Effect of haematinic supplementation and malaria prevention on maternal anaemia and malaria in western Kenya

Summary

Objective  To evaluate the effect of routine antenatal haematinic supplementation programmes and intermittent preventive treatment (IPT) with sulphadoxine–pyrimethamine (SP) in Kenya.

Methods  Anaemia [haemoglobin (Hb) <11 g/dl), severe anaemia (Hb <8 g/dl) and placental malaria were compared among women with known HIV status who delivered at a provincial hospital after study enrolment in the third trimester during three consecutive periods: period 1, no routine intervention (reference); period 2, routine haematinic supplementation (60 mg elementary iron three times/day, folic acid 5 mg once daily) and period 3, haematinics and IPT with SP.

Results  Among 3108 participants, prevalence of placental malaria, anaemia and severe anaemia postpartum was 16.7%, 53.6% and 12.7%, respectively. Compared with period 1, women in period 2 were less anaemic [adjusted odds ratio (AOR), 95% confidence interval anaemia: 0.56, 0.47–0.67; severe anaemia 0.37, 0.28–0.49] and shared a similar prevalence of placental malaria (AOR 1.07, 0.86–1.32). Women in period 3 were also less anaemic (AOR anaemia: 0.43, 0.35–0.53 and severe anaemia: 0.43, 0.31–0.59), and had less placental malaria (AOR 0.56, 0.42–0.73). The effect of intervention did not differ significantly by HIV status.

Conclusion  The haematinic supplementation programme was associated with significant reductions in anaemia in HIV-seropositive and HIV-seronegative women. The subsequent introduction of IPT was associated with halving of malaria, but no additional haematological benefit over haematinics.

Objectif  Evaluer l'effet des programmes de supplémentation hématinique prénatales de routine et celui du traitement préventif intermittent (TPI) au sulfadoxine–pyriméthamine (SP) au Kenya.

Méthodes  L'anémie simple (hémoglobine [Hb] <11 g/dl), l'anémie sévère (Hb < 8 g/dl), et la malaria placentaire ont été comparées chez des femmes au statut VIH connu qui ont accouché dans un hôpital provincial après leur inclusion dans l’étude au cours du 3 trimestre selon trois périodes consécutives. Période 1: aucune intervention de routine (référence), période 2: supplémentation hématinique de routine (60 mg de fer élémentaire 3 fois/jour, 5 mg d'acide folique par jour), période 3: supplémentation hématinique et TPI au SP.

Résultats  Chez les 3108 participants, les prévalences de malaria placentaire, d'anémie simple et d'anémie sévère post-partum étaient de 16,7%; 53,6% et 12,7% respectivement. Comparées à la période 1, les femmes dans la période 2 étaient moins anémiques (rapport de cotes ajustés [AOR] pour l'anémie simple = 0,56 avec une intervalle de confiance [IC] 95%: 0,47–0,67 et pour l'anémie sévère AOR = 0,37, IC95%: 0,28–0,49) et partageaient une prévalence similaire de malaria placentaire (AOR = 1,07, IC95%: 0,86–1,32). Les femmes dans la période 3 étaient également moins anémiques (AOR pour l'anémie = 0,43; IC95%: 0,35–0,5 et pour l'anémie sévère: 0,43; IC95%: 0,31–0,59) et avaient moins de malaria placentaire (AOR = 0,56; IC95%: 0,42–0,73). Il n'y avait pas différence significative dans l'effet de l'intervention selon le statut VIH.

Conclusion  Le programme de supplémentation hématinique était associéà des réductions significatives de l'anémie chez les femmes VIH-séropositives et VIH-séronégatives. L'introduction du TPI en plus a été associée à une réduction de moitié de la malaria, mais aucun avantage hématologique additionnel n'a été observé pour les hématiniques.

Objetivo  Evaluar el efecto de los programas de suplementación hematínica antenatal rutinaria y el tratamiento preventivo intermitente (IPT) con sulfadoxina–pirimetamina (SP) en Kenia.

Métodos  Se compararon la anemia (Hemoglobina [Hb] <11 g/dl), la anemia severa (Hb < 8 g/dl) y la malaria placentaria entre mujeres con estatus de VIH conocido, que hubiesen dado a luz en un hospital provincial después de haber sido enroladas en el estudio durante el 3er timestre de embarazo. La comparación se hizo en tres períodos consecutivos: Período 1: sin intervención de rutina (referencia); Período 2: suplementación hematínica rutinaria (60 mg hierro elemental 3 veces/día, ácido fólico 5 mg una vez al día); Período 3: suplementación hematínica e IPT con SP.

Resultados  La prevalencia de malaria placentaria, anemia y anemia severa entre las 3108 participantes fue del 16.7%, 53.6%, y 12.7%, respectivamente. Con respecto al Periodo 1, las mujeres estaban menos anémicas durante el Periodo 2 (riesgo relativo indirecto [RRI], 95% intervalo de confianza para anemia: 0.56, 0.47–0.67; anemia severa 0.37, 0.28–0.49) y compartían una prevalencia similar para la malaria placentaria (RRI 1.07, 0.86–1.32). Durante el Periodo 3, las mujeres también estaban menos anémicas (RRI anemia: 0.43, 0.35–0.53; anemia severa: 0.43, 0.31–0.59), y tenían menos malaria placentaria (RRI 0.56, 0.42–0.73). El efecto de la intervención no difirió significativamente según el estatus de VIH.

Conclusión  El programa de suplementación hematínica estaba asociado a reducciones significativas de la anemia en mujeres seropositivas y seronegativas para VIH. La introducción posterior del IPT estaba asociada con una disminución de la malaria, sin beneficios hematológicos adicionales a aquellos hematínicos.

Introduction

Anaemia during pregnancy is a well-recognized problem; in developing countries, the prevalence is estimated at 52% (WHO/UNICEF/UNU 2001). Prepartum haemoglobin values <8 g/dl are associated with maternal mortality, low birth weight and infant mortality (Rush 2000; Brabin et al. 2001; Marchant et al. 2004). Anaemia postpartum reduces the maternal capacity to deal with the challenges during the puerperium (Atkinson & Baxley 1994; Corwin et al. 2003). Factors that predispose pregnant women to anaemia include potential exacerbation of micronutrient deficiency as a result of increased demand, insufficient diet and helminth infection (Beard 2000; Bothwell 2006); iron and folic acid supplementation are universally recommended during pregnancy (Stoltzfus & Dreyfuss 1998; WHO/UNICEF/UNU 2001). Infections such as malaria and HIV, and haemoglobinopathies may also contribute (van den Broek 1998; WHO/UNICEF/UNU 2001).

Although anaemia is common among HIV-infected pregnant women in sub-Saharan Africa (Meda et al. 1998; Ramon et al. 1999; Antelman et al. 2000; van Eijk et al. 2001), there is insufficient information about the risks and benefits of iron supplementation in this group (Friis et al. 2003). HIV-associated anaemia in pregnancy may reflect the ‘anaemia of chronic diseases’, which is less responsive to haematinic supplementation than anaemia caused by iron deficiency (Kreuzer & Rockstroh 1997; Means 1997). Other reports suggest that prolonged iron supplementation may increase HIV-1 replication and enhance disease progression, particularly among persons in an advanced stage of disease (Jacobus 1996).

Pregnant women in malaria endemic areas are at increased risk of malaria, compared with non-pregnant women (Brabin 1983). HIV further increases this susceptibility and reduces the efficacy of antimalarial interventions in pregnant women (Parise et al. 1998; ter Kuile et al. 2004). Many countries in sub-Saharan Africa have introduced, or are considering introducing, intermittent preventive treatment in pregnancy (IPTp) with sulphadoxine–pyrimethamine (SP), as national policy for the control of malaria in pregnancy. IPTp-SP consists of two or more presumptive treatment doses of SP after the first trimester through the antenatal clinic (ANC) (WHO/AFRO 2004). Although controlled efficacy studies indicate that IPTp-SP improves birth weight and haemoglobin levels, the impact of the policy on the burden of anaemia and malaria is not well documented (Parise et al. 1998; Shulman et al. 1999).

In Kisumu, western Kenya, we conducted a study to assess the interaction between malaria and HIV in pregnancy (Ayisi et al. 2004). During this 4-year study, a programme of routine haematinic supplementation was implemented in the course of the second study year, followed by IPTp-SP in the ANC involved approximately a year before the end of the study. The successive implementation of these interventions allowed us to assess their effect over time on anaemia and malaria in the antenatal study population. We were also able to evaluate whether their effect differed by HIV status or gravidity.

Materials and methods

Site and subjects

This study was conducted at Nyanza Provincial General Hospital, a large publicly funded hospital in Kisumu (population 300 000), western Kenya. Procedures have been described previously (Ayisi et al. 2004). Briefly, women at the ANC were eligible for participation if they had an uncomplicated singleton pregnancy of >32 weeks’ gestation, resided within Kisumu and had no known underlying chronic illness. Following informed consent, a questionnaire was administered in the local language to obtain information on sociodemographic, health and obstetric factors. A trained HIV counsellor counselled each woman. All screened women were encouraged to deliver at the hospital and approximately 50% did, reflecting the low rate of health facility deliveries in this area (37.4%) (Central Bureau of Statistics et al. 2004). Women who delivered at the hospital were significantly younger, and more likely to have a lower gravidity number and socioeconomic status, but were more likely to have a higher education level than women who delivered at home, and this conforms to the results of a national representative sample (Central Bureau of Statistics et al. 2004). After hospital delivery, a questionnaire was completed obtaining information on the delivery.

During the study, it was noted that routine iron and folic acid supplementation were not always available; therefore, from September 1997 onwards, haematinics were provided by the study clinic free of charge to enrolled clients as 200 mg ferrous sulphate three times and folic acid 5 mg daily for the duration of pregnancy as per MOH recommendation (Ministry of Health, Government of Kenya 1994). Participants received an explanation and leaflet in the local language on the importance and potential side effects of haematinic supplementation. We did not supervise the daily intake.

Because of accumulating evidence over time on the benefits of IPTp-SP (Parise et al. 1998; Shulman et al. 1999), IPTp-SP was introduced in March 1999 in the ANC involved, in accordance with newly adopted policy of the Ministry of Health (Ministry of Health, Government of Kenya 1998). The nurses in the ANC assessed the eligibility of the pregnant women, supervised their intake, and registered their doses on the ANC card. Pregnant women who participated in the study could have received one or more doses of SP before they were enrolled.

Laboratory procedures

Blood samples were taken from participants at enrolment for HIV antibody testing, haemoglobin and malaria, and after delivery in the hospital for haemoglobin and malaria; the last sample was taken as close to the time of delivery (within 24 h) as possible. A placental and a cord blood smear were prepared at delivery. Haemoglobin was measured to the nearest 0.1 g/dl using a HemoCue machine (Mission Viejo, CA, USA). Peripheral and placental thick blood films were stained with 10% Giemsa and examined under oil immersion for malaria parasites. A thick film was considered negative if 100 microscopic fields showed no parasites. Malaria parasites and leukocytes were counted in the same fields until 300 leukocytes were counted. Parasite densities were estimated by using an assumed count of 8000 leukocytes/μl blood. HIV testing of pregnant women used two rapid test methods: an initial Serostrip HIV-1/2 (Saliva Diagnostic Systems Pte Ltd, Singapore) and a confirmatory Capillus HIV-1/HIV-2 (Cambridge Diagnostics, Wicklow, Ireland) on all Serostrip-positive samples. Western blot was performed on discordant samples.

Definitions

Maternal anaemia (any anaemia) was defined as a haemoglobin level below 11 g/dl and severe anaemia as a haemoglobin level below 8 g/dl. Malaria was defined as the presence of asexual-stage parasite in thick smears, independent of any clinical signs or species. Placental parasitaemia was defined as any plasmodial asexual form detected on a thick film. Maternal HIV infection was defined as a positive result on both rapid tests; women not reactive with the initial Serostrip HIV-1/2 test were considered HIV-seronegative. Gravidity was divided into primi- or secundigravidae (G1/G2) vs. gravidae 3 or more (G3+), because of their difference in susceptibility to malaria and anaemia (van Eijk et al. 2001).

An uncomplicated pregnancy was defined as a pregnancy without the presence of AIDS-defining symptoms or other obstetric or chronic medical conditions. Women with a history of severe peri- or postpartum blood loss were defined as women with a peripartum haemorrhage; blood loss was not quantified. An axillary temperature of 37.5 °C or more was defined as documented fever. High rainfall in the previous month was defined as rainfall in the previous month, which was more than the 50 percentile for the entire study period.

Statistical analysis

Differences in proportions were analysed using the chi-square test or Fisher's exact test when appropriate. Differences in mean were compared using Student's t-test. A variable was introduced, called ‘period of enrolment’, indicating enrolment during pregnancy before the introduction of any intervention (period 1: 24 June 1996 to 23 September 1997), enrolment after routine haematinic supplementation but before the implementation of IPTp-SP (period 2: 24 September 1997 to 14 March 1999), and enrolment after routine haematinic supplementation and after the implementation of IPTp-SP in the ANC in the hospital (period 3: 15 March 1999 to 23 June 2000).

Odds ratios (OR) were computed with their 95% confidence interval (CI) using logistic regression models to measure the strength of the associations between the period of enrolment (and therewith the intervention) and anaemia or malaria (Rothman & Greenland 1998). The effects of the period of enrolment on placental malaria were adjusted for potential confounders identified in previous analyses, including gravidity, age, HIV, place of residence (urban vs. semiurban/rural), ethnicity and socioeconomic status (van Eijk et al. 2002). The factors season at the time of delivery and high rainfall in the previous month were not associated with placental malaria and were not included. We did not have information on dietary intake or helminthic infections in the participants, and could not adjust for these potential confounders. Gravidity was a risk factor for placental malaria among HIV-seronegative women but not among HIV-seropositive women. This interaction was statistically significant (P = 0.02) and has been reported in several other studies (ter Kuile et al. 2004). The two-way interaction between HIV and gravidity as well as the main effect terms for HIV and gravidity were therefore entered as covariates in the models for placental malaria. Two- and three-way interactions were examined between HIV, gravidity and period of enrolment to determine the relative effect of each intervention by HIV and gravidity groups, both of which are known to be important determinants of the efficacy of IPT on malaria (Parise et al. 1998; van Eijk et al. 2004b).

A similar approach was used to determine the effect on anaemia with and without interaction terms for HIV and gravidity. The effect of gravidity on anaemia did not vary by HIV status (P-value of interaction term: 0.4). All anaemia models were adjusted for gravidity, age, HIV, placental malaria, ethnicity, mode of delivery (a caesarean section vs. vaginal delivery) and high rainfall in the previous month; place of residence and socioeconomic status were not associated with anaemia. Severe anaemia in the third trimester was included in the model because an unequal distribution of this variable across periods 1 and 2 among HIV-positive women was detected, when no effect of an intervention could account for it (16.9% in period 1 and 24.9% in period 2, P = 0.02); no such difference was seen for any anaemia or among HIV-negative women. Because during enrolment in the third period, IPTp-SP may or may not have been received at the time of enrolment in the ANC, a variable indicating if the enrolment visit was the first antenatal visit was added to the model.

To further assess the relationship between the haemoglobin level at delivery and the period of enrolment, linear regression was used with postpartum haemoglobin as the outcome variable, and the period of enrolment as exposure variable, while adjusted for enrolment at first antenatal visit, mode of delivery, placental malaria, haemoglobin at enrolment, HIV, gravidity, age, ethnicity and high rainfall in the previous month. Haemoglobin is known to decrease during pregnancy, with a nadir at approximately 34 weeks gestation; for this reason, time between third trimester assessment and delivery was calculated in weeks and introduced in the model as well (Subcommittee on dietary intake and nutrient supplements during pregnancy 1990). Statistical analysis revealed no collinearity between haemoglobin at delivery and in the third trimester. spss (spss for Windows version 11, SPSS Inc., Chicago, IL, USA) and sas (sas system for Windows version 8, SAS, Cary, NC, USA) were used for all analyses. All tests were two-sided; P < 0.05 was considered significant.

Ethical review

The study protocol was approved by the institutional review boards of the Kenya Medical Research Institute; CDC, Atlanta, GA, USA; and the Academic Medical Centre (AMC), University of Amsterdam, Amsterdam, the Netherlands. Issues concerning HIV case management during the course of the study have been discussed previously (Ayisi et al. 2004).

Results

Population characteristics

A total of 3155 enrolled women delivered in the provincial hospital during the study period. Of the 3108 women who had haemoglobin level in the third trimester and HIV status available, 37.7% were enrolled in period 1, 36.7% in period 2 and 25.6% in period 3 (Table 1). The median time interval between the enrolment visit in the third trimester and delivery was 4.6 weeks (range 0–18 weeks) and did not vary significantly by HIV status or gravidity. The mean age (standard deviation, SD) was 21.7 (4.6) years and was significantly higher among HIV-seropositive, compared with HIV-seronegative women [22.2 (4.3) years vs. 21.5 (4.6) years, respectively, P < 0.01]. Differences in characteristics between time periods and between HIV-seropositive and HIV-seronegative women over time can be seen in Table 1; haemoglobin levels were significantly lower among HIV-seropositive women, compared with HIV-seronegative women across all time periods.

Table 1.   Characteristics of study women, stratified by HIV status and period of enrolment in the ANC, Kisumu, June 1996 to June 2000

Characteristic	Overall	Period 1: 24 Jun 1996 to 23 Sep 1997 – No intervention		Period 2: 24 Sep 1997 to 14 Mar 1999 – Haematinics		Period 3: 15 Mar 1999 to 23 Jun 2000 – Haematinics and IPTp-SP
	n = 3108 (%)	HIV+ n = 308 (%)	HIV−n = 864 (%)	HIV+ n = 257 (%)	HIV−n = 883 (%)	HIV+ n = 178 (%)	HIV−n = 618 (%)
Haematinic supplementation started at the enrolment visit in periods 2 and 3, whereas IPTp with SP could start at the start of antenatal clinic visits. HIV, human immunodeficiency virus; ANC, antenatal clinic; SES, socioeconomic status; Hb, haemoglobin; SD, standard deviation; IPTp-SP, intermittent preventive treatment with sulphadoxine–pyrimethamine; low/medium SES, based on the absence of electrical supply to the house. *P < 0.05 comparing HIV-positive vs. HIV-negative women in the same time period. †P < 0.05 comparing women of the same HIV status with each of the other time periods. ‡P < 0.05 comparing women of the same HIV status from time period 1 with time period 2. §P < 0.05 comparing women of the same HIV status from time period 1 with time period 3. ¶An axillary temperature at delivery only available for 1855 women (59.7% of the study population).
Enrolment third trimester
Primi- or secundigravidae	70.3	68.5	71.1	66.9	71.1	62.4*	72.5
Age <21 years (median age)	50.0	45.8	52.3	44.4	50.8	33.7*†	54.7
Luo ethnicity	77.6	82.8*	75.0	87.2*	75.8	84.8*	75.1
Low/medium SES§	77.1	78.6	76.9	80.5	77.8	76.8	74.4
Periurban/rural residence	21.7	22.4	21.2	25.3	24.6	17.4	17.5†
Complaint of fever previous week	19.6	28.6*†	22.2†	18.7	16.8	19.8	15.6
Documented fever	1.7	4.2	2.8†	1.6	0.6	1.1	0.7
Antimalarial used before in pregnancy	23.6	25.6	20.8	20.3	21.0	36.2†	28.2†
Haematinics used before in pregnancy	10.9	14.9	12.9†	9.3	8.5	13.5	9.4
Mean haemoglobin (SD), g/dl	9.7 (1.7)	9.3 (1.5)*	9.9 (1.8)	9.1 (1.7)*	9.7 (1.7)†	9.2 (1.8)*	10.0 (1.7)
Hb <11 g/dl	76.7	84.4*	71.4	86.8*	77.8†	88.2*	71.4
Hb <8 g/dl	14.7	16.9†	12.8	24.9*	13.1	25.8*	10.8
Malaria third trimester	16.7	26.4*	16.6	25.5*	16.8	12.6†	9.2†
Enrolment at first ANC visit	15.0	13.3	12.5‡	14.4	17.1	17.4	15.9
BMI (SD), kg/m^2	23.8 (2.7)	23.4 (2.4)*	23.8 (2.6)§	23.6 (2.4)	23.9 (2.8)	23.9 (3.1)§	24.1 (2.9)
Delivery
Complaint of fever previous week	22.2	31.2‡	27.1	23.4	23.3	15.3†	10.6†
Documented fever¶	3.7	5.6	4.2	6.8	3.4	1.8	2.0
Peripartum haemorrhage	0.6	0.6	0.5	2.7*	0.5	0.6	0.0
Caesarean section	3.2	1.3	2.2	2.7	5.7†	1.1	2.6
Mean haemoglobin (SD), g/dl	10.7 (2.5)	9.6 (2.4)*†	10.2 (2.4)‡	10.3 (2.2)*	11.1 (2.3)	10.6 (2.5)*	11.4 (2.5)†
Hb <11 g/dl	53.6	69.5*§	61.1‡	61.9*	46.5	56.2*	40.9†
Hb <8 g/dl	12.7	23.7*†	16.9†	14.4*	7.4	15.7*	7.4
Maternal malaria	13.7	21.5*	14.0	21.6*	13.2	11.4†	7.3†
Placental malaria	16.7	24.8*	16.0	29.3*	16.2	11.8†	10.6†

In the third trimester, overall 76.7% of women were anaemic: 86.1% of the HIV-seropositive women vs. 73.8% of the HIV-seronegative women (OR 2.21, 95% CI: 1.76–2.77). At the time of delivery, the overall prevalence of anaemia had decreased to 53.6%, but a significant difference between HIV-seropositive and HIV-seronegative women remained (63.7%vs. 50.4%, OR 1.72, 95% CI: 1.46–2.04).

Haematinic supplementation before enrolment in the study was used by 10.9% of all women and was similar among HIV-seropositive and HIV-seronegative women, but higher in the first period among HIV-seronegative women. Women who used haematinic supplementation before enrolment in the study were more likely to have been admitted to hospital during the present pregnancy (7.1%vs. 2.8%, OR 2.67, 95% CI: 1.67–4.29), but haemoglobin levels did not differ among women who used and did not use haematinics before enrolment in the study (9.7 g/dl, SD 1.7 vs. 9.7 g/dl, SD 1.7, respectively).

In total, 23.6% of the participants reported the use of any antimalarial prior to enrolment, and 10.4% reported the use of SP. Compared with HIV-seronegative women, HIV-seropositive women tended to be more likely to have used an antimalarial (26.3%vs. 22.8%, respectively, OR 1.21, 95% CI: 1.00–1.46) or SP (12.5%vs. 9.7%, respectively, OR 1.33, 95% CI: 1.03–1.72) before enrolment. As expected, the use of any antimalarial or SP prior to enrolment increased during period 3 (Table 1); however, the implementation of IPT was not optimal as reported before, with 43% of the women only receiving SP at their first ANC visit (van Eijk et al. 2004a).

For 732 of 796 women who participated in period 3 (92.0%), the number of IPTp-SP doses was available as reported by the women at delivery: 194 women received two doses (26.5%), 328 women received one dose (44.8%) and 210 women (28.7%) received no dose of SP. The last or only dose of SP was given with a median of 9 weeks before delivery (range 1–24 weeks). Among the same 732 women, 695 (94.9%) reported at delivery to have used haematinic supplementation during pregnancy, 688 (99.0%) reported to have used iron daily, 608 (87.5%) reported to have used three tablets daily and 565 women (81.3%) reported to have used iron for at least 2 weeks. There was no difference in the number of SP doses or haematinic supplementation by HIV status.

Bed net use was recorded during periods 2 and 3 among a cohort of study women and infants who were in a follow-up study; overall bed net use was 39.3% among 361 women, and was similar during periods 2 and period 3 (40.3% and 37.9%, respectively). Regular retreatment with an insecticide was only reported for 14 nets (10.9%); during the time of the study, insecticide was not routinely commercially available for this purpose in western Kenya. The difference in placental malaria between bed net users and non-users was not statistically significant (9.2%, 95% CI: 0–19, χ² = 3.52, df = 1, P = 0.07); postpartum anaemia was not associated with bed net use.

Malaria

It should be noted that the intervention of haematinic supplementation in periods 2 and 3 started at the moment of enrolment in the study, whereas the intervention of IPTp-SP during period 3 could have occurred before, during or after enrolment in the study. Compared with women enrolled in period 1 (no intervention) or period 2 (haematinics), women enrolled during period 3 (haematinics and IPTp-SP) had a significantly reduced prevalence of peripheral parasitaemia in the third trimester and at delivery, a history of fever in the week preceding delivery, and placental malaria (Table 1; Figure 1).

Figure 1.  Placental malaria for the total study population and stratified by HIV status, and rainfall during the study period, Kisumu, June 1996–June 2000. *Nurses’ strike in the hospital of enrolment during November to December 1997. Twenty-seven women enrolled during this period: prevalence of placental malaria among HIV-seropositive women 4 of 10, among HIV-seronegative women 1 of 17. Periods of enrolment: Period 1, no intervention – 24 June 1996 to 23 September 1997; Period 2, haematinic supplementation – 24 September 1997 to 14 March 1999; Period 3, haematinic supplementation and IPT – 15 March 1999 to 23 June 2000. Routine haematinic supplementation for enrolled women and intermittent preventive treatment with sulphadoxine–pyrimethamine in the ANC.

In the logistic regression model of peripheral parasitaemia in the third trimester, enrolment in period 3 (IPTp-SP) was associated with an adjusted odds ratio (AOR) of 0.48 (95% CI: 0.36–0.65) among all women, an AOR of 0.50 (95% CI: 0.22–1.11) among HIV-positive women and an AOR of 0.47 (95% CI: 0.33–0.68) among HIV-negative women (adjusted for same variables as reported for placental malaria, see also Table 2).

Table 2.   Period of enrolment as risk factors for placental malaria and postpartum anaemia in multivariable analysis in the total study population and by maternal HIV status, Kisumu, June 1996 to June 2000†

	Period of enrolment	All women AOR (95% CI)	HIV [+] women AOR (95% CI)	HIV [−] women AOR (95% CI)
HIV, human immunodeficiency virus; IPT, intermittent preventive treatment; HIV [+], HIV-seropositive; HIV [−], HIV-seronegative; AOR, adjusted odds ratios. Significant odds ratios are given in bold. †Period of enrolment – 24 June 1996 to 23 September 1997: no intervention, reference; 24 September 1997 to 14 March 1999: routine haematinic supplementation for enrolled women; 15 March 1999 to 23 June 2000: routine haematinic supplementation for enrolled women and IPT with sulphadoxine–pyrimethamine through the antenatal clinic. ‡Models 1, 3 and 5 represent the overall effect adjusted for maternal HIV status. Models 2, 4 and 6 represent the effect by maternal HIV status obtained from separate models containing an interaction term between maternal HIV infection and period of enrolment. The P-values for these interaction terms were P = 0.2 for model 2, P = 0.5 for model 4 and P = 0.7 for model 6, respectively, indicating that the difference of haematinic supplementation alone, or haematinic supplementation and IPT did not differ significantly by maternal HIV status. All models for placental malaria were adjusted for gravidity (primi- and secundigravidae vs. gravidae 3 or more), HIV, the interaction term between HIV and gravidity, age, place of residence, ethnicity and socioeconomic status (presence of electricity in the house); models for anaemia at delivery were adjusted for gravidity, age, HIV, placental malaria, ethnicity, mode of delivery, high rainfall in the previous months, severe anaemia in the third trimester (Hb < 8 g/dl) and first ANC visit at enrolment.
Placental malaria		Model 1‡	Model 2	Model 2
	No intervention	Reference	Reference	Reference
	Haematinics	1.07 (0.86–1.32)	1.21 (0.67–2.19)	1.01 (0.78–1.31)
	Haematinics and IPT	0.56 (0.42–0.73)	0.43 (0.20–0.90)	0.61 (0.44–0.84)
Hb <11 g/dl postpartum		Model 3	Model 4	Model 4
	No intervention	Reference	Reference	Reference
	Haematinics	0.56 (0.47–0.67)	0.64 (0.39–1.07)	0.53 (0.44–0.65)
	Haematinics and IPT	0.43 (0.35–0.53)	0.43 (0.24–0.77)	0.43 (0.34–0.55)
Hb <8 g/dl postpartum		Model 5	Model 6	Model 6
	No intervention	Reference	Reference	Reference
	Haematinics	0.37 (0.28–0.49)	0.39 (0.19–0.81)	0.36 (0.26–0.50)
	Haematinics and IPT	0.43 (0.31–0.59)	0.49 (0.21–1.13)	0.40 (0.27–0.60)

In the logistic regression model of placental malaria, the three-way interaction term between HIV, gravidity and period of enrolment was not significant (P = 0.4), and neither was the two-way interaction gravidity and period of enrolment (P = 0.4), indicating that the effect of haematinic supplementation and IPTp-SP on placental malaria did not vary by gravidity. The effect of IPTp-SP on placental malaria was the greatest in HIV-seropositive women (Table 2); however, the difference was not statistically significant (P-value interaction term between HIV status and period of enrolment P = 0.2).

Anaemia in third trimester

Although peripheral malaria in the third trimester decreased in period 3, no decrease in anaemia or an increase in haemoglobin was seen when comparing period 1 and period 3 or period 2 and period 3 (Table 1); this observation did not vary by HIV status (adjusted mean difference in haemoglobin comparing period 3 with period 1 for total study population: 0.03 g/dl, 95% CI: −0.14 to 0.19 g/dl; for HIV-positive women: −0.10 g/dl, 95% CI: −1.07 to 0.87 g/dl; for HIV-negative women: 0.07, 95% CI: −0.13 to 0.26 g/dl; values obtained as described in Table 3; adjusted for gravidity, age, HIV, ethnicity, rainfall in the previous month and enrolment at first ANC visit).

Table 3.   Mean difference in maternal haemoglobin (g/dl) at delivery by period of enrolment overall and by maternal HIV status, Kisumu, June 1996 to June 2000†

Period of enrolment	All women	HIV [+] women	HIV [−] women
	Difference (g/dl) relative to reference (95% CI)	Difference (g/dl) relative to reference (95% CI)	Difference (g/dl) relative to reference (95% CI)
	Model 1‡	Model 2	Model 2
HIV, human immunodeficiency virus; IPT, intermittent preventive treatment; Hb, haemoglobin. Significant differences are given in bold. †Period of enrolment – 24 June 1996 to 23 September 1997: no intervention, reference; 24 September 1997 to 14 March 1999: routine haematinic supplementation for enrolled women; 15 March 1999 to 23 June 2000, routine haematinic supplementation for enrolled women and IPT with sulphadoxine–pyrimethamine through the antenatal clinic. ‡Model 1 represents the overall effect adjusted for maternal HIV status. Model 2 represented the effect by maternal HIV status obtained from a separate model containing an interaction term between maternal HIV infection and period of enrolment. The P-value for this interaction term was 0.3, indicating that the difference of haematinic supplementation alone, or haematinic supplementation and IPT did not differ significantly by maternal HIV status. All models were adjusted for enrolment at first antenatal visit, mode of delivery, placental malaria, haemoglobin at enrolment, HIV, gravidity, age, ethnicity, high rainfall in the previous month and number of weeks between enrolment and delivery.
No intervention	Reference	Reference	Reference
Haematinics	0.94 (0.76–1.12)	0.71 (−0.30 to 1.72)	1.02 (0.81–1.22)
Haematinics and IPT	1.10 (0.89–1.31)	1.03 (−0.08 to 2.14)	1.12 (0.88–1.36)

Anaemia at delivery

Period 2 (the period of haematinic supplementation) and period 3 (the period of the combination of haematinic supplementation and IPTp-SP) were both associated with decreased prevalence of anaemia relative to period 1 (Table 2) and an increase in mean haemoglobin (Figure 2). This effect was observed in both HIV-seropositive and HIV-seronegative women, although the effect was the greatest in HIV-seronegative women. Nevertheless, this difference was not statistically significant (P-value for interaction term between HIV status and period of enrolment 0.5 and 0.7 for any and severe anaemia, respectively).

Figure 2.  Haemoglobin levels at delivery for the total study population and stratified by HIV status, and rainfall during the study period, Kisumu, June 1996 to June 2000. Periods of enrolment: Period 1, no intervention – 24 June 1996 to 23 September 1997; Period 2, haematinic supplementation – 24 September 1997 to 14 March 1999; Period 3, haematinic supplementation and IPT – 15 March 1999 to 23 June 2000. Routine haematinic supplementation for enrolled women and intermittent preventive treatment with sulphadoxine–pyrimethamine in the ANC.

Linear regression models showed that haematinic supplementation was associated with an increase in postpartum haemoglobin concentrations over time (Table 3). The greatest effect was again observed in HIV-seronegative women, but it was also seen in HIV-seropositive women and the difference in effect was not significant (P-value interaction term between HIV and period = 0.3). Enrolment in period 3 among all pregnant women was associated with a larger mean increase in haemoglobin, compared with period 2 (1.10 g/dl vs. 0.94 g/dl, Table 3), but when using period 2 as reference this was statistically not significant (difference 0.15 g/dl, 95% CI: −0.06 to 0.36 g/dl, P = 0.2).

Discussion

The introduction of haematinic supplementation was associated with a concomitant increase in mean haemoglobin levels and decreased prevalence of anaemia in this antenatal population over time. Given the less-than-optimal duration of the haematinic supplementation for many women, it is unlikely that all iron- and folic-acid-deficiency-associated anaemia had been corrected, and a longer course may have improved the results further. The national guidelines recommend relatively high doses of iron (200 mg ferrous sulphate three times/day; this is equivalent to 180 mg elementary iron/day) and folic acid (5 mg/day); these are higher than international recommendations (120 mg elementary iron/day, 0.4–0.6 mg folic acid per day) (Stoltzfus & Dreyfuss 1998). All women received a verbal explanation and an information leaflet in their local language on the use of haematinics and expected side effects, and this may have increased the adherence, but we did not monitor this.

The addition of IPTp-SP in 1999 resulted in a marked reduction in peripheral and placental malaria (45.7% and 42.9%, respectively) and a small additional increase in mean haemoglobin (0.20 g/dl) and a small decrease in any anaemia (11.4%) at delivery, compared with the effect of haematinics alone. However, this did not translate to decreased risks of severe anaemia at delivery. The lack of an observed effect on severe anaemia may reflect the short duration between the last dose of SP and delivery (less than 6 weeks in one third of the participants who received SP) as full recovery from malaria-associated anaemia takes 42 days in successfully treated symptomatic patients (Price et al. 2001). Anaemia is commonly used as an indicator of efficacy or effectiveness of antimalarials in pregnancy (Cot et al. 1998; Parise et al. 1998; Shulman et al. 1999; Kayentao et al. 2005). This retrospective study shows that care should be taken to interpret information on haemoglobin from such studies, if haematinic supplementation has been provided as part of the protocol or in an uncontrolled way by the ANC.

The high prevalence of anaemia among HIV-seropositive women in the third trimester (86.1%) was in line with previous reports from urban ANC populations in Cote d'Ivoire (81.7%), Burkina Faso (78.4%) and Tanzania (83.0%) (Meda et al. 1998; Ramon et al. 1999; Antelman et al. 2000), but considerably higher than that reported from urban ANC populations in Zimbabwe (54%) and Rwanda (6.6% < 10 g/dl) (Leroy et al. 1998; Friis et al. 2001). It has been suggested that iron supplementation over a longer period of time could accelerate progression of HIV (Boelaert et al. 1996; Friis et al. 2001). Micronutrient deficiency studies among HIV-infected women in Zimbabwe revealed that serum folate and serum ferritin levels were significantly lower in HIV-infected women when compared with HIV-uninfected women, and were associated with lower haemoglobin levels, indicative of iron and folic acid deficiency (Friis et al. 2001). The same authors report that storage iron was a predictor of HIV viral load, and that this was not a result of an acute phase response or iron accumulation with advanced HIV infection (Friis et al. 2003). This is in contrast with a cross-sectional study in Malawi, whereby iron status was not associated with viral load (Semba et al. 2001). One observational study reported that 60 mg of elementary iron twice weekly over 4 months did not affect the viral load (Olsen et al. 2004). We do not have data to assess the stage of HIV infection among the HIV-seropositive pregnant participants, and did not monitor viral load changes during pregnancy. The inclusion criteria of the study precluded the enrolment of symptomatic HIV infection, and in a subgroup of 426 women, the median CD4 count 1-month postpartum was 601 cells/μl (interquartile range 413–818). This observational study suggests that these asymptomatic HIV-seropositive pregnant women do respond to haematinic supplementation: the effect of haematinic supplementation was clearly evident in HIV-seropositive women (0.8 g/dl), but greatest in HIV-seronegative women (0.9 g/dl). Our study could not address the question of the potential effect of iron supplementation on HIV disease progression; however, the period of supplementation was relatively short.

This hospital-based observational study enrolled a selected group of relatively healthy women; results cannot be readily extrapolated to the rural or urban community. The prevalence of anaemia at the time of delivery may have been affected by the time between delivery and the maternal blood sample postpartum, when the contraction of the plasma volume immediately postpartum may increase the haemoglobin. We compared haemoglobin during 3 time periods; so we cannot exclude that other factors which changed during these time periods and affected anaemia or malaria have contributed to our results, e.g. an overall decline of malaria or iron and folic acid deficiency in the study area over time would have resulted in an overestimation of the effect of IPTp-SP and haematinic supplementation. However, a significant decrease of placental malaria coinciding with the introduction of IPTp-SP was observed. The use of mosquito nets may have affected our observations; however, as the limited sample of women for whom we had this information available, reported mosquito net use was the same during period 2 and period 3, and the percentage of regularly insecticide-treated nets was low. A change in diet or socioeconomic status coinciding with these periods is unlikely: the body mass index was similar throughout the periods and indicators of socioeconomic status also remained similar.

In summary, anaemia among pregnant women remains an important public health problem. Haematinic supplementation, even when started late in the third trimester of pregnancy, may have a beneficial effect on haemoglobin level. Although the long-term effect of haematinic supplementation in HIV-seropositive women needs to be evaluated, the short-term effect suggests an encouraging reduction in postpartum anaemia. IPTp-SP markedly reduced placental malaria infection and remains important for all pregnant women in malarious areas.

Acknowledgements

We thank the Director of the Kenya Medical Research Institute (KEMRI) for his support and Irene van Laarhoven for her exploratory work. Dr AM van Eijk was supported by the Oak Ridge Institute for Science and Education.