Effects of the injectable contraceptive depot medroxyprogesterone acetate in Thai women with liver fluke infestation: results after six months

Effects of the injectable contraceptive depot medroxyprogesterone acetate in Thai women with liver fluke infestation: results after six months RICHARD A. GROSSMAN,1 VINICH ASSAWASENA,2 SOPON CHALPATI,3 & DILOK TAEWTONG ' The effect of the three-monthly injectable contraceptive depot medroxyprogesterone acetate (DMPA) on liver and lipid function was assessed in Thai women with liver fluke (Opisthorchis viverrini) infestation, DMPA administration being started in the immediate postpartum period. Immediate postpartum IUD and sterilization acceptors with fluke infestation were recruited as a comparison (control) group for the fluke-positive DMPA acceptors. Comparable groups of fluke-negative acceptors were recruited in an area of Thailand free of liver fluke transmission. Results are presented for the first 6 follow-up months for 170 DMPA and 177 controlfluke-positive subjects andfor 153 DMPA and 150 control fluke-negative subjects. Small and similar increases occurred in each of the four groups for alanine amino transferase, isocitrate dehydrogenase, and total bilirubin levels while aspartate amino transferase levels changed less in the DMPA groups than in their respective control groups. None ofthe subjects in either DMPA group had clearly abnormal results in these tests at 6 months. Alkaline phosphatase, cholesterol, and triglycerides levels were markedly lower in each group at 6 months than in the puerperal specimens. There was a greater decrease in triglycerides levels in both DMPA groups than in their respective control groups. However, the decrease in the alkaline phosphatase and cholesterol levels was greater only in the fluke-positive DMPA group than in the fluke-positive control group. None of these biochemical results were related to differences in age, parity, or lactation status between the groups. The results indicate that DMPA did not cause any early deleterious effects in the metabolic factors studied in women with liver fluke infestation. The three-monthly injectable contraceptive DMPA (depot medroxyprogesterone acetate) has been in use since 1965 in an area of northern Thailand where it has been found to be acceptable, safe, and effective (1, 2). Further indication of its potential popularity in Thai women was obtained in a recent study in a rural area of central Thailand when two-thirds of a group of 2000 new contracep- tive acceptors chose DMPA when offered the choice between DMPA, an oral contraceptive, or the IUD (Vichapan, N. & Gray R. H., unpublished results). 1 Medical Officer (Epidemiologist), WHO Research Team on the Clinical Evaluation of Fertility Regulating Agents, Chulalongkorn Hospital Medical School, Bangkok, Thailand. 'Deputy Under-Secretary of State, Ministry of Public Health, Bangkok, Thailand. 'Director, Maternal and Child Health Center, Khon Kaen, Thailand. ' Director, Maternal and Child Health Center, Rajburi, Thailand. In 1974, the Ministry of Health of Thailand ap- proved the use ofDMPA in its national programme and instituted a pilot programme to further evaluate the acceptability of DMPA throughout the country and to assess the possibility of its being administered by nonphysicians. Recent surveys have demonstrated that the liver fluke Opisthorchis viverrini is endemic throughout the large and populous north-east Thailand. Almost 90% of schoolchildren and young adults were found to shed fluke eggs in some areas, including Khon Kaen where this present study was undertaken. No longitudinal clinical studies have been reported while in one cross-sectional study involving large samples no biochemical or clinical differences were found between persons shedding or not shedding fluke eggs (3). However, cross-sectional clinical studies using currently stool-negative groups as controls for stool-positive groups are unsatisfactory in areas where the entire population has probably been 3581 -67- BULL. WORLD HEALTH ORGAN., Vol. 55 1977, RICHARD A. GROSSMAN ET AL. infected at least once and where no satisfactory scientific tool is available to establish the absence of current or past liver fluke infestation. Pathological studies performed in the endemic area have postu- lated that adult flukes living in the intrahepatic biliary tract for many years produce the inflamma- tory and fibrotic changes often observed in autopsy specimens (4). Despite the evidence that liver fluke infestation generally has an essentially benign clinical course in man, concern was expressed as to whether the administration of DMPA, a potent progestogenic agent, was safe for women with liver fluke infesta- tion and thus whether the Ministry of Health should allow the dissemination of DMPA in north-east Thailand. We believed that the risk to the study subjects would be minimal and our study hypothesis was that there would be no significant deterioration in liver function attributable to DMPA usage. This was based on the fact that (a) only a small percent- age of fluke-infested subjects have been found to have clearly abnormal liver function tests (3); (b) studies of DMPA have not shown any adverse effects on liver function (5); and (c) liver function did not change significantly in a small group of 40 fluke-infected Thai women taking a combined oral contraceptive for 12-15 months (6). To avoid the problem of including false-negatives in a control group we have only included women in the endemic area who were shedding liver fluke eggs in their stools. A comparison group of fluke-positive IUD and sterilization acceptors was also recruited. True negative controls for both these groups were obtained by concurrently recruiting DMPA and other contraceptive comparison groups in an area of Thailand known to be free of endemic liver fluke transmission. Because DMPA can be safely given post partum, a time when motivation for family planning is particularly high, we decided to perform the study with immediate postpartum acceptors. This provided the added advantage that all subjects enrolled into the study were in the same physiologi- cal postpartum state, which is also a time immedia- tely following a period of major stress and physio- logical and biochemical alteration in the female. This initial report gives the results up to 6 months post partum. A final report will be prepared after the subjects have been followed for 18 months. MATERIALS AND METHODS The study was undertaken in two large maternal and child health (MCH) centres. One centre is at Khon Kaen in the endemic liver fluke area in north- east Thailand (500 kilometres from Bangkok) and the other centre is at Rajburi 100 kilometres west of Bangkok, an area known to be free of liver fluke transmission. Both centres were performing 400-500 deliveries per month and had active immediate post- partum IUD and sterilization (tubal resection) pro- grammes. Before intake of subjects for the study began, DMPA was added as the third method available to prospective acceptors during the routine motivational discussions held at the antenatal clinic sessions and during confinement. Direct faecal smear stool examinations were per- formed during confinement for all prospective ac- ceptors. At the MCH centre in the endemic fluke area, where the prevalence was around 65 %, women whose stools were positive for liver fluke eggs were invited to join the study. As expected, all the stool examinations performed at the MCH centre in the non-endemic area were negative for fluke eggs. No special intake restrictions were imposed other than those normally used at the centres for women receiv- ing the chosen contraceptive method (DMPA, IUD, or sterilization). Informed verbal consent was ob- tained from all volunteers. Study intake occurred between January and July 1975. Over 80%Y of the DMPA and IUD acceptors at both centres started using the contraceptives within the first three days, and almost all within one week post partum. This was also the case for sterilization acceptors at Rajburi. At Khon Kaen, owing to the overcrowded schedule, about one-half of the tubal resections were performed on day 4 or later, 15% being performed in the second week post partum. Intake interviews, height and weight measure- ment, and the initial collection of 10 ml of venous blood were carried out just prior to receiving the IUD (Lippes loop) or the first 3-monthly 150-mg DMPA injection, or on the morning of, or the day after, the sterilization procedure was performed. (Test results were not significantly different between sterilization subjects who submitted their blood specimen before or after having undergone the pro- cedure). All women were told to return at any time if they had problems. DMPA users were seen at the clinic or in their homes at 3 and at 6 months for their second and third injections; IUD and sterilization subjects were seen at 6 months. Pertinent follow-up information was obtained from all subjects at each follow-up visit. The second sample of venous blood was collected at the 6-month follow-up visit. For the purposes of this paper, only continuing DMPA and 68 INJECTABLE CONTRACEPTIVES 69 IUD users as well as those sterilization subjects who were followed through 6 months (and having paired serum results) are reported. This includes 323 (75 %) of the 420 DMPA subjects and 327 (85 %) of the 375 IUD and sterilization subjects who entered the study. Some DMPA subjects decided to have steril- ization or to stop for personal reasons, while failure to locate subjects during follow-up accounted for the majority of DMPA subjects and almost all of the IUD and sterilization subjects without paired data. None of the DMPA subjects discontinued the drug because of medical reasons. The blood samples were handled in the same way at intake and follow-up. Microhaematocrit values were obtained immediately. Serum was separated following centrifugation and stored in the freezing compartment of a standard refrigerator at each MCH centre. The sera were then shipped to Bang- kok in dry ice every 2 weeks. At the chemistry laboratory in Bangkok the sera were thawed and divided into several fractions. One fraction was immediately used for transaminase determination. The other fractions were stored at -40°C and subsequently thawed to perform the other tests over the succeeding few days. The labels on the serum tubes did not include information on the contracep- tive group to which the subjects belonged. The following biochemical tests were performed on the sera: aspartate aminotransferase a-EC 2.6.1.1 (SGOT), alanine aminotransferase b'EC 2.6.1.2 (SGPT), isocitrate dehydrogenase-EC 1.1.1.42 (ICDH), alkaline phosphatase, total bili- rubin, cholesterol, and triglycerides. These tests were performed with reagent kits from Boehringer Mann- heim GMBH, with strict quality control procedures maintained with the appropriate control sera: Preci- norm E or Precilip (Boehringer Mannheim GMBH), and Seronorm (Nyegaard & Co., As.). SGOT, SGPT, ICDH and alkaline phosphatase determina- tions were performed (at 37°C) automated with the LKB 8600 Reaction Rate Analyzer. Bilirubin, cholesterol, and triglycerides determinations were performed manually, bilirubin with the Hitachi Perkin-Elmer 139 UV-Vis spectrophotometer and cholesterol and triglycerides with a Beckman spec- trophotometer. The upper limits of normal for adults given by the manufacturer for the kits at the temperatures we report our results are as follows: a Also known as glutamic oxaloacetic transaminase. b Also known as glutamic pyruvic transaminase. SGOT and SGPT (30°C): 21 IU/litre ICDH (25°C): 7.0 IU/litre alkaline phosphatase (37°C): 48 IU/litre total bilirubin (25°C): 17.1 ttmol/litre (1.0 mg/ 100 ml) cholesterol (25°C): 6.45 mmol/litre (250 mg/ 100 ml) triglycerides (25°C): 1.94 mmol/litre (172 mg/ 100 ml) In addition, total serum proteins and serum albu- min were manually determined by the Gornall- Bardawill-David method (7), with control serum from Hyland Laboratories, the readings being ob- tained with a Hitachi spectrophotometer. The data for the 6-month period have been as- sessed by comparing the experience of the DMPA group with that of its control (IUD plus sterili- zation) group for each study area. The occur- rence of differences between the DMPA users and their control groups that are similar in both study areas would imply that they are effects attrib- utable directly to DMPA use. Differences between the fluke-positive DMPA subjects and their control group that are significantly larger (or smaller) than those between the fluke-negative DMPA subjects and their control group would imply a differential effect (positive or negative) attributable to DMPA use in women with liver fluke infestation. The changes in the results of the chemical tests between intake and 6-months were first assessed statistically by the Student paired t-test, performed separately for each of the four DMPA and control groups. The DMPA and control groups were then compared after determining whether or not the paired results were affected by the following covari- ables whose distributions differed between the com- parison groups: age, parity; day after pregnancy termination when intake blood specimen was col- lected; type of pregnancy termination; and lactation status at 6 months. For most of the chemical tests, no large or consistent patterns were noted when mean changes were determined within the specific categories of these covariables. Direct adjustment of the data (using the combined DMPA and its control group as the standard population) separately for each of the covariables also showed that differences between the changes in the means of the DMPA and their control groups could not be accounted for by differences in the distributions of these covariables between the groups. For these test results, compari- son of the observed mean changes between the DMPA and their control groups was tested by the unpaired t-test. 70 RICHARD A. GROSSMAN ET AL. The interval between delivery and blood collec- tion, however, was an important modifying variable for the alkaline phosphatase, cholesterol, and tri- glycerides results. Because there were major differ- ences in the distribution of time interval between the groups, adjustment of the mean changes for the effect of this covariable was indicated. Analysis of covariance (8) was used for this purpose. After deter- mining the validity of applying this technique to the data, the derived adjusted mean changes between the DMPA and their control groups were compared statistically. RESULTS Some relevant background data on the 650 sub- jects having paired intake and 6-month data are shown in Table 1. The control groups are composed of both IUD and sterilization subjects, with about 80% being sterilization subjects. The control sub- jects were somewhat older and had about one more living child on the average than the DMPA subjects in both the endemic and non-endemic fluke study areas. A wider age and parity range was included for Table 1. Some intake characteristics of the samples of women followed through 6 months Fluke-positive a Fluke-negative a DMPA Control DMPA Control Samplesize 170 177 153 150 Age (years): range 17-45 17-43 18-35 19-34 mean 26.6 29.1 25.7 26.6 S.D. 7.54 5.00 4.19 3.83 No. living children: range 1-9 1-8 2-6 2-6 mean 2.8 3.9 2.4 3.3 S.D. 2.10 1.41 0.70 0.92 Urban residence (%) 17.0 4.0 12.4 7.3 History of abortion (%) 16.4 15.8 12.4 14.7 History of (%): malaria 3.0 1.1 12.5 5.3 surgery 1.1 3.4 2.0 5.3 transfusions 1.8 0.6 4.7 0.0 jaundice 0.6 0.6 0.7 0.7 a Fluke-positive refers to the subjects with definite liver fluke infestation recruited in the endemic fluke area. Fluke-negative refers to subjects recruited in the non-endemic area. intake at the MCH centre in the endemic fluke area. A larger proportion of rural than urban residents chose the IUD or sterilization rather than DMPA, probably largely owing to their greater difficulty in returning to the MCH centres for continuing care. The proportions of women reporting previous spon- taneous or induced abortions were similar between the groups. Less than 3% gave a history of more than one abortion and less than 2% gave a history of stillbirths. In the fluke-positive area, 15 DMPA and 9 control subjects were enrolled following an abor- tion. Two DMPA subjects, but no control subjects, in the fluke-negative area, were also enrolled after an abortion. Nine women enrolled after premature de- livery. The remaining 615 subjects were enrolled following normal full-term delivery. There was no notable difference between the four groups in the infants' average birthweight (about 3.0 kg) or the subjects' relevant past medical history. A larger proportion of the fluke-negative DMPA group had a past history of malaria (Table 1), but only a few women had had malarial symptoms within the previ- ous five years. Fig. 1-7 show the percentage distributions of the enzyme, bilirubin, cholesterol, and triglycerides values for each study group at intake and at 6 months. Sample sizes vary slightly because of missing 6-month values; post-abortion subjects were excluded from the alkaline phosphatase, cholesterol, and triglycerides analyses. The means and standard deviations were determined from the actual values while the points on each percentage distribution curve, except that for bilirubin, represent the mid- points of grouped data intervals. Although none of the curves shows a normal distribution pattern, the distributions of all intake minus 6-month values were approximately normal in shape, thereby estab- lishing the validity of the t-tests that were performed. Small, but statistically significant, increases occurred between intake and 6 months in most groups in the mean levels of SGOT, SGPT, ICDH, and bilirubin. There is a clear shift to the right in the distributions of SGPT (Fig. 1) and ICDH (Fig. 2) values but the increases were not significantly different between the DMPA and their control groups. This was also the case for the haematocrit, total protein, and albumin levels: in each of the four groups, haematocrit means increased significantly from about 36% at intake to 40%. at 6 months; mean total protein levels in- creased from about 6.8 g/100 ml at intake to 8.0 g/100 ml at 6 months; and mean albumin levels increased from about 3.8 g/100 ml at intake to 5.0 INJECTABLE CONTRACEPTIVES FLUKE-POSITIVE DMPA INT. SIX MOS. N 160 160 8.3 12.1 - s 4.45 5.61 I 91(.01 I I I I I F INTAKE l; f-sSIX|/ \ MONTHS LI9' .~~~~~~~~~~~~~~ FLUKE-POSITIVE CONTROL 1INT. SIX MOS. N 162 162 x 8.7 13.9S.D. 4.46 8.89 S. p(.Ol I I % I % i . I &-- r 50 40 30 20 10 0 50 40 30 20 FLUKE-NEGATIVE DMPA FLUKE-NEGATIVE :CONTROL 1SINT ISIX MOS.I s1 xISIXMOS. N 145 145 N 145 145 50 7.4 10.9 R 7.1 11.3S.D. 4.22 5.48 S.D. 5.54 6.23 40 P<.Ol p<.Ol 30 2011'\ 1\\I- 20 11, \1! ' 0 10 20 30 40 >40 0 10 20 30 40 40 S G P T (U/1) 10 0 FLUKE-POSITIVE DMPA INT. SIX MOS N 156 156 2.4 3.0 S D 1 98 0.88 p<.Ol .|\W INTAKE -S T I MONTHS FLUKE-NEGATIVE DMPA It8 INT . I SIX MOSIt N 144 144\ 2.1 2.7 S D. 1.55 1.03D\\\ ~~P<.01 j1 1' 0 1.0 3.0 5.0 7.0 37.0 ICDH FLUKE-POSITIVE: CONTROL N 167 167 . 2.4 3.6S.D. 1.82 2.15 I \ p<.Ol I11\ '% II" FLUKE -NEG,AT IVE:CONTROL 1. am SIX MOS.A/ 143 14 3 11 2. 3 II I 1.88 1.28 1.0 3.0 5.0 7.0 >7.0(U/ 1) Fig. 1. Percentage distributions of SGPT values at intake and at 6 months for each of the four study groups. N represents sample sizes having paired data; x represents sample means and S.D. sample standard deviations. The level of significance from t-tests performed on each group's intake minus six-month paired values is also shown; NS = P > 0.05. 50 FLUKE-POSITIVE:DMPA FLUKE-POSITIVE :CONTROL A, I.__NT. IXMOSS A II INT |IXMOS| 40 30' 20' 10 60 50 40 % 30 20 10 0 Fig. 2. Percentage distributions of ICDH values at intake and at 6 months for each of the four study groups. Further explanation same as for Fig. 1. N 163 163 N 165 165 FLUKE-POSITIVE:DMPA FLUKE-POSITIVE:CONTROL 17.1 17.6 15.0 20.4 .S.D 8.12 5.26 S.' D. 6.82 8.31 INTAKESN P<.01 40 .I1./LsIx MOS INT. SLIX MOS. N 165 165 N 169 169 INTAKE 0.31 0.42 0.36 0.37 11ix MONTHS 31/ S13D00.172 0.278 S.D.0 144 0.131 P6.01 INS /IMONTHS 10, FLUKE-NEGATIVE:DMPA FLUKE-NEGATIVE:CONTROL 1 w 1'4INT. SIX MOS. INT SIX MOS. 01N 148 148 N 143 1431 15.3 15.0 1X 15.0 16.3 40 FLUKE-NEGATIVE:DMPA FLUKE-NEGATIVE:CONTROL S.D. 5.64 3.86 SI |S. 5.40 5.82 4011 1 \\ NS1 1\p:wI.05 | INT. six MO. INT. SIX MOS. I I \ I \ 34 \IN 145 145 IN 145 14530' R 0.32 0.53 10.39 0.46S.D.O.194 0.254 S.D.O.159 0.167 20, '.<p.01 p(.0 10 I I 20~~~~~~~~~~~~~~~~~~~~)_ _-I ~~~~~~~~~~~~~~~~~~~0.2A .6 .8 O 1-L3 i1A O 2 6.8 L3O la-13 21A0 10 20 30 40 >40 0 10 20 30 40 >40 S G 0 T (U/1) BILIRUBIN (mg/lOOml) Fig. 3. Percentage distributions of SGOT values at intake and at 6 months for each of the four study groups. Further explanation same as for Fig. 1. Fig. 4. Percentage distributions of total bilirubin values at intake and at 6 months for each of the four study groups. Further explanation same as for Fig. 1. (1 mg/ 100 ml = 17.1 jmol/litre). 50 40 30 20 10 0. 71 RICHARD A. GROSSMAN ET AL. g/100 ml at 6 months. For SGOT (Fig. 3), there was a smaller change in the fluke-positive and fluke- negative DMPA groups than in their respective control groups although this was significantly differ- ent (P<0.01) only for the fluke-positive groups. Only for bilirubin (Fig. 4) did the DMPA groups show a greater increase than their control groups and these greater increases were significantly differ- ent (P<0.01) for both the fluke-positive and fluke- negative groups. The mean changes in SGOT, SGPT, ICDH, bilirubin, haematocrit, total proteins and albumin levels were not affected by differences in the distribution between the groups of the sub- jects' age or parity, the day after delivery when the intake blood was collected, the type of pregnancy termination, and whether or not the subject breast- fed her child during the 6 postpartum months. In addition to the changes in the means and the changes in the general shape of the distributions of test values, the upper tails of these distributions also contain obviously important information and were separately examined. The results in Table 2 show that, except for bilirubin, at 6 months in both DMPA groups the proportion with high (not neces- sarily abnormal) values was similar to or smaller than their respective control groups whether or not the values were also high at intake. Almost all of the subjects' SGOT values were higher than their SGPT values and a surprisingly high percentage of fluke- positive subjects had SGOT values above 21 IU/litre at 6 months. This suggests that a higher clinical cut- off point is indicated for our fluke-infested subjects from north-east Thailand. However, using a cut-off point of 30 IU/litre does not change the finding that these high SGOT values at 6 months were less frequent among fluke-positive DMPA subjects (5 %) than fluke-positive control subjects (12 %). Only one woman, a fluke-positive control subject, had transa- minase values higher than 60 IU/litre at 6 months. For bilirubin, only one fluke-positive and one fluke- negative DMPA subject had values over 23.94 ~umol/litre (1.4 mg/100 ml) at six months. None of the women developed jaundice or symptoms sug- gestive of hepatic impairment. At 6 months only one Table 2. Percentage of subjects in each study group having laboratory test values above manufacturer's upper normal limits at intake or at 6 months Test value at Fluke-positive Fluke-negative Intake 6 months DMPA Control DMPA Control SGOT Na Aa 12 27 7 11 A N 12 8 12 6 A A 6 5 1 1 SGPT N A 4 13 6 5 A N 1 13 1 1 A A 0 0 1 0 ICDH N A 0 3 3 1 A N 3 4 0 3 bilirubin N A 3 0 4 0 A N 0 0 1 0 alkaline phosphatase - b A 9 11 5 4 cholesterol - b A 4 4 3 5 triglycerides - b A 12 20 3 6 a N = within manufacturer's normal limits; A = above manufacturer's upper normal limit. b There are no recommended normal limits for immediate postpartum specimens for alkaline phosphatase, cholesterol, and triglycerides. 72 INJECTABLE CONTRACEPTIVES woman-had an albumin level less than 4.0 g/100 ml and no subject had an albumin/globulin ratio less than 1.0. Haematocrit values less than 35% at 6 months were found in less than 1 % of the fluke- positive groups, in 5% of the fluke-negative DMPA group, and in 8% of the fluke-negative control group. The values of alkaline phosphatase (Fig. 5), cholesterol (Fig. 6) and triglycerides (Fig. 7) at intake were generally much higher than the values at 6 months. The high levels at intake were not unex- pected since the levels of these three tests usually rise substantially during the third trimester of preg- nancy (9). Unlike the other tests, these test values were a least partially related to the number of days after delivery that the intake blood was collected. The mean values of alkaline phosphatase and tri- glycerides became progressively lower post partum, the mean values for women providing their initial specimen after postpartum day 6 being only modera- tely higher than the 6-month mean values. Mean cholesterol levels showed a much smaller decrease by lNT. SIX MOS. N 151 151 F~~UFE-~~~ ILAKE S.D.254 12:48 5 20 / 1NT.. siX MOS.I 160 160 ,A\ 0 55.9 36.94OFL 7 ;-F-Ff S ,' \S.D.18.48 12.19 C7-TRr / \L 5 20 _ 0 -- INT. SIX MOS. N 143 143 4 0- 62. 8 31. 2 1 FF1 //\ 0.1.22.56 ~~~~~~~~9.91 20 , 24. SIX MOS. N 142 142 2 67.8 32.5 4 0 FLUYE-NEG. S.D.22.70 9 . 2 5 CONTR>OLX / 9 20 / 10 20 30 40 50 60 70 80 >90 ALKALINE PHOSPHATASE (U/1) Fig. 5. Percentage distributions of alkaline phosphatase values at intake and at 6 months for each of the four study groups. Mean decreases from intake to 6 months were all highly significant (P < 0.001) by paired t-testing. postpartum day. The mean intake values for these three tests in post-abortion subjects were similar to the 6-month means for the control groups, implying that their pregnancies had terminated before the increases in these test values had occurred. They have therefore been excluded from analyses of these data. The range of intake values for these three tests was extremely large on each postpartum day. For the fluke-positive DMPA group, for example, values on postpartum days 0-1 ranged from 14 to 149 IU/litre for alkaline phosphatase, from 3.67 to 9.04 mmol/litre (142 to 350 mg/100 ml) for cholesterol, and from 0.81 to 5.42 mmol/litre (72 to 480 mg/100 ml) for triglycerides, and the values on postpartum days 4-7 ranged from 19 to 93 IU/litre for alkaline phosphatase, from 3.18 to 8.37 mmol/litre (123 to 324 mg/100 ml) for cholesterol, and from 0.59 to 3.48 mmol/litre (52 to 308 mg/100 ml) for triglycerides. The correlations be- tween these test values and the postpartum day when the intake blood was collected were therefore mostly weak and the regressions were nonlinear. 60 FLUKE-POS. INT. SIX MOS. DMPA ./SIX MONTHS N 149 149 40 -INTAKE X 229.3 163.3 %~~~~S.D. 57.97. 37.58. 60- FLUKE-PC CONTROI 40- 20- 60- FLUKEE- DMPA 40 - 20 0 DS. i\ INT. SIX MOS. L // \\\ N 163 163 I 'XX 2 247.4 185.8 !/ 'I. .I=_ -NEG. t' INT. SIO / x N 145 1 X258.7] S.D. 54.78 / *A/ / " %- .m o X MOS. 45 183.1 31. 50 60 FLUKE-NEG. ; INT. SIX MOS. CONTROL / N 146 146 40 / R 269.0 191.4 S.D. 53.27 34.01 20- / 0 , 0 50 100 150 200 250 300 350 400 2400 CHOLESTEROL (mg/1Oml) Fig. 6. Percentage distributions of cholesterol values at intake and at 6 months for each of the four study groups. Mean decreases from intake to 6 months were all highly significant (P < 0.001) by paired t-testing (200 mg/100 ml = 5.18 mmol/litre). 1. n 1- .;. . -~- -=i "<i _- 73 .x7Vfl , I -.1- I- RICHARD A. GROSSMAN ET AL. Because of this large variation in individual values, and the major differences between the groups in the distribution of postpartum days when the intake blood was collected, an appropriate adjust- ment of the data was indicated to be able to compare changes between the intake and 6-month values in the DMPA and their control groups. A valid com- parison was obtained by using the intake minus 6- month values for each subject's paired data and performing grouped regression analyses. This was done separately for each of the four study groups for the alkaline phosphatase, cholesterol, and trigly- cerides data, by computing the regression of the individual differences between intake and 6-month values (Y) on the respective intake values (X). Ten of the 12 estimated least-squares regression lines were found to be statistically linear, the other two show- ing only slight departures from linearity. All of the regression coefficients (b) were highly significant and FLUKE-POS. 60 DMPA FLUKE-POS. INT. SIX MOS. 40 - IX MONTHS N 146 146 / . INTAKE X 228.7 100.2 INAK S.D. 72.706 53.271 20' 0 FLUKE-POS. INT. SIX MOS. 4o - CONTROL -N 15 9 15 9 i_ X~~~~200.1 120. 4 8' ~~~~S. D. 7 2 .70 53 .2 7 20 / . _0_ -- - u - > 80 60 % 40 - 20 60 16 40 20 0 A FLUKE-NEG.I \ DMPA I'% II %INT. SIX M I % N 145 145 R 232.1 80 ,1 % IS.D. 73.46 48 1 I Ne--__~~~~~~~~~~~~~~~ ,A FLUKE-NEG. INT. I \ CONTROL N 143 I R 235. X S.D. 67. 11- I I ".II ~ ~ - Mos. .1 3. 73 SIX MOS. 143 .7 95.5 .73 43.70 0 50 100 150 200 250 300 350 400 )400 TRIGLYCERIDES (mg/lOml) Fig. 7. Percentage distributions of triglycerides values at intake and at 6 months for each of the four study groups. Mean decreases from intake to 6 months were all highly significant (P < 0.001) by paired t-testing. (200 mg/100 ml = 2.26 mmol/litre). the 12 correlation coefficients (r) were all between 0.7 and 0.9. By the analysis of covariance technique (including verifying homogeneity of variance with Bartlett's test), the DMPA and respective control sample regression lines were shown to be statistically paral- lel for the fluke-positive and fluke-negative groups for the alkaline phosphatase, cholesterol, and tri- glycerides data. Since parallelism was established, the mean DMPA and control differences could be validly adjusted and tested for significance. The results are summarized in Fig. 8. (The unadjusted mean differences shown in Fig. 8 differ slightly from what would be obtained by subtracting the respec- tive 6-month from the intake mean values shown in Fig. 5, 6, and 7 because the latter means were derived from the actual test values while the data in Fig. 8 were obtained from grouped data.) For both the alkaline phosphatase and cholesterol data, the 40 ALKALINE PHOSPHATASE 30 (U/i) 20 UNADJUSTED 1JUSTED 36.4 p<.00X NS 32.0 34.0 34..4 26.9 18.7 0- DMPA CON- DMPA CON- DMPA CON- DMPA CON- TROL TROL TROL TROL FLUKE-POS. FLUKE-NEG. IFLUKE-POS. FLUKE-NEG. CHOLESTEROL (mg/lOOml) TRIGLYCERIDES (mg/lOOml) 80 70 60 82.2 P4.001 NO 76.2779 71. 9 70. 3 DMPA CON- DMPA CON- DMPA CON- DMPA CON- TROL TROL TROL TROL FLUKE-POS. FLUKE-NEG. FLUKE-POS. FLUKE-NEG. P<.001 P<.001 L55.0 156 7 150 140.4 138.6 129.1 120 115.3 DMPA CON- DMPA CON- DMPA CON- DMPA CON- TROL TROL TROL TROL FLUKE-POS. FLUKE-NEG . FLUKE-POS. FLUKE-NEG. Fig. 8. Unadjusted and adjusted intake minus 6- month mean differences for alkaline phosphatase, cholesterol and triglycerides data in the four study groups. Data adjustment and significance testing on the adjusted mean differences between the DMPA and their respective control groups was accomplished by analysis of covariance technique (see text); NS = P > 0.05. 74 INTlAKE: MINUS SIX-MONT MEtlAN DIFFErrENCESLbL I _, . : . IFII I I 1 r mr a wy I INJECTABLE CONTRACEPTIVES fluke-positive DMPA sample had a significantly larger adjusted average decrease in values over the 6- month period than did its respective control sample while the adjusted average decreases were virtually the same between the fluke-negative DMPA and control samples. For the triglycerides data, both DMPA groups had larger decreases in adjusted average differences than their respective control groups. Although the mean differences for the fluke- positive groups are smaller than those for the fluke- negative groups for each test, this is largely due to the fact that a larger proportion of fluke-positive than fluke-negative intake specimens were obtained after postpartum day 3 while control group means at 6 months are similar for alkaline phosphatase (Fig. 5) and cholesterol (Fig. 6). Although the numbers of samples were smaller, repeat analyses of intake specimens collected within the first three postpartum days showed the same patterns of differ- ence between the fluke-positive DMPA group and its control group and also showed mean differences for alkaline phosphatase and cholesterol that were about as large as for the fluke-negative groups. Only for the triglycerides data did the mean differences remain moderately smaller for the fluke-positive than for the fluke-negative groups. The data were further adjusted by including age, parity, and lacta- tion as covariables in multiple regression and covariance analyses. The results showed that these other independent variables were not important, none of their contributions appreciably changing the results shown in Fig. 8. The intake sera were also tested for the presence of hepatitis B surface antigen (HBsAg) by the immu- noelectroosmophoresis (IEOP) method. The preva- lence of HBsAg was about 3% in the fluke-positive samples and 6% in the fluke-negative samples. Al- though the number of antigen carriers was small, their liver function test values at intake and the changes over the 6 months of follow-up were not notably different from those of the non-carriers. Except for menstrual changes, the proportion of women who complained of changes in health symp- toms at the follow-up interviews was small and similar in each group. None of the women became pregnant during the period. None of the DMPA subjects experienced a bleeding episode heavy enough to require estrogen treatment. About 90% of the subjects in each of the four groups breast-fed their infants throughout the 6 postpartum months. The proportion of lactating women with amenor- rhoea at 6 months was 63% for both control groups compared with 75% in the fluke-positive DMPA group and 80% in the fluke-negative DMPA group. This difference was primarily due to DMPA subjects who had resumed menstruation during the first 3 postpartum months and who again developed amenorrhoea thereafter. Of the women who were menstruating at 6 months a large proportion (about 60%) in each DMPA group, compared with only 15% in each control group, said that the amount of menstrual flow was less than usual. The number of menstrual bleeding days, the occurrence of spotting, and the occurrence of more discharge than usual was reported with similar frequency by the four groups of women. DISCUSSION To our knowledge, there has been no other report of a comparative longitudinal study of the effects of DMPA on the liver function and lipid factors that we considered in which DMPA administration was initiated within a few days of pregnancy termination. Likewise, no other data are available on the levels of these metabolic factors in Thai women during the early postpartum period or on changes in these levels thereafter. We therefore chose to include a much larger study population than is usually required for a metabolic study of contraceptive users in order to be able to assess the distributions of postpartum levels in Thai women and to ensure that adequate data were available to analyse the results at intake and follow-up by age, parity, and other potentially important intervening variables in women with and without liver fluke infestation. At intake, there was no significant difference between the four study groups with regard to their mean values of haematocrit, transaminases, ICDH, total bilirubin, total proteins, or albumin, while small but statistically significant increases occurred in each of these values (with one exception) in both control groups at 6 months post partum. The post- partum increase in haematocrit, total proteins, and albumin was to be expected. One cross-sectional study found SGOT, SGPT, and ICDH levels to be the same during the puerperium as during pregnancy (10). SGOT levels have also been found sometimes to be lower at delivery than late postpartum (11), possibly due to abnormal pyridoxine metabolism or to pyridoxine deficiency during pregnancy (9). How- ever, the possibility that SGPT, ICDH and total bilirubin levels may also be generally lower than usual shortly after delivery has not previously been reported. A small percentage of both DMPA groups had borderline abnormal bilirubin levels at 6 months 75 RICHARD A. GROSSMAN ET AL. (Table 2). Further follow-up should help to determ- ine whether this is a consistent drug effect. The fact that SGOT levels failed to increase significantly in the two DMPA groups (Fig. 3), despite the propor- tion of each DMPA group having relatively high SGOT levels at intake being larger than that in their respective control groups (Table 2), suggests the possibility of a drug effect independent of liver fluke infestation. The paired results for these tests do not suggest any differential deleterious effects up to 6 months in DMPA users having liver fluke infesta- tion, even after controlling the data for age, parity, lactation, and day after delivery when the intake blood was obtained. Numerous studies have shown that alkaline phos- phatase levels rise in the third trimester of preg- nancy, the rise being due mostly to the presence of the placental isoenzyme in the blood (9). The few studies that have followed the same women after delivery (subjects from the USA and Australia) have reported similar results showing that alkaline phos- phatase levels dropped rapidly within the first 7-10 days after delivery and reached baseline levels by 1 month post partum (12, 13, 14). Our data in Thai women agree with these findings. Our data also agree with other investigators in showing a wide range of alkaline phosphatase values on each testing occasion (12, 13, 15, 16). That the very wide ranges reflect specific individual patterns (possibly genetic) rather than temporal or environmental effects or laboratory test nonspecificity is suggested by the study of Beck et al. (13) who found remarkably similar patterns of alkaline phosphatase change in several women followed through two pregnancies. Some women have dramatic increases in alkaline phosphatase levels during pregnancy while other women show only minimal changes. The reason for the increases in cholesterol and triglycerides levels during late pregnancy is un- known (9) and there has not been any report to show whether the increases that occur during pregnancy are individual-specific as with alkaline phosphatase. It is also not known how soon triglycerides levels return to normal post partum but it has been reported that cholesterol levels remain elevated until several weeks after delivery (9) and one small longi- tudinal study in Seattle, USA, showed cholesterol levels significantly higher 1-6 days post partum and not different 6-7 weeks post partum from the sub- jects' levels in the second trimester of pregnancy (17). Our cholesterol data also show elevated intake levels and little change in levels during the first postpartum week. But, as with the alkaline phos- phatase data, our cholesterol and triglycerides data show marked individual variation. Although our sera were not necessarily collected from subjects after an overnight fast, almost all sera were collected at least 2 hours postprandially, the Thai diet has a relatively low fat content, and there is evidence that cholesterol levels do not vary appreciably even between fasting and 45-minute or 1-hour post- prandial specimens (18, 19). In addition, any effect related to the use of non-fasting values should be present in each group since identical procedures were employed. This would therefore not affect the validity of the DMPA versus control group com- parisons. The differences between the DMPA and control groups in the mean changes between intake and 6 months in the alkaline phosphatase, cholesterol, and triglycerides mean levels (Fig. 8) are therefore probably due to the combination of the large indi- vidual variations in amount of increase in these values during pregnancy and the number of days after delivery that the intake specimen was collected (which was also different between the four groups). Both sources of variation were effectively controlled by adjusting the mean change over the 6-month period in each group for each test by simple linear regression on the intake values and comparison of the adjusted mean differences between the DMPA group and its respective control group by analysis of covariance. This adjustment was particularly impor- tant for the cholesterol data, the adjusted mean differences of the fluke-positive groups being signifi- cantly different while the unadjusted mean differ- ences were almost the same; the reverse situation occurred in the fluke-negative groups. As with the other tests, the lipid and alkaline phosphatase changes were not related to differences in the distri- bution of age, parity, or lactation status between the groups. The results in Fig. 8 indicate that women with liver fluke infestation who started on DMPA imme- diately post partum had larger average decreases in alkaline phosphatase and cholesterol levels over the succeeding 6 months than did fluke-positive control women, while DMPA did not produce any differen- tial effect in alkaline phosphatase or cholesterol levels in fluke-negative subjects. The triglycerides results indicate that DMPA elicits a significantly greater decrease in levels over the first 6 postpartum months than those observed in the control groups regardless of liver fluke status. 76 INJECTABLE CONTRACEPTIVES The only clinical factor of note during the study so far was the expected finding (5) that the proportions of both groups of DMPA users with amenorrhoea at 6 months post partum were greater than those in the control groups. The high probability of amenor- rhoea was explained to the study participants at intake and this has not been a major complaint. Indeed, this is one of the advantages of starting DMPA post partum in societies such as this one where prolonged breast-feeding is the rule and exten- sion of lactational amenorrhoea may be less trouble- some. It is recognized that the study samples at each MCH centre were not randomly allocated to treat- ment groups or otherwise matched to ensure com- parability between the DMPA groups and their respective control groups. However, the numbers were generally large enough to compare results within specific strata (by age, parity, etc.) and, as mentioned, the results were not modified by any of these factors, except that the three tests correlated weakly with the day after delivery on which the intake blood was collected. Although the socioeco- nomic status of the fluke-negative groups and the fluke-positive groups were similar, the subjects being mostly rural Thai residents, and although identical study procedures were followed, the differ- ences between areas are likely to be greater than those between the DMPA and control groups within each area. The dietary habits are obviously different, this accounting for the presence of liver fluke trans- mission in one area but not in the other, although no comparative information is available on the energy contents or quality of the diets. However, the serum determinations (except the levels of alkaline phos- phatase and lipids) were all very similar at intake and similarly unaffected by the covariables consid- ered. In addition, the two control groups responded in a consistent fashion over the 6 postpartum months. In conclusion, the results through 6 months are reassuring. When started immediately post partum, DMPA appears to have had a modifying effect on SGOT levels and caused a larger decrease in trigly- cerides levels in both fluke-positive and fluke-nega- tive users. DMPA may have a singular effect in women with liver fluke infestation in causing a larger than usual decrease in alkaline phosphatase and cholesterol levels. It will be important to observe the course of these presumed effects beyond 6 months. Although statistically significant, it is doubtful whether these relatively small, presumably drug- induced, changes can be considered biologically important. The results up to 6 months certainly provide no statistical or clinical evidence of a delete- rious effect of DMPA on liver function in Thai women with or without liver fluke infestation. ACKNOWLEDGEMENTS This study could not have been achieved without the participation and excellent cooperation of the medical and nursing staff of the Khon Kaen and Rajburi MCH centres. We also acknowledge with much gratitude Mr Porn Israngkun and Mrs Sodsai Tovanabutra for the performance of the serum chemistry tests; Dr R. M. Scott of the SEATO Medical Research Laboratory for the HBsAg determinations; and Mrs Yupha Onthuam and Dr Termsri Chumnijarakij for data processing assistance. RItSUMIt ACETATE DE MEDROXYPROGESTERONE RETARD (DMPA) ET INFESTATION PAR OPISTHORCHIS VIVERRINI: RESULTATS AU BOUT DE SIX MOIS L'acetate de medroxyprogesterone retard (DMPA), contraceptif injectable a effet prolonge de trois mois, s'est revele tout a fait acceptable de meme que sans danger et efficace parmi des femmes thai des regions rurales du nord et du centre de la Thailande. La douve du foie Opisthorchis viverrini est endemique dans toutes les regions fortement peuplees du nord-est de la Thailande. Bien qu'il soit apparu que cette infestation a en general une evolution clinique benigne, une certaine inquietude s'etait fait jour, et l'on s'etait demande si l'administration de DMPA, progestagene puissant, etait sans danger pour les femmes infestees et si le Ministere de la Sante pouvait autoriser que le DMPA soit largement distribue dans le nord-est de la Thailande. La presente etude cherche a repondre a cette question. Elle porte sur quatre groupes de femmes ayant accepte une contraception immediatement apres le postpartum: deux groupes infestes de douves dont l'un recevait le DMPA et l'autre servait de temoin (DIU plus sterilisa- tion), et deux groupes non infestes de douves dont l'un recevait du DMPA et l'autre servait de temoin. Les deux groupes infestes ont et recrutes dans un centre de sante 77 78 RICHARD A. GROSSMAN ET AL. maternelle et infantile situ6 dans une zone oiu cette douve est endemique, et les deux groupes temoins non infestes dans un centre de sante maternelle et infantile situe dans une region du centre de la Thailande ou il n'y a pas de transmission de 0. viverrini. Les resultats presentes concement 170 femmes recevant du DMPA et 177 temoins, toutes infestees, ainsi que 153 femmes rece- vant du DMPA et 150 temoins, toutes non infestees; ces sujets etaient apparies au point de vue des resultats des epreuves cliniques et biochimiques au moment de leur inclusion dans l'enquete et au bout de six mois d'etude. Des augmentations faibles (mais statistiquement signi- ficatives) et similaires se sont produites dans les quatre groupes en ce qui concerne les taux de transaminase glutamopyruvique (SGPT), d'isocitrate-d6shydrogenase (ICDH), de bilirubine totale, de proteines totales, et d'al- bumine, alors que les taux de la transaminase glutamo- oxaloacetique (SGOT) ont moins change dans les groupes recevant le DMPA que dans les groupes temoins corres- pondants. Dans aucun des groupes DMPA, il n'y a eu de femme donnant des resultats nettement anormaux par ces epreuves au bout de six mois, et aucune n'a presente de signe clinique d'un trouble hepatique. Quant aux taux de phosphatase alkaline, de cholesterol et de triglycerides, ils 6taient, dans tous les groupes, nettement plus faibles au bout de six mois que dans les echantillons du post- partum. L'abaissement des triglycerides etait plus marque dans les deux groupes DMPA que dans leurs groupes temoins respectifs. En ce qui concerne la phosphatase alkaline et le cholesterol, cependant, la diminution etait plus marquee seulement dans le groupe DMPA infeste de douves par rapport au groupe infeste temoin. Aucun de ces resultats biochimiques n'etait en rapport avec des differences dans I'age, la parit6 ou l'etat de lactation entre les groupes. Le seul facteur clinique notable 6tait la constatation pr6vue que, dans les groupes d'utilisatrices de DMPA, la proportion d'amenorrh&e au bout de six mois apres l'accouchement etait plus importante que chez les t6moins. Nous pouvons conclure que, d'apres ces resultats, le DMPA n'a eu aucun effet nocif precoce sur les facteurs cliniques ou metaboliques etudies chez les femmes infestees de 0. viverrini. Un rapport final sera prepare apres dix-huit mois d'observation des sujets. REFERENCES 1. MCDANIEL, E. B. & PARDTHAISONG, T. Use of a long- acting injectable contraceptive in a postpartum family planning program. Asian J. Med., 9: 133-135 (1975). 2. PARDTHAISONG, T. ET AL. Acceptance and use of Depoprovera in Chieng Mai. IPPF med. Bull., 9: 1-3 (1975). 3. WYKOFF, D. E. ET AL. Clinical manifestations of Opisthorchis viverrini infections in Thailand. Am. J. trop. Med. Hyg., 15: 914-918 (1966). 4. EVANS, H. ET AL. Biliary tract changes in opis- thorchiasis. Am. J. trop. Med. Hyg., 20: 667-671 (1971). 5. RINEHART, W. & WINTER, J. Injectable progestogens: officials debate but use increases. Washington, DC, George Washington University Medical Center, March 1975 (Population reports, Series k, No. 1). 6. CHULACHARIT, E. ET AL. Oral contraception and liver fluke disease. J. Obstet. Gynaec. Br. Commonwlth., 79: 657-660 (1972). 7. MARK, D. D. & ZIMMER, A. Atlas of clinical laboratory procedures. Clin. Chem. 1: 132 (1967). 8. SNEDECOR, G. W. & COCHRAN, W. G. Statistical methods. Sixth edition. Ames, 10, Iowa State Univer- sity Press, 1967, Chapter 14. 9. DAVIDSOHN, I. & HENRY, J. B., ed. Todd & Sanford's Clinical diagnosis by laboratory methods. Fourteenth edition. Philadelphia, W. B. Saunders Co., 1969, pp. 565, 566, 719. 10. MEADE, B. W. & RoSALKI, S. B. Serum enzyme activity in normal pregnancy and the new born. J. Obstet. Gynaec. Br. Commonwlth, 70: 693-700 (1963). 11. STONE, M. L. ET AL. Glutamic oxaloacetic transami- nase and lactic dehydrogenase in pregnancy. Am. J. Obstet. Gynec., 80: 104-107 (1960). 12. SPEERT, H. ET AL. Serum phosphatase relations in mother and fetus. Am. J. Obstet. Gynec., 59: 148-154 (1950). 13. BECK, E. & CLARK, L. C. Plasma alkaline phospha- tase. II. Normative data for pregnancy. Am. J. Obstet. Gynec., 60: 731-740 (1950). 14. KITCHENER, P. N. ET AL. Alkaline phosphatase in maternal and fetal sera at term and during the puerperium. Am. J. clin. Path., 44: 654-661 (1965). 15. RHONE, D. P. ET AL. The isoenzymes of alkaline phosphatase in sera of normal pregnancy at term. Obstet. Gynec., 43: 31-40 (1974). 16. RoMsLo, I. ET AL. Serum alkaline phosphatase in preg- nancy. Acta obstet. gynec. scand., 54: 437-442 (1975). 17. DEALVAREZ, R. R. ET AL. Serial studies of serum lipids in normal human pregnancy. Am. J. Obstet. Gynec., 77: 743-759 (1959). 18. ANtNIo, J. S. & RELMAN, A. S. The effect of eating on some of the clinically important chemical constituents of the blood. Am. J. clin. Path., 31: 155- 159 (1959). 19. PERIC-GOLIA, L. Postprandial serum cholesterol, Lancet, 2: 876 (1972).