REPEATED SUICIDAL ATTEMPTS BY THE INGESTION OF SHARP FOREIGN BODIES (FB) IN A SEVERELY DEPRESSED PATIENT H Yang, Y Bains. Jersey City Medical Center, Jersey City, NJ and Seton Hall University School of Graduate Medical Education FB ingestion is common in children between the ages of 6 months and 3 years, and coins are the most common FB ingested by children. Adults who intentionally swallow FB arc usually psychiatrically impaired, under the influence of alcohol, or prisoners seeking the relative comfort of hospitalization. We report the unique case of a patient under psychiatric evaluation for severe depression, who ingested multiple sharp FB in an apparent suicidal attempt. 80-90 % of the FB that reach the stomach are expelled without complication. The frequency of perforation due to FB is estimated to be less than 1%, but sharp FB may cause perforation in up to 15 to 35% of patients. The frequency of complication depends on the shape and size, of FB and anatomic variations causing sharp angulations. A 39 year old male resident of a mental institution was hospitalized after he swallowed several razor blades wl)ich were removed surgically from his stomach. He swallowed another set of razor blades which be kept in his pocket at the moment ofdischarge and underwent a second gastrostomy. During recovery, he continued to swallow sharp FB including a nail, a thumb tack, a screw, and a metal plate over a month period. The patient was able to acquire the objects despite being under close observation by staff members. Due to the continual ingestion of FB and multiple recent surgeries, it was decided to withh~ld surgery or other invasive means of recovery. The patient was observed closely for any evidence of sepsis and daily abdominal radiographs were performed to ensure that the FB continued to propagate distally. The.metal plate needed to be removed from the rectum manually, however all the others passed spontaneously without causing perforation or gastrointestinal bleeding. We were unable to find the management of a similar case in a review of the literature. In general, any sharp FB ingested should be removed endoscopically or surgically before they pass the stomach, or if risk of perforation appears to be high. When a patient presents with repeated ingestion of sharp FB despite being under observation, it may be reasonable to monitor them and await spontaneous evacuation. DO GASTROINTESTINAL (GI) SYMPTOM SUBGROUPS EXIST? A CLUSTER ANALYSIS BASED ON SELF-REPORTED SYMPTOMS FROM A COMMUNITY-BASED RANDOM SAMPLE. A.R.Zinsmeister, A.B.An, N.J.Taliey, Dept. of Health Sdenc~s Research, Mayo Clinic and Dept. of Medicine, Univ. of Sydney. Considerable controversy surrounds attempts to document subgroups based on symptoms such as abdominal pain, and altered bowel habit. Identifying underlying symptom complexes or combinations of symptoms that stratify subjects into distinct subgroups is difficult due to the overlapping presence of many individual symptoms, and the uncertainty of exactly how many subgroups exist in a given study group. In addition, studies based on referral practices at large medical centers may not accurately reflect the general population. AIMS Determine whether subgroups in the general population can be identified based on selfreported GI symptoms. METHODS An ageand gender-stratified random sample of healthy residents aged 20-95 of Olmsted Co., MN, were mailed a validated self, report questionnaire; 3022 (74%) responded. Based on the (symmetric) uncertainty measure of association for categorical data (1), a principle components decomposition of the association matrix for 19 GI symptoms (including frequency and severity of pain, the Manning criteria (MC) for irritable bowel syndrome (IBS), nausea, heartburn and dysphagia) and age, gender and n~rital status was used to develop principle components scores (pc-scores) which were thus weighted sums of indi~,idual symptoms and sociodemographic variables. The pc-scores were interpreted using logistic discriminant analysis to distinguish a priori subgroups (e.g., IBS vs. non-lBS). RESULTS Subjects were first stratified by presence of reported abdominal pain during the previous year. In the pain subgroup, five pc-scores provided over 95% concordance between the a priori definition of IBS (2.2 MC with frequent abdominal pain) and the predicted probability of IBS from the logistic model. Their distributions are summarized in the table. PC-Scores (mean :t: SE) non-~BS 1.84+.02 3.225:.03 0.36+.02 -1.11+.04 0.29+.04 IBS 2.55+.03 3.76+.05 0.56+.03 -0.83+.05 -0.67+.04 Although these pc-scores jointly could discriminate subjects with IBScompatible symptoms, their distributions indicated a continuum of underlying (latent) factors rather than distinct clusters of subjects in the general popUlation. CONCLUSIONS Symptom scores can be developed which accurately reflect reported symptom complexes (IBS), but they imply a continuum of underlying symptom composition rather than distinct symptom subgroups in the healthy general population. (1) Goodman LA, Kruskal WH (1972) l Am Statistical Assoc, 67,415-421.