Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Serum Chloride Analysis, Bromide Detection, and the Diagnosis of Bromism R O B E R T E. W E N K , M.D., J A C K A. L U S T G A R T E N , P H . D . , N. J O H N P A P P A S , R O B E R T I. LEVY, M.D., AND R O G E R J A C K S O N , M.D., M.D. Departments of Pathology and Medicine, Sinai Hospital of Baltimore and Crownsville Hospital Center, Baltimore, Maryland ABSTRACT Wenk, Robert E., Lustgarten, Jack A., Pappas, N. John, Levy, Robert I., and Jackson, Roger: Serum chloride analysis, bromide detection, and the diagnosis of bromism. Am J Clin Pathol 64: 4 9 - 5 7 , 1976. Current methods for determining serum chloride concentration vary in specificity. Laboratory detection of bromide in serum is usually dependent on spuriously high chloride values. In the absence of historical information, the chemical identification of bromide in serum may be the only diagnostic sign of bromism. Bromide may be found in as many as 1% of sera obtained from inpatients of a general hospital. Ion-selective electrodes and some thiocyanate methods for chloride analysis can be highly sensitive to bromide (interference). Chloride electrodes are also sensitive to iodide. (Key words: Chloride; Bromide; Bromism; Ion-selective electrode [ISE].) BROMIDE THERAPY has largely been out- tion, which in turn depends on the method moded by the use of other hypnotics and of chloride analysis, and 3) the interference sedatives. Bromide poisoning has therefore of bromide with various methods of become an unfamiliar disorder; a 1975 chloride analysis. review of 100 toxicologic agents 1 that omits bromide from the list attests to the obMethods solescence of bromism as a suspected diagnosis. Nevertheless, occasional cases of Effect of Inorganic Bromide on Methods bromide intoxication (bromism) are still of Chloride Analysis reported. 3,10 Since some cases are first sus(1) Aqueous Br~ solutions (100 mmol. pected as a result of apparently elevated Br~ per 1.) were prepared by weighing serum chloride values, diagnosis is partly out sodium bromide salt. Dilutions yielda function of the chloride method utilized ing solutions of concentrations ranging by clinical laboratories. 2,9 We have evalfrom 1 to 100 mmol. per 1. were made. uated 1) the occurrence of bromide in These solutions, which contained no contemporary serum samples of hospitalchloride, were subjected to chloride analyized patients, 2) the dependence of the sis by five common procedures: ion-selecdiagnosis of bromism on bromide detective electrode (Stat/Ion, Technicon Corp., Received February 18, 1975; accepted for publica- Tarrytown, N.Y.), coulometry (Chlorition April 10, 1975. dometer, American Instrument Co., Silver Address reprint requests to Dr. Wenk: Department of Pathology, Sinai Hospital, Baltimore, Spring, Md.), and thiocyanate-colorimetry Maryland 21215. (Chloride Analyzer, I n s t r u m e n t a t i o n 49 50 WENK ET AL. AJ.C.P.—Vol. 65 were compared with the apparent measured CI - concentrations in all three kinds of solutions. Comparisons were expressed as a ratio of apparent (measured) chloride minus known chloride divided by known bromide: [Apparent Cl~] - [known CI - ] / [known B r - ] 20 40 60 80 100 BROMIDE CONCENTRATION - mmol / liter FIG. 1. Effects of bromide and bromide plus chloride to constant halide of 100 mmol. per 1. in aqueous solution on various methods for chloride. a, DuPont ACA plus chloride; b, DuPont ACA— no chloride; c, Stat/Ion—no chloride; d, Stat/Ion, lus chloride; e, Chloride Analyzer plus chloride; / Chloride Analyzer, no chloride; g, Chloridometer, no chloride; h, Chloridometer plus chloride; i, Theoretical one-to-one response of bromide to chloride. Laboratories, Lexington, Mass.; Automatic Clinical Analyzer, DuPont, Wilmington, Del.; Sequential Multiple Analyzer 6/60, Technicon Corp.).* (2) Similar solutions were prepared in protein-based media and were analyzed by the same procedures. (3) Bromide solutions of similar concentrations were also prepared in saline solution so that total halide, bromide plus chloride, equalled a constant 100 mmol. per 1. T h e concentrations of the two halides together were considered more clinically relevant since physiologically, bromide replaces chloride 12 and because these halides occur together in laboratory samples. These solutions were analyzed for chloride by the procedures noted above. (4) The known concentrations of B r * Apparent chloride test results have not been corrected for those values exceeding the linearity of any particular method. Ratios were plotted graphically against the known bromide concentrations for each method in order to delineate bromide interference with chloride analyses. (5) A study of ion-selective electrodes (ISE) was made because of the relatively recent emergence of ion-selective electrode technology as a routine clinical laboratory tool. To study the effect of loss of selectivity over time, electrode responses using a brand-new CI - sensor electrode and one utilized for several months in our laboratory were compared. Study was made of interelectrode variation, fatigue, change in response time, and the effects of "poisoning" by B r - , I - , and F - . T h e sensors were exposed to pure bromide solutions, to pure chloride solutions, or to mixtures of B r - and CI - in solutions of constant halide concentration. Exposure of the chloride electrodes to bromide (10 mmol. B r - + 90 mmol. Cl per 1.), for periods of 1.5 minutes and for 1 hour, was followed by tests for sensor damage, i.e., refractory or augmented response to aqueous CI - solutions of 100 mmol. per 1. and solutions of CI - + B r (total halide = 100 mmol. per 1., Br~ = 5, 10, 20 mmol. per 1.). The sensors were similarly exposed to aqueous iodide solutions (1 mmol. I + 99 mmol. CI - per 1.) and fluoride solutions (1 mmol. F - + 99 mmol. CI - per 1.). Frequency of Bromide Occurrence in Serum Samples (1) Random sampling was made of hospitalized patients' sera over several months. The samples, 133 sera, were analyzed for January 1976 51 SERUM Cl~ METHODS, BROMIDE AND BROM1SM bromide spectrophotometrically. 13 T h e bromide method was evaluated by calculating the percentage of bromide recovered from aqueous solutions (2-100 mmol. per 1.) in the presence of various concentrations of chloride ( 0 - 9 8 mmol. per 1.). Recovery was also determined from serum pools containing similar concentrations of halides. Precision was determined from statistical analysis of 18 repeated B r - determinations on serum pools containing 1.8 to 3.7 mmol. per 1. (2) It was found that routine analysis of electrolytes by ion-selective electrodes over six months occasionally produced spuriously high chloride values (range 110-306 mmol. per 1.). These values were not accompanied by cation abnormalities and caused a decrease in the normal anion gap ( [ N a + ] minus the [Cl~ + H C 0 3 - ] ) . Spurious chloride elevations occurred in about 0.75% of samples. 9 Each anomalous result initiated a search for historical evidence that the patient used bromide agents. The serum in question was chemically analyzed for bromide. (3) During the course of study, three cases of bromism were diagnosed, and are described. (4) An attempt was made to determine the effect of small amounts of bromide in serum on chloride analysis. One of the authors intentionally ingested 900 mg NaBr (about 14.5 mg. per kg.). Chloride analysis by coulometry, electrode, and thiocyanate (Chloride Analyzer) methods was carried out. Results Method Studies (1) Aqueous bromide solutions, analyzed for chloride by several methods, registered variable apparent chloride concentrations (Fig. 1). All methods showed positive interference—the higher the bromide concentration, the higher the ap- 5.0 A (no chloride) 4.0 3.0 - o ex CO ~ 2.0 01 1.0 - id *e t=*= 3.0 2.0 1.0 20 40 60 80 100 BROMIDE CONCFNTRATION - mmol / liter FlG. 2. Effects of bromide (A) and bromide plus chloride (B) to constant halide of 100 mmol. per 1. aqueous solution on: a, DuPont ACA; b, Stat/ Ion; c, SMA 6/60; d, IL Chloride Analyzer; e, Chloridometer. parent chloride concentration (Figs. 2A and 4). T h e ISE results followed a sigmoidal response curve, while the curves with other methods were linear or curvilinear in response to increasing bromide concentrations. (2) Solutions of bromide in the presence of chloride, with final halide totalling 100 mmol. per 1., when analyzed for chloride, showed similar responses (Figs. 1 and 2B). The electrode response was diminished in magnitude and appeared curvilinear rather than sigmoidal. Other methods showed response curves that were either parallel to, or identical to, the curves derived from pure bromide solutions (coulometry, Chloride Analyzer thiocyanate, ACA thiocyanate). (3) Results of chloride analysis of pooled 52 WENK ET BROMIDE CONCENTRATION - mmol / liter FlG. 3. Analysis of pooled sera containing various bromide concentrations with constant chloride, by various chloride methods, a, DuPont ACA; b, Stat/ Ion; c, 1L Chloride Analyzer; d, Chloridometer. serum in the presence of constant chloride of 97 mmol. per 1. containing bromide over the range of 1-20 mmol. per 1. are shown in Figure 3. The ISE method yielded, at low bromide concentrations, an a p p a r e n t ratio (chloride/bromide) greater than 4, which decreased to approximately 3 over the Br~ range of 10-20 mmol. per 1. For the DuPont ACA, this ratio was also high at low bromide concentrations; however, the ratio decreased to about half that for the ISE method at high B r - concentrations. Other methods also showed initially high ratios, which, however, were approximately 1 or slightly less over the same bromide concentration range. (4) The magnitude of inappropriate response of each chloride measurement to bromide interference is expressed more fully in Figures 2 and 3. In the absence of chloride (Fig. 2A) the Cl~ electrode method is less selective at low concentrations of bromide and more selective at high concentrations. The reverse is true for the ACA thiocyanate procedure. (Some of the curves are foreshortened at low levels since the methods are known not to be linear at these extreme, non- AL. A.J.C.P.—Vol. 65 physiologic concentrations. Manufacturer's recommendations were followed when possible.) Least interference among the methods was noted for the IL thiocyanate procedure, SMA 6/60 method, and coulometry. The coulometric method showed slightly more interference at the lowest bromide levels. Chloride analysis of solutions containing chloride as well as bromide showed similar, but less augmented, interference. Again, the electrode and ACA procedures showed greater responses to bromide than to chloride on a mole-for-mole basis. Other methods signalled one mmol of chloride per liter for each mmol of bromide present. (5) Used and new chloride electrodes did not differ greatly in their responses to pure bromide and bromide plus chloride with constant halide concentrations (Table 1). The greatest difference was found at low concentration of bromide in the absence of chloride. Exposure of electrodes to bromide for short intervals (minutes) caused a refractory or slow response in the one sample cup that followed the bromide-containing solution (Fig. 5). Repeated exposure of an electrode to bromide at intervals of about ten minutes (every six samples on the turntable) appeared to decrease the magnitude of erroneous response to subsequent bromide solutions; however, no predictable magnitude of response was apparent. Exposures of electrodes to bromide, iodide, or fluoride solutions for an hour did not appear to affect subsequent chloride analysis-response further. However, iodide (1 mmol. per 1.) showed interference: a three- to tenfold response as apparent chloride (mmol. per 1.) was signalled for the one mmol of iodide present per liter. Exposure of an electrode to I - for one hour slightly increased the electrode response to bromide, but had no longterm effect. Response to subsequent chloride solutions was appropriate. Fluoride caused no interference either with short or long exposures. January 1976 53 SERUM CI" METHODS, BROMIDE AND BROMISM Patient Studies (1) Of the 133 patients' sera, selected at random over several weeks, two showed the presence of bromide (both approximately 1 mmol. per 1.) by the gold chloride method. This method is considered relatively specific for serum bromide determination. 13 O u r data on constant total halide solutions with bromide in concentrations ranging from 2 to 100 mmol. per 1. showed average recoveries of 100.3%, range 93.8 to 105.0%. Precision and recovery data on serum pools are shown in Table 2. (2) Patients who had spurious chloride elevations in their sera are tabulated below along with historical information and serum B r - concentrations (Table 3). Some of these patients were receiving bromide by prescription; others had taken it in proprietary preparations; still others had no evidence of bromide administration. T h e bromide compounds determined by investigation are listed. (3) The three cases of bromism found in six months of study are summarized as follows: Case 1. R.B., a 78-year-old white woman, was admitted to Sinai Hospital with lethargy, disorientation to time, place, and person; and dehydration. Electrolyte studies using selective ion electrodes revealed , 160 - 5.0 > APPARENT . CHLORIDE / BROMIDE X 3.0 5 140 TO 2.0 1 • 120 SZ ^F ° CD 1.0 o 100 ' APPARENT CHLORIDE < 20 0 20 40 60 80 100 INDIVIDUAL HALIDE CONCENTRATION - mmol / liter FIG. 4. Electrode responses to aqueous solutions containing increasing chloride and bromide concentrations. an initial chloride value of 216 mmol. per 1. Chloride analysis was repeated colorimetrically (IL Chloride Analyzer) and coulometrically (Aminco Chloridometer), yielding values of 106 and 108 Table 1. Responses of Old and New CI Electrodes to Bromide Solutions and to Bromide plus Chloride Solutions (at Constant Halide of 100 mmol. per 1.) New Electrode Br" Concentration Apparent Cl-/Br- Apparent Cl"/Br (in Presence of CI") 10 20 30 40 50 60 70 80 90 100 5.6 5.0 3.7 2.9 2.4 2.1 1.9 1.7 1.6 1.6 2.9 2.5 2.2 2.0 1.9 1.7 1.6 1.5 1.5 1.4 Old Electrode _ CrVBr- Apparent C I ' / B r (in Presence of CI") 4.0 4.1 3.5 3.1 2.5 2.1 1.9 1.7 1.7 1.6 3.0 2.5 2.2 2.0 1.9 1.8 1.6 1.6 1.5 1.5 Apparent 54 WENK ET AL. 20Br UJ oo t— ELE o 210 200 UJ > •— 190 ELE (J 180 o 170 mol / litei (/z> 160 150 E (J z o o UJ o en 140 130 120 9X (_> 110 H- z UJ < Q_ < 100 Q_ 90 5 10 15 20 25 30 35 40 SAMPLE NO. AND SEQUENCE FlG. 5. Effect of increasing concentration of bromide on the subsequent ion-selective electrode response to halide. Test solutions contained increasing concentrations of bromide (mmol. per 1.— perpendicular lines) with chloride to a total halide concentration of 100 mmol. per 1. Wash solution contained chloride, 100 mmol. per 1. Numbers in parentheses represent apparent chloride/bromide. mmol. per 1., respectively. The patient's family was asked to bring all of the patient's medications to the hospital. Search revealed that the patient had chronically ingested a locally-prescribed sedative compound that contained both bromodiethylacetylurea and alpha-bromoisovalerylurea. Serum bromide initially was 10 mmol. per 1. (80 mg. per dl.). Treatment was carried out with intravenous administration of saline solution, diuretics, and withdrawal of bromides. Serum Br" decreased over three weeks' time. The patient was discharged in an alert state of mind. Br~ and CI - in serum were monitored. Values are summarized in Table 4. Case 2. C.S., a 54-year-old black man with a long history of alcoholism, was admitted to Crownsville Hospital Center in an obtunded state. He was weak, drowsy, A.J.C.P.—Vol. 65 confused, and showed marked ptyalism. No history was obtained on admission. Vital signs were normal. Serologic tests for syphilis (RPR, VDRL, and FTA) were positive. A tentative diagnosis of alcohol intoxication was made despite a nurse's note that there was absence of alcoholic breath odor. On the fourth day, the patient was still confused, was unable to swallow, was incontinent of urine and feces, and required restraint. On the fifth day, the pupils were very contracted, plantar reflexes had decreased, and there was a tendency for the head and eyes to turn right. Possible subdural hematoma was considered. At this time, marked elevation of serum chloride was noted (more than 130 mmol. per 1. by Technicon AutoAnalyzer, thiocyanate method). Bromism was suspected; serum bromide analysis revealed a concentration of 62.6 mmol. per 1. (500 mg./dl.). Chloride measurement of the serum, by selective ion electrode, was 209 mmol. per 1.; by Chloride Analyzer, 117 mmol per 1. Treatment with fluids, diuretics, and magnesium sulfate lowered serum bromide to 30.8 mmol. per 1. in two days, and to 17.8 mmol. per 1. in another four days. Serum chloride continued to be greater than 130 mmol. per 1. (AutoAnalyzer thiocyanate). By the eighteenth hospital day, the patient was alert. He admitted taking daily one large bottle of tablets containing triple bromide (NaBr, KBr, NH 4 Br) for two to three weeks before his hospitalization. Bromide concentration decreased to 9 mmol. per 1. by day 23 and to 6.3 mmol. per 1. by day 29. Chloride levels returned to normal. Case 3. H.C., a 94-year-old black woman, was admitted to Sinai Hospital with a five-day history of confusion and disorientation following a fall. The patient was poorly responsive except to loud verbal commands and painful stimuli. Examination showed slight dehydration, decreased breath sounds, and fever (temperature 38 C ) . The pulse rate was 140 per min. Initial laboratory findings included normal January 1976 55 SERUM CI" METHODS, BROMIDE AND BROM1SM plasma glucose and urea concentrations, hemoglobin 13 Gm. per dl., leukocytosis (13,000 per cu. mm.), and positive RPR and FTA tests. Electrolytes were: Na + = 145, K = 4.7, C\~ = 114 (mmol. per 1. by 1SE) and H C 0 3 _ = 27 mmol. per 1. The clinical staff suggested diagnoses that included pneumonia, dehydration, and possible subdural hematoma. The patient was given penicillin and cephalothin when sputum showed Klebsiella on culture, but Gram-positive diplococci on smear. The respiratory symptoms cleared along with the dehydration and fever, but the patient remained confused. Skull series and echoencephalograms disclosed no abnormality. Laboratory follow-up of the initial elevated chloride showed the following data for serum obtained two days later (mmol. per 1.): Chloridometer, 108; ISE,, 183; ISE 2 , 186; ACA, 128; IL, 108. Bromide was 338 mg. per dl. (42 mmol. per 1.). T h e repeat CI" on the initial specimen was again 114 mmol. per 1. The reason for the initially lower CI" serum value is obscure. Cl~ in another serum specimen one week later was as follows (mmol. per 1.): Chloridometer, 98; ISE 1; 131; ISE 2 , 148; ACA, 112; IL, —. Bromide was 23 mmol. per 1. Renewed efforts to obtain a history of Br" ingestion were rewarding. The patient was taking un- Table 2. Measurement of Bromide by Au + + + Method (Summary) Within run Precision (mmol. per 1.) Number/run Mean S.D. Range 18 0.85 0.075 0.70-1.0 Recovery (%) Level of 1.8-3.7 mmol. per 1. Serum 95 93-97 known quantities of a triple-bromide preparation (Nervine, Miles Laboratories), which she had taken for years for insomnia. Diuretics, oral fluids, and withdrawal of medication allowed the sensorium to clear. At discharge B r - was 20 mmol. per 1. (4) Results of one of the authors' ingesting 9 mmol. bromide on several methods are outlined in Table 5. Discussion Bromide is an interfering anion in all of the chloride methods evaluated. The coulometric procedure produced only a relatively small error: 1:1 on a molar basis or 100% interference. Unknown to previous investigators, the colorimetric methods which use thiocyanate reagent vary in their specificity for chloride. Some are relatively specific, yielding 1:1 bromide: chloride interference, while others yielded Table 3. Br in Nontoxic Patient Sera Detected by Chloride Electrode* Apparent ci- Concentration (mmol . per 1.) (mmol. per 1.) Patient Patient Patient Patient Patient Patient Patient Patient Patient Patient 1 2 3 4 5 6 7 8 9 10 12.8 5.5 21.0 9.0 4.0 2.3 1.5 2.0 1.7 13.3 BrMedication Multiple Br~t Multiple Br~t Triple Br"$ Triple Br~J Bromisovalum Triple Br"t Electrode ACA IL 211 138 306 117 134 134 113 126 116 130 130 126 138 — — — — — — — 118 * Found in routine serum electrolyte studies ol'genera) hospital patients as a result of spuriously high CI". Incidence = 0.75£ t NaBr 1,800 mg., KBr 1,800 mg., NH.Br 1,800 mg., ZnBr 2 36 mg. X NaBr 290 mg., KBr 290 mg., NH,Br 30 mg. — — 109 127 122 106 116 107 106 56 WENK ET A.J.C.P.—Vol. AL. 65 Table 4. Patient R.B., Serum Br and CI during Hospitalization Serum Chloride (mmol. per 1.) Hospital Day Serum Bromide (mmol. per 1.) 1 8 9 10 14 22 10.1 Coulometry HgTh locyanate Ion -selective Electrode — 108 113 106 111 216 162 7.5 6.4 5.9 4.8 — — — 103 103 107 101 104 105 140 138 129 much higher bromidexhloride responses. The extent of error may vary directly with the concentration of halides present in solution, and is now known to depend on Hg + + concentration. 8 The ISE also produced falsely elevated chloride values in the presence of bromide. Clinically, bromide in concentrations of as little as 1 mmol. per 1. may elevate chloride values enough to raise suspicion of interference. One cannot predict the actual concentrations of bromide because the curves of the response of the chloride electrode to bromide concentrations in the presence of various concentrations of chloride are complex. A specific bromide electrode method would be useful for confirming suspiciously high chloride values in clinical specimens which the selective (but not specific) chloride sensor detects. 5,11 A specific spot test for bromide is available,4 and quantitation is easily carried out. 6 ' 13 Older and brand-new chloride electrodes respond in similar ways. T h e chloride electrode (as a chloride sensor) appears to be little affected by shortor long-term exposure to bromide, iodide, or fluoride. T h e only exception to this generalization was that the one or two samples following the Br~-contaminated sample showed falsely decreased chloride concentrations (by 4 - 1 0 mmol. per 1.). Samples should be rerun when there is suspicion of Br~ interference in antecedent specimens. There is definite direct interference by iodide in a concentration of 1 mmol. per 1. Clinically, this unphysiologic concentration may be reached on occasion when patients have been given iodide preparations orally or topically. Therefore, Cl~ values will be in error when the serum also contains bromide and/or iodide. There is a small but important proportion of general hospital inpatients whose sera contain bromide.t About a third of these have recorded histories of bromide use for sedative, hypnotic or anticonvulsant purposes. The remainder have no t T h e frequency among psychiatric hospital patients has been much higher in the past, and it may still be so.10,14 Our own preliminary data show that 13% of patients admitted to Crownsville Hospital have serum Br~ concentrations of more than 2 mg. per dl. Table 5. Results of Ingestion of 9 mmol. Bromide Apparent CI (mmol. per 1.) Baseline ingestion One day after ingestion Coulometry ACA IL Stat/Ion True Bromide (mmol. per 1.) 102 103 103 112 101 103 104 112 0 1 January 1976 SERUM CI" METHODS, BROMIDE AND BROMISM initial historical evidence of bromide ingestion. Their identification depends on the chemical detection of bromide in serum. It was in the latter group that we discovered three cases of bromide intoxication in a 12-month interval. Two cases involved elderly women, one of whom had a serum bromide concentration barely in the so-called "toxic" range of more than 9 mmol. per 1. Nevertheless, clinical toxicity was marked: older patients with some organic nervous-system disease, renovascular disease, and ingesting other medications are more sensitive to bromide. 7 The second case demonstrates that extremely high levels that may be lethal in some persons are only toxic (or even nontoxic) in others. Furthermore, a patient may manifest a misleading past history or neurologic signs that may avert suspicion of bromism and may prompt inappropriate therapy such as craniotomy for suspected neoplasm or hematoma (Cases 2 and 3). 7 Elevated serum chloride may be the first or only clue leading to the diagnosis. (The definitive diagnosis, in fact, may be rendered in the laboratory as well.) Therefore, there is some justification for utilizing laboratory chloride methods that are sensitive to bromide interference. T r u e chloride concentration may be obtained from quantitating the Br~ concentration (e.g., colorimetrically) and subtracting the bromide (mmol. per 1.) from the total halide value by one of the several methods that yield a 1:1 response of bromidexhloride. Some physicians still prescribe bromide, and a number of commercial agents (by our count, at least 35 in the United States) in current use still contain bromide, despite the removal of bromide salts from some common antacid-analgesics. (Of interest, Bromo Seltzerf manufactured in the United States contains no Br~, but that manufactured t Warner-Lambert Co. 57 in Canada does.) One may therefore expect continued bromide interference in chloride methods and the appearance of occasional cases of bromism. Acknowledgment. Charles Byrd, B.S., and Barbara Hall, B.S., provided technical assistance. References 1. Baselt RC, Wright J A, Cravey RH: Therapeutic and toxic concentrations of more than 100 toxicologically significant drugs in blood, plasma, or serum: A tabulation. Clin Chem 21:44-62, 1975 2. Blume RS, MacLowry JD, Wolfe SM: Limitations of chloride determination in the diagnosis of bromism. N Engl J Med 279:593595, 1968 3. Carney MWP: Five cases of bromism. Lancet 2:523-24, 1971 4. Claiborne TS: Bromide intoxication. N Engl J Med 212:1214-1216, 1935 5. Degenhart HJ, Abelu G, Bevarrt B, et al: Estimation of Br" in plasma with a Br~selective electrode. Clin Chim Acta 38:217220, 1972 6. Goodwin IF: Colorimetric measurement of serum bromide with a bromate-rosaniline method. Clin Chem 17:544-547, 1971 7. Hanes FM, Yates A: An analysis of four hundred instances of chronic bromide intoxication, South M e d J 31:667-671, 1938 8. Levinson SS, Rieder SV: T h e effect of free mercuric ion on apparent bromide concentrations when measured by the standard AutoAnalyzer method for chloride. Clin Chim Acta 52:249-251, 1974 9. Lustgarten JA, Wenk RE, Byrd C, et al: Evaluation of an automated selective-ion electrolyte analyzer for measuring Na + , K+, and Cl~ in serum. Clin Chem 20:1217-1221, 1974 10. McDanal CE, Owens D, Bolman WM: Bromide abuse: A continuing problem. Am J Psychiatry 131:913-915, 1974 11. Poser S, Poser W, Muller-Oerlinghauseh B: Use of bromide electrodes for rapid screening of elevated bromide concentrations in biological fluids. Z Klin Chem Klin Biochem 12:350-352, 1974 12. Sharpless SK: Hypnotics and sedatives, T h e Pharmacological Basis of Therapeutics. Second edition. Edited by LS Goodman, A Gilman. New York, MacmiUan, 1958, pp 121-123 13. Sunderman FW, Wilkinson J H : Spectrophotometric measurement of bromide in serum and urine, Laboratory Diagnosis of Diseases Caused by Toxic Agents. Edited by FW Sunderman, FW Sunderman Jr. St. Louis, Warren H. Green Inc., 1970, pp 257-258 14. Wagner CP, Bunbury DE: Incidence of bromide intoxication among psychotic patients. JAMA 95:1725-1728, 1930