Field application of a colorimetric method of assaying chloroquine and desethylchloroquine in urine

Bulletin of the World Health Organization, 66 (4): 485-490 (1988) Field application of a colorimetric method of assaying chloroquine and desethylchloroquine in urine R. W. STEKETEE,1 D. L. MOUNT,2 L. C. PATCHEN,2 S. B. WILLIAMS,2 F. C. CHURCHILL,2 J. M. ROBERTS,3 D. C. 0. KASEJE,4 & A. D. BRANDLING-BENNETT5 In a study in western Kenya of malaria-infected adult women who had been treated with chloroquine, we compared the level of chloroquine and its principal metabolite, desethylchloroquine, in urine, measured using a newly developed modified Haskins test, with the level of chloroquine in whole blood, determined by high-performance liquid chromatography. Over a 28-day follow-up period, 277 matched urine and blood samples from 81 women were evaluated. A high correlation was observed between the level of chloroquine in whole blood (in itg/l) and that ofchloroquine +desethylchloroquine in urine(in mg/l). The test was easily performed and may be useful for monitoring use of chloroquine in a community and determining pre-study or post-treatment ingestion or absorption of the drug in in vivo studies ofparasite sensitivity. Information about the use and efficacy of anti- malarial drugs is important for planning and moni- toring malaria control programmes. Monitoring the use of chloroquine in a community is best carried out by quantitative determination of its level in a body fluid in conjunction with administration of a questionnaire about recent drug ingestion. Also, investigations of reports of chloroquine-resistant Plasmodiumfalciparum in new areas and assessment of changes in areas with known chloroquine resist- ance can be aided by measuring the levels of the drug in the body fluids of individuals or populations. Veri- fication of a reported case of chloroquine-resistant P. falciparum in a new area requires confirmation that a patient has taken and absorbed the drug. Furthermore, in vivo drug-sensitivity studies of P. falciparum in a community are best carried out on persons who have been confirmed not to have chloro- quine already in their system and can be shown to have absorbed the study dose of the drug. Chromatographic methods are available for the quantification of chloroquine and its major metab- olite, desethylchloroquine, in whole blood, blood fractions, and urine (1-6). However, the methods for 1 Epidemiologist, Malaria Branch, Division of Parasitic Diseases, Center for Infectious Diseases, Centers for Disease Control, Adanta, GA 30333, USA. Requests for reprints should be sent to this address. 2 Control Technology Branch, Centers for Disease Control, Atlanta, GA, USA. 3 Division of Parasitic Diseases, Centers for Disease Control, Atdanta, GA, USA. 4 Department of Community Medicine, University of Nairobi, Nairobi, Kenya. 5 Kenya Medical Research Institute, Nairobi, Kenya. determining the drug in plasma or whole blood require the use of sophisticated laboratory instru- ments, and this causes delays for workers in the field. Colorimetric methods for the field assay of chloro- quine+desethylchloroquine in urine have been available for many years and include the Dill-Glazko test (7), the Haskins test (8), and the Wilson-Edeson test (9). These tests are, however, less specific and sensitive than analysis of plasma or whole blood and show greater individual variation owing to differ- ences in the rate of drug absorption, tissue distri- bution, and renal excretion. Such tests are neverthe- less, simple to perform in the field, and, to date, have been used qualitatively to determine the presence or absence of chloroquine+ desethylchloroquine in a urine sample. Because of its simplicity, the Dill- Glazko test, in particular, has been used most frequently for this purpose, but it has low sensitivity and is unreliable (10-12). Recently developed colorimetric tests do, however, permit quantification of chloroquine+desethyl- chloroquine in urine in the field (13, 14). One such method developed in our laboratory using a modi- fication of the Haskins urine test (Haskins, MMII) can detect chloroquine+desethylchloroquine to a limit of 1 mg/l (14). We compare here the results obtained using this test for determining chloroquine+ desethylchloroquine levels in urine with those for chloroquine in whole blood determined by high- performance liquid chromatography (HPLC) on samples obtained from malaria-infected adult women (pregnant and non-pregnant) from western Kenya who had been treated with the drug. Possible appli- cations of this test are also discussed. 496 -485- R. W. STEKETEE ET AL. MATERIALS AND METHODS Patient selection The selection of patients for the study is described elsewhere (15). Consecutive pregnant women who attended a district hospital antenatal clinic with Plasmodium falciparum parasitaemia were enrolled and placed into one of two treatment groups: those receiving 25 mg chloroquine per kg body weight over 3 days (10 mg, 10 mg and 5 mg) or those receiving 5 mg chloroquine per kg body weight each week for 4 weeks. All females aged 14-18 years (non- pregnant, as confirmed by urine pregnancy test) attending one large school were screened and enrolled in the study if they exhibited P. falciparum para- sitaemia; they were assigned to one of the two treatment groups for pregnant women described above. Collection ofspecimens Urine and filter-paper-absorbed fmgerstick blood specimens were collected from study participants on enrolment (day 0) and also on days 2, 7, 14, 21, and 28. Processing of specimens Filter-paper blood specimens were air dried, indi- vidually wrapped, and stored in desiccant until processed. The levels of chloroquine and desethyl- chloroquine in whole blood were determined using HPLC (6). Urine specimens were examined at the study site within 6 hours of collection and their specific gravity was determined with a urinometer. A modification of the Haskins test based on color- imetric measurement of ion-pair formation between chloroquine and methyl orange in chloroform (Has- kins, MMII) (14) was used to determine the level of chloroquine+ desethylchloroquine in urine samples. The percentage transmittance of the chloroform layer was read using a hand-held, battery-operated, filter photometer (filter wavelength, 420 nm). For samples with transmittance <0.02 (absorbance > 1.70), specimens were diluted twofold sequentially until a transmittance of 0.10-0.80 was obtained. Absorb- ances were recorded and compared with a prepared standard curve to convert them into concentrations (in mg/i). The standard curve is given by: Chloroquine concentration (mg/i)= -0.348 +8.64 xabsorbance; R2=0.99. The concentrations obtained were corrected for specific gravity (SG) using a standard curve deter- mined in our laboratory to give an adjusted concen- tration in mg/l. The following adjusted-SG standard curve was used: Chloroquine concentration (adjusted-mg/i)= -89.34+ 89.49 x SG; R2=0.99. Measurement units In the test described, the combined level of chloro- quine and desethylchloroquine in urine is expressed in mg/l or adjusted mg/I. Whole-blood determi- nations distinguished between the levels of chloro- quine and desethylchloroquine and, here, we report the measured level of chloroquine in yg/l, where for chloroquine, 1 1g/1=0.003 irmol/l. Data analysis For persons with < 100 tg/l of chloroquine in whole blood on day 0, the correlation between the level of chloroquine in urine and whole blood was analysed in corresponding sample pairs by day of collection. Urine specimens whose volume was insufficient for the required dilutions or specific- gravity measurements were not included in the analy- sis. The levels of chloroquine in urine and whole blood are plotted in Fig. 1 for days 0, 2, 7, 14, 21, and 28 and fitted with a regression line using a least-squares method. Intervals, which included 80% of observed whole-blood chloroquine levels (80%- prediction intervals), were calculated from the regression line for urine values (16). 1300 - 1200 - 1100 - o 1000- * 900- g 800- c 700- O 600- 6 500- m 400- 300 - 200- 1001 10 20 30 40 Urine chloroquine (adjusted-mg/1) 50 60 Fig. 1. Comparison of the level of chloroquine in whole blood and the specific-gravity-adjusted level of chloro- quine + desethylchloroquine in urine from malaria- infected Kenyan women on days 0, 2, and 7 after weekly (5 mg chloroquine per kg body weight) or therapeutic (25 mg chloroquine per kg body weight) doses. Bars represent the 80%-prediction interval. +/ + + / + +++ R n = 2 = + + t + +Regression Iie: Y =48.6 +25.6X (R=0.91) 14U- U sr _ 486 ASSAY OF CHLOROQUINE AND DESETHYLCHLOROQUINE IN URINE Table 1. Mean level of chloroquine in whole blood and urine, by days after treatment for 81 women who received 25 mg of the drug per kg body weight (10 mg, 10 mg, and 5 mg on days 0, 1, and 2, respectively) or 5 mg per kg weekly Day after enrolment Day 0 Day 2 Day 7 Day 14 Day 21 Day 28 25 mg/kg dose Level in whole blood (jtg/l) 26 (31)° 626 (215) 262 (97) 96 (58) 50 (22) 49 (27) Urineb Absorbance 0.01 (0.02) NAc 0.8 (0.4) 0.4 (0.3) 0.1 (0.1) 0.1 (0.3) Level (mg/I) 0.03 (0.10) 34 (23) 9.2 (7.7) 3.3 (2.1) 0.6 (0.8) 0.8 (2.9) Adjusted level (mg/I) 0.02 (0.06) 27 (17) 8.0 (6.0) 3.2 (2.1) 0.6 (0.7) 0.5 (1.8) 5 mg/kg dose Level in whole blood (jsg/l) 30 (21) 112 (55) 65 (25) 92 (41) 98 (40) 112 (91) Urineb Absorbance 0.01 (0.02) 0.3 (0.25) 0.2 (0.23) 0.4 (0.30) 0.2 (0.27) 0.3 (0.23) Level (mg/I) 0.01 (0.05) 4.5 (6.5) 1.4 (1.9) 3.2 (5.3) 2.6 (8.7) 1.8 (2.0) Adjusted level (mg/I) 0.01 (0.02) 2.8 (3.8) 1.1 (1.3) 2.9 (3.6) 2.0 (5.3) 1.5 (1.7) Figures in parentheses are standard deviations. b Results are for chloroquine + desethylchloroquine. Absorbances were measured at X - 420 nm using a filter photometer. The level was determined from a standard curve, and the adjusted level obtained from this by taking also specific gravity measurements into consideration. ' NA - not applicable because specimens were diluted. RESULTS The chloroquine levels in matched samples of urine and blood were obtained for 81 women and 277 matched samples for the 28-day follow-up period. Mean levels of chloroquine in whole blood and of chloroquine+desethylchloroquine in urine for each day of the follow-up and for the two different drug doses are shown in Table 1. A high degree of cor- relation was observed between a plot of the level of chloroquine in blood (ig/l) and that of chloroquine+ desethylchloroquine in urine (adjusted-mg/l) for the 277 samples (Fig. 1) (R=0.91, P<0.0001). The least-squares regression line is given by: Blood-chloroquine level (jug/l) =48.6+25.6 (urine chloroquine+desethylchloroquine [adjusted-mg/l]). The correlation between blood chloroquine levels and urine chloroquine+desethylchloroquine levels without adjusting for specific gravity was only slightly lower (R=0.88, P<0.0001). The 80%-prediction intervals for whole blood chloroquine estimates based on the least-squares regression line are given for specific values of the adjusted level of chloroquine+desethylchloroquine in urine in Table 2. From these prediction intervals, an estimated range for the level of chloroquine in whole blood (jug/l) can be determined for an indi- vidual urine chloroquine+desethylchloroquine (ad- justed-mg/l) value. The plot of the mean level of Table 2. Level of chloroquine in whole blood and esti- .mated number of days since last ingestion of the drug predicted from the adjusted level of chloroquine (CQ) + desethylchloroquine (DES) in urine using the modified Haskins procedure No. of days after taking CQ Adjusted level Associated mean in a dose of: of CQ + DES level of CQ in in urine (mg/l)a whole blood (Jg/1): 5 mg/kg 25 mg/kg 1 74 (0-163)b >7 >12 2 100 (11-189) >7 11-16 3 125(36-214) 5-8 10-14 4 151 (62-240) 4-7 9-13 5 177 (87-266) 4-6 8-12 6 202 (113-291) 3-5 7-11 7 228 (139-317) 2-4 6-10 8 253 (164-343) 2-4 5-10 9 279 (190-368) 1-3 4-8 10 305 (215-394) 1-2 3-7 12 356 (266-445) 1 2-5 15 432 (343-522) 1 2-4 17 484 (394-574) - 1-3 20 561 (470-651) - 1-3 Level of chloroquine + desethylchloroquine in mg/l determined by modified Haskins procedure and adjusted for specific gravity. ' Figures in parentheses are the 80%-prediction intervals for an individual's whole-blood chloroquine level for a given level of the drug in urine. 487 R. W. STEKETEE ET AL. chloroquine in whole blood against number of days since last dose for therapeutic (25 mg chloroquine per kg) or prophylactic (5 mg chloroquine per kg per week) doses (Fig. 2) together with the 80%- prediction intervals from Table 2 can be used to determine that an individual's urine-chloroquine+ desethylchloroquine level is consistent with admin- istration of a known dose of the drug within a given time period. For example, a urine level of chloro- quine+desethylchloroquine (adjusted-mg/l) of7mg/l is consistent with a chloroquine level in whole blood of228 itg/l (80%-prediction interval= 139-317 jtg/l) and with the ingestion of 5 mg chloroquine per kg body weight within 2-4 days or of25 mg chloroquine per kg within 6-10 days previously. The sensitivity and predictive value for a cut-off of A 1 mg/l in urine for the adjusted level of chloro- quine+desethylchloroquine and of <100 jsg/l for chloroquine in whole blood were 84% and 89%, respectively. Since levels of chloroquine in whole blood that are < 100 ,g/l are sub-therapeutic, ad-justed concentrations of chloroquine+desethyl- chloroquine in urine that are A 1 mg/l could be used to determine that an individual has small amounts of chloroquine in his or her system and could be an eligible subject for an in vivo study to assess the therapeutic treatment of malaria infection. Determination of chloroquine+desethylchloro- quine in urine showed an intra-test error of 0.04 of transmittance units for 20 repeated photometric measurements. When the standard curve is corrected by this amount, the error in the predicted con- centration of chloroquine+desethylchloroquine is ± 0.30 mg/l. Comparable errors in the specific gravity measurements were not determined; how- ever, a specific gravity error of only 0.001 would introduce an error of ±0.35 mg/l to the adjusted level of chloroquine+desethylchloroquine in urine. DISCUSSION In the field study ofthe described modification ofthe Haskins test, we found a highly significant correlation between the level of chloroquine+desethylchloro- quine in urine and that of chloroquine in whole blood for 28 days after treatment with the drug. Correlations were slightly improved when the estimated urine level ofchloroquine+desethylchloroquine was adjusted for specific gravity, and the use ofthis adjustment may be important for quantitative measurements. The experimental error in comparisons of the urine chloroquine+desethylchloroquine and whole blood chloroquine tests is low. The method of determining chloroquine in whole blood has a very low intra- and inter-test error and a sensitivity of5 isg/l (6), and thus does not contribute significantly to such errors. Likewise, the intra-test error for the urine test and for the adjusted-mg /l values were acceptably low and well below the range of individual biological variation in chloroquine metabolism. The concentration in whole blood of chloroquine varies in individuals adminis- tered the same dose per kg body weight due to differ- ences in absorption and tissue distribution (17), and measurements of chloroquine in urine introduce ad- ditional biological variability due to renal excretion. Because of such biological variations and the time lag in renal excretion and urinary bladder collection and storage, simultaneous determinations of the concen- tration ofchloroquine in whole blood and urine can be expected to correlate imperfectly. None the less, the ability to rapidly estimate chloroquine+desethyl- chloroquine levels outweighs the disadvantages ofthe lower sensitivity and greater biological variation in the urine measurements made in the field. The Haskins-MMII test for determining chloro- quine+desethylchloroquine in urine was easily performed and readily taught to field technicians. On the average, 30-50 urine specimens could be collected and processed over a 2-hour period. Much of that time was spent in labelling specimens and recording data, operations which are part of any test procedure. The use of a hand-held, battery-operated filter photometer permitted precise quantification of the results; however, the test could also be performed quantitatively by visual comparison of photometer readings with those of known spiked standards. Use of a urinometer or a hand-held refractometer to measure urine specific gravity required an additional 20-30 minutes for 30-50 specimens. The modified Haskins test had a sensitivity of 1 mg/l, which permits estimations of chloroquine+ 14 Days after dosing Fig. 2. Level of chloroquine in whole blood after a therapeutic dose (25 mg chloroquine per kg body weight) or a weekly prophylactic dose (300 mg per week) for at least 10 weeks in adults (18). 488 ASSAY OF CHLOROQUINE AND DESETHYLCHLOROQUINE IN URINE desethylchloroquine in urine or whole blood follow- ing ingestion of the drug in the previous seven days. The specificity of the test was not evaluated during our field studies; however, other investigators (13) have reported that it can detect similar micromolar concentrations of desethylchloroquine, hydroxy- chloroquine, quinine, mefloquine, and proguanil. At much higher levels, pyrimethamine and amodiaquine cross-react under the test conditions; neither of these drugs was, however, available in our study. Other medications that are commonly found in rural areas of developing countries, including analgesics, anti- pyretics, and antibiotics, do not cross-react in the test (13). Determination of the concentration of chloro- quine+desethylchloroquine in urine can assist in assessing drug use practices in a population by establishing whether individuals are responding to febrile illnesses by taking therapeutic doses of chloroquine. Since doses of 25 mg chloroquine per kg body weight would produce blood levels of the drug > 200 Fg/l for approximately 10 days, measure- ment of the level in urine of chloroquine and its principal metabolite during this period would show positive results. Similarly, pregnant women could be assessed for compliance with a chloroquine chemo- prophylaxis programme. For a pregnant woman taking a weekly dose of 300 mg chloroquine base, whole-blood levels of the drug would be expected to exceed 150 itg/l within 5 days of the most recent dose. Such a level in blood would be expected to produce a urine concentration > 1 mg/l, and thus, urine measurements of the type described could verify recent ingestion ofthe drug. The test could also be used to determine the background rate of use of chloroquine in a population where the drug is widely available and taken for a variety of health problems. In in vivo drug sensitivity studies, the test could be used to screen potential study subjects to ensure that only persons who have not recently ingested chloro- quine, as indicated by a chloroquine+desethylchloro- quine level in urine of S 1 mg/l are included. The high negative predictive value of 89% found for the study population suggests that a cut-offof S 1 mg/l is useful in establishing that the whole-blood level ofthe drug is < 100 tg/l. Additionally, the test could be used to monitor enrolled subjects on day 2 or day 7 after ingesting chloroquine in order to assess proper drug absorption. Also, study subjects who failed to clear malaria parasites could be selected for measure- ment of blood chloroquine based on the results of the urine test. During surveillance for in vivo drug resistance of P. falciparum to chloroquine the test could be used to verify recent ingestion of the drug and to help in estimating the amount taken. From observations of whole-blood chloroquine levels after a standard therapeutic dose of 25 mg chloroquine per kg body weight or a weekly prophylaxis regimen of 300 mg chloroquine (Fig. 2), the whole-blood level ofchloro- quine estimated from the urine test result can be used to establish whether an individual has ingested a certain mg/kg-dose of chloroquine within a given period. For example, if a patient has a P. falciparum infection and has taken a therapeutic dose of chloro- quine five days previously, the expected level of the drug in urine could be compared with that measured using the test. If the two values are widely disparate, inadequate drug ingestion, poor drug absorption, or recent additional drug ingestion could explain the observed differences. Finally, the test could also be beneficial in hospitals and clinics in malarious areas. For example, in such areas, where patients often present with malaria and an uncertain history of prior medication, a reliable, quantitative test for urine chloroquine would permit clinical personnel to make more informed decisions about further treatment, allowing them to avoid overdosing patients or to choose alternative drugs, if appropriate. ACKNOWLEDGEMENTS We thank Professor M. Mugambi, Director, Kenya Medical Research Institute, for permission to publish our findings. We also appreciate the assistance of Pius Owuor, James Otieno, and Julius Ochieng in collecting the specimens. James Sande Odera and Tom Allport assisted in the laboratory testing of urine. RtSUME APPLICATION SUR LE TERRAIN D'UNE METHODE COLORIMETRIQUE DE DOSAGE DE LA CHLOROQUINE ET DE LA DESETHYLCHLOROQUINE DANS L'URINE Des methodes colorim6triques de dosage de la chloro- pour evaluer l'utilisation th6rapeutique et chimioprophy- quine et de ses m6tabolites dans l'urine seraient n6cessaires lactique de ce m6dicament; l'epreuve actuellement la plus 489 490 R. W. STEKETEE ET AL. utilis&e (Dill-Glazko) manque a la fois de sensibilite et de fiabilite. Une methode de terrain mise au point dans notre laboratoire d'apres une modification de 1'6preuve urinaire d'Haskins (Haskins-MMII) est capable de d6tecter la pr6sence de chloroquine et de desethylchloroquine a partir de 1 mg/ildans les urines. Lors d'une etude portant sur des femmes presentant une parasitemie palustre dans l'ouest du Kenya, nous avons compare les taux de chloroquine+des- 6thylchloroquine dans les urines, mesur6s par 1'6preuve d'Haskins-MMII et le taux de chloroquine dans le sang total, deternin6 par chromatographie liquide haute perfor- mance (HPLC). Pendant la periode de suivi de 28 jours, on a ainsi examine 277 echantillons apparies d'urine et de sang obtenus chez 81 femmes. On a observe une forte correlation (R=0,91; P<0,0001) entre le taux de chloroquine dans le sang total (en /tg/l) et le taux de chloroquine+d6sethyl- chloroquine dans l'urine (en mg/l apres correction de la densite). La droite de r6gression obtenue par la methode des moindres carres correspondait a la formule: chloroquine dans le sang (jig/i) =48,6+25,6 x (chloroquine+d6sthyl- chloroquine dans l'urine [valeur corrigee en mg/i]). En prenant comme seuil 1 mg/l pour la chloroquine+ d6sethylchloroquine dans l'urine et 100 ug/l pour la chloroquine dans le sang total, on a obtenu pour 1'6preuve une sensibilit6 de 84% et une valeur predictive positive de 89%. On peut savoir si le taux de chloroquine+des6thyl- chloroquine obtenu dans l'urine est compatible avec la prise d'une dose connue de medicament au cours d'une periode donnee d'apres le taux pr6vu de chloroquine dans le sang total d'un sujet un nombre donne de jours apres la prise d'une dose connue et d'apres les intervalles de valeur predictive 80% correspondant au taux de chloroquine dans le sang total obtenu a partir de l'equation de regression. Cette epreuve quantitative de dosage de la chloroquine dans les urines s'est r6velee facile a ex6cuter et peut etre utilis6e pour surveiller l'utilisation de la chloroquine dans une communaute et determiner quelle a ete la prise de chloro- quine avant l'etude ou apres le traitement lors d'etudes in vivo de la sensibilit6 des parasites. REFERENCES 1. BERGQVIST, Y. & FRISK-HoLMBERG, M. Sensitive method for the determination of chloroquine and its metabolite desethylchloroquine in human plasma and urine by high-performance liquid chromatography. Journal of chromatography, 221: 119-127 (1980). 2. BERGQVIST, Y. & DoMEu-NYBERG, B. Distribution of chloroquine and its metabolite desethylchloroquine in human blood cells and its application for the quanti- tative determination of these compounds in serum and plasma. Journal of chromatography, 272: 137-148 (1983). 3. ALVAN, G. ET AL. Determination of chloroquine and its desethyl metabolite in plasma, red blood cells and urine by liquid chromatography. Journal of chroma- tography, 229: 241-247 (1982). 4. CHURCHILL, F. C. ET AL. Determination of chloroquine and its major metabolite in blood using perfluoro- acylation followed by fused-silica capillary gas chromatography with nitrogen-sensitive detection. Journal of chromatography, 274: 111-120 (1983). 5. BERGQVIST, Y. & ECKERBOUR, S. An improved gas chromatography method for simultaneous determi- nation of chloroquine and two metabolites using capillary columns. Journal of chromatography, 306, 147-153 (1984). 6. PATCHEN, L. C. ET AL. Analysis of filter-paper- absorbed, fingerstick blood samples for chloroquine and its major metabolite using high-performance liquid chromatography with fluorescence detection. Journal of chromatography, 278, 81-89 (1983). 7. LELOJVELD, J. & KORTMAN, H. The eosin colour test of Dill and Glazko: simple field test to detect chloro- quine in urine. Bulletin of the World Health Organiz- ation, 42: 477-479 (1970). 8. HAsIcNs, W. T. A simple qualitative test for chloro- quine in urine. American journal of tropical medicine and hygiene, 7: 199-200 (1958). 9. WILSON, T. & EDESON, J. F. B. Studies on the chemo- therapy of malaria: HI. Treatment of acute malaria with chloroquine. Medicaljournal ofMalaysia, 9: 115-131 (1954). 10. VERDIER, F. ET AL. Unreliability of Diil-Glazko test in detecting chloroquine in urine. Lancet, 1: 1282-1283 (1985). 11. RoMBo, L ET AL. Reliability of Dill-Glazko test. Lancet, 1: 1509 (1985). 12. RoMBo, L. ET AL. Evaluation of three qualitative tests for the detection of chloroquine in urine: agreement with plasma concentrations determined with liquid chromatography. Annals of tropical medicine and parasitology, 80: 293-298 (1986). 13. BERGQVIST, Y. ET AL. Determination of chloroquine and its metabolites in urine: a field method based on ion-pair extraction. Bulletin of the World Health Organization, 63: 893-898 (1985). 14. MOUNT, D. L. ET AL. Colorimetric and thin-layer chromatographic methods for field assay ofchloroquine and its metabolites in urine. Bulletin of the World Health Organization, 65: 615-623 (1987). 15. STEKETEE, R. W. ET AL. In vivo response of Plas- modiumfalciparum to chloroquine in pregnant and non- pregnant women in Siaya District, Kenya. Bulletin of the World Health Organization, 65: 885-890 (1987). 16. NETER, J. & WASSERMAN, W. Applied linear statistical models. Homewood, IL, Richard D. Irwin Incorpor- ated, 1974, pp. 70-73. 17. WHITE, N. J. Clinical pharmacokinetics of anti- malarial drugs. Clinical pharmacokinetics, 10: 187-215 (1985). 18. STAIGER, M. A. ET AL. Sensitive high-performance chromatographic analysis for chloroquine in body fluids: application to studies of drug resistance in Plasmodium falciparum. Journal of chromatography, 325: 139-149 (1985).