The majority of OAC aged 6-12 in this study engaged in some sort of work during the previous week (60.7%) and of those who worked, 17.8% (10.5% of the total sample) did so for 28 or more hours. More than one-fifth (21.9%, 13% of the total sample) met UNICEF's child labour definition, excluding hazardous work. The omission of hazardous work from the definition of child labour, the comparatively young age, and high rates of urban residence among OAC in the POFO cohort, suggest that rates of child labour among OAC in the six study sites may be higher than general estimates that indicate that 16% of children aged 5 to 14 in LMICs engage in child labour .
Child work was associated with significantly lower rates of school attendance when the work exceeded 27 hours per week. This relationship was even stronger among OAC younger than 12 years working for pay and OAC ages 12 or above who worked 14 or more hours per week for pay. While these findings support UNICEF child labour definition as a standard for policy makers looking to eliminate child work that might interfere with schooling, it is possible that the lower limit of 21-27 hours suggested by some was not significant because the study cohort is comprised of only OAC who predominantly reside in poorer or otherwise disadvantaged households. The estimated associations between labour and schooling may thus represent estimates at the margin.
The importance of rural vs. urban residence, and income and wealth of the household, indicate the need for policies and programs designed to reduce child labour to focus on family and household support. While wealth was associated with all categories of work and labour, lack of income was associated only with child labour in the full model. It is important to recognize that a household may be very poor, or have little wealth, but the loss of an income earning adult may be what makes a family decide that an OAC must engage in child labour and stop attending school.
The finding that female children were more than twice as likely to be engaged in child labour provides further evidence that that girl children face a particularly heavy burden with respect to household duties, childcare and other kinds of labouring. The finding supports the argument made by child protection and policy making organizations that when unpaid domestic "chores" are not counted as labour, we risk missing the large burden being placed on children, particularly girl children, which can interfere with their educational attainment and future wellbeing.
One unique aspect of this paper is its explicit focus on labour among orphans and abandoned children (OAC); to our knowledge, it is the first of its kind to do so. While point estimates suggest slightly higher odds of child work or labour if the child was not living with a biological parent, the lack of statistical significance suggests being with a biological relative was not protective against labour or work.
Much of the child labour literature focuses on educational attainment and school attendance, as it is widely believed that child labour interferes with schooling [5, 22, 23]. Research conducted in five culturally distinct countries examined the relationship of school attendance, grade attainment and reported working hours, defined as both market and non-market related activity. The authors noted an inverse association with both market-activity defined as..."activities leading to the production of goods and services which are intended for sale or sold on the market" , and non-market activity defined as... "production of goods and services for members of a household for household use", e.g. fetching water, meal preparation, child care, etc...  working hours and school attendance rates in all countries, with drops in attendance around the upper end of working hours (between 25 and 40 or more hours).
Other literature focused on child labour, productivity and the different ways these experiences interacted with school attendance and school performance; a Tanzanian cohort study examined child labour and future productivity saw a reduction in schooling and graduation rates, while research conducted in Lebanon found only "exclusive involvement in economic activity" negatively affected school attendance. In Brazil, children who reported working had higher school performance and less grade repetition. In Sri Lanka, as reported working hours increased, reports of regular attendance decreased although test scores reveal no differences between working and non-working students, while in Turkey the opposite situation was reported . These differences in findings may be due differences in definitions of what constitutes long working hours, or excessive labour. Using a standard definition would improve comparability across studies and sites. This study supports concern being focused on work longer than 28 or more hours per week or paid work outside the home at lower numbers of hours per week, especially for younger children.
While there are distinct and specific negative associations between some kinds of work, school attendance and performance, there also appears to be an overall difference in school attendance, performance, and dropout rates between children who work long hours exclusively in market activity as compared to children who work long hours on household and domestic chores. Indeed, noted in the literature is a reasonable amount of non-market activity work (i.e., chores, light domestic work) is positively associated with better school performance, attendance, and perceptions of self worth [14, 15, 24, 25].
While this manuscript with its focus on OAC adds to the growing body of literature on child labour, there are several important limitations to the study. First, data are reported by the caregivers and the authenticity of their responses cannot be validated. As reported in the literature, it is reasonable to assume that proclamations against child labour have created an atmosphere of anxiety around admitting to engaging in child labour [26–28]. While our interviewers succeeded in developing rapport with caregivers, we cannot rule out the possibility that their responses under-represent the extent of child work and labour practices. While the amount of child labour reported is within reason considering the reported rates for non-OAC in LMICs, it is plausible that the OAC population is at even greater risk of child labour. The results from this study are not generalizable to non-OAC or OAC in "northern" or high income countries.
Additionally, the literature would benefit from specific delineation of the activities that constitute chores, which this study did not do. In order to get at the core of how children are labouring and/or spend time in 'chore/household' activities, we need to collect more detailed information on daily activities. Importantly, because there is no information on the type of work being done, the analyses and child labour definition could not include hazardous work. Finally, the reference period for child work was the week prior to the interview, it is possible that children may engage in significantly different work or labour patterns during other times of the year, especially during the summer.