Inadvertent exposure of pregnant women to ivermectin and albendazole during mass drug administration for lymphatic filariasis


John O. Gyapong (corresponding author), Margaret A. Chinbuah and Margaret Gyapong, Health Research Unit, Ghana Health Service, PO Box GP-184, Accra, Ghana. Fax: +233 21 22 6739; E-mail:


The current strategy for the interruption of transmission of lymphatic filariasis in areas where the disease is co-endemic with onchocerciasis is repeated annual mass treatment of endemic communities with ivermectin and albendazole. These drugs are not recommended for use in pregnancy. Pregnant women are excluded on the basis of their last menses. This exclusion criterion based on recall carries some inherent errors, leading sometimes to inadvertent exposure of foetuses to these drugs. This study set out to document the extent of inadvertent exposure of pregnant women to albendazole and ivermectin and assess the relative risk of congenital malformations because of inadvertent treatment with these drugs in early pregnancy. The study was conducted in the Ahanta West District of Ghana. Local pregnancy revelation norms were studied, followed by a household survey of women aged 15–45 years to assess drug administration coverage. All infants born within 42 weeks of the mass drug treatment were examined to document any congenital malformations. Mothers who had lost any such infants responded to a verbal autopsy to ascertain the probable cause of death. Health facilities and local Traditional Birth Attendants were also visited to review maternity records. Of 2985 women of childbearing age (15–49 years) who were interviewed, 343 were pregnant during the mass drug administration. The sensitivity of the last menstrual period in detecting pregnancy and thus being excluded from treatment was 0.854 (293 of 343). Some pregnant women 50 of 343 (14.6%) had thus been inadvertently treated. This represents 1.7% of women in fertile age group (15–49 years). Of the six children found with some congenital malformations in these communities, one had been exposed to the drugs in-utero. The relative risk for congenital malformation after exposure was 1.05 (P = 1.0). Two of nine reported spontaneous abortions had been exposed to the drugs (P = 0.62). We conclude that the local mode of excluding pregnancy in the current programme, while not perfect, is sufficiently effective and reliable for such a public health intervention; and importantly, that there is no evidence of a higher risk of congenital malformation or abortions in those who are inadvertently exposed.


Filariasis is an endemic disease of public health importance in many parts of the developing world (Gyapong 1999, WHA 1997, Michael et al. 1996). In Ghana it is endemic in the northern sector as well as coastal communities west of Accra (Dunyo et al. 1996; Gyapong et al. 1996a; Gyapong & Remme 2001). In some communities, 32–41% of the population more than 10 years of age are microfilaraemia-positive for Wuchereria bancrofti and 3% have lymphoedema. Although mortality from the disease is low, it is a chronic disease often resulting in deformity and can also be severely debilitating during acute exacerbations of adenolymphangitis (Gyapong et al. 1996b, 1996c).

Repeated annual, mass treatment of eligible community members with appropriate drugs is the current strategy for filariasis control (Ottesen et al. 1997). In the endemic areas in Ghana, Community Directed Treatment is the delivery strategy used (Gyapong et al. 2001). The drug of choice is ivermectin (which has been successfully used for the treatment of onchocerciasis), in combination with albendazole, an anti-helminthic. This combination produces sustained complete clearance of microfilaria for 15 months (Ismail et al. 1998). Both ivermectin and albendazole are not recommended for use in pregnancy and thus pregnant women are excluded from treatment based on their last menstrual period (LMP). Because this (LMP) is based on recall, some pregnant women may be inadvertently included in treatment programmes. Concerns about the possible teratogenic effects of these drugs and the question of whether more definitive screening methods to determine pregnancy could be used have been raised. However, in mass treatment programmes, it is almost impossible to do pregnancy screening on a large scale without concrete evidence to justify the cost. Neither is it possible to exclude all women of childbearing age from treatment programmes because that would severely compromise the goal of interrupting transmission in endemic communities.

Our study was initiated to determine the reliability of LMP as an exclusion criterion for pregnancy in community-based treatment programmes and to assess whether inadvertent treatment of pregnant women with the programme drugs has a higher risk of miscarriages or teratogenicity than the normal population.

Patients and methods

The study was conducted in the Ahanta-West District of the Western Region of Ghana, which lies in the equatorial rain forest belt, with secondary forest vegetation and primary forest reserve. In the southern sector, the vegetation is the coastal Savannah type with stretches of mangrove swamps; thus, its coastline is indented with lagoons and major rivers (Butre, Yanney, Howin and Suni), entering the sea. The low lying nature of the coast line, coupled with the characteristic abundance of mangrove trees, give rise to stagnant pools and ponds, which facilitate the breeding of the Anopheles gambiae and An. melas mosquitoes. The estimated population of the district is 125 000. The inhabitants are mainly fishermen and farmers who grow crops-like cassava, plantain and other vegetables. Microfilaraemia prevalence ranges between 10 and 20%; lymphoedema prevalence is 2% and hydrocele prevalence ranges between 10 and 15% in endemic communities. The health infrastructure is organized around the concept of a district hospital with satellite health centres. The District Health Management Team (DHMT) is responsible for Health Administration and Public Health activities for the district. There are also trained traditional birth attendants and village health workers supporting the health centres in their outreach activities.

The study population included all women of childbearing age (15–45 years) in the district, as well as all infants born to these mothers within 42 weeks of the period following the treatment. A qualitative study was carried out to inform the development of structured questionnaires. Communities were grouped into coastal/fishing communities and farming communities and a random sample of two fishing and three farming communities selected. Focus group discussions were held with young women, older women, young men and older men to collect information on local cultural pregnancy revelation norms and practices. In addition, issues relating to abortions were explored. This information was then used to complement information from other data collection techniques where relevant.

The focus group discussions were followed by a household survey of women of childbearing age normally resident in the communities. These women were interviewed to determine those who were excluded on account of a history of amenorrhoea (reported pregnancy), representing those who should have been unexposed; and those who received treatment as a result of a history of menses (reported not pregnant), representing those exposed. Sixteen communities that had adequately documented information on pregnancy status of women at the time of treatment in their community registers were included in the sample. An electronic database of all women in these communities aged 15–49 was compiled for the study. All women belonging to households within the selected communities were visited and interviewed to find out if they had received the drugs. During the interviews, the women were also asked if they had been pregnant during the time of drug distribution, and/or if they had had a child since. If they had had a child and the infants were alive, then they were asked to bring the child on a specific date for examination and given a card with the name of the child and mother as a reminder and for identification of appropriate children by the physicians. If the child had died, an arrangement was made for a verbal autopsy to be conducted. If the pregnancy was not carried to full term, the circumstances surrounding the loss of the pregnancy were also documented (Figure 1).

Figure 1.

Schematic presentation of study design.

Two paediatricians then examined all infants born to these women within 42 weeks of the period following treatment, for congenital malformations using a standardized instrument.

All health facilities, both public and private, within the district were visited to review records of all deliveries before and after mass treatment with ivermectin or albendazole had begun in any of the communities. This was to document any congenital malformations in the period under review. Practicing traditional birth attendants were also included in this review; they were interviewed and their records were examined as well.

Data management, quality control and analysis

Data collection instruments were standardized through training. All focus group discussions were audio recorded and supplemented with notes taken in longhand. The audio recordings were transcribed and analysed manually. Research Assistants (RA) who conducted interviews were observed during at least one interview by the Co-PI and any mistakes made were corrected. On the first 2 days of data collection during the survey, the Co-PI sat through at least one interview by each RA to ensure consistency. The three supervisors stayed with the RA's throughout the data collection period. All completed instruments were checked while on the field for internal consistency and completeness. Any discrepancies noted were corrected in the field. Data collected were coded using a coding manual, doubly entered, and validated. Internal consistency checks were performed for range as part of data cleaning. Data was analysed electronically using EPI info 6.04.

The assessment of the reliability of the LMP as the exclusion criterion was based on the number of pregnant women who were inadvertently treated compared with those excluded from treatment on account of pregnancy. The number of infants in the exposed group (those whose mothers took the drug) with congenital malformations was then compared with the number of infants in the unexposed group (those whose mothers did not take the drugs) to determine the relative risk of congenital malformation of the two groups. A similar assessment was carried out for the relative risk of exposure of the drugs to spontaneous abortions.

Ethical approval

Clearance was sought and obtained from all the appropriate Health Authorities. Consent of community elders, chiefs and opinion leaders was obtained through personal visits and group discussions. All participants of the focus group discussions and the in-depth interviews, women of childbearing age and household heads were informed about the purpose of the study and their consent sought and obtained. Participants were assured of the confidentiality of any information given. Mothers of infants who were found to have congenital malformations or any other disease that required medical attention were informed about the findings, and the study team facilitated their seeking appropriate care.


From the group discussions, no specific norms or cultural practices prescribing the revelation of pregnancy in its early stages especially to health staff or community volunteers were found. Some respondents mentioned that LMP alone was inadequate to detect early pregnancy. They would usually look for other confirmatory signs such us early morning sickness (23.6%), a feeling of being unwell (17.2%), swollen tender breasts (15.2) and excessive salivation (10.5%).

A total of 2985 interviews with women of reproductive age group (15–45) were conducted. Of these, 1590 (53.3%) received the drugs (Table 1) and 343 (11.5%) were pregnant and should have been excluded from taking the drugs. However, some pregnant women received the drugs. Table 1 shows pregnancy status by those who received and swallowed the drugs. Thus, the proportion of women excluded successfully using the local criteria to determine pregnancy was 293 of 343 (85.4%). However, 50 of 343 (14.6%) received and swallowed the drugs (Figure 1). Figure 2 shows the gestation of their pregnancies at the time they were inadvertently exposed. In these communities, the age at which women first detected their pregnancies varied (Figure 3). By the first month 70.3% had detected their pregnancies, by the second month, 92.4%. Thus, on the whole pregnancy detection was fairly early. This was confirmed during the focus group discussions, by some men and women who mentioned that they could detect the pregnancies almost immediately after conception.

Table 1.  Extent of inadvertent exposure of pregnancy women to programme drugs
StatusExposed to drugsNot exposed to drugsTotal frequency
Figure 2.

Gestation at which pregnant women were inadvertently exposed to the drugs.

Figure 3.

Age at which pregnancies were first detected for all pregnant women in the communities.

Gestation at the time of detection of pregnancy was tabulated against gestation at the time of drug distribution. The results indicate that 24 of the 50 pregnant women who were exposed to the drugs (48%) took the drugs before they knew they were pregnant. Thus, 52.0% of the pregnant women who were exposed knew they were pregnant before they took the drugs. The reasons why are shown in Table 2. Half (25 of 50, 50.0%) did so because they did not know that pregnant women should not take the drugs. Two (4%) took the drugs because they wanted to terminate their pregnancies.

Table 2.  Reasons why pregnant women taken the drugs
Did not know that they should not take drugs when pregnant2550
Wanted to abort the child24
Was not aware of pregnancy2448

Twelve of the 343 pregnant women (3.5%) had lost their pregnancies, nine of these were spontaneous and three were induced. Two of the nine women who had miscarried spontaneously had swallowed the drugs (Table 3). The relative risk of miscarriage following ingestion of the drugs was 1.67 (P = 0.627). But this finding is inconclusive because the numbers are very small.

Table 3.  Risk of miscarriage following ingestion of drugs
 MiscarriageNo miscarriageTotal
  1. Relative risk = 1.67 (0.36 < RR < 7.83); P-value (Fisher exact) =0.62.

Exposed to drugs24850
Not exposed to drugs7286293

Women who had reported they were pregnant at the time of drug distribution were asked if they had delivered babies regardless of whether they had been born alive or dead. A total of 332 of 343 (96.8%) had already delivered at the time of the examination. Fifteen of 343 (4.4%) had lost their babies and seven mothers were available for verbal autopsies. Of the remaining 317 children, 250 (79.9%) were seen and examined by two paediatricians. There were 120 males and 130 females (52.0%). Two hundred and thirty three (93.2%) infants were born at term, i.e. within 37–42 completed weeks. Only one infant (0.4%) was born premature but was found normal on examination and 16 of 250 (6.4%) were born after 42 weeks of gestation. The mean age of the infants examined was 19 weeks and the median 16 weeks.

The infants were weighed in a hanging scale and their head circumference and supine lengths measured with a tape measure. All major body systems were examined and results are shown in Table 4. Standard paediatric examinations for sight and hearing were performed and developmental milestones were assessed (Tanner & Whitehouse 1973). Children who may not have reached the age for particular milestones were excluded from that particular assessment, hence the variations in the denominator. Other details are shown in Table 5. At the end of the evaluation, 244 of 250 (97.6%) infants were assessed as not having any abnormal feature; the remaining six of 250 (2.4%) had one or more abnormal features. One of the six had been exposed to the drug in-utero at 10 weeks. He had presented with a hearing impairment. Table 6 shows the details. The risk of congenital malformation resulting from inadvertent exposure of women while pregnant was found to be no different from the normal population (Table 7).

Table 4.  Findings on examination of the children
FindingsNormal frequency (n = 250) (%)Abnormal frequency (%)Remarks about abnormal features
Head shape246 (98.4)1 (0.4)Prominent occiput
Head circumference246 (98.4)4 (1.6)Small <3rd centile for age
Deformity of head247 (98.8)1 (0.4)Craniostoisis
Protrusions from the scalp250 (100)0Depressed skull in both the frontal (left part) and occipital (right part) bones
Anterior fontanel248 (99.2)2 (0.8) 
Posterior fontanel250 (100)0 
Appearance of face249 (99.6)1 (0.4) 
Eyes249 (99.6)1 (0.4)Mongoloid slant
249 (99.6)1 (0.4)Hypertelorism
Ears235 (94.0)15 (6.0)Low set ears
Nose250 (100)0 
Neck249 (99.6)1 (0.4)Appeared short
Upper limbs250 (100)0 
Lower limbs250 (100)0 
Palmer crease249 (99.6)1 (0.4)Single palmer crease
Foot250 (100)0 
Digits (hands and feet)249 (99.6)1 (0.4)Syndactyly and over-riding digits
General muscle tone250 (100)0 
Chest250 (100)0 
Breast250 (100)0 
Spine250 (100)0 
Hip joint250 (100)0 
Shoulder joint250 (100)0 
Elbow joints250 (100)0 
Knee joints250 (100)0 
Male genitalia120/120 (100)0 
Female genitalia130/130 (100)0 
Abdomen249 (99.6)1 (0.4)Enlarged liver
Anus248 (100)0 
Neurological movements of the eye250 (100)0 
Verbal response/cry250 (100)0 
Best motor response on pain stimulus250 (100)0 
Planter reflexes250 (100)0 
Moro's reflexes248 (99.2)2 (0.8)Present when should have been absent
Sucking reflexes250 (100)0 
Stepping reflexes249 (99.6)1 (0.4) 
Palmer reflexes250 (100)0 
Apex beat – position249 (99.6)1 (0.4)Fifth left intercostal space
Heart rate249 (99.6)1 (0.4) 
Heart beat rhythm250 (100)0 
Murmurs present249 (99.6)1 (0.4)Best heard at apex (VSD)
Table 5.  Developmental milestones of infants
MilestoneOn questioning motherDifficulty found on examination
Normal for ageAbnormalNormal for ageAbnormal
Frequency (%)Frequency (%)Frequency (%)Frequency (%)
Feeding problems248/250 (99.2)2/250 (0.8)249/250 (99.6)1/250 (0.4)
Sight/seeing250/250 (100)0250/250 (100)0
Hearing247/250 (99.1)2/250 (0.8)247/250 (99.2)2/250 (0.9)
Playing242/242 (100)0242/242 (100)0
Crawling 75/78 (96.1)3/78 (3.9) 76/78 (97.4)2/78 (2.6)
Sitting113/114 (99)1/114 (1)113/114 (99)1/114 (1)
Walking  9/12 (75)3/12 (25)  9/12 (75)3/12 (25)
Table 6.  Congenital malformations found on examination of infants
Type of defectNumber of children, N = 234 (%)Exposure to drugsGestation at which mother detected pregnancyGestation at time of exposure to drugs
  1. * This was a 16-week-old infant who did not respond to loud noises such as a clap and jingling of keys. His mother had also noticed that he did not respond to noises. A diagnosis of hearing impairment was made.

  2. † This was an 8 month old female infant with a loud systolic murmur, best heard at the apex of the heart found in the 5th left intercostal space.

Mongolism/Downs1 (0.4)NoneNot applicableNot applicable
Ventricular septal defect†1 (0.4)NoneNot applicableNot applicable
Craniostiosis1 (0.4)NoneNot applicableNot applicable
Syndactyly1 (0.4)NoneNot applicableNot applicable
Hearing impairment*1 (0.4)Yes9 weeks10 weeks
Large umbilical hernia1 (0.4)NoneNot applicableNot applicable
Table 7.  Relative risk of congenital malformation because of inadvertent exposure during pregnancy
  1. Relative Risk = 1.05 (0.13 < RR ≪ 8.75); P-value (Fisher exact) =1.0.

Congenital malformation
Exposure to drugsYesNoTotal
Yes1 39 40

Of the 15 deaths recorded, seven mothers were available for interviews. None had taken the drugs while pregnant. Six indicated that their children had been born without any obvious deformity. One woman, however, said there were deformities of the eyes, arms, hands and fingers, chest, back, legs and toes and genitals. She said the skin of the entire baby seemed rotten and it had difficulty with breathing and died almost immediately after birth. The physician's impression of this description was possibly a dysmorphic baby or a macerated stillbirth.

One hospital and three health centres were visited and records 2 years prior to drug distribution examined. Only one record of a congenital malformation was documented. This was a case of an anencephalic infant born to a woman in February 2001. She however did not belong to any of the study communities. Sixteen traditional birth attendants (TBAs) were visited. Ten of them were active and had delivered children in the year of the study. None recalled delivering an infant with a deformity. All available records were also examined and no deformities had been recorded for the years under review.


Drug treatment of infectious diseases is a vital strategy to bring diseases of public health importance under control. With lymphatic filariasis, this could lead to a substantial reduction in the parasite density in the community and eventually interrupt transmission (Ottesen et al. 1997). The role of the community in the distribution of these drugs in most endemic communities is a vital step in achieving the high treatment coverage essential for the elimination of the disease (Gyapong et al. 2001). In mass treatment programmes, inadvertent treatment of a section of the population is possible irrespective of who does the drug distribution. In these study communities, health volunteers are trained to inform community members that certain persons are not eligible for treatment, notably seriously sick people, pregnant women and children under 5 years (operationalized in the field as children less than 90 cm).

The concept of LMP is used to exclude pregnancies as it is impossible to do pregnancy tests in such mass drug treatments at the community level. We found that in this population, the use of LMP as the pregnancy exclusion criterion inadvertently exposed some 14.6% of pregnant women to the drugs. This is quite high and a lot of effort is needed to reduce the extent of inadvertent exposure in future treatments to exclude pregnant women especially in the early stages of pregnancy since the risk of develop-ment of congenital malformations is highest in the first 3 months of gestation.

Thirty-nine of 50 (76%) of the foetuses were exposed during the first trimester and 11 of 50 (22.0%) after the first trimester, however of these, 15 of 39 (38.5%) of the women had detected their pregnancies before the mass treatment was carried out, thus, 24 of 39 (61.5%) really represents those who were inadvertently treated because they did not know they were pregnant. As a proportion of women who were inadvertently exposed, they represent half, i.e. 25 of 50 (50.0%) and as a proportion of all pregnant women in the communities they represent 25 of 343 (7.3%).

The problems as revealed by this study are twofold: the detection of pregnancy and the knowledge by women and indeed all community members that these drugs are not to be taken while pregnant. Miscarriages are difficult to establish and there is always a lot of secrecy and some stigma associated especially if one is suspected of inducing an abortion. In this study, we did not attempt to validate responses of mothers who said they had miscarried. Mothers were simply asked if they had been pregnant at the time of drug distribution, and the gestation of the pregnancy at that time as well as at detection. They were then asked what the outcome of the pregnancy had been. Eleven women had miscarried and nine of these had been spontaneous. Only two women had taken the drugs and both had miscarried spontaneously. Although this gives some cause for concern, the numbers are too small for statistical inference.

Our study is limited by a small sample size and this is reflected in the lack of precision of our figures through wide confidence intervals. We also accepted mothers’ estimates of the ages of gestation at the time of detection and at the time of mass drug distribution for filariasis. Mass distribution had been carried out in July or November in most of the communities visited. The specific dates of treatment were not known, but the drugs were distributed within 2 weeks of receipt, thus the list of these dates gave us the approximate period for treatment for each community. Since there had been a lapse of 8 and 11 months respectively between drug treatment and the survey, the dates for detection of pregnancy and the gestation of the pregnancy at the time of drug distributions could only be estimates. Within this limitation, the Research Assistants tried to reduce memory bias by using a simple calendar to help confirm the gestation at drug distribution using the child's age if she/he was born and if pregnancy went to term and the number of weeks provided by the mother.

Accurate, early and cheap detection of pregnancy on a mass scale is difficult. Local peoples’ detection of pregnancy was in conformity with basic scientific knowledge. They used mainly LMP (97.7%) with other overt signs of pregnancy such as early morning sickness (23.6%), a feeling of being unwell (17.2%), swollen tender breasts (15.2) and excessive salivation (10.5%). This was confirmed in the focus group discussions where all respondents were emphatic that the last menstrual period alone cannot be used to determine pregnancy. It had to be confirmed with other signs.

The dilemma for programme management is not in how the pregnancy is detected, but how early this can be done using local means without the use of laboratory/scientific testing, which on a large-scale will be too expensive and impractical. According to the men, some can actually tell that their wives are pregnant within a few days even before the woman herself recognizes the pregnancy state. From this survey, 1.8% of women detected their pregnancies within a week of getting pregnant. By the third week only 14.3% had detected their pregnancies, but this improves dramatically so that by 1 month, 70.3% of the women had detected their pregnancy. This however, still leaves 29.7% who had still not detected their pregnancies after 1 month and were thus at risk of inadvertent exposure of their foetuses to ivermectin and albendazole. This is important because congenital malformations tend to occur in the early months of gestation. By the third month, however, 98.3% had detected their pregnancies. In this study the overall relative risk for congenital malformations following exposure was 0.82. This compares favourably with reports in other populations (Pacque et al. 1990).

The second problem revealed by this study is that of information, education and communication (IEC). One would pre-suppose that the various IEC messages sent around would have been enough to inform community members adequately on the exclusion criteria in this treatment programme. However, there are often many reasons why this is difficult to achieve. These may be related simply to the clarity of the message, the media used to inform community members, people's uptake and understanding of the message.

In some communities, there are cultural and language barriers that forbid the discussion of issues relating to pregnancy. In this study, there were no cultural taboos related to the discussion and revelation of pregnancy. Half of the women who took the drug did so because they did not know that the drugs were contraindicated in pregnancy. This is high and must be addressed in further training of the health volunteers, health durbars and in discussions with local people during the mobilization phase of the mass distribution. Reinforcing the messages prior to and during each round could improve community members’ awareness of exclusion criteria.

In conclusion, this study has shown that although the last menstrual period remains the mainstay of pregnancy detection, early pregnancy detection is still a problem in most rural communities in Ghana. About 14.6% of pregnant women in these communities were inadvertently exposed to the programme drugs and this could have been reduced to 7.6% if IEC messages had been adequate. The message of excluding pregnant women from mass drug administration thus needs to be reinforced at all levels. There is however no evidence that the foetuses exposed to ivermectin and albendazole in this study were at a higher risk of abortion nor congenital malformation.

We recommend that these communities be continually monitored for subsequent mass drug administrations for the LF programme. To reduce the likelihood of inadvertent treatment, women should be asked to come for their drugs only after they have had some evidence of their menses for that month.


We are grateful to Drs Jennifer Welbeck, Geoffrey Tsaras, and Kwabena Agyeman of the Korlebu Teaching Hospital in Accra, who did all the paediatric assessment and the immense role played by the health workers and research assistants in the study area during the data collection phase of the study. We acknowledge the data management team especially Ms Gertrude Owusu- Banahene, Dominic A Kobinah, Mrs Cecelia Amoakwao and Mrs Dela Semackor. We also acknowledge the role of the Ahanta West District Health Management Team, Western Region especially the late Dr Emmanuel Asamoah, Mr Joseph Newton and Mr Adomako Asare. This investigation received financial support from the UNDP/World Bank/WHO Special Programme for Research and Training in Tropical Diseases and GlaxoSmithKline.