Association between early alcohol use and academic achievement among primary school children in Uganda: a cross-sectional study

Abstract

Background

Child alcohol use is a public health concern in Uganda. This study assessed academic achievement skills and their association with alcohol use among primary school children in Mbale district, Uganda.

Methods

This cross-sectional study assessed academic achievement skills (word reading, sentence comprehension, spelling, and math computation) in 6–13-year-old primary school children using the Wide Range Achievement Test, Fifth Edition (WRAT-5). Participants from grades 1–7, were randomly selected from 38 primary schools. Raw WRAT-5 scores were standardized to age- and sex-adjusted z-scores to enable meaningful comparisons across subgroups.

Alcohol use was assessed using the validated Uganda–Lumasaaba version of the CRAFFT screening tool, while socio-demographic information was obtained through a structured questionnaire previously employed in studies using the same dataset. Data analysis was conducted in STATA 18. Multiple linear regression models were applied to examine the association between alcohol use and academic outcomes, adjusting for demographic, familial, and school-related factors.

Results

Of the 460 primary school children assessed, 25.2% reported alcohol use in the past 12 months; 56.6% were girls, and 70.9% were aged 10–13 years. Children who reported alcohol use performed significantly worse than non-users in multiple academic domains, with lower scores in word reading (–0.98, 95% CI: − 1.69 to − 0.28), spelling (–1.23, 95% CI: − 2.44 to − 0.14), and sentence comprehension (–0.14, 95% CI: − 0.59 to − 0.02). No significant differences were observed between the two groups in math computation scores. Among alcohol users, male gender, rural residence, and informal caregiver employment were associated with lower academic achievement. Among non-users, harmful parental drinking was a risk factor, while higher parental education appeared to be protective.

Conclusion

Alcohol use among primary school children was associated with poorer academic achievement in word reading, spelling, and sentence comprehension, within the broader context of family and socioeconomic adversity. Targeted school- and community-based interventions are recommended to prevent early alcohol use and promote academic achievement, focusing on rural settings and vulnerable households. Longitudinal studies are needed to clarify the causal link between alcohol use and academic performance.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12889-025-25199-2.

Background

Underage alcohol consumption is a major global public health issue [1]. Data from the Global School-Based Health Survey, which covered 57 countries, indicated that approximately 25% of adolescents consumed alcohol in the past 30 days, with higher prevalence rates observed among boys and older adolescents aged 14–15 years [2]. Uganda’s population is predominantly young, with children aged 0–17 years constituting 50.5% of the total population [3].

In this context, alcohol consumption in Uganda remains highly prevalent. The country ranks among the top ten consumers of alcohol globally and among the highest in Africa, with an estimated per capita intake of 12.2 L, nearly double the continental average of 6.3 L [1]. The widespread use of alcohol is reinforced by strong cultural acceptance [4] and it is easily accessible to children [5]. In line with global trends, previous studies in Uganda found high prevalence rates, with one in four children aged 6 to 13 years consuming alcohol in the past year [6].

Early alcohol use among children poses significant public health risks, with evidence linking underage drinking to impaired cognitive development and academic performance, increased risk of substance use disorders, and long-term mental health challenges [7–9]. Studies indicate that early initiation of alcohol consumption disrupts brain development, as the adolescent brain remains highly vulnerable to neurotoxic effects, potentially leading to deficits in memory, decision-making, and impulse control [9–11]. Additionally, children who consume alcohol are more likely to engage in risky behaviors, including unsafe sexual practices, violence, and delinquency, increasing their susceptibility to adverse life outcomes [12].

Academic achievement during primary school years is a cornerstone of future educational and occupational success, shaping opportunities and outcomes across the lifespan [13]. Basic skills in mathematics, reading, comprehension, and spelling serve as foundational building blocks for cognitive and social development [14]. Studies have shown that early proficiency in basic academic skills especially mathematics and reading skills were a strong predictor of later academic success [15, 16]. Children who excel in reading, math, and writing in primary school are more likely to perform well in secondary education and beyond [17].

However, numerous challenges significantly impede children’s educational success, particularly poverty, a pervasive issue, restricts access to quality education and is associated with high child-to-teacher ratios, overcrowded classrooms, and inadequate provision of scholastic materials and food [18–20]. Also, children’s social conditions are challenged through lack of parental support, including restricted physical and emotional support [21]. These compounding factors create an environment where alcohol use may both stem from and exacerbate the barriers to academic success and benefits thereof [5, 22, 23].

Although research has explored the broader effects of alcohol use among adolescents, its association with academic achievement among primary school children remains underexamined, particularly in low-resource settings such as Uganda. In these contexts, where socio-economic and health-related challenges are prevalent, there is limited understanding of how alcohol consumption and other factors, such as family dynamics, nutrition, and socio-economic status, affect fundamental academic skills.

Addressing these gaps is crucial to inform the development of targeted interventions aimed at reducing the combined burden of alcohol use and educational disadvantage among vulnerable populations. This study aimed to examine the basic academic achievement skills: word reading, sentence comprehension, spelling, and math computation and to investigate their association with alcohol use and other influencing factors among primary school children in Mbale district, Eastern Uganda.

Methods

Study design and setting

This school- based cross-sectional study was conducted between September 2020 and March 2021 during the COVID-19 pandemic among primary school children aged 6 to 13 years from grades 1 to 7 attending urban and rural schools in Mbale District. Mbale district is located in the Elgon region of eastern Uganda. The methodology used for the validation of the Uganda-Lumasaaba version of the CRAFFT screening tool, as well as findings on the prevalence of alcohol use and associated risk factors in this population, have been described in detail in previous publications based on the same dataset [6, 24].

According to the Uganda Bureau of Statistics 2014 Census Report, primary school enrollment was approximately 8 million [25]. In Mbale district, primary school-age children made up 20.9% (n = 100,543) of the population. However, class-specific data showed low primary school completion rates, with 20% of primary going children in P2 and only 8% in P7. Furthermore, just 8% of P7 pupils were 12 years old, the rest were older, indicating delays in progression through the grades [3, 25]. Updated data from the 2024 census was not yet available.

The Ugandan government, through its universal primary education (UPE) policy, abolished tuition fees and parents and teachers association charges, leading to a significant increase in primary school enrollment [26]. Between 2009/2010 and 2015/2016, approximately 12% of the national budget was allocated to education. However, challenges persisted, with an average pupil-to-teacher ratio of 65:1 in Mbale district, far higher than Kampala’s more favorable ratio of 35:1 [3]. Although primary education is supposed to be free, there is still a challenge of parents expected to contribute to their children’s meals and scholastic materials [27].

Sample size and sampling procedures

Yamane’s formula [28], was used to calculate the initial required sample size, which yielded 400 pupils at a 5% level of precision. To account for potential missing data and to increase statistical power, we adjusted the sample size upward by 20%, resulting in a target of 480 participants. Data collection was delayed by COVID-19 school closures, during which some children aged out of eligibility and some girls dropped out due to pregnancy, resulting in a final sample of 470 participants [24].

To ensure representativeness, a stratified two-stage sampling procedure was employed. In the first stage, schools were selected; in the second, individual pupils were randomly sampled. The study population consisted of 176 schools across rural, peri-urban, and urban areas of Mbale District, enrolling 117,556 pupils as of January 2019. From this population, 42 schools comprising 36,522 pupils were selected using stratified random sampling [24], of which 38 schools ultimately participated. Four schools were excluded due to feasibility challenges related to distance and accessibility.

Stratification was based on location (rural, peri-urban, urban) and ownership (government or private), with the number of schools drawn proportionate to the size of eligible children in each stratum. In the second stage, participants were sampled from class lists using computer-generated random numbers. Table 1 presents the distribution of schools and pupils selected by stratum.

Table 1.

Distribution of schools and pupils selected by location and ownership

Location	Ownership	Schools Selected (n)	Eligible Children in Stratum (N)	Children Selected (n)
Rural	Government	20	16,311	199
Urban	Government	3	3556	46
Peri-urban	Government	14	15,381	176
Rural	Private	1	165	10
Urban	Private	3	853	27
Peri-urban	Private	1	256	12
Total		42	36,522	470

Study participants

Children were included in the study if they resided and were studying from the selected primary schools in Mbale district, were within the age range 6 to 13 years, their parents consented, and the children provided assent to study participation. Children who were sick or had left school for some reason, at the time of data collection, were excluded from the study.

Variables and measures

Alcohol use was measured using the CRAFFT screening tool, a widely validated instrument for identifying substance-related risks among adolescents [29]. The Uganda–Lumasaaba version of the CRAFFT, previously adapted and validated within the same dataset [6, 24], was employed in this study. For the present analysis, the tool was applied to assess alcohol use behaviors (any use in the past 12 months and frequency of use), rather than to diagnose alcohol use disorders. The English version of the adapted CRAFFT tool is provided as supplementary File 1.

Dependent variable

The dependent variables in this study were basic academic achievement skills across four domains: Word Reading, Sentence Comprehension, Spelling, and Math Computation, assessed using the Wide Range Achievement Test, Fifth Edition (WRAT-5) [30]. The WRAT series, first developed in 1941 and published in 1946, is widely used to assess foundational academic skills essential for learning and communication. The WRAT-5, a major revision of the WRAT-4, includes updated normative data, enhanced cultural sensitivity, and flexible administration and scoring options (both hand-scored and digital). It is standardized for individuals aged 5 to 85 + and demonstrates high internal consistency, with median coefficient alpha values ranging from 0.87 to 0.93 across its subtests. The WRAT-5 takes approximately 20–30 min to administer [31–33].

While the WRAT-4 has been used in Uganda in studies involving children affected by HIV and malaria [34–37], to our knowledge, this was among the first studies to utilize the WRAT-5 in Uganda for research purposes. The WRAT-5 blue form was used in this study, with grade-based norms applied to accommodate children whose chronological age did not align with their school grade. Assessments with the WRAT-5 [30] were conducted in English, one of Uganda’s official academic languages and the medium of instruction from Primary 1 onward.

The WRAT-5 includes four subtests designed to evaluate distinct domains of academic achievement. The word reading subtest evaluates word recognition and decoding through two components. First, the examinee reads aloud a list of 15 letters, earning one point for each correct response. Then, they proceed to read 55 words of increasing complexity. Notably, dialectal differences and non-native English pronunciations are not penalized.

The sentence comprehension subtest assesses reading comprehension through fill-in-the-blank exercises, where examinees must supply the missing words to demonstrate understanding. The test begins at a grade-appropriate level, with the flexibility to revert to an easier section if needed or to discontinue after a predetermined number of consecutive incorrect responses.

The spelling subtest measures letter recognition and spelling proficiency through two tasks. Initially, the examinee writes their name and 13 dictated letters. Subsequently, they attempt to spell as many words as possible from a list of 42 words, each presented within a contextual sentence to aid comprehension. Finally, the math computation subtest evaluates fundamental problem-solving skills without the use of a calculator. It includes an oral section with 15 math questions ranging from simple counting to basic arithmetic. This is followed by a written component, where the examinee has 15 min to solve up to 40 math problems.

Independent variables

Categorical variables included sex (boy/girl), school grade level (1–7), school geographical location (urban, peri-urban, rural), history of school absenteeism (assessed by asking children if they had missed school in the past month, how many days were missed, and the reasons for absence, using a structured questionnaire). Sociodemographic data were collected using a structured questionnaire developed for the broader project [6], from which this study was drawn. Others were: household food insecurity in the past month, primary caregiver type, family status, parental education level, and employment status. The continuous variables were age (in completed years) and nutritional status, assessed using BMI-for-age z-scores. Age, sex, school grade, and school location were recorded from child interviews, while primary caregiver characteristics, including education level and employment status, were obtained from parents interviews. These questionnaires were administered in the local language. The English versions of the child and parent questionnaires are included as supplementary Files 2 and 3 respectively.

Alcohol use

Child alcohol use was assessed through clinician-administered interviews utilizing the validated Uganda (Lumasaaba) version of the CRAFFT tool [24]. The specific question asked was: “During the PAST 12 MONTHS, on how many days did you consume more than just a few sips (beyond tasting) of beer, wine, locally brewed alcohol, or any other alcoholic beverage?“ [6]. Parental alcohol use was assessed using the Alcohol Use Disorders Identification Test (AUDIT), a validated screening instrument for identifying patterns of alcohol consumption and potential alcohol use disorders [38, 39].

Child nutrition

Anthropometric measurements, including weight and height, were obtained following standardized procedures. Weight was recorded to the nearest 0.1 kg using a calibrated digital scale, and height to the nearest 0.1 cm using a stadiometer. Nutritional status was determined by calculating Body Mass Index (BMI)-for-age z-scores in accordance with the World Health Organization (WHO) Child Growth Standards [40].

Data collection procedures

Prior to the COVID-19 lockdowns, informed consent was obtained from school heads. Following the reopening of schools, additional mobilization was carried out with the support of teachers to recruit participants from lower classes (P1–P6). Parents were contacted by telephone and invited to provide consent.

Data collection was conducted in participating schools, where children were identified and recruited through their respective institutions. Questionnaires were administered by trained interviewers using paper-based forms, which facilitated interaction with participants and allowed verification by colleagues. Adequate classroom space was provided to ensure privacy and a quiet environment for interviews and assessments.

The research team comprised a child psychiatrist, clinical officers, psychologists, teachers, and a statistician. Study activities included anthropometric measurements, academic achievement assessments using the WRAT-5, child and parent interviews, and alcohol use screening with the CRAFFT tool. All data were systematically coded by sub-county, school, and school type, with adherence to privacy and confidentiality throughout the study.

Statistical analysis

Data entry was performed using Epi Info™ 7 and exported to STATA version 18 for analysis. Descriptive statistics summarized categorical variables as frequencies and percentages, while continuous variables were presented as means with standard deviations (SD) when normally distributed, or medians with inter-quartile ranges (IQR) for skewed data.

The primary outcome, academic achievement skills, was assessed across four domains: word reading, math computation, sentence comprehension, and spelling, using the WRAT-5. Raw scores were converted to grade-based standard scores using the WRAT-5 manual, which were then cross-checked against grade-equivalent scores for accuracy. To enable valid comparisons across ages and grade levels, standard scores were transformed into z-scores within six-month age intervals. Z-scores were further classified into three performance levels: low (< −1 SD), medium (−1 to 1 SD), and high (> 1 SD), representing below-average, average, and above-average achievement.

Inferential analysis examined factors associated with academic outcomes. Crude and multivariable linear regression models were fitted to assess associations between academic z-scores and independent variables. Stratified analyses by alcohol use were conducted to explore potential effect modification. Missing data were addressed through imputation under the assumption of missing at random (MAR). Linear interpolation was applied for continuous variables and mode substitution for categorical variables. All subsequent analyses were conducted on the imputed dataset.

Model diagnostics ensured the validity of regression assumptions. Normality of residuals was assessed with Q–Q plots and Shapiro–Wilk tests, while scatterplots of residuals versus fitted values evaluated homoscedasticity (equal variance). Variance Inflation Factors (VIFs) were below 10, indicating no multicollinearity concerns. Model selection was guided by the Akaike Information Criterion (AIC) to ensure a good balance between accuracy and simplicity. Sensitivity analyses tested the robustness of findings, including exclusion of participants with extreme BMI-for-age z-scores or missing key variables. Statistical significance was set at p < 0.05 (two-tailed), and 95% confidence intervals were reported.

Ethical considerations

The study was conducted in accordance with the principles of the Declaration of Helsinki. The study protocol was approved by the Makerere University School of Medicine Higher Degrees Research Ethics Committee (SOMREC # REC REF 2018-095), the Uganda National Council for Science and Technology (UNCST # SS5103), and the Regional Committee for Medical Research Ethics (REC West #107632 and REC South East #50146) in Norway. Permissions were also obtained from the Mbale district administration and the head teachers of participating schools.

Informed consent was obtained from parents, and age-appropriate assent was obtained from child participants. Children identified as having alcohol use problems or mental health concerns were provided with counseling, reassurance, and psychoeducation by the principal investigator, a child and adolescent psychiatrist. Additionally, parents of these children were referred for further support at Mbale Regional Referral Hospital.

Results

Table 2 presents the demographic, educational, and health-related characteristics of the participating children and their primary caregivers. Some of these characteristics have been reported in earlier publications using the same dataset [6, 24]. The study analyzed data from 460 children, the majority of whom were aged 10–13 years (70.9%, n = 320). Girls constituted a slightly larger proportion of the sample than boys (56.6%, n = 260). The median household size was five children (IQR: 4–7).

Table 2.

Child, caregiver, household and school characteristics by child alcohol consumption last 12 months (N = 460)

Factor	Level	Frequency (%) N = 460	Child Alcohol Consumption
			No % (n = 344)	Yes % (n = 116)
Child
Sex a	Male	199 (43.5)	150 (43.6)	50 (43.2)
	Female	260 (56.5)	194 (56.4)	66 (56.9)
Age a	6 to 9	131 (29.1)	109 (31.7)	26 (22.4)
	10 to 13	320 (70.9)	235 (68.3)	90 (77.6)
Class/grade	P1-p4	276 (60.0)	206 (59.9)	70 (60.3)
	P5-p7	184 (40.0)	138 (40.1)	46 (39.7)
Ever missed school a	Yes	195 (42.5)	143 (41.8)	64 (55.2)
	No	263 (57.4)	199 (58.2)	52 (44.8)
Missing meals	Yes	306 (66.6)	224 (65.2)	86 (74.1)
BMI-age-z score	Underweight (<−2)	290 (63.0)	217(63.1)	73 (62.9)
	Normal (−2 ≤ z ≤ 1)	98 (21.3)	69 (20.1)	29 (25.0)
	Overweight (> 1)	72 (15.7)	58 (16.9)	14 (12.1)
Parent/family
Primary care giver a	Mother	336 (73.8)	256 (74.8)	80 (70.8)
	Father	47 (10.3)	38 (11.1)	9 (8.0)
	Other	72 (15.8)	48 (14.0)	24 (21.2)
Caregiver’ s education a	Primary	234 (51.3)	164 (47.8)	70 (61.9)
	Secondary	128 (28.1)	107 (31.2)	21 (18.6)
	Higher institution	94 (20.5)	72 (21)	22 (19.5)
Caregiver’s employment a	Formal employment	100 (21.9)	87 (25.4)	13 (11.2)
	Informal	343 (75.3)	244 (71.1)	99 (85.3)
	Unemployed	12 (2.6)	12 (3.5)	4 (3.4)
Care giver alcohol use (AUDIT) a	Harmful user	22 (4.9)	15 (7.8)	7 (2.8)
	Unharmful user	424 (95.1)	178 (92.2)	246 (97.2)
School location	Urban	208 (45.2)	164 (47.7)	34 (29.0)
	Peri urban	48 (10.4)	38 (11.1)	8 (6.5)
	Rural	204 (44.3)	142 (41.2)	75 (64.5)

^aMissing data < 2%

With regard to school location, 45.2% (n = 208) attended urban schools, 44.3% (n = 204) rural schools, and 10.4% (n = 48) peri-urban schools. School absenteeism at least once per month was reported by 42.5% (n = 195) of the children. The majority of primary caregivers were mothers (73.8%, n = 336), followed by fathers (10.3%, n = 47) and other relatives (15.8%, n = 72). Nearly half of the caregivers (51.3%, n = 234) had attained only primary-level education, and most (75.3%, n = 343) were engaged in informal employment.

Most children (68.6%, n = 313) lived with both parents, while 18.7% (n = 86) had separated parents, and 7.4% (n = 34) were orphaned (i.e., had lost at least one parent). Alcohol use in the past 12 months, assessed using the CRAFFT screening tool, was reported by 116 children (25.2%; 95% CI: 21.4–29.4).

Academic achievement by alcohol use status

Table 3 summarizes academic performance across the four WRAT-5 domains: Word Reading, Sentence Comprehension, Spelling, and Math Computation, categorized as low, medium, and high, and analyzed as continuous outcomes. Results are stratified by alcohol use status and include both unadjusted and adjusted linear regression estimates.

Table 3.

Categories of academic achievement scores and liner regression estimates for the WRAT-5 domains by alcohol use status (N = 460)

Domain	Performance Level	Non-user n (%)	User n (%)	Crude β (95% CI)a	Adjusted β (95% CI)b
Word Reading				−1.04 (−1.73; −0.35)	−0.98 (−1.69; −0.28)
	Low	–	–
	Medium	289 (83.8)	92 (80.0)
	High	56 (16.2)	23 (20.0)
Sentence comprehension				−0.10 (−0.64; −0.03)	−0.14 (−0.59; −0.02)
	Low	69 (20.0)	40 (34.8)
	Medium	200 (58.0)	58 (50.4)
	High	76 (22.0)	17 (14.8)
Spelling				−0.10 (−0.83; 0.62)	−1.23 (−2.44; −0.14)
	Low	0 (0.0)	1 (0.9)
	Medium	283 (82.0)	90 (78.3)
	High	62 (18.0)	24 (20.9)
Math Computation				0.29 (−1.62; 2.20)	0.40 (−1.53; 2.34)
	Low	63 (18.3)	42 (36.5)
	Medium	214 (62.0)	66 (57.4)
	High	68 (19.7)	7 (6.1)

^aThe linear regression coefficient applies to the uncategorized z-scores

^bThe linear regression was adjusted for sex, age and grade level (class)

Word reading

The distribution of performance levels was relatively similar between alcohol users and non-users, with 20.0% of users and 16.2% of non-users scoring in the high-performance category. However, adjusted linear regression analysis indicated a statistically significant negative association between alcohol use and word reading scores (β = − 0.98, 95% CI: − 1.69 to − 0.28, p = 0.006), suggesting that alcohol use may be associated with reduced reading ability after adjusting for potential confounders.

Sentence comprehension

A greater proportion of alcohol users scored in the low-performance category compared to non-users (34.8% vs. 20.0%), and fewer achieved high scores (14.8% vs. 22.0%). The adjusted regression model showed a modest but statistically significant negative association between alcohol use and sentence comprehension (β = − 0.14, 95% CI: − 0.59 to − 0.02, p = 0.022), indicating that alcohol use could be linked to lower comprehension skills.

Spelling

While categorical performance distributions were broadly comparable between users and non-users, adjusted analysis revealed a significant negative association between alcohol use and spelling scores (β = − 1.23, 95% CI: − 2.44 to − 0.14, p = 0.031). This suggests that alcohol use may be associated with lower spelling performance.

Math computation

A larger proportion of alcohol users were classified in the low-performance category (36.5% vs. 18.3% of non-users), with fewer scoring in the high-performance category (6.1% vs. 19.7%). Despite this trend, adjusted regression analysis did not identify a statistically significant association between alcohol use and math computation scores (β = 0.40, 95% CI: − 1.53 to 2.34), indicating no conclusive evidence of an effect in this domain.

Multivariate predictors of basic academic achievement

Table 4 summarizes adjusted regression estimates for multiple socio-demographic and contextual factors associated with academic achievement across WRAT-5 domains. Harmful parental alcohol use significantly predicted lower performance in Word Reading, Spelling, and Math Computation.

Table 4.

Key predictors on the four WRAT-5 academic achievement domains, results from multivariable linear regression models

Factor	Level	Word Reading	Sentence Comprehension	Spelling	Math Computation
Age of assessment	Per year increase	−0.06 (−0.26, 0.13)	−0.03 (−0.19, 0.13)	−0.06 (−0.23; 0.18)	0.007 (−0.54, 0.55)
Gender	Male (r)
	Female	−0.15 (−0.63, 0.33)	−0.07 (−0.47, 0.32)	−0.06 (−0.56; 0.44)	1.52 (0.20, 2.85)*
School location	Urban (r)
	Peri-urban	−0.08 (−0.91, 0.73)	−0.08 (−0.79, 0.91)	-	−3.46 (−5.30, −0.32)*
	Rural	−0.05 (−0.56, 0.46)	-	−0.04 (−0.46; 0.32)
Parent education	Primary (r)
	Secondary	0.24 (−0.33, 0.81)	−0.08 (−0.47, 0.82)	0.14 (0.09; 1.09)*	-
	Higher	0.38 (−0.25, 1.02)	-	-	1.16 (−0.57, 2.90)
Social economic status	Low (r)
	High	0.28 (0.09, 0.77)*	0.17 (−0.22, 0.57)	0.14 (0.03; 2.01)*	−0.35 (−1.68, 0.97)
Parent alcohol consumption	Harmless (r)
	Harmful	−0.06 (−1.07, −0.03)*	−0.08 (−0.97, 0.79)	−1.04 (−2.16; −0.07)*	−3.36 (−6.30, −0.42)*
Child alcohol consumption	No (r)
	Yes	−0.98 (−1.69, −0.28)*	−0.14 (−0.59; − 0.02)*	−1.23 (−2.44; −0.14)*	0.40 (−1.53, 2.34)
Child BMI	Per unit increase	−0.002 (−0.02, 0.02)	−0.002 (−0.02, 0.01)	-	−0.002 (−0.07, 0.06)
Missing meals	No (r)
	Yes	−0.42 (−0.90;0.06)	−0.24(−0.65;0.15)	−0.46 (−1.00; 0.07)	−0.92 (−2.35;0.49)
Missing school	No (r)
	Yes	−0.36 (−0.85;0.11)	−0.22 (−0.60;0.15)	−0.43 (−0.94;0.07)	−0.58 (−1.92;0.74)
Employment status of primary care giver	Formal employment (r)
	Informal employment	0.15 (−0.76;0.44)	0.62 (−1.08;−0.16)*	0.54 (−1.16;0.08)	1.78 (−3.43;−0.14)*
	Unemployed	0.01 (−1.56;1.58)	−0.40 (−1.60;0.79)	−0.26 (−1.89;1.36)	−0.68 (−4.95;3.59)
Primary caregiver	Mother (Ref)
	Father	0.29 (−0.50;1.10)	0.06 (−0.54;0.68)	0.20 (−0.62;1.04)	−2.20 (−4.38;−0.01)*
	Other	0.12 (−0.54;0.79)	0.45 (−0.06;0.96)	0.51 (−0.17;1.20)	−0.39 (−2.21;1.41)

Multivariable linear regression models conducted for each domain and adjusted β, 95% CI presented. Reference values given as (r)

Higher socioeconomic status and greater parental education were positively associated with academic achievement, with the strongest associations observed for Spelling. Female gender was linked to higher performance in Math Computation (β = 1.52, 95% CI: 0.20 to 2.85), while school location in peri-urban areas and having the father as the primary caregiver were associated with lower scores in this domain. Other factors such as age, BMI, missing meals or school, and caregiver employment showed mostly non-significant associations. These findings suggest that alcohol use, and various socio-environmental and family factors are associated with academic outcomes. However, these relationships are complex and multifactorial, with higher socioeconomic status and greater parental education appearing to serve as protective factors.

Stratified multivariable regression analysis of factors associated with academic achievement domains by child alcohol use status

Table 5 presents stratified multivariable regression models examining associations between socio-demographic, contextual, and health-related factors and academic achievement across four WRAT-5 domains (Word Reading, Sentence Comprehension, Spelling, and Math Computation), stratified by child alcohol use status (users vs. non-users). This analysis aimed to explore whether individual, familial, and school-level factors moderated the association between child alcohol use and academic achievement skills.

Table 5.

Key predictors on the four WRAT-5 academic achievement domains by alcohol use

Factor	Level	Word Reading		Sentence Comprehension		Spelling		Math Computation (Adjusted β, 95% CI)
		Non-user	User	Non-user	User	Non-user	User	Non-user	User
Gender	Male (r)
	Female	0.01 (−0.26;0.27)	−0.57 (−2,37;1.22)	−0.09 (−0.21;0.02)	0.63 (−0.88;2.15)	−0.10 (−0.23;0.02)	0.01 (−2.06;2.08)	−0.29 (−0.59;0.01)	7.10 (1.76;12.43)*
Age of assessment	Per year increase	0.03 (−0.07;0.14)	−0.33 (−1.16;0.48)	−0.02 (−0.07;0.02)	0.17 (−0.52; 0.87)	−0.01 (−0.06;0.04)	−0.10 (−1.06;0.84)	−0.01 (−0.13;0.11)	−0.09 (−2.54;2.35)
Child BMI	Per unit increase	0.00 (−0.01;0.01)	0.04 (−0.32;0.40)	0.00 (−0.01;0.01)	0.09 (−0.21;0.40)	0.00 (−0.01; 0.01)	0.24 (−0.17; 0.66)	0.00 (−0.01;0.01)	−0.33 (−1.41;0.74)
Missing meals	No (r)
	Yes	−0.01 (−0.29;0.28)	−1.17 (−3.07;0,72)	−0.07 (−0.19;0.05)	−0.82 (−2.63;0.97)	−0.09 (−0.23; 0.04)	−1.57(−4.03; 0.88)	−0.25 (−0.57;0.06)	−2.71 (−9.08;3.65)
Missing school	No (r)
	Yes	0.00 (−0.27; 0.27)	−0.97 (−2.79;0.89)	0.08 (−0.02; 0.20)	−1.22 (−2.76;0.32)	0.10 (−0.03; 0.23)	−1.66 (−3.78;0.44)	0.28 (−0.01;0.58)	−3.11 (−8.56;2.34)
Primary caregiver	Mother (r)
	Father	0.59 (0.16;1.02)*	−1.08 (−4.53;2.35)	0.04 (−0.14;0.22)	0.45 (−2.38; 3.30)	0.15 (−0.05;0.36)	0.32 (−3.61;4.26)	−0.24 (−0.72;0.23)	−8.98 (−19.0;1.03)
	Other	0.14 (−0.24;0.53)	0.32 (−1.91;2.56)	0.04 (−0.12;0.21)	1.30 (−0.55;3.15)	0.09 (−0.09;0.28)	1.40 (−1.15;3.96)	0.10 (−0.32;0.54)	−1.76 (−8.28;4.74)
Parent alcohol consumption	Harmless (r)
	Harmful	1.46 (0.80;2.12)*	−1.85 (−5.35;1.64)	−0.06 (−0.33;0.20)	−0.72 (−3.76;2.31)	0.20 (−0.11;0.51)	−3.07 (−7.10;0.95)	−1.13(−1.77;−0.48)*	−5.77 (−16.05;4.50)
Parent education	Primary (r)
	Secondary	0.14 (−0.17;0.45)	0.38 (−2.06;2.83)	1.09 (−0.02;0.24)	0.46 (−1.58;2.52)	0.09 (−0.05;0.24)	0.80 (−1.95;3.56)	0.06 (−0.28;0.41)	1.82 (−5.39;9.03)
	Higher	0.09 (−0.25;0.45)	1.57 (−0.86;4.01)	0.10 (−0.05;0.25)	0.40 (−1.64;2.44)	0.12 (−0.04;0.29)	3.17 (0.43;5.92)*	0.23 (−0.16;0.63)	4.91 (−2.26;12.09)
Social economic status	Low (r)
	High	0.09 (−0.17;0.36)	0.99 (−0.88;2.87)	0.02 (−0.09;0.14)	−0.51 (−2.10;1.08)	0.01 (−0.12; 0.13)	0.38 (−1.80;2.56)	0.01 (−0.29;0.30)	−1.94 (−7.54;3.64)
Employment status of primary caregiver	Formal employment (r)
	Informal employment	0.11 (−0.20;0.44)	−2.02 (−4.86;0.81)	−0.05 (−0.19; 0.08)	−3.88 (−6.14;−1.61)*	−0.01 (−0.16;0.14)	−3.86(−7.06;−0.66)	−0.11 (−0.48;0.24)	−11.7(−19.83;−3.56)*
	Unemployed	0.06 (−0.70;0.83)		0.02 (−0.31;0.35)		0.05 (−0.31;0.42)		0.10 (−0.74;0.96)
School location	Urban (r)
	Peri-urban	0.06 (−0.39;0.53)	−1.46 (−4.76;1.83)	−0.002(−0.20;0.19)	−1.31 (−4.08;1.45)	0.03 (−0.18;0.26)	−1.46 (−5.22;2.28)	0.03 (−0.47;0.55)	−1.98 (−11.83;7.85)
	Rural	0.10 (−0.18;0.39)	−1.20 (−3.31;0.90)	−0.11 (−0.24;0.01)	−1.54 (−3.31;0.21)	−0.05 (−0.19;0.08)	−2.60 (−4.96;−0.17)*	−0.25 (−0.57;0.06)	−0.48 (−6.77;5.79)

Results from multivariable linear regression models conducted for each domain stratified by alcohol use according to CRAFFT. Adjusted β, 95% CI presented. Reference values given as (r)

Among users, academic performance was lower in the presence of additional risk factors such as caregiver unemployment, lower parental education, rural school attendance, and harmful parental alcohol use. Female gender was associated with significantly higher math computation scores among users (β = 7.10; 95% CI: 1.76 to 12.43), whereas no significant gender differences were observed across other domains. In the non-users group, gender differences were minimal and statistically non-significant across all academic domains.

Informal caregiver employment was significantly associated with lower sentence comprehension and spelling scores among alcohol users, (β = −3.88; 95% CI: −6.14 to −1.61) and (β = −3.86; 95% CI: −7.06 to −0.66) respectively. This association was not observed among non-users, suggesting that informal employment may increase academic disadvantage among children who consume alcohol. Harmful parental alcohol use was independently associated with lower scores in math computation among children who did not report drinking alcohol (β = −1.13; 95% CI: −1.77 but had no significant effect in those that reported drinking alcohol. Among non-users, having a father as the primary caregiver was associated with higher word reading scores (β = 0.59; 95% CI: 0.16 to 1.02), suggesting a potential protective effect.

Among users, having a parent with higher education was associated with better spelling scores (β = 3.17; 95% CI: 0.43 to 5.92), suggesting a possible buffering effect on literacy-related outcomes. Among users, attending a rural school was significantly associated with lower Spelling scores (β = −2.60; 95% CI: −4.96 to −0.17), an association not evident among non-users.

Child mental health problems, missed meals, missed school days, and BMI were not significantly associated with academic outcomes in either group. However, the direction of associations tended to be negative among users, suggesting that these factors may still contribute to cumulative disadvantage.

Discussion

This study investigated the relationship between alcohol use and academic achievement, as measured by the WRAT-5 domains of word reading, sentence comprehension, spelling, and math computation, among primary school children in Uganda. We found that alcohol use was significantly associated with poorer performance in word reading, sentence comprehension, and spelling. Further, the findings suggest that academic achievement among children, who reported drinking alcohol, was shaped by a combination of interacting factors across individual, family, and community levels, consistent with the socioecological framework [41, 42].

Individual Level factors.

The study found that children who reported alcohol consumption had significantly lower scores in word reading, sentence comprehension, and spelling compared to their peers who did not report alcohol use. In contrast, no significant differences were observed in math computation scores. This pattern suggests a specific vulnerability of language-related cognitive functions to the neurodevelopmental effects of alcohol during middle childhood, particularly those involving attention, working memory, and verbal processing [43, 44].

Unlike math computation, which relies more on procedural and numerical reasoning, Word reading, sentence comprehension and spelling skills are language-dependent and therefore more susceptible to disruption from early alcohol exposure [45]. These findings are consistent with prior research indicating that alcohol use during childhood and adolescence disproportionately impairs verbal and reading-related abilities [46]. While studies in high-income contexts have found that higher academic ability may be associated with increased alcohol use during adolescence [47], our findings contrast with this pattern. This discrepancy may reflect contextual and developmental differences, where early alcohol use in low-resource settings is more likely driven by adversity than experimentation.

The finding that female alcohol users performed significantly better in math computation than their male counterparts is noteworthy and may suggest gender differences in coping strategies, resilience, and access to social or academic support between boys and girls. Previous research has indicated that adolescent girls often engage in more adaptive coping strategies, such as help-seeking and emotional regulation, which may buffer the negative effects of alcohol use on school performance [48]. In contrast, boys are more likely to externalize stress, which can manifest in disruptive behaviors and lower academic engagement.

Family level factors.

The findings underscore the significant association of family-level determinants and academic achievement, particularly among children who reported alcohol use. Within the family context, harmful parental alcohol use emerged as a significant risk factor for lower academic achievement among children, particularly for non-users, suggesting that the home environment exerts an independent influence on learning outcomes regardless of the child’s own alcohol use [49], which is consistent with previous research [50].

The association between informal caregiver employment and lower academic achievement among alcohol-using children in this study may reflect broader socio-environmental vulnerabilities linked to economic precarity. While some evidence from other low-resource contexts, suggests that informal employment may allow parents more time with their children [51], in our study setting, informal employment may increase children’s exposure to alcohol-related environments and reduce structured academic support at home, thereby exacerbating word reading, sentence comprehension and spelling challenges. This points to the role of economic instability and parental work conditions in shaping educational outcomes [52], especially for children already vulnerable due to alcohol use [6].

However, caregiving arrangements showed mixed effects: paternal caregiving was associated with better reading scores among non-users but had no clear benefit among users. Studies have shown that fathers engaged in school-related activities can support children’s reading and learning [53], which may help explain the higher word reading scores observed among children with fathers as primary caregivers in our study. The findings may also reflect the growing significance of fathers’ roles in child development, with a shift in caregiving dynamics within Ugandan households. In addition, higher parental education was linked to better spelling performance among users, reflecting the protective effects of educational capital [54].

Community Level factors.

Children who reported drinking alcohol and attended rural schools showed lower academic performance, particularly in spelling. This may suggest that limited resources and contextual challenges in rural areas may exacerbate the negative effects of alcohol use on learning, a finding consistent with broader research on educational disadvantage in rural settings. These contextual differences appear to influence academic achievement both directly and indirectly, particularly through their impact on parenting practices [55]. In addition, rural residence is often associated with a higher risk of hazardous alcohol use [56], suggesting that community-level inequalities may increase the impact of alcohol use on academic outcomes in primary school children.

Notably, school absenteeism, missed meals, and BMI were not significantly associated with academic performance in this sample, the direction of their effects was consistently negative among users suggesting that results could have been affected by small numbers. It may also suggest that cumulative disadvantage may operate through multiple, reinforcing pathways, an idea central to the socioecological framework.

Strengths of the study

First, the use of the WRAT-5, a well-validated and standardized measure of academic achievement, strengthens the reliability of the outcome assessments across word reading, sentence comprehension, spelling, and math computation. Second, the inclusion of a relatively large and diverse sample of primary school children enhances the representativeness and generalizability of the findings within the Ugandan context. Furthermore, the stratified analyses based on child alcohol use status allowed for nuanced insights into differential associations between socioecological factors and academic outcomes.

Limitations of the study

This study has several limitations. First the cross-sectional design limits causal inferences regarding the relationship between alcohol use and academic achievement skills. To minimize this limitation, we defined study variables, used validated tools, and adjusted for key sociodemographic and family-level confounders in the analyses to strengthen the validity of observed associations. Second, self-reported alcohol use by children may have been affected by recall or social desirability bias, especially given the stigmatized nature of substance use in this age group. To mitigate this, questionnaires were administered anonymously, and trained research assistants emphasized confidentiality to encourage honest reporting. Third, the study lacked neurodevelopmental or cognitive measures that could have provided deeper insights into the mechanisms associating alcohol use and academic performance. However, academic achievement was measured using the WRAT‑5, a standardized, norm-referenced tool, providing an objective outcome measure. Fourth, the exclusion of school-level variables such as teacher quality, classroom resources, or school climate may have limited the analysis of community-level educational disparities. We attempted to minimize this by using stratified sampling across both urban and rural schools to enhance representativeness. Although the CRAFFT tool captures the frequency of alcohol use, this study categorized participants as users or non-users, which limits the ability to examine how varying quantities of alcohol consumption may influence academic performance. Future research should include measures of both frequency and quantity to explore potential dose-dependent effects on academic achievement. Finally, while the use of stratified random sampling enhanced representativeness within the context of Uganda, their generalizability to other regions may be limited.

Conclusions

This study found that children who reported drinking alcohol in the past twelve-months period performed significantly worse in literacy domains (word reading, sentence comprehension, and spelling), compared to their peers who did not drink alcohol, with no significant differences observed in math computation. Contextual vulnerabilities such as informal caregiver employment, harmful parental alcohol use, and rural school location were linked to poorer performance, particularly among alcohol-using children. These findings suggest that early alcohol use not only poses direct neurodevelopmental risks but also interacts with structural disadvantage to compound educational vulnerabilities. In addition, in households without child alcohol use, paternal caregiving was linked to better word reading outcomes, suggesting that active and supportive caregiving, regardless of gender, can foster academic development.

Implications for policy and practice

The findings have important implications for education policy, child protection, and public health practice in Uganda and similar contexts. The observed associations between child alcohol use and poorer performance in literacy domains highlight the need for early identification and prevention efforts targeting substance use among primary school children. Education stakeholders and child welfare advocates should consider integrating school-based screening and referral systems for alcohol use and mental health concerns into existing educational and health frameworks.

At the family level, the negative influence of harmful parental drinking and informal employment on children’s academic outcomes underscores the importance of family-centered interventions, including parental support programs, economic strengthening initiatives, and caregiver education. In addition, the performance disparities observed in rural schools call for enhanced investment in rural education infrastructure and teacher training to reduce contextual inequities that may compound the risks associated with child alcohol use.

Recommendations for future research

Future studies should consider longitudinal designs to better understand the temporal pathways and causal mechanisms linking early alcohol use and academic trajectories. There is also a need for intervention research to evaluate the effectiveness of school- and family-based programs aimed at reducing alcohol use and supporting academic achievement.

Moreover, future research should explore the differential effects of caregiving roles, particularly the influence of paternal versus maternal involvement, in relation to children’s cognitive and academic development. Finally, more detailed examination of school-level variables, including classroom environment, teacher engagement, and availability of psychosocial support, would offer a fuller understanding of how community and institutional factors intersect with individual behaviors to shape academic outcomes.

Abbreviations

HIV

Human Immunodeficiency Virus

MINI KID

Mini International Neuropsychiatric Interview for children and adolescents

WHO

World Health Organization

WRAT 5

Wide Range Achievement Test Fifth edition

Authors’ contributions

JSN, PB, IMSE, JKT, NN, GN conceptualization, JSN data collection, JSN, WS, PB, IMSE, NN, GN data analysis, and interpretation. All contributed to manuscript writing. All authors read and approved the final manuscript.

Funding

This study was supported by:

Grant Number D43TW010132 supported by:

Office Of The Director National Institutes Of Health (D), National Institute Of Dental & Craniofacial Research (NIDCR), National Institute Of Neurological Disorders And Stroke (NINDS), National Heart, Lung, And Blood Institute (NHLBI), Fogarty International Center (FIC), National Institute On Minority Health And Health Disparities (NIMHD). NURTURE partner institutions: Makerere University, JHU, CWRU. The first author was funded by the Research council of Norway, RCN 285489 for periods of the research period. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the supporting offices.

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was conducted in accordance with the principles of the Declaration of Helsinki. It received ethical approval from the Makerere University School of Medicine Higher Degrees Research Ethics Committee (SOMREC #REC REF 2018-095), the Uganda National Council for Science and Technology (UNCST #SS5103), and the Regional Committees for Medical Research Ethics in Norway (REC West #107632 and REC South East #50146). Permission was also obtained from relevant authorities in Mbale district administration and the head teachers of participating schools. Informed consent was obtained from parents, and assent was provided by child participants in an age-appropriate manner. Children identified with alcohol use problems or mental health conditions were offered counseling, reassurance, and psychoeducation, while their parents were referred for further support at Mbale Regional Referral Hospital.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.