Demand for prenatal health care in South America

Abstract

This study extends existing research on prenatal care demand to the South American countries of Bolivia, Columbia and Peru, using data from the Demographic and Health Surveys and employing two measures of prenatal care: whether prenatal care was ever initiated and a measure of prenatal care adequacy that includes information on month of initiation and number of visits. The results indicate that prenatal care demand in South America is significantly affected by a woman's age, previous pregnancy experience, education and marital status. Furthermore, household wealth and the degree of wantedness of the child significantly influence prenatal care demand. Since prenatal care use has been shown to improve infant and maternal health, there may be substantial benefits from economic and public health policies that target these determinants of prenatal care in the countries under study.

Notes

¹Prenatal care use has been studied in other South American countries – Peru (Elo, 1992) and Brazil (Burgard, 2004) – using similar data from older DHS surveys. Other developing countries have been analysed such as China (Anson, 2004), India (Maitra, 2004), South Africa (Burgard, 2004), Taiwan (Lin, 2004), Guatemala (Glei et al., 2003), Mozambique (Chapman, 2003), Turkey (Celik and Hotchkiss, 2000), the Philippines (Hotchkiss, 1998), Thailand (Raghupathy, 1996) and Malaysia (Panis and Lillard, 1994).

²All information attributed to the PHO comes from the PHO website (www.pho.org) and is for the year 2000.

³The prenatal care usage rates for the data used in this study are higher than the 2000 PHO averages (Bolivia, 83%; Columbia, 92% and Peru, 94%).

⁴The data can be downloaded from the DHS website (www.measuredhs.com).

⁵Pregnancies in women of extreme ages involve higher risk of complications (Abel et al., 2002). Due to the differences between women of normal reproductive age and those who are both younger and older, this study concentrates on women between the ages of 15 and 39 years. DHS data also include information on women who were not pregnant during the time span studied. These women are not included.

⁶A discussion of the Kessner index can be found in Kotelchuck (1994) who argues for the use of an alternative measure of prenatal care adequacy. Models are also estimated with Kotelchuck's adequacy index (available from the author); however, estimates with the Kessner index show a better fit in terms of pseudo-R ². Unfortunately, a newborn's gestation is not recorded in the DHS data; therefore, pregnancies are assumed to have gestations of 36–47 weeks. This assumption will result in women with shorter realized gestations being placed in categories below their true level of prenatal care. If the included independent variables are correlated with gestation length, the coefficients and marginal effects reported in this study will be biased.

⁷ European encompasses all European languages, but primarily indicates Castilian Spanish. Quechuan refers to one of several dozen dialects of the Quechua family and Aymaran refers to one of a few dialects of the Aymara family; both Quechuan, the language of the Incan empire, and Aymaran are spoken by indigenous peoples of the Andean region in South America. Castilian Spanish, Quechuan and Aymaran are all official languages in Bolivia and Peru, while Castilian Spanish is the only official language in Columbia. In many countries, race and ethnicity are defined along lines of skin colour, which sometimes intersects with primary language. In the DHS data analysed in this study, no indicators of skin colour are included. The Columbia survey does not ask any question related to ethnicity or language group and, thus, this variable cannot be included.

⁸The vast majority of women have either 0 or 1 previous stillbirths (Bolivia, 98%; Columbia, 99% and Peru, 99%). Thus, a dummy variable indicating the existence of previous inviable births captures most of the variation in the total number of inviable births while limiting the effects of outliers.

⁹Since the number of regions is large for each country (for instance, Columbia has data from 33 different regions), the results with respect to region are not reported in the tables. However, all estimates include dummies for region; in addition, SEs are adjusted for clustering by region. The coefficients on the regional dummies are available from the author.

¹⁰Due to the number of job categories (10) and the difficulty in interpreting coefficients, the results with respect to job category are not reported in the tables. However, all estimates include job categories, so the results are adjusted for these indicators of current income. The coefficients on the job categories are available from the author.

¹¹The statement that the Kessner Index is an ordered categorical variable is based on the theoretical assumption that U′[C_i , ; X_i , Y_i ] exhibits nonsatiation in C_i , i.e. more prenatal care unambiguously increases utility.

¹²Tests (available from the author) indicate that the countries cannot be treated as a combined sample.

¹³Ordered probit gives marginal effects for each of the three categories. For brevity, only those for the adequate category are listed. The full set of marginal effects is available from the author.