definition, remain for it to reach the anterior abdominal wall. Thus, any patient with a mucus fistula will require additional resection of the rectosigmoid down to the proximal rectum at the time of colostomy closure (reversal of Hartmann's procedure) to avoid retention of diverticula that may subsequently cause further symptoms.

SPECIAL CONSIDERATIONS — The approach recommended thus far may need to be amended in selected patients, including those under age 40, those who are immunosuppressed, and those with right-sided diverticulitis.

Young patients — The optimal management of patients with acute diverticulitis who are under age 40 is controversial. Resection had traditionally been recommended after one episode, an approach that was based upon two premises: these patients have more virulent disease; and those who respond to medical therapy are more likely to have recurrent diverticulitis and ultimately require surgical intervention. However, although certain features are common to most series, there is dissent as to the virulence of the disease and its natural history.

Young patients with diverticulitis comprise 12 to 29 percent of large series [52,57-60]. There is a male predominance, ranging from 2:1 to 4:1; in addition, the great majority of patients are obese [24,34].

Large series of all patients with diverticulitis note an operative rate of 27 to 33 percent [21,23]. The thought that diverticulitis is more virulent in the young arises from studies with operative rates of 48 to 88 percent in these patients [52,58,60,61]; however, others describe rates of surgery similar to those in older patients [57,59]. One problem may be that many young patients undergo urgent surgery because of an incorrect preoperative diagnosis [62]; in some series with high rates of surgery, the incidence of incorrect diagnosis has been as high as 41 to 50 percent [52,58]. If CT or contrast studies are used to establish the diagnosis, the operative rate in young patients may actually be lower than that in older patients (15 versus 33 percent in one series) [57].

The natural history after medical management of young patients is also disputed. Some studies report rates of readmission of 55 percent [63,64] and subsequent operative rates of 20 to 41 percent [62,63,65]; others report no subsequent surgery after four years [66]. When total operative rates are examined, the data become more consistent, with an eventual operative risk, on first admission or later, of about 50 percent. Thus, some series have high initial operative rates [52,58,60], while others have a lower rate on admission of 15 to 25 percent, but with poor outcome or later surgery in approximately 30 percent [57,65,67].

In summary, it is unclear if young patients suffer more virulent disease. The literature suggests an ultimate operative rate of about 50 percent; with wider use of CT, it is possible that, although the rate of surgery on the first admission may fall, a larger proportion of patients treated medically may ultimately require resection.

There are no convincing arguments to insist on resection after resolution of uncomplicated diverticulitis in a young patient. Once the patient is advised of risks, the issue becomes one of patient evaluation of risk and lifestyle concerns. In patients with no comorbid conditions, elective surgery after one episode remains a reasonable recommendation; others with the same risk are equally justified in wishing to wait.

Immunosuppression — Immunosuppression is associated with an increased incidence of perforated diverticulitis. Such patients include those on chemotherapy or long-term corticosteroid therapy, but this problem is also apparent in diabetics, those with renal failure, and patients with collagen-vascular disorders such as lupus [68,69].

Immunosuppressed patients may present with minimal symptoms or signs even with frank peritonitis, and diagnosis is frequently delayed. Surgery is necessary in almost all immunocompromised patients, and early surgical intervention should be considered. In one study, for example, medical therapy for acute diverticulitis was successful in 76 percent of 76 nonimmunocompromised patients versus none of 10 immunocompromised patients [70].

Within the limits of currently available data, most surgeons would give serious consideration to surgical resection after a single attack of diverticulitis in any patient who is already immunosuppressed or about to start taking immunosuppressive medications (eg, pretransplant patients). This is probably not the case, however, for patients with HIV/AIDS where further data are needed. It is also not the case for patients undergoing a limited course of chemotherapy, where elective resection would require one month off chemotherapy before operation and another month for recovery before chemotherapy is resumed. In these cases, completion of the course of chemotherapy is probably more pressing.

Right-sided diverticulitis — Right-sided diverticula account for about 5 percent of diverticulosis in Western countries [71] and 1.5 percent of cases of diverticulitis [72]. However, right-sided disease is much more common in Asia, accounting for up to 20 percent of cases of diverticular disease and 75 percent of those with diverticulitis [73,74]. Most right-sided lesions are thought to be false diverticula [75].

Patients with right-sided diverticulitis tend to be younger than those with left-sided disease [73,74]. In published series, the correct preoperative diagnosis is made in only 4 to 16 percent of cases; most patients are thought to have acute appendicitis [74-76].

Recommendations range from medical therapy if the diagnosis is made preoperatively, to appendectomy and antibiotic therapy if the diagnosis is made at laparotomy [73,74], to diverticulectomy for minimal inflammation [73], and to right hemicolectomy for extensive inflammation or a mass suggestive of carcinoma [76]. If carcinoma cannot be ruled out at operation, resection may proceed if there are no contraindications to anastomosis [76]. More extensive evaluation should be performed after recovery from the acute phase. Recommendations for medical therapy were supported by a study from the Netherlands that included 41 patients with right-sided diverticulitis who were treated conservatively [77]. During up to 11 years of follow-up, only five patients developed recurrent symptoms, two of whom underwent elective surgery. Of interest, serial ultrasound and CT examinations demonstrated the eventual spontaneous evacuation of the contents of the inflamed diverticulum into the colonic lumen.

INFORMATION FOR PATIENTS — Educational materials on this topic are available for patients. (See "Patient information: Diverticular disease"). We encourage you to print or e-mail this topic review, or to refer patients to our public web site, www.uptodate.com/patients, which includes this and other topics.

SUMMARY AND RECOMMENDATIONS — From a therapeutic viewpoint, diverticulitis can be divided into complicated and uncomplicated presentations (show figure 1). (See "Clinical manifestations and diagnosis of colonic diverticular disease").

• Complicated diverticulitis refers to the presence of a perforation, obstruction, an abscess, or a fistula.

• Uncomplicated diverticulitis, accounting for 75 percent of cases, refers to diverticulitis without the complications noted above.

Uncomplicated diverticulitis

• We suggest conservative treatment (with bowel rest and antibiotics) rather than surgery in patients with uncomplicated diverticulitis (Grade 2B). Such an approach is successful in approximately 70 to 100 percent of patients. (See "Uncomplicated diverticulitis" aboveSee "Uncomplicated diverticulitis" above).

Several issues must be addressed, including whether treatment can be achieved safely in an outpatient setting, the selection of specific antibiotics, dietary recommendations, the need for follow-up investigation, the likelihood of recurrence, and the choice and timing of operative intervention in those who do not respond adequately.

Outpatients should be instructed to consume clear liquids only. Clinical improvement should be evident after two to three days, after which the diet can be advanced slowly. Patients requiring hospitalization can be treated with clear liquids or NPO with intravenous hydration.

The choice of antibiotics should be based upon the usual bacteria, which are principally Gram negative rods and anaerobes (particularly E. coli and B. fragilis). Our usual outpatient regimen includes ciprofloxacin, 500 mg PO twice daily plus metronidazole, 500 mg PO three times daily, although amoxicillin-clavulanate (875/125 mg twice daily) is an acceptable alternative. Treatment should be continued for 7 to 10 days. (See "Choice of antibiotics" above).

Outpatients should be advised to call for increasing pain, fever, or the inability to tolerate fluids, all of which may be an indication for hospitalization. Complications should be sought in all patients with clinical deterioration or those who fail to improve within two to three days.

• We suggest that two to six weeks after recovery, patients undergo an evaluation of the colon to exclude other diagnostic considerations (such as colonic neoplasia) and to evaluate the extent of the diverticulosis (information that may be important to direct therapy should future complications of diverticulosis occur) (Grade 2C).

Following successful conservative therapy for a first attack of diverticulitis, 30 to 40 percent of patients will remain asymptomatic, 30 to 40 percent will have episodic abdominal cramps without frank diverticulitis, and one-third will proceed to a second attack of diverticulitis.

Complicated diverticulitis — Management of complicated diverticulitis (including obstruction, perforation, abscess, or fistula formation) is described above. (See "Complicated" above").

Surgery — Up to 30 percent of patients with uncomplicated diverticulitis require surgical intervention during the initial attack. The role of surgery in patients who respond to conservative therapy depends upon the clinical setting. Surgery has generally been advised after a first attack of complicated diverticulitis or after two or more episodes of uncomplicated diverticulitis. However, the risk and benefits of surgery should be tailored to the individual patient.

INTRODUCTION — Colonic diverticular disease is common. The prevalence is age-dependent, increasing from less than 5 percent at age 40, to 30 percent by age 60, to 65 percent by age 85. Among all patients with diverticulosis, 70 percent remain asymptomatic, 15 to 25 percent develop diverticulitis, and 5 to 15 percent develop some form of diverticular bleeding (show figure 1). (See "Epidemiology and pathophysiology of colonic diverticular disease").

This topic review will focus on the clinical manifestations and diagnosis of symptomatic diverticular disease. The treatment of diverticulitis and other complications of diverticular disease are discussed separately. (See appropriate topic reviews).

CLINICAL MANIFESTATIONS — The clinical manifestations of diverticulosis depend upon the presence or absence of complications. Uncomplicated diverticulosis is often an incidental finding (show endoscopy 1). Some of these patients complain of symptoms such as cramping, bloating, flatulence, and irregular defecation. However, it is unclear if these symptoms are attributable to the underlying diverticulosis or to coexistent irritable bowel syndrome. (See "Clinical manifestations and diagnosis of irritable bowel syndrome"). The major complications of diverticular disease are diverticulitis and diverticular bleeding. Some authors have noted that complications are more likely in individuals who are obese, but the strength of this association is uncertain [1,2].

Diverticulitis — Diverticulitis represents micro- or macroscopic perforation of a diverticulum (show picture 1). The primary process is thought to be erosion of the diverticular wall by increased intraluminal pressure or inspissated food particles; inflammation and focal necrosis ensue, resulting in perforation.

The inflammation is frequently mild, and a small perforation is walled off by pericolic fat and mesentery. This may lead to a localized abscess or, if adjacent organs are involved, a fistula or obstruction. In comparison, poor containment results in free perforation and peritonitis.

The clinical presentation of diverticulitis depends upon the severity of the underlying inflammatory process and whether or not complications are present. Complicated diverticulitis refers to the presence of an abscess, fistula, obstruction, or perforation while simple diverticulitis refers to inflammation in the absence of these complications.

Left lower quadrant pain is the most common complaint in Western countries, occurring in 70 percent of patients. Pain is often present for several days prior to admission, which aids in the differentiation of diverticulitis from other causes of acute abdominal symptoms. Only 17 percent in one series had symptoms for less than 24 hours [3]. Another helpful diagnostic finding is that up to one-half have had one or more previous episodes of similar pain. Other possible symptoms include nausea and vomiting in 20 to 62 percent, constipation in 50 percent, diarrhea in 25 to 35 percent, and urinary symptoms (eg, dysuria, urgency and frequency) in 10 to 15 percent [4].

Right-sided diverticulitis occurs in only 1.5 percent of patients in Western countries [5] but is more common in Asians (accounting for as many as 75 percent of cases of diverticulitis). Affected patients may present with right lower quadrant pain, often leading to a misdiagnosis of acute appendicitis [6,7]. (See "Treatment of acute diverticulitis", section on Right-sided diverticulitis).

The physical examination usually reveals abdominal tenderness, characteristically in the left lower quadrant [5]. A tender mass is palpable in about 20 percent [8] and abdominal distention is common. Right lower quadrant tenderness usually results from redundant sigmoid colon or right-sided diverticulitis which, as noted above, is rare in the west but common in Asia [5-7]. Generalized tenderness suggests free perforation and peritonitis.

Low grade fever and mild leukocytosis are common. However, their absence does not exclude the diagnosis; in one series, for example, 45 percent had a normal white count [9]. Other blood tests that are commonly ordered in the setting of a patient with acute abdominal pain include liver function tests (LFTs) and amylase. LFTs are usually normal and amylase is either normal or mildly elevated, especially in the patient with perforation and peritonitis. The urinalysis may reveal sterile pyuria induced by adjacent inflammation; the presence of colonic flora on culture suggests a colovesical fistula.

Diverticular bleeding — Diverticular bleeding results from progressive injury to the artery supplying that segment. As a diverticulum herniates, the penetrating vessel responsible for the wall weakness at that point becomes draped over the dome of the diverticulum, separated from the bowel lumen only by mucosa (show endoscopy 2). Over time, the vasa recta is exposed to injury along its luminal aspect, leading to eccentric intimal thickening and thinning of the media. These changes may result in segmental weakness of the artery, predisposing to rupture into the lumen. (See "Epidemiology and pathophysiology of colonic diverticular disease").

The hallmark of diverticular bleeding is painless rectal bleeding, which is usually self-limited. Up to 50 percent give a history of intermittent passage of maroon or bright red blood (hematochezia) [10]. Abdominal discomfort is usually not present, and it is rare for bleeding to coexist with acute diverticulitis [11]. The physical examination is usually unremarkable. (See "Colonic diverticular bleeding").

Approximately 5 percent of patients with diverticulosis present with massive hemorrhage, and hypovolemia. The majority of these patients are over 60 years of age and many have comorbid conditions, which should be sought during the history and physical examination.

Diverticular colitis — Infrequent patients with diverticular disease develop a segmental colitis most commonly in the sigmoid colon. The endoscopic and histologic features vary, ranging from mild inflammatory changes with submucosal hemorrhages (peridiverticular red spots on colonoscopy) to florid, chronic active inflammation resembling (histologically and endoscopically) inflammatory bowel disease. The pathogenesis is incompletely understood. The cause may be multifactorial, related to mucosal prolapse, fecal stasis, or localized ischemia. (See "Diverticular colitis").

DIAGNOSIS OF DIVERTICULITIS — The diagnosis of acute diverticulitis is often made on the basis of the history and the physical examination. In the acute stage, further studies are performed to confirm the diagnosis and to rule out other sources of acute abdominal signs; the elective work-up following resolution of an acute episode of diverticulitis should evaluate the entire colon. Evaluation of patients with suspected colonic diverticular bleeding is discussed separately. (See "Colonic diverticular bleeding").

Radiologic evaluation in the acute setting — Routine abdominal and chest radiographs are commonly performed in the patient with acute abdominal pain, and are most useful in excluding other causes, such as intestinal obstruction, rather than in making the diagnosis of diverticulitis. Free air may be present in patients with a perforated diverticulum. In a stable patient with a previous history of confirmed diverticulitis and a similar current presentation, no further diagnostic work-up may be necessary at this point unless there is failure to improve with conservative management.

CT scan — Computer tomographic (CT) scanning of the abdomen with IV and oral contrast is the diagnostic test of choice in patients suspected of having acute diverticulitis. It is useful for diagnosis, assessment of severity, therapeutic intervention, and quantification of resolution of the disease. The sensitivity, specificity, positive, and negative predictive values of helical CT (with colonic contrast only) were 97, 100, 100, and 98 percent, respectively, in a study that included 150 patients presenting to the emergency department with clinically suspected diverticulitis [12].

CT features of acute diverticulitis include (show radiograph 1A-1C) [13-15]:

• Increased soft tissue density within pericolic fat, secondary to inflammation — 98 percent
• Colonic diverticula — 84 percent
• Bowel wall thickening — 70 percent
• Soft tissue masses representing phlegmon, and pericolic fluid collections, representing abscesses — 35 percent

In approximately 10 percent of patients, diverticulitis cannot be distinguished from carcinoma, since both may show focal thickening of the bowel wall. Differentiating features suggestive of diverticulitis are fluid at the base of the mesentery and mesenteric vascular engorgement [16].

CT can also identify the major complications of diverticulitis, including peritonitis (with diffuse inflammatory changes and scattered loculated fluid collections), fistula formation (usually inferred from extraluminal air collections in the bladder, vagina, and abdominal wall, rather than direct visualization), and obstruction.

CT also stages the extent of pericolic inflammation, which was underestimated by contrast barium enema in 41 percent of patients in one series [14]. Findings on CT have been classified as mild (localized colonic wall thickening and inflammation of pericolic fat) or severe (abscess, extraluminal air, or water soluble contrast); the latter findings have been used as criteria for offering elective resection to patients after successful conservative management [17]. They also predict an increased risk of failure of medical treatment during the first admission [18].

Finally, CT can be used as a therapeutic modality. It permits percutaneous drainage of localized abscesses, thereby downstaging complicated diverticulitis, avoiding emergent surgery, and permitting single-stage elective surgical resection [19]. (See "Treatment of acute diverticulitis").

Contrast enema — Contrast enema is safe in the acute phase if performed by the single contrast technique and if there is no evidence of complications. In the presence of complications, such as pneumoperitoneum or generalized peritonitis, barium is absolutely contraindicated [20], although water soluble contrast remains safe and may demonstrate the site of perforation. However, many prefer to use water soluble contrast even in the absence of these findings in any patient suspected of having acute diverticulitis since an unexpected leak may be found (show radiograph 2A-2B).

Compression ultrasonography — High-resolution, graded, compression ultrasonography is a relatively new method being used to evaluate diverticulitis. Visualization of an abnormal colonic segment (mural thickening greater than 4 mm involving a segment 5 cm or longer) at the point of maximal tenderness is the most common sonographic feature, being present in approximately 85 percent of patients [21]. In cross-section, the thickened colon has a target appearance. Inflamed diverticula, mural abscess, gas bubbles, peridiverticular abscess, and inflammation also may be visualized. Reported sensitivities range from 85 to 98 percent and specificities from 80 to 98 percent [21,22].

Compression ultrasonography is less widely used to evaluate diverticulitis than CT. However, it may be a useful additional tool in certain settings, such as following the resolution of an abscess, or permitting transrectal or transvaginal drainage of abscesses. One report described a potential role in patients presenting with abdominal pain in the emergency room [23]. It is best suited for patients with a thin body habitus. (See "Transabdominal ultrasonography of the small and large intestine").

Evaluation in the elective setting — After resolution of an episode of acute diverticulitis, the colon requires full evaluation to establish the extent of disease and to rule out coexistent lesions, such as polyps or carcinoma. This can be accomplished either with colonoscopy, or with the combination of barium enema plus flexible sigmoidoscopy. Flexible sigmoidoscopy evaluates the distal sigmoid and rectum, which are not well visualized by barium enema, and the barium enema clears the proximal colon of coexistent lesions. In addition, diverticulitis may lead to stricture formation that has the appearance of carcinoma; such lesions should be biopsied. The colon can also be evaluated by CT colography (show radiograph 3), but colonoscopy is still the preferred approach. (See "Computed tomographic colonography").

DIET — As noted above, the mere presence of diverticula in the absence of symptoms does not indicate the need for specific intervention other than recommending that the patient follow a high-fiber diet. Although the role of fiber in the pathogenesis of diverticulosis is controversial (see "Epidemiology and pathophysiology of colonic diverticular disease"), a diet high in fiber appears to be associated with a reduced risk of developing diverticular disease [24,25]. Furthermore, the addition of fiber to the diet of patients with uncomplicated diverticulosis may reduce the risk of subsequent complications [26].

The commonly heard advice to avoid small undigestible foods (such as seeds) because they may theoretically become lodged in a diverticulum is completely unproven and is probably little more than an old wives' tale. The authors have seen tens of thousands of diverticula and never seen a single seed! (See "Treatment of acute diverticulitis" section on dietary recommendations).

INFORMATION FOR PATIENTS — Educational materials on this topic are available for patients. (See "Patient information: Diverticular disease"). We encourage you to print or e-mail this topic review, or to refer patients to our public web site, www.uptodate.com/patients, which includes this and other topics.

SUMMARY AND RECOMMENDATIONS — The clinical manifestations of diverticulosis depend upon the presence or absence of complications. Uncomplicated diverticulosis is usually an incidental finding (show endoscopy 1). Some of these patients complain of symptoms such as cramping, bloating, flatulence, and irregular defecation. It is unclear if these symptoms are attributable to the underlying diverticulosis or to almost invariably coexistent irritable bowel syndrome. (See "Clinical manifestations and diagnosis of irritable bowel syndrome"). The major complications of diverticular disease are diverticulitis and diverticular bleeding.

• Diverticulitis represents micro- or macroscopic perforation of a diverticulum (show picture 1). The primary process is thought to be erosion of the diverticular wall by increased intraluminal pressure or inspissated food particles; inflammation and focal necrosis ensue, resulting in perforation. The clinical presentation of diverticulitis depends upon the severity of the underlying inflammatory process and whether or not complications are present. Left lower quadrant pain is the most common complaint in Western countries.

Complicated diverticulitis refers to the presence of an abscess, fistula, obstruction or perforation while simple diverticulitis refers to inflammation in the absence of these complications.

• The hallmark of diverticular bleeding is painless rectal bleeding. Most patients have minor or occult bleeding, but up to 50 percent give a history of intermittent passage of maroon or bright red blood (hematochezia). Abdominal discomfort is usually not present, and it is rare for bleeding to coexist with acute diverticulitis. The physical examination is usually unremarkable. (See "Colonic diverticular bleeding").

• The diagnosis of acute diverticulitis is often made on the basis of the history and the physical examination. In the acute stage, further studies are performed to confirm the diagnosis and to rule out other sources of acute abdominal signs. Abdominal CT scan with contrast has become the optimal method of investigation in patients suspected of having acute diverticulitis. The elective work-up following resolution of an acute episode of diverticulitis should evaluate the entire colon. (See "Diagnosis of diverticulitis" above).

• Diverticular bleeding is usually diagnosed in the context of patients presenting with acute lower gastrointestinal bleeding. Once the bleeding is determined to be coming from a lower GI source, colonoscopy is the initial examination of choice for diagnosis and treatment. (See "Approach to the adult patient with lower gastrointestinal bleeding" and see "Colonic diverticular bleeding").

INTRODUCTION — Trigeminal neuralgia (TN) is the most well defined and one of the most common causes of facial pain. TN is defined as sudden, usually unilateral, severe, brief, stabbing or lancinating, recurrent episodes of pain in the distribution of one or more branches of the fifth cranial (trigeminal) nerve [1].

An overview of TN is presented here. Other causes of facial pain are discussed separately. (See "Overview of craniofacial pain").

ANATOMY — The trigeminal nerve is the sensory supply to the face and the sensory and motor supply to the muscles of mastication. It has three major divisions:

• Ophthalmic (V1)
• Maxillary (V2)
• Mandibular (V3)

The nerve starts at the midlateral surface of the pons, and its sensory ganglion (gasserian ganglion) resides in Meckel's cave in the floor of the middle cranial fossa.

EPIDEMIOLOGY — The annual incidence of TN is 4 to 13 per 100,000 people [2,3]. Approximately 15,000 new cases occur in the United States each year [4]. TN is one of the most frequently seen neuralgias in the elderly. The incidence increases gradually with age; most idiopathic cases begin after age 50, although onset may occur in the second and third decades or, rarely, in children [5].

The male to female ratio of TN is about 1:1.5 [4]. This slight female predominance may be related to the increased longevity of women compared with men. Rare familial cases have been reported, but the vast majority of patients have sporadic disease [6].

ETIOLOGY AND PATHOGENESIS — Most cases of TN are caused by compression of the trigeminal nerve root, usually within a few millimeters of entry into the pons (the root entry zone) [7]. Compression by an aberrant loop of an artery or vein is thought to account for 80 to 90 percent of cases [7-9]. Idiopathic TN or TN caused by a vascular compression is considered classic TN [10]. (See "Classification" below).

Other causes of TN via nerve compression include vestibular schwannoma (acoustic neuroma), meningioma, epidermoid or other cyst, or rarely a saccular aneurysm or arteriovenous malformation [11-17]. TN caused by structural lesions other than vascular compression is classified as secondary TN [10]. (See "Classification" below).

The mechanism by which compression of the nerve leads to symptoms appears to be related to demyelination in a circumscribed area around the compression [18,19]. Precisely how demyelination results in the symptoms of TN is not entirely clear. Demyelinated lesions may set up ectopic impulse generation, possibly causing ephaptic transmission. Ephaptic cross-talk between fibers mediating light touch and those involved in pain generation could account for the precipitation of painful attacks by light tactile stimulation of facial trigger zones [7]. Furthermore, alteration of afferent input may disinhibit pain pathways in the spinal trigeminal nucleus.

Evidence for a role of central pain mechanisms includes the presence of refractory periods after a triggered episode, trains of painful sensations after a single stimulus, and latency from the time of stimulation to the onset of pain [20]. In addition, electrophysiologic evidence of central sensitization of trigeminal nociceptive processing has been observed in patients with atypical TN who have concomitant chronic facial pain [21].

Demyelination of one or more of the trigeminal nerve nuclei may also be caused by multiple sclerosis or other structural lesions of the brainstem. In multiple sclerosis, a plaque of demyelination typically occurs in the root entry zone of the trigeminal nerve [22] although vascular compression also has been noted in these patients [23].

Classification — TN is divided by presumed etiology into classic and secondary categories by the International Headache Society (IHS) criteria [10].

• Classic TN encompasses both idiopathic TN cases and those related to vascular compression [10]. The rationale is as follows. The IHS acknowledges that TN found to be related to compression of the trigeminal nerve by a vascular loop at surgery should strictly be regarded as secondary. However, most patients do not have surgery, and it is often uncertain as to whether they have primary or secondary TN. For this reason, the IHS uses the term "classic" instead of "primary" for patients with a typical history of TN who have a vascular source of compression as the presumed cause [10].

• Secondary (or symptomatic) TN is reserved for patients with TN caused by structural lesions other than vascular compression.

CLINICAL FEATURES — TN is defined clinically by sudden, usually unilateral, severe, brief, stabbing or lancinating, recurrent episodes of pain in the distribution of one or more branches of the fifth cranial (trigeminal) nerve [1].

The pain of TN tends to occur in paroxysms and is maximal at or near onset. Facial muscle spasms can be seen with severe pain. This finding gave rise to the older term for this disorder, tic douloureux. The pain is often described as electric, shock-like or stabbing. It usually lasts from one to several seconds, but may occur repetitively. A refractory period of several minutes during which a paroxysm cannot be provoked is common. Some patients with longstanding TN may have continuous dull pain that is present between paroxysms of pain. Unlike some other facial pain syndromes, TN typically does not awaken patients at night.

TN is typically unilateral. Occasionally the pain is bilateral, but not on both sides simultaneously [4]. The distribution of pain most often involves the V2 and/or V3 subdivisions of the trigeminal nerve [10]. The V1 subdivision is involved in <5 percent of patients. Of note, V1 is most commonly affected by postherpetic neuralgia. (See "Postherpetic neuralgia").

Trigger zones in the distribution of the affected nerve may be present and are often located near the midline. Lightly touching these zones often triggers an attack, leading patients to protect these areas. Trigger zones can sometimes be demonstrated on physical examination.

Other triggers of TN paroxysms include chewing, talking, brushing teeth, cold air, smiling, and/or grimacing.

Some patients have a history of "pretrigeminal neuralgia," which is said to be dull, continuous, aching pain in the jaw evolving eventually into TN. This brief, milder pain is sometimes suspected to have a dental origin and unnecessary dental procedures have been performed in many cases. On the other hand, TN can be precipitated by dental procedures (eg, dental extraction), resulting in increased confusion about the precise etiology of this problem [24].

The course of TN is variable. Episodes may last weeks or months, followed by pain-free intervals. Recurrence is common, and some patients have continuous pain. Most often, the condition tends to wax and wane in severity and frequency of pain exacerbations. However, there are no pure natural history studies of TN, most likely because the severity of the pain leads to intervention [25].

DIAGNOSIS — The diagnosis of TN is based upon the characteristic clinical features described above, primarily paroxysms of pain in the distribution of the trigeminal nerve.

Once the diagnosis of TN is suspected or confirmed on clinical grounds, a search for secondary causes should be undertaken. Patients with trigeminal sensory loss or bilateral involvement are probably at higher risk of secondary TN [26]. Younger age is also probably associated with a higher risk of secondary TN. However, age is not a clinically useful predictor for distinguishing classic from secondary TN because there is considerable age overlap. In addition, absence of any of these clinical features (sensory loss, bilateral involvement, younger age) does not rule out secondary TN.

Neuroimaging and trigeminal reflex testing are considered useful for distinguishing patients with classic TN (ie, idiopathic or caused by vascular compression) from those with secondary TN (ie, caused by structural brain lesion other than vascular compression) [26].

Diagnostic criteria — The International Headache Society (IHS) diagnostic criteria for classic TN are as follows [10]:

• Paroxysmal attacks of pain lasting from a fraction of a second to two minutes, affecting one or more divisions of the trigeminal nerve

• Pain has at least one of the following characteristics:

      -  Intense, sharp, superficial, or stabbing
      -  Precipitated from trigger areas or by trigger factors

• Attacks are stereotyped in the individual patient

• There is no clinically evident neurologic deficit

• Not attributed to another disorder

Secondary (symptomatic) TN is characterized by pain indistinguishable from classic TN, but is caused by a demonstrable structural lesion other than vascular compression [10]. Unlike classic TN, there is no refractory period after a paroxysm of pain. Secondary TN may exhibit sensory impairment in the distribution of the appropriate trigeminal nerve [10].

Neuroimaging — Neuroimaging with head CT or MRI is useful for identifying the small proportion of patients who have a structural lesion (eg, tumor in the cerebellopontine angle, demyelinating lesions including multiple sclerosis) as the cause of secondary TN [26,27]. In addition, high resolution MRI and magnetic resonance angiography (MRA) may be useful for identifying vascular compression as the etiology of classic TN, but the utility of these studies has not been established.

While specific evidence-based recommendations regarding the indications for neuroimaging in patients with TN cannot be made, we suggest obtaining brain MRI for patients in the following groups to rule out a causative structural brain lesion:

• Patients with trigeminal sensory loss
• Patients with bilateral symptoms
• Young patients (under the age of 40)

Some clinicians obtain an imaging study in all patients who present with TN.

An evidence-based systematic review and practice parameter published in 2008 from the American Academy of Neurology (AAN) and the European Federation of Neurological Societies (EFNS) identified four studies that evaluated consecutive patients with TN and normal neurologic examinations with head CT or MRI [26]. In the pooled data, routine brain imaging identified a secondary cause of TN (other than vascular compression) in 15 percent of patients (95% CI 11-20 percent).

The 2008 AAN/EFNS practice parameter identified seven studies that performed high-resolution brain MRI and/or magnetic resonance angiography (MRA) to identify neurovascular compression in patients with TN [26]. The following observations were made:

• There was wide variation among the included studies for both sensitivity (range 52 to 100 percent) and specificity (29 to 93 percent)

• In three of the five highest quality MRI studies (cohort surveys with prospective data collection), the difference in rate of neurovascular trigeminal nerve compression on the symptomatic side compared with asymptomatic side was statistically nonsignificant.

Given these inconsistent results, the AAN/EFNS concluded that there is insufficient evidence to support or refute the utility of MRI to identify neurovascular compression in classic TN, or to indicate the most reliable MRI technique [26].

Electrophysiologic tests — Electrophysiologic trigeminal reflex testing is probably useful for distinguishing classic TN from secondary TN [26]. Trigeminal reflex tests include the blink reflex, recording from the orbicularis oculi muscles after electrical stimulation of the supraorbital nerve (V1) and the masseter inhibitory reflex after electrical stimulation of the infraorbital (V2) and mental (V3) nerves. The responses are recorded by surface electrodes using standard electromyography equipment [28]. These tests are usually normal in patients with classic TN.

The 2008 AAN/EFNS practice parameter identified five studies that addressed the accuracy of trigeminal reflex testing for distinguishing classic TN from secondary TN [26]. One of the included studies was a prospective evaluation of 120 consecutive patients with TN [29], while the remaining four studies were retrospective in design [30-33]. The pooled sensitivity and specificity of trigeminal reflex testing for distinguishing secondary and classic TN was 94 percent (95% CI 91-97) and 87 percent (95% CI 77-93). Based on high sensitivity and specificity, the AAN/EFNS concluded that abnormal trigeminal reflexes are associated with an increased risk of secondary TN.

Other electrophysiologic tests are less well-studied. The AAN/EFNS practice parameter identified four studies that used trigeminal evoked potentials to distinguish secondary from classic TN and found that the pooled sensitivity and specificity was 84 percent (95% CI 73-92) and 64 percent (95% CI 56-71) [26]. However, there was substantial overlap such that many patients with classic TN had abnormal evoked potentials and many with secondary TN had normal evoked potentials. Thus, trigeminal evoked potentials were not considered clinically useful in the evaluation of patients with TN.

Differential diagnosis — The differential diagnosis of TN includes short-lasting unilateral neuralgiform headache with conjunctival injection and tearing (SUNCT), cluster-tic syndrome, jabs and jolts syndrome, and other neuralgias.

MEDICAL THERAPY — Pharmacologic therapy is the initial treatment of most patients with classic TN (ie, TN that is idiopathic or caused by neurovascular compression). Surgery is reserved for patients who are refractory to medical therapy.

Carbamazepine is the best studied treatment for classic TN and is established as effective [25,26,34]. Side effects can be a problem but are generally manageable, particular if low doses are prescribed initially with gradual titration.

A systematic review and practice parameter published in 2008 from the American Academy of Neurology (AAN) and the European Federation of Neurological Societies (EFNS) concluded that carbamazepine is effective for controlling pain in patients with classic TN, oxcarbazepine is probably effective, and baclofen, lamotrigine, and pimozide are possibly effective [26]. There is limited data and uncertain effectiveness regarding other drugs that have been used for TN, including clonazepam, gabapentin, phenytoin, tocainide, tizanidine, and valproate.

Periodic attempts to gradually withdraw these drugs are warranted in patients achieving relief of pain with oral medications.

No placebo-controlled trials have evaluated the treatment of secondary TN (ie, TN caused by a structural lesion other than vascular compression) and no medications have been established as effective for secondary TN. Treatment of the underlying condition (eg, multiple sclerosis) is recommended, if feasible. In addition, it is reasonable to treat the pain associated with secondary TN using the same medications that are empolyed in classic TN.

Carbamazepine — Four randomized, controlled trials with a total of 147 patients have established the effectiveness of carbamazepine (200 to 2400 mg daily) for TN [35-38].

A systematic review and practice parameter published in 2008 from the American Academy of Neurology (AAN) and the European Federation of Neurological Societies (EFNS) noted that the treatment response in these trials was robust, with complete or near complete pain control attained in 58 to 100 percent of patients on carbamazepine, compared with 0 to 40 percent of patients on placebo [26]. For the outcome of important pain relief, the number needed to treat was <2. However, carbamazepine was sometimes poorly tolerated, with numbers needed to harm for minor and severe adverse events of 3 and 24 respectively.

The usual starting dose of carbamazepine is 100 to 200 mg twice daily. The dose can be increased gradually in increments of 200 mg daily as tolerated until sufficient pain relief is attained. The typical total maintenance dose is 600 to 800 mg daily, given in two divided doses for tablets and extended release capsules, or four divided doses when for oral suspension. The maximum suggested total dose is 1200 mg daily.

Adverse effects of carbamazepine include drowsiness, dizziness, nausea and vomiting; slow titration may minimize these effects. Carbamazepine-induced leukopenia is not uncommon, but it is usually benign; aplastic anemia is a rare side effect. (See "Pharmacology of antiepileptic drugs", section on Carbamazepine).

The HLA-B*1502 allele is a genetic susceptibility marker in Asians that is associated with an increased risk of developing Stevens-Johnson syndrome and/or toxic epidermal necrolysis. For most patients of Asian ancestry, genetic testing for the presence of this marker is recommended by the manufacturer prior to initiation of carbamazepine.

Oxcarbazepine — Oxcarbazepine, an analogue of carbamazepine, was developed to retain the antineuralgic effect of carbamazepine while reducing side effects.

The AAN/EFNS practice parameter identified several randomized controlled trials that compared oxcarbazepine (600 to 1800 mg daily) with carbamazepine in 178 patients with classic TN [26]. In the pooled analysis, both medications were equally effective, with a >50 percent reduction of attacks achieved by 88 percent or more of patients in both treatment groups.

Oxcarbazepine can be started at a total dose of 600 mg daily, given in two divided doses. The dose can be increased as tolerated in 300 mg increments every third day to a total dose of 1200 to 1800 mg daily.

Baclofen — Limited evidence from a small double-blind crossover trial suggests that baclofen is beneficial for TN [39]. Treatment with baclofen 40 to 80 mg daily resulted in a reduction in paroxysms in seven of 10 patients with typical TN, compared with one of 10 who received placebo. [39].

The starting dose of baclofen is 15 mg daily given in three divided doses, with gradual titration to a maintenance dose of 50 to 60 mg per day. Sedation, dizziness, and dyspepsia can occur with treatment, and the drug should be discontinued slowly since seizures and hallucinations have been reported with upon withdrawal.

Lamotrigine — In a small double-blind, placebo-controlled crossover study of patients with TN that was refractory to carbamazepine or phenytoin, adjunct therapy with lamotrigine (400 mg daily) was beneficial for improvement on a composite outcome index [40]. Patients continued taking either carbamazepine or phenytoin for the duration of the trial.

Similarly, an open-label study found that lamotrigine was beneficial 11 of 15 patients with TN once the 400 mg dose was reached [41]. However, the clinical utility of lamotrigine for severe pain is limited by the need to titrate the dose over many weeks [25].

In patients who are not taking other anticonvulsants, lamotrigine is typically started at 25 mg daily for the first two weeks, then increased to 50 mg daily for weeks three and four. The dose is then titrated to effect, increasing by 50 mg daily every one to two weeks. The suggested total dose of 400 mg daily is given in two divided doses.

For patients taking an anticonvulsant drug that induces hepatic enzymes (eg, carbamazepine, phenytoin, or primidone), the initial dose of lamotrigine is 25 mg twice daily, increasing to 50 mg daily after two weeks and titrating upward by 50 mg per day increments every one to two weeks as needed.

For patients taking valproate, the initial dose of lamotrigine is 12.5 to 25 mg every other day, with increases of 25 mg every two weeks as needed to a maximum of 400 mg per day.

Pimozide — Pimozide, a dopamine receptor antagonist, was more effective than carbamazepine in a randomized, double-blind crossover trial of 48 patients with refractory TN [42]. There were no drop-outs among patients taking pimozide. However, pimozide is seldom used because it has many potentially serious side effects, including sedation, arrhythmias, anticholinergic effects, acute extrapyramidal symptoms and parkinsonism.

Other medications — Tizanidine appeared to be more effective than placebo in a small one-week trial, but patients who continued the drug in follow-up developed recurrent attacks of TN within one to three months [43].

Tocainide was as effective as carbamazepine at two weeks in a small cross-over trial [44].

Small open label studies have suggested benefit with a number of medications used for TN [45]:

• Phenytoin [46] and intravenous phenytoin [47]
• Fosphenytoin [48]
• Valproic acid [49,50]
• Gabapentin [51,52]
• Pregabalin [53]
• Clonazepam [54,55]
• Topiramate [56]
• Misoprostol, in patients with TN and multiple sclerosis [57]

However, these agents have not been studied in controlled trials, and their effectiveness in TN is not established.

Although there are no controlled data regarding the efficacy of opioids in TN specifically, we have used opiates in patients with acute exacerbations of pain lasting for days to weeks. Opiates may help make the pain bearable while other, more effective and long-term, treatments take effect. Our experience with opiates suggests partial analgesia with central side effects (particularly sedation) when these drugs are used alone, as high doses of morphine, hydromorphone or oxycodone are usually required. In combination with other neuropathic analgesics, opiates seem to be more effective at lower doses.

Refractory pain — There is little evidence to support treatment alternatives for patients who are refractory to first-line medical therapy. However, some patients who fail carbamazepine monotherapy may benefit from combination therapy with gabapentin, lamotrigine, topiramate, baclofen, or tizanidine. Intravenous infusion of phenytoin, fosphenytoin or lidocaine may provide analgesia while oral medications are titrated [47,58]. Phenytoin and fosphenytoin are dosed at 250 to 1000 mg intravenously [47] at no more than 50 mg/minute and lidocaine is given at 100 to 300 mg [58] over one-half hour while monitoring pulse and blood pressure.

Nevertheless, there are no randomized controlled trials comparing monotherapy with combination therapy for TN.

SURGICAL THERAPY — Patients with TN who are refractory to medical therapy are candidates for surgery. A variety of surgical methods have been employed to relieve the symptoms of TN. The major types of procedures are [25]:

• Microvascular decompression
• Ablative procedures, including:

      - Rhizotomy with either radiofrequency thermocoagulation, mechanical balloon compression, or chemical (glycerol) injection
      -  Gamma knife radiosurgery
      -  Peripheral neurectomy

However, few have been studied in controlled trials, and most of the evidence comes from observational studies.

A systematic review and practice parameter published in 2008 from the American Academy of Neurology (AAN) and the European Federation of Neurological Societies (EFNS) concluded that microvascular decompression, percutaneous procedures on the Gasserian ganglion (rhizotomy), and gamma knife are possibly effective in the treatment of TN [26]. Evidence for peripheral neurectomy was considered negative or inconclusive.

The AAN/EFNS noted that definitive conclusions regarding the relative effectiveness of surgical techniques for TN are precluded by the lack of studies directly comparing them [26]. Indirect comparisons of the findings from different surgical studies suggest that microvascular decompression has a longer duration of pain control than other surgical interventions for TN.

Microvascular decompression is invasive, although the overall mortality and complication rates are low. Ablative procedures are less invasive, but recurrence may be more common. The incidence of facial numbness is higher with rhizotomy procedures than with microvascular decompression or gamma knife.

Although surgical therapy for TN is generally well-tolerated, a feared complication is anesthesia dolorosa, a condition characterized by persistent, painful anesthesia or hypesthesia in the denervated region [10]. It can be more intolerable than the pain from TN itself [59]. This risk warrants careful decision making when considering surgical treatment for TN.

Anesthesia dolorosa most frequently occurs as a complication of rhizotomy or thermocoagulation for TN, but is rarely if ever a complication of gamma knife surgery.

Microvascular decompression — Microvascular decompression is a major neurosurgical procedure that involves craniotomy and the removal or separation of various vascular structures, often an ectatic superior cerebellar artery, away from the trigeminal nerve [60].

The AAN/EFNS practice parameter identified five studies of microvascular decompression for TN [61-65] that used independent outcome assessment [26]. The practice parameter concluded that initial pain relief is attained in 90 percent of patients, but that pain-free rates decline by one, three, and five years to 80, 75, and 73 percent.

The average mortality is approximately 0.2 percent. However, major adverse events, such as cerebrospinal fluid leaks, infarction or hematoma, occur in up to 4 percent of patients [26].

The most common complication is aseptic meningitis in 11 percent of patients [26]. Long term hearing loss occurs in up to 10 percent of patients, and sensory loss is found in 7 percent.

Rhizotomy — Rhizotomy encompasses a number of percutaneous surgical techniques that are performed by passing a cannula through the foramen ovale, followed by lesion of the trigeminal ganglion or root using one of several options [66]:

• Radiofrequency thermocoagulation rhizotomy, which creates a lesion by application of heat

• Mechanical balloon compression, which uses a Fogarty catheter to compress the gasserian ganglion

• Chemical (glycerol) rhizolysis, which involves the injection of 0.1 to 0.4 mL of glycerol into the trigeminal cistern

The 2008 AAN/EFNS practice parameter identified four uncontrolled case series that used independent outcome assessment of these procedures [26], including two reports of radiofrequency thermocoagulation [67,68], one report of glycerol rhizolysis [69], and one of balloon compression [70]. The AAN/EFNS found that initial pain relief is achieved in 90 percent of patients, but that pain-free rates decline by one year to 68 to 85 percent, by three years to 54 to 64 percent, and by five years to approximately 50 percent [26].

The major perioperative complication after rhizotomy procedures is meningitis, mainly aseptic, seen in 0.2 percent [26]. Mortality is rare. Postoperative dysesthesia, described as a burning, heavy, aching, or tired feeling, occurs in 12 percent. Longer-term sequelae include trigeminal distribution sensory loss in nearly one-half of patients, anesthesia dolorosa in approximately 4 percent, and corneal numbness with risk of keratitis in 4 percent.

Gamma knife radiosurgery — Gamma knife therapy produces lesions with focused gamma radiation [71]. (See "Stereotactic cranial radiosurgery and radiotherapy").

The therapy is aimed at the proximal trigeminal root since targeting the gasserian ganglion produced poor results [72]. The aiming of the beams is carried out with a stereotactic frame and MRI. The doses used are 70 to 90 Gy. The beams cause axonal degeneration and necrosis [72]. Pain relief with gamma knife surgery occurs after a lag time of about one month [72,73].

The 2008 AAN/EFNS practice parameter [26] identified one randomized controlled trial of gamma knife surgery for TN that compared two different treatment regimens [74] and found no important differences. In addition, the AAN/EFNS [26] identified three case series with independent outcome assessment [75-77]. Complete pain relief at one year was found in up to 69 percent of patients, and at three years in 52 percent [26]. An earlier systematic review found that approximately 75 percent of patients report complete relief within three months, but the proportion decreases to 50 percent by three years [78].

New or worsened facial sensory impairment occurred in 9 to 37 percent, with more bothersome sensory loss or paresthesia found in 6 to 13 percent of patients [26]. However, anesthesia dolorosa is rarely, if ever, a complication of gamma knife surgery.

Peripheral neurectomy — Peripheral neurectomy can be performed on the branches of the trigeminal nerve, which are the supraorbital, infraorbital, alveolar, and lingual nerves. Neurectomy is accomplished by incision, alcohol injection, radiofrequency lesioning, or cryotherapy. Cryotherapy involves freezing of the nerve using special probes, in theory to selectively destroy the pain fibers.

The AAN/EFNS practice parameter noted that the evidence regarding peripheral techniques for the treatment of TN is either negative or inconclusive [26].

SUMMARY AND RECOMMENDATIONS

• Trigeminal neuralgia (TN) is divided by presumed etiology into classic and secondary categories. Classic TN encompasses both idiopathic TN cases and those related to vascular compression. Secondary (or symptomatic) TN is reserved for patients with TN caused by structural lesions other than vascular compression. (See "Etiology and pathogenesis" above).

• TN is defined clinically by sudden, usually unilateral, severe, brief, stabbing or lancinating, recurrent episodes of pain in the distribution of one or more branches of the fifth cranial (trigeminal) nerve. (See "Clinical features" above).

• The diagnosis of TN is based upon the characteristic clinical features. Sensory loss, bilateral involvement, and younger age are associated with a higher risk of secondary TN, but their absence does not rule out secondary TN. Neuroimaging and trigeminal reflex testing are considered useful for distinguishing classic from secondary TN. (See "Diagnosis" above).

• We suggest obtaining brain MRI to rule out a causative structural brain lesion for all patients with suspected TN who have trigeminal sensory loss, bilateral symptoms, or age <40 years. (See "Neuroimaging" above).

• Pharmacologic therapy is used for initial treatment of most patients with classic TN. No medications have been established as effective for secondary TN. (See "Medical therapy" above)

      - For patients with classic TN who require pain control, we recommend carbamazepine as initial therapy (Grade 1A). (See "Carbamazepine" above).

      - For patients with classic TN who require pain control who do not respond to or tolerate carbamazepine, we recommend oxcarbazepine (Grade 1B). (See "Oxcarbazepine" above).

      - For patients who are refractory to or intolerant of carbamazepine and oxcarbazepine, we suggest switching to treatment with baclofen (Grade 2C). Alternatively, lamotrigine can be used as add-on therapy. (See "Baclofen" above and see "Lamotrigine" above).

      - For patients with TN who are refractory to the first and second-line agents listed above, a number of other medications with limited evidence of benefit may be considered. The choice among these agents is driven by patient preference, side effect profile, cost, and physician familiarity. (See "Other medications" above and see "Refractory pain" above)

• For patients with TN refractory to medical therapy, it is reasonable to discuss options for surgical therapy using microvascular decompression, various types of rhizotomy, or gamma knife radiosurgery. The decision to have surgery and the choice among surgical options will be influenced by individual circumstances including patient preference, adverse effect profile of the available techniques, and expertise of the local center. (See "Surgical therapy" above).

INTRODUCTION — Elevated intracranial pressure (ICP) is a potentially devastating complication of neurologic injury. Elevated ICP may complicate trauma, central nervous system (CNS) tumors, hydrocephalus, hepatic encephalopathy, and impaired CNS venous outflow (show table 1) [1]. Successful management of patients with elevated ICP requires prompt recognition, the judicious use of invasive monitoring, and therapy directed at both reducing ICP and reversing its underlying cause.

The evaluation and management of adult patients with elevated ICP will be reviewed here. Elevated intracranial pressure in children and specific causes and complications of elevated ICP are discussed separately. (See "Elevated intracranial pressure in children" and see appropriate topic reviews).

PHYSIOLOGY — Intracranial pressure is normally ≤15 mmHg in adults, and pathologic intracranial hypertension (ICH) is present at pressures ≥20 mmHg. ICP is normally lower in children than adults, and may be subatmospheric in newborns [2]. Homeostatic mechanisms stabilize ICP, with occasional transient elevations associated with physiologic events, including sneezing, coughing, or Valsalva maneuvers.

In adults, the intracranial compartment is protected by the skull, a rigid structure with a fixed internal volume of 1400 to 1700 mL. Under physiologic conditions, the intracranial contents include (by volume) [3]:

• Brain parenchyma — 80 percent
• Cerebrospinal fluid — 10 percent
• Blood — 10 percent

However, pathologic structures, including mass lesions, abscesses, and hematomas also may be present within the intracranial compartment. Since the overall volume of the cranial vault cannot change, an increase in the volume of one component, or the presence of pathologic components, necessitates the displacement of other structures, an increase in ICP, or both. Thus, ICP is a function of the volume and compliance of each component of the intracranial compartment, an interrelationship recognized over 150 years ago and known as the Monro-Kellie doctrine [4,5].

The volume of brain parenchyma is relatively constant in adults, although it can be altered by mass lesions or in the setting of cerebral edema (show figure 1). The volumes of CSF and blood in the intracranial space vary to a greater degree, and abnormal increases in the volume of either component may lead to elevations in ICP.

CSF is produced by the choroid plexus and elsewhere in the central nervous system (CNS) at a rate of approximately 20 mL/h (500 mL/day) [6]. CSF is normally resorbed via the arachnoid granulations into the venous system. Problems with CSF regulation generally result from impaired outflow caused by ventricular obstruction or venous congestion; the latter can occur in patients with sagittal (or other) venous sinus thrombosis. Much less frequently, CSF production can become pathologically increased; this may be seen in the setting of choroid plexus papilloma. (See "Cerebrospinal fluid: physiology and utility of an examination in disease states").

Cerebral blood flow (CBF) determines the volume of blood in the intracranial space. CBF increases with hypercapnia and hypoxia. Other determinants of CBF are discussed below. Autoregulation of CBF may be impaired in the setting of neurologic injury, and may result in rapid and severe brain swelling, especially in children [7-9].

Intracranial compliance — The interrelationship between changes in the volume of intracranial contents and changes in ICP defines the compliance characteristics of the intracranial compartment. Intracranial compliance can be modeled mathematically (as in other physiologic and mechanical systems) as the change in volume over the change in pressure (dV/dP).

The compliance relationship is nonlinear, and compliance decreases as the combined volume of the intracranial contents increases. Initially, compensatory mechanisms allow volume to increase with minimal elevation in ICP. These mechanisms include:

• Displacement of CSF into the thecal sac
• Decrease in the volume of the cerebral venous blood via venoconstriction and extracranial drainage

However, when these compensatory mechanisms have been exhausted, significant increases in pressure develop with small increases in volume, leading to abnormally elevated ICP (show figure 2).

Thus, the magnitude of the change in volume of an individual structure determines its effect on ICP. In addition, the rate of change in the volume of the intracranial contents influences ICP. Changes that occur slowly produce less of an effect than those that are rapid. This can be recognized clinically in some patients who present with large meningiomas and minimally elevated or normal ICP. Conversely, other patients may experience symptomatic elevations in ICP from small hematomas that develop acutely.

Cerebral blood flow — Following a significant increase in ICP, brain injury can result from brain stem compression and/or a reduction in cerebral blood flow (CBF). CBF is a function of the pressure drop across the cerebral circulation divided by the cerebrovascular resistance, as predicted by Ohm's law [10]:

  CBF   =   (CAP  -  JVP)  ÷  CVR

where CAP is carotid arterial pressure, JVP is jugular venous pressure, and CVR is cerebrovascular resistance.

Cerebral perfusion pressure (CPP) is a clinical surrogate for the adequacy of cerebral perfusion. CPP is defined as mean arterial pressure (MAP) minus ICP.

  CPP   =   MAP  -  ICP

Autoregulation — CBF is normally maintained at a relatively constant level by cerebrovascular autoregulation of CVR over a wide range of CPP (50 to 100 mmHg) (show figure 3) [11,12]. However, autoregulation of CVR can become dysfunctional in certain pathologic states, most notably stroke or trauma. In this setting, the brain becomes exquisitely sensitive to even minor changes in CPP [11-13].

Another important consideration is that the set-point of autoregulation is also changed in patients with chronic hypertension. With mild to moderate elevations in blood pressure, the initial response is arterial and arteriolar vasoconstriction. This autoregulatory process both maintains tissue perfusion at a relatively constant level and prevents the increase in pressure from being transmitted to the smaller, more distal vessels [11]. As a result, acute reductions in blood pressure, even if the final value remains within the normal range, can produce ischemic symptoms in patients with chronic hypertension (show figure 3) [11].

Cerebral perfusion pressure — Conditions associated with elevated ICP, including mass lesions and hydrocephalus, can be associated with a reduction in CPP. This can result in devastating focal or global ischemia. On the other hand, excessive elevation of CPP can lead to hypertensive encephalopathy and cerebral edema due to the eventual breakdown of autoregulation, particularly if the CPP is >120 mmHg [11,14,15]. A higher level of CPP is tolerated in patients with chronic hypertension because the autoregulatory curve has shifted to the right (show figure 3) [11,15]. (See "Hypertensive emergencies: Malignant hypertension and hypertensive encephalopathy", section on Mechanisms of vascular injury).

Ultimately, global or local reductions in CBF are responsible for the clinical manifestations of elevated ICP. These manifestations can be further divided into generalized responses to elevated ICP and herniation syndromes.

CLINICAL MANIFESTATIONS — Global symptoms of elevated ICP include headache, which is probably mediated via the pain fibers of cranial nerve (CN) V in the dura and blood vessels, depressed global consciousness due to either the local effect of mass lesions or pressure on the midbrain reticular formation, and vomiting.

Signs include CN VI palsies, papilledema secondary to impaired axonal transport and congestion (show picture 1), spontaneous periorbital bruising [16] and a triad of bradycardia, respiratory depression, and hypertension (Cushing's triad, sometimes called Cushing's reflex or Cushing's response) [3]. While the mechanism of Cushing's triad remains controversial, many believe that it relates to brainstem compression. The presence of this response is an ominous finding that requires urgent intervention.

Focal symptoms of elevated ICP may be caused by local effects in patients with mass lesions or by herniation syndromes. Herniation results when pressure gradients develop between two regions of the cranial vault. The most common anatomical locations affected by herniation syndromes include subfalcine, central transtentorial, uncal transtentorial, upward cerebellar, cerebellar tonsillar/foramen magnum, and transcalvarial (show table 2) [3,17]. (See "Stupor and coma in adults" sections on the Neurologic examination and on Coma syndromes).

One notable false localizing syndrome seen following neurologic injury, referred to as Kernohan's notch phenomenon, consists of the combination of contralateral pupillary dilatation and ipsilateral weakness [18,19]. Because the diagnostic accuracy of signs and symptoms is limited, the findings described above may be inconstant or unreliable in any given case. Use of radiologic studies may support the diagnosis; however, the most reliable method of diagnosing elevated ICP is to measure it directly.

ICP MONITORING — Empiric therapy for presumed elevated ICP is unsatisfactory because CPP cannot be monitored reliably without measurement of ICP. Furthermore, most therapies directed at lowering ICP are effective for limited and variable periods of time. In addition, these treatments may have serious side effects. Therefore, while initial steps to control ICP may, by necessity, be performed without the benefit of ICP monitoring, an important early goal in management of the patient with presumed elevated ICP is placement of an ICP monitoring device.

The purpose of monitoring ICP is to improve the clinician's ability to maintain adequate CPP and oxygenation. The only way to reliably determine CPP (defined as the difference between MAP and ICP) is to continuously monitor both ICP and blood pressure (BP). In general, these patients are managed in intensive care units (ICUs) with an ICP monitor and arterial line. The combination of ICP monitoring and concomitant management of CPP may improve patient outcomes, particularly in patients with closed head trauma [20-23].

CPP should be kept between 60 and 75 mmHg in patients with elevated ICP, in an attempt to avoid hypoperfusion and ischemic injury [20,24-26]. One study of 158 patients with head trauma and a Glasgow Coma Scale (GCS) <7 found that ICP monitoring and maintenance of CPP >70 mmHg resulted in improved outcomes when compared with historical controls (show table 3). This study also demonstrated that ICP generally did not increase with elevations in CPP until a critical level >110 mmHg was reached [20]. These data support the somewhat counterintuitive concept that pressors can be employed safely in patients with elevated ICP as long as the CPP is not elevated. (See "Use of vasopressors and inotropes").

Indications — The diagnosis of elevated ICP generally is based on clinical findings, and corroborated by imaging studies and the patient's medical history. Closed head injury is one of the most frequent and best-studied indications for ICP monitoring. Much of the current practice of ICP monitoring has been derived from clinical experience with closed head trauma patients [27]. (See "Intensive care unit management of the trauma patient").

Other potential indications for ICP monitoring include: stroke, intracerebral hemorrhage, hydrocephalus, subarachnoid hemorrhage, Reye's syndrome, hepatic encephalopathy, and sagittal sinus thrombosis.

The Guidelines for the Management of Severe Head Injury suggest that ICP monitoring is indicated in comatose head injury patients with Glasgow Coma Score (GCS) 3 to 8, (show table 3) and with abnormal cranial findings on computed tomographic (CT) scan [28-30]. Comatose patients with normal CT scans have a much lower incidence of elevated ICP unless they have the following features at admission:

• Age >40 years
• Unilateral or bilateral motor posturing
• Systolic blood pressure (SBP) <90 mmHg

ICP monitoring in patients with two or more of these risk factors is suggested. As noted above, invasive monitoring is usually not indicated in patients who are awake and able to follow commands. An exception to this rule occurs when an awake patient is at risk of elevated ICP and needs to have general anesthesia for surgery, rendering clinical evaluation impossible during anesthesia. Clinically, this may occur in patients following trauma who have evidence of small petechial hemorrhage on CT and require non-neurologic surgery for orthopedic injuries.

Role of computed tomography — Although CT scans may suggest elevated ICP based on the presence of mass lesions, midline shift, or effacement of the basilar cisterns (show radiograph 1), patients without these findings on initial CT may have elevated ICP. This was demonstrated in a prospective study of 753 patients treated at four major head injury research centers in the United States, which found patients whose initial CT scan did not show a mass lesion, midline shift, or abnormal cisterns had a 10 to 15 percent chance of developing elevated ICP during their hospitalization [31].

Other studies have shown that up to one-third of patients with initially normal scans developed CT scan abnormalities within the first few days after closed head injury [32,33]. Together, these findings demonstrate that ICP can be elevated even in the setting of a normal initial CT, demonstrating the importance of invasive monitoring in high-risk patients and the role of follow-up imaging in patients who develop clinical evidence of increased ICP during hospitalization.

Since ICP monitoring is associated with a small risk of serious complications, including CNS infection and intracranial hemorrhage, it is reasonable to try to limit its use to patients most at risk of elevated ICP [28]. In general, invasive monitoring of ICP is indicated in patients who are [34]:

• Suspected to be at risk for elevated ICP
• Comatose (Glasgow Coma Scale <8) (show table 3)
• Diagnosed with a process that merits aggressive medical care

Types of monitors — There are four main anatomical sites used in the clinical measurement of ICP: intraventricular, intraparenchymal, subarachnoid, and epidural (show figure 4). Noninvasive and metabolic monitoring of ICP has also been studied, but the clinical value of these methods is unclear at present. Each technique requires a unique monitoring system, and has associated advantages and disadvantages.

Intraventricular — Intraventricular monitors are considered the "gold standard" of ICP monitoring catheters. They are surgically placed into the ventricular system and affixed to a drainage bag and pressure transducer with a three-way stopcock. Intraventricular monitoring has the advantage of accuracy, simplicity of measurement, and the unique characteristic of allowing for treatment of some causes of elevated ICP via drainage of CSF.

The primary disadvantage is infection, which may occur in up to 20 percent of patients. This risk increases the longer a device is in place [35,36]. Prophylactic catheter changes did not appear to reduce the risk of infection [36]. (See "Infections of central nervous system shunts and other devices").

A further disadvantage of intraventricular systems includes a small (approximately 2 percent) risk of hemorrhage during placement; this risk is greater in coagulopathic patients. In addition, it may be technically difficult to place an intraventricular drain into a small ventricle, particularly in the setting of trauma and cerebral edema complicated by ventricular compression [37].

Intraparenchymal — Intraparenchymal devices consist of a thin cable with an electronic or fiberoptic transducer at the tip. The most widely used device is the fiberoptic Camino system. These monitors can be inserted directly into the brain parenchyma via a small hole drilled in the skull. Advantages include ease of placement, and a lower risk of infection and hemorrhage (<1 percent) than with intraventricular devices [38-40].

Disadvantages include the inability to drain CSF for diagnostic or therapeutic purposes and the potential to lose accuracy (or "drift") over several days, since the transducer cannot be recalibrated following initial placement. In addition, there is a greater risk of mechanical failure due to the complex design of these monitors. The reliability of intraparenchymal devices has been debated. One group found only a small (1 mmHg) drift in a group of 163 patients [41]; however, a second report found that readings varied by >3 mmHg in more than half of the 50 patients studied [42].

Subarachnoid — Subarachnoid bolts are fluid-coupled systems within a hollow screw that can be placed through the skull adjacent to the dura. The dura is then punctured, which allows the CSF to communicate with the fluid column and transducer. The most commonly used subarachnoid monitor is the Richmond (or Becker) bolt; other types include the Philly bolt, the Leeds screw, and the Landy screw. These devices have low risk of infection and hemorrhage, but often clog with debris and are unreliable; therefore, they are rarely used.

Epidural — Epidural monitors contain optical transducers that rest against the dura after passing through the skull. They often are inaccurate, as the dura damps the pressure transmitted to the epidural space, and thus are of limited clinical utility [43]. They are used in the management of coagulopathic patients with hepatic encephalopathy complicated by cerebral edema. In this setting, use of these catheters is associated with a significantly lower risk of intracerebral hemorrhage (4 versus 20 and 22 percent for intraparenchymal and intraventricular devices) and fatal hemorrhage (1 versus 5 and 4 percent, respectively) [44]. (See "Overview of the treatment of fulminant hepatic failure").

Noninvasive and metabolic systems — A number of devices designed to record ICP noninvasively have been studied, but none have demonstrated reproducible clinical success. However, tissue resonance analysis (TRA), an ultrasound-based method, has shown some promise. In one trial 40 patients underwent both invasive and TRA ICP monitoring, with good correlation between concomitant invasive and TRA measurements [45].

Other proposed noninvasive methods of monitoring ICP include transcranial Doppler (TCD), which measures the velocity of blood flow in the proximal cerebral circulation. TCD can be used to estimate ICP based on characteristic changes in waveforms that occur in response to increased resistance to cerebral blood flow [46,47]. Generally, TCD is a poor predictor of ICP, although in trauma patients TCD findings may correlate with outcome at six months [48-50].

Intraocular pressure can be assessed noninvasively using an ultrasonic handheld optic tonometer, and while some evidence suggests that intraocular pressure correlates with ICP in the absence of oculofacial trauma or glaucoma, other studies' findings disagree [51-53]. Similarly, tympanic membrane displacement (measured using an impedance audiometer) has been compared to direct monitoring, based on the hypothesis that increased ICP will transmit a pressure wave to the tympanic membrane via the perilymph [54,55]. Although promising, neither intraocular pressure monitoring nor tympanic membrane displacement has been studied in large clinical trials.

Finally, the metabolic state of the brain can be assessed using jugular venous oxygen saturation monitoring. This method is a way of quantifying regional oxygen consumption based on the Fick principle. Investigators have used catheters with oxygen monitors in the jugular bulb, as well as experimental sensors that utilize near-infrared spectroscopy, to quantify the oxygen content in blood draining from the CNS [56]. Evidence of jugular venous oxygen desaturation suggests impaired oxygen delivery and an ischemic state in the brain, consistent with elevated ICP [57-59]. (See "Oxygen delivery and consumption").

Waveform analysis — ICP is not a static value; it exhibits cyclic variation based on the superimposed effects of cardiac contraction, respiration, and intracranial compliance. Under normal physiologic conditions, the amplitude of the waveform is often small, with B waves related to respiration and smaller C waves (or Traube-Hering-Mayer waves) related to the cardiac cycle [10].

Pathological A waves (also called plateau waves) are abrupt, marked elevations in ICP of 50 to 100 mmHg, which usually last for minutes to hours. The presence of A waves signifies a loss of intracranial compliance, and heralds imminent decompensation of autoregulatory mechanisms [10,60,61]. Thus, the presence of A waves should suggest the need for urgent intervention to help control ICP.

GENERAL MANAGEMENT — The best therapy for intracranial hypertension (ICH) is resolution of the proximate cause of elevated ICP. Examples include: evacuation of a blood clot, resection of a tumor, CSF diversion in the setting of hydrocephalus, or treatment of an underlying metabolic disorder.

Regardless of the cause, ICH is a medical emergency, and treatment should be undertaken as expeditiously as possible. In addition to definitive therapy, there are maneuvers that can be employed to reduce ICP acutely. Some of these techniques are generally applicable to all patients with suspected ICH; others (particularly glucocorticoids) are reserved for specific causes of ICH.

Resuscitation — The urgent assessment and support of oxygenation, blood pressure, and end-organ perfusion are particularly important in trauma, but applicable to all patients [62-64]. If elevated ICP is suspected, care should be taken to minimize further elevations in ICP during intubation through careful positioning, appropriate choice of paralytic agents (if required), and adequate sedation. Pretreatment with lidocaine has been suggested as a useful intervention to decrease the rise in ICP associated with intubation; however, good clinical evidence supporting this approach is limited [65]. (See "Intensive care unit management of the trauma patient" and see "Overview of advanced cardiovascular life support in adults" and see "Overview of basic cardiovascular life support in adults").

Large shifts in blood pressure should be minimized, with particular care taken to avoid hypotension. Although it might seem that lower BP would result in lower ICP, this is not the case. Hypotension, especially in conjunction with hypoxemia, can induce reactive vasodilation and elevations in ICP. As noted above, pressors have been shown to be safe for use in most patients with intracranial hypertension, and may be required to maintain CPP >60 mmHg [20]. (See "Use of vasopressors and inotropes").

Urgent situations — Life-saving measures may need to be instituted prior to a more detailed workup (eg, imaging or ICP monitoring) in a patient who presents acutely with history or examination findings suggestive of elevated ICP. Many of these situations will rely upon clinical judgment, but the following combination of findings suggest the need for urgent intervention [66,67]:

• A history that suggests elevated ICP (eg, head trauma, sudden severe headache typical of subarachnoid hemorrhage)

• An examination that suggests elevated ICP (unilateral or bilaterally fixed and dilated pupil(s), decorticate or decerebrate posturing, bradycardia, hypertension and/or respiratory depression)

• A Glascow coma scale (GCS) ≤8

• Potentially confounding, reversible causes of depressed mental status, hypotension (SBP <60 mmHg in adults), hypoxemia (PaO2 <60 mmHg), hypothermia (<36ºC), or obvious intoxication are absent

In such patients osmotic diuretics may be used urgently (see "Mannitol" below).

In addition, standard resuscitation techniques should be instituted as soon as possible:

• Head elevation
• Hyperventilation to a PCO2 of 26 to 30 mmHg
• Intravenous mannitol (1 to 1.5 g/kg)

Concomitant with these measures should be aggressive evaluation of the underlying diagnosis, including neuroimaging, detailed neurologic examination, and history gathering. Hyperventilation may be contraindicated in the setting of traumatic brain injury and acute stroke, and is discussed separately (see "Hyperventilation" below). If appropriate, ventriculostomy is a rapid means of simultaneously diagnosing and treating elevated ICP.

Monitoring and the decision to treat — If a diagnosis of elevated ICP is suspected and an immediately treatable proximate cause is not present, then ICP monitoring should be instituted. The use of ICP monitoring is associated with decreased mortality in patients with traumatic brain injury [21]. The type of monitoring device employed should be based on an assessment of the advantages and disadvantages discussed previously (show figure 4).

The goal of ICP monitoring and treatment should be to keep ICP <20 mmHg and CPP between 60 and 75 mmHg [68]. Interventions should be utilized only when ICP is elevated above 20 mmHg for >5 to 10 minutes. As discussed above, brief physiologic elevations in ICP may occur in the setting of coughing, movement, suctioning, or ventilator asynchrony.

Fluid management — In general, patients with elevated ICP do not need to be severely fluid restricted [69]. Patients should be kept euvolemic and normo- to hyperosmolar. This can be achieved by avoiding all free water (including D5W, 0.45 percent (half normal) saline, and enteral free water) and employing only isotonic fluids (such as 0.9 percent (normal) saline). Serum osmolality should be kept >280 mOsm/L, and often is kept in the 295 to 305 mOsm/L range. Hyponatremia is common in the setting of elevated ICP, particularly in conjunction with subarachnoid hemorrhage. (See "Causes of hyponatremia", and see "Treatment of hyponatremia: SIADH and reset osmostat", section on Subarachnoid hemorrhage).

Similarly, the value of colloid compared to crystalloid fluid resuscitation in patients with elevated ICP has been studied, but findings have been inconclusive with respect to the superior approach [70].

Hypertonic saline in bolus doses may acutely lower ICP, but further investigations are required to define a role, if any, for this approach in the management of elevated intracranial pressure. (See "Hypertonic saline bolus" below).

Sedation — Keeping patients appropriately sedated can decrease ICP by reducing metabolic demand, ventilator asynchrony, venous congestion, and the sympathetic responses of hypertension and tachycardia [71]. Establishing a secure airway and close attention to blood pressure allow the clinician to identify and treat apnea and hypotension quickly. Propofol has been utilized to good effect in this setting, as it is easily titrated and has a short half-life, thus permitting frequent neurologic reassessment. (See "Use of sedative medications in critically ill patients").

Blood pressure control — In general, BP should be sufficient to maintain CPP >60 mmHg. As discussed above, pressors can be used safely without further increasing ICP. This is particularly relevant in the setting of sedation, when iatrogenic hypotension can occur. Hypertension should generally only be treated when CPP >120 mmHg and ICP >20 mmHg.

Caution should be taken to avoid CPP <60 mmHg or, as noted above, normalization of blood pressure in patients with chronic hypertension in whom the autoregulatory curve has shifted to the right (see "Autoregulation" above). General issues regarding blood pressure management following stroke are presented elsewhere. (See "Treatment of hypertension following a stroke").

Position — Patients with elevated ICP should be positioned to maximize venous outflow from the head. Important maneuvers include reducing excessive flexion or rotation of the neck, avoiding restrictive neck taping, and minimizing stimuli that could induce Valsalva responses, such as endotracheal suctioning.

Patients with elevated ICP have historically been positioned with the head elevated above the heart (usually 30 degrees) to increase venous outflow. It should be noted that head elevation may lower CPP [20,72]; however, given the proven efficacy of head elevation in lowering ICP, most experts recommend raising the patient's head as long as the CPP remains at an appropriate level [73].

Fever — Elevated metabolic demand in the brain results in increased cerebral blood flow (CBF), and can elevate ICP by increasing the volume of blood in the cranial vault. Conversely, decreasing metabolic demand can lower ICP by reducing blood flow.

Fever increases brain metabolism, and has been demonstrated to increase brain injury in animal models [74]. Therefore, aggressive treatment of fever, including acetaminophen and mechanical cooling, is recommended in patients with increased ICP. Intracranial hypertension is a recognized indication for neuromuscular paralysis in selected patients [75]. (See "Use of neuromuscular blocking medications in critically ill patients").

Antiepileptic therapy — Seizures can both complicate and contribute to elevated ICP [76,77]. Anticonvulsant therapy should be instituted if seizures are suspected; prophylactic treatment may be warranted in some cases. There are no clear guidelines for the latter, but examples include high-risk mass lesions, such as those within supratentorial cortical locations, or lesions adjacent to the cortex, such as subdural hematomas or subarachnoid hemorrhage.

SPECIFIC THERAPIES — As mentioned previously, the best treatment of elevated ICP is to address its underlying cause. If this is not possible, a series of steps should be instituted to reduce ICP in an attempt to improve outcome. In all cases, the clinician should bear in mind the themes of resuscitation, reduction of intracranial volume, and frequent reevaluation discussed above.

Mannitol — Osmotic diuretics reduce brain volume by drawing free water out of the tissue and into the circulation, where it is excreted by the kidneys, thus dehydrating brain parenchyma [78-81]. The most commonly used agent is mannitol. It is prepared as a 20 percent solution, and given as a bolus of 1 g/kg. Repeat dosing can be given at 0.25 to 0.5 g/kg as needed, generally every six to eight hours. Use of any osmotic agent should be carefully evaluated in patients with renal insufficiency.

The effects are usually present within minutes, peak at about one hour, and last 4 to 24 hours [34,82]. Some have reported a "rebound" increase in ICP; this probably occurs when mannitol, after repeated use, enters the brain though a damaged blood-brain barrier and reverses the osmotic gradient [83,84]. Useful parameters to monitor in the setting of mannitol therapy include serum sodium, serum osmolality, and renal function.

Concerning findings associated with the use of mannitol include serum sodium >150 meq, serum osmolality >320 mOsm, or evidence of evolving acute tubular necrosis (ATN). In addition, mannitol can lower systemic BP, necessitating careful use if associated with a fall in CPP. Patients with known renal disease may be poor candidates for osmotic diuresis. (See "Complications of mannitol therapy").

Other diuretics — Furosemide, 0.5 to 1.0 mg/kg intravenously, may be given with mannitol to potentiate its effect. However, this effect can also exacerbate dehydration and hypokalemia [85-87]. Hypertonic saline may also be employed for this purpose (see "Fluid management" above).

Glycerol and urea were used historically to control ICP via osmoregulation; however, use of these agents has decreased because equilibration between brain and plasma levels occurs more quickly than with mannitol. Furthermore, glycerol has been shown to have a significant rebound effect and to be less effective in ICP control [88,89].

Hypertonic saline bolus — Hypertonic saline in bolus doses may acutely lower ICP; however, the effect of this early intervention on long-term clinical outcomes remains unclear [90-97]. One controlled trial randomly assigned 226 patients with traumatic brain injury to prehospital resuscitation with 250 mL hypertonic saline (7.5 percent) or the same volume of Ringer's lactate [90]. Survival until hospital discharge, six-month survival, and neurologic function six months after injury were similar in both groups. The volume and tonicity of saline (7.2 to 23.4 percent) used in these trials varied widely, and further work is required to clarify what role, if any, hypertonic saline infusion has in the management of elevated ICP [98].

Glucocorticoids — The use of glucocorticoid therapy following head trauma has been routine [99]; however, the publication of a large, well-executed trial has cast doubt on the efficacy and safety of this approach [100]. The multinational MRC CRASH (Glucocorticoid Randomization After Significant Head injury) trial enrolled 10,008 patients with head trauma and Glasgow Coma Score <14 (show table 3). Patients were entered into the trial within eight hours of presentation, and were randomly assigned to therapy with methylprednisolone (20 g over 48 hours) or placebo. The primary end point of the trial was death within two weeks after injury.

Patients treated with methylprednisolone experienced increased all-cause mortality at two weeks (21 versus 18 percent; RR 1.18, 95% CI 1.09-1.27). Furthermore, there were no subsets of patients who showed benefit from steroid therapy, including those with more severe injuries, and those who received treatment within one hour of presentation. Subsequent six-month follow-up, which was obtained for 97 percent of the patients enrolled in the trial, demonstrated a persistent increase in the risk of death among patients treated with methylprednisolone (26 versus 22 percent for control patients; relative risk 1.15, 95% CI 1.07-1.24) [101]. These striking results suggest that glucocorticoid therapy is not indicated following head injury, and may be associated with increased acute mortality.

The safety and efficacy of glucocorticoid therapy in the management of patients with nontraumatic causes of increased ICP is discussed separately. In general, glucocorticoids are not considered to be useful in the management of infarction or intracranial hemorrhage. In contrast, glucocorticoids may have a role in the setting of intracranial hypertension caused by brain tumors and CNS infections. (See "Management of vasogenic edema in patients with primary and metastatic brain tumors", see "Treatment and prognosis of brain abscess" and see "Dexamethasone to prevent neurologic complications of bacterial meningitis in adults").

Hyperventilation — Use of mechanical ventilation to lower PaCO2 to 26 to 30 mmHg has been shown to rapidly reduce ICP through vasoconstriction and a decrease in the volume of intracranial blood; a 1 mmHg change in PaCO2 is associated with a 3 percent change in CBF [102]. Hyperventilation also results in respiratory alkalosis, which may buffer post-injury acidosis [102]. The effect of hyperventilation on ICP is short-lived (1 to 24 hours) [103-105]. Following therapeutic hyperventilation, the patient's respiratory rate should be tapered back to normal over several hours to avoid a rebound effect [106].

Therapeutic hyperventilation should be considered as an urgent intervention when elevated ICP complicates cerebral edema, intracranial hemorrhage, and tumor. Hyperventilation should not be used on a chronic basis, regardless of the cause of ICH. Furthermore, hyperventilation should be minimized in patients with traumatic brain injury or acute stroke. In these settings, vasoconstriction may cause a critical decrease in local cerebral perfusion and worsen neurologic injury, particularly in the first 24 to 48 hours [27,103,105,107-110]. Thus, the need for hyperventilation should be carefully considered, and prophylactic hyperventilation in the absence of elevated ICP should be avoided.

Barbiturates — The use of barbiturates is predicated on their ability to reduce brain metabolism and cerebral blood flow, thus lowering ICP and exerting a neuroprotective effect [111-114]. Pentobarbital is generally used, with a loading dose of 5 to 20 mg/kg as a bolus, followed by 1 to 4 mg/kg per hr [115,116]. Treatment should be assessed based on ICP, CPP, and the presence of unacceptable side effects. Continuous EEG monitoring is generally used; EEG burst suppression is an indication of maximal dosing.

The therapeutic value of this maneuver is somewhat unclear. In a randomized trial of 73 patients with elevations in ICP refractory to standard therapy, patients treated with pentobarbital were 50 percent more likely to have their ICP controlled. However, there was no difference in clinical outcomes between groups [117]. In general, the use of barbiturates is a "last-ditch" effort, as several studies show that their ability to lower ICP does not appear to affect outcomes [102,118].

Barbiturate therapy can be complicated by hypotension, possibly requiring vasopressor support. The use of barbiturates is also associated with a loss of the neurologic examination, requiring accurate ICP, hemodynamic, and often EEG monitoring to guide therapy.

Therapeutic hypothermia — First reported as a treatment for brain injury in the 1950s, induced or therapeutic hypothermia has remained a controversial issue in the debate concerning the management of elevated ICP [102,119,120]. It is not currently recommended as a standard treatment for increased intracranial pressure.

Hypothermia decreases cerebral metabolism and may reduce CBF and ICP. Initial studies of hypothermia were limited by systemic side effects, including cardiac arrhythmias and severe coagulopathy. However, later work suggested that hypothermia can lower ICP and may improve patient outcomes [121]. Hypothermia also appeared to be effective in lowering ICP after other therapies have failed [122,123].

Hypothermia can be achieved using whole body cooling, including lavage and cooling blankets, to a goal core temperature of 32 to 34ºC. The best method of cooling (local versus systemic), the optimal target core temperature, and the appropriate duration of treatment are not known [124]. It appears that rewarming should be accomplished over a period of less than 24 hours [125].

The value of therapeutic hypothermia has been best assessed in patients after traumatic brain injury (TBI). A systematic review of 12 randomized controlled trials of mild-to-moderate hypothermia (32 to 33ºC) following TBI noted a small but significant decrease in the risk of death (RR 0.81, 95% CI 0.69-0.96) or poor neurologic outcome (RR 0.78, 95% CI 0.63-0.98) among more than 500 patients treated with hypothermia [125]. Other systemic reviews and meta-analyses found similar but more borderline benefits for death and neurologic outcome as well as an increased risk in pneumonia [126-128]. Substantial variability among studies in the depth and duration of hypothermia, as well as the rate of rewarming limit the ability to draw conclusions from these studies. A subsequent trial of hypothermia therapy in children with TBI showed no improvement in neurologic outcomes and a nonsignificant increase in mortality [129]. (See "Elevated intracranial pressure in children", section on Hypothermia).

Given the uncertainties surrounding the appropriate use of therapeutic hypothermia in patients with elevated ICP, this treatment should be limited to clinical trials, or to patients with intracranial hypertension refractory to other therapies.

Removal of CSF — When hydrocephalus is identified, a ventriculostomy should be inserted. Rapid aspiration of CSF should be avoided because it may lead to obstruction of the catheter opening by brain tissue. Also, in patients with aneurysmal subarachnoid hemorrhage, abrupt lowering of the pressure differential across the aneurysm dome can precipitate recurrent hemorrhage.

CSF should be removed at a rate of approximately 1 to 2 mL/minute, for two to three minutes at a time, with intervals of two to three minutes in between until a satisfactory ICP has been achieved (ICP <20 mmHg) or until CSF is no longer easily obtained. Slow removal can also be accomplished by passive gravitational drainage through the ventriculostomy. A lumbar drain is generally contraindicated in the setting of high ICP due to the risk of transtentorial herniation.

Decompressive craniectomy — Decompressive craniectomy removes the rigid confines of the bony skull, increasing the potential volume of the intracranial contents and circumventing the Monroe-Kellie doctrine. There is a growing body of literature supporting the efficacy of decompressive craniectomy, especially after traumatic brain injury [130-139]. Importantly, it has been demonstrated that in patients with elevated ICP, craniectomy alone lowered ICP 15 percent, but opening the dura in addition to the bony skull resulted in an average decrease in ICP of 70 percent [140].

Observational data suggest that rapid and sustained control of ICP, including the use of decompressive craniectomy, improves outcomes in trauma, stroke, and subarachnoid hemorrhage in carefully selected cases [141-148]. Decompressive craniectomy also appears to improve brain tissue oxygenation [149]. However, in the absence of randomized, controlled trials, it is difficult to make a definitive judgment of the efficacy of craniectomy in these situations [150,151]. Obvious mass lesions associated with an elevated ICP should be removed, if possible.

Paradoxical transtentorial herniation is an uncommon but potentially lethal complication in patients with hemicraniectomy and a large skull defect who subsequently undergo lumbar puncture (LP) or CSF drainage [152,153]. This results from the combined effects of atmospheric pressure with the negative pressure of the LP or ventriculostomy. Marked decompression of the skin and dura over the skull defect accompany neurologic signs of herniation. Standard treatments to lower ICP can hasten herniation. Instead, the patient should be placed supine or in the Trendelenberg position, CSF drains should be clamped, crystalloid fluid should be administered intravenously, and an epidural blood patch placed for patients with dural leak.

SUMMARY — The best therapy for intracranial hypertension is resolution of the proximate cause of elevated ICP. Regardless of the cause, treatment should be undertaken as expeditiously as possible, and should be based on the principles of resuscitation, reduction of the volume of the intracranial contents, and reassessment. The role of evidence-based guidelines in the clinical management of elevated ICP is evolving [154]. However, it is important to remember that individual patients respond differently to different therapies; therefore, interventions should be based on careful assessment of the individual clinical scenario rather than on strict protocols.

Intracranial compensation for an expanding mass lesion

Data from Pathophysiology and management of the intracranial vault. In: Textbook of Pediatric Intensive Care, 3rd ed, Rogers, MC (Ed), Williams and Wilkins 1996. p. 646; figure 18.1.