Examining the relationships between early childhood experiences and adolescent and young adult health status in a resource-limited population: A cohort study.

BACKGROUND: Adolescence is a critical point in the realization of human capital, as health and educational decisions with long-term impacts are made. We examined the role of early childhood experiences on health, cognitive abilities, and educational outcomes of adolescents followed up from a longitudinal cohort study in Pakistan, hypothesizing that early childhood experiences reflecting poverty would manifest in reduced health and development in adolescence. METHODS AND
FINDINGS: Adolescents/young adults previously followed as children aged under 5 years were interviewed. Childhood data were available on diarrhea, pneumonia, and parental/household characteristics. New data were collected on health, anthropometry, education, employment, and languages spoken; nonverbal reasoning was assessed. A multivariable Bayesian network was constructed to explore structural relationships between variables. Of 1,868 children originally enrolled, 1,463 (78.3%) were interviewed as adolescents (range 16.0-29.3 years, mean age 22.6 years); 945 (65%) lived in Oshikhandass. While 1,031 (70.5%) of their mothers and 440 (30.1%) of their fathers had received no formal education, adolescents reported a mean of 11.1 years of education. Childhood diarrhea (calculated as episodes/child-year) had no association with nonverbal reasoning score (an arc was supported in just 4.6% of bootstrap samples), health measures (with BMI, 1% of bootstrap samples; systolic and diastolic blood pressure, 0.1% and 1.6% of bootstrap samples, respectively), education (0.7% of bootstrap samples), or employment (0% of bootstrap samples). Relationships were found between nonverbal reasoning and adolescent height (arc supported in 63% of bootstrap samples), age (84%), educational attainment (100%), and speaking English (100%); speaking English was linked to the childhood home environment, mediated through maternal education and primary language. Speaking English (n = 390, 26.7% of adolescents) was associated with education (100% of bootstrap samples), self-reported child health (82%), current location (85%) and variables describing childhood socioeconomic status. The main limitations of this study were the lack of parental data to characterize the home setting (including parental mental and physical health, and female empowerment) and reliance on self-reporting of health status.
CONCLUSIONS: In this population, investments in education, especially for females, are associated with an increase in human capital. Against the backdrop of substantial societal change, with the exception of a small and indirect association between childhood malnutrition and cognitive scores, educational opportunities and cultural language groups have stronger associations with aspects of human capital than childhood morbidity.

Introduction

Human capital is the knowledge, skills, and health that individuals accumulate over time that enables them to realize their potential as productive members of society [1]. The World Bank created in 2018 a population-level indicator of human capital by indexing child survival and stunting, school attendance and performance, and adult survival. Although essentially an economic construct, its goal is to capture shortfalls in the realization of human development. The Human Capital Index (HCI) describes how improvements to childhood health, nutrition, and early learning lay strong foundations for the future acquisition of cognitive, social, and behavioral skills [2]. As such, the adolescent period is a critical stage between childhood and adulthood where formal education ends and key health decisions are made that have long-term impacts [3,4].

Pakistan ranks 134 of 157 countries in the 2018 World Bank HCI [1]. Located in remote, mountainous northeast Pakistan, Gilgit-Baltistan (G-B) in the 1980s was one of the poorest regions of the country, with a per capita income less than half that of the rest of Pakistan [5], with subsistence farming as the primary occupation, low educational levels, low life expectancy (53 years), and high maternal mortality [6]. Considerable investment in rural support, education, and health changed this situation after the 1978 opening of the Karakoram Highway, which provided increased accessibility to the area, leading to substantial societal change [5]. Communities in G-B were strongly motivated to invest in children’s education for future job opportunities and income [7]. The most recent evaluation in 2010 showed a dramatic secular transformation, resulting in a narrowing of the income gap, to a per capita income of about 90% of the national average [8].

We report the findings of an observational longitudinal study of a cohort of children who were followed up 15–20 years later, when they were adolescents and young adults (hereafter referred to as adolescents) [4]. The childhood cohort came from Oshikhandass, an ethnically mixed, remote rural village with high morbidity and mortality [9], 20 km from the capital of G-B. This study examines individual-level associations between childhood health and family socioeconomic indicators of early life adversity and adolescent health and development. Our study considers the following as outcomes of interest: nonverbal reasoning score, self-reported health status, body mass index (BMI), blood pressure, educational attainment, and employment. We hypothesized that early life experiences reflecting poverty would manifest in reduced health and development in adolescence, which are the foundation of adult human capital. [10].

Methods

This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines (see S1 Appendix) [11]. Prospective protocols (see S2 and S5 Appendices) were used for the collection of data but not for analysis.

Ethics approval and consent to participate

Ethical approval was granted for the 1989–1996 study by the Aga Khan University Human Subjects Protection Committee (15 November 1989 for childhood diarrhea; 3 November 1993 for pneumonia); parents/legal guardians of children provided oral informed consent to participate, given high levels of parental illiteracy. Ethical approval for the adolescent study was granted by the US National Institutes of Health Institutional Review Board (#20TWN071, initially approved by the National Institute of Child Health and Development’s institutional review board), the Aga Khan University Ethics Review Committee (1966-CHS-ERC-11), and Karakoram International University Ethics Review Committee. For the adolescent study, all participants aged 18 years or over provided signed consent to participate; participants under age 18 years provided assent, and their parents provided written consent for their participation.

Data collection for children under 5 years, 1989–1996

From 1989 to 1996, all children under age 5 years in Oshikhandass were enrolled with parental consent in the prospective cohort of the Oshikhandass Diarrhea and Pneumonia Project. Local health workers visited families weekly until the child’s fifth birthday, out-migration, or death; data collection methods and variable definitions are described elsewhere [9]. These data were used to estimate the burden of child morbidity (number of diarrhea and pneumonia episodes divided by days of follow-up). Pneumonia was not included in the main analyses because of a smaller sample size [9]. The mean length/height-for-age z score (HAZ) was calculated for each child from their available quarterly data, using 2006 WHO growth standards. At enrollment, a survey was conducted of household characteristics (e.g., house type based on construction materials, number of rooms and occupants, and type of water and sanitation) and parental educational attainment (illiterate/no formal education, primary to 10 years, or more), occupation, and income [9].

Data collection for follow-up in adolescence and young adulthood, 2011–2014

In July 2011, a village census identified participants from the original cohort who still lived in Oshikhandass, had migrated elsewhere, or died. Cohort members were contacted through their families. After obtaining written informed consent from adolescents, interviews were conducted in Urdu and, rarely, in the participant’s mother tongue (usually Burushaski or Shina), by study staff in person at the study office, at the participant’s residence, or by telephone (<1%).

A questionnaire (see S3 Appendix) was used to collect data on current and past health status. The questionnaire was specifically designed for this study, considering important domains identified in the literature and in consultation with local subject specialists. The questionnaire was translated into Urdu, back-translated, and piloted in a neighboring village. Self-reported current health (SRCH) and self-reported past childhood health (SRPH) (between ages 5 and 15 years) were characterized using a 5-point scale (very poor to excellent), and other lifetime health problems were recorded as “any” versus “none.” The questionnaire recorded educational attainment (years of schooling), whether classes were repeated, employment (student, employed, or not employed), marital status, number of children, current geographic location, age when moved, and languages spoken. Primary language (i.e., mother tongue) was defined as Burushaski or Shina; almost all respondents spoke Urdu, the national language.

The Raven’s Standard Progressive Matrices and Colored Progressive Matrices (PsychCorp) [12] were administered (see S5 Appendix) in a quiet location without distractions to assess nonverbal reasoning skills, which are a critical component of cognitive ability [13]. The Raven’s matrices are widely used internationally because they have minimal language requirements, thus minimizing the impact of cultural and linguistic differences [12]. Scores were calculated by adding the number of correct items and then scaled into a T score (mean 50, standard deviation 10). Scores were analyzed using exploratory and confirmatory factor analytic approaches to ensure construct validity [14] (see S4 Appendix).

Clinical measurements, taken once each by trained master’s level field coordinators, included participant weight (Seca 872 digital scales) without shoes and in light clothing, height (HM200P PortStad Portable Stadiometer, Charder), and abdominal girth (tape measure). BMI was calculated, analyzed as a continuous variable, and for context summarized as underweight (<18.5 kg/m2), normal (18.5–24.9 kg/m2), overweight (25.0–29.9 kg/m2), and obese (≥30 kg/m2); pregnant women were excluded. Blood pressure (in mm Hg) was measured once while the rested participant was sitting cross-legged on the floor using a mercury sphygmomanometer (Yamasu Model 600, Japan) and defined as normal except if elevated (systolic blood pressure 121–130 mm Hg and diastolic blood pressure ≤ 80 mm Hg), stage 1 hypertension (systolic blood pressure 131–140 mm Hg or diastolic blood pressure 81–90 mm Hg), or stage 2 hypertension (systolic blood pressure > 140 mm Hg or diastolic blood pressure > 90 mm Hg). Missing data were excluded from the analysis described below; for example, clinical measurements and Raven’s matrices were not available for adolescents interviewed by telephone.

Data analysis

Univariate linear models were constructed using the Raven’s score (squared to approximate normality) as the outcome. Subsequently, a multivariate Bayesian network (BN, using the R package bnlearn) [15] was constructed to explore the structural relationships between variables; BN analysis determines an optimal statistical description of all the data as it considers many possible model structures [16]. The best-fitting structure to describe the data was sought using a tabu greedy search [17], assuming either linear regression for continuous nodes or conditional probabilities for categorical (including binary) nodes, and a ban list to exclude directed arcs from any variable to the age and sex of the individual (biologically implausible) or from descriptions of their current health, education, or employment status to the variables characterizing their childhood environment (temporally impossible). The single best-fitting model, optimizing the Bayesian information criterion, was bootstrapped (7,500 times), and only arcs present in 50% or more of samples were retained. To estimate confidence intervals for the primary outcomes, regression models were separately constructed using subnetworks identified from the BN.

All analyses were conducted in R (April 2018, release 3.5.0).

Results

Of 1,868 children enrolled in the initial study, 1,463 (78.3%) were interviewed as adolescents (Fig 1). Almost two-thirds (n = 945) were still living in Oshikhandass, 35.1% (347) elsewhere within Pakistan, and 0.3% (5) abroad. Seventy-nine percent had complete information to build the BN (n = 1,165). Compared to the 1,462 individuals who were re-enrolled as adolescents, individuals lost to follow-up (n = 381) tended to be male (57.2% versus 51.0%, p = 0.017, chi-squared test) and to have more diarrheal episodes per child-year (mean [95% CI] 0.54 [0.00, 1.42] versus 0.42 [0.00, 1.02], p = 0.017, Kruskal–Wallis test), less improved household construction (62.2% improved or somewhat improved versus 65.7%, p = 0.002, chi-squared test), and more likely to have improved toilets (3.9% versus 2.3%, p = 0.032, chi-squared test) (see S1 Table). Of 126 (6.8%) children with follow-up who had died, 95 died during the first study and 31 died between studies. Those who did not complete the Raven’s matrices (n = 60/1,462, 4.1%, most because they reported a lack of time), compared to those who did complete the Raven’s matrices, were more likely to live away from Oshikhandass (63.3% versus 34.2%, p < 0.001, chi-squared test), were older (mean 26.7 versus 22.2 years old, p < 0.001, Kruskal–Wallis test), had a higher BMI (mean 23.0 versus 21.7 kg/m2, p = 0.002, t-test), had fewer years of education (mean 7.9 versus 11.2 years, p < 0.001, t-test), and were less likely to speak English (16.7% versus 27.1%, p = 0.10, chi-squared test).

Fig 1

Flowchart of participant follow-up.

Selected cohort characteristics are shown in Table 1. Males outnumbered females in both the original and adolescent cohorts, despite more male deaths (71 male versus 55 female). Mean age at follow-up was 22.6 years (range 16.0–29.3) based on recorded birthdate. Only 14.4% were married; however, this percentage was higher among the female participants, who were married longer (mean 3.8 versus 2.2 years, p < 0.001, t-test), were married at a younger age (mean 21.6 versus 24.7 years, p < 0.001, t-test), and had more children (mean 1.8 versus 1.3, p = 0.01, t-test). At the time of interview, 54 women (7.5%) were pregnant. Females were more likely to remain in Oshikhandass, whereas males were more likely to move for education (30.6% versus 9.6%, p < 0.001, test of proportions) or employment (8.4% versus 0.8%, p < 0.001, test of proportions). In total, 1,240 (67.3%) adolescents had at least 1 childhood HAZ observation (median 8.0, IQR 9), with a population mean HAZ of −2.2, but the mean adolescent HAZ (assuming an age of 19 years, since WHO growth reference standards for calculating age- and sex-standardized HAZ only go up to this age) was −1 (see S1 Fig). Childhood HAZ was positively correlated with adolescent height (Pearson’s rho 0.15 for females and 0.24 for males). Along with Urdu, almost all adolescents spoke Shina (1,414, 96.7%), the language most widely spoken in Gilgit, and about two-thirds also spoke Burushaski, the mother tongue of those originally from Hunza. A quarter (367, 25.1%) of adolescents spoke ≥4 languages, and the fourth most common language was English, spoken more often by males. In our population, based on a linear regression adjusting for age and sex, speaking English was associated with having approximately 2.4 more years of education (95% CI 2.1 to 2.7, p < 0.001); all English speakers had ≥8 years of education.

Table 1

Selected descriptive characteristics of the adolescent cohort.

Characteristic	Total	Male	Female	p-Value
Number of participants	1,463	746 (51.0%); 746 (52.9%)*	717 (49.0%); 663 (47.1%)*	0.3a
Recorded age (years)	22.6 (3.5)	22.5 (3.4)	22.6 (3.6)	0.8b
Height (cm)	162.9 (9.5)	169.5 (7.0)	156.0 (6.4)	<0.001b
Weight (kg)*	57.6 (9.8)	61.5 (9.5)	52.9 (8.1)*	<0.001b
Waist girth (cm)*	78.2 (9.6)	79.8 (8.5)	77.3 (10.1)*	<0.001b
BMI (kg/m2)*	21.7 (3.3)	21.4 (3.1)	21.8 (3.3)*	0.02b
BMI category*
Underweight (BMI < 18.5 kg/m2)	198 (13.5%)	107 (14.3%)	89 (13.4%)*	0.3c
Normal BMI (BMI 18.5–24.9 kg/m2)	1,065 (72.8%)	555 (74.4%)	477 (71.9%)*
Overweight (BMI 25.0–29.9 kg/m2)	171 (11.7%)	74 (9.9%)	83 (12.5%)*
Obese (BMI ≥ 30 kg/m2)	29 (2.0%)	10 (1.3%)	14 (2.1%)*
Systolic BP (mm Hg)	113.2 (10.2)	115.2 (9.5)	111.1 (10.4)	<0.001b
Diastolic BP (mm Hg)	75.8 (8.5)	77.0 (8.4)	74.8 (8.4)	<0.001b
BP category**
Normal BP	1,234 (84.4%)	605 (81.1%)	629 (87.8%)	0.002c
Elevated BP	42 (2.9%)	26 (3.5%)	16 (2.2%)
Stage 1 hypertension	157 (10.7%)	93 (12.5%)	64 (8.9%)
Stage 2 hypertension	29 (2.0%)	22 (2.9%)	7 (1.0%)
Married	211 (14.4%)	41 (5.5%)	170 (23.7%)	<0.001a
Raven’s score (T score)	50.0 (10.0)	51.3 (10.0)	48.6 (9.8)	<0.001b
Currently in Oshikhandass	945 (64.6%)	428 (57.4%)	517 (72.1%)	<0.001a
Student***	984 (67.3%)	515 (69.0%)	469 (65.4%)	0.2a
Employed***	269 (18.4%)	223 (29.9%)	73 (10.2%)	<0.001a
Highest level of education (years)	11.1 (2.8)	11.0 (2.7)	11.2 (2.9)	0.3b
Speaks English	390 (26.7%)	242 (32.4%)	148 (20.6%)	<0.001a
Speaks Burushaski	1,010 (69.0%)	528 (70.8%)	482 (67.2%)	0.2a
Speaks Shina	1,414 (96.7%)	729 (97.7%)	685 (95.5%)	0.03a
Mother illiterate	1,031 (70.5%)	524 (70.5%)	507 (71.2%)	>0.999a
Father illiterate	440 (30.1%)	236 (31.9%)	204 (28.8%)	0.75a
SRCH				<0.001c
Excellent	138 (9.4%)	69 (9.2%)	69 (9.6%)
Good	294 (20.1%)	153 (20.5%)	141 (19.7%)
Satisfactory	766 (52.4%)	427 (57.2%)	339 (47.3%)
Poor	262 (17.9%)	96 (12.9%)	166 (23.2%)
Very poor	3 (0.2%)	1 (0.1%)	2 (0.3%)
SRPH				0.09c
Excellent	117 (8.0%)	57 (7.6%)	60 (8.4%)
Good	388 (26.5%)	194 (26.0%)	194 (27.1%)
Satisfactory	561 (38.3%)	308 (41.3%)	253 (35.3%)
Poor	386 (26.4%)	184 (24.7%)	202 (28.2%)
Very poor	11 (0.8%)	3 (0.4%)	8 (1.1%)

Data are mean (SD) or n (%). BMI, body mass index; BP, blood pressure; SRCH, self-reported current health; SRPH, self-reported past childhood health.

*Fifty-four females were pregnant at the time of the interview. These individuals are not included in the weight, waist girth, and BMI measurements.

**Normal except if elevated (systolic blood pressure 121–130 mm Hg and diastolic blood pressure ≤ 80 mm Hg), stage 1 hypertension (systolic blood pressure 131–140 mm Hg or diastolic blood pressure 81–90 mm Hg), or stage 2 hypertension (systolic blood pressure > 140 mm Hg or diastolic blood pressure > 90 mm Hg).

***Categories are not mutually exclusive.

aTest of proportions.

bt-test for normally distributed data.

cChi-squared test for categorical variables.

Nonverbal reasoning score using the Raven’s matrices

In univariate linear regressions (see S2 Table), the Raven’s score (squared T score) was significantly (p ≤ 0.05) negatively related to childhood household density (−160 [95% CI −220, −91], p < 0.001), inferior household (−400 [95% CI −530, −260], p < 0.001) or toilet construction (−650 [95% CI −980, −320], p = 0.003), lower status paternal occupation (−420 [95% CI −550, −280], p < 0.001), whether they were a Shina speaker (−440 [95% CI −550, −340], p < 0.001), and reported worse health (SRCH, −440 [95% CI −770, −370], p < 0.001; SRPH, −660 [95% CI −850, −460], p < 0.001); scores were positively related to being male (280 [95% CI 180, 370], p < 0.001), height (21 [95% CI 16, 26], p < 0.001), weight (13 [95% CI 7.6, 18], p < 0.001), waist girth (5.2 [95% CI 0.11, 10], p = 0.045), moving and living away from Oshikhandass (730 [95% CI 560, 900], p < 0.001), age at moving (408 [95% CI 249, 567], p < 0.001), whether they spoke English (850 [95% CI 750, 950], p < 0.001), being unmarried (350 [95% CI 200, 490], p < 0.001), being employed or a student (380 [95% CI 220, 540] and 550 [95% CI 410, 690], respectively, p < 0.001), having more years of education (130 [95% CI 110, 140], p < 0.001), not having repeated any classes (150 [95% CI 55, 250], p = 0.002), parents’ educational status (maternal, 870 [95% CI 580, 1,200], p < 0.001; paternal, 610 [95% CI 470, 750], p < 0.001), maternal occupation (occupation other than housewife, 710 [95% CI 520, 900], p < 0.001), and higher parental income (320 [95% CI 150, 500], p < 0.001). The number of variables related to the Raven’s score was substantially reduced in the multivariate BN, with T score directly associated with only an individual’s adolescent height, age at assessment, educational attainment, and whether or not they spoke English (Fig 2). Older participants tended to have a lower Raven’s score (−41.1 squared T score points per year [95% CI −55.8, −26.4], p < 0.001, linear regression), but taller participants and those with more years of education achieved higher scores (13.3 squared T score points per centimeter [95% CI 8.12, 18.5]; 91.2 squared T score points per year of schooling [95% CI 69.1, 113]; both p < 0.001; Fig 3).

Fig 2

Bayesian network showing associations that were robust in ≥50% of 7,500 bootstrap samples.

Continuous nodes are shown as rectangles, and categorical nodes as ellipses; childhood nodes are indicated by dashed outlines, and adolescent nodes by solid outlines. Arc width is proportional to the change in the Bayesian information criterion (BIC): Thicker arrows indicate a larger change in BIC and evidence of a more informative association. Numbers indicate the proportion of bootstrap iterations that supported a given arc. The main outcomes are highlighted (hypertension, Raven’s score, SRCH, BMI, educational level, and employment status). The direction of associations is based on the maximum likelihood. BP, blood pressure; HAZ, length/height-for-age z score; No., number of; SRCH, self-reported current health; SRPH, self-reported past childhood health.

Fig 3

Direct predictors of adolescent Raven’s T score based on the structure identified in the Bayesian network.

The Raven’s score is a function of the participants’ age, current height, their attained level of education, and whether they spoke English. The light grey 95% confidence intervals indicate the full range of each predictor, and the darker shading indicates the 95% confidence interval across the interquartile range of each respective predictor.

Speaking English was strongly positively associated with the Raven’s score as well as related to the current geographic location and SRPH. Participants who spoke English were more likely to have left Oshikhandass (63% [244/390] had moved away compared to 26% [274/1,071] of non-English speakers, p < 0.001, test of proportions) and to report excellent or good childhood health (39% [154/390] of those who spoke English versus 26% [278/1,071] of non-English speakers, p < 0.001, test of proportions). English was also related to participants’ primary language: Burushaski speakers (35%, 286/812) were more likely to speak English than Shina speakers (8%, 29/344, p < 0.001, test of proportions). Speaking English was the only variable with robust evidence to link the childhood home environment (including parental and household characteristics, mediated through maternal education and primary language) and Raven’s score. Adolescent participants who spoke English were more likely to have educated parents (41.1% versus 24.8%, p = 0.024, chi-squared test) and to have been raised in higher income households (37.0% versus 21.6%, p = 0.034, chi-squared test) and households with lower household density (2.6 versus 2.7 people/room, p = 0.013, Kruskal–Wallis test) (see S3 Table).

No relationships were identified in the BN for the number of episodes of childhood diarrhea, and no evidence emerged that either diarrhea or pneumonia was related to the major outcomes in separate regression models (see S4 and S5 Tables). In separate regressions, unemployed individuals tended to have significantly higher rates of childhood diarrhea than those employed (see S4 Table), but this association was not supported in the bootstrapped BN.

Self-reported health status

Most participants reported satisfactory, good, or excellent health for both SRPH (72.8%, 849/1,165) and SRCH (81.9%, 957/1,165). However, females were more likely than males to report poor/very poor SRCH; this difference was less pronounced in the SRPH (Table 1). The single predictor of SRCH was SRPH (Fig 2). The few individuals who reported poor/very poor SRPH were also most likely to report poor SRCH (Fig 4).

Fig 4

The mean probability of the level of self-reported current health (SRCH) as an adolescent as a function of self-reported past childhood health (SRPH).

Whiskers indicate the 95% confidence intervals. Results based on the Bayesian network. The “very poor” category was pooled with the “poor” category due to the small number of responses.

Childhood health status was itself a predictor of lifetime health problems (Fig 2), with participants who characterized their childhood health as poor having an odds ratio of reporting later health problems of 3.4 (95% CI 2.3, 5.3, p <0.001, multinomial regression; Fig 4). Lower SRPH (poor/satisfactory) was associated with repeated years of schooling.

BMI

Height, weight, and waist girth were significantly higher among males than non-pregnant females, but BMI was slightly higher among females (Table 1). Additionally, the BN showed weak support (50% of bootstrap samples) for English-speaking individuals having a higher BMI (Fig 2).

Education and employment

Participants had 11.1 years of education on average, and 984 (67.3%) were still studying. A higher proportion of males were employed (Table 1). Positive associations existed between years of education and speaking English and between education and employment (Fig 2). During childhood, most adolescents’ mothers (1,031, 70.5%), and one-third (440, 30.1%) of fathers were illiterate (Table 1).

Most students (834, 85%) were full-time; a minority (88, 8.9%) were also employed. Of 479 non-students, 283 had completed studies, of whom 208 (73%) were employed, and the remaining 196 (85% of whom, or 166, were female) were doing unpaid family work. Employed participants (n = 313) described a variety of occupations, including teaching (42, 13.4%), business (35, 11.1%), and the army (25, 8.0%), but 106 (33.8%) had higher income roles (e.g., engineering, managerial/financial work, and hotel-related work). In contrast, from the childhood cohort study, their parents described more limited roles for mothers (724/806, 89.8%, were housewives) and for fathers (farming/labor, 35%; the army, 16%; and business, 16%).

Discussion

Human capital is the capability of individuals to contribute to society, and depends, at least in part, on their health, education, employability, and cultural experiences. Adolescents are underrepresented in health research despite their unique role in representing the culmination of childhood life experiences and potentially predicting chronic illness in later life; most relevant literature has focused on the long-term consequences of early childhood malnutrition (see S6 Table). Adolescence also crucially captures the transition from child to actively contributing adult, including becoming a parent. This study provides a unique opportunity to examine the linkages between childhood and adolescent health status. We hypothesized that the participants’ early childhood experiences would affect the development and formation of human capital in this remote Pakistani village.

Underpinning individual human capital potential is nonverbal reasoning. We found that, instead of childhood illness (i.e., diarrhea and pneumonia), nonverbal reasoning was more closely related to a collection of variables that shared an association with language, opportunities for education, and the home environment (e.g., maternal education and parental income). Additionally, we found some evidence to suggest that childhood malnutrition (indicated by HAZ) had a small indirect association with nonverbal reasoning. Two meta-analyses by Sudfeld et al. [18] and Victora et al. [19] reported positive associations between HAZ and cognitive scores, using cross-sectional and longitudinal data, respectively. Like the studies reanalyzed by Victora et al. [19], we found a positive association between childhood HAZ and later height; however, we did not find evidence of an association between undernutrition and cognitive development. Our results are consistent with findings that children who recover from stunting show no difference in cognitive outcomes from those who were not stunted [20-22]. This population underwent substantial secular changes, illustrated by the region’s increases in per capita GDP, diversification of employment opportunities, and access to education. These factors may have alleviated some of the common constraints on both growth and cognitive development during an important period of development, but are also suggestive of opportunities to intervene that extend beyond early childhood [23].

In earlier analyses of a subset of this cohort (n = 107) [24], childhood morbidity was significantly associated with lower educational performance at 7 years old, consistent with the findings of other studies [25,26]. Education is a major component of the HCI. In 2019, the World Bank estimated that a Pakistani child born in 2018 can expect to complete 8.8 years of school, but only 4.8 years when adjusting for the quality of education [1]. In our population, the mean number of years of education was higher for both females and males, at 11.1, and was significantly higher than in the previous generation, where almost three-quarters of mothers and a third of fathers were illiterate or had no formal schooling. Although it is not possible to determine with these data, school performance, rather than attendance, may still be negatively impacted by childhood malnutrition and morbidity.

Childhood mortality in Oshikhandass decreased significantly from 1989 to 1996 [9], allowing families to focus more on helping surviving children grow and thrive. During this time, educational opportunities significantly improved in Oshikhandass, from there being 1 government high school for boys and 1 middle school for girls to the current situation of 9 early childhood education centers, 11 schools, and 2 colleges (up to class 12). One reason that the Burushaski speakers tended to have more years of education is that the leaders of the Ismaili community (primarily Burushaski speakers in Oshikhandass) emphasized its importance, especially for girls [7]. In Oshikhandass, there were major investments in education, especially improving opportunities for females. Notably, many adolescents in this cohort continued their education at university in Gilgit or elsewhere. Given the importance of this age in developing resilience, understanding how experiences and stresses influence development into adulthood is of great importance [27].

Our key hypothesis was that long-lasting associations between childhood health experiences and adolescent nonverbal reasoning and health status would be identified. We found no evidence for direct associations between the childhood and family factors measured and adolescent nonverbal reasoning. However, participants’ mean childhood HAZ, which is an indicator of longer-term nutritional status, was positively correlated with adolescent height (approximately 1 cm per 1 z-score change, approximately a third of the association found in other studies [19]) and thereby indirectly to Raven’s score (3 T score points per 1 cm adolescent height); this is similar to the findings of the other studies described above, but we did not find the strong associations others did. Our finding that neither childhood diarrhea nor pneumonia was associated with adolescent Raven’s score contrasts to studies of cognitive scores in early childhood that report negative impacts of nutrition, infection, and illness [28], but is consistent with findings that such early effects diminish by age 5 years [29]. Given the temporal gap between childhood illness, potential recovery from stunting, and the age at Raven’s test administration, this is perhaps not surprising [30].

Previous studies have examined how the mesoenvironment [31], including a nurturing and stimulating home environment [10,32], becomes increasingly influential in child development as a child ages. Whether or not English was spoken by the participant was identified as a variable that linked both to Raven’s score and to a suite of variables (parental education and occupation, paternal income, dwelling type, household density, type of toilet, and primary language) that characterized a participant’s childhood home environment. We interpreted speaking English as an indication of higher childhood socioeconomic status and greater educational opportunity, as participants from lower socioeconomic backgrounds were less likely to speak English; speaking English was also associated with participants’ opportunities to travel beyond Oshikhandass for either work or education.

The Raven’s matrices are not typically thought to be biased by educational background and are designed to test observation, clarity of thinking, and nonverbal reasoning skills [12]. Here, we found that participants with more years of education tended to achieve higher Raven’s scores. Other studies have similarly shown a positive correlation between educational age (up to a chronological age of 20 years) and Raven’s score that might indicate a familiarity with the type of logic assessed, distinct from familiarity with the test per se [33].

SRCH was related to the participant’s SRPH, but not to objective measures of health (e.g., BMI, anthropometry, or blood pressure). SRCH was also not related to socioeconomic variables, as might be expected based on the theory underpinning human capital [34]. However, self-reported health remains a meaningful tool and predictor of mortality [35].

BMI is one objective measurement of health with correlations to many chronic conditions that manifest in adulthood—such as diabetes, cancer, osteoarthritis, and cardiovascular diseases [36]—that influence adult survival and therefore the HCI. Blood pressure and hypertension are also objective measures of health, but here only related to weight; neither of these metrics was associated to childhood health except through indirect association with HAZ (via weight and height). Hypertension is reported to be a significant problem in northern Pakistan [37]. It is potentially related to salted tea consumption, and our finding of high rates of diastolic hypertension suggests that it will be important to monitor the prevalence of noncommunicable diseases [4,38] in the future in this area.

Childhood experiences varied by sociocultural group, indicated by the participant’s primary language. The Burushaski subpopulation, as compared to the Shina subpopulation, were more likely speak English, to have moved out of the area, to reach higher educational levels, and to have come from families that had a higher socioeconomic status. Similarly, there were notable sex differences, with males tending to have a greater opportunity for continuing their education than females, as they were less likely to have married or to have children, and more likely to have traveled away from Oshikhandass for further study and employment. Additional targeted research would be required to better understand why adolescent females in Oshikhandass reported poorer perceived health, as reported elsewhere in Pakistan [39], and to understand the status of female empowerment.

Following the same cohort from birth to adolescence/young adulthood gives a rare opportunity to understand links from early life experiences to later outcomes, but it comes with the challenge of loss to follow-up. Given that this study was built upon an earlier study, the variables collected were also constrained by what was measured previously (e.g., no cognitive, blood pressure, micronutrient, dietary diversity, or food insecurity indicators were collected during childhood, nor were more detailed parental variables such as maternal mental health). The age at enrollment and duration of initial follow-up also varied. Using the mean HAZ over available HAZ measurements was an attempt to mitigate the variable ages at observation; however, summarizing the data over the first 5 years of life obscures considerable development that takes place in early childhood. Additionally, as the original participants were followed up later, we used retrospective questions to ask about their perceived childhood health at the same time as their perceived current health status. This may lead to biased responses, especially recalling events over several years, and is acknowledged as a subjective rather than objective assessment of health, although we did include other clinical assessments of general health. Another limitation is measurement of school attendance, but not school performance, as has been used elsewhere [26], including in the HCI.

This study does have several strengths; principally, as a prospective longitudinal study spanning over 20 years, it adds significantly to the body of research examining the effects of childhood health and early life exposures on adolescent health status. Additionally, this study examined not only physiological health, but also reasoning abilities and self-perceptions of health.

In summary, we found that adolescents raised in a remote rural village in northeastern Pakistan had high rates of education—significantly higher than their parents—and that most were still students. Early childhood illness (diarrhea or pneumonia) was unrelated to Raven’s score, current self-reported health status, BMI, hypertension, education, or employment. The primary language spoken did link childhood socioeconomic factors, characterized by maternal education, to adolescent outcomes. Sex differences were already apparent in these adolescent outcomes. We present evidence that there are opportunities to improve human capital beyond early childhood, and that with targeted investments such gains can be made even over a single generation.

The distributions of HAZ from childhood measurements and approximated for adolescents.

Childhood measurements (dashed lines) and approximations for adolescents (solid lines). The childhood distribution is the mean value of each child’s available measurements. Adolescents who were older than 19 years when their height was measured were assumed to be 19 years old because this is maximum age in the WHO growth reference to calculate the age- and sex-standardized HAZ. The black vertical lines indicate the mean childhood and adolescent HAZs. Of the 54.1% (630/1,025) of participants who were stunted as a child (HAZ < −2 SD from the median), 84.9% had recovered by adolescence (i.e., 535/630 had HAZ > −2), but 8.4% who had not been stunted as children became stunted as adolescents (45/535).

(DOCX)

Click here for additional data file.

Selected descriptive characteristics comparing those lost to follow-up and those re-enrolled as adolescents.

(DOCX)

Click here for additional data file.

Univariate relationships predicting the squared Raven’s T score assuming a linear regression between each variable and the squared Raven’s T score.

(DOCX)

Click here for additional data file.

Characteristics of the cohort according to whether adolescents spoke English.

(DOCX)

Click here for additional data file.

Regression models of the primary outcomes and diarrheal episodes.

Showing coefficients from linear models for Raven’s T score, BMI, and blood pressure; a Poisson model for education (log count); and a multinomial for employment status (log odds), all as a function of childhood diarrhea episodes. Variable inclusion was based on the Bayesian network.

(DOCX)

Click here for additional data file.

Regression models of the primary outcomes and pneumonia episodes.

Showing coefficients from linear models for Raven’s T score, BMI, and blood pressure; a Poisson model for education (log count); and a multinomial for employment status (log odds), all as a function of childhood pneumonia episodes. Variable inclusion was based on the Bayesian network.

(DOCX)

Click here for additional data file.

Literature search.

Forty-five studies identified using 2 search strategies of the PubMed (US National Library of Medicine) database were manually filtered by the first and last author for relevance to this study. Search 1: (cohort OR longitudinal) AND (childhood OR adolescence) AND (cognitive OR cognition) AND (diarrhea OR diarrhoea OR pneumonia OR “childhood illness*”) AND (BMI OR “health progression” OR growth OR “child development” OR “childhood development”). Search 1 yielded 21 studies. Search 2: ((child* OR infant OR adolescen*) AND (cohort* OR longitudinal OR followup OR “follow up”)) AND cogniti* AND ((“human capital” OR “human capacit*” OR potential OR “child development” OR “health progression” OR “adolescent development”) AND ((adult* OR adoles*) AND outcome*))) AND (undernutrition OR stunt* OR diarrhea OR diarrhoea OR pneumonia OR “childhood illness*”). Search 2 yielded 34 studies. And 30 more studies were identified based on reviewers’ comments.

(DOCX)

Click here for additional data file.

STROBE checklist for observational studies.

(DOCX)

Click here for additional data file.

Protocol (adolescent study).

(DOCX)

Click here for additional data file.

Questionnaire (adolescent study).

(PDF)

Click here for additional data file.

Description of the psychometric analysis of the Raven’s Standard Progressive Matrices and Colored Progressive Matrices.

(DOCX)

Click here for additional data file.

Protocol and scoring form for Raven’s Standard Progressive Matrices and Colored Progressive Matrices administration.

(DOCX)

Click here for additional data file.

1 Mar 2021

Dear Dr Rasmussen,

Thank you for submitting your manuscript entitled "Unlocking human capital: revealing relationships between early childhood experiences and adolescent and young adult health status in a resource-limited population" for consideration by PLOS Medicine.

Your manuscript has now been evaluated by the PLOS Medicine editorial staff and I am writing to let you know that we would like to send your submission out for external peer review.

However, before we can send your manuscript to reviewers, we need you to complete your submission by providing the metadata that is required for full assessment. To this end, please login to Editorial Manager where you will find the paper in the 'Submissions Needing Revisions' folder on your homepage. Please click 'Revise Submission' from the Action Links and complete all additional questions in the submission questionnaire.

Please re-submit your manuscript within two working days, i.e. by March 4, 2021.

Login to Editorial Manager here: https://www.editorialmanager.com/pmedicine

Once your full submission is complete, your paper will undergo a series of checks in preparation for peer review. Once your manuscript has passed all checks it will be sent out for review.

Feel free to email us at plosmedicine@plos.org if you have any queries relating to your submission.

Kind regards,

Beryne Odeny

Associate Editor

PLOS Medicine

10 May 2021

Dear Dr. Rasmussen,

Thank you very much for submitting your manuscript "Unlocking human capital: revealing relationships between early childhood experiences and adolescent and young adult health status in a resource-limited population" (PMEDICINE-D-21-00948R1) for consideration at PLOS Medicine.

Your paper was evaluated by a senior editor and discussed among all the editors here. It was also discussed with an academic editor with relevant expertise, and sent to independent reviewers, including a statistical reviewer. The reviews are appended at the bottom of this email and any accompanying reviewer attachments can be seen via the link below:

[LINK]

In light of these reviews, I am afraid that we will not be able to accept the manuscript for publication in the journal in its current form, but we would like to consider a revised version that addresses the reviewers' and editors' comments. Obviously we cannot make any decision about publication until we have seen the revised manuscript and your response, and we plan to seek re-review by one or more of the reviewers.

In revising the manuscript for further consideration, your revisions should address the specific points made by each reviewer and the editors. Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments, the changes you have made in the manuscript, and include either an excerpt of the revised text or the location (eg: page and line number) where each change can be found. Please submit a clean version of the paper as the main article file; a version with changes marked should be uploaded as a marked up manuscript.

In addition, we request that you upload any figures associated with your paper as individual TIF or EPS files with 300dpi resolution at resubmission; please read our figure guidelines for more information on our requirements: http://journals.plos.org/plosmedicine/s/figures. While revising your submission, please upload your figure files to the PACE digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email us at PLOSMedicine@plos.org.

We expect to receive your revised manuscript by May 31 2021 11:59PM. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

We ask every co-author listed on the manuscript to fill in a contributing author statement, making sure to declare all competing interests. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. If new competing interests are declared later in the revision process, this may also hold up the submission. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT. You can see our competing interests policy here: http://journals.plos.org/plosmedicine/s/competing-interests.

Please use the following link to submit the revised manuscript:

https://www.editorialmanager.com/pmedicine/

Your article can be found in the "Submissions Needing Revision" folder.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

We look forward to receiving your revised manuscript.

Sincerely,

Beryne Odeny,

PLOS Medicine

plosmedicine.org

-----------------------------------------------------------

Requests from the editors:

Thank you for your submission. Before we proceed, please address the following editorial and reviewer comments.

1) Please revise your title according to PLOS Medicine's style. Your title must be nondeclarative and not a question. It should begin with main concept if possible. Please place the study design (e.g. "A prospective cohort study,") in the subtitle (i.e., after a colon).

2) Abstract summary - At this stage, we ask that you reformat your non-technical Author Summary. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract. The summary should be accessible to a wide audience that includes both scientists and non-scientists. Please see our author guidelines for more information: https://journals.plos.org/plosmedicine/s/revising-your-manuscript#loc-author-summary.

3) In the abstract Methods and Findings:

a) Please ensure that all numbers presented in the abstract are present and identical to numbers presented in the main manuscript text.

b) Please quantify the main results with both 95% CIs and p values.

c) Please include the important dependent variables that are adjusted for in the analyses.

d) In the last sentence of the Abstract Methods and Findings section, please describe the main limitation(s) of the study's methodology.

4) Did your study have a prospective protocol or analysis plan? Please state this (either way) early in the Methods section.

a) If a prospective analysis plan (from your funding proposal, IRB or other ethics committee submission, study protocol, or other planning document written before analyzing the data) was used in designing the study, please include the relevant prospectively written document with your revised manuscript as a Supporting Information file to be published alongside your study, and cite it in the Methods section. A legend for this file should be included at the end of your manuscript.

b) If no such document exists, please make sure that the Methods section transparently describes when analyses were planned, and when/why any data-driven changes to analyses took place.

c) In either case, changes in the analysis-- including those made in response to peer review comments-- should be identified as such in the Methods section of the paper, with rationale.

5) Please add the following statement, or similar, to the Methods: "This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 Checklist)."

6) Please include the completed STROBE checklist as Supporting Information. When completing the checklist, please use section and paragraph numbers, rather than page numbers.

7) In the methods, please describe how the adolescent questionnaire was developed and verified. Please provide.

8) If you developed a questionnaire as part of this study and it is not under a copyright more restrictive than CC-BY, please include a copy, in both the original language and English, as Supporting Information, or include a citation if it has been published previously.

9) In statistical methods, please refer to any post-hoc corrections to correct for multiple comparisons during your statistical analyses. If these were not performed please justify the reasons. Please refer to our statistical reporting guidelines for assistance (https://journals.plos.org/plosone/s/submission-guidelines.#loc-statistical-reporting)

10) In statistical methods, please discuss how you accounted for clustering of repeated measurements in this cohort.

11) Your study is observational and therefore causality cannot be inferred. Please remove language that implies causality, such as “greater influence on”, “have greatly increased”, or “investments in education have unlocked...” Refer to associations instead. Please temper the last sentence of your conclusion to avoid overreaching what can be concluded from the data. For example, refer to phrases such as “education has the potential to unlock…” and so forth.

12) In the Methods and Results section:

a) Please provide 95% CIs and p values for estimates in the main text and tables

b) When a p value is given, please specify the statistical test used to determine it.

13) Figures and tables:

a) Please indicate in the figure caption the meaning of the whiskers in Fig 4

b) Please define the following abbreviations in your tables. For example, IQR, SRCH, HAZ, BP, BMI

14) The terms gender and sex are not interchangeable (as discussed in http://www.who.int/gender/whatisgender/en/ ); please use the appropriate term.

15) Please use the "Vancouver" style for reference formatting and see our website for other reference guidelines https://journals.plos.org/plosmedicine/s/submission-guidelines#loc-references.

Comments from the reviewers:

Reviewer #1: See attachment

Michael Dewey

Reviewer #2: Rasmussen and colleagues use longitudinal data show that investments in education of females is associated with adult human capital in a Pakistani village decades later. Research of merit and has broad implications. A few points.

1.Secular trends - How was this addressed in this work. Many populations improve in health/well being over time(in fact from S1, your cohort got healthier between childhood and adolescence). I am curious of how was these shifts were addressed in your analyses and how if might impact your overall findings.

2.The role of undernutrition in early life it seems to me is under-emphasized by authors. We know growth, brain development and other systems can all impact outcomes later in life -- all need proper nutrition. Similar work in Guatemalan birth cohort, and other LMICs showed that nutritional investments in females had impact on adult human capital and intergenerational benefits, with regards to offspring birth size. Underweight children have been found to be associated with failing a school grade in early life. It would be good if these points are discussed in this work to improve its contributory value.

3. Measurements - What was done to ensure reliability of clinical measurements(WT, HT, BP, WC etc) ?

4. Raven score tests were not given to adolescents who had a phone interview. What proportion of the of your sample were these? How different were completers vs those who weren't interviewed?

Reviewer #3: The study tries to examine the role of early childhood experiences on health, cognition, and education outcomes of adolescents/young adults in Gilgit Baltistan, Pakistan. Childhood data collected was compared with data collected in adulthood to examine the relationship.

Strong introduction and methods sections. As the authors also highlight in line 64/65 that female empowerment and mental health variables were not available, yet the result section provides interesting insights.

If available, reference number 5 needs to be updated to a recent study/report.

Reference numbers for the ethics approval need to be reported in ethics approval section.

Although authors discuss bias in lines 423-424, some discussion on recall bias needs to be reported as a limitation for SRPH.

The manuscript has the potential to add to literature on examining the relationship of early and childhood health on adolescent/young adult health outcomes/status.

Reviewer #4: This well-structured manuscript describes an important study which investigated the non-verbal cognitive outcomes of a group of adolescents and young adults followed up following an early life cohort study of under-5 mortality and morbidity in Pakistan. This study leverages the value created by a large initial cohort study from the 1990's and spans of period of rapid socio-economic mobility due to increased accessibility of one of the poorest regions in Pakistan. While the high level results reported here of higher SES (in particular, maternal education), predicting better cognitive outcomes are not highly novel, they are nevertheless important given the region on which the study reports, the authors do include discussion on the context specific elements at a more granular level which are highly relevant and important in thinking about potential of such findings to inform policy.

The inclusion of measures of physical health (growth metrics, BMI, hypertension) as signals of potential risk for future NCD's is an additional strength. The methods and analysis approach is well described including specific reasons for children lost to follow up.

As noted in the abstract, measures on maternal mental health does seem like an important limitation and the potential role that this may have played deserves some discussion in the limitations section. Further discussion on the critical role this period of life represents in terms of being the crucible for intergenerational risk and resilience would add value.

No major revisions needed from my perspective, though suggest a formal statistical review and publication of analysis plan if possible

Any attachments provided with reviews can be seen via the following link:

[LINK]

Submitted filename: rasmussen.pdf

Click here for additional data file.

28 May 2021

Submitted filename: Responses_to_Reviewers_Final.docx

Click here for additional data file.

8 Jul 2021

Dear Dr. Rasmussen,

Thank you very much for re-submitting your manuscript "Examining the relationships between early childhood experiences and adolescent and young adult health status in a resource-limited population:  A prospective cohort study" (PMEDICINE-D-21-00948R2) for review by PLOS Medicine.

I have discussed the paper with my colleagues and the academic editor and it was also seen again by three reviewers. I am pleased to say that provided the remaining editorial and production issues are dealt with we are planning to accept the paper for publication in the journal.

The remaining issues that need to be addressed are listed at the end of this email. Any accompanying reviewer attachments can be seen via the link below. Please take these into account before resubmitting your manuscript:

[LINK]

***Please note while forming your response, if your article is accepted, you may have the opportunity to make the peer review history publicly available. The record will include editor decision letters (with reviews) and your responses to reviewer comments. If eligible, we will contact you to opt in or out.***

In revising the manuscript for further consideration here, please ensure you address the specific points made by each reviewer and the editors. In your rebuttal letter you should indicate your response to the reviewers' and editors' comments and the changes you have made in the manuscript. Please submit a clean version of the paper as the main article file. A version with changes marked must also be uploaded as a marked up manuscript file.

Please also check the guidelines for revised papers at http://journals.plos.org/plosmedicine/s/revising-your-manuscript for any that apply to your paper. If you haven't already, we ask that you provide a short, non-technical Author Summary of your research to make findings accessible to a wide audience that includes both scientists and non-scientists. The Author Summary should immediately follow the Abstract in your revised manuscript. This text is subject to editorial change and should be distinct from the scientific abstract.

We expect to receive your revised manuscript within 1 week. Please email us (plosmedicine@plos.org) if you have any questions or concerns.

We ask every co-author listed on the manuscript to fill in a contributing author statement. If any of the co-authors have not filled in the statement, we will remind them to do so when the paper is revised. If all statements are not completed in a timely fashion this could hold up the re-review process. Should there be a problem getting one of your co-authors to fill in a statement we will be in contact. YOU MUST NOT ADD OR REMOVE AUTHORS UNLESS YOU HAVE ALERTED THE EDITOR HANDLING THE MANUSCRIPT TO THE CHANGE AND THEY SPECIFICALLY HAVE AGREED TO IT.

Please ensure that the paper adheres to the PLOS Data Availability Policy (see http://journals.plos.org/plosmedicine/s/data-availability), which requires that all data underlying the study's findings be provided in a repository or as Supporting Information. For data residing with a third party, authors are required to provide instructions with contact information for obtaining the data. PLOS journals do not allow statements supported by "data not shown" or "unpublished results." For such statements, authors must provide supporting data or cite public sources that include it.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Please review your reference list to ensure that it is complete and correct. If you have cited papers that have been retracted, please include the rationale for doing so in the manuscript text, or remove these references and replace them with relevant current references. Any changes to the reference list should be mentioned in the rebuttal letter that accompanies your revised manuscript.

Please note, when your manuscript is accepted, an uncorrected proof of your manuscript will be published online ahead of the final version, unless you've already opted out via the online submission form. If, for any reason, you do not want an earlier version of your manuscript published online or are unsure if you have already indicated as such, please let the journal staff know immediately at plosmedicine@plos.org.

If you have any questions in the meantime, please contact me or the journal staff on plosmedicine@plos.org.

We look forward to receiving the revised manuscript by Jul 15 2021 11:59PM.

Sincerely,

Beryne Odeny,

Associate Editor

PLOS Medicine

plosmedicine.org

------------------------------------------------------------

Requests from Editors:

1) Please remove the term “prospective” from the title as it is not clear that the research was prospectively planned, and there was no prespecified analysis plan.

2) Adolescent participants provided “written informed consent” - would assent and parental consent not be more usual? Either way, please clarify under which circumstances they offered assent or consent (e.g., married, or emancipated minors consented, while assent and/or parental consent was sought for younger adolescents?)

3) For references #3 & #10, please provide access dates for the referenced weblinks. Please ensure that all weblinks are accessible and access dates updated.

Comments from Reviewers:

Reviewer #1: The authors have addressed all my points.

Michael Dewey

Reviewer #2: No further comments

Reviewer #4: my comments have been adequately addressed by the authors

Any attachments provided with reviews can be seen via the following link:

[LINK]

15 Jul 2021

Submitted filename: Responses_to_Reviewers_Final.docx

Click here for additional data file.

28 Jul 2021

Dear Dr Rasmussen,

On behalf of my colleagues and the Academic Editor, Dr. Kathryn Mary Yount, I am pleased to inform you that we have agreed to publish your manuscript "Examining the relationships between early childhood experiences and adolescent and young adult health status in a resource-limited population:  A cohort study" (PMEDICINE-D-21-00948R3) in PLOS Medicine.

Before your manuscript can be formally accepted you will need to complete some formatting changes, which you will receive in a follow up email. Please be aware that it may take several days for you to receive this email; during this time no action is required by you. Once you have received these formatting requests, please note that your manuscript will not be scheduled for publication until you have made the required changes.

In the meantime, please log into Editorial Manager at http://www.editorialmanager.com/pmedicine/, click the "Update My Information" link at the top of the page, and update your user information to ensure an efficient production process.

PRESS

We frequently collaborate with press offices. If your institution or institutions have a press office, please notify them about your upcoming paper at this point, to enable them to help maximise its impact. If the press office is planning to promote your findings, we would be grateful if they could coordinate with medicinepress@plos.org. If you have not yet opted out of the early version process, we ask that you notify us immediately of any press plans so that we may do so on your behalf.

We also ask that you take this opportunity to read our Embargo Policy regarding the discussion, promotion and media coverage of work that is yet to be published by PLOS. As your manuscript is not yet published, it is bound by the conditions of our Embargo Policy. Please be aware that this policy is in place both to ensure that any press coverage of your article is fully substantiated and to provide a direct link between such coverage and the published work. For full details of our Embargo Policy, please visit http://www.plos.org/about/media-inquiries/embargo-policy/.

To enhance the reproducibility of your results, we recommend that you deposit your laboratory protocols in protocols.io, where a protocol can be assigned its own identifier (DOI) such that it can be cited independently in the future. Additionally, PLOS ONE offers an option to publish peer-reviewed clinical study protocols. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols

Thank you again for submitting to PLOS Medicine. We look forward to publishing your paper.

Sincerely,

Beryne Odeny

Associate Editor

PLOS Medicine