This study assessed the prevalence of AIP and its correlates at various stages of hemoglobin assessment during ANC. The findings have established the prevalence of AIP at ANC registration as 35.8% without and 25.3 with adjustment for haemodilution; at 28 weeks of gestation as 53.1% without and 37.5% with adjustment for haemodilution; and at 36 weeks as 44.8%. Variations in the AIP statuses of the respondents could be explained by a number of factors including maternal age, religion, education, household wealth, place of residence, type of healthcare facility, and timing of ANC registration, which were found to be statistically significantly associated with AIP at various stages of hemoglobin assessment.
Without adjustment for haemodilution, the observed AIP prevalence reported at ANC registration, was lower compared to similar studies both in Ghana [36, 52] and in neighboring countries [53–55]. Also, the AIP prevalence at 36 weeks in this study was lower compared to a similar study in northern Ghana  but higher than that of the finding of a study in the middle belt of Ghana . Contextual differences in study settings [7, 58–60] are more likely to account for these observed differences in AIP prevalence. While maternal healthcare is free in Ghana, the availability, access to, utilization and the quality of healthcare obviously varies  between the northern and middle belt of Ghana. Also, a relatively higher availability and accessibility to nutritious foods in the middle belt compared to the north is likely to account for the observed differences in the AIP prevalence. The interventions in ANC are expected to correct anaemia identified at registration and at 28 weeks and prevent non-anaemic registrants from becoming anaemic. AIP prevalence should therefore be lesser in subsequent points of test especially at ≥36 weeks of gestation. This study however, illustrates worsening AIP prevalence after ANC registration. Some previous studies assessing changes in anaemia status after some interventions have also reported worsening situations [54, 62] although others reported reduced prevalence [46, 56, 63]. A worsening situation suggest little or unsuccessful correction of AIP during ANC as evidenced in the non-significant changes between 28 weeks and 36 weeks after accounting for haemodilution. Despite the lower prevalence in the study area, AIP is still of great concern, especially as it is one of the leading cause of maternal mortality . As a moderate public health problem at registration which transitions into a severe one at 36 weeks , its impact on maternal outcomes cannot be overemphasized. Stepped up efforts should therefore be taken to address it. Preconception nutrition and healthcare which will prevent or reduce it at registration is recommended.
In current practice, anaemia is diagnosed using a single cut off point of 11g/dl throughout pregnancy regardless of the time of assessment. As seen in this study, this practice presents a totally different pattern of anaemia prevalence along the course of pregnancy taken into account the haemodilution theory (reference). Without adjusting for haemodlition, AIP peaks at 28 weeks with a much higher prevalence. However, after adjusting for haemodilution, a rising straight-line curve pattern, demonstrating a continuous increasing prevalence from registration to 36 weeks, was observed. This could have some serious implication for planning for anaemia control and prevention in pregnancy. Without haemodilution, healthcare providers would expect the highest prevalence of AIP around 28 weeks and may place so much efforts on it. However, this could be deceptive and could lead to poor prioritization of care. Even though the rates of AIP as observed in this study are very high and under no circumstance requires urgent attention, the adjustment changed the prevalence pattern which could affect decision making. For instance, wrong conclusion could have been drawn relying on the observation that AIP prevalence peaks at 28 weeks and declines afterwards without haemodilution, which would have suggested some false level of success in correction. Haemodilution is a natural physiological process in pregnancy that leads to decrease haemoglobin levels by 1 - 2g/dl but does not imply a disease state . Indeed, there is an argument that haemoglobin above 14.30g/dl (43% haematocrit) may signal poor haemodilution which poses danger to mother and baby . Hence, the argument for an adjustment of 0.5g/dl to be made for diagnosis at the 2nd trimester where haemodilution peaks could not be more apt. [4, 25]. Thus, there is the need to review the anaemia diagnosis at 28 weeks of gestation. Perhaps it is time to adopt the cutoff point of 10.5g/dl as recommended in Ghana [4, 25]. The different prevalence pattern of AIP as observed vis a vis haemodilution may also be a diagnosing error in anaemia research where all pregnant women regardless of their state of haemodilution are assessed using a common cutoff and group prevalence is reported.
Age offered significant consistent protection at 36 weeks, as older age groups had reduced risk of AIP. While this finding may be limited to women at 36 weeks of gestation, it agrees with other studies that reported increased AIP risk for younger mothers [42, 55, 66–68] and protection for older mothers . Also, the women in the middle age group had some protection against anaemia at 28 weeks when haemodilution is not taken into consideration. This also falls in line with some studies that found isolated risk for specific age groups [70, 71]. Also, the lack of association between anaemia at registration and age groups as observed in this study has also been reported in other studies [35, 36, 72].
The trimester of booking was a significant correlate of anaemia status throughout the three stages of assessment. The consistence was however achieved after adjusting for haeodilution at 28 weeks. This finding agrees with some assertions of increased risk of anaemia for registering after 1st trimester [36, 44, 73] and echoes the report of increasing risk with increasing trimester of registration . It is however, in variance with Ampiah and Colleagues finding of reduced risk for registering in the 2nd trimester. Several reasons may account for this. Even though feeding issues such as appetite, nausea and vomiting issues are said to predominate in the 1st trimester , early booking enable early intervention, thereby enhancing nutrient intake. However, these likely determinants were not assessed in this study. Also, the effects of these are also most likely to be felt in later trimesters than in 1st trimester, thereby increasing the risk in those times. Furthermore, early registration could increase the likelihood of more doses of iron and folic acid consumed thereby reducing the risk of anemia even in later points of haemoglobin check. This suggests the need to enable and encourage women to register early for ANC as a strategy for anaemia control and prevention in pregnancy.
The type of health care provider had significant influence on anaemia status at registration only. Attending ANC at the private hospitals was associated with reduced risk of anaemia at registration but attending ANC at the regional hospital was associated with increased risk of anaemia at registration. Very important to note is that, more women (64%) registered early in the private hospitals than all other facilities. In practice and as may be observed by those not in the healthcare setting, the cost of services at the private facilities are higher than public facilities in Ghana even when the client has NHIS coverage. People registering at private hospital are therefore likely to be more committed to achieving good pregnancy outcomes or are wealthy women with the needed resources. Hence, they will most likely register early and comply with recommended behaviours. On the other hand, more (53.2%) women booked late at the Regional Hospital. Also, as a public facility and referral centre, clients booking there may be less likely to be committed and/or referred there due to previous risky (including anaemia risk) pregnancies. Hence, such women could have chosen the regional hospital because of pre-known anaemia issues before booking.
Resident of urban areas had some protection against anaemia at 28 weeks after adjusting for haemodilution. Previous studies have reported similar findings [12, 42, 77, 78]. Availability, access and utilization of highly skilled and comprehensive ANC or anaemia-related care; and arguably, at a lower cost is an advantage to urban women over their rural counterparts. Certainly, travel expenses disrupts care access and utilization by pregnant women  and so will place rural women at the disadvantage side. For instance, NHIS accredited laboratories, pharmacies, drug stores are all located in the urban area of the municipality. Indeed Ayoya and colleagues report that inaccessibility to IFA supplementation is far more common in rural area than in urban areas . Also, women in urban areas are more likely to have higher (tertiary) education and be members of richest households which are both protective factors of anaemia at 28 weeks reported in this study . As such they stand more empowered to carry out recommended anaemia prevention and control behaviours that offer protection.
Christianity offered statistically significant protection against anaemia over non-Christian at 28 weeks. The difference in anaemia in pregnancy risk determined by religion has also been reported in India . Religion as a social determinant of health shapes the beliefs, norms, myths and conventions surrounding health and health-related behaviours . These health-related behaviours could range from food and hygiene norms to health seeking ones that can impact on the incidence and recovery from anaemia. Even though not assessed in this study, the more liberal Christian norms regarding animal foods which are high bioavailable sources of protein, iron and other micronutrients needed for haemoglobin formation could account for this.
Similarly, educational status of spouses offered some protection against anemia at 36 weeks but only at the secondary level. Generally, education as social determinant has been linked to higher nutrition and health literacy [81, 82] and higher economic fortunes [80, 81, 83] which directly impact on healthy behaviour leading to better health outcomes. Women with highest level of education are therefore more likely to have the cognitive and affective ability to access, understand, appraise anaemia prevention and control information as well as the material resources to use that information to stay non-anaemic or quickly recover from anaemia at 28 weeks or 36 weeks. Moreover, such women have good social standing and are therefore more likely to receive the needed attention. They are also more likely to be able to navigate effectively in the healthcare settings to demand care that meets their needs. This therefore calls for the education of the female child for better nutrition and anaemia outcomes in pregnancy.
Last but not the least, women in the first two wealthier households as expected were protected against anaemia at 28 weeks and 36 weeks. Previous studies have also documented similar findings [12, 78]. Wealth has been identified as a determinant of several health outcomes [81, 84].. While how wealth impact on anaemia status may be complex , the increase likelihood to have effective demand to anaemia prevention and control related commodities such as drugs, nutritious food, etc. is an advantage for women from wealthy households. This therefore may be the pathway to protection against anaemia in pregnancy. Empowering women economically may therefore be a means to the prevention and control of anaemia.