This article has Open Peer Review reports available.
Measuring progress in maternal and newborn health care in Mexico: validating indicators of health system contact and quality of care
© The Author(s). 2016
Received: 9 June 2016
Accepted: 15 August 2016
Published: 30 August 2016
The majority of births in Mexico take place in a health facility and are attended by a skilled birth attendant, yet maternal mortality has not declined to anticipated levels. Coverage estimates of skilled attendance and other maternal and newborn interventions often rely on women’s self-report through a population-based survey, the accuracy of which is not well established.
We used a facility-based design to validate women’s report of skilled birth attendance, as well as other key elements of maternal, newborn intrapartum, and immediate postnatal care. Women’s reports of labor and delivery care were collected by exit interview prior to hospital discharge and were compared against direct observation by a trained third party in a Mexican public hospital (n = 597). For each indicator, validity was assessed at the individual level using the area under the receiver operating curve (AUC) and at the population level using the inflation factor (IF).
Five of 47 indicators met both validation criteria (AUC > 0.60 and 0.75 < IF < 1.25): urine sample screen, injection or IV medication received during labor, before the birth of the baby (i.e., uterotonic for either induction or augmentation of labor), episiotomy, excessive bleeding, and receipt of blood products. An additional 9 indicators met criteria for the AUC and 18 met criteria for the IF. A skilled attendant indicator had high sensitivity (90.1 %: 95 % CI: 87.1–92.5 %), low specificity (14.0 %: 95 % CI: 5.8–26.7 %) and was suitable for population-level estimation only.
Women are able to give valid reports on some aspects of the content of care, although questions regarding the indication for interventions are less likely to be known. Questions that include technical terms or refer to specific time periods tended to have lower response levels. A key aspect of efforts to improve maternal and newborn health requires valid measurement of women’s access to maternal and newborn health interventions and the quality of such services. Additional work on improving measurement of population coverage indicators is warranted.
KeywordsMaternal and newborn health Quality of care Intervention coverage Validation Measurement Indicators Skilled birth attendance
The maternal mortality ratio in Mexico has declined by less than 7 % in the past two decades . As part of its efforts to improve maternal health, the Mexican government has emphasized increasing access to skilled obstetric care as a central focus of the national health plan [2, 3]. To scale-up access, government-led initiatives have sought to promote contact with the health system and skilled birth attendance, establishing policies such as universal health coverage for pregnant women and no-cost emergency obstetric services .
To track such efforts, the Mexican government has emphasized monitoring women’s contact with the health system . Indicators routinely collected as proxy measures for skilled obstetric care include the proportion of births delivered in health facilities and the proportion of births attended by a skilled birth attendant. In the absence of health monitoring systems that can provide accurate data on population coverage, these indicators often rely on women’s self-reports collected in household surveys such as the National Health and Nutrition Survey (Encuesta Nacional de Salud y Nutrición) in Mexico or international programs including the Demographic and Health Surveys (DHS) and Multiple Indicator Cluster Survey (MICS). However, the accuracy of women’s reports on these indicators is not well understood .
Insufficient progress in improving maternal health in Mexico and elsewhere has called into question reliance on indicators of health system contacts, rather than the content and quality of care. For example, researchers have noted that while physician-attended births were reported to be near universal in Mexico in 2012 (96 %), maternal deaths have not declined to the anticipated level [6–9].
A small, but emerging evidence base has sought to validate which aspects of maternal and newborn care women are able to report with accuracy. Validation research to date has largely taken place in settings where maternal mortality rates are low and often relies on hospital records, which often are subject to incomplete and inaccurate reporting as the reference standard. Although three rigorous studies were recently published in this area, to our knowledge, one of the most widely used indicators of health service contact – skilled attendance at birth– remains to be empirically validated [10–12]. In addition, apart from research by Tuncalp and colleagues which focused on validating women’s reports of cesarean section in the Dominican Republic , we know of little research to validate indicators of the quality and content of maternal and newborn health care in the Latin America and Caribbean region.
This study responds to the need for enhanced data related to the measurement of maternal and newborn health intervention coverage. We assessed the validity of a set of intrapartum and immediate postnatal indicators among women who delivered in a large Mexican public hospital. Our goal is to inform which aspects of intrapartum and immediate postnatal care women are able to report with accuracy and have the potential to be included in routine population-based surveys to monitor the coverage of lifesaving maternal and newborn health interventions.
As part of this work, a parallel study validated a set of quality of care indicators in Kenya . Given variation in the status of maternal health and the organization of maternal health services, these results are reported separately.
Data for the validation analysis were drawn from women’s reports on the maternal and newborn health services they received during labor and delivery gathered by exit interviews prior to their discharge from the hospital. Women’s reports were compared against observations by a trained third party using a structured checklist. Direct observation was chosen as the reference standard as it was considered to yield the most accurate reflection of all facets of the care-giving process. In the event that clarification was needed (e.g., in a few instances the newborn and mother were taken into separate rooms and the observer remained with the mother) observations were supplemented by checking medical records or by consulting providers.
Study setting and population
The study took place in a large public hospital in Mexico City where approximately 4,000 deliveries (54 % vaginal) take place annually. Hospital services include comprehensive obstetric care to women with normal pregnancies who are self-referred for admission, in combination with women with high-risk pregnancies who are referred from other public primary or secondary health care institutions. The population served by the study hospital tends to have a lower-than-average socioeconomic status and characteristically lacks health insurance. Patients may travel large distances to arrive at the study hospital with a large proportion from the neighboring State of Mexico (37 %) and 5 % from the rest of the country.
All pregnant women aged 15−49 who were admitted for delivery at the study hospital and were able to provide consent were eligible for participation. Women were excluded if they were unable to provide consent, if they presented with a complication, or if their stage of labor was considered too advanced by medical personnel. Study observation was discontinued if a cesarean section was indicated at any point.
Ethical approval and consent
Written informed consent was obtained from all participants prior to observation. For women under the age of 18, written consent was also obtained from their spouse/ common law partner or parent (as responsible parties) in accordance with local ethical guidelines . Provider consent was obtained to observe health care workers in the labor and delivery ward. A list of providers who gave their consent was provided to observers before they commenced observation. Very few providers refused participation.
The study protocol was approved by the Population Council’s Institutional Review Board and the Ethics and Research Committee of the participating hospital prior to participant enrollment.
Data collection took place between November 2013 and April 2014 and occurred over a 24-h period, using three daily observation shifts. Observations included all interactions between women and provider (s) upon admission to the labor ward until one hour following delivery. Client exit-interview questionnaires reflected the same elements of care and patient-provider interactions as the observation checklist (Additional files 1, 2, 3 and 4). Observers were general medical practitioners or nurses with sufficient clinical training to accurately record delivery room practices. Interviewers were social workers and psychologists with prior research experience. To minimize the potential for bias, study interviewers and observers were not the same individuals and all data collectors were external to the study facility. We recruited only female interviewers to facilitate rapport with participating women and interviews were conducted in Spanish. Data collectors received intensive training on the study protocol, including detailed explanation of each aspect of the client questionnaire and observation checklist to ensure full understanding, how to record responses, and the procedures for ethical research.
Indicators selected for validation were identified through a landscaping scan of published and grey literature conducted between April and July 2012, a detailed description of which is published elsewhere . Inclusion criteria for indicator selection were: reflection of the content or quality of maternal and newborn health services relating to the early labor and immediate postnatal period (up to one hour following delivery) , current or proposed use in household survey programs (e.g., DHS or MICS) or the ability to reflect critical elements of care and be recalled by women. A total of 82 indicators were selected for validity testing (Additional file 5: Table S1).
Interview questionnaires were translated into Spanish and underwent minor modifications to improve participant understanding.
The target sample size for this study was 600 women. Since no information on the coverage of maternal and newborn interventions in the facility was available, we assumed 50 % prevalence for all indicators, with 60 % sensitivity ± 6 % precision, 70 % specificity ± 6 % precision, and with α = 0.05 assuming normal approximation to the binomial distribution. Using these assumptions, a target sample size of 500 was calculated. To allow for non-response, the sample size was increased to approximately 600 women.
Analyses were performed using Stata Version 12 (StataCorp, College Station, TX). We assessed two aspects of indicator validity, accuracy at the individual level and population level. To measure individual level reporting accuracy, we compared women and observer responses to a single or combination of questions computed for each indicator by constructing two-by-two tables. Missing and “Don’t Know” responses were excluded from the analysis. Where there was sufficient sample size (i.e., at least five counts per cell), we plotted the receiver-operating curve for each indicator. Receiver operating curve (ROC) analysis provides a global statistic of indicator accuracy by plotting the tradeoff between indicator sensitivity (i.e., the true positive rate), against its false positive rate (or 1 – specificity) . To summarize the average accuracy of each indicator, we calculated the area under the receiver-operating curve (AUC). AUC scores range from 0 to 1, with an AUC of 0.5 representing a random guess and an AUC of 1 representing perfect diagnostic accuracy. For the purposes of this study we used an AUC of 0.6 or greater as an a priori benchmark of validity . We present estimates of sensitivity, specificity and the AUC with accompanying 95 % confidence intervals (CIs) assuming a binomial distribution.
To estimate the prevalence that would be obtained for each indicator if assessed in a population-based survey (Pr), we applied the sensitivity (SE) and specificity (SP) calculated for each indicator to its true prevalence (P) (i.e., observer report) using the following equation: Pr = P x (SE + SP-1) + (1-SP) . We then calculated the inflation factor (IF), or the ratio of the survey-based prevalence to the true prevalence, to estimate the degree to which each indicator would be over or under-estimated if assessed at the population level. A priori validation criteria for the IF was set at 0.75 < IF < 1.25 .
We defined overall acceptable indicator performance in terms of meeting both the individual (AUC) and population-level (IF) criteria (0.60 < AUC and 0.75 < IF < 1.25). However, we note that indicator validity criteria should be weighed according to the indicator’s intended use. For example, an indicator with low accuracy at the individual-level may produce an acceptable estimate of population-level coverage if the false positive reports and false negative reports yield a ratio of approximately one. Given this, we refer readers to the full validation results.
Sample descriptive characteristics
Percent distribution of women by background characteristics
Percentage (%) N = 597
Prior parity (Total number live births)
4 or more
Single, never married
Percentage of women who responded “Don’t Know” to survey questions
% “Don’t Know”
Immediately after the delivery of your baby, did anyone give you a medication or injection called oxytocin to help your uterus contract/become firm?
Did anyone give you medication or an injection called ‘oxytocin’ before you delivered the placenta?
While you were at the health facility for the birth of your baby, did you receive a HIV test?
Did you or anyone else give anything to the baby to eat or drink within the first hour after delivery?
Did the health provider(s) wash their hands with soap and water or use antiseptic before examining you?
Just after the delivery of your baby, in the first few minutes after the delivery of your baby, did anyone give you medication intravenously (through a through a tube in your arm)?
Was your baby dried off with a towel immediately, within a few minutes after his/her birth ?
About how long after birth was your baby bathed for the first time?
In your first physical examination/check after delivery, did a health provider check your belly to see whether your womb was becoming firm after the birth of your baby?
The highest percentage of women responded “Don’t Know” to questions regarding administration of oxytocin (i.e., uterotonic for the prevention of postpartum hemorrhage) at any time or before the delivery of the placenta (53 % and 38 %, respectively). An alternate proxy indicator regarding receipt of a uterotonic for prevention of postpartum hemorrhage – whether, immediately after delivery, “anyone [gave] medication intravenously through a tube in your arm” – also exceeded 5 % (8 %) “Don’t Know” responses. High percentages of “Don’t Know” responses were also recorded for several immediate postnatal interventions for the mother and newborn. In addition, nearly one-quarter of women did not know whether they received an HIV test. Anecdotal evidence from data collectors indicated that many women did not know what HIV was, or confused it with human papillomavirus (HPV).
Validation results for maternal and newborn intrapartum and immediate postnatal care indicators †
True Preva (%)
Sensitivityb (95 % CI)
Specificityc (95 % CI)
Self-report survey estimate (%), Based on sensitivity and specificityd
AUC (95 % CI) (>0.60)
IFe (0.75 to 1.25)
Indicators of Health Service Contacts
Main provider labor- doctor or medical resident
Main provider delivery– doctor or medical resident
Indicators of Content of Care
Takes urine sample
Injection or IVg medication received at some time during labor, before birth of baby (general)
Blood products given
Receives injection or IVg medication for induction or augmentation of labor
Augments labor with uterotonic
Membranes ruptured (labor induction or augmentation)
Breastfeeding initiated within first hour of birthi
First post-delivery exam, provider checks for involution
Palpates uterus after delivery of placenta
Uterotonic received 1-3 min after birth
3 AMTSLh elements: prophylactic uterotonic + controlled cord traction + uterine massage following delivery of placenta
HIV status checked
Induces labor with uterotonic
Woman received pain relief medication
First post-delivery exam, provider ask/checks for bleeding
First post-delivery exam, provider examines perineum
First post-delivery exam, provider takes temperature
Encourages/assists woman to ambulate during labor
Woman asked for pain relief medication at some time
Something other than breastmilk given to baby within first hour of birthi
Uterotonic received 1-3 min after delivery of placenta
Baby given to mother immediately after birthi
Severe bleeding (hemorrhage)
High blood pressure/convulsions (eclampsia)
Complication (yes to any)
Contact with the health system
Indicators of contact with the health system were: the type of provider involved in labor and delivery care and the type of facility where the delivery took place.
Skilled birth attendance
Cross-tabulation of women’s response and observer report of the main provider during delivery
Self- report (Number)
Observer report (Number)
Although this study was not designed to assess recall of the type of institution for delivery given that it took place in one facility only, for exploratory purposes women were asked to identify where they gave birth and whether it was a public or private sector facility. Replicating the methodology of the DHS and MICS surveys, women were also asked to specify the facility type (e.g., hospital, health clinic/center, health post, or other location). If women were unable to provide this information, they were asked to name the place.
Among women who responded to the question on whether the facility was classified as public or private sector and then specified its type, 85 % were correct. Among women who responded to the open-ended component of the question only: 83 % correctly named the facility, 9 % identified the facility as a hospital but did not identify the sector, and four percent identified both elements (e.g., reporting “Hospital público”, “Hospital federal”, or “Hospital de asistencia pública”); the open-ended probe elicits additional correct information not captured initially in the categorical response.
Content of care
We validated 21 indicators related to interventions received throughout labor and the immediate postnatal period. Indicators of content of care with the highest individual-level accuracy were episiotomy (AUC: 0.80, 95 % CI: 0.77–0.84), followed by receiving an injection for the induction or augmentation of labor (AUC: 0.69, 95 % CI: 0.65–0.74), urine sample screen upon hospital admission, (AUC: 0.67, 95 % CI: 0.67–0.71) and receipt of blood products (AUC: 0.67, 95 % CI: 0.63 – 0.71). Indicators related to the third stage of labor, including receiving an injection or medication following delivery (i.e., uterotonic for the prevention of postpartum hemorrhage) and immediate postnatal care for the mother had moderate or low overall individual level validity.
Population-level accuracy for 13 content of care indicators were within 25 % of the true prevalence (0.75 < IF < 1.25). Overestimation was nearly double or greater the observed prevalence for four indicators: the infant was placed with the mother immediately after birth (5.8), receipt of an injection or medication following delivery of the placenta (i.e., uterotonic for the prevention of postpartum hemorrhage) (4.7), something other than breastmilk was given to the infant within the first hour of birth (2.7), and breastfeeding initiated in the first hour of birth (1.9).
Uterotonic for prevention of postpartum hemorrhage
A critical component of maternal health care is receipt of a uterotonic for the prevention of hemorrhage, a leading cause of maternal death in Mexico and globally [18, 19]. Women were asked whether, “Immediately after the delivery of your baby, did anyone given you” (1) “an injection in your thigh or buttocks”, (2) “medication intravenously (through a tube in your arm”, or (3) “tablets to swallow or hold in your mouth”, or (4) “tablets placed in your rectum”. As the standard of care in the study facility was administration of the uterotonic via intravenous injection (100 % of cases observed), there was insufficient sample size to assess each of these indicators independently. Given this, we compared a combined indicator comprised of women who reported “Yes” to any method of uterotonic administration within the first few minutes of birth against the observer report of whether the woman received a uterotonic within three minutes of birth. The observed prevalence was 61 % and the reported prevalence was 64 %. The indicator did not meet the AUC (0.51, 95 % CI: 047–0.55) but did meet the IF criterion (0.95).
Women were asked the same set of questions about the first few minutes after the delivery of the placenta. A combined indicator constructed from women’s reports of whether she received a uterotonic via any method met neither criteria (AUC 0.51, 95 % CI: 0.47–0.55; IF 4.7). The specificity of this indicator (6 %, 95 % CI: 4–9 %) was low.
We attempted to validate an alternative indicator that used the name of the uterotonic drug and its purpose. Women were asked whether, “Immediately after the birth of your baby, did anyone give you a medication or injection called oxytocin to help your womb contract/become firm?” In the study facility, oxytocin was the standard uterotonic (administered in 97 % of cases). However, of women who reported “Yes” to any of the methods of uterotonic administration, 62 % responded either “Don’t Know” or “No” to the oxytocin question. While there was insufficient sample size to robustly analyze this indicator, descriptive results show that, among women who were observed to receive oxytocin and who gave a valid response to the question, 50 % accurately reported receiving the medication by its name.
Women were asked whether they had experienced a list of several obstetric complication symptoms (e.g., excessive bleeding, prolonged labor, high blood pressure, convulsions). There was sufficient sample size to assess the validity of four indicators based on these questions. One indicator- excessive bleeding (hemorrhage), met both acceptability criteria (IF 1.0, AUC 0.73, 95 % CI: 0.69–0.76). This indicator high specificity (95.2 %, 95 % CI: 93.1–96.9) but moderate sensitivity (50.0 %, 95 % CI: 35.2–64.8 %); nearly half of women who experienced excessive bleeding would not be captured in a survey question. The other three indicators met the individual criteria only and were overreported by women at the population level. For example, experiencing high blood pressure and convulsions (eclampsia) was overestimated by over 9 times at the population level (IF 9.1), although the low prevalence means that even slightly less than perfect reporting can lead to large overestimation.
Indicators that measure contact with the health system in addition to the quality of received care are critical for measuring progress in improving maternal and newborn health. This is the first study to empirically validate indicators related to the contact, quality and outcomes of maternal and immediate newborn postnatal care in Mexico. A strength of this study is the use of direct observation as the reference standard.
Few indicators assessed in the present study met both study validity criteria. This occurred in part because many interventions were routine practice and there was insufficient variation in the observed prevalence for robust analysis. However, an additional 9 indicators met the individual-level criteria and 18 met the population-level criteria. Indicators that did not meet both criteria are not necessarily invalid for all measurement purposes, but should be used in accordance with the rationale for their use. For example, indicators that do not meet criteria for the AUC but do meet the IF criterion may be suitable to measure intervention coverage at the population-level as false positive and negative reporting balance each other at the aggregate level.
Our findings also highlight the challenges of measuring low prevalence indicators accurately. Given that the calculation of the IF depends upon the indicator’s observed prevalence, even a small number of false positive responses can result in overestimation as measured by the IF. The implications of varying ‘true’ prevalence rates on the estimated prevalence that would be obtained from women in a household survey is illustrated in a previous article . Additional validation work is needed to identify appropriate strategies for measuring low prevalence indicators, such as maternal complications and non-indicated practices.
The consistency of our results with prior findings is mixed. For example, the moderate sensitivity and high specificity for reported symptoms of hemorrhage (excessive bleeding) corresponds with levels found among women delivering in Indonesia, Benin and the Philippines [20–22]. However, these results differ from women’s reporting in Taiwan and Ghana . We also found high sensitivity and low specificity for an indicator of whether or not the woman received a uterotonic within the first few minutes of birth; which differs from the low sensitivity and moderate specificity documented in a study conducted in Mozambique . Taken together, these findings suggest that women’s understanding and recall of interventions received and obstetric complications experienced may vary by clinical and cultural context, and reinforces the importance of validating women’s reporting in different settings.
No contact with care indicators that we were able to assess (i.e., indicators of skilled birth attendance) met both validation criteria. However, a combined indicator of whether the primary birth attendant during delivery was a doctor/ob-gyn or medical resident did meet the criteria for valid estimation at the population level. Given that broad categorizations of provider types are likely to be more programmatically meaningful than the ability of women to delineate between finer categories of qualified health professionals (e.g., doctors versus medical residents), a composite indicator is likely to have utility for assessing population-level coverage. It is important to recognize, however, that skilled attendance is not synonymous with receiving appropriate care. Even a competent provider may not be able to provide high quality care if necessary supplies and other aspects of the enabling environment are inadequate [24, 25]. When possible, self-reported data should be triangulated with other data sources, such as stock-outs of essential medicines [26, 27].
Given that “skilled attendance” cannot be used reliably as a proxy for quality of care, indicators that measure the content of care administered by providers should also be assessed. Across content of care indicators assessed in this study, however, we observed a tendency for women to overreport some standard prevention practices, particularly related to the initial client assessment and immediate postnatal care for the mother and newborn. The high sensitivity and low specificity of many of these indicators suggests that in the few instances when an intervention was not administered, women tended to falsely report that it had occurred.
One potential explanation for this pattern is facility bias, or the assumption by women that they received quality care due to the fact that they delivered in a health facility [10, 13]. For example, more careful probing of some indicators of care, such as whether an injection or IV medication was received before birth (i.e., uterotonic for induction or augmentation of labor), or following delivery (i.e., uterotonic for prevention of postpartum hemorrhage) suggests that women did not know the specific indication for the care received, just that an intervention occurred. Validation results for the uterotonic indicators may reflect more general knowledge women had about having an IV line set, which was standard hospital practice for all women entering the labor and delivery ward, rather than knowledge of which medication was given or why. For example, neither an indicator of whether “an injection or IV medication was received to speed up labor” (i.e., uterotonic induction of labor) or whether “an injection or IV medication was received to strengthen labor” (i.e., uterotonic for augmentation of labor) met both validation criteria, while a general indicator of receiving an injection or IV medication at some time before the birth of the baby did. Similarly, most women correctly reported receipt of a uterotonic within the first few minutes after delivery of the baby, but also falsely reported the receipt of a uterotonic within the first few minutes after the delivery of the placenta, suggesting women may be able to report that some intervention occurred, but not be able to report accurately on aspects of its timing or purpose.
In light of the tendency for overreporting in a facility setting, high specificity for an indicator of receipt of a uterotonic may be favored at the cost of sensitivity. The near universal administration of oxytocin limited our ability to determine if an alternate uterotonic indicator that used the name of the drug and its purpose would achieve higher specificity as opposed to an indicator that asks women about the route of drug administration. Our descriptive results, however, did show that use of an indicator that uses the drug name captured substantially fewer true cases and fewer women were able to answer the question at all. This suggests that asking about receipt of a uterotonic using the drug name and reason for its use may not be a better alternative to asking women about whether they received an injection or some other method of drug administration.
An alternate explanation for positive reporting bias by women is that mothers misunderstood or conceptualized the meaning of questions differently than the observer definitions. For example, indicators with the highest “Don’t Know” responses generally included (1) technical terms, such as medication or disease names, related to the (2) timing of interventions received (including the sequence of newborn thermal care), or (3) the immediate postnatal period. Anecdotal evidence from study interviewers shows that additional explanation was needed to describe the difference between HIV vs. HPV, intravenous vs. intramuscular drug administration, and to define epidurals. Questions to which a substantial percentage of women responded “Don’t Know” (e.g. 20 % or more) are unlikely to yield valid information via self-reports and are not recommended for inclusion in population-based surveys.
Our findings highlight the importance of a two-part question with an open-ended component to improving reporting for complex indicators, such as the type of institution where the woman delivered. The fact that additional women could identify key facility attributes in an open-ended format suggests that providing the opportunity for women to name the specific facility, as is done in both the DHS and MICS surveys, is advantageous for capturing information on institutional deliveries.
While the study provides useful insight regarding the validity of contact and content of care indicators, there were some limitations. Due to budget limitations, the validation exercise took place in one type of facility only. Our study results are therefore reflective of women who delivered in a large public hospital. For example, although institutional delivery is near universal in Mexico (94 %), it is higher among women who reside in urban relative than rural areas (96 % and 88 %, respectively) and our results may be more reflective of this type of population . The minimal variation in standard hospital procedures limited our ability to assess all indicators, which may have otherwise been proven valid. Additionally, in the study setting relatively few women experienced complications such as eclampsia, a leading cause of maternal death in Latin America and the Caribbean. It is possible that with a greater sample size, different combinations of questions regarding the symptoms of eclampsia could be validated to identify question wording with the highest accuracy.
Another limitation it that, due to hospital policy, it was not possible to include women who became indicated for a cesarean section delivery in the study. Future research should extend this work to a broader range of types of health facilities, including health centers and clinics, and may have the potential to validate the reporting accuracy of women who underwent a cesarean delivery. Validation research in other types of facilities may also reduce the potential for positive facility bias in reporting. Finally, exit interviews with women occurred shortly following delivery, prior to hospital discharge. The short recall time may cause women’s self-reports to be more accurate than in population level surveys such as the DHS and MICS when women are typically interviewed about a birth that took place one to three years prior. As part of an overall program of work on indicator validation, a separate follow-up study is underway to re-interview women approximately one year following delivery in Kenya.
As attention in maternal health shifts to a new emphasis on ending preventable maternal mortality (EPMM) and addressing inequities, the tracking of progress will remain important . A key aspect of these future efforts will require the valid measurement of not only contact with maternal and newborn health services but also coverage of key interventions that reflect the quality of services received. Until recently, however, little attention has been paid to improving the measurement of intervention coverage and quality at the population level. This information is needed to set programmatic priorities and to allocate resources effectively. Given the effect that this information can have on improvements in the lives of women and newborns, priority should be given to work that ensures that indicators of progress are selected, measured, and interpreted accurately.
We acknowledge Hannah Taboada, who developed the data collection tools used in the study. We also thank Tahilin Karver and Alberto Martinez Polis who provided oversight for the facility-based data collection. We appreciate the advice of the maternal health experts who advised on the study and are grateful to the women and providers who allowed us to share their experiences. This study was supported by the Maternal Health Task Force through a grant from the Bill & Melinda Gates Foundation.
Availability of data and materials
Data are available by request from the corresponding author.
Conceived of the study design and selected indicators for analysis: AKB, CD. ICMJE criteria for authorship read and met: AKB CD KM KB. Agree with manuscript and conclusions: AKB CD KM KB. Collected the data: CD KB. Analyzed the data: KM. Wrote the paper: AKB KM KB. All authors read and approved the final manuscript.
The authors declare that they have no competing interests.
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.
- Lozano R, Wang H, Foreman KJ, Rajaratnam JK, Naghavi M, Marcus JR, Dwyer-Lindgren L, Lofgren KT, Phillips D, Atkinson C, et al. Progress towards millenium development goals 4 and 5 on maternal and child mortality: an updated systematic analysis. Lancet. 2011;378:1139–65. doi:10.1016/S0140-6736(11)61337-8.View ArticlePubMedGoogle Scholar
- Frenk J, Gómez-Dantés O, Langer A. A comprehensive approach to women’s health: lessons from the Mexican health reform. BMC Womens Health. 2012;12:42. doi:10.1186/1472-6874-12-42.View ArticlePubMedPubMed CentralGoogle Scholar
- Secretaría de Salud. El Convenio General de Colaboración Interinstitucional para la Atención de la Emergencia Obstétrica, un camino hacia la universalización de los servicios de salud. Mexico City. 2012. http://www.ccinshae.salud.gob.mx/descargas/Atencion_de_la_Emergencia_Obstetrica.pdf. Accessed 21 Jan 2015.
- Evidence and Lessons from Latin America (ELLA): Observatory of maternal mortality in Mexico: a civil society-led initiative. 2013. https://www.gov.uk/dfid-research-outputs/ella-policy-brief-observatory-ofmaternal-mortality-in-mexico-a-civil-society-led-initiative. Accessed 25 May 2015.
- Bryce J, Arnold F, Blanc A, Hancioglu A, Newby H, Requejo J, Wardlaw R, CHERG Working Group on Improving Coverage Measurement. Measuring coverage in MNCH: New findings, new strategies and recommendations for action. PLoS Med. 2013;10(5):e1001423. doi:10.1371/journal.pmed.1001423.View ArticlePubMedPubMed CentralGoogle Scholar
- Gutiérrez JP, Rivera-Dommarco J, Shamah-Levy T, Villalpando-Hernández S, Franco A, et al. Encuesta Nacional de Salud y Nutrición 2012. Resultados Nacionales. Instituto Nacional de Salud Publica. 2012. http://ensanut.insp.mx/informes/ENSANUT2012ResultadosNacionales.pdf. Accessed 21 Jan 2015.
- United Nations, Population Division: World Population Prospects, The 2012 Revision. (2014). http://esa.un.org/wpp/. Accessed 24 Nov 2014.
- Walker D, De María L, Campero L, González D, Suárez L, et al. El uso y rol de proveedores no-médicos para la atención prenatal y obstétrica en México: parteras profesionales técnicas y enfermeras obstetras. Evidencias y áreas de oportunidad. Cuernavaca, Mexico: Instituto Nacional de Salud Pública; 2011.Google Scholar
- Demaria LM, Campero L, Vidler M, Walker D. Non-physician providers of obstetric care in Mexico: Perspectives of physicians, obstetric nurses and professional midwives. Hum Resour Health. 2012;10:6. doi:10.1186/1478-4491-10-6.View ArticlePubMedPubMed CentralGoogle Scholar
- Liu L, Li M, Yang L, Ju L, Tan B, Walker N, Bryce J, Campbell H, Black RE, Guo Y. Measuring coverage in MNCH: A validation study linking population survey derived coverage to maternal, newborn and child health care records in rural China. PLoS One. 2013;8(5):e60762. doi:10.1371/journal.pone.0060762.View ArticlePubMedPubMed CentralGoogle Scholar
- Tuncalp O, Stanton C, Castro A, Adanu R, Heymann M, Adu-Bonsaffoh K, Lattof SR, Blanc A, Langer A. Measuring coverage in MNCH: Validating women’s self-report of emergency cesarean sections in Ghana and the Dominican Republic. PLoS One. 2013;8(5):e60761. doi:10.1371/journal.pone.0060761.View ArticlePubMedPubMed CentralGoogle Scholar
- Stanton CK, Rawlins B, Drake M, dos Anjos M, Cantor D, Chongo L, Chavane L, da Luz VM, Ricca J. Measuring coverage in MNCH: Testing the validity of women’s self-report of key maternal and newborn health interventions during the peripartum period in Mozambique. PLoS One. 2013;8(5):e60694. doi:10.1371/journal.pone.0060694.View ArticlePubMedPubMed CentralGoogle Scholar
- Blanc AK, Warren C, McCarthy KJ, Kimani J, Ndwiga C, RamaRao S. Assessing the validity of indicators of the quality of maternal and newborn health care in Kenya. J Glob Health. 2016;6(1):010405.View ArticlePubMedPubMed CentralGoogle Scholar
- Secretaría de Salud, Mexico. Programa de acción específico 2007-2012: salud sexual y reproductiva para adolescentes. Mexico City. 2008. http://www.cdi.salud.gob.mx:8080/BasesCDI/Archivos/Saludsexualyreproductiva/110801_pae_adolescentes.pdf. Accessed 12 Apr 2014
- Austin A, Salam RA, Lassi LS, Das JK, Bhutta ZA. Approaches to improve the quality of maternal and newborn health care: an overview of the evidence. Reprod Health. 2014;11 Suppl 2:S1. doi:10.1186/1742-4755-11-S2-S1.View ArticlePubMedPubMed CentralGoogle Scholar
- Eng J. Receiver operating characteristic analysis: a primer. Acad Radiol. 2005;12:909–16.View ArticlePubMedGoogle Scholar
- Vecchio TJ. Predictive value of a single diagnostic test in unselected populations. N Engl J Med. 1966;274(21):1171–3.View ArticlePubMedGoogle Scholar
- Schiavon R, Tronsco E, Polo G. Analysis of maternal and abortion-related mortality in Mexico over the last two decades, 1990-2008. Int J Gynaecol Obstet. 2012;118 Suppl 2:S78–86. doi:10.1016/S0020-7292(12)60004-6.View ArticlePubMedGoogle Scholar
- Say L, Chou D, Gemmill A, Tuncalp O, Moller AB, Daniels J, Gulmezoglu AM, Temmerman M, Alkema L. Global causes of maternal death: a WHO systematic analysis. Lancet Glob Health. 2014;2:e323–33. doi:10.1016/S2214-109X(14)70227-X.View ArticlePubMedGoogle Scholar
- Ronsmans C, Achadi E, Cohen S, Zazri A. Women’s recall of obstetric complications in South Kalimantan, Indonesia. Stud Fam Plann. 1997;28(3):203–14.View ArticlePubMedGoogle Scholar
- Filippi V, Ronsmans C, Gandaho T, Graham W, Alihonou E, Santos P. Women’s reports of severe (near-miss) obstetric complications in Benin. StudFam Plann. 2000;31(4):309–24.View ArticleGoogle Scholar
- Stewart KM, Festin M. Validation study of women’s reporting and recall of major obstetric complications treated at the Philippine General Hospital. Intl J Gynaecol Obstet. 1995;48:S53–66.View ArticleGoogle Scholar
- Souza JP, Parpinelli MA, Amaral E, Cecatti JG. Populations surveys using validated questionnaires provided useful information on the prevalence of maternal morbidities. J Clin Epidemiol. 2008;61:169–76. doi:10.1016/j.jclinepi.2007.04.009.View ArticlePubMedGoogle Scholar
- Hussein J, Bell J, Nazzar A, Abbey M, Adjei S, Graham W. The skilled attendance index: proposal for a new measure of skilled attendance at delivery. Reprod Health. 2004;12(24):160–70.Google Scholar
- Harvey S, Wong Blandon YC, McCaw-Binns S. Are skilled birth attendants really skilled? A measurement method, some disturbing results and a potential way forward. Bull World Health Organ. 2007;85(10):783–90.View ArticlePubMedPubMed CentralGoogle Scholar
- Winter R, Pullum T, Florey L, Hodgins S. Impact of scale-up of maternal and delivery care on reductions in neonatal mortality in USAID MCH priority countries, 2000-2010. Rockville, Maryland: ICF International; 2014. http://dhsprogram.com/publications/publication-as46-analytical-studies.cfm Accessed 12 Apr 2015.Google Scholar
- Hodgins S. Achieving better maternal and newborn outcomes: coherent strategy and pragmatic, tailored implementation. Glob Health Sci Pract. 2013;1(2):146–53. doi:10.9745/GHSP-D-13-00030.View ArticlePubMedPubMed CentralGoogle Scholar
- UNICEF Global Database: Delivery care: institutional deliveries- percentage. 2014. http://Data.unicef.org/maternal-health/delivery-care. Accessed 19 May 2015
- World Health Organization: Strategies toward ending preventable maternal mortality (EPMM). 2015. http://who.int/reproductivehealth/topics/maternal_perinatal/epmm/en/. Accessed 4 June 2015.