Protocol for a drugs exposure pregnancy registry for implementation in resource-limited settings

Background: The absence of robust evidence of safety of medicines in pregnancy, particularly those for major diseases provided by public health programmes in developing countries, has resulted in cautious recommendations on their use. We describe a protocol for a Pregnancy Registry adapted to resource-limited settings aimed at providing evidence on the safety of medicines in pregnancy.Methods/Design: Sentinel health facilities are chosen where women come for prenatal care and are likely to come for delivery. Staff capacity is improved to provide better care during the pregnancy, to identify visible birth defects at delivery and refer infants with major anomalies for surgical or clinical evaluation and treatment. Consenting women are enrolled at their first antenatal visit and careful medical, obstetric and drug-exposure histories taken; medical record linkage is encouraged. Enrolled women are followed up prospectively and their histories are updated at each subsequent visit. The enrolled woman is encouraged to deliver at the facility, where she and her baby can be assessed.Discussion: In addition to data pooling into a common WHO database, the WHO Pregnancy Registry has three important features: First is the inclusion of pregnant women coming for antenatal care, enabling comparison of birth outcomes of women who have been exposed to a medicine with those who have not. Second is its applicability to resource-poor settings regardless of drug or disease. Third is improvement of reproductive health care during pregnancies and at delivery. Facility delivery enables better health outcomes, timely evaluation and management of the newborn, and the collection of reliable clinical data. The Registry aims to improves maternal and neonatal care and also provide much needed information on the safety of medicines in pregnancy. © 2012 Mehta et al.; licensee BioMed Central Ltd. This Pregnancy Registry is a prospective observational cohort study that enrols pregnant women at their first antenatal visit to a selected health facility. To obtain a more accurate medical and drug exposure history, early enrollment is desirable. The population of interest is pregnant women at selected antenatal clinics (ANCs). These clinics are chosen because of their location in areas with a high prevalence of the disease of interest (i.e. where women are likely to be exposed to the drug/s in question). In addition, women presenting to these facilities should be very likely to initiate antenatal care early in pregnancy and deliver at the facility. All pregnant women seeking care for the first time during their current pregnancy are eligible. However, if there is a very high case-load at a facility, a random sampling method might be devised. Thus, the number of women recruited at any facility will depend on practical considerations such as available staff capacity, attendance rates, facility delivery rates, and willingness of women to consent to being followed to term. Country-sites fulfilling the following criteria would be ideally placed to contribute to the Registry database. • Strong commitment to the Registry from the relevant departments of the Ministry of Health, particularly Reproductive health and disease control programmes (eg malaria and HIV/AIDS). • Written agreement to pool all data derived from the Pregnancy Registry into the WHO International Registry database. • A moderate to high prevalence of the disease of interest with drugs of interest in common use for these diseases (e.g. HIVAIDS, malaria, etc). • Good antenatal, labor/delivery and diagnostic services, and a willingness of staff to be trained to improve reproductive care. • Early ANC care attendance at the clinic. • High attendance at delivery at the health facility. (This increases the likelihood of a skilled assessment of the child at birth). OR • A reliable plan to follow-up women delivering at home, and assess the child (e.g community health nurses/midwives to monitor home deliveries). • Identification of a local pediatrician or doctor and commitment of such a person to review births and photographs of births and advise on management of infants. OR • Identification, where possible, of a local specialist in the diagnosis and management of congenital anomalies. • Identification of record-linkage systems between general out- or in-patient records and specialized clinics such as HIV clinics. The approach described in this protocol establishes the baseline risk of poor birth outcomes (miscarriages, deaths, congenital anomalies, premature delivery, low birth weight) and enables calculation of additional risk of mortality or morbidity associated with specific diseases with or without exposure to specific drugs. Pooling data from different countries strengthens interpretation of findings beyond a single population or geographical location and enables detection of small or moderate risks in a larger population. A sound obstetric, medical and drug history requires a systematic approach by health staff. A training package has been developed and includes modules to address the following: – How to obtain informed consent, if required – How to obtain a comprehensive medical, drug and obstetric history from pregnant women – How to perform and record the findings of maternal physical exams – How to encourage women to attend follow-up visits as recommended in national guidelines – How to encourage women to deliver at the health facility – How to capture data in the Registry database – Procedures for tracing and following up recruited women and newborns – How to examine a newborn infant and take a digital photograph for expert assessment (An instructional video has been developed as a tool to teach the systematic newborn surface examination). – How and when to seek or refer a child for specialist clinical or surgical management. In the African region, 18 countries report ANC attendance rates of ≥ 90% for at least one prenatal visit, with only 4 countries reporting a rate below 50% [16]. Hence, women in many parts of Africa can be enrolled at their first ANC visit as a representative sample of the pregnant population seeking care. Figure 1 provides a flow diagram of the study procedures from the initial visit at the sentinel site to the national/regional registry. Flow diagram of study procedures. Depending upon staff capacity, attendance rates and case-load, a decision should be made on whether to enroll all women who meet the inclusion criteria or a random sample of these women (e.g. 1 day per week, or first 2 new ANC cases per day). A second decision is whether all consenting pregnant women exposed to a drug of particular current interest, or with the particular condition of interest, should be considered for enrollment irrespective of the random selection process. At her first ANC visit, a woman’s obstetric and medical history is documented, including treatments and laboratory tests for acute or chronic conditions (e.g. diabetes, epilepsy, HIV, malaria), and referrals to higher levels of care made for specific checks as necessary. Data on routine treatments to be provided during pregnancy, including iron and folic acid, are collected and the woman is encouraged to return at scheduled ANC appointments for further checks and updated documentation on the pregnancy. She is encouraged to give birth at the healthcare facility to reduce the risk of adverse outcomes for herself and her baby, and to enable examination of the newborn by trained staff. If enrolled women do not return for ANC visits or deliver at the sentinel health facility within the expected period of time after a scheduled ANC visit or delivery date, they are followed up by phone or at home. This encourages facility attendance during the pregnancy. In the case of a home delivery it enables the pregnancy outcome to be documented. After home births, women are encouraged to return to the health facility with their infants for an examination and, if needed, medical care. Loss to follow-up of enrollees constitutes a limitation of the Registry because it will not be known whether the participants lost have the same outcomes as those who remain under observation. Losses to follow-up can be minimised by training health staff responsible for intake to record detailed contact information of the women, counselling them about the importance of follow-up antenatal visits and delivering at the facility and providing women with a contact number at the clinic to inform nurses about any moves to travel, changes in their condition). A final requirement is to identify a national, or preferably a local, neonatologist who can be contacted (by telephone) to advise on the management of a newborn requiring immediate clinical or surgical attention. Participants voluntarily agreeing to participate should be enrolled at the chosen sentinel sites using the following inclusion and exclusion criteria: All of the following conditions must be met for enrolment: • A confirmed pregnancy (by physical examination, pregnancy test or ultrasound) • Presentation at the first antenatal visit for the current pregnancy • Information on presence/absence of fetal abnormality unknown at the time of enrolment. • Voluntary agreement by the woman to be followed up to term. Meeting any of the following criteria would exclude a woman from enrolment: • Refusal to participate in the Registry • Presence of a medical, psychiatric, or social condition that interferes with the woman’s ability to provide an accurate medical or drug history or give informed consent (e.g. mentally disabled patients) • Women who report that they will not give birth at the health facility. A coloured, visible sticker may be placed on the ANC card of women meeting the selection criteria. The sticker signals that facility staff must fill out CRF1 (see Additional file 1) at the first visit and update it at each subsequent visit and perform a physical exam on the infant when the woman delivers. The Registry CRF1 includes a more detailed history than standard ANC information, but otherwise all procedures routinely carried out at ANC in accordance with national guidelines are conducted without modification. It documents demographic data, medical history, obstetric history, infections, drugs and vaccine use, intermittent preventive treatments and previous birth outcomes. The CRF1 establishes whether the woman has had exposure to the medicine/s of interest, the timing of exposure and whether the exposure took place on the basis of a clinical or laboratory confirmed diagnosis. Various aids are introduced at the sentinel sites to improve women’s recall of drug exposures during pregnancy- including tablet and treatment package visual identification kits, treatment diaries or medicine storage sleeves. These recall aids are utilized at the initial and follow-up ANC assessments. Where possible, drug exposure is verified through record linkage with other systems. Results of any tests or clinical assessments are recorded. These could include voluntary counseling and testing for HIV, assessment of weight, height, gestational age, presence of chronic diseases or conditions such as diabetes, syphilis and anaemia, and pharmaceutical medications taken during or just before the current pregnancy. Table 1 provides an overview of the procedures and assessments to be conducted during each of the facility visits during a woman’s pregnancy. Review of Pregnancy Registry-related assessments and procedures conducted during antenatal and intrapartum care At the end of each ANC visit, an appointment for the next ANC visit is made and contact details of the woman are taken and updated when necessary. Women are asked to return for follow-up assessments as normally scheduled, generally at least four ANC visits for each pregnancy. Updated information on the woman’s obstetric and medical conditions, medication use, smoking and alcohol consumption are also obtained during follow-up ANC visits (CRF 1). At each ANC visit the woman is counseled to return to the facility for delivery. The labor/delivery ward staff identifies Registry enrollees by the brightly coloured labels on the ANC cards. Immediately after delivery, the baby (whether alive or stillborn) will undergo a careful physical surface examination and the details recorded on a Pregnancy Outcome CRF (CRF2 – see Additional file 2). This form also records information on any abortions or miscarriages which terminate the pregnancy. The vital status of the infant is recorded as well as sex, head circumference, length and weight. The date, place and time of examination and the age of the baby at examination (number of days since birth) is documented on CRF2 , so that analysis can separate babies assessed at birth from those assessed some time later. Any abnormalities observed and any adverse outcomes in the mother or baby are noted and reported in the CRF. Training materials have been developed to support antenatal facility staff and delivery staff, to complete the CRFs and to assess a newborn. The systematic surface examination, as outlined in the training video, identifies congenital anomalies which are visible to the examiner. Some of these could be of major surgical or clinical importance (refer to the section below on ascertainment and classification of outcome). Digital photographs of any abnormality are taken for the purposes of confirmation and classification by an international Birth Defect Panel. For the process of confirming birth defects, see the section on ascertainment of outcome and Figure 2. Assessment process for congenital anomalies. Photographs will not be taken if a mother refuses to have a photograph taken of her infant at the time of birth even if she had agreed to do so earlier. Her decision is recorded on CRF2. Infants with major anomalies are referred to a specialist for confirmation of the diagnosis (see CRF 3 – Additional file 3) and further management. It is not expected that all staff performing deliveries in resource-limited settings, usually nurses/midwives, can auscultate for heart murmurs nor is it assumed that assessment of other internal problems through ultrasound equipment will be routinely available. The Registry does not require that the palate be examined for oral clefts nor that the hips be examined for congenital hip dysplasia. Consequently, these important malformations will not be detected unless clinical signs are noted and further tests are performed. If a woman does not deliver at the facility, arrangements need to be made to carry out the surface examination as soon as possible after birth. In resource-limited settings, where home-based delivery rates can be high and referral systems inefficient, completion of the initial examination and follow-up on findings within 12 weeks is considered acceptable, especially if the examination is done at a facility. A community health nurse can be taught to conduct home assessments for women who do not come to the facility to deliver. Community-based staff would need to be adequately trained and equipped to ensure that off-site assessments are equivalent to facility-based assessments, especially if many assessments are likely to be home-based. Although home-based follow-up is important, the high rates of facility-based BCG vaccinations even in the poorest areas of Africa, suggests that a facility assessment can be undertaken at the time of the first vaccination. Losses to follow-up and un-assessed or uninvestigated neonatal deaths or birth defects represent a limitation, of the Registry as the possibility of infanticide or neglect could occur if the newborn has a visible abnormality such as clubfoot, cleft lip or myelomeningocele [17]. The WHO Ethics Review Committee and the ethics committees of the 5 participating countries (Brazil, Ghana, Kenya, Tanzania, and Uganda) implementing the pilot study have reviewed and approved the protocol for use. Any site or country wanting to implement the protocol can do so and establish whether local/national ethics approval is required. Investigators, in consultation with public health officials and ethics review committees, should determine whether the Registry will be required to take informed consent from enrollees and, if so, whether written or verbal consent is appropriate. Some ethics committees may waive the need for written informed consent on the basis that: i) the research is observational, part of standard of care and involves no additional risk; ii) a waiver does not adversely affect the rights and welfare of the subjects; AND iii) the subjects will be provided with pertinent information, if appropriate, after analysis [18]. In settings where the Registry is seen as part of a government program evaluating treatment policy for high priority diseases in the country, a waiver may be sought since written informed consent increases the time spent with each woman at ANC and can reduce time for patient care. Participation in the Registry is voluntary. Hence, if a woman refuses to be part of the Registry, she can do so without consequence to her medical care. If a woman refuses to have a digital photograph taken of her infant (i.e. in the case of a malformation), her data may be included without photographs, with her agreement. The woman must be assured that all information is kept confidential and that her identity will be anonymized when data are analyzed. A guidance note is provided to women on the safe use of medicines in pregnancy in order to engender confidence in the medicines being prescribed for her during her pregnancy and the need for exercising caution when using over-the-counter and other un-prescribed medicines (Additional file 4). Clinical staff at participating health facilities are trained to rapidly identify and refer infants with adverse birth outcomes or birth defects to the appropriate medical providers or healthcare facilities, to improve neonatal outcomes. However, each participating facility must decide how to cover referral and treatment costs. All significant findings from analyses need to be communicated to sentinel site staff on an ongoing basis. Should any information come to light, which may directly or indirectly benefit or harm the patient, the attending clinician or health care provider should be informed as soon as possible (e.g. if a baby is born with a hereditary or potentially preventable birth defect the mother may need counselling, or if it is noticed that a clinic is prescribing inappropriate medication to pregnant women). The woman should be informed that anonymized information from all countries will be pooled in a relational online database and the findings shared with the facility. An International Birth Defects Assessment Panel has been established by WHO to review, confirm and classify major birth defects reported by the participating sites. Although all birth defects are documented, exclusion criteria have been developed by the Panel on the categories of birth defects which are not relevant. All defects that do not constitute “major malformations” defined as a structural abnormality with surgical, medical or cosmetic importance and hereditary and chromosomal disorders that are not caused by exposure to drugs, are excluded. The following features/anomalies are therefore excluded from the primary endpoint of assessing drug-related risk [19]: • minor anomalies ( e.g. transverse palmar crease) • normal variations (e.g. umbilical hernia in African infants) • hereditary disorders (e.g. polydactyly, postaxial Type B) • birth marks (e.g. haemangiomas; congenital moles) • chromosome abnormalities (e.g. Down syndrome) • positional deformities (e.g. congenital hip dislocation in infants who are born in breech position; • features of prematurity (e.g. undescended testes and patent ductus arteriosus in infants less than 37 weeks gestational age; • biochemical abnormalities(e.g. carriers of cystic fibrosis, and abnormal haemoglobins that may be identified during newborn screening. Progress and training of staff needs to be monitored regularly for quality and reliability of procedures, examinations and data. This will reduce the possibility of biases in enrolment and the risk of false negative results (e.g. when a surface examination of a baby was not complete). Detection of the presence/absence of malformations depends upon the training, skill, and experience of the assessor. Absence of malformations will not be reassuring, nor will data be reliable if assessors have not looked for them or are unable to assess a child. Only reliable data should contribute to national policy. For this reason it is recommended that a random sample of 5% of all pregnancy outcomes be re-evaluated by a second nurse and the comparison of these separate findings be recorded. The cohort study design allows for analysis of the data using three approaches – 1) as a complete cohort, 2) using a case–control approach and 3) using a case-cohort approach. The cohort analysis facilitates examination of the demographic and clinical information about the mothers and infants and prevalence of risk factors such as drug exposures during a particular trimester. Case–control studies can be performed to determine the risk of specifically defined birth defect/s (e.g. neural tube defects) and whether or not these are associated with in-utero exposure to particular medicines. A case-cohort analysis approach can be used to test hypotheses not initially considered when the cohort was initiated or to analyze data where additional assessments have been carried out. Using this approach, a randomly selected sub-cohort of non-cases will be selected from the main cohort and used to compare risk factors for any or all adverse outcomes. Case–control studies enable the assessment of whether exposures to specific drugs are associated with an increased frequency of specific, major malformations. The protocol enables assessment of the additional risk of adverse outcomes consequential to drug exposure. Confounding is addressed through the use of multivariate analysis, where the sample size is adequate. Alternatively, cases and controls can be matched in order to address the potential for confounding. Subsequent analyses will therefore use paired statistical methods. In the uncommon situation where the underlying condition being treated and the suspected drug/s give rise to the same adverse birth outcome or birth defect, the issue of confounding by indication arises. In such instances, assessing the differences in risk profile for different drugs used to treat the same condition could assist in addressing this potential confounding. At the point of analysis, for both the exposed and the unexposed comparison group, the pre-defined inclusion/exclusion criteria for major congenital anomalies need to be used to specify which features count as major malformations. To include minor features as “major” abnormalities unfairly and inappropriately exaggerates the apparent fetal effects of an exposure [20]. This makes it essential that studies of potential teratogens set inclusion/exclusion criteria in advance of analysis to evaluate the test group. The same criteria must be used to evaluate their unexposed comparison group. Therefore, inclusions and exclusions are not conducted at sentinel sites but by the International Birth Defects Panel of experts, without prior knowledge of group/exposure. Sentinel sites are encouraged to report all anomalies (major and minor) and all birth outcomes regardless of the type or severity of such anomalies, whereas the Birth Defects Panel adjudicates the defect as major or minor. The background risk of major congenital anomalies and other adverse birth outcomes (e.g. stillbirths/miscarriage/abortion) at sites is estimated from the proportion of major anomalies and other adverse outcomes identified in newborns of women who report no inter-current clinical condition or exposure to potentially teratogenic medication during the relevant stage of pregnancy. Analyses should be carried out for drug exposures in all three trimesters as well as for the first trimester only. As the number of enrolled women increases over time, sub-analyses that assess the safety of individual medicines or combination therapies becomes possible, as well as detailed analysis based on dose and duration of exposure. Descriptive statistics are the primary approach for summarizing data from pregnancy exposure registries with frequencies of outcomes expressed via absolute risk, relative risk, and population attributable risk, with 95% confidence intervals. Data collected through the Registry should be evaluated with standard periodic reviews of the database to assess the primary and secondary endpoints as well as to assess the quality of the information submitted by contributing sites based on pre-defined indicators (e.g. missing data fields, inconsistencies between fields). Statistical analysis should consider heterogeneity within the data set in terms of methods employed, quality of the data and other site-specific differences. If different methods of assessing drug exposure are used, analysis of outcomes should be stratified by method (e.g. ANC card, prescription, record linkage or self report), as some are more precise than others. The estimated sample size depends on the strength of the teratogenic effect, the background frequency of adverse pregnancy outcomes in the population of interest as well as the number expected to be exposed to the drugs under study, and the choice of comparator group (background rate, or contemporaneous internal comparator) [21]. Much of this information is unknown in the proposed settings for such a registry [22]. Therefore, sample size should be finalized after the potential for harm of a particular substance is characterized by an initial cohort of recruited women. The background incidence of congenital anomalies is established through a contemporaneous internal comparator group. Table 2 below describes two options for estimating sample size based on case/comparators ratio: 1 case/ 1 comparator and 1 case/ 4 comparators. Two options are also presented regarding the relative risk to be detected (i.e. 2 or 10). Three different incidences of risks are chosen in the comparator group. They correspond to a baseline incidence of major malformations of 1% or 5%, and an incidence of 0.1% (which is generally accepted incidence of the more common malformations) [21]. A continuity correction of 10% for the chi-squared test has been included in the estimation. Sample size estimations based on background incidence, case/comparator ratio and anticipated relative risk including continuity correction Based on the calculations provided in Table 2, it can be estimated that a cohort of at least 2515 exposed women per drug of interest in the 1st trimester and an equal number of comparison women would achieve a power of 80 % to detect a doubling of risk of major birth defects for any given drug of interest, assuming a background incidence of birth defects of 1% (excluding congenital anomalies, cardiovascular malformations and genetic disorders which may not be detected) and a significance level of P < 0.05. The sample size was calculation was calculated using WinPepi version 11.4 with continuity correction. The WHO Pregnancy Registry database has been developed using the free access software OpenClinica. The database has been designed to accommodate electronic and paper-based CRFs depending on the preferences of the contributing sites. Data entered on paper CRFs by health staff can be subsequently entered into an electronic database. Investigators have experience with direct data entry into the online OpenClinica database and with offline EPIdata databases, with periodic transfer of both into a common statistical package. The EPIdata programme is also an open access standalone database but does not have an audit trail. This can be provided to sites not requiring an audit-trail and agreeing to routinely download the data into the pooled Registry database. This multi-pronged approach for data capture is intended to facilitate data capture at sites where online connectivity and feasibility of electronic data capture approaches may be limited, while ensuring ease of electronic data capture where such infrastructure exists. Future steps envisaged are to ensure that CRFs are adapted for data capture using mobile phones and other hand held electronic devices. Depending on the approach adopted by the site, systems can be supported at the sites to ensure the integrity of the data capture and contributions to a pooled Registry database. Pre-defined searches, reports and analyses from the pooled Registry will be developed in collaboration with sites and countries in order to facilitate routine analysis of data by sites, countries and internationally. Pre-defined analyses for determination of drug effect will be agreed by contributing sites and investigators before publication. National coordinators and research sites agreeing to pool data into the database can be trained to import data from the database into statistical packages such as Stata and SPSS. The Registry is an opportunity to build trust and confidence in the communities it serves. ANC staff should inform communities that they will be monitoring safety of treatments during pregnancy. This will reduce the likelihood of rumors and allay fears. Through the automated online login system, pre-designed tables and reports can be generated to provide indicators of maternal and child health to the relevant authorities and interest groups on a regular basis. Similarly, routine progress reports by the registry investigators to the facility staff can ensure ongoing feedback.