Introduction
In the absence of accurate Vital Registration Systems, developing countries have to resort to alternatives that merely provide periodic estimates of maternal mortality ratio (MMR), which makes it difficult for programme managers to measure and improve the effectiveness of safe-motherhood initiatives, and to monitor progress being made toward achieving the fifth Millennium Development Goal (MDG-5). Two proxy indicators used for this purpose are: the percentage of institutional deliveries and the percentage of deliveries attended by skilled birth attendants. These indicators are easy to monitor from periodic cross-sectional surveys like the Demographic and Health Survey (DHS). Cross-country comparisons also show a negative relationship between MMR and each of these two indicators. The percentage of deliveries attended by skilled birth attendants is also included as one of the indicators under MDG-5. However, there are a few issues with these indicators.
First, there is a big overlap between these two indicators. For example, while the proportion of births attended by skilled birth attendants is usually higher than the proportion of institutional deliveries, the difference between the two was less than 5 percentage points in 27 out of 40 developing countries with a DHS conducted since 2005 (Macro International, 2012). This happens because the number of home deliveries attended by skilled birth attendants is usually low due to the lack of such attendants in rural communities. Second, even if all home deliveries are attended by trained midwives deployed in rural communities, this strategy alone is unlikely to substantially reduce MMR because these skilled birth attendants are neither equipped nor trained to deal with obstetric complications that require surgical interventions. They can identify such cases in a timely fashion and refer them to the appropriate facility. However, women with obstetric complications still have to reach a health facility with appropriate care and be treated in time in order to be saved after arriving at the facility. Third, the basic assumption underlying the negative relationship between MMR and the percentage of institutional deliveries may be violated in situations where institutional deliveries are promoted through the use of vouchers or cash incentives because such incentives may disproportionately attract women without obstetric emergencies. Moreover, not all health facilities are equipped to treat obstetric emergencies requiring surgical intervention. Under these circumstances, an increase in institutional deliveries alone may not result in a concomitant decline in MMR.
Other alternatives to supplement these proxy indicators must be explored. Two such indicators are: (1) the number of women with obstetric complications as a percentage of all deliveries at health facilities offering comprehensive emergency obstetric care, and (2) the case fatality ratio among these women with obstetric complications (Jain, Reference Jain2010). These two woman-level indicators can only be monitored from facility-level data about women reaching these facilities.
The purpose of this paper is to demonstrate the impact of monitoring these woman-level outcomes at health facilities providing comprehensive emergency obstetric care (EmOC) and also to demonstrate the importance of these indicators in understanding the impact of improvements in quality of services, especially in dealing with emergencies, on the reduction in MMR in communities served by these facilities. The case study presented in this paper uses data from Pakistan. However, the outcome and the implications are equally applicable to other developing countries interested in monitoring the progress towards achieving MDG-5. It highlights the importance of a facility-based monitoring system in providing a gauge for measuring an important component of changes in maternal mortality risks among communities served by these facilities. Such a monitoring system could also help to keep a check on the time and condition of the women on arrival, whether they faced any complications prior to arrival at the facility or when at the facility, time elapsed since arrival before receiving treatment, the type of treatment provided and the outcome of the treatment. While this case study used data from public sector health facilities, the methodology could also be applied to private sector facilities.
Context
The MMR in Pakistan is estimated to have decreased from about 500 maternal deaths per 100,000 live births in 1990 to about 276 in 2007 (NIPS & Macro International, Inc., 2008) and 260 in 2008 (140–490; World Health Organization, 2010). In order to achieve the MDG-5, the country has to reach a MMR level of about 140 by 2015. While the MMR in Pakistan has declined during 1990 and 2007, policymakers and programme managers do not know how they are doing towards reaching that goal. The tragedy of maternal deaths in Pakistan is that so many are avoidable if women have timely access to quality essential obstetric and emergency care (Islam & Yoshida, Reference Islam and Yoshida2009). The factors contributing to high maternal mortality in the country include the lack of timely and appropriate care in case of obstetric emergencies, with 66% of the deliveries taking place at home (NIPS & Macro International, Inc., 2008), mostly attended by traditional birth attendants. While most of these are normal deliveries, the delays encountered by women while deciding to seek care only after the onset of obstetric complications (first delay) and getting to the right facility in time (second delay) further exacerbate their condition, with many women reaching it too late to be saved (Jafarey & Korejo, Reference Jafarey and Korejo1993; Thaddeus & Maine, Reference Thaddeus and Maine1994). Moreover, once women reach a health facility, they may not receive appropriate treatment (quality of care) in a timely fashion (third delay) and the maternal outcome can be dire.
A review by Jafarey et al. (Reference Jafarey, Kamal, Qureshi and Fikree2008) of initiatives undertaken to reduce maternal mortality in Pakistan included the Pakistan Initiative for Mothers and Newborn (PAIMAN) project implemented in ten districts between 2005 and 2010. Efforts to reduce maternal and neonatal mortality under the PAIMAN project included community-level inputs to increase the demand for safe delivery and facility-level inputs to ensure that these facilities are equipped to provide emergency obstetric care (EmOC). The community-level inputs included women's support group meetings; training and involvement of traditional birth attendants and community midwives; dramas and advertisements on TV and radio; puppet shows; and involvement of ulamas and religious scholars. Facilities were provided with equipment and supplies such as partographs, and staff were trained in important maternal and neonatal health areas, including client-centred approaches, AMTSL (active management of the third stage of labour) and the use of a partograph. Project-supported interventions at these facilities were implemented during 2006 and completed by the end of September 2006 (Mahmood, Reference Mahmood2010).
The important issue in any large safe-motherhood intervention designed to reduce maternal and neonatal mortality such as PAIMAN is how to evaluate the impact of improvements in the service quality and demand environments on maternal mortality. A proper design for evaluating the impact of an intervention demands the inclusion of experimental and control areas and observations prior to and after the intervention. However, this type of design was not used in this case because PAIMAN's implementation strategy was based on responding to empirically based key constraints that lead to high maternal and neonatal mortality. Research was instituted to refine its strategy and enhance its impact. The ultimate purpose of research in PAIMAN was to understand the situation on the ground for making informed decisions and effective planning.
Two cross-sectional surveys were conducted before and after the end of the project implementation to assess the impact of PAIMAN inputs. These surveys indicated that the percentage of institutional deliveries increased from 38% in 2005 to 50% in 2010. This increase of 12 percentage points was entirely attributed to the increase in deliveries in private health facilities (Mahmood, Reference Mahmood2010). Thus it is established that women were seeking facility care in higher proportions. The ratio of Caesarean sections (C-sections) to deliveries also increased by about 5 percentage points in both the private (from 13% to 18%) and public (from 11% to 16%) sector facilities. However, there are indications that women with obstetric complications are more likely to be eventually referred to the district-level or teaching (public sector) hospitals. The purpose of this paper is not to accurately assess the impact of interventions implemented through the PAIMAN project on MMR. Instead it focuses on illustrating the use of facility-level indicators for estimating the effect of quality of care at health facilities on the change in mortality risk – an important component of community-level MMR.
Methods and Data
Data
This paper is based on data especially collected through the monitoring system set up under the PAIMAN project from 31 upgraded health facilities. Three public sector facilities per district (one district-level [DHQ] hospital, one tehsil-level [THQ] hospital and one rural health centre [RHC]) were selected under the PAIMAN project for improvements in services.
The objective of selecting three facilities per district was to systematically and geographically link women in each of the ten districts to a lower-level health facility (RHC), THQ hospital or DHQ hospital. However, exceptions had to be made. The DHQ hospital in District 10 was already a well equipped tertiary-level teaching hospital. Hence one additional THQ hospital was selected from this district. In addition, a Basic Health Unit (BHU) was also selected to develop linkages between the remote areas and the THQ hospital in the hilly tehsil (sub-district) of this district. There was no THQ hospital in District 1; instead one additional RHC was selected. Thus the 31 health facilities upgraded under the PAIMAN project included nine DHQ hospitals, eleven THQ hospitals, ten RHCs and one BHU.
Data on several indicators were compiled from those already being collected at health facilities as part of the government's Health Information System (HMIS). These data were recorded in various hospital registers and case reports (out-patient, obstetric, labour room, emergency and operating theatre). PAIMAN introduced a ‘Performa’ (data collection instrument) for compiling these data monthly. A doctor at each facility entered data from various registers and case reports on to this Performa, and reported it to the PAIMAN headquarters on a monthly basis. These data were checked for completeness and entered into an Excel spreadsheet. The accuracy of these data was assured by randomly checking the information provided through field visits to these facilities. These data included the following:
• Number of women with obstetrical complications treated
• Number of births
• Number of Caesarean sections
• Number of intra-uterine fetal deaths
• Number of obstetrical deaths
• Number of newborn deaths
A uniform definition of obstetric complications was used across all facilities. Obstetric complications included pregnant women admitted due to any complication related to or aggravated by the pregnancy, and occurring during pregnancy, delivery or the post-partum period (within 42 days after delivery/termination of pregnancy). All these women, irrespective of whether or not they required surgical intervention or the outcome or the mode of delivery, were included. However, obstetric complications did not include those women who were admitted without complications and delivered through normal vaginal delivery without complications.
The data being collected at these health facilities were not linked to individual woman. Consequently, the collected data referred to the number of episodes of obstetric complications and not the number of women with obstetric complications. The number of women with obstetric complications will be less than the number of episodes of obstetric complications because the same woman may be admitted more than once to the same or different facility due to the same or different complications. The data collection system did not include the type of obstetric complication and as such the case fatality ratio (CFR) for each complication cannot be calculated.
The monthly data were collected from 31 upgraded health facilities for 36 months starting in January 2007 and ending in December 2009. Data for the months of April and May 2009 were not available for the DHQ hospital in District 1. Moreover, because of disturbances in District 8, data collection for RHCs in this district started in September 2007; data were also missing for one RHC in this district for 9 months for 2008–09. It is quite possible that the numbers of obstetric deaths are not fully recorded. Moreover, there is a possibility of differential degree of under-reporting among districts as well as among health facilities. Overall, these data included information on 59,572 births, 23,778 episodes of obstetric complications, 6545 C-sections, 2509 intra-uterine fetal deaths, 1524 newborn deaths and 164 obstetric or maternal deaths.
All the sampled health facilities were eligible to have emergency obstetric care. However, the data indicate a disproportionate use of district hospitals for emergency care. For example, close to half of all live births and about 60% of obstetric complications were recorded in the nine district hospitals, but they performed about 90% of all C-sections and about 90% of all maternal deaths also occurred in these hospitals. It is possible that the lower level facilities referred the more complicated deliveries to the district hospitals.
Methods and rationale
The following five indicators are computed: the annual facility-level MMRs, C-sections as a percentage of live births, episodes of complications as a percentage of live births, CFR based on C-sections and CFR based on episodes of complications. Facility-level MMR is computed for comparisons with other studies. Caesarean sections as a percentage of live births and episodes as a percentage of live births provide a range for the indicator ‘obstetric complications as a percentage of live births’. Similarly, CFR based on C-sections and CFR based on episodes of complications provide a range of the indicator ‘CFR among deliveries with obstetric complications’. Two indicators for each of the two proposed indicators became essential to compute because of the limitation of the data mentioned above.
United Nations process indicators (Fauveau & Donnay, Reference Fauveau and Donnay2006) recommend five indicators for monitoring the progress towards reduction in MMR: (1) number of facilities that provide EmOC per 500,000 population (one comprehensive and four basic); (2) proportion of all births in EmOC facilities; (3) proportion of women with obstetric complications delivered at EmOC facilities (met need); (4) Caesarean deliveries as a proportion of all births (5–15%); and (5) CFR among women with obstetric complications admitted to a facility (<1%). The two indicators used in this analysis are a slightly modified version of two of these five indicators. The theoretical link between these two proposed indicators and MMR is postulated below.
A large majority (85%) of deliveries are expected to be ‘normal’, with close to zero risk of maternal death irrespective of the place of delivery (home or facility) or type of attendant (skilled or not) at delivery. The percentage of women who potentially need emergency obstetric care appears to be quite similar in developing and developed countries, at around 15% (Bang et al. (Reference Bang, Bang, Reddy, Deshmukh, Baitule and Filippi2004) for rural community in India; Lobis et al. (Reference Lobis, Fry and Paxton2005) for the United States). The UN process indicators also used 15% to estimate the need for comprehensive EmOC in a geographic area (Lobis et al., Reference Lobis, Fry and Paxton2005). This means that almost all maternal deaths would be concentrated among the small fraction (15%) of women who experience obstetric complications around childbirth and require surgical intervention. Hence, the overall MMR in any geographic area (e.g. district, province, country) can be estimated by multiplying (a) CFR among women with obstetric complications and (b) obstetric complications as a proportion of all deliveries, usually assumed to be 0.15. Thus, community-level MMR would simply be equal to 0.15×CFR.
The CFR among all complicated deliveries can also be expressed as the weighted average of CFR among complicated deliveries occurring in facilities with comprehensive EmOC services and CFR among complicated deliveries occurring outside these facilities at home or on the way to these facilities. The CFR among complicated deliveries occurring in facilities with comprehensive EmOC services is expected to be lower than the CFR among complicated deliveries occurring outside these facilities because of the differences in the quality of available services. Thus the community-level MMR will decrease only if complicated deliveries are shifted in a timely fashion directly from home to facilities with comprehensive EmOc services; and if CFR among complicated deliveries in facilities with comprehensive EmOC services is reduced (Jain, Reference Jain2010).
Analysis of facility-level data usually measures maternal mortality risk by including all deliveries in the denominator of MMR. For example, Ronsmans et al. (Reference Ronsmans, Chowdhury, Kobilinsky and Ahmed2010) estimated facility-level MMRs (based on all deliveries) to study the associations between the use of skilled attendance and maternal and perinatal mortality in Bangladesh. They found that while MMR among those who delivered at facilities with comprehensive obstetric care declined over time it remained higher than MMR among those who did not receive skilled care at birth. They attributed this unexpected finding to the possibility that women seek care only when they are ill. They also postulated that facility-level MMR may decline as ‘more women who are less at risk seek care’. The unexpected result in this analysis, however, could be due to the possibility that the unskilled obstetric care group included mostly normal deliveries with almost zero risk of maternal death. A more appropriate comparison for assessing the effect of the quality of obstetric care would be to estimate MMR separately for normal and complicated deliveries among each group.
Facility-level MMR would certainly decline as more women who are at less or no risk of maternal death deliver at these facilities. However, such a decline would be a statistical artifact. It may not indicate improvements in quality of care or survival probability among those who need care because its denominator would also include a large proportion of women with normal deliveries with close to zero risk of maternal death. Thus, the facility-level MMR can decrease with an increase in the number of normal deliveries as a percentage of all deliveries in health facilities even if service improvements had no effect on CFR or mortality among women with obstetric complications. This point is illustrated in this paper by comparing MMR from two district hospitals. For this reason, this analysis proposes the calculation of CFR among those with obstetric complications as an indicator of quality of services.
While it can be reasonably assumed that all maternal deaths in 31 upgraded health facilities occurred among women who experienced obstetric complications, a limitation, as mentioned above, is that the number of women who experienced complications around childbirth at these facilities is not known because this information was not recorded. Instead, the data were collected for the number of episodes of complications and the number of C-sections. The number of women with obstetric complications would lie between two extremes. It would be lower than the episodes of complications because the same woman may visit these facilities more than once for the same complication or for different complications. It would be higher than the number of C-sections because not all obstetric complications require C-sections. However, the number of women with obstetric complications in a population can be even lower than the number of C-sections if large numbers of C-sections are not warranted but are performed on demand. Similarly, the actual CFR among women with complications is expected to be higher than the CFR based on the number of C-sections, and lower than the CFR based on the episodes of obstetric complications.
This analysis is undertaken to illustrate the utility of the two proposed facility-level indicators and to illustrate the limitation of using facility-level MMR as an indicator of quality of obstetric care. For this reason, no effort is made to test the statistical significance of any particular hypothesis concerning changes over time or differences among sub-groups. This analysis is based on secondary data without individual identifiers and did not require Institutional Review Board (IRB) approval.
Results
The analysis of facility-level data shown in Table 1 indicates a decline of about 18% in the facility-level MMR from 329 in 2007 to 271 maternal deaths per 100,000 births in 2009. The results also indicate that these upgraded facilities attracted a higher proportion of women with complicated deliveries. For example, the number of C-sections as a percentage of live births increased from about 10% in 2007 to 12% in 2009; and the episodes of complications as a percentage of live births also increased during this period from about 36% in 2007 to 47% in 2009 (Table 1). The actual increase in the percentage of women with complications delivering at these facilities would lie between these two extremes. The cross-sectional surveys also showed that the number of C-sections as a percentage of all deliveries in public sector facilities increased from 11% in 2005 to 16% in 2010. This is reassuring since the objective of the PAIMAN project was precisely that high-risk pregnancies reach facilities in greater numbers than the normal ones.
Table 1. Obstetric events and ratios for 31 public sector health facilities between 2007 and 2009 in Pakistan
Upgraded health facilities included nine district-level (DHQ) hospitals, eleven tehsil-level (THQ) hospitals, ten rural health centres (RHC) and one basic health unit.
The CFR, a key indicator of the effect of improved quality on the risk of maternal death among women with obstetric complications, also declined. However, the level remains high. For example, CFR based on C-sections decreased by 29% from 3233 to 2291 maternal deaths per 100,000 C-sections; and CFR based on episodes of complications declined by 37% from 927 to 581 maternal deaths per 100,000 episodes of complications (Table 1). The actual value of CFR based on the number of women with complicated deliveries is expected to lie between these two extremes, and the decline in actual CFR would lie between 29% and 37%.
The district hospitals attracted a higher proportion of complicated deliveries requiring surgical interventions. Table 2 shows district-level indicators of obstetric complications and CFR based only on data from nine district hospitals. The CFR also appears to be a better way to compare quality of services across district hospitals. A comparison of data for DHQs in District 1 and District 9 (Table 2) illustrates this point. The facility-level MMR in District 9 is higher than in District 1 (1265 vs 477). This difference, however, does not accurately reflect the effect of differences in the quality of obstetric care available at these two hospitals, because DHQ in District 9 attracted a higher proportion of complicated deliveries than did the DHQ in District 1 (C-section rate: 40% vs 6%; complication rate 45% vs 7%). Consequently, CFR in District 9 is lower than in District 1 (for C-sections: 3164 vs 8176; for complications: 2787 vs 6897). While the numbers are small, CFR based on C-sections in DHQ in District 9 declined by 58% from 4990 in 2007 to 2091 in 2009, indicating a substantial improvement in quality of obstetric care (data not shown).
Table 2. Indicators of complications and case fatality ratio for 2007–2009 period among nine district hospitals in Pakistan
a Data for April and May 2009 were not available for DHQ hospital in District 1.
b No C-section and no maternal death recorded in this DHQ hospital.
c One C-section and one maternal death recorded in this DHQ hospital.
Discussion
This analysis illustrated the importance of monitoring woman-level indicators based on facility-level data when gauging changes in community-level maternal mortality ratios. The data from 31 upgraded health facilities in Pakistan indicated changes in the facility-level indicators between 2007 and 2009: the number of births recorded in these facilities increased by 32%; the facility-level MMR (based on all births) decreased by 18%; the ratio of C-sections to live births increased from 10% to 12%; and the ratio of the episodes of complications to live births increased from 36% to 47%. Furthermore, both CFRs – maternal deaths to C-sections and maternal deaths to episodes of complications – declined for all health facilities as well as for nine district hospitals, which accounted for over 90% of C-sections as well as maternal deaths. In the absence of any control group and in the absence of any observations prior to the intervention, it is difficult to conclude with certainty the effect upgrading these health facilities under PAIMAN has had on maternal mortality. However, since these results satisfy both conditions – an increase in the number of women with complications reaching health facilities and a decrease in CFR among them – they reflect a potential decrease in the level of MMR in districts covered by these facilities.
Within the three delays framework (Thaddeus & Maine, Reference Thaddeus and Maine1994), the observed increase in the ratio of episodes of complications (and number of C-sections) to live births points to a decrease in the first and the second delay, which occur in the communities, in reaching the appropriate facility. The observed decrease in CFRs points to an improvement in treating complications and a decrease in the third delay, which occurs once women arrive at the health facility. These effects cannot be estimated separately from the available data.
A limitation of the data is that how often the same woman visited the same or different health facility for the same or different complications was not known, and therefore the actual value of the ratio of obstetric complications to live births and that of CFR among women with obstetric complications could not be estimated. Moreover, the conditions in which women arrived at the health facilities or elapsed time since their arrival before they received treatment were not collected. Consequently, cause-specific CFRs could not be estimated. Nevertheless, these results are reflective of an improved quality response by the health system to deal with high-risk pregnancies and reduce their risks of mortality. These results also point to the fact that more needs to be done to improve services at these hospitals to reduce CFR among women with obstetric complications; and in communities to transfer women who develop obstetric complications directly to these hospitals.
In most developing countries, the data at the health facilities are recorded in a series of registers and case notes, including admission, delivery, discharge, referral, and rarely kept surgical registers on obstetric complications. Moreover, clinical recommendations endorse these types of data collection activities (Ali et al., Reference Ali, Hotta, Kuroiwa and Ushijima2005; Farooq et al., Reference Farooq, Jadoon, Masood, Wazir, Farooq and Lodhi2006; World Health Organization, 2009). However, a study in Pakistan highlighted that record-keeping was non-standardized and records were of poor quality and that important information was often missing from records, even in teaching hospitals in Karachi (Aziz & Rao, Reference Aziz and Rao2002). Fauveau & Donnay (Reference Fauveau and Donnay2006) also mentioned many limitations of facility-level record-keeping in developing countries. The quality of data is particularly crucial for the definition of obstetric complications (Aziz & Rao, Reference Aziz and Rao2002; Fauveau & Donnay, Reference Fauveau and Donnay2006) but this information is often not recorded, as in many cases there is no column in the register that specifically asks whether any obstetric complication occurred (Aziz & Rao, Reference Aziz and Rao2002; Ali & Kuroiwa, Reference Ali and Kuroiwa2007). Furthermore, these records are rarely linked at the woman level. These deficiencies in hospital record-keeping are not limited to Pakistan (Fauveau & Donnay, Reference Fauveau and Donnay2006). Furthermore, they compromise the calculation of such indicators as the case fatality rate and the Caesarean section rate, highlighting the fact that a standardized and integrated system for collecting and recording data is the need of the hour. This absence or poor quality of record-keeping undermines evidence-based decision-making, especially pertaining to maternal health.
The incidence of complications during pregnancy is an important indicator of the need for emergency obstetric care. However, it is generally under-reported or excluded altogether in facility-level data and over-reported in cross-sectional surveys. It is essential to know the condition a woman is in when she arrives at a health facility and the complications she presents, and those she experiences once admitted to the health facility. Furthermore, it is imperative for improving quality of care to document each woman's experience during the entire period of her stay at the health facility, ranging from the type of procedures conducted and the type of medications prescribed. Such documentation will enable standardization of care, while also providing facts and figures to monitor episodes of births, complications during pregnancy, and maternal and neonatal deaths.
In brief, maintaining hospital records, which are linked for each woman, is important to monitor the effect of improvements in the quality of care provided by these facilities; to monitor the effect of improvements in the referral systems; and to gauge changes in community-level maternal mortality ratios. The data used in this analysis came from existing record-keeping, which did not collect information on each woman from the time of arrival at these facilities to the time they leave, alive or dead. In order to better understand this dynamic process, the following type of data need to be collected from each woman reaching a facility equipped to provide comprehensive EmOC services:
• Whether or not the woman is experiencing any obstetric complications upon arrival.
• Type of complications: haemorrhage; sepsis, eclampsia, obstructed/prolonged labour.
• Time since complication started before reaching the facility.
• Did she come directly from home or from other health facility?
• Management of complication, e.g. treatment provided, time elapsed since arrival before treatment was provided.
• Outcome of treatment for each woman.
• Whether woman survived or not.
• Whether newborn survived or not.
The following indicators can then be computed for each (or a group of) facility offering comprehensive EmOC services:
1. Complicated deliveries as percentage of all deliveries.
2. Percentage of women with complications arriving directly at the facility in time for treatment.
3. Case fatality ratio among women with obstetric complications reaching such a facility.
An increase in the first indicator would suggest that this facility is attracting a higher proportion of complicated deliveries. An increase in the second indicator would suggest that they are reaching directly the appropriate facility in time for treatment. Both indicators together would suggest improvements in the referral system in the community and at other health facilities that are not equipped to provide comprehensive EmOC services. A decrease in the third indicator would suggest an improvement in the quality of care provided to these women. All these changes together would suggest a decline in the maternal mortality ratio in the community.
Setting up a well functioning information-gathering and analysis system in health facilities with comprehensive obstetric care will help challenge policymakers and programme managers to refocus on programme content; to focus on development of organizational strategies to provide access to essential emergency obstetric care 24 hours a day 7 days a week, and to account for every birth and every death (Islam et al., Reference Islam, Malik and Basaria2002; Jafarey et al., Reference Jafarey, Kamal, Qureshi and Fikree2008). Implementation of an individual-level data recording system will aid in fostering the completeness, accessibility and reliability of the data collected, both for clinicians at the point of care as well as researchers monitoring statistics such as MMR and CFR. The collection of woman-level data could be facilitated by an electronic record-keeping system, which could also be used to rapidly generate aggregate reports, and which should be more complete and accurate because users will more likely recognize errors regarding their own patients (Fraser et al., Reference Fraser, Biondich, Moodley, Choi, Mamlin and Szolovits2005). An electronic recording system is ideal to collect woman-level data for settings like District Headquarter (DHQ) hospitals in Pakistan where a limited number of skilled health care providers attend to a large patient population with critical illnesses and emergencies.
Conclusions
Reduction in maternal mortality ratio (MMR) in Pakistan or in any other developing country will require improvements in the availability and quality of comprehensive EmOC services, at least at the district-level hospitals, and reduction in unnecessary delays occurring in the communities by referring/transferring women with obstetric complications directly to these hospitals. In addition, a system of woman-level data collection needs to be instituted at these hospitals to monitor changes in the effects of any reductions in the ‘three delays’ and any improvement in quality of care or the effectiveness of treating pregnancy-related complications among women reaching these facilities. Such a system of information gathering at health facilities with comprehensive emergency obstetric care will help policymakers and programme mangers to measure and improve the effectiveness of safe-motherhood initiatives and to monitor progress being made toward achieving the fifth Millennium Development Goal.
Acknowledgments
An earlier version of this paper was presented at the Global Maternal Health Conference held in New Delhi, India, on 1st September 2010. The PAIMAN project was funded by the United States Agency for Development (USAID) by a co-operative agreement with JSI Research and Training Institute, Inc. The authors take full responsibility for the analysis and interpretation of the results. The authors do not have any conflicts of interest.
