HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

Electronic Letters to:

RESEARCH AND PRACTICE: Isabelle L. Horon Underreporting of Maternal Deaths on Death Certificates and the Magnitude of the Problem of Maternal Mortality Am J Public Health 2005; 95: 478-482 [Abstract] [Full text] [PDF]

eLetters: Submit a response to this article

Electronic letters published:

THE MAGNITUDE OF THE PROBLEM OF MATERNAL MORTALITY

Oscar J Mujica   (31 March 2005)

THE MAGNITUDE OF THE PROBLEM OF MATERNAL MORTALITY 31 March 2005
Oscar J Mujica, Epidemiologist Pan American Health Organization, PAHO/WHO Send letter to journal: Re: THE MAGNITUDE OF THE PROBLEM OF MATERNAL MORTALITY mujicaos{at}paho.org Oscar J Mujica	THE MAGNITUDE OF THE PROBLEM OF MATERNAL MORTALITY   Dear Editor,   In the March issue of the Journal, Horon1 suggests the possibility that not all maternal deaths were identified in her study “even by using additional data sources”. Accordingly, the adjusted maternal mortality ratio in Maryland for the years 1993 through 2000 of 22.2 per 100 000 live births —which is 60.9% higher than the ratio based only on information reported in death records— may still underestimate the true figure.   It is, in fact, possible to estimate the magnitude of this ‘residual under-registration’ by directly fitting Horon’s data into a log-linear capture-recapture model,2 an epidemiological tool for estimating or adjusting for the extent of incomplete ascertainment using information from overlapping lists of cases from distinct sources.3   According to the results of this analysis, obtained with the help of EpiDat ® —a software in the public domain4 (Table 1), an additional 12 maternal deaths may have been expected, raising the total to 141 (95% confidence interval 132 to 152) maternal deaths between 1993 and 2000. Therefore, the magnitude of the problem may well be closer to 24.3 per 100 000 [95%CI: 22.7 to 26.2] in Maryland. If maternal deaths are assumed to be underreported at the same level nationally as they are in Maryland, under similar source dependency assumptions the maternal mortality ratio for the United States for 2001 would have been 17.4 per 100 000 live births.   Dr. Horon’s study succeeds in providing strong evidence for the critical need of improving data collection procedures and analysis to better understand this worldwide devastating problem as well as to monitor progress towards its control.    Oscar J Mujica, MD MPH PHE   Acknowledgment The author would like to thank Cesar B. Vieira MD MPH MPHI, for his comments on a draft of this letter.   References 1.    Horon IL. Underreporting of Maternal Deaths on Death Certificates and the Magnitude of the Problem of Maternal Mortality. Am J Public Health 2005;95(3):478-82. 2.    Chao A, Tsay PK, Lin SH, Shau WY, Chao DY. The applications of capture-recapture models to epidemiological data. Statistics in Medicine 2001;20:3123-57. 3.    Hook EB, Regal RR. Capture-recapture methods in epidemiology: methods and limitations. Epidemiol Rev 1995;17(2):243-64. [errata: Am J Epidemiol 1998;148(12):1219.] 4.    EpiDat: programa para análisis epidemiológico de datos tabulados, versión 3.0 [EpiDat: software for epidemiological analysis of tabulated data, version 3.0]. Consellería de Sanidade de la Xunta de Galicia y Organización Panamericana de la Salud. Santiago de Compostela & Washington DC; 2003 (downloadable at: http://dxsp.sergas.es)     Table 1.            Capture-recapture three-source log-linear model applied to Horon’s data.       source 3     yes   no     source 2   source 2     yes no   yes no source 1 yes 26 21   23 10 no 18 18   13 X   source 1: death certificates (n=80) source 2: linkage of records (n=80) source 3: medical examiner records (n=83)    dependency models X N CI (95%) G2 df BIC completeness of source (%) 1 2 3 1,2 1,3 2,3 1,2,3 1, 2 & 3 independent 10 139 132 147 3.07 3 -5.99 57.6 57.6 59.7 79.9 83.5 85.6 92.8 1 & 2 independent from 3 12 141 132 152 1.72 2 -4.33 56.7 56.7 58.9 78.7 82.3 84.4 91.5 1 & 3 independet from 2 9 138 130 147 2.99 2 -3.05 58.0 58.0 60.1 80.4 84.1 86.2 93.5 2 & 3 independent from 1 7 136 129 143 0.78 2 -5.26 58.8 58.8 61.0 81.6 85.3 87.5 94.9 1 independent from 2 5 134 128 141 0.23 1 -2.79 59.7 59.7 61.9 82.8 86.6 88.8 96.3 1 independent from 3 8 137 129 147 0.19 1 -2.83 58.4 58.4 60.6 81.0 84.7 86.9 94.2 2 independent from 3 13 142 129 155 1.71 1 -1.31 56.3 56.3 58.5 78.2 81.7 83.8 90.8 1, 2 & 3 dependent (saturated) 6 135 126 146 0.00 0 0.00 59.3 59.3 61.5 82.2 85.9 88.1 95.6   X: unobserved cell (expected number of cases undetected by the three sources) N: total population (estimated ‘true’ total number of cases or population size) G2: likelihood ratio estimator df: degree of freedom BIC: Bayesian information criterion