AJTMH ASTMH MEMBERSHIP INFORMATION: astmh@astmh.org
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

Am. J. Trop. Med. Hyg., 69(1), 2003, pp. 92-97
Copyright © 2003 by The American Society of Tropical Medicine and Hygiene

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by EISELE, T. P.
Right arrow Articles by SLUTSKER, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by EISELE, T. P.
Right arrow Articles by SLUTSKER, L.

EVALUATING THE COMPLETENESS OF DEMOGRAPHIC SURVEILLANCE OF CHILDREN LESS THAN FIVE YEARS OLD IN WESTERN KENYA: A CAPTURE-RECAPTURE APPROACH

THOMAS P. EISELE, KIM A. LINDBLADE, DANIEL H. ROSEN, FRANK ODHIAMBO, JOHN M. VULULE, AND LAURENCE SLUTSKER
Division of Parasitic Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia; Department of International Health and Development, Tulane School of Public Health and Tropical Medicine, New Orleans, Louisiana; Center for Vector Biology and Control Research, Kenya Medical Research Institute, Kisumu, Kenya


ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
We evaluated the completeness and differential ascertainment of vital events in children less than five years old registered in two rounds of a demographic surveillance system (DSS) in western Kenya using a two-sample capture-recapture. The primary lists consisted of births and child deaths identified by two rounds of the DSS conducted in October 2000 and August 2001. The secondary lists consisted of births and child deaths identified independently from two surveys of 5,000 randomly selected households conducted immediately after each DSS round, covering the same population over the same time period. Analysis of the overlap between lists yielded the following sensitivities for the two DSS rounds: 62% and 49%, respectively, for identifying neonatal deaths (<1 month); 72% and 78%, respectively, for post-neonatal child deaths (1–59 months); and 88% and 78%, respectively, for identifying newborns. Female deaths were less likely to be reported than male deaths. The primary limitation of using capture-recapture in this setting was difficulty in matching between lists due to inconsistent dates of birth and death and variability in spelling of names. Assuming limitations of current methods are sufficiently addressed, capture-recapture appears to be a useful tool in evaluating DSS completeness and differential ascertainment of vital events.


INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Reduction in child mortality is the goal of many community-based public health interventions in developing countries, including vitamin A supplementation1–3 and the use of insecticide-treated bed nets (ITNs) in malaria-endemic areas.4–9 Such community-based interventions typically use all-cause mortality rates to measure impact. Recently, a number of sites in developing countries have established demographic surveillance systems (DSS) to measure mortality rates and other vital events through continuous monitoring of a defined population.10 Currently, there are at least 20 DSS sites worldwide.11

Although individual DSS sites may use different methods to capture births and deaths, events are likely to be missed no matter what method is used. Additionally, there may be differential ascertainment of vital events based on age, sex, geographic location, or season. A potentially effective means of assessing both the completeness of a DSS and differential ascertainment of demographic events is the use of capture-recapture methods. Capture-recapture techniques, first developed to estimate the size of wildlife populations, have been adapted to epidemiologic and demographic data to estimate the true number of some event of interest. By using the overlap between two or more independent lists of a particular event, capture-recapture techniques have been used to estimate the numbers of hard-to-reach populations,12–14 the number of disease events,15,16 as well as to evaluate completeness of disease surveillance systems.17,18

We used a two-sample capture-recapture method to evaluate the completeness of two rounds of a DSS in western Kenya in identifying births and deaths among children from birth to 5 years of age (0–59 months old). The purpose of this exercise was both to determine the feasibility of using capture-recapture techniques to evaluate our data collection methodology, and to identify births and deaths among particular subgroups that were likely to be undercounted. We discuss here the strategies used to meet the underlying assumptions of a two-sample capture-recapture, the difficulties of matching individuals between lists in areas where birth and death certificates are not regularly obtained and spelling of names is fluid, and methods to reduce differential ascertainment of demographic events by age and sex.


MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 REFERENCES
 
Study site. Approximately 60,000 people living in the rural Rarieda Division (Asembo) of Kenya, which is located along the shores of Lake Victoria approximately 45 km from Kisumu City, and 70,000 people in the adjacent Wagai and Yala Divisions (Gem), were under demographic surveillance (beginning in 1997) to evaluate the impact of (ITNs) on all-cause child mortality.19 The characteristics of the population of this area have been described in detail elsewhere.20 Briefly, the Luo are the predominant tribe and most are subsistence farmers, although fishing is also a significant source of income. Polygamy is common and a husband and his wives generally live in separate houses within a single compound.

Mapping. All houses in the Asembo study area were mapped using a global positioning system between April and November 1999, and updated in 2001. Details of the mapping methodology have been described elsewhere.20,21 Between January and November 2000, only compounds in Gem, rather than individual houses, were mapped due to the size of the population. In both Asembo and Gem, each house was identified by a unique code comprised of village, compound and household identifiers. Data were entered into ArcInfo (Environmental Systems Research Institute, Redlands, CA), a geographic information system (GIS). The GIS databases were created independently and maintained separately from databases generated by the DSS.

The capture: DSS. The baseline census of the population was conducted in February–April and July–November 1997 in Asembo and Gem, respectively. Demographic changes in the study population were then monitored through biannual enumerations of residents of the study area. Details of the methods are given elsewhere.22 Briefly, after all households and individuals were enumerated in the baseline survey, households were visited twice per year to record any migrations, births or deaths within the study area since the previous DSS round. New households were also registered at the time of each DSS round. Households were visited by local residents used and trained by the project as enumerators. The same unique household codes used in the GIS database were used to identify the homes of individuals enumerated in the DSS. Three names were recorded for each household resident: Christian name, traditional Juok (dhoLuo) name, and Juok name of the individual’s father or husband. Dates of birth and/or death were obtained from the household respondent because neither birth nor death certificates are commonly obtained in this area. Preprinted census forms containing details of all residents alive and present at the time of the previous DSS round were used for identifying and updating individuals at each subsequent DSS round. All new births, as well as any new immigrants who satisfied residency requirements, were registered during each DSS round.

The list of births and deaths in children less than five years old occurring between March 1 and September 30, 2000 in Asembo, and those between January 1 and July 31, 2001 in Gem, served as the capture lists. Births and deaths in the Asembo DSS round were recorded in October 2000, while births and deaths in the Gem DSS round were recorded in August 2001.

The recapture. The recapture list is intended to be an independent register of the event of interest, such that the probability of an event being recorded on the first list is independent of the event being recorded on the second list. Due to the known undercount of neonates in civil registration data in Kenya,23 we chose not to use an existing list of births and deaths from civil registration data, but to create a recapture list by conducting a survey of 5,000 randomly selected households in both Asembo and Gem. To maximize independence between the list generated from the DSS and the recapture list, households registered in the Asembo GIS database served as the sampling frame for Asembo. However, because GIS mapping in Gem was conducted at the compound level, a household sampling frame for Gem was created by merging the GIS compound database with the DSS household database.

Based on the number of births and deaths reported in previous DSS rounds, a sample of 5,000 households in each study area was intended to yield approximately 100 deceased children and 250 new births over a seven-month interval. The recapture surveys were conducted by a different group of trained enumerators from those who had participated in the DSS. All new births and deaths in children less than five years old that occurred during the same interval covered by the DSS round were recorded during the corresponding recapture survey in the same manner as for the DSS rounds in November 2000 in Asembo and August–September 2001 in Gem.

Matching events between lists. Estimation of the total number of events occurring in a population requires the total number of events identified on each list separately, as well as the number of events identified by both. Matching newborn or deceased children between each DSS round and corresponding recapture survey was first done using a matching program created in SAS (Version 8.1; SAS Institute, Inc., Cary, NC). The program was first set to identify individuals present on both lists via exact matches on a child’s three names, household identifiers, sex, dates of birth (and death), and two of the child’s mother’s names. Increasingly lenient SAS matching programs were also used to allow less strict criteria for matching individual characteristics between lists. However, because of the variability in spelling of names and poor date recall, visual matching of individuals between lists was ultimately undertaken using a preset algorithm. A match between newborns across lists required the following: two of three names of the child; same date of birth within two months; and at least one of the mother’s Christian or Juok names. A match between deceased children required the following: two of three names of child; same year of birth; date of death within two months; and at least one of the mother’s Christian or Juok names. A match between names did not require exact spelling. Because children in this area frequently spend time outside their compound of residence with extended family members, a match did not require the same household address. However, only 3% of all matches did not have the same village and compound.

Estimation methods. We assumed that the probability that a vital event was missed by a DSS round was independent of the probability of it being missed by the corresponding recapture survey, and thus PB = PC x PR, where PB is the probability that an event was missed by both systems, PC is the probability that an event was missed by the DSS round, and PR is the probability that it was missed by the recapture survey. We estimated the total number of events missed by each DSS round and corresponding recapture survey using the Peterson maximum likelihood estimator for two independent samples,24,25 calculated as

where C is the number of events identified only by the DSS round, R is the number of events identified only by the recapture survey, and B is the number of events identified by both systems. Thus, estimates of the total number of events that actually occurred over the designated time periods were obtained from the sum of all events identified, plus those estimated to have been missed. Due to the relatively small number of deaths on each list when disaggregated by age and sex, we used the nearly unbiased estimator for small sample sizes to obtain estimates for the number of deaths missed at each DSS round by substituting B + 1 in the denominator of the Peterson estimator, as has been previously recommended.26,27

Because dates of birth and death were matched within two months, age at death was calculated as the month of death minus the month of birth. When a child matched between lists had conflicting dates of birth or death that resulted in the child falling into two different age categories, age was based on dates listed on the DSS database. Neonates are defined here as less than one month old.

To minimize the effect of heterogeneity, the total number of missing births and deaths at each DSS round were estimated using the sum of stratified data (by sex for births and by age and sex for deaths), as has been previously recommended.24 Ninety-five percent confidence intervals (CIs) were derived using a goodness-of-fit method described by Regal and Hook.28

The sensitivity of each DSS round in capturing births and deaths was estimated as the number of events identified by the DSS round divided by the total number of events estimated by the overlap between lists.

Ethical clearance. This research was approved by the institutional review boards of the Kenya Medical Research Institute (Nairobi, Kenya) and the Centers for Disease Control and Prevention (Atlanta, GA).


RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 REFERENCES
 
The Asembo DSS round identified 218 child deaths and 1,260 births, while the Asembo recapture survey identified 74 child deaths and 456 births over the same time period. The Gem DSS round identified 371 child deaths and 1,364 births, while the Gem recapture survey identified 60 child deaths and 245 births over the same time period.

Overall, the distribution of deaths by age was similar between the Asembo and Gem DSS rounds and recapture surveys (Table 1Go). While the distribution of deaths by sex was identical between the Asembo DSS round and recapture survey, the Gem recapture survey identified a higher proportion of male deaths compared with the Gem DSS round. The proportion of male and female births identified by the DSS rounds and recapture surveys was similar in both study sites.


View this table:
[in this window]
[in a new window]
 
TABLE 1
Age and sex of deceased children (0–59 months old) and newborns identified by the demographic surveillance system (DSS) and recapture surveys, Asembo and Gem, western Kenya, 2000–2001
 
Matching via computer program was difficult due to the variability in the spelling of names and inaccurate recall of dates of birth and death. At best, only 2% of deaths between each DSS round and corresponding recapture survey could be matched in this manner. Matching via visual inspection yielded a total of 400 and 190 matches of newborns, and 52 and 45 matches of deceased children less than five years old, between the DSS and recapture lists in Asembo and Gem, respectively.

We estimated that a total of 1,436 (95% CI = 1,400–1,481) and 1,769 (95% CI = 1,662–1883) births occurred during the periods covered by the Asembo and Gem DSS rounds, respectively (Table 2Go). The sensitivities of the DSS rounds at identifying births were therefore estimated to be 88% and 77%, respectively. Sensitivity of the DSS rounds did not differ significantly according to the sex of newborns.


View this table:
[in this window]
[in a new window]
 
TABLE 2
Number of births registered by the demographic surveillance system (DSS) and recapture surveys, estimated total number of births, and sensitivity of the DSS by sex in Asembo and Gem, western Kenya, 2000–2001*
 
We estimated that 306 (95% CI = 278–360) child deaths (0–59 months old) occurred during the period covered by the Asembo DSS round, yielding a sensitivity of 71% (Table 3Go). In Gem, results suggested that 492 (95% CI = 442–581) child deaths occurred, yielding a similar sensitivity of 75%.


View this table:
[in this window]
[in a new window]
 
TABLE 3
Number of deaths registered by the demographic surveillance system (DSS) and recapture surveys, estimated total number of deaths, and sensitivity of the DSS by age and sex in Asembo and Gem, western Kenya, 2000–2001*
 
The sensitivity of the two DSS rounds generally increased with age at death (Table 3Go). Deaths among neonates were poorly identified in both the Asembo and Gem DSS rounds; calculated sensitivities were 62% and 49%, respectively. When neonatal deaths were excluded from the stratified analysis, the sensitivity of the Asembo DSS in identifying post-neonatal child deaths was 72% while the sensitivity of the Gem DSS round was 78%. In both the Asembo and Gem DSS rounds, the DSS was more sensitive in identifying male deaths (sensitivities of 81% each) than female (62% and 70%, respectively).

We investigated why some child deaths had been missed in the DSS rounds but were found by the recapture survey. Most (92%) of the deceased children missed by the two DSS rounds had never been registered in previous rounds as either new births or new residents, and thus were not present on the pre-printed census forms. There were no clear or consistent reasons why these children were left out of the DSS registration because in many cases, both their parents and siblings had been registered.


DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
REFERENCES
 
A number of different methods have been used to obtain all-cause mortality rates, including indirect methods such as birth histories,29,30 monitoring mortality events in cohorts,31–33 and community-based vital registration.34,35 However, these methods may not be satisfactory because of selection bias, small sample sizes, or inadequate coverage. For this reason, DSS, which involve longitudinal monitoring of a large, geographically defined population, have become the gold standard for evaluating the impact of community-based health interventions aimed at reducing mortality in developing countries.1–9

Although DSS perform very well in registering vital events over time, they are not 100% accurate and may ascertain events differentially according to demographic characteristics, geographic location, or season. We used a two-sample capture-recapture to evaluate the completeness of two rounds of a DSS in western Kenya and to determine if there was any bias in ascertainment of vital events by age and sex. The sensitivities of the Asembo and Gem DSS rounds in identifying child deaths (0–59 months old) were estimated to be 71% and 75%, while the sensitivities of the two rounds in identifying recent births were estimated to be 88% and 77%, respectively. In neither location did the DSS perform well in identifying neonatal deaths, and sensitivities generally increased with the age at which a child died. Additionally, female deaths were more likely to go unreported.

Although several studies have attempted to estimate the completeness of vital event registration systems,36,37 we found only one that has used a capture-recapture approach.38 As part of a large-scale ITN trial in Burkina Faso,8 Diallo and others used overlap between two concurrent demographic registration systems, a DSS and village-based vital registration, to evaluate the completeness of each method for ascertainment of births and child deaths. They found that the sensitivity of their DSS for identifying births and child deaths (<5 years old) was 94% and 88%, respectively. Similar to our results, they also reported that the DSS performed less well at identifying deaths among children less than six months old.

The use of a two-sample capture-recapture to determine the sensitivity of a DSS requires several assumptions. First, the population under study must be closed, meaning there can be no entries or losses of individuals during the study period. This assumption is unlikely to hold true in most human populations. However, by assessing DSS rounds that occur over relatively short time intervals within a well-defined area, and having recapture surveys that take place immediately after the conclusion of the DSS round, the likelihood of in- or out-migration among children and family informants is minimized.

Second, the two lists must be independent with respect to the probability of ascertainment. To maximize independence between lists, different interviewers were used to collect the data for the recapture surveys from those used during the DSS rounds. Additionally, households were selected for the recapture survey from a list of households created separately from the DSS database. Although it is impossible to identify any specific effects of dependence when using only two lists,24,25 we found no evidence to suggest negative or positive dependence between being identified by the DSS rounds and their corresponding recapture surveys.

Third, each individual/event in the population must have an equal probability of ascertainment, oftentimes referred to as a lack of heterogeneity. To limit the potential effect of heterogeneity, we report results from a stratified analysis. Because the probability of capturing births was affected by sex, and both sex and age affected the probability of capturing deaths, the estimated total number of newborns was based on the sum of estimates among males and females, while the total number of child deaths was based on the sum of estimates disaggregated by sex as well as age.

Last, it must be possible to match events between lists to use a capture-recapture method. Matching newborns and deceased children between lists proved extremely challenging in our study population. We found that using a computer algorithm to match individuals between lists using unique name and demographic identifiers was not useful due to the inconsistency of spelling and inaccurate recall of dates. We therefore matched children between lists by visual inspection on a case-by-case basis. This method was tedious and time-consuming, and would most likely be impracticable for frequent monitoring of a DSS.

There are several steps that could be taken to facilitate matching of individuals between DSS and recapture surveys. Distribution of study identification cards (IDs) would permit consistent identification of study subjects between the DSS and recapture samples. Study IDs have been used successfully in a number of DSS, but implementing them may depend on national laws, local customs, and linkage of services with the DSS, as well as logistics. An alternative to DSS ID cards could be the use of a name dictionary to standardize the spelling of common names. In areas without widespread registration of births and deaths, DSS study staff could also facilitate obtaining birth and death certificates to minimize the problem of date recall.

The finding of biased ascertainment by age and sex pointed to several potential means of improving DSS methodology. Identifying and monitoring pregnancies among women enrolled in this DSS, which was not done during the two DSS rounds included in this analysis, would increase the likelihood of identifying both births and neonatal deaths. Furthermore, the addition of an independent, real-time village reporting system may also improve the likelihood of identifying births and deaths of children that occur between DSS rounds. The problem of differential ascertainment of deaths by sex is not one that is easily solved. However, improving registration of births through monitoring of pregnancies would likely improve the registration of all children who are born and die between DSS rounds.

The results of this study should not be interpreted to suggest that child mortality rates obtained from a DSS with <100% ascertainment are necessarily incorrect. We found that nearly all (92%) individuals whose deaths were missed by the DSS were themselves missing from the population database. Our investigation of the children missed by the DSS could find no specific reasons for their absence besides random data collection errors. Therefore, a mortality rate derived from these data would be less precise than if all deaths and individuals had been registered, but likely would not be biased because there was no indication that the mortality rate among children missed completely by the DSS was different from those included.

Results presented here suggest that capture-recapture may prove to be a worthwhile tool for assessing completeness and differential ascertainment of vital events between subgroups of the population monitored with a DSS. In many developing countries, care will have to be taken to improve the facility with which names and dates can be matched. Information from capture-recapture analyses may then suggest changes to DSS methodology to improve registration of vital events in specific segments of the study population, thereby improving precision in measuring mortality rates overall.

Received February 4, 2003. Accepted for publication April 26, 2003.

Acknowledgments: We express our gratitude to the villagers of Asembo and Gem for their participation in this research. George Olang, James Kwach, Michael Onyango, Richard Otieno, and Maurice Ombok are thanked for their field and data management skills. We are grateful to the Centers for Disease Control and Prevention/Kenya Medical Research Institute administrative team for their support. We thank Dr. William A. Hawley for reviewing the manuscript, and the Director of the Kenya Medical Research Institute for his permission to publish this manuscript.

Financial support: This research was funded by the United States Agency for International Development.

Disclaimer: The opinions or assertions contained in this manuscript are the private ones of the authors and are not to be construed as official or reflecting the views of the U.S. Public Health Service or Department of Health and Human Services.

Authors’ addresses: Thomas P. Eisele, 3243 Dellwood Road, Cleveland Heights, OH 44118. E-mail: teisele{at}tulane.edu. Kim A. Lindblade, Daniel H. Rosen, Frank Odhiambo, John M. Vulule, and Laurence Slutsker, Centers for Disease Control and Prevention/Kenya Medical Research Institute, PO Box 1578, Kisumu, Kenya.


REFERENCES
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 

  1. Rahmathullah L, Underwood BA, Thulasiraj RD, Milton RC, Ramaswamy K, Rahmathullah R, Babu G, 1990. Reduced mortality among children in southern India receiving a small weekly dose of vitamin A. N Engl J Med 323: 929–935.[Abstract]
  2. Vijayaraghavan K, Radhaiah G, Prakasam BS, Sarma KV, Reddy V, 1990. Effect of massive dose vitamin A on morbidity and mortality in Indian children. Lancet 336: 1342–1345.[ISI][Medline]
  3. Anonymous, 1993. Vitamin A supplementation in northern Ghana: effects on clinic attendances, hospital admissions, and child mortality. Ghana VAST Study Team. Lancet 342: 7–12.[ISI][Medline]
  4. Alonso PL, Lindsay SW, Armstrong JR, Conteh M, Hill AG, David PH, Fegan G, de Francisco A, Hall AJ, Shenton FC, Cham K, Greenwood BM, 1991. The effect of insecticide-treated bed nets on mortality of Gambian children. Lancet 337: 1499–1502.[ISI][Medline]
  5. D’Alessandro U, Olaleye BO, McGuire W, Langerock P, Bennett S, Aikins MK, Thomson MC, Cham MK, Cham BA, Greenwood BM, 1995. Mortality and morbidity from malaria in Gambian children after introduction of an impregnated bednet programme. Lancet 345: 479–483.[ISI][Medline]
  6. Binka FN, Kubaje A, Adjuik M, Williams LA, Lengeler C, Maude GH, Armah GE, Kajihara B, Adiamah JH, Smith PG, 1996. Impact of permethrin impregnated bednets on child mortality in Kassena- Nankana district, Ghana: a randomized controlled trial. Trop Med Int Health 1: 147–154.[ISI][Medline]
  7. Nevill CG, Some ES, Mung’ala VO, Mutemi W, New L, Marsh K, Lengeler C, Snow RW, 1996. Insecticide-treated bednets reduce mortality and severe morbidity from malaria among children on the Kenyan coast. Trop Med Int Health 1: 139–146.[ISI][Medline]
  8. Habluetzel A, Diallo DA, Esposito F, Lamizana L, Pagnoni F, Lengeler C, Traore C, Cousens SN, 1997. Do insecticide-treated curtains reduce all-cause child mortality in Burkina Faso? Trop Med Int Health 2: 855–862.[ISI][Medline]
  9. Schellenberg JA, Abdulla S, Nathan R, Mukasa O, Marchant T, Kikumbih N, Mushi A, Mponda H, Minja H, Mshinda H, Tanner M, Lengeler C, 2001. Effect of large-scale social marketing of insecticide-treated nets on child survival in rural Tanzania. Lancet 357: 1241–1247.[ISI][Medline]
  10. Phillips J, 1998. The "Perfect DSS". INDEPTH Network Constituting Conference. Dar es Salaam, Tanzania, 1–16.
  11. Indepth-Network, 2002. Indepth-network website <http://www.indepth-netwrok.net/>.
  12. Doscher ML, Woodward JA, 1983. Estimating the size of sub-populations of heroin users: applications of log-linear models to capture/recapture sampling. Int J Addict 18: 167–182.[ISI][Medline]
  13. Mastro TD, Kitayaporn D, Weniger BG, Vanichseni S, Laosunthorn V, Uneklabh T, Uneklabh C, Choopanya K, Limpakarnjanarat K, 1994. Estimating the number of HIV-infected injection drug users in Bangkok: a capture—recapture method. Am J Public Health 84: 1094–1099.[Abstract/Free Full Text]
  14. Maxwell JC, 2000. Methods for estimating the number of "hard-core" drug users. Subst Use Misuse 35: 399–420.[ISI][Medline]
  15. Rubin G, Umbach D, Shyu SF, Castillo-Chavez C, 1992. Using mark-recapture methodology to estimate the size of a population at risk for sexually transmitted diseases. Stat Med 11: 1533–1549.[ISI][Medline]
  16. Gill GV, Ismail AA, Beeching NJ, 2001. The use of capture-recapture techniques in determining the prevalence of type 2 diabetes. QJM 94: 341–346.[Free Full Text]
  17. Modesitt SK, Hulman S, Fleming D, 1990. Evaluation of active versus passive AIDS surveillance in Oregon. Am J Public Health 80: 463–464.[Abstract/Free Full Text]
  18. Berghold A, Stronegger WJ, Wernecke KD, 2001. A model and application for estimating completeness of registration. Methods Inf Med 40: 122–126.[ISI][Medline]
  19. Anonymous, 2000. 1999 Population and Housing Census: Volume 1. Nairobi, Kenya: Republic of Kenya: Central Bureau of Statistics, Ministry of Finance and Planning.
  20. Phillips-Howard PA, Nahlen BL, Alaii JA, ter Kuile FO, Gimnig JE, Terlouw DJ, Kachur SP, Hightower AW, Lal AA, Schoute E, Oloo AJ, Hawley WA, 2003. The efficacy of permethrin-treated bed nets on child mortality and morbidity in western Kenya. I. Development of infrastructure and description of study site. Am J Trop Med Hyg 68(Suppl 4): 3–9.[Abstract/Free Full Text]
  21. Hawley WA, Phillips-Howard PA, ter Kuile FO, Terlouw DJ, Vulule JM, Ombok M, Nahlen BL, Gimnig JE, Kariuki SK, Kolczak MS, Hightower AW, 2003. Community-wide effects of permethrin-treated bed nets on child mortality and malaria morbidity in western Kenya. Am J Trop Med Hyg 68(Suppl 4): 121–127.[Abstract/Free Full Text]
  22. Phillips-Howard PA, Ter Kuile FO, Nahlen BL, Alaii JA, Gimnig JE, Kolczak MS, Terlouw DJ, Kariuki SK, Shi YP, Kachur SP, Hightower AW, Vulule JM, Hawley WA, 2003. The efficacy of permethrin-treated bed nets on child mortality and morbidity in western Kenya. II. Study design and methods. Am J Trop Med Hyg 68(Suppl 4): 10–15.[Abstract/Free Full Text]
  23. Arudo J, Gimnig JE, ter Kuile FO, Kachur SP, Slutsker L, Kolczak MS, Hawley WS, Orago ASS, Nahlen BL, Phillips-Howard PA, 2003. Comparison of government statistics and demographic surveillance to monitor mortality in children less than five years old in western Kenya. Am J Trop Med Hyg 68(Suppl 4): 30–37.[Abstract/Free Full Text]
  24. Hook EB, Regal RR, 1995. Capture-recapture methods in epidemiology: methods and limitations. Epidemiol Rev 17: 243–264.[Free Full Text]
  25. Chao A, Tsay PK, Lin S, Shau W, Chao D, 2001. Tutorial in biostatistics: the application of capture-recapture models to epidemiological data. Stat Med 20: 3123–3157.[ISI][Medline]
  26. Chapman DG, 1951. Some properties of the hypergeometric distribution with applications to zoological sample censuses. Univ Calif Public Stat 1: 131–160.
  27. Wittes JT, 1972. On the bias and estimated variance of Chapman’s two-sample capture-recapture population estimate. Biometrics 28: 592–597.
  28. Regal R, Hook E, 1984. Goodness-of-fit based confidence intervals for estimates of the size of a closed population. Stat Med 3: 287–291.[ISI][Medline]
  29. Becker SR, Diop F, Thornton JN, 1993. Infant and child mortality in two counties of Liberia: results of a survey in 1988 and trends since 1984. Int J Epidemiol 22(Suppl 1): S56–S63.[Abstract]
  30. Nannan N, Bradshaw D, Mazur R, Maphumulo S, 1998. What is the infant mortality rate in South Africa? The need for improved data. S Afr Med J 88: 1583–1587.[ISI][Medline]
  31. McElroy PD, ter Kuile FO, Hightower AW, Hawley WA, Phillips-Howard PA, Oloo AJ, Lal AA, Nahlen BL, 2001. All-cause mortality among young children in western Kenya. VI: the Asembo Bay Cohort Project. Am J Trop Med Hyg 64: 18–27.[Abstract/Free Full Text]
  32. van den Broeck J, Eeckels R, Massa G, 1996. Maternal determinants of child survival in a rural African community. Int J Epidemiol 25: 998–1004.[Abstract/Free Full Text]
  33. Watts T, Ng’andu N, Wray J, 1989. Mothers in an urban township in Zambia. J Trop Pediatr 35: 117–120.[Abstract/Free Full Text]
  34. De Francisco A, Hall AJ, Schellenberg JR, Greenwood AM, Greenwood BM, 1993. The pattern of infant and childhood mortality in Upper River Division, The Gambia. Ann Trop Paediatr 13: 345–352.[ISI][Medline]
  35. Salum FM, Wilkes TJ, Kivumbi K, Curtis CF, 1994. Mortality of under-fives in a rural area of holoendemic malaria transmission. Acta Trop 58: 29–34.[ISI][Medline]
  36. Mung’ala VO, Snow RW, 1994. Death registration on the Kenyan Coast. East Afr Med J 71: 747–750.[ISI][Medline]
  37. Binka FN, Maude GH, Gyapong M, Ross DA, Smith PG, 1995. Risk factors for child mortality in northern Ghana: a case-control study. Int J Epidemiol 24: 127–135.[Abstract/Free Full Text]
  38. Diallo DA, Habluetzel A, Esposito F, Cousens SN, 1996. Comparison of two methods for assessing child mortality in areas without comprehensive registration systems. Trans R Soc Trop Med Hyg 90: 610–613.[ISI][Medline]




This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via ISI Web of Science (1)
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by EISELE, T. P.
Right arrow Articles by SLUTSKER, L.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by EISELE, T. P.
Right arrow Articles by SLUTSKER, L.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS