Study design

A cohort study with both retrospective and prospective arms was conducted.

Study setting

The study was conducted in the Greater Accra Region (GAR) of Ghana. The region is about 90% urban, and is served by both public and private facilities. About 97% of pregnant women here attend antenatal clinic at least once during their pregnancy and skilled attendance at delivery is around 84%. About 62% of all health facility deliveries in the region take place in the public sector. The public sector comprises of one teaching hospital, a regional hospital and nine district and sub-metropolitan hospitals. There are ten polyclinics, 31 health centres, some community clinics and three Community Health Planning and Services (CHPS) compounds. The polyclinics are primary health care facilities like the health centres, usually sited in urban and populous areas and provide both general and specialist services. They also have more than one doctor unlike the health center which usually has one doctor or in some cases manned by a medical assistant. CHPS compounds operate at the community level, with midwives or community health nurses, either through home visits or clients going to the compounds. There are several private hospitals operating in the region. All these health facilities offer antenatal and delivery services, though the smaller facilities refer complications to the bigger health facilities. The National Health Insurance Scheme (NHIS) is operational in all the public as well as in most of the private facilities.

Selection of districts and facilities

We randomly selected participants from the different levels of public health care (variable “type of facility”) across the Greater Accra region. All districts and sub-metropolises in the region were grouped into those with a district or sub-metropolitan hospital (eight in number) and those without hospitals (12 in total) and then five districts were randomly selected by balloting with replacement from each group. By this process, the names of the districts in each group were written on a piece of paper, folded and put in a box. After shuffling by an independent person, five districts were picked one at a time, replacing the picked district in the box each time before picking the next to ensure equal chance of selection for all. The district hospitals in the selected districts with hospitals were included in our study. In those districts without hospitals all the primary level care facilities offering both antenatal and delivery services were included for random sampling and one was selected from each district.” We also included the regional hospital which is in a sub-metropolis different from the selected districts/metropolises. Thus in all, 11 health facilities (one regional and five district/sub-metropolitan hospitals, four polyclinics and a health center) were randomly selected from 11 districts/sub-metropolises for the study.

Sample size and recruitment of women

The sample size for the cohort study was based on a prevalence of pregnancy complications of 6% [29] using Open Epi calculator for estimation. In the absence of any documented evidence in the setting, we assumed that the complication rate will be twice as high amongst the unexposed group (incomplete adherence). To detect a two-sided significance difference at 95% confidence interval at a power of 80%, and a one-to-one ratio of exposure to non-exposure, a sample of 372 women was required per exposure group. The minimum total sample size therefore required for both arms was 744 women.

Participants were recruited at delivery and once they met the inclusion criteria, informed consent was obtained. The inclusion criteria included the following: participant 18 years or older; participant had at least one antenatal clinic visit in a health facility during the current pregnancy; participant had first antenatal clinic visit at gestational age less than or equal to 5 months and participant had first antenatal clinic visit at the facility of delivery or in one of the sampled facilities for this study.

From each of the eleven (11) facilities sampled for the study, we recruited a minimum of 68 women, a minimum of 34 women exposed to complete provider adherence and a minimum of 34 unexposed women. We continued recruitment irrespective of adherence status until we had at least 34 women in either group.

Data collection processes and tools

Every woman delivering at the facility on any day and who met the inclusion criteria, and provided informed consent for participation, was enrolled into the study.

At recruitment, a record review of their first antenatal clinic visit from the maternal health record book was retrospectively carried out using a checklist. Data on socio-demographic characteristics, potential confounders such as timing of first antenatal visit, number of antenatal clinic attendance and previous pregnancy history, as well as the 13 variables on guideline requirements were collected. Records of subsequent antenatal clinic visits were also reviewed for any complications developed and identified during the pregnancy. Finally, data on delivery outcomes was collected from both the maternal health record book and delivery register and notes. Participants’ telephone numbers were linked to their study identification numbers (IDs) for follow up.

In the prospective arm, all women and their neonates were followed up till 6 weeks postpartum to complete data collection on outcomes. Follow up was at the health facility during post-natal care visit at 6 weeks postpartum and also by phone. Participants received a phone call 3 weeks post-partum, and at 6 weeks, they were followed up at the postnatal clinic. If they were not available there, they received another phone call. Those who could not be reached by any of these means were treated as lost to follow up. A facility audit was conducted to assess facility factors such as the availability of personnel, services, infrastructure, logistics and supplies that are needed to support adherence to the guidelines at facility level.

The assumption was made that any information on history, examination, laboratory examination and treatment available is what was recorded in the maternal health record book. Information on any service not recorded, was deemed not to have been delivered [30].

Variables

The outcome variables studied were all maternal and neonatal complications during antenatal, delivery and post-partum periods. Maternal complications were i. any antenatal complication (defined as having at least one of the following: anemia in pregnancy, pregnancy induced hypertension/pre-eclampsia/eclampsia, malaria in pregnancy and antepartum vaginal bleeding), ii. Any antenatal complication excluding anemia in pregnancy,, iii. Caesarean section, iv. Delivery complication (defined as having at least one of the following: pregnancy induced hypertension/pre-eclampsia/eclampsia, ruptured uterus, obstructed labour, vaginal tears and perineal tears), v. post-delivery complication (defined as having at least one of the following: postpartum haemorrhage, anemia, malaria and sepsis), vi. Maternal mortality, vii. Any maternal complication (which is defined as having at least one of the listed maternal complications). Neonatal complications were preterm, post-maturity, low birth weight, neonatal jaundice, asphyxia, still birth and neonatal mortality.

The determinants were type of facility, client’s socio-demographics (age, educational level, marital status and employment status), and client’s prenatal factors (parity, trimester at first antenatal care visit, previous pregnancy history and number of times antenatal clinic was attended during pregnancy). Provider adherence levels were defined as complete or incomplete.

Data analysis

Descriptive analysis of participants’ socio-demographic information and previous pregnancy history was conducted by use of frequencies and chi-square analysis. Adherence to guidelines was computed by calculating the proportion of women whose first antenatal clinic visits met the criteria for complete adherence. Possible associations of adherence status with some patient characteristics were assessed using chi-square. Incidences of maternal and neonatal complications were estimated in percentages. Determinants of maternal and neonatal complications, as well as the effect of complete adherence on pregnancy outcomes were estimated and expressed as relative risks (RRs) with their 95% confidence intervals (CI). To correct for potential clustering effect due to our sampling strategy, we used generalized linear equation estimation in calculating RRs. We also adjusted for variables such as trimester of first ANC visit, maternal age, parity, previous pregnancy complication and any antenatal complication where appropriate. Significance was determined at p-value <0.05. Data analysis was carried out using IBM SPSS Statistics for Windows, Version 20.0. Armonk, NY: IBM Corp.

Adherence to first antenatal care guidelines

Overall, complete adherence to guidelines pertained to the care of only 48.5% [95% CI (45.3–51.7%)] of all participants in our sample, during their first antenatal clinic visit. The range was 33.8 to 61.7% amongst the facilities sampled. Complete adherence to guidelines was higher amongst women seen at the polyclinics [52.9% (47.5–58.2%)] than those seen in hospitals [47.8% (43.5–52.1%)] and the health center [34.2% (22.8–44.8%)].

Participants’ baseline characteristics

Risk of maternal and neonatal complications

Main findings

Key findings of this study are that provider adherence to first antenatal care guidelines is low (48.5%), most women register for first antenatal care in the second trimester and complete adherence to first antenatal guidelines reduced the risk of delivery and any neonatal complication.

Strengths and limitations

Adherence to first antenatal care guidelines was retrospectively scored from participants’ records. It is therefore not possible that provider practice changed because of the study, thus reflecting everyday provider practice of adherence to guidelines. Loss of follow-up was minimal at 2.1%. We however recognize some limitations for our study. The fact that we recorded zero maternal deaths in our study is not reflective of the situation in Greater Accra, and might purely be due to the fact most of our participants are those who survived pregnancy and delivery. Also, since only women receiving care in public facilities were included in the study, generalizability of the results to service delivery in private practice is limited.

Interpretation

Antenatal care as well as skilled attendance at delivery (which all participants in our study had) have been shown to improve maternal and neonatal outcomes [35]. Provider adherence to first antenatal care guidelines was observed to be low in our study in contrast to what other studies [36–38] conducted in advanced settings found. However, one earlier study found limited use of maternal health guidelines in Ghana, Burkina Faso and Tanzania [39]. Previous studies from Ghana identified gaps in quality of care of maternal, new-born and child health, including antenatal care [40, 41], which may be explained by low adherence to guidelines and subsequent suboptimal quality of care provided to pregnant women.

The current study demonstrates the significant role of quality of care in translating access to antenatal care to good outcomes. Complete adherence to first antenatal guidelines resulted in reduced risk of delivery and neonatal complications. Guidelines can only lead to improved or quality care if they are translated into daily provider practice [42]. As they continue to provide service, clinicians may not even be aware of their lack of adherence to available guidelines [43]. It has been established that active steps, such as reminders, are necessary to translate clinical practice guidelines into daily practice [9, 42, 44]. The process of engagement of and support for providers of maternal and neonatal care with regards to utilization of clinical guidelines can be enhanced through a mechanism of regular monitoring and providing feedback to them as has been done in other contexts [45]. Programs to improve adherence to guidelines that are fashioned around a continuous educational framework for providers are likely to transform provider practice [46] and will subsequently improve the outcome of neonatal service delivery.

Provider adherence to guidelines however did not significantly influence the risk of any maternal complications, although it specifically reduced the risk of delivery related complications. Perhaps other mechanisms better explain maternal outcomes apart from adherence. It must also be recognized that perhaps the frequent antenatal clinic attendance allows providers to make up for what they missed out during the first antenatal visit. Basic, skilled and ultimately safe intrapartum care is noted to be important in addressing maternal and neonatal morbidity and mortality [2, 47], and any intervention of the process that results in better outcomes should be encouraged. Evidence based practice intrapartum is essential as one systematic review has shown [48]. Thus mechanisms to enhance provider adherence to both ANC and delivery guidelines should be instituted at all levels of care.

Although poor quality of care decreases utilization of services [41], we found that most women in our study attended antenatal clinics at least four times during their pregnancy despite low provider adherence to guidelines during their first visit. It is uncertain how much of guideline requirement is known by clients, but one study has described high satisfaction for antenatal care services amongst attendants in Ghana [49]. Frequent antenatal attendance provides enough contact between the provider and the pregnant woman to identify and address any challenges of the pregnancy [50]. It is also important that during each of these contacts quality of care is assured.

We also find that women are reporting late for their first antenatal visit. Most women reported during the second trimester, as previously noted in studies both in Ghana and other low and middle-income countries [6, 51–55]. This is a source of concern as pregnancy complications may be recognized too late and the opportunity for timely interventions missed.

About 70% of the study participants had at least one form of complication during the antenatal, delivery or post-partum period. The majority of these complications occurred during antenatal period and anemia in pregnancy was high on the list. This is worrying in a setting where hemorrhage is a common cause of maternal deaths [56], since anemia will affect the woman’s ability to adequately compensate physiologically in case of bleeding during delivery. We are not sure what proportion of this risk of anemia is due to physiological hemodilution in pregnancy [57], but malaria is a common cause of anemia in pregnancy in our setting [57–59], the incidence being 12.2% amongst our study participants. Both malaria in pregnancy and anemia potentially increase the risk of low birth weight and prematurity [58]. Compared to other studies, risk of hemorrhage was low (2.1%) amongst our study participants compared to what other studies have found [56, 60]. The relatively high rate of caesarean section in our study may partly be due to the fact many of our facilities were hospitals which receive referrals from lower level facilities. However, this was within the range of globally the acceptable rates of up to 15% [29, 61].

Previous pregnancy history was found to be an important determinant of maternal outcome as shown in our study and other studies [62, 63] and so providers can use this information in their clinical decision making during health care provision to pregnant women.

In the year 2013, the Greater Accra region recorded as many as 201 institutional maternal deaths [64]. For the 6 months of data collection within 11 facilities in the region no maternal death was recorded amongst our participants post-delivery. Maternal deaths are usually a result of direct causes like hemorrhage, pre-eclampsia and eclampsia and sepsis as well as indirect causes like anemia [65]. These occur mostly during delivery or within 24 h afterwards [65, 66]. Although no maternal deaths occurred amongst our participants during the 6 weeks postpartum, it is possible that we missed some deaths that may have occurred before or during delivery since we recruited women who had already survived delivery or that our sample was not large enough to detect such an outcome in the postpartum period.