Temporal trends in childhood mortality in Ghana: impacts and challenges of health policies and programs

Background Following the adoption of the Millennium Development Goal 4 (MDG 4) in Ghana to reduce under-five mortality by two-thirds between 1990 and 2015, efforts were made towards its attainment. However, impacts and challenges of implemented intervention programs have not been examined to inform implementation of Sustainable Development Goal 3.2 (SDG 3.2) that seeks to end preventable deaths of newborns and children aged under-five. Thus, this study aimed to compare trends in neonatal, infant, and under-five mortality over two decades and to highlight the impacts and challenges of health policies and intervention programs implemented. Design Ghana Demographic and Health Survey data (1988–2008) were analyzed using trend analysis. Poisson regression analysis was applied to quantify the incidence rate ratio of the trends. Implemented health policies and intervention programs to reduce childhood mortality in Ghana were reviewed to identify their impact and challenges. Results Since 1988, the annual average rate of decline in neonatal, infant, and under-five mortality in Ghana was 0.6, 1.0, and 1.2%, respectively. From 1988 to 1989, neonatal, infant, and under-five mortality declined from 48 to 33 per 1,000, 72 to 58 per 1,000, and 108 to 83 per 1,000, respectively, whereas from 1989 to 2008, neonatal mortality increased by 2 per 1,000 while infant and under-five mortality further declined by 6 per 1,000 and 17 per 1,000, respectively. However, the observed declines were not statistically significant except for under-five mortality; thus, the proportion of infant and under-five mortality attributed to neonatal death has increased. Most intervention programs implemented to address childhood mortality seem not to have been implemented comprehensively. Conclusion Progress towards attaining MDG 4 in Ghana was below the targeted rate, particularly for neonatal mortality as most health policies and programs targeted infant and under-five mortality. Implementing neonatal-specific interventions and improving existing programs will be essential to attain SDG 3.2 in Ghana and beyond.


Introduction
Early childhood mortality continues to remain a prominent global health issue even though Millennium Development Goal 4 (MDG 4) was universally adopted to reduce underfive mortality by two-thirds between 1990 and 2015 (1). Also, several 'calls for action' to reduce neonatal mortality have been made (2Á5), and in response, both governmental and non-governmental bodies have committed considerable resources to this public health challenge. Similar to other low-and middle-income countries (LMICs), postadoption of MDG 4 in Ghana has witnessed formulation and implementation of maternal and child health policies and intervention programs towards actualizing MDG 4. Program (SMP) (6), Life Saving Skills (LSS) program (7), and Integrated Management of Childhood Illness (IMCI) program (8) were initiated. The SMP aims to secure safe delivery for women and improve child health services while the LSS (7) seeks to sharpen the clinical skills of midwives. Similarly, the IMCI (8) targets to improve child survival through the provision of clinical guidelines for management of childhood illnesses, health system strengthening, and improving community health practices. In the subsequent decade, from 1998 to 2008, some additional intervention programs and policies implemented were the Community-Based Health Planning and Services (CHPS) (9), User Fees Exemption for Delivery (UFED) (10), Focused Antenatal Care (FANC) (11), and National Health Insurance Scheme (NHIS) (12). The CHPS program aims to bring healthcare closer to the people through primary health care service while the UFED (10) and NHIS programs seek to ease the financial burden of healthcare service and reduce inequality in healthcare uptake. The FANC pursues improvement in maternal and child survival through individualized antenatal care that entails a comprehensive assessment of pregnant women in terms of their socio-cultural beliefs, lifestyle, and medical characteristics to improve early detection and treatment of illness and pregnancy complications. In addition to these national programs and policies, various regions also implemented different intervention programs, for example, the Kybele program in the Greater Accra region (13,14), Accelerated Child Survival and Development (ACSD) (15) sponsored by United Nation Children and Education Fund (UNICEF) in the Northern, Upper East, and Upper West regions, and Kangaroo Mother Care (KMC) (16) which commenced in six regions in 2007. Although the deadline for the attainment of MDG 4 has elapsed, 99% of childhood mortality still occurs in LMICs (5,17), with Africa accounting for about 50% (18). Assessment of progress made so far is of utmost importance to inform policy makers and healthcare planners tasked to realize the newly adopted Sustainable Development Goal 3.2 (SDG 3.2) that seeks to end preventable deaths of newborns and under-five children by 2030.
Thus, this study aimed to 1) compare the temporal trends in neonatal, infant, and under-five mortality in Ghana from 1988 to 2008; 2) describe the trends in the proportion of infant and under-five mortality attributed to neonatal deaths in Ghana over the same period; 3) compare national and regional trends in neonatal mortality over the same period; and 4) identify the impact and challenges of health policies and intervention programs implemented in Ghana during this time period.

Setting
Ghana is located in sub-Saharan Africa, along the Gulf of Guinea with a total population of about 24.4 million (19).
It has an annual growth rate of about 2.4% per year (20). Ghana has 10 administrative regions, namely Greater Accra, Western, Central, Volta, Eastern, Ashanti, Brong-Ahafo, Northern, Upper East, and Upper West. It has about 100 ethnic groups with different languages but the major ethnic groups are Akan, Ewe, Mole-Dagbane, Guan, and Ga-Adangbe (21).

Design of data collection
This longitudinal study compared the trends in neonatal mortality, infant and under-five mortality, and described the trends in the proportion of infant and under-five mortality attributed to neonatal deaths in Ghana, from 1988 to 2008, using Ghana Demographic and Health Survey (GDHS) datasets obtained in 1988, 1993, 1998, 2003, and 2008 (22). These datasets were collected by the ICF Macro in conjunction with the Ghana Statistical Service and the Ministry of Health/Ghana Health Service. All the GDHSs followed the same sampling technique; households were randomly sampled for interview by applying a stratified, two-stage cluster random sampling technique. All women and men in all the selected households, within the age range 15Á49 and 15Á59 years, respectively, were targeted for face-to-face interview using questionnaires. Prior to the interview, informed consent was obtained from every participant. The datasets are nationally representative with an individual response rate of 95Á97% and a household response rate of 97Á99%. The datasets were weighted to have a better representation of the study population. In GDHS, neonatal mortality was defined as the probability of dying within the first month of life, infant mortality was defined as the probability of dying before the age of 12 months, and under-five mortality was defined as the probability of dying before the age of 60 months. Detailed information on the sampling techniques and procedures for the data collection has been published elsewhere (22). In order to highlight the impact and challenges of health policies and intervention programs implemented in Ghana from 1988 to 2008, MEDLINE, EMBASE, Google Scholar, African Index Medicus, and Ghana Medical Journal were searched, and the articles that assessed the impact and challenges of these interventions implemented from 1988 to 2008 in Ghana were identified and reviewed.

Statistical analysis
Neonatal, infant, and under-five mortality rates estimated at national and regional level from each GDHS were used to perform trend analysis. Temporal trend patterns were depicted by plotting the number of neonatal deaths per 1,000 live births against the year when the data were captured; infant and under-five mortality underwent a similar analysis at national level. Also, temporal trend pattern of neonatal mortality at the national level was compared with that of regions. Likewise, the proportion of infant and under-five mortality attributed to neonatal mortality was examined by plotting the percentage of infant and under-five mortality attributed to neonatal death against the year when the data were captured. In order to quantify the trends objectively, a Poisson regression analysis was applied to quantify the incidence rate ratios of the trends. Statistical significance was determined by two-tailed Wald test at significant level of alpha equal to 5%; all analyses were performed in Stata statistical software package version 11 (23).

Ethical approval
Anonymous publicly available data were utilized in this study. Thus, no ethical approval is required. Table 1 shows the total number of live births captured per each GDHS and the number of neonatal, infant, and under-five deaths. Over this period, five demographic and health surveys were conducted in Ghana for which a total of 16,474 live births (average 3,295 live births per GDHS) were captured. Total neonatal, infant, and under-five deaths captured over this period was 673 (average 135 deaths per GDHS), 1,013 (average 203 deaths per GDHS), and 1,378 (average 276 deaths per GDHS), respectively. The average rates of decline per year for neonatal, infant, and underfive mortality were 0.6, 1.0, and 2.1%, respectively.

Descriptive statistics
National trends in neonatal, infant, and under-five mortality Figure 1 shows the trends in neonatal, infant, and underfive mortality from the 1988 to the 2008 GDHS while Table 2 reports the results of Poisson regression analysis that quantified the changes in the trends observed in Fig. 1. From 1988 to 1998, neonatal mortality declined from 47.9 per 1,000 to 33.1 per 1,000 and by 2008 neonatal mortality increased to 35.4 per 1,000. Considering the results in Table 2, neonatal mortality has not witnessed any significant decline over this period.
Infant mortality declined from 72.3 per 1,000 to 58.2 per 1,000 from 1988 to 1998 and by 2008 infant mortality dropped to 52.5 per 1,000. However, the results in Table 2 shows that the decline observed in infant mortality from 1988 to 2008 was not statistically significant. From 1988 to 1998, under-five mortality declined from 107.8 per 1,000 to 82.5 per 1,000 and by 2008 under-five mortality had further declined to 66.2 per 1,000. Over the same period, the results in Table 2 shows that under-five mortality was significantly lower in 1993 and 2008 when compared with 1988. In 1993 and 2008, the risk of under-five death was reduced by 38% (IRR 00.62; 95% CI: 0.46Á0.84) and 39% (IRR 00.61; 95% CI: 0.45Á0.83), respectively, when compared with that of 1988. Figure 2 depicts the trends in the proportion of infant and under-five mortality attributable to neonatal deaths. From 1988 to 1998, the percentage of infant mortality attributed to neonatal mortality declined from 66 to 57%; however, by 2008, it increased to 67%. Likewise, from 1988 to 1998, the proportion of under-five mortality attributable to neonatal deaths reduced from 44 to 40%; however, by 2008 it increased to 53%.

Regional trends in neonatal mortality
The regional trends of neonatal mortality are shown in Fig. 3. In 1988, neonatal mortality rates in the Central, Volta, and Ashanti regions were above the national rate (48 neonatal deaths per 1,000 live births); in 1998, Central, Eastern, Brong Ahafo, Upper East, and Upper West regions had a higher neonatal rate than the national average (33 neonatal deaths per 1,000 live births). By 2008 the Central, Upper West, and Northern regions exceeded the national neonatal mortality rate (35 neonatal deaths per 1,000 live births). The neonatal mortality in the Central region was persistently higher than the national average, whereas neonatal mortality in the Greater Accra region (GAR) stayed below the national average from 1988 to 2008.

Impact of implemented health policies and intervention programs on MDG 4 in Ghana
Following the adoption of the MGDs in Ghana, the Ghanaian government in collaboration with international donors implemented several intervention programs and health policies aimed at accelerating attainment of MDG  Table 3. At the national level, the LSS (7), SMP (6), and IMCI (8) programs were initiated between 1988 and 1998 and subsequently scaled up thereafter. Additional interventions such as the CHPS (9), User Fees Exemption for Delivery Care (UFEDC) (10), FANC (11), and the NHIS (12) were implemented from 1999 onward to complement the impact of the existing programs so as to accelerate attainment MDG 4 and MDG 5. Maternal and child policies reviewed (Table 3) showed that most of the policies were directed at maternal, infant, and under-five mortality rather than neonatal mortality. Results presented indicate that these policies seem to have a greater effect on maternal healthcare utilization and maternal and childhood mortality and morbidity during the initiation phase than the scale-up phase (24Á26, 48). Factors consistently identified to have a negative impact on the effectiveness of the various interventions were: deviation from good standard practice in policy formulation and implementation, erratic funding, insufficient community engagement, lack of proper monitoring, and inadequate manpower and equipment.

Discussion
This study compared the trends in neonatal, infant, and under-five mortality from 1988 to 2008 in Ghana. It also identified the impact and challenges of various health policies and programs implemented during this time period to attain MDG 4. Despite the global attention on childhood mortality, we noticed that from 1988 to 2008 in Ghana, the decline rates in neonatal, infant, and under-five mortality were far below the expectation of a 4% annual decline to attain MDG 4 globally (49) and less than the 7% annual reduction stipulated to achieve MDG 4 in sub-Saharan Africa (50).  (7), and IMCI (8)) were observed to have a larger effect on childhood mortality than those implemented from 1998 to 2008 (UFEDC (10), Focus Antenatal Care (FANC) (11), NHIS (12), and CHPS (9)).
Generally, the decline rates in neonatal, infant, and under-five mortality were far below expectations, and the implemented health policies and intervention programs appeared to have had more impact on under-five mortality than on neonatal and infant mortality. Due to the paltry decline in neonatal mortality, the proportion of infant and under-five mortality attributed to neonatal mortality has increased; this mimics global and SSA observations (49,53). In addition, we identified in our review factors that were responsible for the slow decline observed in neonatal, infant, and under-five mortality. Studies that have evaluated health policies and intervention programs implemented in Ghana repeatedly showed that factors  such as deviation from good standard practice in policy formulation and implementation (27,28), erratic funding (29,30), insufficient community engagement (9), inadequate monitoring (31,32), and inadequate manpower and equipment (29,33) are major challenges of health policies and programs that might have hindered a more pronounced decline in childhood mortality. Our observation was corroborated by a previous multi-country study that identified factors such as inadequate policy formulation and implementation, poor financing, shortage of health human resources, lack of re-training of staff, inadequate medical products and technologies as the major constraints to scale up intervention programs to improve survival in early life (54). At the regional level, we observed some degree of variation in neonatal mortality trends. This observation may partly be explained by differences in implementation of national health policies and programs in conjunction with the disparities in additional programs implemented in the regions; examples of such regional differences are the Kybele program in the Greater Accra region (13,14), kangaroo mother care (55), UNICEF-sponsored ACSD (15) in Northern Ghana, High Impact Rapid Delivery (HIRD) (56), and Project Five Alive (57,58). The variation may also be driven by differences in baseline rate of neonatal mortality across the regions.

Recommendation
Considering the slow rate of decline in childhood mortality, particularly in neonatal mortality, implementation of cost-effective, neonatal-specific interventions, such as newborn resuscitation, exclusive breastfeeding, use of partograph, kangaroo mother care, use of micronutrients, tetanus toxoid immunization, will be needed to successfully address attainment of SDG 3.2 (2,5,59). In addition, implemented interventions to tackle childhood mortality should be reformed based on the recurrent defects identified in policy formulation and implementation to accelerate attainment of SDG 3.2 (9, 31, 34).

Study limitations and strengths
This is the first study in Ghana that utilized nationally representative data to examine trends in childhood mortality, allowing us to generalize our findings. GDHS data are generally regarded as high-quality data because   Consequently, reimbursement was erratic and insufficient (30).   of the sampling technique and the excellent household and respondent response rates (22). We went beyond the traditional graphical description of the mortality trends by applying Poisson regression to quantify the risk of dying over time. However, we are aware that there may have been the possibility of underreporting and misclassification in childhood mortality as a result of recall bias (60). In addition, non-sampling error such as misunderstanding of the question on the part of the participant or the interviewer could have occurred. As the current study was based on published articles, some valuable information on the impact and challenges of the intervention programs implemented and reported in the grey literature may not have been fully captured in this study. Also, most articles that assessed the implemented intervention programs were not properly designed to evaluate the effectiveness of these intervention programs.

Conclusion
This study compared the trends in neonatal, infant, and under-five mortality over two decades in Ghana. The observed decline rates were generally slow, particularly for neonatal mortality. This could be attributed to the shortcomings identified for health policies and intervention programs formulation and implementation, particularly with regard to neonatal mortality. Implementation of a sustainable evidence-based neonatal-specific intervention and improving other existing interventions will be a prerequisite to actualize SDG 3.2 in Ghana and beyond.

Summary
What's known: Interventions were implemented in Ghana to achieve MDG 4 but the impact and challenges have not been assessed to inform SDG 3.2.
What's new: Since 1988, the decline in childhood mortality in Ghana was below the expected rate and the proportion of infant and under-five mortality attributed to neonatal death has increased because implementation of most intervention programs was suboptimum and newborns less considered. Implications: Implement neonatal-specific interventions and improve existing programs.