Life Table Techniques Research Articles

Age-specific incidence and clearance of high-risk human papillomavirus infections in women in the former Soviet Union

Recently, conflicting results on human papillomavirus (HPV) clearance have been reported and the data on the accumulation of incident HPV infections are still fragmentary. Thus, we completed an analysis of the age-specific incidence and clearance rates of high-risk (HR) HPV infections in 448 women participating in a multi-centre screening study in three New Independent States countries. At baseline, 239 of the 448 women were negative for HR HPV DNA, whereas 209 were HR HPV-positive and cleared HR HPV during the prospective follow-up. The cumulative incidence and clearance of HR HPV were modelled using life-table techniques. The monthly incidence rates of HR HPV were significantly age dependent (p = 0.0001), whereas monthly clearance rates remained constant across the nine age groups (p = 0.920). The incidence rates (3.04% and 2.65%) exceeded the clearance rates in the two youngest age groups only, 15-20- and 21-25-year-old women, and remained lower (0-0.84%) in all other age groups. The cumulative rate of incident HR HPV infections (1.0%) was significantly lower than the overall clearance rate (1.9%) (p = 0.001). In life-table analysis, incident HR HPV infections between the nine age groups were significantly different (p = 0.0001), while cumulative HR HPV clearance was identical in all groups (p = 0.822). The accumulation of incident HR HPV infections is significantly age-related, whereas virus clearance remains constant between 15 and 60 years of age. These distinct age-specific incidence and clearance rates explain the differences in age-specific prevalence of HR HPV infections in the study population.

Clearance of high-risk human papillomavirus (HPV) DNA and PAP smear abnormalities in a cohort of women subjected to HPV screening in the New Independent States of the former Soviet Union (the NIS cohort study)

Background: We analysed the temporal relationships of the clearance of human papillomavirus (HPV) DNA and cytological abnormalities in women participating in a screening study in three NIS countries. Methods: The 274 patients included in this analysis were prospectively followed-up for 21.6 months (range: 0.5–42.9). All 274 women had abnormal PAP test (ASC-US or higher) and high-risk HPV-positive test (HCII) at baseline. Two groups were compared: 132 women who cleared both tests (Group 1), and 142 women who cleared either HPV or abnormal PAP test (Group 2). The first clearance during the follow-up, and the last visit clearance were modeled using life-table techniques, and the predictive factors were analysed using univariate (Kaplan–Meier) and multivariate (Cox) survival analysis. Results: There was no difference in the mean clearance time for the abnormal PAP test (14.4 months; 0.7–40.5 and 12.6 months; 0.5–35.0) and high-risk HPV DNA (12.67 months; 0.6–33.5 and 10.8 months; 0.7–33.4) in Group 1 and Group 2 (Mann–Whitney: P = 0.107 and P = 0.082, respectively). Clearance times for HPV DNA and abnormal PAP test did not deviate from each other in either groups (Wilcoxon: P = 0.063 and P = 0.088). The monthly clearance rates for the abnormal PAP test are 1.32 and 1.38%, and those for the HPV DNA 1.62 and 1.61%, in Groups 1 and 2, respectively. Of the factors predicting the last visit clearance, the issues related to smoking are of particular interest. Conclusions: The clearance of high-risk HPV type and abnormal PAP test shows a close temporal relationship, the former preceding the latter, however, by an interval of 1.0–2.0 months.

Risk for Relatives of Patients with Multiple Sclerosis in Central Sardinia, Italy

Multiple sclerosis (MS) is a chronic, inflammatory, disabling disease of the central nervous system, known for its complex interplay between genetic and environmental factors. We used life table techniques to calculate age-adjusted recurrence risks for different categories of relatives of MS patients from Central Sardinia (Italy), a genetically homogeneous, stable population with a high degree of consanguinity. We included 313 probands and a total of 12,717 relatives in the analysis. The overall age-adjusted recurrence risk for relatives of MS probands is 1.90% [95% confidence interval (CI): 1.57–2.30]. The age-adjusted recurrence risk in parents was 1.26% (95% CI 0.60–2.63), in children 2.33% (95% CI 0.09–5.56), in sibs 4.76% (95% CI 3.57–6.32), in second-degree relatives 0.72% (95% CI 0.42–1.22), and in third-degree relatives 1.79% (95% CI 1.27–2.51). The sex of the probands (male) and of the relatives (female), and the number of affected relatives in the family significantly increase the risk of MS in relatives.

Measuring the impact of hepatitis A virus in Chinese cities: a life table approach.

The infection of hepatitis A virus is an important public health problem in China. In 1990, a survey on seroprevalence of anti-hepatitis A was conducted in eight large cities of population size between one to four million. The results were used to analyze the impact of hepatitis A. In our study, we used life table techniques to determine the impact that the hepatitis A virus (HAV) has on these eight cities by computing the expected years of life with anti-hepatitis A virus. We compared the results from our approach with those from a previous method. The expected years of life with anti-hepatitis A virus were 53.82, 54.65, 44.26, 53.89, 46.18, 55.77, 55.31 and 49.89 for the eight cities (Xi'an, Huhehaote, Chongqing, Nanjing, Jinan, Ha'erbin, Fuzhou and Nanchang), respectively. The hepatitis A virus had the greatest impact in Ha'erbin in northeastern China; whereas, the impact was lowest in Chongqing in southwestern China. However, based on the previous indicator, Chongqing was ranked the highest among these cities. Through our comparative study of the methods, our approach provides a better measure on the burden of the disease than the previous method. It also identifies subgroups of the population where the disease has its greatest impact on the population.

Tailoring of minimal immunosuppression long term

WITH INCREASED EFFICACY of initial immunosuppressive therapy the proportion of patients experiencing acute rejection has been substantially reduced. It is also evident that the events have been delayed in time. The incidence of acute rejection in clinical trials on new immunosuppressive drugs are often depicted using the life table technique. Following an initial period of time with a rising curve there are no more early episodes of acute rejection and the curve becomes flat. In control groups, the first period of time is usually about 4 months. With reduction of acute rejection the point in time when there are no more events is often delayed until about 6 months. One example is the pivotal trial on daclizumab. The two mentioned time periods have been called preadaptation and postadaptation, respectively. During preadaptation it is vital to give effective immunosuppression with multiple drugs, including induction, a calcineurin inhibitor, mycophenolate mofetil (MMF) or azathioprine (Aza), and steroids. In postadaptation, it is just as important to minimize immunosuppression to reduce long-term toxicity. This may be achieved by selective withdrawal, using two instead of three maintenance drugs, and by keeping doses low. This article focuses on the question of whether all double combinations of immunosuppressive drugs are equally safe and, secondly, how to tailor immunosuppression to the individual patient’s needs. Factors negatively affecting graft and patient survival in postadaptation include chronic nephrotoxicity, hypertension, hyperlipidemia, and diabetes—all leading to cardiovascular disease and malignancy. The increased incidence of malignancy after renal transplantation is one sign of toxicity of immunosuppression. Invasive, life-threatening tumors occur in about 17% of transplant patients at 20 years. Nonmelanoma skin cancer occurs in 40% to 60% at 20 years. These findings support arguments to reduce immunosuppression during postadaptation. The major cause of graft loss after renal transplantation is patient death with a functioning graft, which occurs in 40% to 50% of patients. The major cause of death in these patients is cardiovascular disease, namely in about 50% of cases similar to the general population. However, renal transplant patients die from cardiovascular disease at younger ages than the general population. Consequently, the objective of long-term management should be to reduce the level of immunosuppression by selecting drugs that minimize risk factors for cardiovascular disease. Particularly patients with overt drug side effects, for example, risk factors such as hypertension and hyperlipidemia, would benefit from drug modification. The various toxicity profiles of immunosuppressive drugs may be the instrument at hand. Recently, we performed a survey at the Transplant Unit in Malmo, Sweden, to investigate how many of our patients receive too much immunosuppressive therapy. Among 1500 kidney transplant patients 550 were surviving with functioning grafts more than 1 year after transplantation. More than half of them (57%) were receiving a triple-drug combination, including cyclosporine (CsA) or tacrolimus, MMF or Aza, and prednisone at a mean of 7 years after transplant. Reduction of their immunosuppression could be achieved by discontinuing one drug, leaving eight different doubledrug combinations. The major question was: do we have evidence that all of these combinations are equally safe? A great number of clinical trials have been reported in the literature with different designs to evaluate the safety of double-drug combinations. They may have been designed for avoidance, sparing, or withdrawal, mainly focusing on calcineurin inhibitors or steroids. Some trials are randomized and some involve only selected patients. Kasiske and coworkers recently published a meta-analysis of randomized trials on steroid or CsA withdrawal. They reported an increased risk of acute rejection (14%) and graft loss (38%) after steroid withdrawal. However, most of the patients in these trials were maintained on CsA monotherapy or CsA Aza after steroid withdrawal. Consequently, this latter double-drug combination should be marked as doubtful since it does not achieve a sufficient level of immunosuppression during the postadaptation period. Another double-drug combination, CsA MMF, was tested by Grinyo et al in a pilot study of 26 selected patients. No acute rejections were reported at 10 months after steroid withdrawal. Using the same double-drug combination for maintenance therapy, Vanrenterghem et al tested continued steroids vs steroid sparing (1/2 dose ini-

Shifting Childrearing to Single Mothers: Results from 17 Western Countries

We investigate how recent changes in the Western family have affected childhood living arrangements. For 17 developed countries, we use multistate life table techniques to estimate childhood trajectories of coresi‐dence with biological fathers versus other maternal partners. In all countries childhood exposure to single parenting is more often caused by parental separation than out‐of‐partnership childbearing. Both exposure to single parenting and expectancy of childhood spent with a single non‐cohabiting mother vary widely across countries, with the United States exhibiting the highest levels of each at early 1990s rates. The greatest international variations concern parental cohabitation—its prevalence, durability, and the degree to which its increase has compensated for a decrease in the expectancy of childhood spent with married parents. Overall, we find little evidence of international convergence in childrearing arrangements, except that in countries where parental marriage has declined over time, childrearing has predominantly shifted to single mothers.

The incidence of varicella: correction of a common error.

Interest in the incidence of varicella (chickenpox) has increased since the discovery of an effective vaccine, but calculations to date have incorrectedly ignored the question of susceptibility. We studied the occurrence of varicella in Italy on the basis of 33,343 children (age 0-14 years) cared for by 35 pediatricians between 1 October 1997 and 30 September 1998. The life-table technique was used to calculate the number of susceptible children. On this basis, we estimated the corrected age-specific and cumulative incidence. We identified 1749 cases among the estimated 21,783 susceptible children, for a crude incidence of 8.0% (95% confidence interval [CI] = 7.7-8.4). The rate age-standardized to the Italian population 0-14 years old was 6.8 (CI = 6.5-7.2). The incidence was more than 16% among children age 3-4 years and more than 4% for those age 1-10 years. Comparison of the usual method and our corrected method showed that the uncorrected method underestimates the crude annual incidence (5.2% 8.0%), shifts the peak incidence to earlier ages, and underestimates cumulative incidence (at age 14, 49% 67%). The use of our corrected method provides more valid estimates of the incidence of varicella than the ones that are currently available. Corrected estimates should be preferred to uncorrected ones in models to study the cost-effectiveness of universal vaccination against varicella.

Analysis of ante-partum maternal morbidity in rural Bangladesh.

This paper presents the results of a prospective study of maternal morbidity during the ante-partum period in rural areas of Bangladesh. The data came from a survey of Maternal Morbidity in Bangladesh, conducted by the Bangladesh Institute of Research for Promotion of Essential and Reproductive Health and Technologies (BIRPERHT) during the period from November 1992 to December 1993. Since then no such national level survey has been conducted in Bangladesh. This paper employs multiple-decrement life table technique, a convenient way of analysing the risks of different types of disease conditions that women experience during the antenatal period for different age categories. The high-risk complications such as ante-partum haemorrhage, excessive vomiting, fits/convulsion and oedema were considered in this study. In this study a cause specific model was applied to explore the differences in the risks exerted at different ages of reproductive life attributable to some selected complications of pregnancy. The results of this study indicate that women of age 25-29 years are less susceptible to most of the selected life-threatening and high-risk complications during pregnancy such as haemorrhage, fits/convulsion and oedema. However, younger women (age < 25 years) are more likely to have excessive vomiting during pregnancy, and older women (age > or = 30 years) are at greater risk of haemorrhage, fits/convulsion and oedema.

Life-table representations of family dynamics in Sweden, Hungary, and 14 other FFS countries

In this study, we present a system of descriptions of family-demographic behavior in developed countries. We use life-table techniques in order to describe the experience of men, of women, and of children in processes related to family formation and fa

Segmental resection or subtotal colectomy in Crohn's colitis?

Segmental resection for Crohn's colitis is controversial. Compared with subtotal colectomy, segmental resection is reported to be associated with a higher rate of re-resection. Few studies address this issue, and postoperative functional outcome has not been reported previously. This study compared segmental resection to subtotal colectomy with anastomosis with regard to re-resection, postoperative symptoms, and anorectal function. Fifty-seven patients operated on between 1970 and 1997 with segmental resection (n = 31) or subtotal colectomy (n = 26) were included. Reoperative procedures were analyzed by a life-table technique. Segmentally resected patients were also compared separately with a subgroup of subtotally colectomized patients (n = 12) with similarly limited colonic involvement. Symptoms were assessed according to Best's modified Crohn's Disease Activity Index and an anorectal function score. The re-resection rate did not differ between groups in either the entire study population (P = 0.46) or the subgroup of patients with comparable colonic involvement (P = 0.78). Segmentally resected patients had fewer symptoms (P = 0.039), fewer loose stools (P = 0.002), and better anorectal function (P = 0.027). Multivariate analysis revealed the number of colonic segments removed to be the strongest predictive factor for postoperative symptoms and anorectal function (P = 0.026 and P = 0.013, respectively). Segmental resection should be considered in limited Crohn's colitis.

De novo malignancies after liver transplantation: a major cause of late death.

1. Recurrent and de novo malignancies are the second leading causes of late death in liver transplant recipients, following age-related cardiovascular complications. 2. The increased incidence of de novo malignancies in liver transplant recipients compared with the general population reflects their demographic makeup, known preexistent risk factors for cancer, greater rate of chronic viral infection, and actions of exogenous immunosuppression. 3. The greatest incidence of de novo malignancies is seen in cancers associated with chronic viral infections, such as Epstein-Barr virus-associated posttransplant lymphoproliferative disease, and skin cancers, including squamous cell carcinoma and Kaposi's sarcoma. 4. Although a greater incidence of such malignancies as oropharyngeal malignancy and colorectal cancer was noted, there did not appear to be an increased risk for liver transplant recipients matched for age, sex, and length of follow-up using modified life-table technique and Surveillance Epidemiology End Result data with a similar at-risk group. However, they may present with more advanced stages of disease. 5. An increased incidence of de novo cancers in chronically immunocompromised liver transplant recipients demands careful long-term screening protocols to help facilitate diagnosis at an earlier stage of disease.

Small bowel obstruction after appendicectomy.

This study analysed the risk of surgically treated small bowel obstruction after open appendicectomy. This was a historical cohort study of 245 400 patients who underwent open appendicectomy and population-based matched controls, identified by linkage of computer registries. Analyses were made with the life-table technique, Kaplan-Meier plots and Cox proportional hazards regression analysis. The cumulated risk of surgically treated small bowel obstruction after appendicectomy was 0.41 per cent after 4 weeks, 0.63 per cent after 1 year and 1.30 per cent after 30 years of follow-up, compared with 0.003 per cent at 1 year and 0.21 per cent after 30 years of follow-up among the non-operated controls. The highest risk was found after operation for other diagnoses (adjusted hazard ratio 5.2 (95 per cent confidence interval 4.6-5.8)), followed by operation for perforated appendicitis (adjusted hazard ratio 3.5 (3.1-3.8)), non-specific abdominal pain (adjusted hazard ratio 2.6 (2.3-3.0)) and mesenteric lymphadenitis (adjusted hazard ratio 2.4 (2.0-2.8)) compared with operation for non-perforated appendicitis. The relation with age was J shaped, with the lowest risk at 20-39 years. Women had a slightly lower risk than men. The risk of postoperative small bowel obstruction needing surgical treatment after open appendicectomy is lower than previously thought. Perforated appendicitis, negative appendicectomy and high age are the risk factors.

Childbearing in Cohabiting Unions: Racial and Ethnic Differences

Cohabitation provides a two-parent family union in which to have and raise children outside of marriage. Little is known, however, about the conditions under which cohabiting couples conceive and decide to have children. The National Survey of Family Growth provides detailed data on the cohabitation and fertility histories of American women. Life-table techniques, event-history analyses and logistic regression were employed to understand the racial and ethnic differences in the timing of childbearing within cohabiting unions and whether childbearing within cohabiting unions is more acceptable to members of minorities than to whites. In multivariate models, Hispanic women were found to be 77% more likely than white women to conceive a child in cohabitation and black women were 69% more likely than white women to do so. Among women who became pregnant while cohabiting, Hispanic women were almost twice as likely and black women were three times as likely as white women to remain cohabiting with their partner when their child was born. In addition, children born to Hispanic women in cohabiting unions were found to be 70% more likely to be intended than were those born to cohabiting white women. In terms of fertility, cohabitation does not maintain the same place in the American family system for all racial and ethnic groups. These racial and ethnic differences in fertility-related behavior are not explained by socioeconomic differences. Based on levels of childbearing during cohabitation, relationship status at time of birth and intention status of children, it appears that cohabitation is a more acceptable arena for family building among Hispanic women than among whites or blacks.

Childbearing in five Arab countries.

The life-table technique is used in this study to describe childbearing patterns in five Arab countries (Algeria, Egypt, Morocco, Sudan, and Yemen). Two summary measures are estimated: the cumulative proportion of women of a given parity having a subsequent birth within 60 months of the previous birth (quintum), and the median length of the interval between successive births. The analysis indicates that in Yemen (where the total fertility rate exceeds seven children per woman of reproductive age), women begin childbearing at a relatively early age and a large proportion of ever-married women reach high parity at a relatively fast pace. Morocco and Egypt (with TFRs of 3.3 and 3.8 children per woman, respectively) are characterized by delay in the onset of childbearing, a slow pace of childbearing, and a smaller proportion of ever-married women reaching high parity. In Algeria and Sudan (with TFRs of 4.4 and 4.6 children), delay in the onset of childbearing plays a larger role in fertility reduction than do the tempo or the quantum of childbearing.

Insect biodemography.

Biodemography is an emerging subdiscipline of classical demography that brings life table techniques, mortality models, experimental systems, and comparative methods to bear on questions concerned with the fundamental determinants of mortality, longevity, aging, and life span. It is important to entomology because it provides a secure and comprehensive actuarial foundation for life table and mortality analysis, it suggests new possibilities for the use of model insect systems in the study of aging and mortality dynamics, and it integrates an interdisciplinary perspective on demographic concepts and actuarial techniques into the entomological literature. This paper describes the major life table formulae and mortality models used to analyze the actuarial properties of insects; summarizes the literature on adult insect life span, including a discussion of basic concepts; identifies the major correlates of extended longevity; and suggests new ideas for using demographic concepts in both basic and applied entomology.

Cell dose and speed of engraftment in placental/umbilical cord blood transplantation: graft progenitor cell content is a better predictor than nucleated cell quantity

There is evidence that the total cellular content of placental cord blood (PCB) grafts is related to the speed of engraftment, though the total nucleated cell (TNC) dose is not a precise predictor of the time of neutrophil or platelet engraftment. It is important to understand the reasons for the quantitative association and to improve the criteria for selecting PCB grafts by using indices more precisely predictive of engraftment. The posttransplant course of 204 patients who received grafts evaluated for hematopoietic colony-forming cell (CFC) content among 562 patients reported previously were analyzed using univariate and multivariate life-table techniques to determine whether CFC doses predicted hematopoietic engraftment speed and risk for transplant-related events more accurately than the TNC dose. Actuarial times to neutrophil and platelet engraftment were shown to correlate with the cell dose, whether estimated as TNC or CFC per kilogram of recipient's weight. CFC association with the day of recovery of 500 neutrophils/μL, measured as the coefficient of correlation, was stronger than that of the TNC (R = −0.46 and −0.413, respectively). In multivariate tests of speed of platelet and neutrophil engraftment and of probability of posttransplantation events, the inclusion of CFC in the model displaced the significance of the high relative risks associated with TNC. The CFC content of PCB units is associated more rigorously with the major covariates of posttransplantation survival than is the TNC and is, therefore, a better index of the hematopoietic content of PCB grafts.

Cell dose and speed of engraftment in placental/umbilical cord blood transplantation: graft progenitor cell content is a better predictor than nucleated cell quantity

AbstractThere is evidence that the total cellular content of placental cord blood (PCB) grafts is related to the speed of engraftment, though the total nucleated cell (TNC) dose is not a precise predictor of the time of neutrophil or platelet engraftment. It is important to understand the reasons for the quantitative association and to improve the criteria for selecting PCB grafts by using indices more precisely predictive of engraftment. The posttransplant course of 204 patients who received grafts evaluated for hematopoietic colony-forming cell (CFC) content among 562 patients reported previously were analyzed using univariate and multivariate life-table techniques to determine whether CFC doses predicted hematopoietic engraftment speed and risk for transplant-related events more accurately than the TNC dose. Actuarial times to neutrophil and platelet engraftment were shown to correlate with the cell dose, whether estimated as TNC or CFC per kilogram of recipient's weight. CFC association with the day of recovery of 500 neutrophils/μL, measured as the coefficient of correlation, was stronger than that of the TNC (R = −0.46 and −0.413, respectively). In multivariate tests of speed of platelet and neutrophil engraftment and of probability of posttransplantation events, the inclusion of CFC in the model displaced the significance of the high relative risks associated with TNC. The CFC content of PCB units is associated more rigorously with the major covariates of posttransplantation survival than is the TNC and is, therefore, a better index of the hematopoietic content of PCB grafts.

A 10-year follow-up study of contraceptive Norplant implants

Objectives: To evaluate the long-term efficacy and health problems associated with Norplant implants and re-implants after 5 full years of first implants. Method: From 1984 to 1988, 1657 women accepted first implants of Type I and Type II of Norplant, and 394 first acceptors had a re-implant, at a clinic for study. The clinical records and annual follow-up data of acceptors were analyzed. The life-table technique was the main method used. Results: The continuation rates were very high. The cumulative pregnancy rates at 1, 3 and 5 full years of use were 0.0%, 0.1% and 0.7%, respectively. Users with heavier body weight had higher pregnancy rates. The rate of menstrual disturbances peaked at 73% after 3 months and consistently decreased to 20% at 5 years of use. Rates of menstrual disturbances associated with re-implants were much lower. Conclusion: Norplant is extremely effective and safe for long-term use.

Lack of improvement of life expectancy at advanced ages in The Netherlands.

Several countries have reported an increase in life expectancy at advanced ages. This paper analyses recent changes in life expectancy at age 60 and 85 in The Netherlands, a low mortality country with reliable mortality data. We used data on the population and the number of deaths by age, sex and underlying cause of death for 1970-1994. Life expectancy at age 60 and 85 was estimated using standard life-table techniques. The contribution of different ages and causes of death to the change in life expectancy during the 1970s (1970/74-1980/84) and the 1980s (1980/84-1990/94) were estimated with a decomposition technique developed by Arriaga. Life expectancy at age 60 increased in the 1970s and 1980s, whereas life expectancy at age 85 decreased (men) and stagnated (women) in the 1980s, and has decreased in both sexes since 1985/89. The decomposition by age showed that constant mortality rates in women aged 85-89, and increasing mortality rates at ages 85+ (men) and 90+ (women) have caused this lack of increase in life expectancy. The decomposition by cause of death showed that smaller mortality reductions from other cardiovascular and cerebrovascular diseases, which contributed most to the increase in life expectancy at age 85 in the 1970s, and mortality increases from, amongst others, chronic obstructive pulmonary disease (COPD), mental disorders and diabetes mellitus produced the decrease (men) and plateau (women) in life expectancy at age 85. Life expectancy at advanced ages stopped increasing during the 1980s in The Netherlands due to mortality increases at ages 85+ (men) and 90+ (women). Cause-specific trends suggest that, in addition to (past) smoking behaviour in men, changes in the distribution of morbidity and frailty in the population might have contributed to this stagnation.

Contraceptive failure: levels, trends and determinants in Matlab, Bangladesh.

This study investigated the levels, trends and determinants of contraceptive use-failure in Matlab, Bangladesh, using a set of prospective data on 25,960 women of reproductive age. The data were extracted from the Record Keeping System (RKS) of Matlab for the period 1978-94. If there was any live birth during the use or within 7 months after the discontinuation of use, it was considered as a failure. The life table technique and hazard model were used as analytical tools. The results suggest that use-failure for pills, IUDs (TCu 200) and injectables and other temporary methods increased from 1978 to 1988, but began to decline after 1988. The cumulative probability of first-method failure within 1 year of method acceptance of the cohort of 1990-94 acceptors was 12.9% for pills, 2.0% for IUDs, 0.5% for injectables, 22.0% for condoms and 13.4% for 'other' methods (sampoon, foam, jelly and traditional methods). For pills, condoms and 'other' methods, the likelihood of failure declined with the duration of use; by contrast, the probability of an IUD failure increased over time, peaking at 3 years of use. The injectables maintained a low likelihood of failure regardless of the duration of use. The quality of Community Health Workers' (CHWs) performance was associated with the risk of failure of all temporary methods except condoms; women's background characteristics associated with failure varied by method. The effect of the quality of the CHWs' performance and the background variables on failure did not change much over time. It is felt that contraceptive failure deserves the serious attention of programme managers and policy makers to make the Bangladesh national family planning programme more successful.