Effectiveness of intermittent preventive treatment with sulphadoxine-pyrimethamine for control of malaria in pregnancy in western Kenya: a hospital-based study

Summary

Objective  To monitor the effectiveness of intermittent preventive treatment (IPT) with sulphadoxine-pyrimethamine (SP) for the control of malaria in pregnancy at delivery in the Provincial Hospital in Kisumu, Kenya, and to assess the effect of IPT in participants in a cohort study.

Methods  Between June 1999 and June 2000, information on IPT and birth outcome was collected in 2302 consecutive deliveries. A group of 889 women, who were enrolled in a cohort to assess the interaction between malaria and HIV, were analysed separately because of the enrolment criteria and different access to health care.

Results  The prevalence of placental malaria was 13.8% and of low birthweight (LBW) was 12.2%. In multivariable analysis, IPT (≥1 dose of SP) was associated with a reduction in placental malaria and LBW [adjusted odds ratio (OR) 0.56, 95% confidence interval (CI) 0.39–0.83 and OR 0.65, 95% CI 0.45–0.95, respectively]. An adjusted mean increase in birthweight of 61 g was seen (95% CI 22–101 g) for each increment in number of SP doses (≥2 doses grouped together). IPT was associated with a reduction in placental malaria in HIV-seronegative women (OR 0.49, 95% CI 0.28–0.86) but this was not significant among HIV-seropositive women (OR 0.45, 95% CI 0.20–1.05). A significant effect on birthweight could not be detected among participants in the HIV-cohort.

Conclusions  This evaluation confirms that IPT with SP is effective in reducing placental malaria and LBW. It will be important to increase coverage of IPT and to extend IPT to antenatal clinics in peri-urban and rural areas.

Introduction

Malaria in pregnancy is associated with maternal anaemia, premature delivery and intrauterine growth retardation, resulting in delivery of a low birthweight (LBW) infant, the single greatest risk factor for infant mortality (Brabin 1983; Menendez 1995; Steketee et al. 1996). Control of malaria during pregnancy remains problematic. Weekly prophylaxis with chloroquine (CQ) proved unsuccessful because of logistical constraints, the spread of high-grade CQ-resistant malaria parasites, pruritis (itching) following CQ use, and sociobehavioural barriers (Kaseje et al. 1987).

Intermittent preventive treatment (IPT) with sulphadoxine-pyrimethamine (SP) in the second and third trimester gave striking results: in comparison with CQ prophylaxis, IPT reduced placental malaria by 72% (from a prevalence of 32% to 9%) in primi- and secundigravidae (Schultz et al. 1994). Similar studies in Kenya showed a beneficial effect of IPT on haemoglobin level in the pregnant woman (Shulman et al. 1999), and confirmed the decrease in placental malaria (Parise et al. 1998). SP, when taken under supervision in the antenatal clinic (ANC), was well accepted and tolerated; side-effects were mild and no severe adverse skin reactions were seen. The use of SP was not associated with perinatal complications. In areas of CQ-resistant malaria the two dose regimen with SP was shown to be a cost-effective intervention to reduce LBW incidence (Schultz et al. 1995). After Malawi, which introduced IPT with SP in 1993, the Kenya Ministry of Health introduced this policy in 1998, and since 2000 the World Health Organization has recommended this policy (WHO 2000).

A monitoring system was set up in the labour ward of the Nyanza Provincial General Hospital (NPGH) as part of a study of the interaction of malaria and HIV in pregnancy, to monitor birthweight and placental malaria. This allowed us to assess the effectiveness of IPT with SP on birthweight and placental malaria under programmatic conditions and taking practical issues, like drug delivery and compliance, into account. Here, we describe the effects on delivery outcome as seen in this hospital. Because women enrolled in the cohort study had a different degree of access to health care because of the nature of study, they were not included in the analysis of effectiveness under programmatic conditions. They were considered separately to assess the effect of SP in women with known HIV-status and their results are reported separately.

Patients and methods

Study site

This study was conducted at the NPGH in Kisumu, a town located on the shores of Lake Victoria in western Kenya with a population of approximately 300 000. Malaria transmission is perennial and Plasmodium falciparum accounts for 98% of the infections. CQ resistance is widespread; in the early 1990s 75% of P. falciparum infections were reported to have an RII/RIII resistance to CQ (Bloland et al. 1993). NPGH is a 400-bed government referral hospital, providing health care mostly to the local low-income population. Between June 1999 and June 2000, the period of evaluation, records on the number of ANC visits to the NPGH were available for 10 of the 12 months: on average 1441 women visited per month and 469 of them (32.5%) did so for the first time.

This study formed part of an evaluation of the implementation of IPT with SP in this hospital, details of which have been described elsewhere (A. M. van Eijk, J. G. Ayisi, F. O. ter Kuile et al. 2002, personal communication). Briefly, in cooperation with the Ministry of Health of Kenya, the Centers for Disease Control and Prevention, and the Kenya Medical Research Institute, the staff of the ANC received training on the delivery of IPT. Drugs were supplied by UNICEF and a supply for 1 year was available in the hospital. IPT with SP was implemented as part of the ANC routine of NPGH on 15 March 1999.

Procedures

From June 1999 to June 2000, information on birth outcome, including vital status, birthweight, gender and congenital abnormalities, was systematically collected for all women who delivered in the labour ward of the NPGH after informed consent was obtained. Birthweight was measured on an electric weighing scale to the nearest gram (Ohaus Florham Park, NJ, USA). Information was copied from the ANC card on number of ANC visits, date and gestational age, as assessed by palpation, at first ANC visit, date of first day of last menstruation (LMP) if known, and recordings of IPT, which were confirmed with the client. A placental smear was obtained.

Procedures at delivery were the same for women enrolled in the cohort study. However, only women with an uncomplicated pregnancy of a gestational age of ≥32 weeks, who were residing in Kisumu municipality, and who were willing to be tested for HIV, were enrolled in the ANC in the cohort study; health care was provided free-of-charge to these women and drugs, including iron and folate supplementation in pregnancy, were supplemented by the HIV-cohort study if they were not available in the ANC.

Laboratory procedures

Placental smears were stained with Giemsa and examined under oil immersion for malaria parasites. A thick smear was considered negative if 100 microscopic fields revealed no parasites. Malaria parasites were counted against 300 leucocytes. Parasite densities were expressed per microlitre using an assumed count of 8000 white blood cells per microlitre of blood. A second microscopist read 10% of the blood smears, and in case of discordant results a third reader was used as a tiebreaker. HIV testing involved the sequential use of two rapid test methods: Serostrip HIV-1/2 (Saliva Diagnostic Systems Pvt. Ltd, Singapore) for screening and Capillus HIV-1/HIV-2 (Cambridge Diagnostics, Wicklow, Ireland) for confirmation.

Definitions

Placental malaria was defined as the presence of asexual stage parasites in placental thick smears, independent of species. Young maternal age was defined as an age less than the median age of 21 years. Rainy season included the months April, May and June (long rains) and October and November (short rains). LBW was defined as a birthweight <2500 g. HIV seropositivity was defined as a reactive result on both rapid tests; women not reactive on Serostrip HIV-1/2 were considered HIV seronegative. In case of an inconclusive result, a Western blot test was carried out to establish HIV infection status (R. W. Steketee, J. G. Ayisi, B. L. Nahlen et al., personal communication).

Data analysis

Differences in mean values were compared using the Student's t-test. The Mann–Whitney U-test was used for non-parametric comparison of two groups. Differences in proportions were analysed using the chi-square test or Fisher's exact test when appropriate. Because primi- and secundigravidae are at higher risk for malaria and its consequences in an endemic area, and to be able to compare with a previous study (Parise et al. 1998), gravidity was divided into primi- or secundigravidae (G1/G2) vs. gravidae 3 or more (G3+).

Gravidity, ethnicity (Luo vs. non-Luo) and location of residence (urban vs. peri-urban or rural) were identified as potential confounders of the association between number of SP doses and placental malaria. Age and season of delivery were not associated with the number of SP doses. In analyses involving birthweight, only singleton live-born infants were included. The following factors were identified as potential confounders of the association between birthweight and number of SP doses: ethnicity, the use of haematinics in pregnancy, gravidity, and gender of the infant. Season of delivery, age and location of residence were not confounders of the association between IPT and birthweight. Logistic regression was used to examine the association between IPT and placental malaria, or IPT and LBW in multivariable analysis, and linear regression was used to assess the adjusted effect of SP doses on mean birthweight.

Because of the eligibility criteria and the different access to health care and medication for women participating in the cohort study, we considered that delivery outcome in these women may not reflect the ‘programmatic conditions’ of IPT. However, they allowed an opportunity to assess the effect of SP in women with known HIV status. For these reasons, results for women are presented, stratified by participation in the cohort study.

The statistical programs SPSS (SPSS for Windows 9.0, SPSS Inc., Chicago, IL, USA) and SAS (SAS system for Windows, version 6.12, SAS Inc., Cary, NC, USA) were used for analyses. For all statistical tests a two sided P-value <0.05 was considered significant.

The study protocol was approved by the institutional review boards of the Kenya Medical Research Institute, the Centers for Disease Control and Prevention, Atlanta, USA, and the Academic Medical Centre of the University of Amsterdam, the Netherlands.

Results

During the study period, information was collected for 2302 consecutive deliveries in the hospital. Of these, 1674 women (72.7%) had attended the ANC of NPGH, 573 women (24.9%) attended an ANC elsewhere, and 55 women (2.4%) had not attended an ANC. Fifty-six women delivered twins. The birth outcome among the 55 women who did not attend an ANC was poor, with 13 stillbirths (23.6%, all LBW), and 11 (26.2%) LBW infants among the 42 live-born infants. In this subgroup, 70.9% were primigravidae and 54.5% were younger than 20 years. Of the 47 women with a placental smear result available, 15 placentas of 41 live-born infants (36.6%) and three of the six stillbirths contained parasites. Because of the small number and because this group was distinctly different from the remainder of the study sample, it was excluded from the analysis. Of the 2247 deliveries whereby the mother had attended an ANC, 889 women participated in the cohort study; the remaining 1358 women were included in the analysis of the effectiveness of IPT with SP under programmatic conditions (Table 1).

Table 1.  Characteristics of women who delivered in the Provincial Hospital, June 1999–June 2000*

Maternal or infant characteristics	Women not in HIV-cohort (N = 1358) mean (SD) or n (%)	Women in HIV-cohort (N = 889) mean (SD) or n (%)
ANC, antenatal clinic; PH, Provincial Hospital; SP, sulphadoxine-pyrimethamine; CS, caesarean section; LBW, low birthweight; GMPD, geometric mean parasite density. * Women who never attended an ANC were excluded for this table. † There were missing values for number of ANC visits (240 women,10.4%), gestational age at first visit (302 women, 13.1%), total number of SP doses (90 women, 3.9%), haematinic supplementation (82 women, 3.6%), and placental smears (144 women, 6.3%). For 2114 of the 2119 singleton life-born infants a birthweight was available. ‡ P < 0.05 comparing women not participating in the cohort with participants of the cohort. § Assessment either by palpation or by first day of last menstrual period.
Mean age (SD) in years	22.5 (5.4)	21.8 (4.6)‡
Gravidity
Primigravidae	641 (47.3)	428 (48.1)
Secundigravidae	281 (20.7)	200 (22.5)
Gravidae ≥ 3	433 (32.0)	261 (29.4)
Luo ethnicity (vs. other tribes)	1086 (80.1)	725 (81.6)
Peri-urban and rural residence (vs. urban residence)	647 (47.7)	214 (24.1)‡
ANC card present	1132 (84.5)	866 (97.4)‡
Trimester of first visit†§
First trimester	29 (2.6)	9 (1.0)‡
Second trimester	489 (43.1)	348 (40.2)
Third trimester	616 (54.3)	509 (58.8)‡
Median number of ANC visits (range)	4 (1–13)	5 (1–15)‡
SP dose(s) received in the ANC†	498 (37.5)	605 (68.4)‡
Use of iron and/or folic†	252 (18.9)	740 (83.4)‡
Complaint of fever in pregnancy	415 (31.1)	233 (26.3)‡
Reported use of medication for fever	353 (26.4)	169 (19.1)‡
CS, vacuum or forceps delivery	114 (8.4)	58 (6.5)
Still birth (singletons only)	67 (5.1)	5 (0.6)
Mean birthweight in g, singleton life-born infants (SD)†	3048 (513)	3159 (429)‡
LBW, singleton and live-born infants†	151 (12.2)	48 (5.5)‡
Placental malaria†	177 (13.8)	85 (9.7)‡
GMPD (positive smears) parasites/μl (95% CI)	2791 (2058–3785)	3362 (2101–4559)

The use of bed nets was not systematically recorded in the present study, but was approximately 35% among women participating in the HIV-cohort during the study period. In addition, almost all of these bed nets were untreated, because insecticide was not routinely commercially available for this purpose during this period in western Kenya. IPT was not associated with bed net use among this group (data not shown).

For the analysis of birthweight, 56 twin deliveries, 72 stillbirths, three deliveries with an unknown outcome, and two infants with unknown birthweight were excluded.

Intermittent preventive treatment under programmatic conditions, placental malaria and birthweight

Among the 1358 ANC attendees, one dose of SP was received by 333 women (24.5%), two or more doses were received by 163 women (12.0%), three doses by two women, and 829 women (61.0%) did not receive IPT. For 31 women (2.3%) the number of SP doses was not known. The use of SP in pregnancy from a source other than the ANC was reported by 108 women (8.0%).

The use of IPT with SP was associated with a reduction in placental malaria (Table 2), and this effect was similar regardless of the number of doses of IPT with SP taken. When comparing one or more doses of SP to non-IPT, the adjusted odds ratio (OR) for placental malaria was 0.56 [95% confidence interval (CI) 0.39–0.83]. Although we observed a substantial reduction in the frequency of parasitaemia with IPT with SP, among women who remained parasitaemic no significant differences were observed in the geometric mean parasite density (GMPD) of infections: these were 5148 (95% CI 2032–14445)/μl among women who had received two or more doses, 2900 (95% CI 1201–6999)/μl among women who had received one dose of IPT, and 2581 (95% CI 1834–3633)/μl among women who had not received IPT. Among 1237 women (91.1%) who had the date of IPT documented, women who received one dose of IPT within the last 6 weeks before delivery had the lowest prevalence of placental malaria. Risks associated with the timing of the last SP dose varied somewhat by gravidity and HIV-status (Table 4), but none of the differences were statistically significant. LBW was associated with SP doses (Table 3), and a trend was seen towards decreasing prevalence of LBW with increasing number of doses (trend test in multivariable analysis: P = 0.02). The trend test in multivariable analysis among G1/G2 was borderline significant (P = 0.05), and not significant among G3+ (P = 0.11). When comparing one or more doses of SP to non-IPT among all women, the adjusted OR on LBW was 0.65 (95% CI 0.45–0.95).

Table 2.  The association between number of SP doses, placental malaria, and LBW, among ANC attendees not participating in the cohort, Kisumu, Kenya, June 1999–June 2000

	Placental malaria				Low birthweight
	N	Number with placental malaria (%)	Adjusted OR (95% CI)*	P-value	N	Number with LBW (%)	Adjusted OR (95% CI)*	P-value
OR, odds ratio; CI, confidence interval; SP, sulphadoxine-pyrimethamine; G1/G2, primi- and secundigravidae; G3+, gravidae 3 or more; HIV[+], HIV-seropositive women; HIV[−], HIV-seronegative women; LBW, low birthweight. Significant odds ratios are presented in bold. For 103 women no placental smear available or number of SP doses unknown or both. * Adjustment for ethnicity (Luo vs. other tribes), gravidity (where applicable), and place of living (urban vs. peri-urban/rural residence). For low birthweight also adjusted for gender of the child, and use of haematinics. Unadjusted odds ratios were very similar to the adjusted odds ratios and are not presented for this reason. † Trend test for all women: P = 0.015, odds ratio for each increase in number of doses: 0.72, 95% CI 0.55–0.94; trend test for G1/G2: P = 0.052, odds ratio 0.75, 95% CI 0.56–1.00.
All (SP doses)
0	784	132 (16.8)	1		745	101 (13.6)	1†
1	314	28 (8.9)	0.55 (0.35–0.86)	0.009	313	32 (10.2)	0.73 (0.48–1.12)	0.148
≥2	157	15 (9.6)	0.59 (0.33–1.05)	0.074	158	12 (7.6)	0.50 (0.27–0.94)	0.031
G1/G2 (SP doses)
0	530	112 (21.1)	1		513	78 (15.2)	1†
1	212	24 (11.3)	0.56 (0.35–0.91)	0.018	212	28 (13.2)	0.85 (0.53–1.36)	0.500
≥2	120	12 (10.0)	0.52 (0.27–0.99)	0.049	122	10 (8.2)	0.50 (0.25–0.99)	0.049
G3+ (SP doses)
0	254	20 (7.9)	1		232	23 (9.9)	1
1	102	4 (3.9)	0.49 (0.16–1.51)	0.215	101	4 (4.0)	0.35 (0.12–1.04)	0.058
≥2	37	3 (8.1)	1.11 (0.30–4.13)	0.880	36	2 (5.6)	0.59 (0.13–2.65)	0.491

Table 4.  Timing of last SP dose and placental malaria in women who delivered in the Provincial Hospital, Kisumu, June 1999–June 2000

Time between last SP dose and delivery	n	Placental malaria (%)	Odds ratio (95% CI)
SP, sulphadoxine-pyrimethamine; CI, confidence interval; G1/G2, primi- and secundigravidae; G3+, gravidae 3 or more; HIV[+], HIV-seropositive women; HIV[−], HIV-seronegative women. Significant odds ratios are presented in bold. * For 121 women not participating in the cohort (8.9%) and 29 women participating in the cohort (3.3%) the date of the last SP dose, number of SP doses, or a placental smear result was not known.
ANC-attendees not participating in the cohort*
All women
Received 0 SP	781	132 (16.9)	Reference
<6 weeks	139	9 (6.5)	0.34 (0.17–0.69)
6–11 weeks	217	24 (11.1)	0.61 (0.39–0.97)
≥12 weeks	100	10 (10.0)	0.55 (0.28–1.08)
G1/G2
Received 0 SP	529	112 (21.2)	Reference
<6 weeks	99	9 (9.1)	0.37 (0.18–0.76)
6–11 weeks	152	19 (12.5)	0.53 (0.32–0.90)
≥12 weeks	71	8 (11.3)	0.47 (0.22–1.02)
G3+
Received 0 SP	252	20 (7.9)	Reference
<6 weeks	40	0
6–11 weeks	65	5 (7.7)	0.97 (0.35–2.68)
≥12 weeks	29	2 (6.9)	0.86 (0.19–3.88)
ANC attendees who participated in the cohort*
HIV[+] women
Received 0 SP	72	14 (19.4)	Reference
<6 weeks	39	2 (5.1)	0.22 (0.05–1.04)
6–11 weeks	66	8 (12.1)	0.57 (0.22–1.47)
≥12 weeks	30	3 (10.0)	0.46 (0.12–1.74)
All HIV[−] women
Received 0 SP	205	26 (12.7)
<6 weeks	112	5 (4.5)	0.32 (0.12–0.86)
6–11 weeks	239	17 (7.1)	0.53 (0.28–1.00)
≥12 weeks	94	8 (8.5)	0.64 (0.28–1.47)
HIV[−] G1/G2
Received 0 SP	143	24 (16.8)	Reference
<6 weeks	82	5 (6.1)	0.32 (0.12–0.88)
6–11 weeks	174	15 (8.6)	0.47 (0.24–0.93)
≥12 weeks	67	5 (7.5)	0.40 (0.15–1.10)

Table 3.  The association between number of SP doses, placental malaria, and LBW, among ANC attendees participating in the cohort, Kisumu, Kenya, June 1999–June 2000

	Placental malaria				Low birthweight
	N	Number with placental malaria (%)	Adjusted OR (95% CI)*	P-value	N	Number with LBW (%)	Adjusted OR (95% CI)*	P-value
OR, odds ratio; CI, confidence interval; SP, sulphadoxine-pyrimethamine; G1/G2, primi- and secundigravidae; G3+, gravidae 3 or more; HIV[+], HIV-seropositive women; HIV[−], HIV-seronegative women; LBW, low birthweight. Significant odds ratios are presented in bold. For 18 women no placental smear available or number of SP doses unknown or both. * Adjustment for ethnicity (Luo vs. other tribes), gravidity (where applicable), and place of living (urban vs. peri-urban/rural residence). For low birthweight also adjustment for gender of the child, and use of haematinics. Unadjusted odds ratios were very similar to the adjusted odds ratios and are not presented for this reason.
HIV[+] women (SP doses)
0	72	14 (19.4)	Reference		70	2 (2.9)	1
1	84	6 (7.1)	0.33 (0.12–0.93)	0.035	86	5 (5.8)	2.69 (0.48–15.25)	0.263
≥2	55	7 (12.7)	0.67 (0.24–1.83)	0.433	55	4 (7.3)	2.68 (0.45–15.98)	0.278
All HIV[−] women (SP doses)
0	206	26 (12.6)	1		203	11 (5.4)	1
1	290	18 (6.2)	0.45 (0.24–0.86)	0.015	293	16 (5.5)	1.02 (0.46–2.27)	0.958
≥2	164	13 (7.9)	0.56 (0.27–1.14)	0.107	161	10 (6.2)	1.12 (0.46–2.74)	0.799
HIV[−] women G1/G2 (SP doses)
0	144	24 (16.7)	Reference		143	11 (7.7)	1
1	204	13 (6.4)	0.34 (0.17–0.70)	0.004	205	13 (6.3)	0.83 (0.36–1.92)	0.660
≥2	127	13 (10.2)	0.57 (0.27–1.18)	0.129	125	9 (7.2)	0.99 (0.39–2.49)	0.979

One dose of IPT was associated with an adjusted mean increase in birthweight of 54 g (95% CI 12–120 g, P = 0.11), and two or more doses with an adjusted mean increase of 128 g (95% CI 42–213 g, P = 0.004) compared with birthweights of infants of mothers who had not used IPT. The trend test was significant with an adjusted mean increase of 61 g (95% CI 22–101 g, P = 0.002) for each increment in number of SP doses (two or more doses grouped together). For G1/G2, the adjusted mean increases were 65 g (95% CI 15–144 g, P = 0.11) for one dose, and 158 g (95% CI 60–256 g, P = 0.002) for two or more doses of IPT, the adjusted mean increase per increment of SP dose was 76 g (95% CI 30–122 g, trend test P = 0.001). No significant difference was seen for G3+ women.

Women in the rural and peri-urban areas who had not received IPT were more likely to have placental malaria than residents of urban areas who had not received IPT (20.4%vs. 12.0% respectively, OR 1.88, 95% CI 1.26–2.81, P = 0.002); however, the proportion of LBW was similar among rural/peri-urban and urban women (12.1%vs. 15.4%, OR 0.76, 95% CI 0.50–1.15, P = 0.2). Women living in rural and peri-urban areas were less likely to receive IPT than urban women (22.8%vs. 50.6%, respectively, OR 0.29, 95% CI 0.23–0.37, P < 0.001) and were more likely to use SP from a source other than the ANC (10.6%vs. 5.8%, respectively, OR 1.91, 95% CI 1.28–2.86, P = 0.002). Analyses repeated taking these extra doses into account did not change the results.

Intermittent preventive treatment, placental malaria and birthweight among women participating in the cohort study

Among 889 delivering women participating in the cohort study, 217 women were HIV-seropositive (24.4%). Women enrolled in this cohort differed from the other women who delivered in the provincial hospital (Table 1); they were younger, were less likely to come from a rural area, presented more frequently with an ANC card, visited the ANC more frequent, and were more likely to have received IPT, and iron and/or folic acid. The mean birthweight was significantly higher compared with the mean birthweight of babies of women who were not enrolled.

Among women in the HIV-cohort, 384 women (43.2%) had received one dose, 217 women (24.4%) had received two doses, four women had received three doses, and 280 women had not received SP (31.5%). For only four women (0.4%) the number of doses was unknown. Thirty women (3.4%) reported the use of SP in pregnancy from a source other than the ANC.

When comparing one or more doses of SP to non-IPT, the adjusted OR for placental malaria was 0.49 (95% CI 0.28–0.86) for HIV-seronegative women and 0.45 (95% CI 0.20–1.05) for HIV-seropositive women. Differences in GMPD among parasitaemic HIV-seropositive and HIV-seronegative women by number of SP-doses were not significant (data not shown). No association was seen between LBW (Table 3) or mean birthweight (Figure 1) and number of SP doses. Among HIV-seronegative multigravidae (gravidae 3+), seven women had placental malaria and four women had LBW; differences in birthweight and placental malaria by number of SP dose or timing of the last SP dose were not significant. Women who delivered within 6 weeks of their last SP dose had a lower prevalence of malaria compared with women who had received SP 6 weeks or more before delivery, but this difference was not significant (Table 4).

Figure 1. Mean birthweights and 95% confidence intervals by number of SP doses for women who delivered in the Provincial Hospital in Kisumu, Western Kenya, June 1999–June 2000. SP, sulphadoxine-pyrimethamine; G1/G2, primi- and secundigravidae; G3+, gravidae 3 or more; HIV[+], HIV-seropositive women; HIV[−], HIV-seronegative women. * P < 0.05, 0 doses of SP as reference.

Discussion

In 1998, the Kenya government adopted IPT with SP as the national strategy for malaria prevention in pregnancy following results of two randomized controlled trials showing the beneficial effect of IPT on placental malaria (Parise et al. 1998) and maternal anaemia (Shulman et al. 1999). We evaluated the operational effectiveness of IPT in women who delivered in the NPGH in Kisumu. The prevalence of placental malaria was reduced in women who had used IPT and the mean birthweight was increased in a dose-dependent manner.

Placental malaria can be used as a proxy-indicator for malaria control during pregnancy, but prevalence will depend on the efficacy of the antimalarial drug, the time elapsed between last malaria treatment and delivery, and the re-infection rate during this period, which depends on malaria transmission intensity. Additional host factors may include HIV seropositivity, young age, and an early pregnancy number. Similar to our results, a previous study into the effectiveness of IPT with SP in Blantyre, Malawi, an area with seasonal transmission, showed a significant decrease of placental malaria in women who had used one or more doses of IPT (Rogerson et al. 2000). Although not significant, a decrease in placental malaria with two doses of IPT was seen in Chikwawa district in Malawi, an area with perennial transmission; the authors suggest that the high re-infection rate in this area may explain this result; an efficacy study in children in the same area during the same period showed 90% sensitivity to SP (Verhoeff et al. 1999). In our study, we saw the lowest prevalence of placental malaria in women who had received the last IPT dose within 6 weeks before delivery. However, even in women who received IPT before this interval, a protective effect was seen, indicating that re-infection may be less a problem in women delivering in the NPGH in Kisumu who had received IPT. In all the effectiveness studies mentioned here, a high prevalence of HIV infection was reported (30% in Blantyre, 22.3% in Chikwawa district and 24.4% in Kisumu), demonstrating that the benefit of IPT in pregnancy is seen even in settings with high HIV infection rates.

Because birthweight is a reflection of events during pregnancy and a strong predictor of infant health, birthweight may be an important indicator to monitor malaria control in pregnancy. In multivariable analysis, a decrease in the proportion of LBW was seen in women of all gravidities who had used IPT; when stratified by gravidity this was borderline significant in G1/G2 women (P = 0.05). A significant trend of decreasing proportion of LBW with increasing number of SP doses was seen in all women, and babies born to women who had used ≥2 doses of IPT were on average 128 g heavier than babies of women who had attended an ANC but had not received IPT. In Blantyre, Malawi, IPT was associated with an increase in mean birthweight of 195 g in infants of women receiving ≥2 doses IPT (Rogerson et al. 2000). In the last study, prevalence of placental malaria (33.2%) and LBW (22.9%) in women who had not received IPT were approximately twice the prevalence of placental malaria and LBW in women who had not received IPT in Kisumu (16.2% and 11.1%, respectively). The mean birthweight in untreated women in Blantyre was significantly lower compared with the mean birthweight in untreated women in Kisumu (2802 g SD 584 and 3069 g, SD 509, respectively, P < 0.001), and the absolute impact of IPT on birthweight may be greater in malaria endemic areas with more LBW. All the women in the hospital-based study in Chikwawa district, Malawi, received at least one dose of SP, making a direct comparison not possible (Verhoeff et al. 1998). However, in the last study, a strong beneficial effect on birthweight was seen with a mean increase of 257 g in primigravidae and 142 g in multigravidae when comparing women who received one dose of IPT with women who received two doses (Verhoeff et al. 1998).

Although several studies indicated that HIV-seropositive women may benefit from more frequent doses of SP to reduce placental malaria (Parise et al. 1998) and increase birthweight (Verhoeff et al. 1998), we failed to observe this specific difference in our analysis of HIV-seropositive women. Women enrolled in our cohort study were different on a number of characteristics including being more likely to have received IPT and iron/folic acid supplementation, and the prevalence of LBW did not differ significantly by the number of IPT doses received. It is likely that the enrolment criteria for this study (only women with an uncomplicated pregnancy and delivery, residing in Kisumu municipality, willing to be tested for HIV) and the additional service to this population (free access to health care, iron and folic tablets) selected for a population at low risk of delivery of LBW babies.

There are some limitations to this study, including limitations inherent to effectiveness studies in general. We only included women who delivered in the hospital, and from previous studies in this setting, we know that compared with women who deliver elsewhere, women who deliver in the hospital are younger, more likely to be primigravidae, of relative higher socio-economic status, to have more years of education, and are more likely to live near the hospital (van Eijk et al. 2002). However, socio-economic status and education were not associated with peripheral parasitaemia among women who attended the ANC of the Provincial hospital (van Eijk et al. 2002). In the same study, the nutritional status was not associated with placental parasitaemia among women who delivered in this hospital. The distribution of IPT was not randomly allocated; women attending the ANC of NPGH were more likely to receive IPT. Women from rural and peri-urban areas who did not receive SP had a different risk of placental malaria compared with urban residents who had not received SP, indicating a different risk of background exposure to malaria. Under the programmatic conditions of our study, SP doses were obtained from the ANC card, and non-recording of doses may have led to an underestimation of doses received. Non-recording of doses has been observed in the ANC and was estimated to occur in 2.2–7.5% and the possible underestimating of doses received would be expected to either not alter our findings or to result in a reduction of a true effect of SP on malaria and birthweight.

Among women who received IPT, 28% received the last dose within 6 weeks of delivery, despite the official recommendations not to give SP after 34 weeks of gestation. This may indicate the difficulties in this particular setting with accurate assessment of gestational age, as reported elsewhere (A. M. van Eijk, J. G. Ayisi, F. O. ter Kuile et al. 2002, personal communication). Assessment of gestational age of the infant at the time of delivery was not performed as part of the programme monitoring, so we cannot assess how prematurity was adding to the short-time period between last SP dose and delivery, or to birthweight.

In summary, we evaluated the effect of IPT with SP when delivered under ‘real life conditions’. A significant decrease in placental malaria and an increase in birthweight were seen in women who had visited an ANC and received IPT. In addition, we were able to determine effectiveness of IPT in pregnant women enrolled in a HIV-cohort study. No effect on birthweight was seen in this selected group. Given the higher prevalence of placental malaria in peri-urban/rural areas combined with the lower use of IPT of their pregnant residents, efforts should be made to increase coverage of IPT in ANCs serving these areas in particular.

Acknowledgements

The authors thank all the participating women for their help. We are grateful to the staff of NPGH, especially the ANC staff and labour ward staff, for their cooperation and patience, and thank Dr Larry Slutsker for his critical review of the manuscript and the Director of the Kenya Medical Research Institute (KEMRI) for his support. This work was supported by United States Agency for International Development (USAID). A. M. van Eijk, J. G. Ayisi, and F. O. ter Kuile acknowledge support from the Dutch Foundation for the Advancement of Tropical Research (WOTRO), The Hague, the Netherlands.