Video Abstract

Video Abstract

OBJECTIVES:

To test associations between onset of formal child care (in infancy or as a toddler), high school graduation, and employment earnings from ages 18 to 35 years.

METHODS:

A 30-year prospective cohort follow-up study, with linkage to government administrative databases (N =3020). Exposure included formal child care, if any, by accredited caregivers in centers or residential settings at ages 6 months and 1, 1.5, 2, 3, and 4 years. A propensity score analysis was conducted to control for social selection bias.

RESULTS:

Of 2905 participants with data on child care use, 59.4% of male participants and 78.5% of female participants completed high school by age 22 to 23. Mean income at last follow-up (n = 2860) was $47 000 (Canadian dollars) (SD = 37 700) and $32 500 (SD = 26 800), respectively. Using group-based trajectory modeling, we identified 3 groups: formal child care onset in infancy (∼6 months), formal child care onset as a toddler (after 2.5 years), and never exposed. After propensity score weighting, boys with child care started in infancy had greater odds of graduating than those never exposed (odds ratio [OR] 1.39; 95% confidence interval [CI]: 1.18–1.63; P < .001). Boys attending child care had reduced odds of low income as young adults (infant onset: OR 0.60 [95% CI: 0.46–0.84; P < .001]; toddler onset: OR 0.63 [95% CI: 0.45–0.82; P < .001]). Girls’ graduation rates and incomes revealed no significant association with child care attendance.

CONCLUSIONS:

For boys, formal child care was associated with higher high school completion rates and reduced risk of adult poverty. Benefits for boys may therefore extend beyond school readiness, academic performance, and parental workforce participation.

What’s Known on This Subject:

Correlational studies suggested that early child care services with population-wide availability can potentially be effective tools for improving developmental trajectories. However, the association of child care attendance with outcomes beyond adolescence has been documented in few studies. Long-term outcomes remain to be explored.

What This Study Adds:

With this study, we provide the first prospective evidence indicating that for boys, regulated child care services initiated in infancy may increase high school completion and prevent poverty in adulthood.

Research suggests that educational attainment and earning capacity are rooted in early childhood.14  Poor quality of care and inadequate learning environments in early childhood have been identified as risk factors for poor academic performance and mental health problems in children and adolescents. Adverse childhood experiences are also linked to unhealthy or risk-taking behaviors that place children at increased risk of educational underachievement, unemployment, poverty, health-related problems, or premature mortality.411 

In line with experimental evidence,1215  correlational studies have suggested that early child care services (ie, day care) with population-wide availability can potentially be effective tools for improving developmental trajectories.16,17  However, social selection (ie families using child care services tend to be of higher socioeconomic status because they need to send their children somewhere during working hours) has limited the conclusions that could be drawn from correlational research. Yet, over the past decade, the use of statistical methods, such as propensity score weighting, and instrumental variables have strengthened the possibility of drawing inferences on the impact of child care services. A number of outcomes then remain to be explored.

First, few studies have documented the association, if any, of child care attendance with outcomes assessed beyond middle childhood or adolescence. Correlational studies of economically and ethnically diverse populations have revealed that child care services were associated with better cognitive and academic functioning and mental health during childhood and adolescence, especially in settings with high-quality child care.1623  These associations were shown to persist until the end of elementary school24  and into high school.2022  However, several studies indicated that these effects may fade over time.25,26  To our knowledge, there is, as yet, no evidence that child care attendance is linked to high school completion or to future economic success.

Second, authors of few studies have examined whether starting child care at an earlier or later age is more beneficial, and none have included follow-up into adulthood. Timing studies have yielded mixed results, some reporting earlier,16,24,27,28  and others later,29,30  onset as related to positive outcomes. The question of optimal timing for starting child care has important implications for child development, mothers in the workforce, duration of parental leave, and public investment in these and other child-related services.

Third, there is limited evidence of sex differences as reported in associations between child care, academic achievement, and later economic outcomes.31  The lack of subgroups limits the application of findings to health promotion and economic equality interventions.

Lastly, long-term studies that have linked early intensive child care programs to earned income in adulthood have generally relied on self-reported incomes at a single time point or for a short time frame. This has yielded mixed results.32  Self-reported outcomes potentially introduce response bias and/or measurement error; in addition, these studies failed to capture more complete information on income dynamics and fluctuations in adulthood.

In the current study, we examined the associations between attendance at child care before school entry, high school graduation (official records by age 22–23 years), and employment earnings from ages 18 to 35 years (from government tax records data) in a random sample of births in Quebec, Canada. We specifically tested for differential associations between earlier versus later age of onset of child care services and adult academic and economic outcomes, including sex differences.

Data were obtained from the Quebec Longitudinal Study of Kindergarten Children (QLSKC), an ongoing prospective, population-based cohort study of 3020 children (1420 girls [47.2%] and 1600 boys [52.8%]) born in 1980 and 1981.33 

We conducted a database linkage study using the records of the QLSKC cohort to determine participants exposed and not exposed to child care, as originally reported by the mother. Data regarding high school graduation were provided by the Quebec Ministry of Education records for all participants at ages 22 to 23 years. Personal income data were obtained from Statistics Canada tax records for years 1998 through 2015, with 18 measurement points at ages 18 to 35 years. A description of the linkage methods was recently published.34  Participants with no information on child care attendance from 6 months to age 4 years (n = 112; 3.7%) or with no filed tax returns data from age 18 to 35 years (n = 50; 1.7%) were excluded from the trajectory analysis. Sample sizes varied thereafter for specific outcomes and predictors. The final samples included 2905 participants with complete data for both child care attendance and high school graduation and 2860 participants with data available on both child care attendance and employment earnings.

The study was approved by the Research Ethics Boards of the Sainte-Justine Research Center and the University of Montreal. Written informed consent was provided by parents.

Child Care Services

Child care was defined in accordance with the Act Respecting Child Day Care (Revised Statutes of Quebec c.S-4.1, 1979) (an act now replaced by the Educational Child Care Act, c.S-4.1.1, 2005) and regulations from the Quebec Office of Child Care Services (1980). For further details, see the Supplemental Information. Mothers originally reported type and quantity of child care use from 6 months to age 4 years (at 6 months and 1, 1.5, 2, 3, and 4 years). Child care types included (1) center-based care, (2) family-based care, (3) home care in the child’s home by a nonrelative, and (4) home care provided by the mother or a relative. For this study, we defined formal child care services as regulated center-based and family-based child care settings operating weekdays and required to conform to legislative provincial standards. Children were classified into the child care type, if any, in which they spent the most hours per week; quantity of use was originally coded as follows: 0, never; 1, 1 to 10 hours; 2, 10 to 20 hours; 3, 20 to 40 hours; and 4, >40 hours.

Outcomes and Control Variables

Primary outcomes were high school completion by age 22 to 23 from official records and adult income at ages 18 to 35, as measured by government tax return data. Income was defined as the total of all personal pretax wages, salaries, and commissions, excluding capital gains. Low-income threshold was defined as the 20th percentile for the study sample. A total of 16 covariates representing a well-documented source of social selection bias were selected for propensity score weighting (Supplemental Information).35  These included baseline indicators of family socioeconomic status and structure, perinatal child characteristics, and stimulation received at home, as measured on family survey questionnaires.

Group-based developmental trajectory modeling36  was used to estimate child care attendance pattern over time, with probability of child care attendance specified as following a cubic function of age.37  The same analysis was performed to model employment earnings for each year from age 18 to 35 years. The aim was to identify groups that followed a distinctive low-income trajectory. Final model selection was based on the Bayesian information criterion (BIC), Akaike information criterion (AIC), model-adequacy tests, quality of classification, and parsimony.

The unadjusted association between child care trajectory and outcomes was estimated by using binary logistic regression (high school completion) or multinomial logistic regression (income trajectory). Missing data on covariates were handled by using multiple imputation by chained equations. Ten complete data sets were generated, a number generally considered sufficient.38  For the adjusted model, propensity score weighting was used to control for social selection bias and to isolate the putative effect of child care on high school graduation and income trajectories by using the 16 covariates mentioned earlier that were associated (P < .10) with either child care attendance trajectory or outcomes or both.35  Propensity score weighting was applied by using multinomial logistic regression and inverse probability of treatment weighting (IPTW) after multiple imputations.3840  Covariance balance was assessed to obtain the standardized mean difference with and without propensity score weighting. The associations between child care onset and outcomes of interest were then reestimated. Adjusted models of each outcome were stratified by sex.

To verify robustness of the results, we repeated the analyses, introducing additional covariates such as attention deficit and IQ, as measured at ages 12 and 13 years, respectively (Supplemental Information). Attention deficit and IQ were shown to be crucial to the prediction of educational and economic outcomes.4145 

All analyses were conducted by using Stata version 14 (Stata Corp, College Station, TX) and/or R for statistical computing versions 3.5.2 and 3.5.3 (64 bit). Significance was set at 0.05, and all tests were 2-tailed.

The best model identified the following 3 patterns of attendance (Fig 1): infant onset (n = 547; 18.8%) included children with a high likelihood of starting formal child care at ∼6 months of age and continuing through to age 4 to 5 years; toddler onset (n = 422; 14.5%), after 2.5 years; and never exposed (n = 1936; 66.6%) included children who had never attended formal child care in their preschool years. Model fit statistics are reported in Supplemental Table 4.

FIGURE 1

Trajectories of formal child care attendance before school entry. The y-axis represents the probability of attending child care services from 6 months to age 4 years. Boxes represent observed values, and lines represent the fitted regressions slopes. Fit indices for the model included the lowest BIC (−5431.01) and mean odds of correct classification (31.03) (ie, the model classified the participants 31.03 times better than classification by chance, which is adequate if >5.0). Details, including model fit statistics, are provided in Supplemental Table 4.

FIGURE 1

Trajectories of formal child care attendance before school entry. The y-axis represents the probability of attending child care services from 6 months to age 4 years. Boxes represent observed values, and lines represent the fitted regressions slopes. Fit indices for the model included the lowest BIC (−5431.01) and mean odds of correct classification (31.03) (ie, the model classified the participants 31.03 times better than classification by chance, which is adequate if >5.0). Details, including model fit statistics, are provided in Supplemental Table 4.

In Table 1, we report baseline child and family characteristics by child care trajectory as well as statistical differences between the 3 groups before propensity score weighting. Children who started formal child care in infancy, as compared with starting as toddlers or never exposed, were less likely to come from families with low levels of education or low occupational prestige. Mothers were more likely to be working; they were also less likely to have had a teenage pregnancy. After propensity score weighting, differences between groups decreased substantially, and similarities on personal characteristics and family backgrounds increased. The standardized mean difference for covariates was below the conservative threshold of 0.10, representing balanced grouping46  (Supplemental Tables 5 through 7, Supplemental Fig 4).

TABLE 1

Baseline Child and Family Characteristics by Child Care Trajectory (Before Adjustment)

Formal Child Care Attendance
Never Exposed (n = 1936)Infant Onset (n = 547)Toddler Onset (n = 422)P
Child characteristics, No. (%)     
 Male participants 1016 (52.5) 285 (52.1) 233 (52.2) .56 
 Birth order     
  Firstborn 811 (42.1) 344 (63.8) 195 (47.3) — 
  Second born 755 (39.2) 159 (29.5) 161 (39.1) .000 
  Third born or higher 359 (18.6) 36 (6.7) 56 (13.6) — 
 Low birth wt, <2500 g 105 (8.0) 26 (7.1) 20 (7.6) .85 
 Preterm birth, <37 wk gestation 297 (23.4) 73 (19.9) 69 (25.3) .24 
Family characteristics     
 Mother born in Canada, No. (%) 1869 (96.9) 517 (94.9) 392 (93.8) .003 
 Father born in Canada, No. (%) 1658 (96.1) 453 (92.6) 335 (89.10) .000 
 Lowa maternal education, No. (%) 610 (31.8) 96 (17.8) 91 (22.1) .000 
 Mother employed during child’s preschool years, No. (%) 715 (37.4) 424 (78.6) 299 (71.9) .000 
 Teenaged mother at child’s birth, No. (%) 62 (3.2) 8 (1.5) 21 (5.02) .007 
 Lowa paternal education, No. (%) 603 (35.5) 127 (26.2) 99 (27.9) .000 
 Family structure, No. (%)     
  Intact family unit 1235 (87.7) 308 (80.0) 194 (67.8) — 
  Single-parent family 132 (9.4) 53 (13.8) 70 (24.5) .000 
  Blended family 42 (3.0) 24 (6.2) 22 (7.7) — 
 Maternal age at child’s birth, y, mean (SD) 26.6 (4.6) 26.6 (3.9) 26.5 (4.8) .80 
 Paternal age at child’s birth, y, mean (SD) 29.20 (5.6) 28.67 (5.5) 29.07 (4.9) .14 
 Maternal occupational socioeconomic index,b mean (SD) 42.43 (12.8) 47.23 (12.7) 44.90 (12.3) .000 
 Paternal occupational socioeconomic index,b mean (SD) 43.06 (14.5) 44.54 (14.5) 44.60 (14.3) .05 
 Mother-child interactions, mean (SD)     
  Stimulating −1.82 (3.8) −1.46 (3.9) −1.45 (3.9) .06 
  Pleasurable 14.35 (6.3) 13.30 (6.7) 13.51 (6.7) .07 
Formal Child Care Attendance
Never Exposed (n = 1936)Infant Onset (n = 547)Toddler Onset (n = 422)P
Child characteristics, No. (%)     
 Male participants 1016 (52.5) 285 (52.1) 233 (52.2) .56 
 Birth order     
  Firstborn 811 (42.1) 344 (63.8) 195 (47.3) — 
  Second born 755 (39.2) 159 (29.5) 161 (39.1) .000 
  Third born or higher 359 (18.6) 36 (6.7) 56 (13.6) — 
 Low birth wt, <2500 g 105 (8.0) 26 (7.1) 20 (7.6) .85 
 Preterm birth, <37 wk gestation 297 (23.4) 73 (19.9) 69 (25.3) .24 
Family characteristics     
 Mother born in Canada, No. (%) 1869 (96.9) 517 (94.9) 392 (93.8) .003 
 Father born in Canada, No. (%) 1658 (96.1) 453 (92.6) 335 (89.10) .000 
 Lowa maternal education, No. (%) 610 (31.8) 96 (17.8) 91 (22.1) .000 
 Mother employed during child’s preschool years, No. (%) 715 (37.4) 424 (78.6) 299 (71.9) .000 
 Teenaged mother at child’s birth, No. (%) 62 (3.2) 8 (1.5) 21 (5.02) .007 
 Lowa paternal education, No. (%) 603 (35.5) 127 (26.2) 99 (27.9) .000 
 Family structure, No. (%)     
  Intact family unit 1235 (87.7) 308 (80.0) 194 (67.8) — 
  Single-parent family 132 (9.4) 53 (13.8) 70 (24.5) .000 
  Blended family 42 (3.0) 24 (6.2) 22 (7.7) — 
 Maternal age at child’s birth, y, mean (SD) 26.6 (4.6) 26.6 (3.9) 26.5 (4.8) .80 
 Paternal age at child’s birth, y, mean (SD) 29.20 (5.6) 28.67 (5.5) 29.07 (4.9) .14 
 Maternal occupational socioeconomic index,b mean (SD) 42.43 (12.8) 47.23 (12.7) 44.90 (12.3) .000 
 Paternal occupational socioeconomic index,b mean (SD) 43.06 (14.5) 44.54 (14.5) 44.60 (14.3) .05 
 Mother-child interactions, mean (SD)     
  Stimulating −1.82 (3.8) −1.46 (3.9) −1.45 (3.9) .06 
  Pleasurable 14.35 (6.3) 13.30 (6.7) 13.51 (6.7) .07 

P values are based on analysis of variance for continuous variables or Pearson’s χ2 test for categorical variables. —, not applicable.

a

No high school diploma.

b

Based on and occupational prestige scale (Supplemental Information).

Educational, economic, and family characteristics at follow-up are shown in Table 2. When compared with male participants, female participants had higher high school graduation rates (78.5% vs 59.4%), had lower personal earnings ($32 500 vs $47 000), were more likely to be married or cohabiting, and were more likely to have children living in the household by age 35 years.

TABLE 2

Education, Earnings, and Family Characteristics at Age 35 Years

CharacteristicsaFemale Participants (n = 1370)Male Participants (n = 1530)
High school completion by age 22–23 y, No. (%) 1080 (78.5) 910 (59.4) 
Personal income at age 35 y, n 1360 1500 
 $ CAD, mean (SD) 32 500 (26 800) 47 000 (37 700) 
 $ CAD, median (IQR) 31 100 (7290–50 200) 44 500 (19 300–68 300) 
 $ CAD, range 0–190 000 0–370 300 
Family characteristics at age 35 y, No. (%)   
 Married or cohabiting 870 (64) 820 (55) 
 Divorced, separated, or single 440 (32) 530 (35) 
No. offspring in household, No. (%)   
 0 children 310 (23.2) 520 (34.3) 
 1–2 children 760 (56) 680 (45.4) 
 3+ children 230 (17) 160 (11) 
CharacteristicsaFemale Participants (n = 1370)Male Participants (n = 1530)
High school completion by age 22–23 y, No. (%) 1080 (78.5) 910 (59.4) 
Personal income at age 35 y, n 1360 1500 
 $ CAD, mean (SD) 32 500 (26 800) 47 000 (37 700) 
 $ CAD, median (IQR) 31 100 (7290–50 200) 44 500 (19 300–68 300) 
 $ CAD, range 0–190 000 0–370 300 
Family characteristics at age 35 y, No. (%)   
 Married or cohabiting 870 (64) 820 (55) 
 Divorced, separated, or single 440 (32) 530 (35) 
No. offspring in household, No. (%)   
 0 children 310 (23.2) 520 (34.3) 
 1–2 children 760 (56) 680 (45.4) 
 3+ children 230 (17) 160 (11) 

$ CAD, Canadian dollar; IQR, interquartile range.

a

Up to 10% missing. In accordance with Statistics Canada data protection requirements, displayed counts are rounded to base 10, and percentages are rounded to 1 decimal point; earnings are rounded to the nearest 100, and ranges represent the mean of the 5 lowest and 5 highest scores, respectively.

Bivariate association and adjusted models with IPTW propensity scoring revealed that the time of onset of formal child care was significantly associated with high school graduation in the overall study sample (Supplemental Table 8). After IPTW propensity score adjustment, boys who initiated formal child care in infancy had higher odds of high school graduation than those who were never exposed (odds ratio [OR] 1.39; 95% confidence interval [CI]: 1.18–1.63; P < .001) (Table 3) and a 7.7-point increase (ie, 14% higher) in the rate of graduation (Fig 2A). This pattern was not found in girls. Girls graduated from high school at higher rates than boys, and the rate did not differ according to child care attendance or timing thereof.

TABLE 3

High School Graduation and Risk of Poverty at Ages 18–35 Years by Child Care Attendance

Child Care AttendanceEducationRisk of Poverty
Completed High School (n = 2905)Low (n = 1840)High (n = 340)
OR (95% CI)POR (95% CI)OR (95% CI)P
Female Participants      
 Never exposed 1 (reference) — 1 (reference) 1 (reference) — 
 Onset in infancy 1.11 (0.91–1.35) .31 — 0.82 (0.64–1.06) .13 
 Onset as toddler 1.10 (0.91–1.35) .29 — 0.92 (0.71–1.18) .50 
Male Participants      
 Never exposed 1 (reference) — 1 (reference) 1 (reference) — 
 Onset in infancy 1.39 (1.18–1.63) .001 — 0.60 (0.46–0.84) .001 
 Onset as toddler 1.06 (0.90–1.24) .44 — 0.63 (0.45–0.82) .001 
Child Care AttendanceEducationRisk of Poverty
Completed High School (n = 2905)Low (n = 1840)High (n = 340)
OR (95% CI)POR (95% CI)OR (95% CI)P
Female Participants      
 Never exposed 1 (reference) — 1 (reference) 1 (reference) — 
 Onset in infancy 1.11 (0.91–1.35) .31 — 0.82 (0.64–1.06) .13 
 Onset as toddler 1.10 (0.91–1.35) .29 — 0.92 (0.71–1.18) .50 
Male Participants      
 Never exposed 1 (reference) — 1 (reference) 1 (reference) — 
 Onset in infancy 1.39 (1.18–1.63) .001 — 0.60 (0.46–0.84) .001 
 Onset as toddler 1.06 (0.90–1.24) .44 — 0.63 (0.45–0.82) .001 

Results were adjusted by inverse probability weighting by using propensity scores from all 16 confounders related to selection bias. —, not applicable.

FIGURE 2

Adjusted rates for high school completion and risk of poverty in adulthood by child care attendance. A, High school completion. B, Risk of poverty. Values are adjusted by IPTW by using propensity scores. Error bars indicate 95% CIs.

FIGURE 2

Adjusted rates for high school completion and risk of poverty in adulthood by child care attendance. A, High school completion. B, Risk of poverty. Values are adjusted by IPTW by using propensity scores. Error bars indicate 95% CIs.

Trajectories of low income from 18 to 35 years are shown in Fig 3. The best model identified the following patterns: stable, low risk of poverty (n = 1900; 64.4%), moderate risk of poverty (n = 710; 24%), and high, rapidly increasing risk of poverty (n = 360; 12%). The difference in median income at age 35 between the low- and high-risk trajectories was $44 100. Model fit indices are presented in Supplemental Table 9.

FIGURE 3

Risk of poverty at ages 18 to 35 years. The y-axis represents the probability of earning a low income (being at the 20th percentile) for each year of follow-up, as reported in individual tax return records. Fit indices for the model included the lowest BIC (−18073.04) and mean odds of correct classification (the model classified the participants 19.1 times better than a classification by chance, which is adequate if >5.0). In accordance with Statistics Canada data protection requirements, displayed counts in the figure are rounded to base 10, and percentages are rounded to 1 decimal point; medians and interquartile ranges (IQRs) are rounded to base 100. Details, including model fit statistics, are provided in Supplemental Table 9.

FIGURE 3

Risk of poverty at ages 18 to 35 years. The y-axis represents the probability of earning a low income (being at the 20th percentile) for each year of follow-up, as reported in individual tax return records. Fit indices for the model included the lowest BIC (−18073.04) and mean odds of correct classification (the model classified the participants 19.1 times better than a classification by chance, which is adequate if >5.0). In accordance with Statistics Canada data protection requirements, displayed counts in the figure are rounded to base 10, and percentages are rounded to 1 decimal point; medians and interquartile ranges (IQRs) are rounded to base 100. Details, including model fit statistics, are provided in Supplemental Table 9.

Both unadjusted and adjusted models revealed a lower risk of poverty when the participant had been enrolled in a formal child care service (Supplemental Table 10). In the sex differences analysis, however, only male participants showed a significant association between formal child care and lower likelihood of poverty (infant onset: OR 0.60 [95% CI: 0.46–0.84; P < .001]; toddler onset: OR 0.63 [95% CI: 0.45–0.82; P < .001]) (Table 3, Fig 2B).

Attention deficit and IQ at ages 12 and 13 years, respectively, are significantly associated with high school graduation and earned income. After introducing these additional covariates into the main adjusted model, the association with child care attendance remained significant across outcomes. The covariate-adjusted model yielded slightly higher estimates of an association between infant onset of child care and high school completion compared with the adjusted model without the additional covariates (Supplemental Tables 8 and 11).

Little is known about the long-term outcomes of child care attendance and whether early gains translate into benefits later in life. Using a large and economically diverse population-based sample, we explored whether child care before school entry (ie, before age 5 years) was associated with educational attainment and employment earnings in young adulthood up to 3 decades later. Results were adjusted by using propensity score weighting to control for social selection bias and reduce differences in family background between children who did or did not attend child care. Adjusted results revealed that, for boys, early enrollment in formal child care was associated with higher rates of high school graduation and a lower likelihood of poverty at ages 18 to 35 years. For girls, graduation rates and incomes revealed no significant association with child care.

With an OR of 1.39 for high school graduation and a protective OR of 0.60 against the risk of poverty, male participants attending formal child care since infancy were ∼8 percentage points more likely to graduate from high school and 4 percentage points less likely to follow a trajectory of persistent low income in adulthood, as compared with those with no exposure to child care. Starting as a toddler also conferred protective advantages against poverty (∼4 percentage points). Although the effect sizes were small, it is important to keep in mind, as previously noted,47  that because of the widespread use of child care, those small effects may be highly relevant at the population level.

Given that failure to complete a high school education carries with it substantial economic costs to individuals and society,48,49  our findings provide evidence that the provision of early child care can be a cost-effective means of increasing social development, economic opportunity, and protection against poverty.

Our results are consistent with previous experimental studies demonstrating long-term gain for children enrolled in intensive preschool intervention programs.26,5052  Three studies from the Chicago Longitudinal Study on the large-scale Child-Parent Center Education Program reported higher rates of high school graduation for male participants along with higher socioeconomic levels (including income), with benefits increasing with intervention duration.1214  It is noteworthy that these structured programs were conducted on a population from low-income inner city schools in Chicago. This limits their generalizability to community-based child care services, like ours, that cover the entire income spectrum in the population, with child care services varying from home based to center based and of varying levels of quality.

Caution should be used in generalizing the results to cohorts of participants born in other time periods. However, we note that authors of previous studies reported benefits of early formal child care up to adolescence using distinct birth cohorts separated by at least 10 years, namely the British Millennium Cohort Study,16  the Eunice Kennedy Shriver National Institute of Child Health and Human Development Study of Early Child Care,2022,27  and the Quebec Longitudinal Study of Child Development.19,24,28  The US Early Childhood Longitudinal Study–Kindergarten class of 1998–1999 revealed that children who started center care programs at ages 2 to 3 years had better reading and math scores at school entry than those who started at younger or older ages.29  Methodologic differences between the current study and the Early Childhood Longitudinal Study–Kindergarten include assessment of outcomes, follow-up duration, and statistical methodology, limiting a direct comparison.

Several plausible mechanisms may account for the associations observed in this study. There is already a well-established literature documenting the association between early child care attendance, school readiness, and better academic performance in literacy and mathematics. The 2 latter are likely to mediate the association with high school graduation or later employments earnings. Early development of socioemotional competence53,54  could also function as an important mediator. Social-emotional learning and a play-centered early child care environment that includes ample opportunities for relational and active learning may be particularly effective in fostering school adaptation for boys, thereby increasing higher educational attainment and contributing to later earning capacity.

Our findings provide information about the age at which onset of child care services may be the most beneficial and to whom. In particular, our findings indicate the importance of such services for boys, who are generally at higher risk of problem behavior and school dropout.

The benefits observed for boys are not generalizable to girls. Future research that is aimed at understanding why the association between early formal child care and long-term outcomes differs for boys and girls is needed. Potential explanations include differences in the socialization process of boys and girls,5557  with boys benefiting more from early socialization with same-aged peers.55,58,59 

This study has several strengths. First, we used a large population-based sample from the French-language public school system in Quebec, Canada. Second, we used 2 independent government administrative databases for long-term 30-year follow-up. Third, we examined 2 outcomes at different developmental time points: (1) high school graduation, a key path to entry into the workforce, and (2) employment earnings during adulthood, an important indicator of economic status and well-being. Fourth, our results were stratified by sex. Fifth, we used a robust method to address social selection bias, namely IPTW (in this case, exposure to regulated child care services) by using propensity scores. Finally, our analyses included timing of onset and type of child care, which are important for several reasons: (1) the putative combined effects on child development and total time spent in child care and (2) the impact on parental participation in the workforce and the implications thereof on social policy.

We also note some limitations. This was an observational study; causal inferences are therefore not possible. After propensity scoring, the groups were made as equivalent to each other as possible with respect to observed baseline covariates. Even if our findings were consistent with previous experimental studies, they may not be generalizable to contexts with different child care systems or to populations with different cultural and ethnic backgrounds. Studies in other contexts and populations are needed to test the generalizability of results. No differentiation between levels of child care quality and educator’ level of training is also a limitation that should be addressed in future large-scale population-based child care studies.

To our knowledge, this study provides the first evidence of an association between early child care, increased rates of high school graduation, and lower risks of poverty in young adulthood, particularly for boys. The findings suggest that benefits of formal child care may extend well beyond school readiness and parental participation in the workforce, including a reduction in high school drop-out rates for boys, for long-term socioeconomic gain.

We thank all participating families for their continued commitment to the study. We thank Helene Beauchesne and Lucille David for supervision of data collection; Lyse Desmarais-Gervais, Pierre McDuff, Muriel Rorive, Hélène Beaumont, and Qian Xu for management of the data bank; and Danielle Buch, medical writer and researcher, for critical revision and substantive editing of the manuscript.

Dr Domond contributed to the conception and design of the study, obtained funding, conducted all analyses, interpreted the results, drafted the manuscript, and revised the manuscript critically for important intellectual content; Dr Orri assisted with the analyses and interpretation of results and revised the manuscript critically for important intellectual content; Dr Algan contributed to the conception and design of the study and acquisition of data, obtained funding for the study, and reviewed the manuscript; Drs Kohen and Findlay contributed to the acquisition of data and revised the manuscript critically for important intellectual content; Dr Vitaro contributed to the conception and design of the study, designed the data collection instruments, obtained funding for the study, and revised analyses, interpretation of the results, and the manuscript critically for important intellectual content; Dr Tremblay contributed to the conception and design of the study and acquisition of data, designed the data collection instruments, obtained funding for the study, and revised the interpretation of results and the manuscript critically for important intellectual content; Dr Côté contributed to the conception and design of the study and acquisition of data, obtained funding for the study, and critically revised the analyses, interpretation of results, and the manuscript for important intellectual content; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: Supported by the Canadian Institutes for Health Research, Canada’s Social Science and Humanities Research Council, Statistics Canada, and the Department for Woman and Gender Equality funding agencies. Dr Domond is funded by the Department for Woman and Gender Equality. Dr Côté received funding from the Initiative of Excellence, University of Bordeaux via the Origin Chair. The sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

COMPANION PAPER: A companion to this article can be found online at www.pediatrics.org/cgi/doi/10.1542/peds.2020-0483.

     
  • AIC

    Akaike information criterion

  •  
  • BIC

    Bayesian information criterion

  •  
  • CI

    confidence interval

  •  
  • IPTW

    inverse probability of treatment weighting

  •  
  • OR

    odds ratio

  •  
  • QLSKC

    Quebec Longitudinal Study of Kindergarten Children

1
Tremblay
RE
,
Vitaro
F
,
Côté
SM
.
Developmental origins of chronic physical aggression: a bio-psycho-social model for the next generation of preventive interventions
.
Annu Rev Psychol
.
2018
;
69
:
383
407
2
Poulton
R
,
Caspi
A
,
Milne
BJ
, et al
.
Association between children’s experience of socioeconomic disadvantage and adult health: a life-course study
.
Lancet
.
2002
;
360
(
9346
):
1640
1645
3
Duncan
GJ
,
Ziol-Guest
KM
,
Kalil
A
.
Early-childhood poverty and adult attainment, behavior, and health
.
Child Dev
.
2010
;
81
(
1
):
306
325
4
Caspi
A
,
Bradley
REW
,
Moffitt
TE
,
Silva
PA
.
Early failure in the labor market: childhood and adolescent predictors of unemployment in the transition to adulthood
.
Am Sociol Rev
.
1998
;
63
(
3
):
424
451
5
Heckman
JJ
.
Skill formation and the economics of investing in disadvantaged children
.
Science
.
2006
;
312
(
5782
):
1900
1902
6
Jaffee
SR
,
Ambler
A
,
Merrick
M
, et al
.
Childhood maltreatment predicts poor economic and educational outcomes in the transition to adulthood
.
Am J Public Health
.
2018
;
108
(
9
):
1142
1147
7
Hillemeier
MM
,
Morgan
PL
,
Farkas
G
,
Maczuga
SA
.
Perinatal and socioeconomic risk factors for variable and persistent cognitive delay at 24 and 48 months of age in a national sample
.
Matern Child Health J
.
2011
;
15
(
7
):
1001
1010
8
Walker
D
,
Greenwood
C
,
Hart
B
,
Carta
J
.
Prediction of school outcomes based on early language production and socioeconomic factors
.
Child Dev
.
1994
;
65
(
2 Spec No
):
606
621
9
Goodman
SH
,
Rouse
MH
,
Connell
AM
,
Broth
MR
,
Hall
CM
,
Heyward
D
.
Maternal depression and child psychopathology: a meta-analytic review
.
Clin Child Fam Psychol Rev
.
2011
;
14
(
1
):
1
27
10
Hart
B
,
Risley
TR
.
Meaningful Differences in the Everyday Experience of Young American Children
.
Baltimore, MD
:
Paul H. Brookes Publishing Co
;
1995
11
Chetty
R
,
Hendren
N
,
Katz
LF
.
The effects of exposure to better neighborhoods on children: new evidence from the moving to opportunity experiment
.
Am Econ Rev
.
2016
;
106
(
4
):
855
902
12
Reynolds
AJ
,
Ou
SR
,
Temple
JA
.
A multicomponent, preschool to third grade preventive intervention and educational attainment at 35 years of age
.
JAMA Pediatr
.
2018
;
172
(
3
):
247
256
13
Reynolds
AJ
,
Temple
JA
,
Ou
SR
,
Arteaga
IA
,
White
BA
.
School-based early childhood education and age—28 well-being: effects by timing, dosage, and subgroups
.
Science
.
2011
;
333
(
6040
):
360
364
14
Reynolds
AJ
,
Temple
JA
,
Ou
SR
, et al
.
Effects of a school-based, early childhood intervention on adult health and well-being: a 19-year follow-up of low-income families
.
Arch Pediatr Adolesc Med
.
2007
;
161
(
8
):
730
739
15
Campbell
F
,
Conti
G
,
Heckman
JJ
, et al
.
Early childhood investments substantially boost adult health
.
Science
.
2014
;
343
(
6178
):
1478
1485
16
Côté
SM
,
Doyle
O
,
Petitclerc
A
,
Timmins
L
.
Child care in infancy and cognitive performance until middle childhood in the millennium cohort study
.
Child Dev
.
2013
;
84
(
4
):
1191
1208
17
Côté
SM
,
Mongeau
C
,
Japel
C
,
Xu
Q
,
Séguin
JR
,
Tremblay
RE
.
Child care quality and cognitive development: trajectories leading to better preacademic skills
.
Child Dev
.
2013
;
84
(
2
):
752
766
18
Geoffroy
MC
,
Côté
SM
,
Giguère
CE
, et al
.
Closing the gap in academic readiness and achievement: the role of early childcare
.
J Child Psychol Psychiatry
.
2010
;
51
(
12
):
1359
1367
19
Côté
SM
,
Geoffroy
MC
,
Pingault
JB
. Early Child Care Experiences and School Readiness. In:
Boivin
M
,
Bierman
KL
, eds.
Promoting School Readiness and Early Learning: Implications of Developmental Research for Practice
.
New York, NY
:
Guilford Press
;
2014
:
133
162
20
Vandell
DL
,
Belsky
J
,
Burchinal
M
,
Steinberg
L
,
Vandergrift
N
;
NICHD Early Child Care Research Network
.
Do effects of early child care extend to age 15 years? Results from the NICHD study of early child care and youth development
.
Child Dev
.
2010
;
81
(
3
):
737
756
21
Vandell
DL
,
Lee
KTH
,
Whitaker
AA
,
Pierce
KM
.
Cumulative and differential effects of early child care and middle childhood out-of-school time on adolescent functioning
.
Child Dev
.
2020
;
91
(
1
):
129
144
22
Vandell
DL
,
Burchinal
M
,
Pierce
KM
.
Early child care and adolescent functioning at the end of high school: results from the NICHD study of early child care and youth development
.
Dev Psychol
.
2016
;
52
(
10
):
1634
1645
23
Pingault
JB
,
Tremblay
RE
,
Vitaro
F
,
Japel
C
,
Boivin
M
,
Côté
SM
.
Early nonparental care and social behavior in elementary school: support for a social group Adaptation hypothesis
.
Child Dev
.
2015
;
86
(
5
):
1469
1488
24
Laurin
JC
,
Geoffroy
MC
,
Boivin
M
, et al
.
Child care services, socioeconomic inequalities, and academic performance
.
Pediatrics
.
2015
;
136
(
6
):
1112
1124
25
Bailey
D
,
Duncan
GJ
,
Odgers
CL
,
Yu
W
.
Persistence and fadeout in the impacts of child and adolescent interventions
.
J Res Educ Eff
.
2017
;
10
(
1
):
7
39
26
Deming
D
.
Early childhood intervention and life-cycle skill development: evidence from Head Start
.
Am Econ J Appl Econ
.
2009
;
1
(
3
):
111
134
27
Li
W
,
Farkas
G
,
Duncan
GJ
,
Burchinal
MR
,
Vandell
DL
.
Timing of high-quality child care and cognitive, language, and preacademic development
.
Dev Psychol
.
2013
;
49
(
8
):
1440
1451
28
Geoffroy
MC
,
Côté
SM
,
Borge
AI
,
Larouche
F
,
Séguin
JR
,
Rutter
M
.
Association between nonmaternal care in the first year of life and children’s receptive language skills prior to school entry: the moderating role of socioeconomic status
.
J Child Psychol Psychiatry
.
2007
;
48
(
5
):
490
497
29
Loeb
S
,
Bridges
M
,
Bassok
D
,
Fuller
B
,
Rumberger
RW
.
How much is too much? The influence of preschool centers on children’s social and cognitive development
.
Econ Educ Rev
.
2007
;
26
(
1
):
52
66
30
Brooks-Gunn
J
,
Han
WJ
,
Waldfogel
J
;
National Institute of Child Health and Human Development
.
Maternal employment and child cognitive outcomes in the first three years of life: the NICHD study of early child care
.
Child Dev
.
2002
;
73
(
4
):
1052
1072
31
Anderson
ML
.
Multiple inference and gender differences in the effects of early intervention: a reevaluation of the abecedarian, perry preschool, and early training projects
.
J Am Stat Assoc
.
2008
;
103
(
484
):
1481
1495
32
Campbell
FA
,
Pungello
EP
,
Burchinal
M
, et al
.
Adult outcomes as a function of an early childhood educational program: an Abecedarian Project follow-up
.
Dev Psychol
.
2012
;
48
(
4
):
1033
1043
33
Rouquette
A
,
Côté
SM
,
Pryor
LE
,
Carbonneau
R
,
Vitaro
F
,
Tremblay
RE
.
Cohort profile: the Quebec Longitudinal Study of Kindergarten Children (QLSKC)
.
Int J Epidemiol
.
2014
;
43
(
1
):
23
33
34
Findlay
L
,
Beasley
E
,
Park
J
, et al
.
Longitudinal child data: what can be gained by linking administrative data and cohort data?
Int J Popul Data Sci
.
2018
;
3
(
1
). doi:
35
Brookhart
MA
,
Schneeweiss
S
,
Rothman
KJ
,
Glynn
RJ
,
Avorn
J
,
Stürmer
T
.
Variable selection for propensity score models
.
Am J Epidemiol
.
2006
;
163
(
12
):
1149
1156
36
Nagin
DS
.
Analyzing developmental trajectories: a semiparametric, group-based approach
.
Psychol Methods
.
1999
;
4
(
2
):
139
157
37
Nagin
DS
.
Group-Based Modeling of Development
.
Cambridge, MA
:
Harvard University Press
;
2005
38
Van Buuren
S
,
Brand
JPL
,
Groothuis-Oudshoorn
CGM
,
Rubin
DB
.
Fully conditional specification in multivariate imputation
.
J Stat Comput Simul
.
2006
;
76
(
12
):
1049
1064
39
Austin
PC
,
Stuart
EA
.
Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies
.
Stat Med
.
2015
;
34
(
28
):
3661
3679
40
Penning de Vries
BBL
,
Groenwold
RHH
.
A comparison of approaches to implementing propensity score methods following multiple imputation
.
Epidemiol Biostat Public Health
.
2017
;
14
(
4
):
12630
41
Pingault
JB
,
Tremblay
RE
,
Vitaro
F
, et al
.
Childhood trajectories of inattention and hyperactivity and prediction of educational attainment in early adulthood: a 16-year longitudinal population-based study
.
Am J Psychiatry
.
2011
;
168
(
11
):
1164
1170
42
Vergunst
F
,
Tremblay
RE
,
Nagin
D
, et al
.
Association of behavior in boys from low socioeconomic neighborhoods with employment earnings in adulthood
.
JAMA Pediatr
.
2019
;
173
(
4
):
334
341
43
Moffitt
TE
,
Arseneault
L
,
Belsky
D
, et al
.
A gradient of childhood self-control predicts health, wealth, and public safety
.
Proc Natl Acad Sci USA
.
2011
;
108
(
7
):
2693
2698
44
Furnham
A
,
Cheng
H
.
Factors influencing adult earnings: findings from a nationally representative sample
.
J Socio Econ
.
2013
;
44
:
120
125
45
Fergusson
DM
,
Horwood
LJ
,
Ridder
EM
.
Show me the child at seven II: childhood intelligence and later outcomes in adolescence and young adulthood
.
J Child Psychol Psychiatry
.
2005
;
46
(
8
):
850
858
46
Harder
VS
,
Stuart
EA
,
Anthony
JC
.
Propensity score techniques and the assessment of measured covariate balance to test causal associations in psychological research
.
Psychol Methods
.
2010
;
15
(
3
):
234
249
47
Belsky
J
.
Classroom composition, childcare history and social development: are childcare effects disappearing or spreading?
Soc Dev
.
2009
;
18
(
1
):
230
238
48
Krueger
PM
,
Tran
MK
,
Hummer
RA
,
Chang
VW
.
Mortality attributable to low levels of education in the United States
.
PLoS One
.
2015
;
10
(
7
):
e0131809
49
Krohn
MD
,
Thornberry
TP
,
Collins-Hall
L
,
Lizotte
AJ
. School Dropout, Delinquent Behavior, and Drug Use: An Examination of the Causes and Consequences of Dropping out of School. In:
Kaplan
HB
, ed.
Drugs, Crime, and Other Deviant Adaptations: Longitudinal Studies
.
Boston, MA
:
Springer
;
1995
:
163
183
50
Schweinhart
LJ
,
Montie
J
,
Xiang
Z
,
Barnett
WS
,
Belfield
CR
,
Nores
M
.
Lifetime Effects: The High/Scope Perry Preschool Study Through Age 40
.
Ypsilanti, MI
:
High/Scope Press
;
2005
51
Magnuson
KA
,
Kelchen
R
,
Duncan
GJ
,
Schindler
HS
,
Shager
H
,
Yoshikawa
H
.
Do the effects of early childhood education programs differ by gender? A meta-analysis
.
Early Child Res Q
.
2016
;
36
:
521
536
52
Garces
E
,
Thomas
D
,
Currie
J
.
Longer-term effects of Head Start
.
Am Econ Rev
.
2002
;
92
(
4
):
999
1012
53
Peisner-Feinberg
ES
,
Burchinal
MR
,
Clifford
RM
, et al
.
The relation of preschool child-care quality to children’s cognitive and social developmental trajectories through second grade
.
Child Dev
.
2001
;
72
(
5
):
1534
1553
54
Orri
M
,
Tremblay
RE
,
Japel
C
, et al
.
Early childhood child care and disruptive behavior problems during adolescence: a 17-year population-based propensity score study
.
J Child Psychol Psychiatry
.
2019
;
60
(
11
):
1174
1182
55
Rose
AJ
,
Rudolph
KD
.
A review of sex differences in peer relationship processes: potential trade-offs for the emotional and behavioral development of girls and boys
.
Psychol Bull
.
2006
;
132
(
1
):
98
131
56
Diprete
TA
,
Jennings
JL
.
Social and behavioral skills and the gender gap in early educational achievement
.
Soc Sci Res
.
2012
;
41
(
1
):
1
15
57
Tremblay
RE
,
Côté
SM
.
Sex differences in the development of physical aggression: an intergenerational perspective and implications for preventive interventions
.
Infant Ment Health J
.
2019
;
40
(
1
):
129
140
58
Benenson
JF
,
Apostoleris
NH
,
Parnass
J
.
Age and sex differences in dyadic and group interaction
.
Dev Psychol
.
1997
;
33
(
3
):
538
543
59
Benenson
JF
,
Morganstein
T
,
Roy
R
.
Sex differences in children’s investment in peers
.
Hum Nat
.
1998
;
9
(
4
):
369
390

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data