Home Contact us Site Map

Cette page existe uniquement en anglais

Home > Activities > Committees >Reproductive Health (1996-99) > Seminar Report

Seminar on Innovative Approaches to the Assessment of Reproductive Health

Manila, Philippines, 24-27 September 1996
Organised by the IUSSP Committee of Reproductive Health


The IUSSP Committee on Reproductive Health planned a series of three conferences on methodological and critical issues confronting this field to be held over a two-year period. The first of these was co-sponsored by the university of the Philippines Population Institute and held in Manila, Philippines from 24 to 27 September 1996. The seminar was devoted to reproductive health measurement and sessions were organized to address methodological and conceptual challenges within three main areas of reproductive health: maternal mortality, obstetric and gynaecological morbidity and infertility.

The development of measurement tools is vital to determine levels and trends of different reproductive health conditions, for the identification of the various determinants that influence outcomes as well as for monitoring and evaluating the impact of programmes addressing these issues. The rationale for the conference was based on the need to review and discuss existing reproductive health indicators as well as to explore new measurement approaches that have been used recently in the field. In the opening ceremony, participants were welcomed by Aurora Perez, Director of the Population Institute of the university of the Philippines, Amaryllis Torres, Vice-Chancellor for Academic affairs, university of the Philippines, Carla Makhlouf Obermeyer and Axel Mundigo, co-chairs of the Committee on Reproductive Health and Oona Campbell, Seminar Scientific Co-ordinator.

The scientific programme of the Seminar consisted of the following sessions : Maternal mortality

The set of papers presented under this topic examined the substantial body of work that has accumulated on the measurement of maternal mortality, estimated by WHO to be approximately 500,000 deaths globally every year. Two overview papers presented in the first session examined basic issues inherent in maternal mortality estimation. Oona Campbell and Wendy Graham’s overview of measurement issues (Measuring Maternal Mortality: direct and indirect approaches) analyzed the relative advantages and limitations of various definitions, data sources and techniques used in direct and indirect measurement approaches of maternal mortality and discussed feasibility problems in the absence of solid pre-existing data infrastructures. A very serious limitation is the difficulty of using existing techniques to monitor or evaluate programme impact and attribute change to a specific intervention. Reasons such as the rarity of maternal deaths, problems of ascertainment and lack of cause-specific data, raise problems for assessing the impact of specific interventions.

Cynthia Stanton’s overview (Strategies for Model-Based Estimates of Maternal Mortality by Stanton, Hill, AbouZahr and Wardlaw) presented a recently developed model-based approach to maternal mortality, an approach which estimates the proportion maternal of deaths of females (PMDF) of reproductive age and is then applied to estimate the numbers of such deaths. Developed by uNICEF, WHO and Johns Hopkins university, the model differs from others in that it does not depend on female life expectancy at birth and therefore is not affected by variations in child mortality. In addition, it is able to provide national level estimates for nearly all countries, rather than single national estimates or regional aggregates.

Josefina Cabogan (Indirect Estimation of Maternal Mortality Ratios for 1990-1995 at the National, Regional, Provincial and City Levels in the Philippines) described how the subnational estimates of maternal mortality ratios were derived in the Philippines as part of the work of the Technical Working Group on Maternal and Child Mortality. The use of the Rashomon Technique, an innovative multi-perspective qualitative method developed by Terence Hull to understand the factors that lead to maternal deaths in rural Indonesia was described by Meiwita Iskandar (Reexamining the Witnesses of Maternal Death in Indonesia: was it simply because her time had come?). This policy-oriented technique, using in-depth interviews of several witnesses to community maternal deaths, deepens understanding of the multiple factors that determine exposure to risk and eventually affect outcomes.

using data from metropolitan Guatemala, Edgar Kestler (A Surveillance System in Maternal Mortality: the Central American experience) described an Active Surveillance System established to investigate maternal deaths. The use of multiple sources of information yielded a detailed analysis of each maternal death identified, and highlighted the inadequacy of existing vital registration systems. John Ballweg and Imelda Pagtolun-an (The Risk of Childbearing: a Philippines study of pregnancy complications and maternal mortality) examined the impact of maternal care on the incidence of pregnancy complications and on identifying factors associated with maternal death. The fact that many women experienced complications despite receiving prenatal care suggests that for many, such care is seen as a curative service rather than a preventive exercise.

By examining data from Morocco, Vincent De Brouwere, Ahmed Laabid and Wim Van Leberghe (An Alternative for the Maternal Mortality Ratio: the coverage of obstetric interventions need) presented a new indicator to yield information on health system effectiveness. Defining 'need' as the absolute maternal indication for obstetric intervention, rates can be calculated with numbers of mothers delivered with an intervention in response to such a need as the numerator and the total number of births as the denominator, for any given area. In the formula, 'unmet need' is the gap between the actual rate and a reference rate for major obstetrical interventions. The method gives a spatial mapping of geographical areas indicating where unmet needs are greatest. The indicator operationalises the concept of unmet need and could be a potential tool for resource allocation.

Micah Madzima (Community-based Case-control Studies on Maternal Mortality: risk factors and preventability of maternal death, by Mbizvo, Fawkus, Lindmark and Nystrom) examined the social, operational and reproductive factors contributing to maternal mortality in urban and rural Zimbabwe. Avoidable factors were identified in 90% of the rural deaths and 85% of the urban deaths, and were further classified into those occurring at the community level as compared to within various levels of the health care system.

These papers generated a number of questions around the utility of using estimates which are orders of magnitude to assess impact of programmes. Mortality has proven a difficult yardstick with which to measure the progress of maternal health programmes- because mortality is a comparatively rare event, large sample sizes and considerable resources are needed to yield findings of statistical confidence. In developing countries where the statistical infrastructure is weak it is even more difficult and impractical. There was general agreement that while national estimates of maternal mortality are important for advocacy purposes their utility for evaluating programmatic impact and attributing change is doubtful. New approaches are very much needed obstetric and gynaecologic morbidity

Studies measuring different aspects of reproductive morbidity are less numerous than those that have addressed maternal mortality. For this reason reproductive morbidity indicators are at an earlier stage of development and present substantial methodological challenges. The second grouping of presentations discussed both objective as well as subjective indicators of obstetric and gynaecologic morbidity. Marie-Helene Bouvier-Colle (Frequency and Characteristics of Obstetric Patients who Need Treatment in Intensive Care units) examined hospital records of obstetric patients in an attempt to understand why maternal mortality in France is relatively higher than in comparable countries. The hospital data yielded wide discrepancies both in terms of the prevalence of obstetric morbidities as well as the fatality rates for different conditions and highlighted problems inherent in defining morbidity. Similar conclusions about definition criteria were reached by Timothy Gandaho, who presented an argument for collecting information on severe maternal morbidity as an alternative to measuring mortality (The 'Near Misses': are life threatening complications practical indicators for safe motherhood programmes?, by Filippi, Gandaho, Ronsmans, Graham and Alihonou). While such an approach is conceptually attractive as a substitute for or as a complement to mortality estimates, the authors caution that there is considerable work needed for the development of definition criteria and survey instruments, both at the hospital and community levels. The concept of reproductive morbidity and its sub-categories (obstetric, gynaecologic and contraceptive) is still in flux, with fewer than a dozen community-based studies carried out on the subject to date. A framework of reproductive morbidity was presented by Judith Fortney (Reproductive Morbidity: a conceptual framework) with the aim of simplifying and clarifying issues of definition for each category. The presentation illustrated categories that are not clear-cut and drew attention to areas where more research is needed, such as gynaecological morbidity resulting from cultural practices. Rosario Cardenas (Are RTIs a PUBLIC Health problem?) analyzed vital statistics and epidemiologic information to estimate the impact of sexually transmitted diseases and some reproductive pathologies in Mexico. By examining the disease burden not only in terms of premature mortality and disability but also in terms of pressure on the system, consultations, hospital bed days and costs of treatment, Cardenas illustrated the PUBLIC health toll in a variety of ways.

The possibility of assessing morbidity using survey questionnaires was a key concern during the seminar. Anrudh Jain (Measuring Reproductive Morbidity Through a Sample Survey in Peru, by Jain, Stein, Arends-Keunning and Garate) raised the question whether an alternative approach for measuring reproductive morbidity can be accomplished by questioning women about past diagnoses received during previous hospitalizations and health centre visits. In the absence of laboratory testing, survey methodology could come to provide an alternative with a few key questions added to questionnaires commonly used in cross-sectional fertility and contraceptive use surveys. Methodological lessons from population-based morbidity research were the subject of several papers. Michael Koenig’s review of the Indian experience (undertaking Community-based Research on the Prevalence of gynaecological Morbidity: lessons from India) raised issues central to fieldwork of this type. Low participation rates in such studies are often a problem, making investment in rapport and confidence-building in the community essential for successful implementation. Equally important is the committment to resolve or facilitate resolution of study participants gynaecological problems which are diagnosed by the study.

Longitudinal community-based studies of pregnancy are rare, but may be used increasingly to measure programmatic impact. Elizabeth Goodburn (Methodological Lessons from a Study of Post-Partum Morbidity in Rural Bangladesh by Goodburn and Graham) reviewed possible sources of bias from one of the few longitudinal studies of maternal morbidity conducted to date. Loss of respondents because of migration, accuracy of recall over time and the sensitizing effect of the study itself upon participants’ reporting are all biases that can potentially arise from prospective studies, and must be taken into account for accurate interpretation of results. Loretta Brabin (Measuring the Reproductive Health Problems of Adolescent Girls) proposed that improving malaria control for teenage primiparae in malaria endemic regions can have a major impact on their reproductive health. In addition, regarding gynaecological morbidity, more data on STDs amongst adolescents is needed than is currently available.

Much discussion revolved around using self-reported/perceived morbidity as a measurement surrogate for case identification. Huda Zurayk (Measurement of Reproductive Morbidity : the usefulness of perceived/reported morbidity on reproductive tract infections, by Zurayk and Kabakian) argued the need for expanding the conceptual understanding of morbidity beyond the bio-medical perspective in order to reflect different 'meanings' or 'realities' that are distinguishable in measurement terms. Questionnaires asking women about symptoms of biomedical conditions have shown poor validity. They have however yielded very useful information on women’s understanding of ill-health and the constraints surrounding utilization behaviour. Her presentation stressed the importance of understanding women’s perceptions of ill-health within their socio-cultural context for the development of measurement strategies.

Isabella Danel, Rebeca Ponce de Leon and Patricio Lozado (Validation Study of Responses to the Maternal Morbidity Module of the 1994 Ecuadoran Reproductive Health Survey) compared obstetric morbidity reported by women with their hospital records for labour and delivery. While the proportion of agreement was good for pregnancy outcomes, it was low for less severe complications. On the other hand, the inadequacy of the medical record as a tool was an observation shared by many of the seminar participants, emphasizing its inadequacy as a gold standard. While medical records are particularly unreliable in documenting less severe complications there is still a need for further refinement and validation of questions on obstetric morbidity, worded in locally appropriate terminologies. Esi Amaoful (Validation of Self Reported Symptoms of Severe Obstetric Complications by Amoaful, Sloan, Arthur, Winikoff and Adjei) discussed using the management of major obstetric complications as a measure of impact of safe motherhood programmes in Ghana. The results of interviewing inpatients cast doubt on the validity of using self-reported morbidity to identify life threatening complications.

Issues related to using self-reported symptoms as surrogates for case-status identification and the role played by selection bias in affecting the magnitude of estimates of conventional measures of validity were also discussed by Carine Ronsmans (Studies validating women’s reports of reproductive ill health : how useful are they?). Her paper served as a reminder that the magnitude of the discrepancy between self-reports and medically diagnosed conditions depends to a large extent on the actual but unknown prevalence of disease. Questions used to estimate morbidity in surveys may have high diagnostic accuracy but a low predictive value. Without knowing the actual prevalence of a disease, it is impossible to determine how well a questionnaire will predict the prevalence of disease in unselected populations.

The acceptability of newly developed diagnostic technologies applied in uganda by Ronald Gray and Maria Wawer (Clinical/Laboratory Methods for the Diagnosis of Reproductive Morbidity inĘPopulation-Based Studies) has very interesting implications for identification of RTIs in population-based studies. In a community-based trial of STD control for AIDS, the compliance rates with provision of self-administered vaginal swabs in the home were over 90 percent, suggesting that this mode of sample provision is much more acceptable to women than undergoing pelvic examination.

During the discussion that followed these papers it was pointed out that distinguishing between reproductive and non-reproductive conditions can sometimes be difficult. Methodological questions were also raised regarding the use of case-control designs in studying severe morbidity. It was stressed that much thought and care should go into their design, to avoid confusing between the aetiology of a condition and the factors which may play a role in its severity. Questions were also raised regarding how to identify predictors of disease and understand determinants of fatality. Infertility

The final set of presentations revolved around an issue which is often neglected in discussions of reproductive health measurement, namely infertility. These papers addressed levels, risk factors and consequences measurement as well as infertility trends and differentials. Amadou Noumbissi (Multivariate Approach to Analyse Primary and Secondary Infertility: applications using DHS data from Cameroon) proposed data classification using survival techniques to identify primary and secondary infertility. Arvind Pandley and Samuel Otor’s findings (Infertility in Sub-Saharan Africa: a case of Sudan) suggest that early menarcheal women are more likely to incur reduced fertility and not a longer period of fecundity, as had previously been thought. Their data also indicate that while the relationship between primary and secondary infertility may be linear for low and moderate levels of primary infertility, it seems to be curvilinear where primary infertility is high as in parts of sub-Saharan Africa. Johanne Sundby (Some Challenges in Estimating the Burden of Infertility for the Incorporation into a Reproductive Health Approach: an example from the Gambia by Sundby, Mboge and Sonko) reviewed questions of definition and issues relating to a population-based estimate of sub-infertility. The paper documented coping mechanisms and treatment-seeking patterns which indicate that child fostering is a frequent solution to childlessness and that modern health care is often sought as the last resort after traditional care has failed.

Innovative approaches to examining the relationship between infertility and STD/HIV were also presented in this session. Infertility as a risk factor for HIV/STD infection was discussed by Ties Boerma (Infertility and its Association with Sexual Behaviour, STD and HIV Infection in Tanzania, by Boerma, urassa and Isingo). The paper also illustrated how infertility affects measurement of various reproductive health elements, like HIV and STD, in different ways. There is evidence that HIV infection has an effect on fertility, with fertility being lower among HIV positive women. Therefore, estimates of HIV prevalence collected through surveillance of antenatal women are likely to be underestimates, since subfertile women would not be included in the samples. Measurement of HIV prevalence in surveys may also be biased by underrepresentation of more mobile childless women. There is a need for longitudinal studies to disentangle the effect of infertility to the increased risk of HIV independent of early onset of intercourse and high risk sexual behaviour unrelated to the fertility status of the woman.

In closing, it was felt that the seminar had provided an opportunity for rich discussion and exchange in several areas: in the conceptualisation of definitions, conditions, alternative indicators, determinants and consequences of reproductive health including also an examination of different data sources and the scrutiny of different measurement methodologies. The question of how to best inform and evaluate policy was a continuing theme over the four days. There was general agreement that the search for a core of reproductive health indicators continues that can address and evaluate issues from these diverse perspectives.

Karima Khalil

IUSSP, 3-5 rue Nicolas, F-75980 Paris cedex 20, France
Tel  +33 1 56 06 21 73 - Fax +33 1 56 06 22 04 - contact us