Assessing antenatal care utilization, socioeconomic inequalities and low birth weight outcomes: a population-based analysis using the 2022 Ghana demographic and health survey data Authors: Michael Arthur Ofori, Diana Maame Agyeiwaa Agyei, Abena Abokoma Asemanyi, Sesi Sedegah, Theophilus Asamoah, David Kwamena Mensah Published: 2026 ABSTRACT BackgroundSub-Saharan Africa continues to face significant challenges with low birth weight (LBW). Factors such as antenatal care (ANC) utilization and socioeconomic status play critical roles in birth outcomes. It is therefore imperative to understand these relationships, which are crucial to developing tailored, workable interventions to improve maternal and child health outcomes in Ghana.ObjectiveThis study aimed to examine the association between ANC utilization patterns, socioeconomic disparities, and low birth weight outcomes among women in Ghana.MethodsThe study included 6,965 observations, with 4,056 complete cases. Due to substantial missingness in key variables (LBW, adequate ANC), a monotone missing data pattern consistent with Missing at Random (MAR) was observed. Multiple Imputation by Chained Equations (MICE) was employed using Fully Conditional Specification across 20 imputed datasets. Survey-weighted logistic regression was performed on each imputed dataset, with estimates pooled via Rubin’s Rules. Model fit was assessed using the Hosmer–Lemeshow goodness-of-fit test. The slope index of inequality was calculated to examine socioeconomic disparities.ResultsThe prevalence of LBW was 10.0% in complete cases and 19.2% in imputed data. Socioeconomic inequality analyses revealed significant pro-rich gradients in ANC utilization. The disparities were more pronounced for optimal ANC (8 + visits; SII: 0.431; RII: 1.538) than adequate ANC (4 + visits; SII: 0.203; RII: 1.224). LBW was disproportionately concentrated among women of lower socioeconomic status (SII: −0.159; RII: 1.172). The adjusted multivariable model showed inadequate ANC (aOR: 1.97; 95% CI: 1.34–2.90), suboptimal ANC (aOR: 1.73; 95% CI: 1.27–2.36), and poor wealth status (aOR: 1.79; 95% CI: 1.18–2.72) were independently associated with higher odds of LBW.ConclusionANC visit frequency and household wealth are independently associated with LBW among Ghanaian women. These findings underscore the need for targeted interventions to improve ANC attendance and address socioeconomic disparities to reduce the burden of LBW in Ghana. Introduction Over the years, the relationship between antenatal care (ANC) utilization and birth outcomes, especially low birth weight (LBW), has been a key focus of inquiry in maternal and child health research. In Ghana, the importance of ANC cannot be downplayed, as it functions as an initial and essential point of contact for expectant mothers and has a direct influence on maternal and neonatal outcomes. Previous scholarship has illuminated persistent socioeconomic disparities in accessing quality ANC services, which aggravate adverse birth outcomes such as LBW [4, 13, 19, 25]. With the release of the 2022 Ghana Demographic and Health Survey (GDHS), a comprehensive assessment of ANC utilization, socioeconomic inequalities, and their effects on birth weight outcomes is now feasible. A growing body of research indicates a strong correlation between socioeconomic status and ANC service utilization, with those lower socioeconomic groups faced greater constraints in accessing these essential health services [1, 3, 5]. Findings from these studies intimate that people with a higher level of education are more likely to use maternal healthcare services, which will eventually improve their overall birth outcomes [3, 11, 13]. Additionally, LBW remains a topical and pressing public health challenge, especially in regions such as Sub-Saharan Africa. The estimated prevalence of LBW in SSA stands at approximately 13%, with considerable variation across individual countries [23]. LBW contributes significantly to neonatal mortality, accounting for more than one-third of neonatal deaths (34%) and increasing the risk for surviving infants of long-term health and developmental issues [2]. These consequences span the life course: LBW neonates who survive infancy are at increased risk for health, growth, and developmental problems, and those who are small for their gestational age may be predisposed to developing chronic adult disorders such as hypertension, type 2 diabetes, and heart disease [16]. In Ghana specifically, the national prevalence of LBW was reported at approximately 10% in 2014, though prevalence in some parts of the country is considerably higher, with studies reporting rates as high as 26% in parts of Northern Ghana [6]. Studies have shown that woefully inadequate access to quality ANC, coupled with maternal health issues including nutrition and overall health status, considerably contributes to the incidence of LBW [24, 26]. There are notably several determinants, including maternal age, education level, health insurance coverage, and geographical location, which play central roles in shaping patterns of ANC utilization [1, 4, 11]. It is worth noting that successive governments in Ghana have continued to implement health policies aimed at improving access to maternal health care. Notably, Ghana introduced the National Health Insurance Scheme (NHIS) in 2005. They subsequently launched the Free Maternal Healthcare Policy (FMHCP) in July 2008 through the NHIS. This was designed to provide comprehensive antenatal, delivery, and postnatal care services to pregnant women at no cost [7]. Additionally, the Community-based Health Planning and Services (CHPS) initiative has been deployed to extend primary health care access to underserved communities. Evidence suggests the fee exemption policy produced modest gains, yielding increases of 8%, 9%, and 21% in timely first ANC visit, completion of recommended ANC visits, and facility-based delivery, respectively [20]. However, these improvements have not translated uniformly into improved maternal and neonatal outcomes. Despite the implementation of the NHIS, Ghana still reports a universal health coverage service index below the global average. Persistently high rates of LBW and incomplete ANC coverage across socioeconomic groups underscore the gap between policy intent and actual outcomes [13, 25]. Geographic and financial barriers, supply-side deficiencies, and inequities in service quality continue to limit the reach of these policies, particularly among the rural poor. Therefore, understanding these dynamics is essential for informing effective health interventions that will be workable within these communities to improve maternal health as a whole. Methods Study design and data source This study employed a cross-sectional analytical design and used secondary data from the 2022 Ghana Demographic and Health Survey (GDHS), conducted between 17 October 2022 to 14 January 2023. The GDHS is a representative of a national household survey conducted periodically to provide current information on demographic, health, and nutrition indicators in Ghana. This 2022 GDHS was executed by the Ghana Statistical Service in partnership with the Ghana Health Service and ICF International, with technical assistance from the DHS Program. Sampling strategy The 2022 GDHS used a two-stage stratified cluster sampling design from the 2021 Ghana Population and Housing Census frame. In the initial stage, enumeration areas (EAs) served as primary sampling units and were selected with probability proportional to size. In the second stage, households were systematically sampled within each selected EA. All 16 administrative regions of Ghana were covered in this survey and provided distinct estimates for urban and rural areas. This sampling approach ensured equal national representation and adequate statistical power for key health indicators. Study population and sample The study population consisted of women of reproductive age (15–49 years) who had a live birth within five years preceding the 2022 GDHS. In all, the data comprised of 6965 data points. Among this, low birth weight had 2854 (40.97%) missing observations and adequate ANC had 1760 (25.27%) missing. Also, early ANC had 1858 (26.68%) data points missing. The missingness had a monotone missing data pattern. This missingness is more likely Missing at Random (MAR) or by design than Missing Completely at Random (MCAR). Most analyses of incomplete data take a complete case analysis approach. This is often based on the assumption that missing data are MCAR. The total complete cases were 4056. Due to the monotone nature of the missingness, we used Multiple Imputation by Chained Equations (MICE) to estimate the missing data. Three predictors and the response had missing data. We used fully conditional specification (FCS) multiple imputation with M = 20 imputed datasets under a missing-at-random assumption. Survey-weighted logistic regression was run on each imputed dataset separately, and estimates were combined using Rubin’s Rules via PROC MIANALYZE with degrees of freedom adjusted for the complex survey design. All results presented are pooled across the 20 imputations. The binary variables; lbw adequate ANC, optimal ANC, early ANC initiation, and residence were engineered under the FCS LOGISTIC with LIKELIHOOD=AUGMENT (augmentation stabilizes estimates in sparse cells). Also, the polytomous/ordinal variable; ANC categories was engineered under the FCS DISCRIM with CLASSEFFECTS=INCLUDE. Lastly, the continues variable; birthweight was categorized under the FCS REG. FCS was used rather than listwise deletion to reduce selection bias. Also, seed was fixed for reproducibility. Variables and definitions The primary outcome variable was LBW, which, according to World Health Organization standards, is defined as birth weight less than 2,500 g. As part of the 2022 GDHS data collection protocol, birth weight information was obtained either from health facility records or, where records were unavailable, from maternal recall. The variable was coded as a binary outcome: 1 for low birth weight (less than 2,500 g) and 0 for normal birth weight (2,500 g or more). This dichotomization conforms to international standards, which makes it easier to compare with other population-based studies. Four key ANC utilization variables were examined as primary exposures: Early ANC initiation Defined as the maiden ANC visit happening within the first trimester of pregnancy (first 12 weeks of gestation). This variable was dichotomized as “yes” (early initiation) or “no” (late initiation). This is in accordance with WHO recommendations that early contact enables timely risk assessment, nutritional supplementation, and infection screening. Adequate ANC visits This is based on the World Health Organization’s previous recommendation of at least four ANC visits during pregnancy. Women were categorized as having adequate ANC if they had attended four or more visits, and inadequate ANC if they had attended fewer than four visits. Optimal ANC visits Following the WHO’s updated 2016 recommendation of at least eight ANC contacts for a positive pregnancy experience. Women were categorized as having optimal ANC if they completed eight or more visits, and suboptimal ANC if they completed fewer than eight visits. Wealth status This refers to the household wealth index, derived from a principal component analysis of household assets, dwelling characteristics, and access to utilities and services. The DHS wealth index was categorized into three groups: poor (the lowest and second quintiles), middle (the middle quintile), and rich (the fourth and highest quintiles). Educational attainment Maternal education was categorized into four levels based on the highest education completed: no formal education, primary education, secondary education, and higher education (post-secondary or tertiary). Residence Place of residence was classified as urban or rural according to the Ghana Statistical Service’s definitions of urban and rural areas. Data analysis For this study, all statistical analyses were completed using SAS version 9.4 and incorporated survey weights, clustering, and stratification. This was done to account for the complex survey design and ensure equal national representative estimates using the Proc surveyfreq command with weight = weight, cluster = psu, and strata = strata. Weighted frequencies and percentages were also computed using both complete dataset and the imputed data stratified by low-birth-weight status. Simple logistic regression models were used to examine the bivariate relationship between each exposure variable and low birth weight in order to account for their individual contributions using thecrude odds ratios (cOR) and its 95% confidence intervals (CI). This was done using both complete cases and multiple imputation data. In order to examine the magnitude of socioeconomic inequalities in both ANC utilization and LBW outcomes, we calculated the Slope Index of Inequality (SII) using logistic regression models. Fractional rank (ri) was derived for each wealth quintile as the midpoint of the cumulative weighted population share. This placed the poorest at ≈ 0 and the richest at ≈ 1. PROC SURVEYLOGISTIC was used to regress each outcome on ri. Eventually, separate models were created for adequate ANC visits and low birth weight as outcomes. For each outcome in this study, adjusted odds ratios (aOR) and 95% confidence intervals were estimated for wealth status, educational accomplishment, and residence while controlling for other socioeconomic variables. This SII approach affords a comprehensive measure of absolute inequality across the entire socioeconomic gradient. Eventually, a multiple logistic regression model was conducted to ascertain independent predictors of low birth weight while concurrently controlling for all ANC utilization indicators and socioeconomic variables. Variables included in the final model were adequate ANC visits, early ANC initiation, wealth status, educational attainment, and residence. Adjusted odds ratios, 95% confidence intervals, and p-values were calculated for each predictor. Model fit and multicollinearity were evaluated using variance inflation factors and the Hosmer–Lemeshow goodness-of-fit test. The multivariable model enabled estimation of the independent impact of each exposure variable on LBW likelihood while accounting for possible confounding. Ethical considerations Since this study is a secondary analysis of publicly available, de-identified data from the 2022 Ghana Demographic and Health Survey, no additional ethical approval was required. Ethical clearance for the original survey was obtained from the Ghana Health Service Ethical Review Committee and the Institutional Review Board of ICF International. Informed consent was obtained from all participants prior to data collection by the DHS field teams. Data management and quality control To ensure data quality, the DHS Program’s standardized data collection protocols were followed, including interviewer training, field supervision, and data validation procedures. Data cleaning was done for this analysis by checking for logical inconsistencies, identifying and handling missing values, and authenticating the appropriate application of survey weights. Variables were recoded, which was essential to facilitate analysis, and all transformations were documented to ensure reproducibility. Results Descriptive statistics (weighted analysis) Table 1 presentsA total of 4,056 births were included in the complete case analysis, of whom 380 (10.0%) were low birth weight (LBW). Following multiple imputations, the analytical sample expanded to 6,965 births, with 1,349 (19.2%) classified as LBW. Among complete cases, the majority of mothers attended early antenatal care (ANC) (67.6%). and had adequate ANC of 4 or more visits (93.1%). Although, fewer than half achieved optimal ANC of 8 or more visits (43.1%). Table 1Descriptive StatisticsVariablesLow Birth WeightTotalNoYesComplete casesImputed dataComplete casesImputed dataComplete casesImputed data Total 3676(90.0) 5616(80.2) 380(10.0) 1349(19.2) 4056(100) 6965(100) Early ANC No 1190(28.8)1908(27.3)146(3.6)541(7.5)1336(32.4)2449(34.8) Yes 2486(61.2)3707(53.5)234 (6.4)808(11.7)2720(67.6)4516(65.2)Adequate ANC (4+) visits No 268(5.6)681(9.7)46(1.3)307(4.2)314(6.9)988(13.9) Yes 3408 (84.4)4934(71.6)334(8.7)1042(15.0)3742(93.1)5976 (86.1)Optimal ANC (8+) visits No 2198(49.8)3536(47.5)262(7.1)967(13.5)2460(56.9)4503(61.0) Yes 1478(40.2)2080(33.3)118(2.9)382(5.7)1596(43.1)2462(39.0)Residence Rural 1955(41.5)3154(39.3)212(5.1)835(10.5)2167(46.6)3989(49.8) Urban 1721(48.5)2461(41.5)168(4.9)515(8.7)1889(53.4)2976(50.2)Wealth Poor 1826(33.7)2983(32.7)209(4.4)842(9.7)2035(38.0)3825(42.4) Middle 720(18.0)1052(15.9)79(2.3)240(3.9)799(20.3)1292(19.8) Rich 1130(38.4)1581(32.2)92(3.3)267(5.7)1222(41.6)1848(37.9)Education No Education 889(15.5)1585(16.7)95(1.8)475(5.1)984(17.3)2060(21.8) Primary 537(12.6)892(12.1)51(1.4)234(3.2)588(14.0)1126(15.3) Secondary 1895(51.5)2690(44.0)199(5.9)558(9.5)2094(57.4)3248(53.5) Higher 355(10.4)448(7.9)35(0.9)83(1.5)390(11.3)531(9.4) Table 1Descriptive StatisticsVariablesLow Birth WeightTotalNoYesComplete casesImputed dataComplete casesImputed dataComplete casesImputed data Total 3676(90.0) 5616(80.2) 380(10.0) 1349(19.2) 4056(100) 6965(100) Early ANC No 1190(28.8)1908(27.3)146(3.6)541(7.5)1336(32.4)2449(34.8) Yes 2486(61.2)3707(53.5)234 (6.4)808(11.7)2720(67.6)4516(65.2)Adequate ANC (4+) visits No 268(5.6)681(9.7)46(1.3)307(4.2)314(6.9)988(13.9) Yes 3408 (84.4)4934(71.6)334(8.7)1042(15.0)3742(93.1)5976 (86.1)Optimal ANC (8+) visits No 2198(49.8)3536(47.5)262(7.1)967(13.5)2460(56.9)4503(61.0) Yes 1478(40.2)2080(33.3)118(2.9)382(5.7)1596(43.1)2462(39.0)Residence Rural 1955(41.5)3154(39.3)212(5.1)835(10.5)2167(46.6)3989(49.8) Urban 1721(48.5)2461(41.5)168(4.9)515(8.7)1889(53.4)2976(50.2)Wealth Poor 1826(33.7)2983(32.7)209(4.4)842(9.7)2035(38.0)3825(42.4) Middle 720(18.0)1052(15.9)79(2.3)240(3.9)799(20.3)1292(19.8) Rich 1130(38.4)1581(32.2)92(3.3)267(5.7)1222(41.6)1848(37.9)Education No Education 889(15.5)1585(16.7)95(1.8)475(5.1)984(17.3)2060(21.8) Primary 537(12.6)892(12.1)51(1.4)234(3.2)588(14.0)1126(15.3) Secondary 1895(51.5)2690(44.0)199(5.9)558(9.5)2094(57.4)3248(53.5) Higher 355(10.4)448(7.9)35(0.9)83(1.5)390(11.3)531(9.4) Descriptive Statistics LBW was more prevalent among neonates born to mothers in the poorest wealth quintile (4.4% in complete cases and 9.7% in imputed data) compared to wealthier counterparts. Also, among those with no formal education (1.8% in complete cases 5.1% in imputed data) relative to mothers with higher education. Rural residence was associated with a slightly higher proportion of LBW births compared to urban settings in both complete (5.1% vs. 4.9%) and imputed (10.5% vs. 8.7%) samples. Bivariate analysis of low birth weight Table 2 presents the bivariate analysis using both complete and imputed datasets. The unadjusted analyses with the complete case data revealed that inadequate ANC was significantly associated with nearly twice the odds of LBW (cOR: 2.17; 95% CI: 1.41–3.35; p = 0.0004). Likewise suboptimal ANC of fewer than 8 visits (cOR: 2.02; 95% CI: 1.50–2.72; p < 0.0001) was also significantly associated with nearly twice the odds of LBW. Following multiple imputation, pooled estimates attenuated and lost statistical significance for both adequate ANC (pooled cOR: 2.17; 95% CI: 0.97–4.82; p = 0.0579) and optimal ANC (pooled cOR: 1.72; 95% CI: 0.60–4.96; p = 0.2949). This suggests sensitivity to missing data. Poor wealth status was associated with significantly elevated odds of LBW in both complete (cOR: 1.53; 95% CI: 1.10–2.13) and imputed (pooled cOR: 1.69; 95% CI: 1.24–2.29) analyses. Rural residence was associated with higher odds of LBW in imputed data (pooled cOR: 1.28; 95% CI: 1.02–1.62; p = 0.0368). Meanwhile, this association did not reach significance in complete case analysis. No education was associated with elevated LBW odds in imputed analyses (pooled cOR: 1.68; 95% CI: 1.08–2.61; p = 0.0212). Lastly, early ANC timing showed no statistically significant association across either dataset. Table 2Bivariate analysis of Low Birth Weight, ANC utilization and Socioeconomic FactorsParameterComplete DataImputed DatacOR (95% CI)Pr > |t|Pooled cOR (95% CI)Pr > |t|Early ANC visit (ref = yes)1.20 (0.92, 1.58)0.18401.26 (0.86, 1.86)0.2222Adequate ANC visit (4+) (ref = yes)2.17 (1.41, 3.35)0.00042.17 (0.97, 4.82)0.0579Optimal ANC visit (8+) (ref = yes)2.02 (1.50, 2.72)< 0.00011.72 (0.60, 4.96)0.2949Wealth (ref = rich) Poor 1.53 (1.10, 2.13)0.01261.69 (1.24, 2.29)0.0012 Middle class 1.51 (0.98, 2.32)0.06271.39 (0.97, 1.97)0.0696Education (ref = Higher) No education 1.30 (0.74, 2.29)0.36431.68 (1.08, 2.61)0.0212 Primary education 1.23 (0.66, 2.28)0.51421.46 (0.93, 2.30)0.0997 Secondary education 1.28 (0.76, 2.16)0.34751.18 (0.79, 1.77)0.4063Residence (ref = Urban)1.24 (0.95, 1.63)0.11451.28 (1.02, 1.62)0.0368 Table 2Bivariate analysis of Low Birth Weight, ANC utilization and Socioeconomic FactorsParameterComplete DataImputed DatacOR (95% CI)Pr > |t|Pooled cOR (95% CI)Pr > |t|Early ANC visit (ref = yes)1.20 (0.92, 1.58)0.18401.26 (0.86, 1.86)0.2222Adequate ANC visit (4+) (ref = yes)2.17 (1.41, 3.35)0.00042.17 (0.97, 4.82)0.0579Optimal ANC visit (8+) (ref = yes)2.02 (1.50, 2.72)< 0.00011.72 (0.60, 4.96)0.2949Wealth (ref = rich) Poor 1.53 (1.10, 2.13)0.01261.69 (1.24, 2.29)0.0012 Middle class 1.51 (0.98, 2.32)0.06271.39 (0.97, 1.97)0.0696Education (ref = Higher) No education 1.30 (0.74, 2.29)0.36431.68 (1.08, 2.61)0.0212 Primary education 1.23 (0.66, 2.28)0.51421.46 (0.93, 2.30)0.0997 Secondary education 1.28 (0.76, 2.16)0.34751.18 (0.79, 1.77)0.4063Residence (ref = Urban)1.24 (0.95, 1.63)0.11451.28 (1.02, 1.62)0.0368 Bivariate analysis of Low Birth Weight, ANC utilization and Socioeconomic Factors Slope index of inequality assessment for adequate ANC visits and low birth weight Table 3 presents the Slope Index of Inequality Assessment for Adequate ANC visits and Low Birth Weight. The Slope Index of Inequality and Relative Index of Inequality revealed significant pro-rich inequalities across all three outcomes. For adequate ANC attendance (4 + visits), the SII was 0.203 (95% CI: 0.189–0.216), indicating that the most advantaged women had a 20.3% point higher probability of attending adequate ANC compared to the least advantaged. The inequality was even more pronounced for optimal ANC attendance (8 + visits), with an SII of 0.431 (95% CI: 0.408–0.454), reflecting a 43.1% point absolute gap between the wealthiest and poorest women. For low birth weight, the SII was − 0.159 (95% CI: -0.176 - -0.142), indicating that the poorest women faced a 15.9% point higher absolute probability of low birth weight compared to the wealthiest, representing a wealth-related inequality. Table 3Slope Index of Inequality Assessment for Adequate ANC visits and Low Birth WeightOutcomeMean PrevalenceStandard errorSII (95% CI SII)RII (95% CI RII)ANC 4 + visits0.8920.0020.203 (0.189, 0.216)1.224(1.208, 1.241)ANC 8 + visits0.4400.0040.431 (0.408, 0.454)1.538(1.503, 1.574)Low Birth Weight0.1630.003-0.159 (-0.176, -0.142)1.172(1.153, 1.192) Table 3Slope Index of Inequality Assessment for Adequate ANC visits and Low Birth WeightOutcomeMean PrevalenceStandard errorSII (95% CI SII)RII (95% CI RII)ANC 4 + visits0.8920.0020.203 (0.189, 0.216)1.224(1.208, 1.241)ANC 8 + visits0.4400.0040.431 (0.408, 0.454)1.538(1.503, 1.574)Low Birth Weight0.1630.003-0.159 (-0.176, -0.142)1.172(1.153, 1.192) Slope Index of Inequality Assessment for Adequate ANC visits and Low Birth Weight Multiple logistic regression model for low birth weight Table 4 shows the results of the multiple logistic regression model examining independent predictors of low birth weight while concurrently controlling for all ANC utilization and socioeconomic variables. The pooled imputed data was used for the multivariable model. The result showed that inadequate ANC remained independently associated with elevated odds of LBW after adjustment for socioeconomic and residential covariates (aOR: 1.97; 95% CI: 1.34–2.90; p = 0.0007). Similarly, suboptimal ANC of fewer than 8 visits was associated with 73% higher odds of LBW (aOR: 1.73; 95% CI: 1.27–2.36; p = 0.0005). This underscores the independent contribution of ANC frequency beyond adequacy thresholds alone. Early ANC initiation was not independently associated with LBW (aOR: 0.93; 95% CI: 0.71–1.24; p = 0.6263) after adjustment. Table 4Multiple Logistic Regression model for Low Birth weightParameterPooled aOR95% CIPr > |t|VIFIntercept< 0.0001 Adequate ANC visit (4+) (ref = yes)1.9681.3372.8980.00071.175 Early ANC visit (ref = yes)0.9330.7051.2350.62631.209 Optimal ANC (ref = yes)1.7321.2742.3550.00051.206Wealth (ref = rich)1.724 Poor1.7901.1762.7240.0067 Middle class1.5030.9642.3430.0720Education (ref = Higher)1.332 No education1.0180.5481.8910.9561 Primary education0.9230.4981.7090.7985 Secondary education0.9630.5581.6620.8935Residence (ref = Urban)1.0930.8081.4800.56281.405 Table 4Multiple Logistic Regression model for Low Birth weightParameterPooled aOR95% CIPr > |t|VIFIntercept< 0.0001 Adequate ANC visit (4+) (ref = yes)1.9681.3372.8980.00071.175 Early ANC visit (ref = yes)0.9330.7051.2350.62631.209 Optimal ANC (ref = yes)1.7321.2742.3550.00051.206Wealth (ref = rich)1.724 Poor1.7901.1762.7240.0067 Middle class1.5030.9642.3430.0720Education (ref = Higher)1.332 No education1.0180.5481.8910.9561 Primary education0.9230.4981.7090.7985 Secondary education0.9630.5581.6620.8935Residence (ref = Urban)1.0930.8081.4800.56281.405 Multiple Logistic Regression model for Low Birth weight Among socioeconomic determinants, poor wealth status was associated with significantly higher odds of LBW relative to the richest quintile (aOR: 1.79; 95% CI: 1.18–2.72; p = 0.0067). Meanwhile, the middle-class wealth did not reach statistical significance (aOR: 1.50; p = 0.0720). Educational attainment and rural residence showed no statistically significant independent associations with LBW in the fully adjusted model. Variance inflation factor (VIF) values across all covariates were below 2.0, indicating no meaningful multicollinearity. The Hosmer–Lemeshow goodness-of-fit test indicated adequate model fit (χ² = 7.999; df = 8; p = 0.4335). It suggests that there was no significant departure between observed and predicted probabilities. Discussion The results of this study provide vital insight into public health challenges in Ghana. The findings emphasize the association of socioeconomic factors on maternal health, particularly concerning the prevalence of LBW among newborns. The prevalence of LBW was 19.2% (95% CI: 18.3–20.2%) in imputed data. This exceeds the WHO global estimate of approximately 15%. This is higher than the 13.5% reported in the 2017 GDHS and the 14.8% reported in the 2014 GDHS [2], suggesting a possible upward trend in LBW prevalence in Ghana. This finding is higher than previous reports in Burkina Faso (13.4%), Ghana (10.2%), Malawi (12.1%), Senegal (15.7%) and Uganda (10.0%) [14]. Meanwhile, the prevalence of 10.0% (95% CI: 9.0–10.9%) in complete cases data was lower than the WHO threshold. The higher prevalence observed in this study compared to earlier rounds may be attributed to missing data imputation, self-reported birth weight, or survey design and population. In addition, this can also be due to the changes in data collection methods, birth weight recording practices, or sample composition across survey. This highlights a growing public health burden that demands targeted and community-level policy interventions. It is obvious that the trends of ANC utilization among pregnant women are significantly associated with the occurrence of LBW. Although early ANC visit was not found to be statistically significant yet women who initiated ANC late had higher odds of LBW compared to those who initiated ANC early. The associations of early ANC and LBW has been reported in prior studies [18]. Also, early ANC initiation is linked with improved maternal and infant health outcomes [1, 12]. Furthermore, the data uncover that a vast majority of women managed to get adequate visits. This shows a clear disparity in ANC quality that potentially aggravates health outcomes, as insufficient ANC visits strongly correlate with LBW likelihood (Nketiah‐Amponsah et al., 2013; [4, 17]). Previous scholarship corroborates this by validating that access to adequate ANC services significantly decreases LBW incidence [15, 22]. The current analysis reflects a solid socioeconomic gradient in both ANC utilization and LBW outcomes. Women from poorer households demonstrated significantly higher LBW prevalence than their counterparts who are richer, a growing trend that aligns with bigger patterns spotted in other developing nations [10, 21]. Precisely, the study findings reveal that poverty is associated with health disparities, with 44.1% of the sample belonging to poor households and suffering disproportionately higher LBW rates [18]. This is supported by multiple studies demonstrating that wealth-based disparities in maternal health service uptake are prevalent, further underscoring the need for targeted interventions for impoverished populations [8, 9]. These current findings present the complex relationships between ANC utilization trends, socioeconomic status, educational factors, geographical disparities, and LBW outcomes. The total socioeconomic gradient in LBW, as captured by the SII reflects the cumulative burden of poverty operating through all pathways including constrained ANC access. The persistence of wealth as an independent predictor in Table 4, even after adjusting for ANC utilization, indicates that poverty is associated birth outcomes through pathways such as nutritional deficiency, chronic stress, environmental exposures, and limited access to emergency obstetric care. None of these are fully captured by ANC attendance alone. Limitation We acknowledge the following as limitation to the study. First, the cross-sectional nature of this study limits causal inference. Second, we defined Optimal ANC visits as eight or more visits following the 2016 WHO recommendation. The 2016 WHO guidelines shifted terminology from “visits” to “contacts” to reflect a broader, more flexible model of care. A contact encompasses any interaction between a pregnant woman and a healthcare provider. This may include in-person clinic visits, home visits by community health workers, telephone or telehealth consultations, and outreach or community-based encounters. The GDHS captures facility-based ANC visits, potentially missing non-facility contacts. This implies that the “8 + visits” threshold may underestimate the true number of contacts women had. Policy implications The study’s findings have several significant implications for maternal and child health policy in Ghana. First, efforts to reduce LBW must focus on improving both the coverage and quality of ANC services, particularly on early initiation, and ensure that women complete at least four visits. This calls for the to address structural barriers including transportation costs, clinic accessibility, and service availability, especially in underserved and marginalized areas. Second, the persistent wealth gradient in LBW outcomes, independent of healthcare utilization, underscores the exigency of poverty alleviation intervention and economic support programs for pregnant women. Interventions such as conditional cash transfers, nutritional supplementation programs, and social protection schemes targeting poor pregnant women should be implemented as they may help address the upstream determinants of LBW. The results reflect a fundamental health communication and literacy challenge that must be addressed. This will make interventions and programs in the communities successful. We acknowledged that health literacy and communication effectiveness were not assessed. We suggest that future work looks into communication, literacy and health outcomes. When expectant mothers are unable to understand the health information, it makes it difficult to translate it into action. Conclusion This study showed that ANC visit frequency and household wealth are important independent factors associated with LBW in Ghana. In addition, socioeconomic inequalities also add to the burden of the poorest women. The significant pro-rich gradient in both adequate and optimal ANC use points to structural challenges which go beyond individual health-seeking behaviors. The results of these findings call for specific interventions that would promote attendance at ANC that are targeted and aimed at equity. It must also tackle the socio-economic factors that underpin poor birth outcomes. Quality and accessibility of ANC services should be improved, with a special focus on rural and economically deprived women. This is very important to complement efforts at reducing maternal and neonatal health outcomes as well as prevalence of LBW in Ghana.