Assessing the continuum of care for maternal health in Mexico, 1994–2018

Abstract Objective To describe the temporal and geographical patterns of the continuum of maternal health care in Mexico, as well as the sociodemographic characteristics that affect the likelihood of receiving this care. Methods We conducted a pooled cross-sectional analysis using the 1997, 2009, 2014 and 2018 waves of the National Survey of Demographic Dynamics, collating sociodemographic and obstetric characteristics of 93 745 women aged 12–54 years at last delivery. We defined eight variables along the antenatal–postnatal continuum, both independently and conditionally. We used a pooled fixed-effects multivariable logistic model to determine the likelihood of receiving the continuum of care for various properties. We also mapped the quintiles of adjusted state-level absolute change in continuum of care coverage during 1994–2018. Findings We observed large absolute increases in the proportion of women receiving timely antenatal and postnatal care (from 48.9% to 88.2% and from 39.1% to 68.7%, respectively). In our conditional analysis, we found that the proportion of women receiving adequate antenatal care doubled over this period. We showed that having social security and a higher level of education is positively associated with receiving the continuum of care. We observed the largest relative increases in continuum of care coverage in Chiapas (181.5%) and Durango (160.6%), assigned human development index categories of low and medium, respectively. Conclusion Despite significant progress in coverage of the continuum of maternal health care, disparities remain. While ensuring progress towards achievement of the health-related sustainable development goal, government intervention must also target underserved populations.


Introduction
Despite significant progress in the provision of maternal health care in low-and middle-income countries as a result of the millennium and sustainable developments goals (SDGs), 1 significant challenges in the provision of both maternal and universal health coverage (UHC) remain. Gaps in the coverage of important maternal health interventions [2][3][4][5] that are closely associated with social vulnerability have also been identified. [6][7][8][9][10][11] Effective policies to improve health outcomes and promote the full implementation of UHC require changes in procedures from monitoring crude coverage to quality-adjusted coverage. 12 Recent studies have suggested that current methods of measuring intervention coverage for reproductive and maternal health do not adequately determine the quality of services delivered; without information on the quality of care, it is difficult to assess expected health improvements. 13,14 These recent studies have also quantified the alarming discrepancies between the impact on women's health as measured from crude coverage indicators and the impact as measured from contact coverage indicators (i.e. that represent the delivery and benefits from high-quality services). 12 Widely accepted as a proxy for quality-adjusted coverage indicators, the continuum of care principle for maternal, newborn and child health aims to reduce the burden of maternal and child mortality by integrating health services throughout the life cycle. [15][16][17][18] According to Kerber's framework, the continuum of care has two dimensions: 15 (i) time, which refers to the linking of health care during adolescence and prepregnancy through childbirth, the immediate postnatal period and childhood; and (ii) place, which refers to the linking of health care that is provided across different environments, including households, communities and clinical care at different levels. 15,19,20 The continuum of care therefore aims to provide women with reproductive health services and newborns with the opportunity of a healthy childhood, 15,20 but also to ensure that services are delivered in an integrated way to avoid inefficiency, control associated costs, 15,20 and minimize maternal and neonatal mortality. 16 Over the past 25 years, 21 the priorities of maternal, newborn and child health have emerged as key within the Mexican health-care system, and financial protection was provided in the form of the Seguro Popular de Salud. The now-obsolete government-financed Seguro Popular de Salud, a voluntary family health insurance programme for those without social security (i.e. the self-employed, underemployed and unemployed), was operational in 2003 [22][23][24] until it was replaced by the Health Institute for Wellbeing by a new administration in early 2020. To inform health-care policies as we work towards the 2030 target of the SDGs, we need a comprehensive assessment of progress made and challenges remaining in monitoring the quality-adjusted coverage of maternal health care. Our aims are therefore to: (i) describe the temporal and geographical trends in the provision of the continuum of maternal health care in Mexico during the past 25 years; and (ii) determine the sociodemographic and obstetric characteristics that affect the likelihood of this continuum of care being received.
Continuum of maternal health care, Mexico Edson Serván-Mori et al.

Study design
We conducted a pooled cross-sectional analysis using the 1997, 2009, 2014 and 2018 waves of the National Survey of Demographic Dynamics. 25 Implemented by the National Institute of Statistics and Geography of Mexico, these crosssectional, probabilistic, retrospective population-based surveys are representative at both the national and state level and across different residential areas. 25 The four surveys include the sociodemographic and reproductive characteristics We obtained data on the sociodemographic characteristics of the survey participants at an individual and place of residence or contextual level, including age at time of last delivery, whether at least one indigenous language spoken, marital status, level of education, whether recently employed and health insurance status at the time of the survey. We also recorded obstetric information, such as: whether primiparous; whether the woman had experienced an infant death, miscarriage or abortion, or a health problem during pregnancy or childbirth; and type of delivery. At the household level, we included a factorial asset and housing material index as a measure of socioeconomic status. We used this index to stratify participants over five categories according to the method of Dalenius and Hodges, 26 where the higher categories indicate a greater number of assets and better housing conditions. We classified type of residence as either rural (< 2500 inhabitants), urban (2500-100 000 inhabitants) or metropolitan (> 100 000 inhabitants).

Continuum of maternal health care
Our approach focuses on the routine processes that are recommended during contact between mother and healthcare provider. However, we defined our main outcome variable as the qualityadjusted conditional coverage indicator, a measure of the receipt of high-quality services and not simply contact with a health-care provider. 12 First, we defined our eight independent coverage indicators of the continuum of care in terms of antenatal and postnatal health-care processes, that is, whether: (i) antenatal care was received; (ii) antenatal care was provided by a skilled birth attendant (doctor or nurse); Continuum of maternal health care, Mexico Edson Serván-Mori et al.
(iii) the first medical visit occurred during the first 8 weeks of pregnancy (timely antenatal care); (iv) at least five antenatal consultations were received (frequent antenatal care); (v) antenatal care included at least 75% of recommended care 17,27 (adequate antenatal care; defined as receipt of 60-80% of recommended care elsewhere 28,29 ); (vi) the delivery was attended by skilled personnel; (vii) a postnatal consultation was received; and (viii) postnatal care occurred within 15 days after delivery (timely postnatal care), according to Mexican health-care system guidelines 30 and the outcomes of our previous research. 17,31 We define all coverage indicators according to international recommendations made by the World Health Organization (WHO), 32,33 with some minor exceptions. The WHO guidelines suggest a minimum of eight antenatal care visits, 32,33 whereas Mexican health-care system guidelines recommend at least five visits. 30 Additionally, Mexican guidelines recommend that the first prenatal care visit takes place during gestational weeks 6-8, whereas WHO references the first trimester. 32,33 Similarly, the number of postnatal appointments recommended nationally are a minimum of two clinic visits, one within 15 days of the birth and the second at the end of the puerperium. 30 In contrast, the WHO guidelines recommend three visits over the same time period. 32,33 Worth noting is that the definition of what is included within recommended antenatal care has changed in Mexico during the 25-year study period. An abdominal examination was performed in the 2009 survey only, while mother's weight measurement was excluded from the 2014 survey. The 2009, 2014 and 2018 surveys included ultrasound, blood and urine tests, the prescription of vitamins and/or mineral supplementation, and human immunodeficiency virus testing. The 2018 survey included height measurement, data on fetal movement and mental health services.
We defined a further eight binary outcome variables indicating the incremental access to interventions (i)-(viii) along the antenatal-postnatal continuum. We constructed these conditional coverage indicators using the coverage cascade principle, in which receiving the care described by each separate independent indicator is conditional on receiving the care described by the preceding independent indicator. 12 We defined the proportion of women who were considered to have received a continuum of care as the proportion who received all eight antenatal-postnatal interventions.

Statistical analysis
We performed all analyses using the "svy" module and sampling weights of the statistical software Stata version 15.1 (StataCorp, College Station, United States of America). We calculated the sociodemographic and obstetric charac- CI: confidence interval. a The data set includes women who have experienced at least one pregnancy. b We classified areas with less than 2500 inhabitants as rural, areas with 2500-100 000 inhabitants as urban and areas with more than 100 000 inhabitants as metropolitan.

Research
Continuum of maternal health care, Mexico Edson Serván-Mori et al.
teristics of the 93 745 surveyed women according to the period of last delivery (1994-1997, 2004-2009, 2010-2014 and 2015-2018), then converted the survey data to weighted populationlevel estimates (population, 29 822 452) with 95% confidence intervals (CIs). 25 We then converted these estimates to percentages of the relevant population receiving each of the individual healthcare interventions. Our modelled conditional coverage indicators refer to compliance with all eight health-care interventions.
We used a pooled fixed-effects multivariable logistic model to determine which sociodemographic and obstetric characteristics affect the likelihood of receiving the continuum of maternal health care. We adjusted our model for all covariates recorded in the surveys (except for type of delivery, because of its temporality), including survey year and a binary variable for each state (i.e. state fixed effect). We reported adjusted odds ratios with their 95% CIs. We then adjusted the prevalence of receiving the continuum of care according to health insurance. We also mapped the quintiles of adjusted absolute change in receipt of the continuum of care between 1994 and 2018 at the individual state level.

Results
From the first study wave to the most recent, we observed a decrease in the proportion of women who were either married or cohabiting with their partner from 89.4% (95% CI: 88.7-90.0) in 1994-1997 to 83.0% (95% CI: 82.2-83.8) in 2015-2018 (Table 1). We also observed an increase in the proportion of women who were head of the household, from 4.8% ( We can also quantify the proportion who started to receive, but did not complete, the continuum of maternal health care during each period. For example, during 1994-1997 the proportion of women who received frequent antenatal care (defined as at least five antenatal consultations) was only 43.2% compared with the 92.9% who received at least one antenatal consultation (Table 2).
Our regression analysis showed that, compared with women aged 12-19 years, being in any of the older age groups (20-29, 30-39 and 40-54 years) was associated with a greater likelihood of receiving continuum of care coverage, after controlling for sociodemographic and obstetric characteristics (Table 3). Having social security and a level of education beyond elementary school is also associated with a greater likelihood of receiving continuum of care coverage.
In our temporal analysis, we observed an increase in national continuum of care coverage by around 30% from 1994-1997 to 2015-2018, regardless of health insurance status (Fig. 1). Assuming that most women covered by Seguro Popular de Salud were previously uninsured, Fig. 1 also highlights the large increase of 28.3% in the growth of continuum of care for uninsured women (from 19.0% to 47.3%) and of 31.1% for those changing from presumably no  We mapped the wide geographical distribution of the quintiles of absolute increase in continuum of care coverage in Fig. 2. We list both absolute and relative increases in continuum of care coverage by Mexican state in Table 4, in which we observe the largest relative increases in the states of Chiapas (assigned a human development index, HDI, of low) and Durango (medium HDI).

Discussion
This population-based study illustrates a notable improvement in the continuum of maternal health-care coverage over the last 25 years, with coverage more than doubling in all Mexican states. The data analysed also illustrate a social transformation in the living conditions of the female population in Mexico, with increased participation in higher levels of education and paid employment, and the greater economic independence these factors will bring. We observed an increase in the proportion of heads of households who are female, as well as in the single, divorced or widowed proportion of this population facing a pregnancy. This trend has also been accompanied by a decrease in total fertility rate, which is partly due to women's greater engagement in reproductive decision-making. [35][36][37] Our results showing the high numbers of women who begin but fail to complete the continuum of antenatal care are similar to a previous study, which showed that, although 84% of women received at least one antenatal care visit, only 38% received at least four visits. 7 Similar levels of loss to follow-up in the continuum of care, associated with a higher risk of maternal and neonatal complications, have been reported elsewhere. 16,38 In terms of the sociodemographic characteristics that increase the likelihood of receiving continuous care during pregnancy, childbirth and the puerperium, we observed that a higher level of educational attainment, recent employment and access to health insurance were all predictors. A study from Egypt showed similar results, and the authors suggested that educated women are more familiar with the meaning and importance of maternal health services, and may have better employment opportunities and a greater likelihood of access to medical insurance. 39 As education level increases, the social gap  between pregnant women and service providers decreases; women also become more aware of maternal health and experience an improved engagement with health-care services. 38 Our calculation of the increasing proportion of women who received the continuum of care (from less than one fifth 25 years ago to over one half in 2018) highlights the achievements of Mexico's maternal health-care policies. Since 1997, the anti-poverty programme Prospera (formerly Progresa or Oportunidades) has aimed to improve the provision and quality of basic social services, including reproductive health. Positive synergies between Prospera and Seguro Popular de Salud in the reduction of gaps in effective coverage for maternal health services have recently been assessed. 40 Another successful health-care policy is the 2001 Arranque Parejo en la Vida programme that aimed to improve access to specialized delivery care, particularly in rural areas where the highest numbers of maternal mortality are reported. The government introduced the Seguro Popular de Salud in 2003 as part of the efforts to expand health coverage for those members of the population without social security; 22,41,42 by 2018, about 45% of the Mexican population were covered by the Seguro Popular de Salud. 43 This expansion in coverage financed the trebling of the health ministry budget from 2000 to 2018, 43 allowing the provision of healthcare services to be greatly enhanced.
Despite the significant progress achieved in maternal health care over the last 25 years in Mexico, important challenges remain. First, we documented a sustained and non-desirable increase in the proportion of deliveries by caesarean section over the study period. Our finding is consistent with a global increase in caesarean section deliveries, particularly in Latin America and the Caribbean. 44,45 Although this issue in Mexico is beyond the scope of this study, some factors that may explain these results include the potential role of market forces, economic incentives and medico-legal issues in decision-making processes. 46 Second, notwithstanding the increase in coverage in the continuum of maternal care seen with the introduction of the Seguro Popular de Salud, disparities remain between those with and without access to this programme. Third, improvements are needed in the efficiency and management of resources, the quality of services, the transparency of budgeting exercises and the expansion of existing coverage. 47 This improvement is particularly important for the most vulnerable populations, such as indigenous women, those of lower socioeconomic status and adolescent women, for whom effective access to the continuum of care was the lowest.
Our study has several limitations. First, although the National Survey of Demographic Dynamics is a highquality population-based survey, our analysis is subject to potential omitted variable bias, meaning that the conclusions reached here do not have the same strength as causal inference. Second, we used self-reported measures of outcome as variables and covariables; these may be subject to memory and interpretation bias, however, particularly for the timing of the initial prenatal visit variable. Third, the temporalities of our outcome and covariates were measured at the time of the survey and not at the time of the last delivery, meaning that some outcomes may have been subject to recall bias. Fourth, the surveys did not collect information about the exact locations at which prenatal, delivery and postpartum services were provided; it is therefore possible that women with social security may have received a particular type of health care at a facility not generally associated with this form of health care, potentially biasing our estimates of state-level continuum of care. Fifth, the changing definition of antenatal care as maternal health policies were updated and improved during our study period means that our antenatal care coverage figures for the earlier survey waves may be overestimated, indicating that our calculated increases in continuum of maternal health-care coverage over the 25-year period may be underestimated. Sixth, we have explored some objective indicators of quality of care focused on the care process, but have not provided evidence on subjective indicators of quality; however, prior studies have shown that large populations receiving reproductive health services in Mexico perceive that interactions with health personnel are inadequate. 48,49 Regardless of the achievements depicted by our results, the ability to improve health care is also dependent upon the performance of health personnel. Finally, our data on postnatal care are limited. For instance, women may confuse a visit for their new infant with their own postpartum visit; in addition, we did not consider the satisfaction of care received at postpartum visits.