Estimation of self-funded human papillomavirus vaccine recipients from Japan's previously assumed "unvaccinated generation".

Precise vaccination data is essential to accurately estimate the effectiveness of the human papillomavirus (HPV) vaccine against HPV-related cancers. In Japan, the number of subsidized HPV vaccinations can be tracked through registries, but the number of self-funded vaccinations has not been tracked. The number of individuals who chose to receive the vaccine at their own expense, despite being ineligible for public subsidies due to their age, is unknown and has been nominally considered to be zero. Our aim is to produce a more accurate estimate of this number using recently released proprietary data. First, we estimated the total number of self-funded HPV vaccinations occurring from 2010 to 2012 using public data from the Ministry of Health, Labour and Welfare and our previously reported data on the number of HPV vaccinations eligible for public subsidy. Second, using proprietary data from the vaccine manufacturer, we calculated the distribution of self-funded vaccination shots by age. Finally, we combined these data to estimate the number of self-funded HPV vaccinations by birth fiscal year (FY) relative to a yearly reference population. We found that 78,264 individuals born in FY1993 and 58,190 born in FY1992 self-funded their vaccinations, representing 13.6% and 10.0% of the reference population, respectively. Additionally, we found that 5%-10% of individuals born from FY1986 to FY1991 self-funded their vaccinations. Our study revealed for the first time that a certain number of individuals from the "HPV unvaccinated generation," ineligible for subsidies due to age restrictions, chose to self-fund their vaccinations.

Prediction of deciduous teeth eruption in Brazilian children: A cross-sectional study nested in a prospective birth cohort (BRISA).

BACKGROUND: Dental eruption is part of a set of children´s somatic growth phenomena. The worldwide accepted human dental eruption chronology is still based on a small sample of European children. However, evidence points to some population variations with the eruption at least two months later in low-income countries, and local standards may be useful. So, this study aimed to predict deciduous teeth eruption from 12 months of age in a Brazilian infant population. METHODS: We developed a cross-sectional study nested in four prospective cohorts - the Brazilian Ribeirão Preto and São Luís Cohort Study (BRISA) - in a sample of 3,733 children aged 12 to 36 months old, corrected by gestational age. We made a reference curve with the number of teeth erupted by age using the Generalized Additive Models for location, scale, and shape (GAMLSS) technique. The explanatory variable was the corrected children´s age. The dependent variable was the number of erupted teeth, by gender, evaluated according to some different outcome distributional forms. The generalized Akaike information criterion (GAIC) and the model residuals were used as the model selection criterion. RESULTS: The Box-Cox Power Exponential method was the GAMLSS model with better-fit indexes. Our estimation curve was able to predict the number of erupted deciduous teeth by age, similar to the real values, in addition to describing the evolution of children's development, with comparative patterns. There was no difference in the mean number of erupted teeth between the sexes. According to the reference curve, at 12 months old, 25% of children had four erupted teeth or less, while 75% had seven or fewer and 95% had 11 or fewer. At 24 months old, 5% had less than 12, and 75% had 18 or more. At 36 months old, around 50% of the population had deciduous dentition completed (20 teeth). CONCLUSION: The adjusted age was an important predictor of the number of erupted deciduous teeth. This outcome can be a variable incorporated into children's growth and development curves, such as weight and height curves for age to help dentists and physicians in the monitoring the children's health.

Implications of population growth on prevalence of diabetes. A look at the future.

The purpose of this study was to present projections of the future population of diabetes patients, to discuss policy implications of these projections, and to suggest ways that these projections might be made more useful to medical professionals. Under the assumption that the incidence of diabetes in four age-groups will remain constant in future years, previous estimates of the incidence of diabetes will be applied to Bureau of the Census population projections to project the number of new cases of diabetes that can be expected in future years in each of these age-groups. The prevalence of diabetes will remain relatively constant at approximately 1 million patients in younger populations (less than 45 yr old) through the middle of the next century. As the post-World War II baby boom ages, the number of older diabetes patients (45 and older) will almost double from 6.5 million in 1987 to an estimated 11.6 million in the year 2030. Although there is little doubt that the aging of the population will increase the number of diabetes patients, the assumption of constant incidence rates is a very limiting one. These projections would be more useful for the planning of research and training if the incidence of diabetes could be estimated for more refined categories of demographic and medical characteristics.

PIP: A 1987 US study cited the prevalence of diabetes as 26.8 patients/1000 population. Using the Census Bureau's population projections of these same age groups until the year 2050 based on their middle mortality assumptions, projections of the number of diabetes patients by age-group are obtained by multiplying with the prevalence rates. These projections indicate that the number of diabetes patients 25 years of age will remain almost constant in the next 1/2 century, whereas the number between 25 and 45 will decline from a high of 983,000 in 1995 to a low of 870,000 in 2040. The number of patients in the 45-64 age group is projected to rise from 2.4 million in 1990 to 4.1 million in 2015, i.e., about 69,000 additional patients/year for the next 25 years. Those aged or= 65 years with the highest prevalence are projected to increase by an average of 55,000 new patients/year. For the following period of 2015-35 this oldest group of patients is projected to increase by an average of 120,000 patients/year. This is under the overly pessimistic assumption that there will be no scientific or medical discoveries to reduce the prevalence of diabetes. There are some implications for the diabetes community if the population of patients increases by over 1 million in the next decade and by over 3.7 million by 2020. Although the number of diabetes patients 65 will grow steadily for the next 20 years, the highest growth rates will be in the 45-64 age group. This implies that the number of qualified professionals may have to be increased and more emphasis directed to initial diagnosis and treatment. After the year 2010, the rate of growth of diabetes patients age 65 or older can be expected to accelerate compared with other age groups. These projections stress the importance of prevention and education. The requisite change in life style, exercise, or nutrition habits will be more difficult than if a drug is developed for treatment.

Preparation for phase III HIV vaccine efficacy trials: methods for the determination of HIV incidence.

OBJECTIVE: Accurate estimates of HIV incidence that reflect the effect of non-vaccine interventions (education, counselling, condom promotion, and possibly sexually transmitted disease treatment) and that may be provided in a Phase III vaccine efficacy trial, are needed so that vaccine trial population sample sizes can be accurately determined. In order to avoid delays in the implementation of efficacy trials, well characterized cohorts must also be developed and available to participate in such trials. We reviewed the potential study populations, the epidemiologic methods for the determination of HIV incidence (using open cohort, closed cohort, and seroprevalence data methods), and the need for the development of population cohorts in preparation for Phase III HIV vaccine efficacy trials. SETTING: Phase III trials in developed and developing countries. METHODS: Comparison of open and closed cohorts and those using seroprevalence data to estimate HIV incidence. RESULTS: Open and closed cohorts each have disadvantages and advantages. However, the open cohort may be more suitable for determining estimates of HIV incidence that reflect non-vaccine interventions and for the development of a well characterized cohort available to participate in efficacy trials. CONCLUSION: Careful preparation of research infrastructures and population cohorts will help ensure the successful conduct of scientifically and ethically sound HIV vaccine efficacy trials in the future.

Projecting the impact of AIDS on mortality.

OBJECTIVE: To illustrate the magnitude of the impact of AIDS on projections of mortality, to explain the reasons for the differences in projections by major international organizations and to provide a simple approach to estimating the impact of AIDS on life expectancy. RESULTS: AIDS has already had significant impacts on mortality in a number of countries in the developing world and this impact is expected to grow substantially in the next 10 years. By 2005 the population of the most severely affected countries in Africa will be 13-59 million less than it would have been without AIDS. Life expectancy may decline by as much as 27% in these countries. Country specific projections made by the United Nations (UN) and the US Census Bureau differ significantly in their estimates of the impact of AIDS. The UN projects that AIDS will reduce the population growth of the most severely affected countries in Africa by 13 million by 2005 and 30 million by 2025, while the US Census Bureau projects the reduction to be four times larger (59 million by 2005 and 120 million by 2025). These differences are due largely to the use of different methodologies for projecting future levels of adult HIV prevalence. Other factors contributing to the different projections are different estimates of current levels of HIV prevalence and different assumptions about the length of the incubation period (from initial infection until death from AIDS) and the perinatal transmission rate. In addition to the number of deaths caused by AIDS, useful indicators of mortality include life expectancy at birth, the under five mortality rate and the life-time risk of dying from AIDS. An equation for estimating the impact of AIDS on life expectancy is presented and its use is illustrated. CONCLUSION: It is clear that AIDS has already increased mortality significantly in many countries and will continue to do so in the coming decades. Uncertainty about current and future levels of HIV prevalence among adults leads to differences in the projections of future AIDS-related mortality. As data and projection methodologies improve, the differences in projections may be reduced for sub-Saharan Africa, but the growing epidemic in some of the largest countries of Asia may increase uncertainty about future global impacts.

PIP: The authors describe the projected impact of AIDS on mortality, explain the reasons for the differences in mortality projections by major international organizations, and offer an approach to estimating the impact of AIDS upon life expectancy. Although AIDS has already significantly increased mortality in many countries, the impact of AIDS upon mortality is expected to grow substantially over the next 10 years and into future decades. By 2005, the population of the most severely affected countries in Africa will be 13-59 million less than it would have been without AIDS, and life expectancy may decline by as much as 27% in those countries. Uncertainty about current and future trends in HIV prevalence among adults has led to differences in the projections of future AIDS-related mortality. For example, while the UN projects that AIDS will reduce the population growth of the most severely affected countries in Africa by 13 million by 2005 and 30 million by 2025, the US Census Bureau projects the reduction to be 59 million by 2005 and 120 million by 2025. These differences are largely due to the use of different methodologies for projecting future levels of adult HIV prevalence. There are also differing estimates of current levels of HIV prevalence and differing assumptions about the length of the incubation period and the perinatal transmission rate. As data and projection methodologies improve, the differences in projections may be reduced for sub-Saharan Africa, while the growing epidemic in some of the largest countries of Asia may increase uncertainty about future global impacts.

Demographic approaches to the estimation of incidence of HIV-1 infection among adults from age-specific prevalence data in stable endemic conditions.

OBJECTIVE: To develop methods for estimating the incidence of HIV-1 infection among adults from age-specific prevalence data derived in stable endemic conditions. METHODS: Two methods are proposed. The first method is the Cumulative Incidence and Survival Method which treats HIV-1 prevalence at any given age as the cumulative incidence of new infections at each preceding age, adjusted for mortality. A model for age-specific incidence is fitted to the data using maximum likelihood techniques. The other method is the Constant Prevalence Method whereby the incidence of new infections within a time interval (t-r, t) is calculated as the difference, after adjusting for mortality, between observed prevalence levels at two successive age intervals, whose mean ages are r years apart. The two methods were applied to data from Kampala, Uganda. RESULTS: Plausible estimates of age-specific and cumulative HIV-1 incidence were obtained from each of the methods. Estimates of HIV-1 incidence are sensitive to assumptions regarding the length of the survival period after infection and the stability of the epidemic. CONCLUSIONS: Reasonable estimates of HIV-1 incidence can be obtained from prevalence data derived in near-stable conditions. With the Constant Prevalence Method, these conditions may be relaxed if large sample sizes are available and age-reporting is good. The methods proposed could be used in the design and implementation of HIV-1 prevention trials. Cumulative incidence is a better indication of demographic impact than average age-specific incidence.

PIP: Two methods to estimate the incidence of HIV-1 infection among adults from age-specific prevalence data derived in stable endemic conditions are presented and tested on data from Kampala, Uganda. The Cumulative Incidence and Survival Method is first proposed which treats HIV-1 prevalence at any given age as the cumulative incidence of new infections at each preceding age, adjusted for mortality. A model for age-specific incidence is fitted to the data using maximum likelihood techniques. The Constant Prevalence Method is then presented in which the incidence of new infections within a given time interval is calculated as the difference, after adjusting for mortality, between observed prevalence levels at two successive age intervals, whose mean ages are a specified number of years apart. Plausible estimates of age-specific and cumulative HIV-1 incidence were obtained from each method.

Estimating the seroincidence of HIV-1 in the general adult population in Kigali, Rwanda.

OBJECTIVE: To estimate the seroincidence of HIV-1 infection in the general adult population of Kigali, Rwanda. DESIGN: Repeated standardized cross-sectional studies. SETTING: Two urban prenatal clinics. PATIENTS: A total of 4486 consecutive pregnant women consulting in 1989 and 1990. MAIN OUTCOME MEASURES: Prevalence of HIV-1 antibodies. RESULTS: HIV seropositivity increased by 3-5% annually over this period, indicating that it has not reached a plateau in this sentinel population. The percentage infection rates, calculated using two complementary methods, were 26.2-30.7% in 1990. Extrapolating these results to the general population of Kigali, we estimate that 2300-3800 new infections in young women and 3600-6100 new infections in young men occur annually among the total population of 350,000 in Kigali. CONCLUSIONS: A new HIV infection occurred in an adult every 50-90 min, on average, in Kigali during 1989-1990, while every 6-7 h a baby with maternally acquired HIV infection was born. Our HIV surveillance system, which is based on prenatal sentinel posts, is a useful tool for monitoring the progression of the HIV epidemic in Kigali.

PIP: This study aimed to estimate the seroincidence of HIV-1 infection in the general adult population of Kigali, Rwanda, by examining a total of 4486 consecutive pregnant women consulting in 1989 and 1990 at 2 urban prenatal clinics. Via repeated standardized cross-sectional studies the authors aimed to measure prevalence of HIV-1 antibodies. HIV seropositivity increased by 3-5% annually over this sentinel population. The percentage infection rates, calculated using 2 complementary methods, were 26.2-30.7 in 1990. Extrapolating these results to the general population of Kigali, the authors estimate that 2300-3800 new infections in young women and 3600-6100 infections in young men occur annually among the total population of 350,000 in Kigali. A new HIV infection occurred in an adult every 50 through 90 minutes, on average, in Kigali, during 1989 and 1990, while every 6 to 7 hours a baby with maternally acquired HIV infection was born. This HIV surveillance system, which is based on prenatal sentinel posts, is a useful tool for monitoring the progression of the HIV epidemic in Kigali.

Estimating the incubation period of paediatric AIDS in Rwanda.

OBJECTIVE: To estimate the distribution of the incubation period of paediatric AIDS in Rwanda. DESIGN: Data were collected between February 1984 and December 1990 at the Centre Hospitalier de Kigali (CHK), the capital city of Rwanda, Central Africa. PATIENTS: We used a sample of 685 AIDS cases registered consecutively in the Department of Paediatrics of the CHK, in which the proportion of perinatally acquired HIV-1 infection was estimated to be 98.6%. METHODS: We performed both non-parametric and parametric analyses. The methods of estimation were adapted to truncated data, using essentially the same methods as Auger et al. in their analysis of data from the New York City and the New York State AIDS case registries in 1988. RESULTS: We found that a double Weibull model fitted the data very well and that the risk of developing AIDS was high for subjects under 18 months of age, but lower for older subjects. CONCLUSIONS: Our results were qualitatively similar to those of Auger et al.. There were quantitative differences between the two studies, but it was not possible to compare median survival periods. Parameters such as median or mean survival times cannot be validly estimated using only data from registers because these data exclude infected subjects who have not yet developed AIDS.

PIP: The authors used nonparametric and parametric methods and data on 685 AIDS cases at the Centre Hospitalier de Kigali, collected between February 1984 and December 1990, to estimate the distribution of pediatric AIDS in Kigali, Rwanda. 98.6% of the cases probably acquired AIDS via vertical transmission. A combination of the 2 Weibul distributions (parametric method) resulted in a good fit, suggesting that the sample population consisted of a subpopulation with a short incubation period and an other with a longer incubation period. The researchers could not deduce proof of heterogeneity from the shape of the distribution, however. The probability of developing AIDS during the first year of life was 0.29, which corresponded with that of the European Collaborative Study (0.26). The risk of developing pediatric AIDS increased considerably for children less than 18 months old but fell and became constant for older children. The qualitative findings matched those of a study in New York City. Even though quantitative differences between this study and the other study existed, the researchers could not compare median survival times. Since data from registers did not include HIV-infected children who had not yet developed AIDS, the researchers were not able to estimate median and mean survival periods. A possible source of bias was that the data were from a surveillance system based on cases at just 1 hospital, which probably did not see all pediatric AIDS cases. In conclusion, truncated data determined rather well the distribution of incubation periods, but could not provide much information about the scale parameters of the model.

Transition dynamics of HIV disease in a cohort of African prostitutes: a Markov model approach.

The progression of HIV-related disease from infection to death is represented as a staged Markov model. Transitions between stages are considered reversible. The model is fitted to data from a cohort of African prostitutes by means of maximum likelihood. It appears that the progression to symptomatic disease (Centers for Disease Control stage IV) in this population is considerably more rapid than that reported from studies in Western countries.

PIP: Identifying the incubation period of HIV infection is important for individual prognoses, for developing and testing intervention strategies, for determining the reproductive rate of the disease, and for prevalence of the disease. Mathematical modeling of HIV infection in Africa is necessitated because the disease is more widespread and the immune system is constantly active due to the exposure to diseases such as malaria and tuberculosis. The Markov model for this analysis was selected because parametric estimation is not based on the time a stage is entered, but on the duration between observations and the stages at the time of observation. The HIV infected female prostitutes in the Pumwani area of Nairobi, Kenya (a population primarily of Tanzanian origin) have been identified as a study population since 1985, and seen every 6 months in clinic, or as needed. Data are constricted by the movement out of the area in the end stage of disease, which is only partially solved by tracking with community health workers. The stages identified in incubation estimation are stage 1: seropositive but symptom free (CDC stage II); stage 2: generalized lymphadenopathy (CDC stage III); stage 3: symptomatic disease (CDC stage IV); and stage 4: death. Data reflect the movement back and forth between stage 1 and 2, between 2 and 3, so the model is not a pure Longini model but rather a timed homogeneous staged model with reversible stages called transition parameters computed in a numerical differentiation. The Fortran computer program for the analyses is available from the authors. The results suggest a quick transition between seroconversion and lymphadenopathy (2.4 months) and unlikely reversal, with the mean waiting time until passage to stage 3 is approximately 2.6 years and conversions are common. Since opportunistic infections are treatable, this makes sense. Assuming a correct model, the estimation of the transition time of 20 months of h34 value of .01 and .05, the mean passage time from stage 1, 2, 3 to 4 (death) is 9.1, 8.9, and 6.2 years 12.9, 12.7, and 10.1 years respectively. The implications are that 1) when infectiousness is hypothesized to be not uniform, peak infectivity occurs earlier in Africa than in the West at least among prostitutes, or 2) if infectivity is constant throughout the incubation period, then HIV transmission must be higher in Africa to explain the high rate of infection.

Preventing the heterosexual spread of AIDS: what is the best advice if compliance is taken into account?

It has been recently advocated that avoiding partners who may be at high risk of carrying HIV provides 5000-fold better protection against HIV infection than usage of condoms [1]. In this paper, it is demonstrated that this guideline is largely based on unrealistic assumptions. If the sensitivity of identifying high-risk partners, realistic estimates of the efficiency of mechanical and chemical barrier methods, and the compliance in following either strategy are taken into account, use of condoms and/or suppositories containing nonoxynol-9 might be more effective than the attempt to avoid high-risk partners. Thus, both barrier methods should be strongly recommended for casual sexual heterosexual contacts.

PIP: Generalizing the results of Fienberg, the effects of optimistic assumptions on prevalence and infectivity and pessimistic assumptions about the effectiveness of barrier methods on estimates of the risk of HIV infection are demonstrated. Using the Pearl index as a basis for estimating the efficiency of a barrier method, it is shown that condoms -- preferably used together with spermicides -- and suppositories containing the spermicide nonoxynol-9 should be recommended for casual contacts among heterosexuals, The risk of HIV infection depends on the prevalence of HIV and its infectivity, the compliance of the individual to the strategy, i.e., the probability that the guidelines are followed, and the effectiveness of the strategy. It is misleading to conclude that "choosing a partner who is not in any high-risk group provides almost 4 orders of magnitude (5000-fold) of protection, compared with choosing a partner who is in the highest-risk category (with a prevalence of 5%.) With realistic assumptions, the protection cannot be in excess of 230-fold -- prevalence among randomly selected partners divided by prevalence among low-risk partners. Further, one cannot ignore the sensitivity of identifying high-risk partners. Finally, the influence of emotions or alcohol may even lead to sexual encounters with identified high-risk partners. Avoiding high risk partners with a compliance of 90% cannot provide more than 5-fold protection. For a given sensitivity and compliance rate, the use of barrier methods provides more protection than choice of low-risk partners. Due to the fact that barrier methods also provide protection against other sexually transmitted diseases which might be cofactors for HIV infection, the actual protective effect of these methods most likely will be even higher.

Comparing standardized mortality ratios.

It is well known that the ratio of two standardized mortality ratios (SMRs) is not in general an unbiased estimate of the underlying within-stratum ratio of rates of one subcohort relative to another. It is also established, although less well known, that a sufficient condition for unbiasedness is that the underlying stratum-specific rates in each of the two subcohorts be proportional to the reference population. Further, the ratio of SMRs is more precise than the wholly internal (Poisson regression) estimate of rate ratio. In data that are compatible with the proportionality assumption, use of the ratio of SMRs thus buys precision at the cost of increased vulnerability to bias. To further elucidate choice between methods, we derive expressions for the asymptotic precision of each. These show that improved precision of ratio of SMRs depends on the extent to which the distribution of expected deaths over strata is different in the two cohorts, or equivalently, on the variance over strata of the proportion of expected deaths falling in the first cohort. The results are illustrated by hypothetic examples.

Applying the sisterhood method for estimating maternal mortality to a health facility-based sample: a comparison with results from a household-based sample.

BACKGROUND: The sisterhood method is an indirect technique used to estimate maternal mortality in developing countries, where maternal deaths are often poorly registered in official statistics. It has been used successfully in many community-based household surveys. Because such surveys can be costly, this study investigated the suitability of using data collected in outpatient health facilities. METHODS: Adults visiting any one of 91 health centres or posts in a rural region of Nicaragua were randomly sampled and interviewed by health personnel. A sample size, proportional to the population served, was assigned to each facility and 9232 adults were interviewed. Characteristics of health facility users were compared with the general population to identify factors that would allow generalization of results to other settings. RESULTS: Based on these data, the lifetime risk of maternal death was 0.0144 (1 in 69). This estimate is essentially identical to that from a household-based survey in the same region 8 months earlier, which obtained a lifetime risk of 0.0145 (1 in 69). These findings correspond to a maternal mortality ratio of 241 and 243/100000 livebirths, respectively. CONCLUSIONS: This is the first report comparing results of the sisterhood method from household and health facility-based samples. The sisterhood method provided a robust estimate of the magnitude of maternal mortality. Results from the opportunistic health facility-based sample were virtually identical to results from the household-based study. Guidelines need to be developed for applying this low-cost and efficient aproach to estimating maternal mortality in suitable opportunistic settings at subnational levels.

PIP: Researchers compared maternal mortality estimates using the sisterhood method in a household survey conducted in November 1991 and in an outpatient health facility survey conducted in July 1992. Both surveys were conducted in Region I, a predominantly rural, mountainous area in northern Nicaragua. They analyzed data from 9232 interviews with adults younger than 49. The estimated lifetime risk of maternal death and the corresponding maternal mortality ratio were essentially identical for both the household and health facility surveys (0.145 and 0.144 [i.e., 1 in 69 of reproductive age died due to pregnancy-related events] and 243 and 241/100,000 live births, respectively). The estimates were similar for both surveys, even when the results were standardized for age, residence, and socioeconomic characteristics. An important limitation to the sisterhood method of estimating maternal mortality is that it estimates maternal mortality for a period about 10-12 years before the study and therefore cannot be used to assess the immediate effect of interventions to reduce maternal mortality. Nevertheless, in areas with poor maternal mortality surveillance or where no alternative exists to collecting population-based data, the sisterhood method can reliably estimate maternal mortality. These findings suggest that health facilities-based studies using the sisterhood method is a feasible, low-cost, and efficient method to estimate maternal mortality in certain settings at subnational levels.

Maternal mortality in developing countries.

PIP: The authors of this letter respond to earlier letters prepared in response to their article on maternal mortality in developing countries. It is conceded that maternal mortality is high in India and Bangladesh; however, statistics from Gambia are based on small populations and are therefore inconclusive. It is noted that a 7-year survey of 4000 households in Machakos, Kenya, where 73% of deliveries occurred at home, yielded a maternal mortality rate of only 0.8/1000 deliveries. Finally, it is asserted that the measurement traditionally used in estimating maternal mortality for many African countries (ratio of recorded maternal deaths to recorded deliveries) is misleading. Maternal deaths are more likely than deliveries to be recorded. In Niger, the number of maternal deaths increased from 1980 (374) to 1982 (484). The ratio of maternal deaths to expected live births also increased from 135 to 166/100,000, whereas the traditionally calculated maternal mortality rate decreased from 519 to 420/100,000 due to changes in the denominators. It is recommended that health authorities of African countries such as Niger consider setting an absolute number of maternal deaths below which they would try to bring the current toll.^ieng

Child and maternal mortality during a period of conflict in Beira City, Mozambique.

BACKGROUND: Child mortality rates have been declining in most developing countries. We studied child and maternal mortality risk factors for child mortality in Beira city in July 1993, after a decade of conflict in Mozambique. METHODS: A community-based cluster sample survey of 4609 women of childbearing age was conducted. Indirect techniques were used to estimate child mortality ('children ever born' method and Preceding Birth Techniques (PBT) and maternal mortality (sisterhood method). Deaths among the most recent born child, born since July 1990, were classified as cases (n = 106), and two controls, matched by age and cluster, were selected per case. RESULTS: Indirect estimates of the probability of dying from birth to age 5 (deaths before age 5 years, (5)q(0) per 1000) decreased from 246 in 1977/8 to 212 in 1988/9. The PBT estimate of 1990/91 was 154 (95 percent confidence interval [CI]: 124-184), but recent deaths may have been underreported. Lack of beds in the household (odds ratio [OR] = 2.0, 95 percent CI: 1.1-3.8), absence of the father (OR = 2.4, 95 percent CI : 1.2-4.8), low paternal educational level (OR = 2.1, 95 percent CI: 0.8-5.4), young maternal age (OR = 2.0, 95 percent CI: 1.0-3.7), self-reported maternal illness (OR = 2.4, 95 percent CI : 1.2-4.9), and home delivery of the child (OR = 2.3, 95 percent CI : 1.2-4.5) were associated with increased mortality, but the sensitivity of risk factors was low. Estimated maternal mortality was 410/100 000 live births with a reference date of 1982. CONCLUSIONS: Child mortality decreased slowly over the 1980s in Beira despite poor living conditions caused by the indirect effects of the war. Coverage of health services increased over this period. The appropriateness of a risk approach to maternal-child-health care needs further evaluation.

PIP: In July 1993, public health specialists conducted a cluster sample survey of 4609 women aged 15-49 living in 3190 houses in Beira city to determine child and maternal mortality after 10 years of internal conflict in Mozambique and a nested case control study of risk factors for child mortality. The indirect estimate techniques were child ever born and preceding birth techniques for child mortality and the sisterhood method for maternal mortality. The case control study compared 106 deaths among the most recent born child born since July 1990 with two age- and cluster-matched controls. The proportion of dead sisters who died of pregnancy-related causes was only 10.3% compared to 25-33% in developing countries. In 1982, the estimated maternal mortality ratio was 410/100,000 live births. The lifetime risk of maternal mortality was 263/1000. The preceding birth technique obtained a much lower child mortality estimate than the child ever born technique (154 vs. 212/1000). The child ever born technique analyzed data from 1977-1978 to 1988-1989 and found that the probability of dying from birth to age 5 fell 14% (246-212). During this period, coverage of health services improved. Even though the preceding birth technique is usually more reliable for recent estimates, underreporting of recent child deaths likely contributed to the lower child mortality estimate. Risk factors for child mortality included no beds in the household (odds ratio [OR] = 2.02), absence of the father (OR = 2.43), low paternal educational level (OR = 2.08), young maternal age (OR = 1.96), self-reported maternal illness since birth of child (OR = 2.43), and home delivery (OR = 2.31). Yet the sensitivity of these risk factors was rather low (15-57%). These findings show that child mortality fell slowly during the 1980s despite the poor living conditions brought about by the indirect effects of the civil war. They point to the need to further evaluate the appropriateness of a risk approach to maternal and child health care needs.

Estimating maternal mortality in Djibouti: an application of the sisterhood method.

In many developing countries even crude estimates of the level of maternal mortality are lacking and the prospects of fulfilling this need using conventional sources of vital registration and health service statistics are not encouraging. The constraint this imposes on the effective planning, management and evaluation of the programmes now being launched to reduce these neglected deaths is self-evident. It is less obvious how the majority of developing countries can be expected to meet the call for reliable estimates of maternal mortality by 1995. The sisterhood method provides a means of obtaining population-based estimates using household surveys for data collection. This paper describes the application of the method in Djibouti in the context of a rapid multi-purpose household survey in difficult field circumstances. In recent years the reduction of the level of maternal mortality in developing countries has become a priority for both national governments and international agencies. Attention has been drawn to the wide range of levels within and between countries and to the huge discrepancies in the lifetime risk of maternal death for women in the developed compared with the developing world. This risk has been estimated to range from 1 in 19 in West Africa to almost 1 in 10,000 in Northern Europe.

PIP: The sisterhood method, developed in 1987, of estimating maternal mortality is applied to Djibouti population data. The method is based on the reports of the proportion of adult sisters dying during pregnancy, childbirth, or the puerperium. The method is used where date collection is limited or the field circumstances are difficult. The advantage is that it can be used with small sample sizes to produce stable estimates. It is useful in gauging the extent to which there has been underreporting of maternal deaths. 4 questions and the 5 year age group of the respondent are required. Precise working is a requirement. The survey was implemented in February, 1989 using a modified version of the EPI/CDD cluster sampling method. Comprising the national sample, 30 clusters were involved in Djibouti City and 30 clusters in 4 rural areas using probability proportional to size (PPS) methods and the 1983 population census sampling frame with a 3% increase. Within 100 ever married women were selected per cluster and 4000 respondents were desired. The assumption was about 500 per 100,000 live births, a total fertility rate of about 6, and an error rate of 6%. The questionnaire was translated from French into Afar, Arabic, and Somali. The method of training interviewers, the group management, and pretesting are described. The 4 questions in English were 1) How many sisters have you ever had who were born to your mother?. 2) How many of these sisters reached age 15, including those now alive now? 3) How many of these sisters reaching age 15 are dead? and 4) How many of these dead sisters died while pregnant, or during childbirth or in the 6 weeks after delivery, an abortion, or a miscarriage? The results of the 7408 females 15-49 years interviewed were that the lifetime risk of dying of maternal causes were found to be 0.049 or 1 in 20. Using a total fertility rate of 6.8, the maternal mortality ratio was calculated as 740 maternal deaths per 100,000 live births 11.6 years prior to the survey. The results of the assessment of the quality of the data showed underreporting of the 2 youngest age groups, which suggests misreporting. Severe age heaping on ages of respondents ending in 0 and 5 was also apparent. In spite of the difficulties, the results are plausible and lend support to the method.

Specificity of association in analysis of mortality and inference on causality.

Specificity of association between putative risk factor and disease under study is important to inference on causality. Nevertheless many studies investigate mortality of a single disease without comparison with a control. Age-standardized proportional mortality ratios make single disease studies into case-control studies and thus demonstrate whether or not associations are disease specific. Comparison of disease-specific with all-cause mortality experiences of whole populations classified by exposure, clearly distinguishes between exposures associated with more death and with earlier/younger death, thereby overcoming an important limitation of the familiar standardized mortality ratio (SMR). Smoking is associated with more death from lung cancer (lifetime cause-specific proportions, never 1%, light 6%, moderate 8% and heavy 12%) and with earlier/younger death from ischaemic heart disease (never 35%, light 34%, moderate 32% and heavy 29%).

Inappropriate use of daily mortality analyses to estimate longer-term mortality effects of air pollution.

BACKGROUND: To avoid the usual problems of multi-population correlation studies of air pollution and mortality, and for reasons of convenience, daily time-series mortality studies within single populations have recently become popular in air pollution epidemiology. Such studies describe how the short-term distribution of deaths relates to short-term fluctuations in air pollution levels. The regression-based risk coefficients from these acute-effects studies have been widely used to estimate the excess annual mortality within a population with a specified average level of air pollution. Such calculations are inappropriate. Since daily time-series data provide no simple direct information about the degree of life-shortening associated with the excess daily deaths (many of which are thought to be due to exacerbation of well-advanced disease, especially cardiovascular disease), such data cannot contribute to the estimation of the effects of air pollution upon chronic disease incidence and long-term death rates. Yet it is that category of effect that is of most public health importance. CONCLUSION: Such effects are best estimated from long-term cohort studies that incorporate good knowledge of local (or personal) exposure to air pollutants and of potential confounders. Time-series studies, properly evaluated, can identify the existence of acute toxic effects of transient peak levels of air pollution; they are thus useful for monitoring acute toxicity and for identifying the most noxious pollutants. However, to quantify the long-term health impacts of air pollution we cannot use acute-effects data.

Estimating deaths and injuries due to road traffic accidents in Karachi, Pakistan, through the capture-recapture method.

BACKGROUND: Road traffic accidents (RTA) are an important yet preventable cause of death and disability in developing countries like Pakistan. Yet accurate epidemiological data on injuries in developing country injuries is often difficult to obtain. We applied the capture-recapture method to estimate the death and injury rates due to RTA in Karachi. METHODS: We applied the two-sample capture-recapture method using traffic police records as one source of capture and the logs of a non-government ambulance service as the second capture source for the same 10 months and 20 days for which 1994 data were available. We generated a conservative adjusted estimate of injuries and deaths by considering entries in the two sources as matched if they reported the same date, time, and place, and at least one of the other matching variables, of name, vehicle registration number, vehicle types or patient outcome. We then compared the estimated rates with the police rates. RESULTS: In 1994 police reported 544 deaths and 793 injuries due to RTA while ambulance records noted 343 deaths and 2048 injuries. The capture-recapture analysis estimated at least 972 (95% CI: 912-1031) deaths and 18,936 (95% CI: 15,507-22,342) injuries attributable to RTA during the study period. Official sources counted only 56% of deaths and 4% of serious injuries. The estimated rates for the year 1994 were 185 injuries and 11.2 deaths per 100,000 population. CONCLUSION: Road traffic injuries and deaths in Karachi are a much more substantial health problem than is evident from official statistics.

Provisional country estimates of prevalent adult human immunodeficiency virus infections as of end 1994: a description of the methods.

BACKGROUND: A country-by-country review of human immunodeficiency virus (HIV)/acquired immunodeficiency syndrome (AIDS) data was undertaken by the World Health Organization. This paper describes the methods used to make estimates of HIV prevalence. RESULTS: It is estimated that, globally, approximately 16.9 million adults were living with HIV infections at the end of 1994. The majority (66%) of the infections were in sub-Saharan Africa (over 11000000), followed by South and South East Asia (over 3000000). Estimated prevalence rates for HIV infection ranged from less than 1 per 100000 sexually active adult population to 18 per 100 (18%), with a median prevalence of 14 per 10000. In 50 countries the estimated HIV prevalence rate was less than 5 per 10000 sexually active adults. In 15 countries (all in sub-Saharan Africa) the prevalence rate was above 5%. The lowest estimated prevalence rates were seen in Central and East Asia and the highest in Central and Southern Africa. CONCLUSIONS: Estimates of prevalent HIV infections are intended to give an indication of the magnitude of the HIV pandemic but, due to the difficulties in accurately assessing the levels of HIV infections in national populations, should be considered provisional.

PIP: During 1995, the World Health Organization (WHO) reviewed country-level data on adult HIV/AIDS cases and, in consultation with local and regional experts, revised its estimates of HIV prevalence as of the end of 1994. Data sources included cases reported to WHO by Member States, routine HIV surveillance, published studies, and blood banks. Globally, an estimated 16.9 million adults were infected with HIV at the end of 1994. About 66% of HIV infections were in sub-Saharan Africa and 18% in South and South East Asia. HIV prevalence ranged from less than 1/100,000 sexually active adults to 18/100, with a median of 14/10,000. 43% of adult HIV infections were in the least developed countries and another 50% in developing and middle-income countries. Although these statistics should be considered provisional, they provide an indication of the magnitude of HIV infection, provide a baseline for future monitoring, and may facilitate rational decision making regarding resource allocation. This paper includes a country-by-country listing of the estimated 1994 adult HIV prevalence rate per 100 population and the number of adult HIV infections, working estimates of adult HIV prevalence by subcontinent, and a case study (India) illustrating WHO's estimation technique.

A comparison of sisterhood information on causes of maternal death with the registration causes of maternal death in Matlab, Bangladesh.

BACKGROUND: To explore whether causes of maternal death can be investigated using the sisterhood method, an indirect method for providing a community-based estimate of the level of maternal mortality, this study compares the sisterhood causes of maternal death with the Matlab Demographic Surveillance System's (DSS) causes of maternal death. METHODS: Data for this study came from the Matlab DSS, which has been in operation since 1966 as a field site of the International Centre for Diarrhoeal Disease Research, Bangladesh. The maternal deaths that occurred during the 15-year period from 1976 to 1990 in the Matlab DSS area are the basis of this study. A sisterhood survey was conducted in Matlab in November and December 1991 to collect information on conditions, events and symptoms that preceded death. The collected information was evaluated to assign a most likely cause of maternal death. The sisterhood survey cause of maternal death was then compared with the DSS cause of maternal death. RESULTS: Cause of death could not be assigned with reasonable confidence for 34 (11%) of the 305 maternal deaths for which information was collected. For the remaining deaths, the agreement between the two classification systems was generally high for most cause-of-death categories considered. CONCLUSIONS: Though cause-of-death information obtained by the sisterhood method will always be subject to some error, it can provide an indication of an overall distribution of causes of maternal deaths. This data can be used for the planning of programmes aimed at reducing maternal mortality and for the evaluation of such programmes over time.

PIP: This study compared the sisterhood method of determining causes of maternal death, an indirect method for providing a community-based estimate of the level of maternal mortality, with the Matlab Demographic Surveillance System's (DSS) causes of maternal death. Data were derived from the Matlab DSS, which has been in operation since 1966 as a field site of the International Center for Diarrheal Disease Research, Bangladesh. The maternal deaths that occurred during the 15-year period from 1976 to 1990 in the Matlab DSS area were the basis of this study. A sisterhood survey was conducted in Matlab in November and December 1991 to collect information on conditions, events, and symptoms that preceded death. The collected information was evaluated to assign a most likely cause of maternal death. The sisterhood survey cause of maternal death was then compared with the DSS cause of maternal death. Of the 510 deaths identified as maternal deaths by the DSS, 384 siblings, 1 for each deceased woman, was interviewed. 305 of these correctly reported that they had a sister who died during pregnancy or childbirth. 16 reported that they did not know whether their sister died during pregnancy or after termination of a pregnancy. The remaining 63 respondents misreported their sisters' deaths as nonmaternal deaths. Cause of death could not be assigned with reasonable confidence for 34 (11%) of the 305 maternal deaths for which information was collected. For the remaining 271 deaths, the agreement between the 2 classification systems was generally high for most cause-of-death categories considered. The overall rate of agreement between DSS cause and survey cause was 82%. For the direct obstetric deaths as a group, the agreement was 86%, while it was around 76% for indirect obstetric deaths, and 71% for abortion-related deaths. Though the sisterhood method will always be subject to some error, it can provide an indication of an overall distribution of causes of maternal deaths.