study in Mexico that it is often sold on a pill by pill basis, which may explain the drop in sales. 22 The present study provides new national and regional estimates for 2006. It uses the same methodology as the 1990 study, but adapts those methods when necessarymost importantly by incorporating the use of misoprostol to induce abortion, a practice that was rare in the earlier survey period. This approach provides for comparability, and enables us to assess trends in induced abortion in Mexico over the past decade and a half. In addition to looking at changes in abortion incidence between 1990 and 2006, we examine patterns in hospitalization due to abortion-related complications, a key indicator of morbidity resulting from unsafe abortion. Finally, we explore the relationship between contraceptive use and differences in abortion incidence among the four regions and at the national level, and discuss the broader relevance and implications of our findings.
DATA AND METHODS
Data SourcesWe used two data sources for estimating abortion incidence: hospital discharge data on the number of women treated for abortion complications in 2006, and a survey of key informants who were knowledgeable about abortion provision in Mexico to obtain an estimate of the proportion of women who get abortions who are hospitalized.•Hospital discharge data. Data from Mexico's National System of Health Information (Sistema Nacional de Información en Salud) on the number of women treated in publicsector hospitals for abortion complications in 2006 were aggregated for seven hospital systems (see Web site Appendix at http://www.guttmacher.org/pubs/ifpp/appendix/ 3404.pdf). 27 We examined the data for quality and completeness and to ensure comparability with data for 1990. To obtain a count of patients treated for postabortion complications in 2006 that was comparable to the count used in 1990, we selected the appropriate diagnostic codes from the new ICD-10 classification system that matched those from the earlier ICD-9 system. The previous study had made other adjustments to the hospital discharge data to account for misclassification of codes; 16 because the 2006 data are of higher quality, these adjustments were not necessary.The sources of health systems data on hospital care changed between 1990 and 2006: In 2006, the number of women hospitalized for abortion complications was obtained from three sources (outpatient, inpatient and emergency cases), whereas a single source was used in 1990 (only inpatient cases were available at that time). The total number of women treated for abortion complications (resulting from spontaneous or induced abortion) in all components of the public-sector hospital system in 2006 was 194,774 (112,978 reported inpatients, 26,823 reported outpatients and 54,973 estimated emergency cases; see Appendix Table 1).•Health Professionals Survey. The Health Professionals Survey (HPS) was designed to assess the conditions of induced abortions; given the rise in use of contraceptives since the 1980s and the increase ...