Review

A Practical and Evidence-Based Approach to Common Symptoms A Narrative Review Kurt Kroenke, MD

Physical symptoms account fo r more than half o f all outpatient visits, yet the predominant disease-focused model o f care is inad­ equate fo r many o f these symptom-prompted encounters. M ore­ over, the amount o f clinician training dedicated to understanding, evaluating, and managing common symptoms is disproportionally small relative to their prevalence, impairment, and health care costs. This narrative review regarding physical symptoms addresses 4 common epidemiologic questions: cause, diagnosis, prognosis, and therapy.

Important findings include the follow ing: First, at least one third of common symptoms do not have a clear-cut, disease-based ex­ planation (5 studies in primary care, 1 in specialty clinics, and 2 in the general population). Second, the history and physical examina­ tion alone contribute 73% to 94% o f the diagnostic information, w ith costly testing and procedures contributing much less (5 studies of multiple types o f symptoms and 4 o f specific symptoms). Third,

physical and psychological symptoms commonly co-occur, making a dualistic approach impractical. Fourth, because most patients have multiple symptoms rather than a single symptom, focusing on 1 symptom and ignoring the others is unwise. Fifth, symptoms im ­ prove in weeks to several months in most patients but become chronic or recur in 20% to 25% . Sixth, serious causes tha t are not apparent after initial evaluation seldom emerge during long-term follow-up. Seventh, certain pharmacologic and behavioral treat­ ments are effective across multiple types o f symptoms. Eighth, measuring treatment response w ith valid scales can be helpful. Finally, communication has therapeutic value, including providing an explanation and probable prognosis w ithou t "normalizing" the symptom.

Ann Intern Med. 2014;161:579-586. doi:10.7326/M14-0461 www.annals.org For author affiliation, see end of text.

Symptoms account for over half of all outpatient visits or more than 400 million office visits annually in the

United States alone (1). Yet those who seek care represent a minority of symptomatic persons in the general popula­ tion: 80% of persons have at least 1 distressing symptom in a given month, yet fewer than 1 in 4 persons visit a health care provider for their symptoms (2). Thus, we must re­ frain from overmedicalizing symptoms in the community at large and excessively testing and treating the subset who present clinically. This is not to say that symptoms are minor, trivial, or unimportant; indeed, they cause greater distress and impairment than many of the asymptomatic risk factors (for example, hypertension, hyperlipidemia, and obesity) that we target for health care. Most symptom­ atic persons are currently suffering, whereas only a fraction of those with medicalized risk factors will ultimately be­ come ill and often not until decades later. Moreover, symp­ toms are associated with substantial impairments in health- related quality of life, work-related disability, and increased health care costs (1, 3, 4). Further, patient and clinician dissatisfaction can occur when there are multiple symp­ toms or symptoms that are unexplained (5).

This article focuses on the 4 common epidemiologic questions about a clinical condition: cause, diagnosis, prog­ nosis, and therapy. A symptom is operationally defined as an uncomfortable or distressing bodily sensation experi­ enced by a person that is not observable by the clinician (those that are observable are signs). For example, cough, emesis, edema, and syncope are all symptomatic but also observable by clinicians and other persons besides the pa­ tient. The focus is further restricted to physical (also called somatic) symptoms. Although psychological and cognitive symptoms (for example, depression, anxiety, and impaired memory or concentration) frequently co-occur with phys-

www.annals.org

ical symptoms, the patient in the medical setting often presents with physical symptoms that prompt a biomedi- cally oriented search for medical causes and treatments. O f physical symptoms presented in practice, about 50% are pain, 25% to 30% are respiratory (usually upper respira­ tory), and 20% to 25% are nonpain and nonrespiratory in nature (for example, fatigue, sleep symptoms, gastrointes­ tinal symptoms, or dizziness). Because symptoms related to upper respiratory infections are often self-limited and diag­ nostically less challenging, our main attention is on the three quarters of symptom-related office visits triggered by non—upper respiratory physical symptoms. Finally, the fo­ cus is not on a specific approach to a particular symptom but on generic principles that apply to common symptoms as a whole. Although there are symptom-specific issues, and in some cases guidelines, there are also cross-cutting epidemiologic themes that broadly apply across most symptoms.

The literature cited in this review includes articles fa­ miliar to the author complemented by relevant papers identified by a bibliographic search of those articles. The breadth of this review precluded a more formalized litera­ ture search. Also, some studies included had small samples, short follow-ups, single raters using unstructured assess­ ments, and other methodological limitations highlighted in Tables 1 to 3.

S ee a ls o :

W eb-O nly CM E qu iz

21 October 2014 Annals of Internal Medicine Volume 161 • Number 8 5 7 9

R e v i e w A Practical and Evidence-Based Approach to Common Symptoms

Key Summary Points

At least one third of common symptoms do not have a clear-cut, disease-based explanation.

The patient's history alone yields 75% of the diagnostic information.

Physical and psychological symptoms commonly co-occur.

Most patients have multiple symptoms rather than a single symptom.

Symptoms become chronic or recur in 20% to 25% of patients.

Serious causes that are not apparent after initial evaluation seldom emerge later.

Some medications and behavioral interventions are effec­ tive for multiple types of symptoms.

Measuring treatment response with valid scales can be helpful.

Communication has therapeutic value, including providing an explanation and probable prognosis w ithout "normaliz­ ing" the symptom.

C ause: Sym p to m s A re Sufficient

Symptoms transcend disease. The subjective is not inferior to the objective.

The dominant clinical paradigm is that symptoms are a derivative of disease and that optimal symptom manage­ ment will naturally follow once the causative disease is identified. A corollary is that the “subjective” (what pa­ tients experience and report) depends on and is inferior to the “objective” (what clinicians or testing find). An alter­ native model is that symptoms are a higher-order phenom­ enon that come from a varying mix of disease and nondis­ ease input (for example, biological factors that modulate symptoms and mediate symptom perception; cognitive processes, such as symptom attributions, amplification, at­ tention, and affect; and external interpersonal and socio­ cultural influences). This model favors an integrative ap­ proach wherein symptoms are the most human expression of clinical medicine and do not lend themselves to overly simplified, reductionistic, or mechanistic explanations.

A t least one third o f symptoms evaluated in primary care are medically unexplained.

As shown in Table 1, studies conducted in primary care (3, 6 -9), specialty settings (10), and the general pop­ ulation (11, 12) have consistently shown that a substantial proportion of somatic symptoms are medically unex­ plained. O f the 8 studies, 5 showed that 31% to 37% of symptoms were medically unexplained. The study with the highest rate (74%) may have overestimated because it de-

580 21 October 2014 Annals of Internal Medicine Volume 161 • Number 8

pended on the ratings of 1 physician reviewer using im­ plicit judgment rather than explicit criteria (6). Conversely, the study reporting only a 20% rate might have underesti­ mated because certain somatic symptoms were not counted as medically unexplained if they were diagnostic criteria for patients who qualified for a depressive or anxiety disorder (3). The lack of a definitive explanation for many symp­ toms is further underscored by the use of adjectival modi­ fiers indicating what a symptom is not (“noncardiac” chest pain or “nonulcer” dyspepsia) or implying causal explana­ tions that are weakly defensible (“tension” headache, “me­ chanical” low back pain, or “irritable” bowel syndrome) (1). Also, some purported explanations for symptoms have become extinct (hypoglycemia, mitral valve prolapse, or chronic brucellosis), controversial (for example, multiple chemical sensitivity or sick building syndrome), or event- triggered but complex in cause (for example, Gulf War or other postwar syndromes or World Trade Center syn­ drome) (1, 13).

Dualistic (physical vs. psychological) explanatory models are particularly problematic.

A binary approach to classifying symptoms as medical, physical, or organic in cause or psychological, mental, or functional is neither evidence-based nor patient-centered. For example, when depression coexists with chronic pain, is it the cause, consequence, or product of a common path­ way? Rather than a chicken-egg conundrum, longitudinal studies of pain and depression have consistently shown that their effects are reciprocal rather than unidirectional (14). This interactive influence of physical and psychological symptoms is true of other nonpain somatic symptoms and other psychological symptoms, such as anxiety (1, 15).

A more useful classification scheme considers cause along a spectrum from medical to mental disorders with 5 salient nodes (16). First, there are the symptoms clearly attributable to a specific medical disease, such as dyspnea in a wheezing asthmatic patient or substernal chest pain in the patient with an acute myocardial infarction. Second, there are the less well-understood functional somatic syn­ dromes, such as irritable bowel syndrome, fibromyalgia, and chronic fatigue syndrome. Third, there are symptom- only diagnoses, such as low back pain, nonmigraine head­ ache, nonspecific dizziness, and many other symptoms that cannot be ascribed to an obvious disease. Fourth, there is somatic symptom reporting seen in patients with depres­ sion and anxiety as either core diagnostic criteria (for ex­ ample, fatigue and insomnia in depression or cardiopulmo­ nary symptoms in panic disorder) or, more often, the increased reporting of both general (4, 17) and disease- specific (18) somatic symptoms associated with psycholog­ ical conditions. Fifth, there are the medically unexplained symptoms associated with dysfunctional illness behavior classified as somatoform disorders.

Symptoms may often be multifactorial in cause. Efforts to pinpoint a single cause for a symptom can

be disappointing. For example, it may be difficult to deter-

w w w .a n n a ls .o rg

A Practical and Evidence-Based Approach to Com m on Symptoms R e v i e w

mine what proportion of the fatigue in a patient with ma­ jor depression and congestive heart failure is due to each disorder. Although a reduction in fatigue with disease- specific therapeutic trials may be informative, such treat­ ments may only partially alleviate disease-related symp­ toms, and some symptoms may be less responsive than others (for example, fatigue may be more refractory than cardiopulmonary symptoms or mood symptoms). The cause of persistent dizziness may be multifactorial up to half of the time (19), and dyspnea may be due to more than 1 condition in a third of patients (20).

Multiple rather than solitary symptoms are the norm. Symptoms commonly travel in company rather than

solitude. In 2 studies totaling 1500 primary care patients who completed a checklist of 15 common physical symp­ toms, the proportion that endorsed 0 to 1, 2 to 3, 4 to 5, 6 to 8, and 9 or more symptoms was 21%, 23%, 21%, 22%, and 12%, respectively (3, 21). In a third study of 338 primary care patients, the proportion endorsing 0 to 1, 2 to 3, 4 to 6, and 7 or more symptoms was 25%, 30%, 31%, and 14%, respectively (8). Thus, multiple symptoms are the rule rather than the exception. Although symptom checklists (not unlike the traditional review of systems) might lead to an overendorsement of symptoms that are less clinically relevant, relying exclusively on the chief com­ plaint may underestimate symptoms (1).

A related topic is symptom clustering, which is studied most extensively in cancer (22) but also found in other diseases (23). A common cancer cluster is the sleep-pain- anxiety-depression-energy pentad, known as SPADE, wherein insomnia, pain, fatigue, and mood symptoms fre­ quently co-occur. The somatic-anxiety-depressive symp­ toms triad, known as SAD, is a related cluster consistently found across various medical populations (4, 15, 24). At the level of functional somatic syndromes, such as irritable

bowel syndrome, fibromyalgia, chronic fatigue syndrome, and others, not only do individual symptoms (25) fre­ quently overlap but syndromes often co-occur (26).

D i a g n o s i s : L i s t e n i n g t o t h e P a t i e n t

Most diagnoses for common symptoms can be made on the basis o f the patient’s history alone.

Empirical studies of patients presenting with general (various) somatic symptoms (6, 27-30) and particular symptoms (19, 20, 31) have suggested that most final di­ agnoses can be derived from the history (in about 75%) and physical examination (in about 10% to 15%), whereas diagnostic testing infrequently contributes essential infor­ mation (Table 2). The central diagnostic role of the history and physical examination has also been shown in other studies (32, 33). This is confirmed by surveys showing that physicians (34) and medical students (35) attribute more than 80% of diagnostic information to the history and physical examination. Ironically, the hierarchical ordering of reimbursement in the United States (tests are more costly than physical examination, which is more expensive than the history) is converse to the diagnostic value of these services. Although billing practices disproportionately in- centivize tests and procedures, the physical examination garners more financial reward than a detailed interview according to evaluation and management coding rules that pay for examining more bodily parts regardless of their relevance to the patient’s medical problems.

Clinical examinations should be symptom-focused and evidence-based rather than complete.

We have to make the interview and physical examina­ tion efficient by gathering data that, like a good diagnostic test, have reasonable operating characteristics (sensitivity, specificity, and predictive value) for classifying the patient’s

Table 1. P ro p ortio n o f S o m atic S ym pto m s T h a t A re M e d ic a lly U n e x p la in e d

Study, Y e a r (R e fe rence ) Study S e ttin g Study D esign P atients , n

M e th o d fo r C lass ify in g S ym p tom s as

M e d ic a lly U n e xp la in ed M e d ic a lly U n e xp la in ed

S ym p tom s (9 5 % C l), %

Kroenke and M ange lsd o rff, 1989 (6)

Prim ary care C hart review 1000 O ne physician cha rt au d ito r using im p lic it crite ria 74 (7 1 -7 8 )

Khan e t al, 2003 (7) Prim ary care C hart review 4 5 0 T w o physician cha rt aud ito rs using exp lic it criteria; exce llen t in te rra te r re liab ility ( k = 0 .75)

34 (3 0 -3 8 )

M a rp le e t al, 1997 (8) Prim ary care Prospective cohort 338 Clinical ju d g m e n t o f pa tien t's p rim ary care physician 33 (2 8 -3 8 ) S te inbrecher e t al, 2011 (9) Prim ary care Survey 620 Clinical ju d g m e n t o f pa tien t's p rim ary care physician 37 (33—41) Kroenke e t al, 1994 (3) Prim ary care Survey 1000 Clinical ju d g m e n t o f pa tien t's p rim ary care physician 2 0 * (1 8 -2 2 ) Reid e t al, 2001 (10) Specialty c lin ics t C hart review 361 O ne physician ra te r review ed consu lta tions on fre q u e n t

a ttenders to 12 c lin ic types; exce llen t ra te r re liab ility ( k = 0 .7 6 -0 .8 8 )

27 (2 2 -3 2 )

Kroenke and Price, 1993 (11)

General population Survey 13 328 S tructured in te rv iew using the D iagnostic In te rv iew Schedule

35 (3 4 -3 6 )

Escobar e t al, 2 0 10 (12) General population Survey 4 8 64 T w o physician raters independen tly review ed structured 31 (3 0 -3 2 ) in te rv iew data; bo th had to agree th a t sym p tom was unexpla ined

* Certain somatic symptoms were not counted as medically unexplained if they were part of the diagnostic criteria for patients who qualified for a depressive disorder (e.g., fatigue or insomnia) or an anxiety disorder (e.g., chest pain or palpitations in panic disorder). t “Frequent attender” sample defined as persons in the top 5% of outpatient use. Rates of medically unexplained symptoms were particularly high in 5 of the 12 clinics, including gastroenterology (54%), neurology (50%), cardiology (34%), rheumatology (33%), and orthopedics (30%).

w w w . a n n a ls .o r g 21 October 2014 Annals of Internal Medicine Volume 161 • Number 8 581

R e v i e w A Practical and Evidence-Based Approach to Common Symptoms

Table 2. D ia g n o s tic Y ie ld o f H is to ry and Physical E xam in a tio n in P atien ts W ith C o m m o n S ym pto m s

Study, Year (Reference) Symptom Study Setting Patients, n Follow-up, mo History, % Physical Examination, %

Hampton et al, 1975 (27) General Primary care 80 2 82 9 Sandler, 1980 (28) General Primary caret 630 18-30 56 17 Kroenke, 1989 (6) General Primary care 382* 11§ – –

Gruppen et al, 1988 (29) General Primary care 119 0 94 –

Peterson et al, 1992 (30) General Primary care 80 2 76 12 Schmitt et al, 1986 (20) Dyspnea Hospital inpatients 146 0 74 –

Kroenke e ta l, 1992(19) Dizziness Various hospital clinics 102 12 76 4 Martina et al, 1997 (31) Abdominal pain Primary care 112 29|| – –

Martina et al, 1997 (31) Chest pain Primary care 78 29|| – m m m

* A = retrospective; B = single rater per case (or for all cases) using unstructured assessment; C = no explicit criteria for diagnostic classification; D = poor description of sample; E = diagnostic evaluation varied considerably among patients or was not well-described, t Setting had a special interest in cardiologic conditions. t Unit of analysis was the symptom rather than the patient. A minority of patients had more than 1 symptom. § Mean follow-up; range was not provided. || Mean follow-up; range was 18 to 56 mo. H Combined proportion may be an underestimate because the study reported only the diagnostic contribution of the history.

symptoms. The standard mantra handed down to medical students of “do a complete history and physical” is not cost-effective in most instances. Instead, a symptom- focused clinical examination is preferable. For example, a 5-minute evaluation targeting a few key items from the history and physical examination is an evidence-based ap­ proach to the initial evaluation of dizziness in primary care (36). Clinical time comes at a premium and cannot be squandered.

Besides a low diagnostic yield, testing has other important limitations.

The likelihood of detecting a serious condition may be as low as 0.5% to 3.0% when diagnostic tests are ordered in patients with a low probability of disease (37, 38). This means that a diagnostic test with 90% sensitivity and 90% specificity would yield 4 to 19 false-positive results for ev­ ery true-positive result in patients for whom the test is ordered simply to rule out a disease for which clinical sus­ picion is already low. False-positive results may trigger ad­ ditional and sometimes invasive procedures as well as anxiety, which may linger for several months or more. False-negative results can also be a concern. For example, the negative predictive value of abdominal computed to­ mography in patients presenting to the emergency depart­ ment with undifferentiated upper abdominal pain is only 64%, which means up to 1 of every 3 normal scans in this population may be a false-negative result (38). A meta­ analysis of 14 randomized trials that examined the utility of diagnostic tests in patients with a low pretest probability of disease found no benefits on reducing symptom persis­ tence, illness worry, or anxiety (37).

P r o g n o s i s : F o l l o w i n g t h e P a t i e n t

Serious diseases not initially expected seldom emerge dur­ ing long-term follow-up.

582 21 October 2014 Annals of Internal Medicine Volume 161 • Number 8

Serious diseases that are unsuspected in the initial eval­ uation of common symptoms seldom emerge in long-term follow-up (Table 3) (31, 39-49). Medical textbooks that provide exhaustive tables of the differential diagnosis of common symptoms, such as headache or fatigue, rarely provide an epidemiologic rank ordering of particular causes. Such compendiums or “laundry list tables” include many conditions that are infrequent or rare causes of a particular symptom. We may also be unduly influenced by the Sherlock Holmes approach exemplified in academic clinicopathologic conferences or in popular television se­ ries, such as House, in which medical sleuths track down the needle-in-the-haystack diagnosis. What is glossed over in these glamorous depictions is the rarity of the villain relative to the “usual suspects.”

A quarter o f symptoms become chronic. Longitudinal studies have shown that approximately

25% of somatic symptoms persist at 1 to 2 weeks (8, 50, 51), 3 months (51), 12 months (52), and up to 5 years (53) after a patient presents in primary care with a symp­ tom. Indeed, 1 study followed the same cohort of 500 primary care patients presenting with a somatic symptom and found the proportion with symptom persistence to be similar at 2 weeks (29%), 3 months (21%), and 5 years (24%) (53). Thus, a rule of thumb would be that although most patients presenting with symptoms in primary care improve within weeks to several months, about a quarter develop chronic symptoms. Even patients with somato­ form disorders, originally considered to have high persis­ tence rates over time, show improvement rates of 50% to 75% (54). This can inform diagnostic testing and clinical management in that a conservative approach (symptom- specific management and limited testing) is sufficient for most patients, whereas a more extensive work-up can be reserved for the fraction of patients with persistent symp-

w w w .a n n a ls .o rg

A Practical and Evidence-Based Approach to Common Symptoms Review

Table 2—Continued

Combined (95% Cl), %

Study Sample Study Limitations

A B c D E 91 (85-97) New patients referred by family physicians to a general medicine clinic – 7 7 a i .7 73 (70-76) New patients referred by family physicians to a general medicine clinic / 7 7 7 90 (87-93) Symptoms documented in chart review of 1000 internal medicine clinic patients; included are

382 o f 567 symptoms tha t had testing beyond history and physical examination / 7 7 7

94H (90-98) Primary care walk-in clinic; correct diagnosis produced by chief complaint alone in 79% o f cases 7 7 7 7 88 (81-95) Internal medicine clinic patients w ith a new or previously undiagnosed condition ' Spy; . .7 V 7411 (67-81) Patients hospitalized w ith dyspnea 7 80 (72-88) Structured assessment o f patients w ith persistent dizziness isifsimiRW 72 (64-80) Consecutive clinic patients w ith chief complaint o f abdominal pain 7 88 (81-95) Consecutive clinic patients w ith chief complaint o f chest pain 7

toms. Moreover, chronic or recurring symptoms may re­ quire a different management approach.

T herapy: Caring for the Patient

“Management” may be a preferable term to “therapy” or “treatment” because the latter terms tend to connote greater symptom specificity or targeting of particular mechanisms. The emphasis here is on strategies that cross symptom boundaries rather than those unique to a partic­ ular symptom. Nevertheless, the suggestions are selective rather than comprehensive, with the intent to highlight several principles that tend to be overlooked or devalued. Other strategies for managing poorly explained symptoms are reviewed elsewhere (17, 55-57).

Communication is therapeutic. Symptom-related concerns and expectations may be as

important as symptom severity or duration in prompting a health care visit for the subset of persons who actually seek care for their symptoms. Common patient expectations in­ clude provider answers to questions (for example, “What is causing my symptom?” and “How long is it likely to last?”) and subsequent actions (treatments, tests, and referrals) (58-60). However, the most common unmet expectations after symptom-related visits relate to insufficient provider explanations about diagnosis and prognosis rather than in­ adequate physician actions. Consequently, 2 useful ques­ tions a provider might consider in closing a symptom- related encounter relate to patient-specific worries and wants: “Was there anything else you were worried about?” and “Was there anything else you thought might be helpful?”

What kind of diagnosis should be offered to the sub­ stantial proportion of patients in whom the symptom is poorly explained? First, one should maintain etiologic neu­ trality and feel comfortable with symptom-only diagnoses (headache, fatigue, and vertigo) rather than modifiers that are unsupported by mechanistic evidence. Second, prema­ ture psychologization should be avoided; the absence of a physical disease that definitively accounts for the symptom should not lead a physician to automatically default to a psychological explanation. Instead, positive evidence of de-

w w w . a n n a ls .o r g

pression, anxiety, or other mental disorders should be elic­ ited. However, patients often volunteer clues to psychoso­ cial factors that clinicians can pick up on and incorporate into their diagnostic explanations (55). Third, avoid nor­ malization. Although clinical examination and testing may not uncover findings to substantiate a specific diagnosis, patients do not like to hear that “everything is normal” (61). Fourth, providing a mechanistic explanation, even if tentative, may be useful (for example, central sensitization contributing to chronic widespread pain, neurotransmitter imbalances accounting for the somatic symptoms associ­ ated with depression or anxiety, or neurally mediated co­ lonic contractions in irritable bowel syndrome).

Some treatments may be effective across various symptoms. Cognitive behavioral therapy and antidepressants have

proven beneficial across various symptoms and symptom syndromes and have an effect that is independent of the patient’s depression status (62). Likewise, exercise has proven beneficial in pain conditions (63), chronic fatigue (64), depression (65), and anxiety (66). Further, there is emerging evidence for the benefits of other types of psy­ chotherapy (67), mindfulness-based stress relaxation (68), and some types of complementary and alternative medicine therapies (69) for various symptoms. Treatments that are effective for multiple types of symptoms suggest that symp­ toms may share a common etiologic pathway or that some treatments may have more than 1 mechanism of action.

Measuring symptoms is important for monitoring out­ comes and tailoring treatment.

Medical treatment is typically guided by measurement that, for some diseases, consists of findings on physical examination (for example, heart failure, hypertension, or neurologic conditions) or laboratory tests (for example, di­ abetes, hyperlipidemia, or anemia). Patient report is the fundamental metric for symptoms, and the use of validated measures has proven helpful for some symptom-based con­ ditions, such as depression, to adjust, switch, or combine treatments (70). The clinical utility of a patient-reported outcome measure is enhanced by it being brief, self- administered, easy to score, freely available (that is, public

21 October 2014 Annals of Internal Medicine Volume 161 • Number 8 583

R e v i e w A Practical and Evidence-Based Approach to Common Symptoms

Table 3 . R arity o f U nsuspec ted Serious D iag n o s es E m erg ing A fte r In itia l E va lua tio n o f C o m m o n S ym ptom s

Study, Year (Reference)

Symptom Study Setting Patients, n

Follow-up, m o

M ain Results Study Lim itations* * * §

A B D E o W asson e t al,

1981 (39) A bdom ina l

pain Prim ary care 552 4 M a le ou tpa tie n ts w ith abdom ina l pain (m edian du ra tion , 3

w k); specific diagnosis usually m ade in 1 w k (8 1 % ), w ith diagnosis ta k in g long e r than 3 m o in on ly 3 patients; o f the 4 3 8 patien ts w ith id iop a th ic pa