European Journal of Internal Medicine 13 (2002) 240–245

www.elsevier.com / locate / ejim

Original article

Assessing clinical probability of organic disease in patients with

involuntary weight loss: a simple score

J. Bilbao-Garay*, R. Barba, J.E. Losa-Garcıa,

´ H. Martın,
´ G. Garcıa´ de Casasola, V. Castilla,
´
I. Gonzalez-Anglada, A. Espinosa, C. Guijarro
´ Hospital Alcorcon
Unidad de Medicina Interna, Fundacion ´ , C /Budapest no. 1, 28922 Alcorcon
´ , Madrid, Spain

Received 13 August 2001; received in revised form 17 January 2002; accepted 21 January 2002

Abstract

Background: Involuntary weight loss (IWL) is a frequent complaint with a difficult diagnosis. Any one of a number of different
diseases may be the source of the symptom. However, there is no universal clinical protocol that can help physicians study this complex
syndrome. Methods: In March 1998, we defined a diagnostic protocol for the study of IWL. IWL was defined as an involuntary and
documented weight loss of at least 5% of the usual body weight in the previous 3 months. We analyzed 78 consecutive patients with IWL
who came to our clinic between March 1998 and December 2000. Results: An organic disease was found in 56% of cases; cancer,
metabolic and digestive diseases were the most common entities. Psychiatric problems were found in 33% of cases. After extensive study,
an idiopathic group of 11% was identified. The variables that were independently predictive of a final diagnosis of organic disease were:
age.50 years (OR: 8.6, CI 95%: 1.7–43.6), psychiatric symptoms (OR: 0.2, CI 95%: 0.1–0.8), smoking (OR: 14.3, CI 95% 2.3–74), the
presence of guide symptoms (OR: 8.0, CI 95%: 1.8–34.4), and anemia (OR: 3.1, CI 95%: 2.5–387). Sixteen percent of the patients died,
more often those suffering from organic diseases. Based on multivariate regression coefficients, a clinical risk score was established.
Conclusions: IWL is a complex and frequent syndrome with a 16% rate of mortality during the first year. A protocol based on clinical
data can help in the management of IWL. Our clinical prediction rule may help physicians to identify those patients with IWL who are
likely to have an underlying organic disease.  2002 Elsevier Science B.V. All rights reserved.

Keywords: Weight loss; Clinical protocol; Clinical prediction rule; Diagnosis; Prognosis

1. Introduction presenting with IWL; (2) to define a clinical protocol for

the study of IWL; and (3) to develop a clinical prediction
Involuntary weight loss (IWL) is a frequent complaint rule that might help in the management of these patients.
that causes concern to both the patient and the attending
physician because of the possibility of a severe underlying
disease. The best clinical approach to IWL remains 2. Patients and methods
unclear, especially in the absence of specific clinical clues
[1]. A few studies analyzed this problem, most of them in 2.1. Setting
the 1980s [2–6].
We undertook this study with the following objectives: Fundacion´ Hospital Alcorcon
´ is a public general hospital
(1) to evaluate the underlying clinical disorders in patients in the southwest of Madrid that serves about 250 000
inhabitants. Our internal medicine department has an
*Corresponding author. outpatient clinic where nearly 1400 new patients are seen
E-mail address: [email protected] (J. Bilbao-Garay). every year.

0953-6205 / 02 / $ – see front matter  2002 Elsevier Science B.V. All rights reserved.
PII: S0953-6205( 02 )00032-8
 J. Bilbao-Garay et al. / European Journal of Internal Medicine 13 (2002) 240 – 245 241

Table 1
Protocol for the study of patients with involuntary weight loss
1. Complete medical history: previous diseases, psychiatric problems, and
surgical procedures. Social problems (isolation, poverty, unavailability of
preferred foods, inadequate education) especially in the elderly. Drug history,
toxic habits, cigarette smoking, alcohol and illicit drugs

2. Physical examination: General appearance of the patient: mood and attitude.

Active search for lymphadenopathy, breast and thyroid abnormalities, organomegaly
or masses. Rectal and genital exam

3. Basic laboratory work: see Table 2

4. Additional tests: done as a function of the clinical data

(a) Abdominal CT scan and / or abdominal ultrasound if abdominal pain
(b) Mammography: if abnormal findings on breast examination
(c) Serum immunoglobulins and protein electrophoresis if hypercalcemia or increased
globular sedimentation rate
(d) Colonoscopy (or barium enema) if iron deficiency anemia, melena or inflammatory
bowel disease is suspected
(e) Upper endoscopy (or upper GI barium) for upper GI bleeding, or dysphagia
(f) ACTH test for weakness, pigmentation or hyponatremia / hyperkalemia
(g) Blood cultures for fever of unknown origin
(h) 72-h stool fat for chronic diarrhea
(i) Head CT for headache or neurological symptoms
(j) Temporal artery biopsy if giant cell arteritis is suspected
(k) Formal psychiatric consultation if no disease found or depressive symptoms

2.2. Study protocol illness in the previous month that could explain the weight
loss.
In March 1998 we defined a diagnostic protocol for the We considered a causal relationship between the weight
study of involuntary weight loss (IWL) adapted from loss and a given disease when a neoplasm was found or
Foster [7] (Tables 1 and 2). From March 1998 to De- when changes in weight loss paralleled changes in disease
cember 2000, all patients referred to our clinic with IWL status, as previous authors have done [3].
as the main clinical complaint were included in the study. Thirteen of the 91 patients were lost after the first visit
IWL was defined as an involuntary and documented and were not included in the analysis. We were thus left
weight loss of at least 5% of the usual body weight in the with 78 patients to analyze in whom a clinical diagnosis
previous 3 months or more than 10% during the previous 6 was reached or who were followed for at least 6 months.
months, as reported in other studies [1–6]. From their clinical charts we were able to obtain and
record demographic information (age, sex), toxic habits
2.3. Patients (smoking, alcohol consumption in g / day, use of illicit
drugs), previous diseases (hypertension, diabetes, other
Between March 1998 and December 2000, 91 of 2968 medical diseases, and previous psychiatric disorders),
patients over 15 years of age met the above-mentioned duration of the weight loss (number of days from the
criteria for IWL (3%). Patients were excluded from the beginning of the IWL to the first visit in the office),
study if: (1) they had voluntary weight loss or (2) if they accompanying symptoms, abnormal physical findings on
had been diagnosed as having a medical or psychiatric physical examination, degree of weight loss, complemen-

Table 2
Basic laboratory data for patients with involuntary weight loss
Complete blood cell count
Erythrocyte sedimentation rate
Urinalysis
Multiphase chemical screen: glucose, creatinine, SGOT, SGPT, alkaline
phosphatase, bilirubin, albumin, calcium, electrolytes, cholesterol, triglycerides
Ultrasensitive thyrotropin assay
Chest film
HIV testing if risk factors are present
Fecal occult blood testing at 40 years or older
242 J. Bilbao-Garay et al. / European Journal of Internal Medicine 13 (2002) 240 – 245

tary data (blood cell count, blood chemistry, liver en- The most frequent symptoms seen were asthenia (64%),
zymes, thyrotropin assay, radiological studies, and other anorexia (62%), depressive symptoms (sadness, insomnia,
complementary data registered in the clinical record), final easy crying, loss of interest) (27%), abdominal pain
diagnosis, and the number of days of follow-up. (26%), nausea and vomiting (5%), arthralgia (3%), and
For the purpose of analysis, patients were divided into miscellaneous symptoms (3%). Patients were followed up
two groups: group 1, which included patients with organic for 376.36244 days (range: 9–982 days).
disease (neoplasia or other medical diseases), and group 2, A diagnosis that could explain the IWL was reached in
patients with no organic disease (psychiatric illness and 69 (88%) patients (Table 3). Eighteen patients (23%) were
patients with unknown diagnosis). The idiopathic group, diagnosed as having neoplasia, 25 patients (32%) had
hereafter referred to as weight loss of unknown origin other medical conditions, and 26 patients (33%) had
(WLUO), was diagnosed after an extensive negative psychiatric disorders; in nine patients (11%) we did not
clinical study and a minimum follow-up of 3 months. find any disease to explain the IWL (WLUO). One patient
Patients finally diagnosed as having psychiatric disease did initially diagnosed as having WLUO was finally diagnosed
not have any other known diagnosis that justified IWL, nor with cancer at 6 months of follow-up.
had they had any acknowledged psychiatric illness in the Clinical symptoms and laboratory data are shown in
previous month that could have been linked to IWL. Table 4. The variables most frequently associated with
organic disease (group 1) were: male sex, age over 50
2.4. Statistics years, alcohol consumption.40 g / day, smoking, abnor-
malities in basic laboratory data, and the need for hospital
Patients were divided into two groups according to the admission (P,0.05). Depressive symptoms (sadness, easy
diagnosis of organic or non-organic disease. All demo- crying, insomnia, apathy) and previous psychiatric disease
graphic, clinical, and complementary data were compared were more frequent in patients in group 2 (no organic
between these groups. The two-tailed Student’s t-test for disease). Anorexia was presented more frequently in the
quantitative variables and the x 2 -test for dichotomous organic disease group, especially in cancer patients, al-
variables were used. Odds ratios and 95% confidence though this symptom did not reach a significant difference
intervals were estimated from the regression coefficients. (Table 4). Mean weight loss (9.265.1 kg; 1360.6% of
The most clinically relevant variables and those statisti- body weight) and the duration of symptoms (35% more
cally significant were analysed according to a multiple than 6 months; 39% less than 3 months) were similar in
logistic model to identify independent correlates of organic both groups. As a group, the patients with WLUO were
disease (a P-value,0.1 was used to include variables in very similar to the patients with a final diagnosis of
the model). A backward, stepwise procedure, with P.0.10 psychiatric disease, except for sex, with male patients
as the criterion for exclusion, was used to find the best clearly predominating in the WLUO subgroup (77 vs.
predictive models for organic disease. 11%, P,0.05), presence of psychiatric symptoms (11 vs.
We developed a clinical model to stratify the probability 50%, P,0.05), and previous psychiatric disease (11 vs.
of organic disease into high, moderate, and low categories. 61%, P,0.05). Age, physical examination, and laboratory
A simple, integral diagnostic score was computed from the data were similar in both subgroups.
multivariate logistic regression model, assigning points in Only 25 patients (34%) had a combination of clinical
proportion to the regression coefficients. The ideal score symptoms that helped in the diagnosis. These symptoms,
was defined as a score based on objective and reproducible defined as chief complaints pointing to disease of a specific
variables that were clinically relevant. organ (‘guide symptoms’), such as gastrointestinal or
All statistical analyses were done with the Statistical pulmonary symptoms, were present more frequently in
Package for the Social Sciences (SPSS, version 8.0) [8]. patients with organic disease (50 vs. 14%, P,0.001).
Patients in group 1 (organic disease) had lower hemo-
globin and cholesterol levels than patients with no organic
3. Results disease (group 2). In addition, globular sedimentation rate
(GSR) and liver enzymes were higher in patients with
Of the 91 patients with IWL seen at our internal organic disease (Table 4).
medicine clinic, 78 (86%) were included in the study. The Twelve patients (16%) died during the follow-up period,
13 patients who were not included because they did not and mortality was higher in group 1 than in group 2 (23 vs.
have enough follow-up to achieve a diagnosis did not 4%, P,0.001). Patients diagnosed as having WLUO did
differ in sex or age from the patients who were included. not deteriorate during the follow-up period, except for one
Fifty-one percent of the patients were women, and their patient who was eventually diagnosed as having dissemi-
mean age was 59619 years. nated cervical cancer at 6 months after the first visit and
The primary care physician referred 85% of the patients. who died 2 months later.
Eighteen patients (23%) were hospitalized for study due to The following variables were included in a logistic
a deterioration in their clinical conditions. regression model: age, sex, tobacco use, depressive symp-
 J. Bilbao-Garay et al. / European Journal of Internal Medicine 13 (2002) 240 – 245 243

Table 3
Diseases found in 78 patients with involuntary weight loss
Medical diseases: 25 (32%) Cancer 18 a : (23%) Psychiatric diseases: 26 (33%)
Digestive diseases: 5 (6%) Lymphoma 5 Depression 17 (22%)
Inflammatory bowel disease 2 Renal cancer 2 Anxiety 7 (9%)
Gastritis 1 Gastric cancer 2 Anorexia nervosa 2 (3%)
Chronic hepatitis 1 Colon carcinoma 2
Malabsorption 1 Uterine cancer 2 a
Metabolic diseases: 8 (10%) Lung cancer 1
Hyperthyroidism 3 Pancreas cancer 1
Diabetes mellitus 2 Prostate cancer 1
Alcohol abuse 3 Liposarcoma 1
Infections: 4 (5%) Laryngeal cancer 1
Urinary infection 1
Lung abscess 1
Lung tuberculosis 1
HIV infection 1
Miscellaneous: 7 (9%)
Giant cell arteritis 2
Systemic lupus erythematosus 1
Emphysema 1
Cerebrovascular disease 1
Non-specific lymphadenopathy 1
Amyotrophic lateral sclerosis 1
a
One patient originally classified as having weight loss of unknown origin was diagnosed 6 months later with uterine cancer.

toms, previous disease, the presence of guide symptoms, Based on multivariate regression coefficients, a clinical
normal physical examination, and abnormalities in labora- risk score ranging from 24 to 7 points was established.
tory data. The variables that were independently predictive The best cut-off score for identifying a high probability
of a final diagnosis of organic disease were: the presence group was 3 points. Patients with a total score of 1–3
of guide symptoms (OR: 8.0, CI 95%: 1.8–34.4); anemia points have a moderate probability and patients with less
(OR: 3.1, CI 95%: 2.5–387); age.50 years (OR: 8.6, CI than 1 point have a low risk of organic disease (Table 5).
95%: 1.7–43.6); psychiatric symptoms (OR: 0.2, CI 95%: The sensitivity of the formula is 94% with a specificity of
0.1–0.8), and smoking (OR: 14.3, CI 95% 2.3–74). 68%.

Table 4
Clinical and laboratory variables in 78 patients with involuntary weight loss
Group 1 Group 2 P-value
N543 N535
Age (years6S.D.) 67616 50619 ,0.001
Previous psychiatric disease (%) 9 48 ,0.001
Male sex (%) 68 28 0.001
Smokers (%) 52 23 0.008
Alcohol (ingestion more than 40 g / day) (%) 25 6 0.02
Presence of guide symptoms (%) 50 14 0.001
% kg lost 8.964 9.566 0.601
Abdominal pain (%) 20 34 0.167
Sadness, anxiety, and other psychiatric 16 40 0.01
symptoms (%)
Normal physical examination (%) 41 97 ,0.001
Anorexia (%) 68 54 0.206
Asthenia (%) 68 60 0.450
Hemoglobin (mean6S.D., g / dl) 1262.5 1461.2 0.001
Alkaline phosphatase (mean6S.D., U / l) 2506158 145646 ,0.001
SGOT (mean6S.D., U / l) 32629 1765 0.005
SGPT (mean6S.D., U / l) 45681 1668 0.03
Cholesterol (mean6S.D., mg / dl) 174643 209640 0.001
Erythrocyte sedimentation rate, 45639 1463 ,0.0001
(mean6S.D., mm / h)
Need for hospital admission (%) 36 6 ,0.001
Mortality (%) 23 4 ,0.001
244 J. Bilbao-Garay et al. / European Journal of Internal Medicine 13 (2002) 240 – 245

Table 5 veloped a clinical prediction rule which, by combining

Clinical prediction rule for calculating the organic disease risk in patients simple parameters, can help predict the evolution of
with involuntary weight loss
patients with IWL. We feel this formula needs to have a
Clinical variables Score very good sensitivity, even if this means a loss in
Age (.50 years) 13 specificity, in order to avoid classifying low-risk patients
Smoking 12 with those who really present an organic disease. This
Presence of guide symptoms 11 system can help the physician in the differential diagnosis
Presence of psychiatric symptoms 24
Anemia 11 of IWL and contribute to the use of diagnostic procedures
only in patients with a high risk of suffering an organic
High probability: .3 points. Moderate probability: 1–3 points. Low
disease. This formula should be validated in a prospective
probability: ,1 point. Sensitivity 94% (proportion of people with medical
or neoplastic disease classified within the pathologic group). Specificity way in another set of patients with IWL before expanding
68%. its use, as we have only created or derived the rule from
our own patients. This validation process may require
several studies to fully test the accuracy of the rule in
4. Discussion different clinical settings [12].
Marton et al. [2] developed a different formula, but
In general, weight loss occurs because of decreased given that nearly all of their patients were male, the
caloric intake, alterations in energy metabolism, or loss of applicability of their prediction rule is lower than ours.
calories in the urine or feces. Different abnormalities in Sullivan et al. found that oral problems are the best
energy metabolism have been described and multiple predictor of significant weight loss, but their study was
factors contributing to weight loss may be present in a conducted among elderly, male patients [13]. In the future,
particular patient with cancer, AIDS, or in elderly patients the application of a validated formula like the one we
[1,9]. The precise mechanisms underlying weight loss are propose, based on clinical and basic laboratory data, will
currently being elucidated and probably involve the action enable diagnostic procedures to be used only for patients
of different hormones and cytokines [1,9]. with a high risk of organic diseases. Until then, such
A wide variety of clinical conditions and procedures as endoscopy and computed tomography will
pathophysiological mechanisms may lead to IWL: tumors, continue to be used for most patients with IWL, despite
infections, metabolic and psychiatric diseases. Fifty-six our knowledge that nearly half of them do not require
percent of our patients had organic diseases, 33% psychiat- them.
ric dysfunctions, and 11% had no identifiable cause. Weight loss of unknown origin (WLUO) was diagnosed
Similar figures have been published by others [2–6]. in 11% of our patients. These patients, in whom no entity
IWL is not an uncommon complaint: in our internal responsible for the IWL can be found, represent 10–35%
medicine clinic, it represents 3% of the patients, and it can of all cases of IWL [2–6]. Clinical management of WLUO
reach 13% in older patients [10]. Weight loss is a matter of is controversial: some authors recommend an extensive
concern to most physicians for it is well known that weight study and others a meticulous observation [1]. We think
loss is associated with mortality [9–11]. Although some- that patients diagnosed as having WLUO may in fact have
times it is not possible to document the weight loss, a an occult psychiatric disorder. Most of our patients with
convincing history may be sufficient to pursue the study WLUO were male, and in our culture male patients do not
[1,2]. easily recognize depressive symptoms. One of our patients
Weight loss is a potentially serious symptom: 16% of initially diagnosed as having WLUO was later found to
our patients died in the first year of follow-up, most of have a disseminated cancer at 6 months of follow-up. This
them suffering from organic diseases. On the other hand, is why we follow patients with WLUO for a minimum of 6
44% of our patients had no organic disease, 33% had months to rule out an occult disease. In Marton et al.’s
psychiatric diseases and, in 11% of cases, no problem was experience, four patients (4%) in this idiopathic group
identified after a prolonged follow-up. It is our experience were also diagnosed with neoplasia in the follow-up period
that patients with no organic disease have a better prog- [2].
nosis. Being able to quickly differentiate these groups can Our cases were outpatients referred to an internal
be useful, to both patients and clinicians. Although at times medicine clinic. This can explain some of the differences
diagnosis is easy, it can be difficult with most patients. The observed with other studies [2–6]. Patients from the
presence of a guide symptom (34% in our patients, 50% in Marton et al. study [2] came from a Veterans hospital.
Marton et al. [2]) may be useful, but it may also be Virtually all patients were male, with a high frequency of
misleading, as it may be present in patients without organic diseases, particularly neoplasia. In our sample,
organic diseases and vice-versa. organic diseases were also more common among males.
We have demonstrated that a combination of clinical and Similarly, the Rabinovitz et al. study [3], conducted in a
basic laboratory data obtained during the first visit can be hospital ward, found a high prevalence of organic diseases,
useful to classify patients with IWL. We have also de- as we did for those patients of ours requiring hospital
 J. Bilbao-Garay et al. / European Journal of Internal Medicine 13 (2002) 240 – 245 245

admission. Thompson and Morris [4] limited their study to we recommend a follow-up of at least 6 months in patients
elderly patients, and thus their results cannot be extrapo- with no apparent diagnosis (WLUO) in order to rule out an
lated to younger patients. This age group can have other organic disease.
etiologies of involuntary weight loss, the most common
conditions appearing to be related to dementia, depression,
drug reactions, oral problems and chronic diseases [13,14]. Acknowledgements
Leduc et al. [5] found a high proportion of psychiatric and
idiopathic patients in their study; organic diseases were We thank Alejandra Perez del Real for her helpful
found in no more than 29% of their IWL patients. In their technical assistance.
study, they only included patients with a negative basic
study, thus representing a group similar to the one we have
defined as WLUO. Not surprisingly, as with our own
group, their WLUO group had a low prevalence of serious References
organic disorders. Therefore, their results cannot be ex-
[1] Reife CM. Involuntary weight loss. Med Clin North Am
trapolated to the whole range of patients with IWL seen in
1995;79:299–313.
a general internal medicine clinic. Huerta and Viniegra [6] [2] Marton KI, Sox Jr. HC, Krupp JR. Involuntary weight loss:
limited their study to patients who lost weight in less than diagnostic and prognostic significance. Ann Intern Med
6 months. Thirty-five percent of our patients would have 1981;95:568–74.
been excluded if we had applied this criterion. From our [3] Rabinovitz M, Pitlik SD, Leifer M et al. Unintentional weight loss: a
retrospective analysis of 154 cases. Arch Intern Med 1986;146:186–
experience, as well as that of Rabinovitz et al. [3], we are
7.
inclined to say that patients who lose weight within 6 [4] Thompson MP, Morris LK. Unexplained weight loss in the ambulat-
months do not differ from patients who lose weight more ory elderly. J Am Geriatr Soc 1991;39:497–500.
slowly. However, additional studies are needed to confirm [5] Leduc D, Rouge PE, Rousset H et al. Clinical study of 105 cases of
this. Quick weight loss could suggest an organic disease as isolated weight loss in internal medicine. Rev Med Interne
1988;9:480–6.
the cause, but our data do not support this hypothesis.
[6] Huerta G, Viniegra L. Involuntary weight loss as a clinical problem.
Limitations of our study could be related to the number Rev Invest Clin 1989;41:5–9.
of patients seen. It is possible that if we had had a larger [7] Foster DW. Gain and loss in weight. In: Harrison’s principles of
sample of IWL patients, other diseases might have ap- internal medicine, 14th ed., McGraw-Hill, 1998, pp. 244–6.
peared. However, our results are in accordance with other [8] Statistical Package for Social Sciences (computer program). Win-
dows version, release 8.0. SPSS, Chicago, 1998.
studies [2–6], especially with regard to the idiopathic
[9] Wise GR, Craig DC. Evaluation of involuntary weight loss. Where
group. Our sample represents a group of patients seen by a do you start? Post Medicine 1994;95:143–50.
general internist, and this sample may differ from those [10] Wallace JI, Schwartz RS, LaCroix AZ, Uhlmann RF, Pearlman RA.
seen by general practitioners. Our patients may be sicker Involuntary weight loss: incidence and clinical significance. J Am
than those seen by general practitioners, but we think that Geriatr Soc 1995;43:329–37.
[11] Yaari S, Goldbourt U. Voluntary and involuntary weight loss:
the protocol we propose can also be useful for them.
associations with long term mortality in 9228 middle aged and
IWL is a complex syndrome with 16% mortality in the elderly men. Am J Epidemiol 1998;148:546–55.
first year in our experience, a figure similar to that found in [12] McGinn TG, Guyatt GH, Wyer PC, Naylor CD, Stiel IG, Richardson
other studies [2]. Since this is a frequent syndrome that WS. User guides to the medical literature. How to use articles about
consumes considerable diagnostic resources (8% of the clinical decision rules. J Am Med Assoc 2000;284:79–84.
patients consult a general internist for IWL [1,9]), IWL [13] Sullivan DH, Martin W, Flaxman N, Hagen JE. Oral health problems
and involuntary weight loss in a population of frail elderly. J Am
must receive more attention. We think that a protocol based Geriatr Soc 1993;41:725–31.
on clinical data can help in the management of IWL, and [14] Palmer R. Failure to thrive in elderly: diagnosis and management.
we present a prediction rule here for that purpose. Finally, Geriatrics 1990;45:47–55.