Poverty and intellectual development in childhood

Abstract

Introduction
Childhood exposure to and duration of poverty can affect several individual characteristics related to intellectual development.

Objectives
This paper examines the implications of movement in and out of childhood poverty using a unique linkable database from the Canadian province of Manitoba. Differences in measurement of poverty and intellectual development are explored.

Methods
Almost 90,000 children were followed using two definitions of poverty -- neighborhood and household poverty. The large database permitted exploring the role of another variable -- maternal mental health.

Results
The association of household poverty with poorer intellectual outcomes has been shown to be stronger than the association of neighborhood poverty with such outcomes. This was true using various outcome measures appropriate across childhood (from age 5 to age 17). Comparisons with the role of maternal mental health were made and further analyses suggested.

Conclusion
The richness of the data has facilitated the study of childhood intellectual development. Household poverty appears to play an important role; neighborhood poverty and maternal mental health also seem to influence such development, but less strongly.

Introduction

Children born into poverty face many challenges that can negatively impact their long-term social, educational, and health outcomes [1-4]. A lack of school readiness and poor academic achievement have been associated with poverty which may be influenced, in part, by factors such as atypical structural brain development, limited language development, increased out-of-home care, and a greater likelihood of experiencing food insecurity [5-11]. Furthermore, previous studies have shown that higher household family income and neighborhood socioeconomic status lead to greater school readiness in numeracy knowledge, communication, vocabulary, and attention [12, 13]. Childhood exposure to poverty has also been associated with several health outcomes including externalising mental conditions, attention-deficit/hyperactivity disorder, asthma, and injuries resulting in hospitalisations [14-20].

Although many negative outcomes have been linked to childhood poverty, less is known about the effects of exposure to and the duration of poverty. Children may be exposed to different forms of economic deprivation, and some may transition in and out of poverty throughout their developmental years. Previous studies have suggested that both household and neighborhood poverty can negatively affect childhood outcomes and that an increased duration of poverty leads to more risk factors for poor outcomes [17, 21-23]. Additionally, moving out of childhood poverty leads to better long-term outcomes; however, the specific implications of such movement are unclear [21, 24].

Measuring the timing and duration of economic deprivation facilitates the study of several individual characteristics—intellectual development, mental health, sociability, openness to new ideas—and so on [22]. The effects of the timing and duration of deprivation can be researched using a unique linkable database following almost 90,000 children born in the Canadian province of Manitoba. Poverty remains a major issue in Manitoba, which has the highest child poverty rates among all Canadian provinces. According to the Census Family Low Income After-Tax Measure, which adjusts for family size and composition, 20.68% of children in Manitoba live in poverty – 7.21 percentage points higher than the national average. Manitoba also has the highest market income child poverty rate in Canada with two of the 14 Manitoba federal electoral ridings among the country’s poorest 5 ridings as to rate of poverty [25].

While poverty has been linked to developmental disparities, less is known about how intellectual differences between poor and affluent children can be measured and how these differences change throughout childhood. This paper addresses this knowledge gap by studying questions such as: How can childhood differences in intellectual development between the poor and the affluent be measured? How important are differences in measurement of poverty in assessing intellectual development? How do the differences change over childhood?

The Manitoba data allow comparison of the growth among children of different economic levels; panel data allow assessing intellectual development in children at six ages. Two definitions of poverty are used. Family or household poverty was evaluated using variables relating to receipt of Employment and Income Assistance (EIA – analogous to welfare); neighborhood poverty was defined as living in a neighborhood with median income in the lowest income quintile [26].

Methods

Setting and data

Manitoba has previously ranked in the mid-range of a series of Canadian indicators of health status and health care expenditures [27, 28]. In 2021, the provincial population was over 1.3 million with more than half (n = 749,607) living in Winnipeg, Canada’s eighth largest metropolitan area [29]. Manitoba has a comparatively large Indigenous population (18%) [30]. Canadian scores on the Programme for International Student Assessment (PISA), a standardised educational test administered on 15 years olds, have been above the international averages since the tests’ inauguration in 2000, but scores in all three subjects (reading, mathematics, and science) declined over the 2000–2018 period [28, 31]. Analyses of the 2018 PISA scores across 79 participating countries/provinces showed students in four provinces (Prince Edward Island, New Brunswick, Manitoba, and Saskatchewan) below the Canadian average in reading [32]. Approximately 30% of Manitoba’s five-year olds have been shown to be vulnerable in at least one developmental area; this exceeded the Canadian average of 26% [33]. Overall, Canada was outperformed by only three countries/regions employing the PISA test – China, Singapore, and Macao (China) [32].

We used administrative data in the Population Research Data Repository housed at the Manitoba Centre for Health Policy. A scrambled personal health number allowed linking across multiple de-identified datasets; information on linkage methods, confidentiality and privacy, and validity is reviewed elsewhere [34, 35]. Population registry data have been combined with individual-level information from: hospital discharge abstracts (containing International Classification Of Diseases [ICD] diagnosis codes; International Classification Of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes before April 1, 2004 and International Classification Of Diseases, 10^th Revision, Canada [ICD-10-CA] codes after April 1, 2004), physician visits (ICD-9-CM codes), Families First screens (filled out during routine home visits by Public Health Nurses for most births and include information on various aspects of the mother’s social circumstances), and monthly receipt of Employment and Income Assistance data - EIA (basically welfare). Canadian census data provided neighborhood-level median income.

Sample (from ages 5 through 17)

The number of children studied at each point in time varied across measures. Concentrating on those born in 2000-2002 generated a total of 7,424 children having scores on all six measures of intellectual development. Such person-specific information controlled for several known/unknown family and individual factors and extended from ages 5 through 17. Comparisons building on the very large number of cases available for less than all six time periods changed the results only minimally. Children who are home schooled or frequently absent from school would seem less likely to have been included in the sample.

Exposure to poverty

Household poverty

The analysis of social assistance payments provides the measure of exposure to household poverty. The database was searched for monthly involvement by both the mother and child. For counting years experiencing household poverty (years on EIA), we looked for monthly involvement (i.e., receipt of EIA payment) by the mother from one year before the child’s birth up to the date the child reaches a certain age, or until a particular childhood event, as well as involvement by the child from date of birth up to the same target age or event. Examples of such events would be the estimated date that the child had an assessment or test, such as the Early Development Instrument (EDI) in kindergarten or a grade-specific evaluation, or the date the child reaches a specific age (usually associated with the calendar year the child is expected to reach a specific grade) [36, 37]. Years were typically indexed at the child’s date of birth, and these years were counted for every year of age the child reaches to the event date. The year was counted as experiencing household poverty (being on EIA) if an EIA payment was received in any month during that year, i.e., the involvement date fell between the two anniversary dates of the child’s birthdate.

Neighborhood poverty

Using census data, neighbourhoods have been ranked from 1 (lowest median income) to 5 (highest median income); rankings were created separately for rural and urban neighbourhoods. Quintiles are based on census dissemination areas including ~400 individuals. Urban was specified as any neighborhood in Winnipeg or Brandon; rural was any other neighbourhood. Change of neighbourhood is defined in a manner parallel to the definition of change of household.

Minority and majority of years

The determination of a majority of years was based on a simple majority, i.e., a ratio of 0.5 or more. The numerator in this ratio was the count of years in poverty as described above. The denominator was the total count of years in the time period, starting from a year before the child’s birth up to the event date. Using the receipt of EIA as the indicator for household poverty, the event date was the child’s 18^th birth date; the total denominator was 19, to count the gestational year in which the mother may have received EIA. If any EIA payment was made to the mother or child during half or more of these years, the child was placed in the group having received EIA in the majority of years to the event or assessment date. If EIA payments were received in less than half of these years, but not zero years, the child was placed in the group having received EIA in the minority of years to the target date. Lastly, if no EIA payment was received for the entire time period, the child was placed in a third group labeled “None” or “No EIA”. A similar approach was used for counting the minority/majority of years in neighborhood poverty.

Measures of intellectual development

The measures were grouped together in their similarity:

The measures of intellectual development described below come from largely independent assessments used for government reports and comparative studies. The mean and standard deviation of the logit scores for each measure were first calculated. Each individual score was scaled by subtracting the mean score and then dividing by the standard deviation. This yields scores with a mean of 0 and a standard deviation of 1 for the cohort year. Appendix 1 describes each measure more fully.

Early development instrument index – mean early development score (MEDS)

The Early Development Instrument Index assesses “school readiness” (i.e. a child’s level of preparedness for the formal educational environment) in 5 developmental domains: physical health and well-being, social competence, emotional maturity, language and cognitive development, and communication skills and general knowledge [37, 38]. The EDI questionnaire is completed by kindergarten teachers for children in their classroom, typically in the second half of the school year. The index has acceptable interrater reliability and high internal consistency [21]. A child is considered developmentally vulnerable if scoring in the lowest 10^th percentile according to national norms in 1 or more developmental domains [39, 40]. Children having all 5 EDI scores (one for each domain) are used for the Early Development Instrument Index. Mean Early Development Score scaling in a manner consistent with the measures presented below has been created from this index.

Grade level assessments (Grades 3, 7, and 8)

These grade level assessments are designed to identify individual strengths and needs as well as classroom engagement. This information helps judge the student’s foundational knowledge and skills needed to support learning across the curricula.

Grade 3 assessment index

Each of two Grade 3 assessments is associated with specific competency measures. The reading assessment focuses on goals, strategies to make sense of texts, and comprehension. The numeracy assessment emphasises repeating patterns, equalities, representing whole numbers in a variety of ways, and using mental strategies for addition and subtraction. Outcome measures relating to both assessments concerned needing help, approaching and meeting expectations, and ‘out of range’ due to various problems.

Grade 7 assessment index

Two types of Grade 7 assessments were based on specific competencies:

One assessment involved numeracy: ordering fractions and decimal numbers, understanding numerical representation in a variety of ways, solving problems using number patterns, and adapting a variety of strategies to calculate and explain a mental math problem.

A second assessment concerned student engagement: competencies related to demonstrating an interest in personal learning, engaging in self-assessment, awareness of learning goals, participation in lessons, and accepting responsibility for assignments.

Numeracy relevant outcome measures were centered around mid-Grade 7 performance level: “not meeting, approaching, meeting, and out of range.” Measures for engagement were: “emerging, developing, establishing, inconsistent, and out of scope (mental health concerns).” The grade 7 assessment index score was based only on the set of numeracy competencies.

Grade 8 assessment index

One set of competencies associated with Grade 8 was assessed:

Reading Comprehension and Expository Writing: understanding, interpreting, and responding to texts; organising, choosing, and editing to clarify ideas; generates, selects and organises ideas to support reader understanding; chooses languages (word choices, sentence patterns) to make an impact on the reader; and uses conventions (spelling, grammar and/or punctuation) and resources to edit and proofread to clarify meaning.

Achievement Indices (Grades 9 and 12)

Grade 9 and Grade 12 Achievement Indices were developed along parallel lines. Data come from the Manitoba Health Insurance Registry (e.g.: demographic information on age and gender) held by the Manitoba Centre for Health Policy, and from Enrollment, Marks, and Assessments information on high school (grade 9-12) class enrollment and course marks received from Manitoba Education and Training.

Results

Exposure to poverty

Tables 1 and 2 show similar standard scores and standard errors for children with no family poverty and those experiencing no neighbourhood poverty. However, major differences were found in the scores associated with exposure to the two kinds of poverty.

Table 1: Standardised indices by exposure to household poverty (Employment Income Assistance (EIA) to mother – panel data). Standard errors [SE] are shown in brackets. **denotes statistical significance at the 5 percent level. 7,424 children had information on all six educational indices. Of these indices, 6,711 had a mother not receiving EIA in the year before birth of her child; 713 had a mother receiving EIA.

Composite educational	No EIA,	EIA,	Arithmetic difference
index	mean [SE]	mean [SE]	of means [SE]
Age 5 Mean Early Development Score	0.068 [0.012]	–0.543 [0.042]	0.611** [0.044]
Age 8 Grade 3 Competencies Index	0.086 [0.012]	–0.534 [0.039]	0.620** [0.041]
Age 12 Grade 7 Competencies Index	0.105 [0.012]	–0.693 [0.040]	0.798** [0.042]
Age 13 Grade 8 Competencies Index	0.085 [0.011]	–0.643 [0.041]	0.728** [0.043]
Age 14 Grade 9 Achievement Index	0.235 [0.010]	–0.559 [0.025]	0.794** [0.027]
Age 17 Grade 12 Achievement Index	0.242 [0.010]	–0.521 [0.029]	0.763** [0.031]

Table 2: Standardised indices by exposure to neighbourhood poverty (residence in low-income neighbourhood – panel data. Standard errors [SE] are shown in brackets. **denotes statistical significance at the 5 percent level. 6,183 children had no exposure to a low-income neighbourhood in their first year of life. 1,241 resided in a low-income neighbourhood in their first year of life.

Composite educational	No exposure to low-income	Exposure to low-income	Arithmetic difference of
index	neighbourhood, mean [SE]	neighboruhood, mean [SE]	means [SE]
Age 5 Mean Early Development Score	0.059 [0.012]	–0.240 [0.031]	0.299** [0.033]
Age 8 Grade 3 Competencies Index	0.080 [0.012]	–0.241 [0.029]	0.322** [0.032]
Age 12 Grade 7 Numeracy Competencies Index	0.092 [0.012]	–0.286 [0.031]	0.378** [0.033]
Age 13 Grade 8 Reading/Writing Competencies Index	0.063 [0.012]	–0.221 [0.031]	0.284** [0.033]
Age 14 Grade 9 Achievement Index	0.216 [0.011]	–0.128 [0.025]	0.344** [0.027]
Age 17 Grade 12 Achievement Language Arts & Math Index	0.229 [0.011]	–0.132 [0.026]	0.361** [0.028]

For example, the MEDS (Mean Early Development Score) for children with no exposure to household poverty was 0.068 (Table 1); for those with no exposure to neighborhood poverty the comparable figure was quite similar–0.059 (Table 2). The picture changes on comparing children experiencing the two types of poverty. Children with a mother receiving EIA had a MEDS score of – 0.543 (Table 1) while their counterparts exposed to neighborhood poverty recorded quite a different score of – 0.240 (Table 2). Household poverty is more strongly associated with lower scores on the Early Development Index than is neighbourhood poverty. Other educational indices showed parallel differences.

Intellectual development in the categories related to poverty was measured six times. Taking household poverty as an example (Table 1).

1. Differences among children with no EIA exposure and with EIA exposure (differences between Mean Early Development Scores of 0.611) are striking (by kindergarten entry – about age 5).
2. Differences among children with and without EIA exposure increase very slowly after age 8 when indices of development are compared. With the indices constructed differently, the findings must be interpreted cautiously.
3. Differences among children whose mothers received EIA and those who did not were significant at the 5 percent level for all ages. The difference between the Mean Early Development measure and the Grade 3 assessment was not significant at the 5 percent level, neither for children whose mothers received EIA nor those who did not. Differences between children aged 8 and age 12 are substantial; the increase in the difference of means between those ages is significant at the 5 percent level.

In summary, differences emerge by age 5, with administration of the Early Development Index. These differences are largely the same at age 8, even though another measure (the Grade 3 Competencies Index) is used. Such findings reinforce efforts at better understanding the early stages of childhood and call attention to facilitating development at these stages.

Comparison of Tables 1 and 2 show expected differences between children growing up in poverty and those who did not. The third column reports the average differences in outcomes between the columns for each measure. As expected, the differences were substantially greater for those children experiencing household poverty than those experiencing neighborhood poverty.

Introducing maternal mental health problems

The administrative data facilitate introducing new variables into the research. For example, maternal mental health data highlight the importance of this variable. Chartier et al. (2018) coded maternal diagnosis of mood and anxiety disorders, substance use disorders, psychotic disorders (including schizophrenia), or personality disorders in detail in the year before birth [41].

Examining an index using the maternal mental health diagnoses generates a different picture than does analysing the family poverty data. Comparing tables with mothers receiving EIA in the year before their birth (N=713) (Table 1) with those using mothers with a mental health diagnosis in this period (N = 527) (Table 3), the composite educational indices differed markedly between the two groups. The 713 children with a history of suffering from household poverty averaged lower standardised scores on the six educational indices (from – 0.543 to – 0.693) (Table 1) than did the 527 children whose mothers were diagnosed with mental health issues (from – 0.083 to – 0.237) (Table 3). The gap between Mean Early Development Scores (MEDS – measured about age 5) and the Grade 7 Numeracy Competency Index was somewhat greater for children suffering from household poverty (0.798–0.611=0.187) (Table 1).

Table 3: Standardised indices by exposures to mother with mental health diagnoses – panel data. Standard errors [SE] are shown in brackets. **denotes statistical significance at the 5 percent level. 6,897 children had mothers with no mental health problems in the year before birth; 527 children had mothers with mental health problems in the year before birth.

Composite educational	No exposure to mother	Exposure to mother	Arithmetic
index	with mental health	with mental health	difference of
	diagnosis mean [SE]	diagnosis mean [SE]	means [SE]
Age 5 Mean Early Development Score	0.025 [0.012]	–0.196 [0.052]	0.221** [0.053]
Age 8 Grade 3 Competencies Index	0.045 [0.012]	–0.218 [0.044]	0.263** [0.044]
Age 12 Grade 7 Numeracy Competencies Index	0.049 [0.012]	–0.237 [0.046]	0.286** [0.048]
Age 13 Grade 8 Reading/Writing Competencies Index	0.033 [0.012]	–0.221 [0.046]	0.254** [0.048]
Age 14 Grade 9 Achievement Index	0.178 [0.010]	–0.090 [0.036]	0.268** [0.039]
Age 17 Grade 12 Achievement (Language Arts & Math Index)	0.188 [0.010]	–0.083 [0.036]	0.271** [0.039]

Duration of poverty

Additional analyses present the number of years a child experienced household poverty and scores on the educational indices. Table 4 puts together the “no EIA, minority of years with EIA, and majority of years with EIA” categories for each of the six measures. Table 5 highlights similar data using the large N with single index comparisons.

Table 4: Standardised development indices by duration of household poverty (receipt of EIA to mother, in years to assessment - three categories, panel data). Standard errors [SE] are shown in brackets. Columns 1-6 report the mean index value for no years, for a minority or for a majority of years to each assessment where Employment Income Assistance (EIA) was received, and number in each category. Means in each category differed substantially from each other at the 5 percent level, through a series of two-sample T-tests.

Composite educational index	No EIA, mean [SE]	N	Minority of years, mean [SE]	N	Majority of years, mean [SE]	N
Age 5 <6 years used > Mean Early Development Score	0.108 [0.012]	6,194	–0.430 [0.052]	499	–0.531 [0.040]	731
Grade 3 Competencies Index	0.131 [0.012]	6,140	–0.343 [0.042]	606	–0.590 [0.040]	678
Grade 7 Competencies Index	0.164 [0.012]	6,080	–0.467 [0.037]	797	–0.756 [0.046]	547
Grade 8 Competencies Index	0.139 [0.012]	6,069	–0.414 [0.036]	832	–0.733 [0.049]	523
Grade 9 Achievement Index	0.299 [0.011]	6,058	–0.349 [0.023]	852	–0.660 [0.029]	514
Grade 12 Achievement Index	0.312 [0.011]	5,995	–0.325 [0.026]	931	–0.638 [0.035]	498

Table 5: Standardised development indices by duration of household poverty (receipt of EIA to mother, in years to assessment - three categories, using all cases available for a single time period). Standard Errors [SE] are shown in brackets. Columns 2-6 report the mean index value for no years, for a minority or for a majority of years to each assessment where Employment Income Assistance (EIA) was received, and number in each category. Means in each category differed statistically from each other at the 5 percent level through a series of two-sample T-tests.

Composite educational index	No EIA, mean [SE]	N	Minority of years, mean [SE]	N	Majority of years, mean [SE]	N
Mean Early Development Score	0.113 [0.006]	24,568	–0.398 [0.025]	2,091	–0.607 [0.021]	3,201
Grade 3 Competencies Index	0.128 [0.005]	35,464	–0.431 [0.018]	3,454	–0.690 [0.016]	4,396
Grade 7 Competencies Index	0.151 [0.003]	70,576	–0.444 [0.010]	10,025	–0.837 [0.012]	7,434
Grade 8 Competencies Index	0.141 [0.003]	70,352	–0.416 [0.010]	10,386	–0.797 [0.013]	7,065
Grade 9 Achievement Index	0.165 [0.004]	77,831	–0.538 [0.007]	13,170	–0.725 [0.008]	7,894
Grade 12 Achievement Index	0.172 [0.004]	54,249	–0.510 [0.009]	10,520	–0.776 [0.012]	5,125

The duration of poverty is important. Across all six indices representing varying ages and somewhat differing data, the children with most of their years spent in household poverty showed less intellectual development than their counterparts with fewer such years. Those living without experiencing poverty showed the greatest intellectual development.

For example, counting the number of years of assessment with the panel data (Table 4), children (age 5, 6 years of assessment) scored average values of 0.108 on the Mean Early Development Score with no years of EIA, (n=6,194). Children having a minority of years on EIA (n=499) averaged -0.430 while those spending most years on EIA (n=731) averaged -0.531. For each age-related index, the more years of poverty, the lower the educational index. Table 4 presents a fuller picture of household poverty across multiple years and all six indices. Similar trends were found for the duration of exposure to neighbourhood poverty.

Correlations among developmental indices

Indices developed for the six-wave panel study (N = 7,424) were also created using the largest available number of cases for each comparison. The correlations in Table 5 are remarkably stable (generally within .02) when the two approaches (panel study and largest available numbers) were compared; such stability is typical when the different approaches are compared. The pairs using largest available numbers varied between 15,015 and 88,035 in size.

Table 6 presents the moderate correlations among most developmental indices, regardless of the details in their construction. The correlation between the Grade 9 and Grade 12 Achievement Indices provides one exception. These two indices were built using very similar assessment methods and scaled logit scores calculated from rank categories based primarily on high school marks; this doubtless contributed to the higher correlation (723) between the Grade 9 and 12 indices.

Table 6: Correlations among standardised development indices of household poverty – panel data. 7,424 children provided information on all six measures. Correlations among indices are significant at the .05 level.

	Age 5	Age 8	Age 12	Age 13	Age 14
Composite educational index	mean early development	grade 3 competencies	grade 7 competencies	grade 8 competencies	grade 9 achievement
	score	index	index	index	index
Age 8 Grade 3 Competencies Index	0.464	—	—	—	—
Age 12 Grade 7 Competencies Index	0.432	0.518	—	—	—
Age 13 Grade 8 Competencies Index	0.459	0.510	0.578	—	—
Age 14 Grade 9 Achievement Index	0.410	0.415	0.537	—	—
Age 17 Grade 12 Achievement Index	0.385	0.408	0.505	0.527	0.723

Table 6 also shows a generally slow decline in correlations as the years between measurement increase. The correlations seem remarkably strong, given the number of years involved. Such correlations suggest that the developmental indices can be used for multiple purposes without worrying unduly about measurement issues. Change in circumstances with age and, perhaps, changes in measurement would seem responsible for the observed differences.

Maternal mental health, poverty, and educational outcomes

To further explore the relationship between maternal mental health and child educational outcomes, we conducted additional linear regression analyses using the panel study subset. These models examined the independent effects of specific maternal mental health diagnoses, poverty indicators, and sociodemographic factors on educational achievement at Grades 3 (Table 7), 9 (Table 8), and 12 (Table 9).

Table 7: Maternal mental health and associated sociodemographic covariates with outcome - grade 3 competencies index.

Covariates	Cases	Prevalence rate	Estimate (95% CLs)	Pr > ChiSq
Maternal ADHD in year before birth of child to 9th birthday	40	5.388	0.0000 (–0.2946, 0.2946)	0.9998
Maternal mood/anxiety disorder in year before birth of child to 9th birthday	3,033	408.54	–0.0568 (–0.1014, –0.0122)	0.0125
Maternal psychotic disorder in year before birth of child to 9th birthday	43	5.792	–0.0771 (–0.3616, 0.2073)	0.5951
Maternal substance abuse disorder in year before birth of child to 9th birthday	680	91.595	–0.1915 (–0.2679, –0.1151)	<.0001
Received EIA in majority of years to grade 3 assessment	1,962	264.278	–0.2731 (–0.3143, –0.2319)	<.0001
Low-income neighbourhood majority of years from child’s birth to age 8	3,540	476.832	–0.1490 (–0.1804, –0.1176)	<.0001
Mother not married/common-law at birth of child	3,540	476.832	–0.0574 (–0.1037, –0.0111)	0.0152
Mother below age 20 at birth of child	454	61.153	–0.0176 (–0.1147, 0.0795)	0.7224
Child was Small for Gestational Age	524	70.582	–0.1123 (–0.1965, –0.0281)	0.0089

Table 8: Maternal mental health and associated sociodemographic covariates with outcome - grade 9 achievement index.

Covariates	Cases	Prevalence rate	Estimate (95% CLs)	Pr > ChiSq
Maternal ADHD in year before birth of child to 15th birthday	89	11.988	–0.0890 (–0.2554, 0.0774)	0.2944
Maternal mood/anxiety disorder in year before birth of child to 15th birthday	3,822	514.817	–0.0592 (–0.0962, –0.0223)	0.0017
Maternal psychotic disorder in year before birth of child to 15th birthday	70	9.429	–0.0877 (–0.2750, 0.0995)	0.3583
Maternal substance abuse disorder in year before birth of child to 15th birthday	975	131.331	–0.2728 (–0.3284, –0.2172)	<.0001
Received EIA in majority of years to grade 9 assessment	1,880	253.233	–0.3599 (–0.3975, –0.3223)	<.0001
Low-income neighbourhood majority of years from child’s birth to age 14	3,514	473.33	–0.1452 (–0.1737, –0.1167)	<.0001
Mother not married/common-law at birth of child	3,540	476.832	–0.1525 (–0.1912, –0.1138)	<.0001
Mother below age 20 at birth of child	454	61.153	–0.0299 (–0.1107, 0.0510)	0.4689
Child was Small for Gestational Age	524	70.582	–0.0053 (–0.0758, 0.0652)	0.8828

Table 9: Maternal mental health and associated sociodemographic covariates with outcome - grade 12 achievement language arts and math index.

Covariates	Cases	Prevalence rate	Estimate (95% CLs)	Pr > ChiSq
Maternal ADHD in year before birth of child to 19th birthday	145	19.531	–0.0547 (–0.1884, 0.0789)	0.4223
Maternal mood/anxiety disorder in year before birth of child to 19th birthday	4,236	570.582	–0.0511 (–0.0892, –0.0131)	0.0084
Maternal psychotic disorder in year before birth of child to 19th birthday	87	11.719	0.0172 (–0.1540, 0.1885)	0.8436
Maternal substance abuse disorder in year before birth of child to 19th birthday	1,098	147.899	–0.2228 (–0.2772, –0.1683)	<.0001
Received EIA in majority of years to grade 12 assessment	1,927	259.564	–0.3722 (–0.4105, –0.3338)	<.0001
Low-income neighbourhood majority of years from child’s birth to age 17	3,601	485.048	–0.1396 (–0.1686, –0.1107)	<.0001
Mother not married/common-law at birth of child	3,540	476.832	–0.1271 (–0.1666, –0.0877)	<.0001
Mother below age 20 at birth of child	454	61.153	–0.0224 (–0.1045, 0.0598)	0.5934
Child was Small for Gestational Age	524	70.582	–0.0188 (–0.0906, 0.0529)	0.6072

Both maternal mood/anxiety and substance use disorders were associated with lower educational achievement in children across all three grade levels, even after adjusting for poverty and other sociodemographic factors. Children growing up in poverty, especially those in families receiving EIA, showed poorer educational outcomes, with the effect becoming more pronounced by Grade 12 (Table 9). Being born small for gestational age may affect early achievement, but its impact diminished by adolescence. Maternal marital status remained a modest but consistent predictor across the three grade levels.

Discussion

Earlier research has shown household poverty to be associated with worse early childhood outcomes than neighbourhood poverty [17, 21]. This study extends such work across childhood up to grade 12, looking at a series of indicators of intellectual development. Specifically, this paper has introduced analyses of 1) panel information on poverty and maternal mental health within large administrative data sets and 2) measures of intellectual development made as equivalent as possible given their collection in different ways. This study is specific to Manitoba, which has unique demographic and socioeconomic characteristics along with a strong data infrastructure. While our findings may not generalise directly to other regions, they demonstrate the value of using detailed population data to understand child development and poverty. These insights can inform similar work in comparable settings.

Tables 1 and 2 illustrate the same general pattern – the children experiencing poverty show lower scores on the measures of intellectual development than do their more affluent counterparts. Furthermore, household poverty is demonstrated to be more strongly associated with lower scores on the EDI than is neighborhood poverty, aligning with previous research. A UK scoping review found that individual-level measures of economic disadvantage were more strongly linked to child health outcomes than area-level measures [42]. Similarly, another UK study found household income to be a stronger predictor of academic achievement and various health outcomes compared to neighbourhood level socioeconomic indicators [43]. A British Columbia study also observed that household poverty was more consistently associated with poorer developmental outcomes in Canada [17, 21]. Comparisons with Table 3 highlight the importance of coding decisions in deciding the importance of variables. If household poverty is taken as a definition of poverty, poverty seems much more important than maternal mental health in affecting a child’s intellectual development. However, if neighbourhood poverty is used to define poverty, maternal mental health and poverty appear to have effects on childhood intellectual development which are more similar. Definition plays a critical role in interpretation.

To build on these findings, we disaggregated maternal mental health into specific diagnostic categories and found mood/anxiety disorders and substance use disorders to be consistently associated with lower educational achievement across Grades 3, 9, and 12, even after adjusting for poverty and other sociodemographic factors. These results highlight the importance of considering both the definition and the granularity of variables when interpreting their effects. It is also important to recognise that socioeconomic status may influence not only the prevalence of mental health conditions but also the likelihood of receiving a diagnosis and accessing treatment [44]. As a result, the observed associations may underestimate the true burden of maternal mental health challenges in lower-income populations.

Policies targeting household-level poverty appear to have a greater impact on early childhood outcomes than those aimed at neighborhood-level poverty. Direct economic support such as income supplements, housing assistance, or food programs may be particularly effective in improving children’s intellectual development [43, 45]. The variation in findings based on how poverty is defined also underscores the importance of using precise, household-level measures, in addition to neighborhood-level factors, to better identify families most in need [43]. While maternal health remains important, alleviating household poverty may be a more foundational intervention for supporting early development.

Some additional analyses are relatively simple. Dramatic differences in performance on the educational indices are seen when boys and girls are considered separately. Girls do much better than boys. But the two sexes show similar trends; experiencing poverty is consistently associated with lower index scores, no matter the age. Dividing the panel study participants into six age groups opens the door for other research on within-family influence. For example, instead of time spent in poverty, time spent with a mother having mental health issues could be studied. Alternatively, child’s mental health – particularly after the first few years – might be taken as an outcome of poverty.

Population-based data suggest many analytical possibilities. Additional variables based on various behaviors and medical diagnoses are available in the data base; these include maternal drug or alcohol use and maternal smoking during pregnancy. If childhood conditions are included, the list of medical conditions which might correlate with intellectual development increases markedly. Measures of overall health and of specific conditions such as diabetes might prove as or more important than poverty. Using measures of intellectual development for the older children allows selecting health diagnoses suitable for this group.

Such longitudinal information would also allow analysing both the larger effects of the COVID-19 pandemic in school districts with different characteristics [46] and the effects of varying educational interventions [47]. The effects of traumatic childhood events – such as family breakup or death of a family member – can be studied using baseline data from the previous assessment period. Work among children under the age of five to try to better understand how these differences become established needs to be increased [48]. More broadly, being able to use population-based data in various ways, incorporating panels, siblings, and instrumental variables, can be critical in performing nonexperimental research to investigate developmental differences [49].

Conclusion

The richness of the data has facilitated the study of childhood intellectual development. Household poverty appears to play an important role in the intellectual growth of children; neighbourhood poverty and maternal mental health also seem to influence such development.

Acknowledgements

The authors acknowledge the Manitoba Centre for Health Policy for use of data contained in the Manitoba Population Research Data Repository under project #2019-024 (PHRPC 2018/2019-70). Data used in this study were derived from databases provided by Manitoba Health, Manitoba Education, and Statistics Canada. The results and conclusion are those of the authors and no official endorsement by Manitoba Health or other data providers should be inferred.

Territory acknowledgement

The University of Manitoba campuses are located on original lands of Anishinaabeg, Ininewak, Anisininewuk, Dakota Oyate, and Dene, and on the National Homeland of the Red River Metis. We respect the Treaties that were made on these territories, we acknowledge the harms and mistakes of the past and we dedicate ourselves to move forward in partnership with indigenous communities in a spirit of reconciliation and collaboration.

Statement on conflict of interest

The authors have no conflicts of interest or relevant financial relationships to disclose.

Funding

This research was supported by funding from the Canadian Institutes of Health Research (grant no. 162111)

Ethics statement

This study was approved by the University of Manitoba Health Research Ethics Board (H2019:110) and the Health Information Privacy Commission at Manitoba Health, Seniors and Active Living (2018/2019 – 70). Using deidentified administrative data files did not require participants’ informed consent.

Data availability statement

The source data used in this study were originally collected during the routine administration of health and social services in Manitoba and were provided to the Manitoba Centre for Health Policy (MCHP) for secondary use in research under specific data sharing agreements between the data trustees and MCHP. The data are approved for use at MCHP only. They are not owned by the researchers or by MCHP and cannot be deposited in a public repository. To review source data specific to this article or project, interested parties should contact the MCHP Repository Access & Use team at MCHP.Access@umanitoba.ca. The team will then facilitate data access by seeking consent of the original data holders and the required privacy and ethics review bodies.

References

1. Power C, Kuh D, Morton S. From Developmental Origins of Adult Disease to Life Course Research on Adult Disease and Aging: Insights from Birth Cohort Studies. Annu Rev Public Health 2013; 34: 7–28. 10.1146/annurev-publhealth-031912-114423c

   10.1146/annurev-publhealth-031912-114423c

2. Wood D. Effect of child and family poverty on child health in the United States. Pediatrics 2003; 112: 707–11. 10.1542/peds.112.S3.707

   10.1542/peds.112.S3.707

3. Brooks-Gunn J, Duncan GJ. The effects of poverty on children. Future of Children 1997; 7: 55–71. 10.2307/1602387

   10.2307/1602387

4. Aber JL, Bennett NG, Conley DC, et al. The effects of poverty on child health and development. Annu Rev Public Health 1997; 18: 463–483. 10.1146/annurev.publhealth.18.1.463

   10.1146/annurev.publhealth.18.1.463

5. Putnam-Hornstein E, Needell B. Predictors of child protective service contact between birth and age five: An examination of California’s 2002 birth cohort. Child Youth Serv Rev 2011; 33: 2400–2407. 10.1016/j.childyouth.2011.04.006

   10.1016/j.childyouth.2011.04.006

6. Casey PH, Szeto KL, Robbins JM, et al. Child health-related quality of life and household food security. Arch Pediatr Adolesc Med 2005; 159: 51–56. 10.1001/archpedi.159.1.51

   10.1001/archpedi.159.1.51

7. Cook JT, Frank DA, Levenson SM, et al. Child food insecurity increases risks posed by household food insecurity to young children’s health. Journal of Nutrition 2006; 136: 1073–1076. 10.1093/jn/136.4.1073

   10.1093/jn/136.4.1073

8. Hart B, Risley TR. Meaningful Differences in the Everyday Experience of Young American Children. Remedial and special education 1998; 19: 371–371. 10.1007/s00431-005-0010-2

   10.1007/s00431-005-0010-2

9. Hair NL, Hanson JL, Wolfe BL, et al. Association of Child Poverty, Brain Development, and Academic Achievement. JAMA Pediatr 2015; 169: 822–829. 10.1001/jamapediatrics.2015.1475

   10.1001/jamapediatrics.2015.1475

10. Feijó DM, Pires JF, Gomes RMR, et al. The impact of child poverty on brain development: does money matter? Dement Neuropsychol 2023; 17: e20220105. 10.1590/1980-5764-DN-2022-0105

    10.1590/1980-5764-DN-2022-0105

11. Blair C, Raver CC. Poverty, Stress, and Brain Development: New Directions for Prevention and Intervention. Acad Pediatr 2016; 16: S30. https://doi.org/10.1016/j.acap.2016.01.010

12. Bradley RH, Corwyn RF. Socioeconomic status and child development. Annu Rev Psychol 2002; 53: 371–399. 10.1146/annurev.psych.53.100901.135233

    10.1146/annurev.psych.53.100901.135233

13. Browne DT, Wade M, Prime H, et al. School readiness amongst urban Canadian families: Risk profiles and family mediation. J Educ Psychol 2018; 110: 133–146. 10.1037/edu0000202

    10.1037/edu0000202

14. Duncan GJ, Dowsett CJ, Claessens A, et al. School Readiness and Later Achievement. Dev Psychol 2007; 43: 1428–1446. 10.1037/edu0000202

    10.1037/edu0000202

15. Shaw DS, Shelleby EC. Early-starting conduct problems: Intersection of conduct problems and poverty. Annu Rev Clin Psychol 2014; 10: 503–528. 10.1146/annurev-clinpsy-032813-153650

    10.1146/annurev-clinpsy-032813-153650

16. Currie J, Stabile M, Manivong P, et al. Child Health and Young Adult Outcomes. Journal of Human Resources 2010; 45: 517–548. https://doi.org/10.3368/jhr.45.3.517

17. Gill R, Karim ME, Puyat JH, et al. Childhood poverty, social support, immigration background and adolescent health and life satisfaction: A population-based longitudinal study. J Adolesc. Epub ahead of print 1 January 2024. 10.1002/jad.12408

    10.1002/jad.12408

18. Yang-Huang J, van Grieken A, You Y, et al. Changes in family poverty status and child health. Pediatrics 2021; 147: 2020016717. https://doi.org/10.1542/peds.2020-016717

19. Janus M, Brownell M, Reid-Westoby C, et al. Neighbourhood-level socioeconomic status and prevalence of teacher-reported health disorders among Canadian kindergarten children. Front Public Health 2023; 11: 1295195. 10.3389/fpubh.2023.1295195

    10.3389/fpubh.2023.1295195

20. Renneberg CK, Brund RBK, Heuckendorff S, et al. Preschool children from lower household incomes experience inequality in asthma treatment: findings from a Danish nationwide cohort study. Eur J Public Health 2024; 34: 85–90. 10.1093/eurpub/ckad187

    10.1093/eurpub/ckad187

21. Roos LL, Wall-Wieler E, Lee JB. Poverty and Early Childhood Outcomes. Pediatrics; 143. Epub ahead of print 1 June 2019. https://doi.org/10.1542/peds.2018-3426

22. Green MJ, Stritzel H, Smith C, et al. Timing of poverty in childhood and adolescent health: Evidence from the US and UK. Soc Sci Med 2018; 197: 136. 10.1016/j.socscimed.2017.12.004

    10.1016/j.socscimed.2017.12.004

23. Gill R, Karim ME, Puyat JH, et al. Childhood poverty and school readiness: Differences by poverty type and immigration background. SSM Popul Health 2024; 25: 101563. 10.1016/j.ssmph.2023.101563

    10.1016/j.ssmph.2023.101563

24. Dreyer BP. What Canada can teach us about addressing childhood poverty. Pediatrics; 143. Epub ahead of print 1 June 2019. https://doi.org/10.1542/peds.2019-0195

25. Social Planning Council of Winnipeg. Poverty, the pandemic, and the province: Manitoba child and family poverty, February 2023. Winnipeg, February 2023.

26. Government of Manitoba. Employment and Income Assistance (EIA). Government of Manitoba, https://www.gov.mb.ca/fs/eia/ (accessed 24 September 2018).

27. Oreopoulos P, Stabile M, Walld R, et al. Short-, Medium-, and Long-Term Consequences of Poor Infant Health. Journal of Human Resources 2008; 43: 88–138. 10.3368/jhr.43.1.88

    10.3368/jhr.43.1.88

28. O’Grady K, Deussing M, Scerbina T, et al. Measuring up: Canadian results of the OECD PISA study. Toronto, Canada, https://www.cmec.ca/Publications/Lists/Publications/Attachments/365/Book_PISA2015_EN_Dec5.pdf (2016, accessed 1 December 2024).

29. Statistics Canada. Census Profile, 2021 Census Population. Statistics Canada.

30. Statistics Canada. Aboriginal Peoples: Fact Sheet for Manitoba. Statistics Canada.

31. Richards J. Student performance in PISA 2018: Nettlesome questions for Canada. Toronto, Canada, July 2020.

32. O’Grady K, Rostamian A, Monk J, et al. Measuring up: Canadian Results of the OECD PISA 2018 Study. Toronto ON, https://cmec.ca/Publications/Lists/Publications/Attachments/419/PISA2018_GC_Report_EN.pdf (2021, accessed 27 May 2025).

33. Canadian Institute for Health Information. Children Vulnerable in Areas of Early Development: A Determinant of Child Health. Ottawa, 2014.

34. Roos L. A Research Registry Uses, Development, and Accuracy. J Clin Epidemiol 1999; 52: 39–47. 10.3368/jhr.43.1.88

    10.3368/jhr.43.1.88

35. Roos LL, Gupta S, Soodeen R-A, et al. Data Quality in an Information-Rich Environment: Canada as an Example. Can J Aging 2005; 24: 153–170. 10.1353/cja.2005.0055

    10.1353/cja.2005.0055

36. Manitoba Centre for Health Policy. Concept: Early Development Instrument (EDI) Outcomes. Manitoba Centre for Health Policy, http://mchp-appserv.cpe.umanitoba.ca/viewConcept.php?conceptID=1423 (2022, accessed 3 December 2024).

37. Janus M, Offord DR. Development and psychometric properties of the Early Development Instrument (EDI): A measure of children’s school readiness. Can J Behav Sci 2007; 39: 1–22. 10.1037/cjbs2007001

    10.1037/cjbs2007001

38. Government of Manitoba. Assessment and Evaluation. Kindergarten: Early Development Instrument. Government of Manitoba, https://www.edu.gov.mb.ca/k12/assess/kindergarten/index.html (accessed 23 April 2025).

39. Janus M, Brinkman S, Duku E, et al. The Early Development Instrument: A Population-Based Measure for Communities. A Handbook on Development, Properties, and Use. Hamilton, Canada: Offord Centre for Child Studies, 2007.

40. Snow KL. Measuring School Readiness: Conceptual and Practical Considerations. Early Educ Dev 2006; 17: 7–41. 10.1207/s15566935eed1701_2

    10.1207/s15566935eed1701_2

41. Chartier M, Bolton J, Mota N, et al. Mental illness among adult Manitobans. Winnipeg, MB, 2018.

42. Clery A, Grant C, Harron K, et al. Measuring disadvantage in the early years in the UK: a systematic scoping review. The Lancet; 400. https://doi.org/10.1016/S0140-6736(22)02242-5

43. Skarda I, Cookson R, Gilbert R. Does household income predict health and educational outcomes in childhood better than neighbourhood deprivation? J Public Health (Bangkok) 2025; 47: 62–73. 10.1093/pubmed/fdae283

    10.1093/pubmed/fdae283

44. Hodgkinson S, Godoy L, Beers LS, et al. Improving Mental Health Access for Low-Income Children and Families in the Primary Care Setting. Pediatrics; 139. Epub ahead of print 1 January 2017. 10.1542/peds.2015-1175

    10.1542/peds.2015-1175

45. Goldfeld S, Downes M, Gray S, et al. Household income supplements in early childhood to reduce inequities in children’s development. Soc Sci Med 2024; 340: 116430. 10.1016/j.socscimed.2023.116430

    10.1016/j.socscimed.2023.116430

46. Kane T, Reardon S. Parents don’t understand how far behind their kids are. New York Times, 11 May 2023.

47. Karoly L, Kilburn M, Cannon J. Early childhood interventions: Proven results, future promises. Rand, Santa Monica CA, 2005.

48. Hahn RA, Barnett WS. Early Childhood Education: Health, Equity, and Economics. Annu Rev Public Health 2023; 44: 75–92. https://doi.org/10.1146/annurev-publhealth-071321-032337

49. Shadish W, Cook T, Campbell D. Experimental and quasi-experimental and design for generalized casual interference. Boston, New York: Houghton Mufflin Company, 2002.

50. Manitoba Centre for Health Policy. Concept: Early Development Instrument (EDI) Outcomes. Manitoba Centre for Health Policy.

51. Duku E, Pottruff M, Janus M. Creating and Evaluating Two Cumulative Developmental Vulnerability Risk Measures. Int J Popul Data Sci; 7. Epub ahead of print 25 August 2022. 10.23889/ijpds.v7i3.1813

    10.23889/ijpds.v7i3.1813

52. Manitoba Centre for Health Policy. Concept: Grade Level Assessments. Manitoba Centre for Health Policy.

53. Manitoba Centre for Health Policy. Concept: Education Indices. Manitoba Centre for Health Policy.

54. Manitoba Centre for Health Policy. Concept: Grade 9 Achievement Index. Manitoba Centre for Health Policy.

55. Willms JD. Social Class Segregation and Its Relationship to Pupils’ Examination Results in Scotland. Am Sociol Rev 1986; 51: 224. https://doi.org/10.2307/2095518