Sie sind auf Seite 1von 17

REDUCING MATERNAL MORTALITY THEORIES, MODELS, AND PRACTICE

Ravi Pamnani MS&E 408

Pamnani

MS&E408

Prof J Pietzsch

2/17

Pamnani

MS&E408

All models are wrong, but some are useful. George E. P. Box ABSTRACT Of all health indicators, the maternal mortality ratio shows the greatest discrepancy between developed and developing countries. Current global health programs have not been able to meet international goals to reduce maternal mortality. Analytical models have been used to address the complex, multifactorial etiology of maternal death and injury. Although past models can elucidate the inter-relationships between factors affecting maternal health, there is no existing model which can help evaluators compare different programs prior to implementation. In this report, I propose a framework for a predictive model which can be used to quantitatively assess alternative maternal health strategies and therefore optimize programs prior to implementation.

3/17

Pamnani

MS&E408

BACKGROUND Every minute, one woman dies because of preventable complications during pregnancy and child birth nearly 15,000 deaths per day, or about 530,000 deaths per year. Additionally, almost 10 million women each year suffer injury, infection, or disease from pregnancy- or birth-related complications.1 Maternal mortality and morbidity is a worldwide epidemic, displaying the greatest discrepancy between developed and developing countries.2 Nearly 99% of deaths occur in the developing world.3 In subSaharan Africa, it is estimated that one in 22 women will die of complications from pregnancy and childbirth. In South Asia, one in 200 will die. Compare this with one in 7,300 in developed countries.4 The maternal mortality ratio (MMR) is measured by the number of maternal deaths per 100,000 live births in each country (Figure 1). Because of the stark differences in MMR, maternal health has evolved into a development measure used by many international agencies. Figure 1. Global Maternal Mortality Ratios Deaths per 100,000 live births, 2005.4

Beyond fulfilling humanitarian objectives, there are significant societal benefits for pursuing the improvement of maternal health in developing countries. The loss of the mother represents the loss of a productive member of society whose labor and activities are essential to families and communities. Additionally, studies show that the health of the mother considerably influences that of their children. Children without mothers die more frequently, are at a higher risk of malnourishment, and are less likely to obtain an education, seriously impacting their future productivity. 5,6 Therefore, policymakers need to prioritize public and private investment in maternal health as major part of an integrated development strategy. Many government programs, NGOs, and socially-minded for-profit enterprises have committed to tackling this growing issue. In 1987, the World Bank, in a partnership with the World Health Organization (WHO) and the United Nations Population Fund (UNFPA), launched the Safe Motherhood Initiative.7 In 2000, the UN reaffirmed its international commitment by setting a key Millennium Development Goal to reduce maternal mortality from its 1990 level by three-quarters in the year 2015.4 However, little progress has been made. In the areas where there is the largest concernSub-Saharan

4/17

Pamnani

MS&E408

Africa and South AsiaMMR has dropped by 2% and 20%, respectively, a far cry from the UN goal of 75%.4 Without significant acceleration in public health investments or new policies in the developing How should policymakers decide on how to achieve maternal health improvements? The root causes of maternal morbidity and mortality are multifactorial and include many clinical and socioeconomic aspects. Programs that address these aspects have high opportunity costs that require significant financial, time, and human investment. To mitigate these risks, programs have run as small pilot phases to demonstrate cost-effectiveness, prior to implementation at larger-scale scales. However, these pilot phases may be lengthy and too narrowly-focused. If one pilot phase is limited in its efficacy, program managers have little choice but to implement the mildly effective program or start over with a different approach. I propose a framework for a predictive, decision-making model that has the potential to evaluate optimal interventions. An effective model can help steer public health programs toward effective, site-specific interventions based on local epidemiological, socioeconomic, and geographical characteristics. In this report, I will examine the existing analytical models for maternal mortality in the developing world, and then propose a framework for the development of a predictive model to evaluate maternal health programs. OVERVIEW OF MATERNAL MORTALITY AND MORBIDITY Maternal death is defined by the WHO as a death of a woman while pregnant or within 42 days of termination of pregnancy, irrespective of the duration and site of the pregnancy, from any cause related to or aggravated by the pregnancy or its management but not from accidental or incidental causes.8 The selection of 42 days is historical, and is theorized to originate with specific Western religious and cultural beliefs, as opposed to a study on the timing of maternal deaths. With the ability of modern medicine to delay death following significant complications, the latest WHO guidelines have also introduced the concept of late maternal death up to one year following pregnancy.9 Maternal deaths occur throughout pregnancy, labor, childbirth, and the postpartum period. The majority of maternal deaths (50-71%) occurs postpartum. Within the postpartum period, 45% of maternal deaths occur within the first 24 hours, and more than 65% happen within the first week. Although most causes of postpartum death result from complications that resulted during childbirth, the postpartum period is too often neglected by caregivers, resulting in a large number of preventable deaths.10 Many different types of complications can lead to maternal death (Figure 2). Causes are categorized as direct or indirect. Indirect causes of maternal death make up 20% of the worldwide total and are typically pre-existing or concurrent diseases that are not related to pregnancy. These conditions can either complicate the pregnancy or can worsen as a result of the pregnancy. Malaria, anemia, HIV/AIDS, and cardiovascular disease are common indirect causes of maternal death. Direct causes make up the remaining 80% of worldwide maternal deaths; the top five of which are hemorrhage, infection, eclampsia, obstructed labor, and unsafe abortion.10

Figure 2. Causes of maternal death10,a

5/17

Pamnani

MS&E408

Total is more than 100% due to rounding

For each maternal death, at least 30 women suffer injuries or permanent disability due to childbirth.3 Although international organizations estimate 15% of pregnancies develop complications, many local studies suggest a higher rate of morbidity. In one prospective observational study from Maharashtra, India, the incidence of maternal morbidity requiring medical care was estimated at 56.2%, with 15.3% requiring emergency obstetric care.11 High rates of mental illness in pregnant women of the developing world have also been reported.12 Preexisting psychiatric disorders surface as depression, substance abuse, or attempts at suicide, especially in conjunction with unwanted pregnancies.10 Other indirect factors may contribute to longer-term morbidities, as well. For example, malnutrition can be exacerbated by pregnancy, leading to a cascade of detrimental health effects.9 Maternal mortality and morbidity are the results of a complex array of factors. Any model which attempts to provide a complete understanding of the issue will inevitably be complex and multifactorial. EXISTING ANALYTICAL MODELS The purpose of a maternal health model is to help researchers and policymakers approach the issue, understand how different causes and effects are interconnected, and design solutions to address the cause using a rigorous, scientific methodology. For a model to be useful, it must provide the users of the model some kind of insight into the system being analyzed, either through the visualization of the connections, or through the development of analytic relationships between connections. Furthermore, a model is most useful when it has predictive qualities; namely, a user can understand how modifying one area of the model will affect other areas. These predictive abilities are most powerful when the model can distinguish the relative impacts or weights of different factors. This is what distinguishes a qualitative modelone that demonstrates highlevel connections between factorsand a quantitative oneone that demonstrates the ability to weigh measures and predict outcomes. Additionally, using sensitivity analysis, a quantitative model can predict how well the program must perform in order to achieve the desired effect on maternal mortality. This will be illustrated in the PROPOSED MODEL section on page 10.

6/17

Pamnani

MS&E408

Two analytical models for maternal health have been discussed in the literature: the McCarthy/Maine Model and the Three Delays Model. The McCarthy/Maine Model In 1992, McCarthy and Maine presented a qualitative framework to determine the relationship between factors contributing to maternal mortality (Figure 3). The model was developed in response to an increase in maternal health programsaddressing a wide range of issues from womens status to emergency obstetric carewithout explicit or systematic consideration of the mechanism in which these various factors impacted maternal mortality. Although McCarthy and Maine acknowledged that other authors have attempted to understand the complete process that results in maternal mortality and morbidity, (including the Three Delays Model described in the next section), they felt none have been comprehensive or fully-developed. 13 Figure 3. The McCarthy/Maine Model: A detailed framework for analyzing the determinants of maternal mortality and morbidity.13

The McCarthy/Maine Model is purposefully broad in scope. The model is meant to show that any improvement in Distant Determinants, i.e., socioeconomic and cultural factors, must operate through

7/17

Pamnani

MS&E408

the closer Intermediate Factors, in order to impact maternal health outcomes.14 Specifically, McCarthy and Maine claim that all determinants of maternal mortality (and therefore, all efforts to reduce it) must operate through three intermediate factors: (1) the likelihood that a woman will become pregnant, (2) the likelihood that a pregnant woman will experience a serious complication of pregnancy or childbirth, and (3) the likelihood of an adverse outcome for women with complications.13 McCarthy and Maine admit the model is simplistic in nature and that the components can be divided into many subcomponents. However, they emphasize the importance of the model to provide structure for researchers and program planners to discuss and consider various new intervention programs. They also acknowledge that not all of the relationships are represented in this modelbut instead, only those most important to the discussion.14 The strength of the model is its focus on the big picture, allowing the planner to compare and contrast more distant factors from more intermediate ones. McCarthy and Maine argue that the consideration of interventions in the framework of their model compels the planner to specify the chain of events by which a program might reduce maternal mortality. Therefore, when comparing two different interventions, policymakers can use this model to see how they are related and how they will reduce maternal mortality. However, the model does not quantitatively predict how two interventions might perform against each other. Besides the qualities of distant versus intermediate, there are no other relative measures in the McCarthy/Maine Model against which to compare programs. The Three Delays Model The Three Delays Model is based upon the following two assumptions: (1) about 80% of maternal deaths result from direct obstetric causes (Figure 2) and (2) the majority of deaths can be prevented with timely medical treatment.10,15 The Three Delays Model employs a different view of maternal mortality by taking on the perspective of a woman who is experiencing an obstetric complication (Figure 4). The model states that once an obstetric complication occurs, the primary factor attributed to maternal deaths per the above assumptions is delay. These delays are separated into three distinct, chronological phases: 1. Phase I. The first phase of delay is in the individuals or familys decision to seek care. 2. Phase II. The second phase of delay is identifying and reaching an adequate health care facility. 3. Phase III. The third phase of delay is receiving adequate care at the facility. A break anywhere along the chain of these three phases can result in increased likelihood of maternal mortality and morbidity.16 The model also describes how the three phases are influenced by the following factors: socioeconomic and cultural factors, accessibility of facilities, and the quality of care.16 Socioeconomic and cultural factors, as well as perceived accessibility and perceived quality of care impact Phase I, which is the individuals or familys decision to seek care. This first step is critical, because unless the mother or family decides to seek care, the other phases of delay (identifying and reaching a facility, and receiving quality care once there) are irrelevant. Once a decision is made to seek care, the accessibility of the facilities (the actual distance, the available transportation options, and the costs of transportation) impacts Phase II, the likelihood of reaching the facility. Finally, if a woman is able to reach a facility, the quality of care (quality and availability of trained staff, the availability of equipment, blood, drugs, etc, and the competency of the physicians) impact Phase III, the likelihood of receiving adequate and appropriate treatment.

8/17

Pamnani

MS&E408

Figure 4. The Three Delays Model16

The Three Delays Model is narrower in scope than the McCarthy/Maine Model, as it focuses on the interval between the onset of an obstetric complication and its outcome, rather than the prevention of the obstetric complication itself. It also does not include consideration of other, non-emergent factors such as reproductive behavior (e.g., condom use, womens status) and health status (e.g., nutrition). Although experts debate the relative importance of socioeconomic and purely medical interventions, there is evidence that even among women who are well-nourished and well-educated, a large percentage undergo serious complications during childbirth.14 Therefore, a focused emphasis on the management and treatment of women with obstetric complications may lead to wider benefits for women across all socioeconomic levels. Like the McCarthy/Maine Model, the Three Delays Model does not allow for quantitative comparisons between alternative interventions. One advantage of the Three Delays Model is the framing of the problem in terms of the decision and experiences of the pregnant woman facing a potential complication. In this way, alternative interventions can be interpreted through the lens of the pregnant woman herself. For example, even if an NGO builds a new hospital in a village district (addressing accessibility and quality of care), if there exists a major Phase I delay based on cultural stigmatization of hospitals, then the presence of a new, advanced hospital may not reduce maternal mortality as intended. However, there may be some reduction, as Phase II and Phase III delay may be mitigated by the presence of the new hospital (depending on the cost of treatment, etc). Additionally, because the Three Delays Model is built upon the initial decision to seek care, the use of more sophisticated techniques (e.g., decision analysis) may be able to incorporate quantitative elements into the model, allowing for a more comprehensive understanding of how different programs might impact overall maternal mortality.

9/17

Pamnani

MS&E408

WHY USE DECISION ANALYSIS? To develop a predictive model for maternal health, I propose using decision analysis. The decision analysis technique combines logic and probability to assess alternative strategies and make decisions based on the best evidence available. Using Bayesian probabilities, decision analysis is capable of handling significant uncertainty, which is often the case in developing countries, where accurate data collection is challenging. Decision analysis also provides for an explicit, reproducible process that allows other program evaluators to navigate the logic behind the decision, thereby ensuring transparency. The technique requires a meticulous, methodological approach, structuring complex alternatives in a rational way, and avoiding inconsistencies and errors. Decision analysis allows for the incorporation of expertise and information from a variety of consultants without yielding the decision to any single person. Finally, as mentioned above, decision analysis can provide predictive insight into a problem, by incorporating existing data to project future outcomes.17 Researchers have advocated the use of decision analysis in public health for at least 20 years.18 For example, the techniques have been employed for disease screening, especially in problems where the patients are asymptomatic (e.g., cancer screening, glaucoma screening, hypertension screening, toxic substance exposure screening, etc).17 Decision analysis is particularly useful in modeling these scenarios because screening tests are rarely 100% sensitive or specific, and therefore may not be cost-effective (depending on the cost of the test and the prevalence of the disease) or clinically-effective (depending on any potential complications from the diagnostic test itself).19 Researchers have also used decision analysis to study the effectiveness of immunization practices (e.g., rubella, swine flu, pertussis, hepatitis) for the same reasons.17 In the developing world, researchers have employed decision analysis techniques to explore a wide array of topics, ranging from female sterilization in Bangladesh20 to reducing mother-to child transmission of HIV.21 An effective decision analysis-based model for maternal mortality would be able to help policymakers achieve their goals. The model must be able to answer questions such as: (1) What is the likelihood that a proposed program will achieve the communitys maternal health goals? (2) What is the minimum level of effectiveness that the program needs to reach to achieve these goals (for example, if the proposed program is a better way to transport women to the hospitalwhat percentage of women need to participate to reduce the MMR to a certain target level)? (3) How does this compare to other potential programs? PROPOSED MODEL To begin developing a decision analysis-based model, I have reconfigured the Three Delays Model as a decision diagram (Figure 5). I have selected the Three Delays Model as the starting point because it lends itself to a decision analysis interpretation due to the formulation of the maternal health problem as a decision by the mother/family. In this section, I will outline the underlying framework for the development of a broad, customizable maternal health model. The foundation of the model is the decision diagram, which maps the decision to be made, the factors influencing the decision, and the potential outcome. The decision diagram consists of nodes arranged in a network connected by arrows. The rectangular node indicates a decision, which in this case, is the decision made by the mother/family to seek medical care. An elliptical node indicates an uncertainty or probability distribution. The uncertainties in the diagram influence the decision to be made and other uncertainties. For example, the likelihood of reaching a medical facility influences the perceived accessibility of care, which in turn influences the decision of the mother/family to seek care. The

10/17

Pamnani

MS&E408

octagonal node represents a value, which in many decision analysis scenarios is a monetary value, but in the case of this public health decision diagram, is the MMR. Figure 5. Decision diagram for maternal mortality programs, based on the Three Delays Model
Accessibility of facilities Perceived accessibility of care

Reaching a medical facility

Cultural & Socioecono mic Factors

Seek medical care?

Clinical outcome

MMR

Perceived quality of Care Quality of care

Receiving quality care

For the purposes of the model, the nodes identified as uncertainties can be modeled as probability distributions as related to the target locality (the locality of interest to the program evaluator), or similar locality, if data for the target locality are not available. Note that if the programs under evaluation do not impact one or more of the uncertainties (for example, a program aimed at improving public transportation to the health facility does not necessarily impact cultural or socioeconomic factors), then those parts of the model can be ignored. This is discussed further in the Customizing and implementing the model section on page 14. Cultural and socioeconomic factors Cultural and socioeconomic factors influence the decision to seek medical care. In the Three Delays Model, these factors are broken down into the following four sub-factors16: 1. Illness factors. The way individuals in the target locality perceive the severity and etiology of the illness significantly influences health-seeking behavior. Individuals must also be able to recognize their illness, as some women may believe that important signs of complications are part of a normal pregnancy. This is also related to educational status. 2. Womens status. Constraints on womens independence and preferential treatment to males limit womens access to health services. 3. Economic status. Economic status correlates to utilization of health services. Higher illness prevalence was found in rural areas and in the low socioeconomic areas of cities. 4. Educational status. Higher educational status generally correlates to increased health-seeking behavior. This is sub-factor is tied heavily to womens status, as well. In order to model these factors as uncertainties, it is preferable to leverage existing research, either from the target locality or a similar region. For example, one study links socioeconomic status to average number of doctor visits: the higher the socioeconomic status, the more likely an individual is to visit the doctor.22 Therefore, as the model is developed, real-world data from the literature can be used to support probability assignment (e.g., an individual/family in X socioeconomic category is X likely to seek health care). If relevant data is not available, expert opinions can be substituted instead.

11/17

Pamnani

MS&E408

Accessibility of facilities As described in the Three Delays Model and mapped in the proposed decision diagram (Figure 5) , the accessibility of the facilities affects the model in two ways: (1) influencing the decision to seek care, and (2) the actual delay in reaching a facility and receiving adequate care. Multiple factors, in turn, influence access to facilities (Figure 6). Three sub-factorscost of travel, distance to the facility, and availability of transportimpact a womans ability to physically reach a medical facility. These factors can be further broken down, depending on the level of detail of the programs to be evaluated and the availability of the probability distributions for the relevant information. Determining or estimating the proportion of expecting mothers who actually reach a facility can yield a simplified probability distribution. For example, one rural study in Kenya indicated that only 36% of women who intended to deliver in a hospital actually did.23 Therefore, once the decision to seek health care has been made, a simplified model may conjecture that there is a 36% chance of reaching the hospital for the target locality. Though this example is over-simplified, it applies even in cases where more specific attributes are under evaluation (e.g., improving roads, lowering the costs of buses, expanding the bus route, providing specialized transportation, etc). Alternatively, expert opinion or other types of modelssuch as locationallocation models24may substitute for data, and is necessary when data is difficult to obtain or unreliable. Figure 6. Factors influencing accessibility of facilities
Cost of travel Distance Availability of transport

Reaching a medical facility

Influences Receiving

Perceived accessibility of care

Influences Seek

Quality of care After deciding to seek care and reaching a medical facility, the final delay is in receiving quality care itself. As with the accessibility of facilities, there are two dimensions to the quality of care per the Three Delays Model: (1) the perceived quality of care, which influences the decision to seek care, and (2) the actual quality of carelabeled as receiving quality carewhich influences the clinical outcome. There are a wide variety of contributing factors that affect the quality of care (Figure 7). The relative impacts of these factors are well-documented in the literature,16 and therefore provide substantial data for modeling purposes. In the case of perceived quality of care and its contributing factors (satisfaction with service and previous experience/reputation), the best way to model these factors may be through a probability distribution across the population, similar to the probability distributions mentioned in the previous section.

12/17

Pamnani

MS&E408

For Receiving quality care, contributing factors like the availability of staffing and availability of equipment are not true uncertainties, as the presence of staffing or equipment in a given facility is determinable. These characteristics can best be modeled in a binary fashion (1 or 0is it available or not). This method is applicable, even if the contributing factors are broken down further. For example, equipment can be broken down into: availability of gauze or availability of drugs, which can be represented in the model as either available (1) or not available (0). The uncertainty is captured later on in the model, when estimations on how the availability of staff or equipment impacts the outcome. The literature or expert opinion can provide supporting data for these portions of the model (Clinical outcome section). However, the adequacy of management is dependent on the training of the clinical staff, and may still be subject to uncertainty. A probability distribution may be useful, if the quality of the team is variable (for example, if one physician is available for every fifty nurses). The literature may suggest that even the most well-trained team may perform diagnostic mistakes or management mistakes. For example, the distinctions for the Adequacy of management node may state that there is a 5% chance of being treated by a physician, and a 95% chance of being treated by a non-physician. Then, based on these two distinctions, the evaluator can map the likelihood of clinical outcomes for each of these alternatives, which is captured in the next phase. Figure 7. Factors influencing the quality of care
Availability of staffing Availability of equipment Adequacy of managemen t

Satisfaction with service

Receiving quality care

Influences Clinical

Previous experience/ reputation

Perceived quality of care

Influences Seek

Clinical outcome The clinical outcome will result in at least three possibilities: (1) death (maternal mortality), or (2) survival, or (3) disability, which can be split up into an infinite array of categories, depending on the morbidities of interest to the evaluator. The clinical outcome node represents the uncertainty inherent in any medical treatment. Even with perfect accessibility and perfect quality of care, there may still be some uncertainty in the outcome, simply based on the variations in illness and recovery. Therefore, this uncertainty node captures the probability of the potential outcomes, given the prior information (reaching a facility within X time interval, obtaining care at X quality, etc). The clinical outcome measure is the maternal mortality rate, captured in a value node MMR, which is used because it is the public health metric of interest when comparing alternative maternal health strategies.

13/17

Pamnani

MS&E408

Customizing and implementing the model The basic framework for the model is outlined in Figure 5. As particular areas become relevant to a program evaluator, those elements can be fleshed out with additional uncertainties and influence diagrams. Using standard decision analysis techniques, the influence diagrams and uncertainties can be modeled into decision trees (or their mathematical equivalents), with source data from the literature or expert opinions. The model can then be used in the following two ways: 1. Assessing how much change a program must achieve to achieve an MMR goal. In the first round of setting up the model for a specific analysis, the evaluator will arrive at an estimated MMR result. Using sensitivity analysis, the evaluator can see how the MMR will be affected by modifying the inputs to the model. This may be particularly useful if the target MMR is outside of the estimated MMR achieved in the first-pass of the model. 2. Comparing two, completely different programs. One program is to obtain a five year supply of much-needed drugs, while a second program is to develop a long-term educational program on maternal health danger signs during pregnancy and delivery. By using a comprehensive and customizable model, the evaluator can elaborate on certain sections, using existing data from their locality. They walk through the model three timesfirst, using a base case; second, using program alternative #1; and third, using program alternative #2. By comparing the estimated MMR outcomes and costs, the evaluator can see if there is a clearly superior choice. As before, the evaluator can then perform sensitivity analyses to validate the robustness of the model, and to establish his or her confidence in the choice. SHORTCOMINGS If the model is developed further, there are a couple of important shortcomings which need to be addressed. Reliance upon existing data Using existing data provides the most real-world evidence to support the model. At the same time, there are significant limitations to such an approach. Because maternal mortality and morbidity are difficult to measure, reliable epidemiological information is scarce, particularly in low-income, developing countries. Registration systems in developing countries are often absent or inadequate.25 Data regarding maternal mortality and morbidity are mostly collected from hospital deliveries or womens self-reports through questionnaires.26 Sampling hospital deliveries introduces significant selection bias, as an estimated 63% of deliveries globally occur at home.27 Questionnaires may also be unreliable, particularly with morbidity information. Validation studies have shown disagreement between womens self-reports of obstetric complications and medical data.28 The unreliability of the information may be mitigated by considering a range of values (e.g., through a sensitivity analysis), but this reduces the confidence in the model, and as a result, its usefulness. Existing data may only exist on different localities, or different continents altogether. The relevance of a given data set in this case is questionable. Additionally, the data may be ten or fifteen years old. The unreliability of the data in these cases may be mitigated through extrapolation based on prior knowledge. For example, the evaluator should consider certain cultural, socioeconomic, or geographical differences that are known between the target locality and the origin of the source data set. Additionally, if certain trends in the population are well-understood, data from ten or fifteen years ago may be extrapolated and projected to have a relatively accurate estimation of the target localitys current maternal health status.

14/17

Pamnani

MS&E408

Practicality The model may be attempting to take on too much at once, and therefore may be impractical from a program evaluators point of view. If the model is too detailed, it may not be universal. If the model is too broad and general, it may require too much specialized knowledge for the program evaluator to implement in his or her local setting. The model must be scalable, and in its ideal form, exist as a software tool with a simple user interface. This will yield the highest potential for adoption. The model must maintain complete transparency, so that adept users can experiment and make modifications in the field, which would provide the most useful feedback in later iterations. FUTURE STEPS The next steps to continue developing this model to be a useful, predictive tool for program planners are: 1. Use aggregate data from the literature to build specific quantitative impacts of each uncertainty node. 2. Obtain expert opinion and conduct surveys to fill in any gaps, or validate any assumptions. 3. Validate the model by comparing the models predicted results to actual results of an intervention (an intervention not used in building the model itself). 4. Establish the model as an open source, software tool and disseminate widely. Set up a forum to obtain feedback for continuous improvement of the model. As increasing evidence becomes available from small, village-scale studies on the effectiveness of a variety of interventions, a single, integrated approach to compare program alternatives in areas where maternal mortality still remains a significant issue, could provide an effective public health policy tool for program planners. CONCLUSION The significance of maternal health cannot be exaggerated and is especially important in the developing world. The maternal mortality ratio is a powerful development metric to assess the overall status of a regions health care system since prevention and treatment of obstetric complications is dependent on so many aspects of society. Current programs to reduce the MMR to the UN Millennium Development Goal levels have failed. Existing analytical models to assess maternal health are descriptive, but cannot predict the efficacy of potential programs to succeed in a given region. A new framework, using decision analysis, combines the existing understanding of maternal health determinants with data weighing the degree to which these determinants interact. Once this quantitative, predictive model is refined with further research and field validation, it can help to optimize maternal health programs prior to implementation, as well as compare alternative programs when resources are scarce. With implementation of a predictive model based on the proposed framework, successful reduction of maternal mortality is achievable. WORKS CITED

15/17

Why do so many women still die in pregnancy or childbirth? Ask the expert. World Health Organization. <http://www.who.int/features/qa/12/en/index.html>. 2 Nestel P. Introduction. Am J Clin Nutr 2000;72(suppl):209S-11S. 3 The World Bank. Safe motherhood and The World Bank: lessons from ten years of experience. Washington, DC: The World Bank, June 1999. 4 The Millennium Development Goals Report 2008. United Nations Department of Economic and Social Affairs (DESA). August 2008. 5 Strong MA. The effects of adult mortality on infant and child mortality. Unpublished paper presented at the Committee on Population Workshop on the Consequences of Pregnancy, Maternal Morbidity and Mortality for Women, their Families, and Society, Washington, DC, 1920 October 1998. 6 Ainsworth M. The impact of adult deaths on the nutritional status of children. In: Coping with AIDS: the economic impact of adult mortality on the African household. Washington, DC, World Bank, 1998. 7 Safe Motherhood: Overview and Lessons Learned. RHO Archives. <http://www.rho.org/html/sm_overview>. Last updated: 2005. Accessed: 2008-12-01. 8 World Health Organization. ICD-10: International Statistical Classification of Diseases and Health Related Problems. Geneva: WHO, 1992. 9 Hj L. Maternal mortality: only 42 days? BJOG. Nov 2003. 110:995-1000. 10 The World Health Report 2005: Make every mother and child count. WHO, 2005. 11 Bang RA. Maternal morbidity during labour and the puerperium in rural homes and the need for medical attention: A prospective observational study in Gadchiroli, India. BJOG. Mar 2004. 111:231-238. 12 Cox J. Psychiatric morbidity and pregnancy: a controlled study of 263 semi-rural Ugandan women. B J Psych 1979;134:401-5 13 McCarthy J. A Framework for Analyzing the Determinants of Maternal Mortality: Implications for Research and Programs. Studies in Family Planning. 1992. 12(1):23-33. 14 Maine, D. Safe Motherhood Programs: Options and Issues. Prevention of Maternal Mortality, Center for Population and Family Health. 1991 (Second printing: 1993). 15 World Health Organization. Prevention of Maternal Mortality: Report of a World Health Organization Interregional Meeting. 11-15 Nov 1985. Geneva, 1986. 16 Thaddeus S. Too far to walk: maternal mortality in context. Soc. Sci. Med. 1994. 8: 1091-1110. 17 McNeil S. Decision analysis for public health: principles and illustrations. Ann Rev Public Health. 1984. 5:135-161 18 Alemi F. Decision analysis in health administration programs: an experiment. J Health Adm Educ. Winter;4(1):45-61. 19 Berwick D. Cost-effectiveness of lead screening. 1982. N Engl J Med. 306:1392-98. 20 Rosenberg M. Decision analysis for assessing the impact of female sterilization in Bangladesh. Stud Fam Plan. Feb 1982. 13(2):59-63. 21 Bertolli J. Decision analysis to guide choice of interventions to reduce mother-to-child transmission of HIV. AIDS 2003, 17:2089-2098. 22 Kloos H. Illness and health behaviour in Addis Ababa and rural central Ethiopia. Soc Sci Med, 1987. 25(9):1003-1019. 23 Voorhoeve A. Modern and traditional antenatal and delivery care. Maternal and Child Heatlh in Rural Kenya, p. 309. Croom Helm, London, 1984. 24 Rahman S. Use of location-allocation models in health service development planning in developing nations. Eur J of Op Res, 2000. 123:437-452. 25 AbouZahr C. Maternal mortality at the end of a decade: signs of progress? Bulletin of the World Health Organization, 2001. 79(6): 561-73.
26

Fortney JA. Measuring maternal morbidity. In: Berer M, Sundari TK, editors. Safe Motherhood Initiatives: Critical Issues, Reproductive Health Matters. Oxford: Blackwell Science, 1999:43-50.
27

WHO. Essential Newborn Care: Report of a Technical Working Group 1994. Geneva: WHO, 1996.

28

Filippi V. Womens reports of severe (near-miss) obstetric complications in Benin. Stud Fam Plann 2000;31(4):309 324.

Das könnte Ihnen auch gefallen