Patterns of Substance Use Among Street-Involved Children and Youth in Kenya: the Roles of Street-Exposure, Migratory Factors, Family Deprivation, and Geographic Sub-location

Abstract

Substance use among street-involved children and youth (SICY) in low- and middle-income countries is common. Using data abstracted from program intake forms (2016–2022) for an intervention to reintegrate SICY with their communities, we assess which individual, family, and geographic characteristics are predictive of substance use, and specifically inhalant use and non-inhalant use among a sample of 227 SICY in Meru County, Kenya. Assessed determinants include age, geographic location of home community, years on street, family deprivation prior to street-migrating, motivation for street migrating, abuse experiences on the street, and activities on the street. Number of years lived on the street, experiencing abuse on the street, and citing peer-self relations as the reason for street migration were all associated with significantly higher odds of reporting substance use, and inhalant use specifically. Inhalant use was also significantly associated with peer socialization and specific street activities. Further research should explore the role of peer-self dynamics in substance use among SICY and how it can inform approaches to reintegrating children from street situations and sustaining their development in non-street environments.

Global estimates of the numbers of children who live on the streets, separated from parents or guardians, range widely and are without robust empirical support (UNICEF, 2005). Estimates for Kenya based on global and national censuses range from approximately 46,000 children to 300,000 (UNICEF, 2005; IRIN, 2007; Kenya Ministry of Labour and Social Protection, 2018). The challenges of enumerating street-involved children are multiple and well-documented (c.f. Kenya Ministry of Labour and Social Protection, 2018).

Research has illumined multiple challenges experienced by street-involved children and youth (SICY). SICY are at greater risk of engaging in substance use and of experiencing abuse and neglect, social stigma, sexually-transmitted infections, and worse mental health compared to children in nurturing home environments (Embleton et al., 2020; Sitienei & Pillay, 2019; Omari et al., 2021; Gayapersad et al., 2020). These street-related adversities compound life histories of maltreatment prior to migrating to the streets (Sitienei & Pillay, 2019; Seidel et al., 2018).

Substance use (both inhalant and non-inhalant substances) among SICY is common. A 2013 systematic review of literature revealed the pooled prevalence of substance use among SICY in resource-constrained countries was 60%, with inhalant use contributing the highest burden at 47% (Embleton et al., 2013b). Cross-sectional studies have identified risk factors for substance use among SICY in lower- and middle-income countries including: family substance use history, peer and sibling influence, spending more time on the streets, experiencing abuse, and poor mental health (Hills et al., 2016; Tyler et al., 2016; Embleton et al., 2013a; Ayenew, Kabeta & Woldemichael, 2020). Additionally, research in higher income countries has also confirmed the association of additional family factors such as poor family functioning and family conflict (Milburn et al. 2017) and of street victimization (Bender et al., 2015) with substance use among homeless youth.

Substance use represents a major health risk factor for SICY and is associated with poor physical, neurological and psychiatric health outcomes and impairments in cognitive functioning (Yücel et al., 2008; Lees et al., 2020). In addition to undermining healthy physiological and psychological development, substance use appears to mark social identification with other SICY (Woan, Lin & Auerswald, 2013). Studies have demonstrated that risk factors for adolescent substance use are different than those for adult substance use, namely social and developmental factors that are relevant to adolescence: peer social relationships, peer substance use, and conflict with parents and parental monitoring (Allen et al., 2021). Socialization with other SICY is a critical component of their lives on the street. Because of the adversities faced prior to and once on the street, SICY often have lengthy and deep experiences with trauma that require strong social support and feelings of solidarity to survive (Reza & Henly, 2018). Despite the documented risks for substance use among SICY, there are still gaps in understanding how their home environments prior to street migration and their lived experiences (specifically their relationships with peers, activities, and victimization) on the street are related to substance use in general, and in particular inhalant use.

Study Aim

Therefore, this study aims to assess similarities and differences in street exposures, street behaviors, motivations for street-migration, family of origin challenges, and geographic sub-locations among substance users, and specifically inhalant and non-inhalant users, among a sample of 227 SICY living in small cities across three sub-counties of Meru County. Additionally, we aim to determine which individual, family, and geographic characteristics predict reporting substance use, inhalant use and non-inhalant use. Assessed determinants include age, geographic location of home community, years on street, family deprivation prior to street-migrating, motivation for street migrating, abuse experiences on the street, and activities on the street. This study uses participant intake data collected as part of an ongoing program to reintegrate SICY with their home communities or enroll them in boarding or polytechnical schools.

Methods

Program Description

The Kenyan National Council for Children’s Services promotes a reintegration-oriented strategy for children living in street situations (NCCS, 2022). Consistent with this strategy, the program assessed in this study utilizes a “4R + P” strategy—rescue children from street situations, rehabilitate children from substance use and other street-specific behaviors, reintegrate children with their communities of origin, resocialize children within their home contexts, and prevent children from migrating to the streets. The program, Watoto wa Ahadi (“Children of Promise”) Rescue Center run by Sodzo Kenya, began interventional work in April 2016. The intervention utilizes a 79-acre farm with dormitories, dining hall, remedial school, and meeting space for children and youth who need to transition off the street prior to reintegration. The program collects baseline data on all program participants and data from family retracing efforts to inform an exit strategy for each child. The program operates in Meru County, Kenya, which contains 2.8% of all street-living persons according to the Kenyan government’s 2018 census (Kenya Ministry of Labour and Social Protection, 2018). Reflecting gendered patterns of street-living persons within Meru County and nationally, all but one participant was male, thus preventing a gendered analysis of findings.

Sample Selection

Data from all participants in the Watoto wa Ahadi Rescue Center (ARC) program are included in this study. The program data span a 6-year period, with data collected upon entry to the program. As some participants return more than once, we have included only their initial entry to the program. These data include 227 unique children.

Measures

Child Self-reported Data

Recruitment to the program begins with employed social workers establishing rapport with children identified to be living on the street. Children who are interested in enrolling in the program then complete an intake form (read aloud) with the social worker. Intake forms are kept in files onsite. Data for this study were abstracted from the intake forms and entered into Excel.

Substance Use

Substance abuse was measured by self-reported use of glue huffing/inhalant, miraa/khat, bhang, alcohol, and other substances. Glue inhalants are toluene-based and are procured from cobblers working in the towns. Toluene-based inhalants severely impair neurological and neuropsychological development and can permanently damage white matter, slowing cognitive processing speed, sustained attention, memory retrieval, executive function and language (Yucel, Takagi, Walterfang & Lubman, 2008). Miraa (a.k.a. Khat or Catha edulis) is a stimulant with a globally contentious status as either drug of abuse or a mild stimulant similar to coffee (Gebissa, 2008). Miraa production is a substantial driver of agro-economic development in Meru County (Krueger & Mutyambai, 2020). Bhang is a cannabis-derived substance typically ingested to induce euphoria, and may also cause panic, depression, psychomotor impairment, or immune system dysregulation (El-Gohary & Eid, 2004). Other substances were not defined on the form, and there was no place for notes to write in a specific type. Inhalant use was defined as self-reported use of glue; non-inhalant use was defined as self-report of miraa/khat, bhang, alcohol, or other (could be in addition to use of glue). Use of each substance was coded as a binary variable—reported vs. reported as not occurring or having nothing recorded in the data field.

Age, Years on Street, Age at First Street-Migration

Youth indicated their age in years, and the number of years they have lived on the street. Age of street migration was then calculated by subtracting years on the street from age. These three datapoints were used as continuous variables in bivariate analysis. Age and years on the street were used in regression analyses.

Motivation to Migrate to the Streets

While migration dynamics are complex and interconnected, youth were asked to provide their leading motivation to migrate to the streets. These responses were initially categorized into nine categories when they were abstracted from the intake data forms, then further categorized into four categories. The four categories (nine initial categories shown) included peer-self relations (boredom, influence by friends, and personal choice), parent challenge (family conflict at home, parental death, and parental separation), economic challenge (inadequate food, or inadequate housing), and other. Initial bivariate statistics indicated peer-self relations as the most prevalent reason for migrating to the street among each category of substance users compared to the other three reasons. In our initial simple logistic regression models to predict substance use from the four categories, peer reasons were associated with a significantly higher odds of substance use; thus, in our final model, we collapsed all categories for street migration motivation into a binary variable—migration due to peer-self relations or due to another reason.

Abuse Experiences on the Street

Youth were asked about challenges they experienced on the street and recorded as notes. These notes were then codified into four forms of abuse—physical, emotional, economic, and sexual. These responses do not differentiate based on who perpetrated the abuse. Physical abuse included instances where the child was physically beaten by other SICY, adults, or police. Emotional abuse included instances where the child was called names, yelled at, or otherwise embarrassed by someone with more social, economic, or physical power (could include other SICY). Economic abuse included instances where SICY were promised some payment but were not subsequently provided due wages. Sexual abuse included instances where the child was forcibly penetrated or forced to provide some sex act to another person. Each abuse was coded as present in the report or not. In multivariate analyses, we chose to use a dichotomous variable for reporting experiencing abuse on the street or not reporting abuse.

Interest in Socializing with Other Children

Children were asked what they enjoyed doing on the streets (sports/games; social activities with other children; domestic activities; or enrichment hobbies). Based on these four response options, we created a binary variable for indicating an interest in socializing or spending time with other SICY.

Street Activities

Children were asked what activities they engage in on the street. Categories for begging for food or money, rummaging through garbage, pushing a handcart, peddling drugs, cleaning shops, and other were initially determined. We chose to focus on begging for money and searching garbage for resources because they indicate a deeper level of street-involvement and have previously been associated with abuse experiences (e.g. Pinzon-Rondon et al., 2010). A dichotomous variable for children who reported begging for money/food or rummaging through garbage versus those who did not was created and used in multivariate analyses.

Home Sub-location

As part of program family retracing efforts, respondents were asked how to find their family and communities of origin. Home communities were confirmed in each case by social workers who visited the home village to retrace the child’s family. We used the 2019 Kenya census as the standard for the names and taxonomy (sub-county > location > sub-location > village; Kenya National Bureau of Statistics, 2019). Some villages named by child and confirmed by staff did not match a village name in the taxonomy, but did match a location or sub-location name. In these cases, the largest village in the taxonomy was used. Villages of origin were determined to be nested within 11 sub-locations defined by Kenyan geographic boundary definitions. Three sub-locations with fewer than 10 observations were censored to limit participant identifiability. Eight sub-locations comprising 184 of the SICY remained after geographic data suppression.

Form of Family Deprivation

Program social workers recorded the challenges faced by families after visiting families of origin, including substance use, inadequate housing, family instability, food insecurity, water insecurity, and foster care situation. The presence/absence of each type of deprivation was coded as a dichotomous variable.

Statistical Analysis

Descriptive Analysis

The mean or percentage reporting different categorical responses was calculated for all children and youth and for each substance use group. Bivariate tests of independence were used to assess statistical difference in participant characteristics, family of origin characteristics, street experiences, and street activities between youth who reported substance use, inhalant use, or non-inhalant user and those who did not. For pairwise comparisons, we utilized ANOVA (continuous), Wilcoxon rank-sum (ordinal), and for categorical variables, X² (cells with larger frequencies) and Fisher exact (cells with smaller frequencies smaller) to determine statistical significance.

Geospatial information was used to visualize substance use (any substance, inhalant and non-inhalant) by sub-location where sub-location counts were greater than 10 observations. Eleven sub-locations (see the “Form of Family Deprivation” section) were mapped in ArcMap 10.7.1 (Esri, 2019) using a shapefile of Kenyan administrative boundaries, with the prevalence of glue, non-inhalants, and substance use overall symbolized. Sub-locations were assessed for heterogeneity within each substance use category.

Hypothesis

We hypothesized that longer street involvement, experiencing abuse while on the street, participating in specific street activities (begging/rummaging), family factors prior to migration (such as substance abuse, economic insecurity, and family conflict), and individual-peer factors (migrating to the street due to self-peer influences, and an interest in socializing with other children) would predict SICY substance abuse. Though this region of Kenya shares a common language and cultural practices, we hypothesized that different places of origin could be associated with reporting substance abuse.

Inferential Analysis

Multivariate logistic regression using the maximum likelihood estimator was used to predict the odds of reporting substance use, inhalant or non-inhalant use by model variables. Initial models included age and geolocation only to determine if sub-location of origin was associated with reporting drug use. After assessing the heterogeneity of odds of reporting each category of substance use by sub-location, we decided not to include the geographic variables in the final models. Three regression models were then developed to predict substance use overall, inhalant use, and non-inhalant use from the abovementioned variables. Model building proceeded as follows: following initial models with age and geolocation, we regressed each substance use variable on each predictor variable and assessed unadjusted associations. Final models included age (controlled for), years on the street, experiencing abuse on the street, migration to the street for individual/peer reasons, interest in socializing with other children, and family substance use.

All statistical analyses were conducted using STATA v.16.1 (StataCorp, 2019).

Ethical Consideration

Court committals or proxy consent were given for all children within programmatic care. All data were collected following a period of rapport building on streets by social workers. Parental consent was attained whenever possible immediately following rescue from street situations. Child assent was provided in the conduct of programmatic activities. The Institutional Review Board exemption for publishing deidentified data was provided by the lead author’s academic institution. All geographic information was censored when individuals could represent more than 10% of sub-group data.

Results

Table 1 displays characteristics of street-involved children in youth who reported using any substance, inhalants, or any non-inhalant. Inhalant (or glue) use was reported by 56% of respondents, while non-inhalant use was reported by 50%. Of non-inhalants reported, “other” was the highest (35%), followed by miraa (32%), bhang (27%), and alcohol (12%). Of youth who reported glue use, 71% also reported using a non-inhalant; of youth who reported non-inhalant use, 79% reported also using glue. Approximately 40% of all respondents reported using both inhalants and non-inhalants, and 43% reported using more than one substance (polysubstance use).

Table 1.

Description and bivariate analysis of model variables of children and youth living on the streets of Meru County Kenya (N=227)

	All children & youth		Substance users			Non-inhalant substance users (may also use inhalants)			Inhalant substance users (may use non-inhalants too)
	N	% or M(SD)	N	% or M(SD)	p-value*	N	% or M(SD)	p-value*	N	% or M(SD)	p-value*
Substance Use on Street
Any substance	227	67.0%	-	-	-	-	-	-	-	-	-
Glue inhalant use	227	56.4%	152	84.1%	-	114	78.9%	-	-	-	-
Alcohol	227	11.5%	152	17.1%	-	114	22.8%	-	128	17.2%	-
Khat/miraa	227	32.2%	152	48.0%	-	114	64.0%	-	128	47.7%	-
Bhang	227	27.3%	152	40.8%	-	114	54.4%	-	128	42.2%	-
Other	227	35.2%	152	52.6%	-	114	70.2%	-	128	50.8%	-
Any non-inhalant (alcohol, khat, bhang, other)	227	50.2%	152	75.0%	-	-	-	-	128	70.3%	-
Polysubstance	227	42.7%	152	63.8%	-	114	85.1%	-	128	70.3%	-
Age	226	13.0 (2.16)	152	13.0 (2.0)	0.709	114	13.1 (2.1)	0.603	128	13.0 (1.8)	0.998
Years on street	227	1.3 (1.9)	152	1.8 (2.1)	<0.001	114	1.8 (2.1)	<0.001	128	1.9 (2.2)	<0.001
Age at first migration to streets	183	11.4 (2.5)	141	11.1 (2.4)	0.006	107	11.2 (2.5)	0.175	121	11.0 (2.4)	0.010
Number of forms of abuse	227	1.1 (0.9)	152	1.4 (0.8)	<0.001	114	1.4 (0.9)	<0.001	128	1.5 (0.9)	<0.001
Abuse experience on street
Physical	227	37.4%	152	48.7%	<0.001	114	45.6%	0.011	128	54.7%	<0.001
Emotional	227	36.6%	152	46.1%	<0.001	114	48.2%	<0.001	128	42.2%	0.045
Economic	227	27.8%	152	38.8%	<0.001	114	38.6%	<0.001	128	41.4%	<0.001
Sexual	227	5.7%	152	8.6%	0.006	114	10.5%	0.003	128	7.8%	0.156
Street activities
Beg for money/food or rummage through garbage	227	39.2%	152	52.0%	<0.001	114	51.8%	<0.001	128	58.6%	<0.001
Push handcart	227	20.7%	152	28.9%	<0.001	114	26.3%	0.036	128	31.3%	<0.001
Peddle drugs	227	3.1%	152	3.9%	0.284	114	5.3%	0.119	128	3.1%	0.967
Clean shops	227	33.0%	152	45.4%	<0.001	114	48.2%	<0.001	128	49.2%	<0.001
Reasons for migrating to street
Child’s choice/peer influence	182	41.2%	152	42.8%	0.207	114	39.5%	0.721	128	46.1%	0.029
Parental conflict/instability	182	29.7%	152	29.6%	0.003	114	30.7%	0.014	128	28.9%	0.039
Economic hardship	182	11.0%	152	10.5%	0.224	114	10.5%	0.483	128	9.4%	0.0816
Other	182	18.1%	152	15.8%	0.129	114	17.5%	0.903	128	14.8%	0.092
Interest in socializing with other children on street	227	41.4%	152	50.7%	<0.001	114	50.0%	0.008	128	58.6%	<0.001
Form of family challenge
Substance use	227	33.5%	152	39.5%	0.006	114	36.0%	0.426	128	40.6%	0.009
Inadequate housing	227	67.0%	152	73.7%	0.002	114	70.2%	0.301	128	71.9%	0.073
Family instability	227	70.0%	152	81.6%	<0.001	114	78.9%	0.003	128	81.3%	<0.001
Food insecurity	227	59.9%	152	67.1%	0.002	114	66.7%	0.037	128	63.3%	0.239
Water insecurity	227	18.9%	152	23.0%	0.03	114	19.3%	0.891	128	25.8%	0.003
Foster care	227	11.9%	152	8.6%	0.027	114	9.6%	0.294	128	5.5%	0.001

Substance use, age, years on street, abuse experience, reasons for street migration and interest in socializing with other children on the street provided by child self-report during intake and confirmed where possible by adults familiar with the child. Form of family challenge recorded by social worker upon family retracing.

^*p-value provided for bivariate tests for street exposure, street experiences, street activity, migration reason, and family challenge variables; for continuous variables, ANOVA; for ordinal, Wilcoxon rank sum; for categorical variables (%), chi^2 test with Fisher’s exact for small cell size. Comparisons are for any substance use, any non-inhalant, and inhalant use versus not reporting substance use, not reporting non-inhalant use, and not reporting glue use, respectively

The mean age of respondents was 13 years (SD = 2.2), which did not differ significantly between those who reported substance abuse, non-inhalant, or inhalant use and those who did not. The mean number of years on the street was 1.3 years (SD = 1.9), which was significantly higher among youth who reported substance use (1.8), non-inhalant use (1.8), and inhalant use (1.9) compared to youth reported no substance use, non-inhalant use, or inhalant use, respectively. The mean age at first migration to the streets was 11.4 years (SD = 2.5) and was significantly lower among youth who reported substance use (11.1) and among youth who reported inhalant use (11.0) compared to youth reported no substance abuse or inhalant use, respectively.

The percentage of youth reporting physical abuse (37%), emotional abuse (37%), and economic abuse (28%) was significantly higher among SICY reporting substance abuse, non-inhalant use, and inhalant use compared to youth reported no substance abuse, non-inhalant use, or inhalant use, respectively. Youth who reported substance use, non-inhalant use, or inhalant use all had a significantly higher mean number of types of abuse experienced on the street compared to youth who did not report use.

Overall, 39% of respondents reported begging for money or rummaging through garbage, and this percentage was significantly higher among respondents who reported substance use, non-inhalant use, and non-inhalant use (compared to not reporting use).

Over 40% of children reported they migrated to the streets due to their choice, influence from peers, or boredom at home. Nearly one-third of children reported family dysfunction (30%) motivated their move to the streets, while 11% reported economic hardship as the motivating factor, and 18% reported some other cause.

Over 40% of children reported they were interested in socializing with other street-involved children, with the highest percentage being among inhalant users (59%). In each category of substance use, a significantly higher percentage of youth reported interest in socializing with peers compared to youth who did not report use. The most frequent challenges observed by social workers at home retracing visits were family instability (70%), inadequate housing (67%), and food insecurity (60%).

Figure 1 shows the variation in inhalant use by sub-location based on a participant’s village of origin. There is substantial heterogeneity in inhalant use by sub-location, ranging from 10% in one sub-location to 76% in the highest sub-location. Similar heterogeneity is also found among substance use and non-inhalant use by sub-location (cf. Supplemental Figures 1 and 2; Supplemental Table 2).

Fig. 1.

Percentage of SICY reporting inhalant use by sub-location

Table 2 depicts the logistic regression models for substance abuse, non-inhalant use, and inhalant use (respectively) on age and predictive model variables. Initial models including only age and geographic sub-location are provided in Supplemental Table 1. Heterogeneity of reported substance use by sub-location is more pronounced for participants reporting any substance use and inhalant use than for those reporting non-inhalant use. Sub-location #4 shows higher odds of non-inhalant use relative to sub-location #6 (reference), but this pattern is opposite for any substance use and inhalant use. Sublocation 5 had the lowest prevalence of reporting any of the substance categories (Supp Table 2). Sub-locations 5, 7, and 8 had a significantly lower odds of reporting any substance use or inhalant use than sub-location 6 (reference). The number of years on the street predicted significantly higher odds of substance use and inhalant use (aOR: 1.43 and 1.61, respectively). The number of abuse experiences on the street predicted a significantly higher odds of substance abuse, non-inhalant use, and inhalant use (aORs: 2.83, 1.54, and 2.29, respectively). Higher odds of substance use, and specifically inhalant use but not non-inhalant use, was associated with peer-self related motivations to migrate to the street (aOR: 2.95 and 4.83). Additionally, begging for money or rummaging through garbage to acquire resources and an interest in socializing with other youth on the street predicted a significantly higher odds of inhalant use (aOR: 3.93 and 2.70, respectively). Coming from a family where substance abuse was an issue was not significantly associated with a higher odds of use in any substance use category.

Table 2.

Logistic regression of any substance use, non-inhalant use, and inhalant use on model variables

	Model 1: substance use				Model 2: non-inhalant use				Model 3: inhalant use
	OR	95% CI		p-value	OR	95% CI		p-value	OR	95% CI		p-value
Current age	1.06	0.87	1.29	0.572	1.09	0.93	1.27	0.283	1.04	0.86	1.26	0.671
Years on street	1.44 *	1.02	2.04	0.038	1.15	0.96	1.38	0.117	1.58 **	1.18	2.12	0.002
Forms of abuse experienced on streets	2.83 **	1.50	5.35	0.001	1.54 *	1.03	2.30	0.036	2.29 **	1.30	4.03	0.004
Reason for migration (peer-self relations)	2.95 *	1.39	2.29	0.022	1.09	0.57	2.07	0.800	4.83 ***	2.03	11.50	<0.001
Interest in socializing with peers	0.94	0.47	−0.13	0.899	0.88	0.43	1.80	0.720	2.70 *	1.07	6.81	0.036
Beg for money and/or rummage through garbage	1.85	0.90	1.26	0.206	1.15	0.55	2.41	0.714	3.93 **	1.66	9.32	0.002
Family substance abuse	1.27	0.60	0.49	0.621	0.73	0.38	1.40	0.339	1.21	0.52	2.80	0.661

Logistic regression of substance use category on model variables. OR odds ratio. 95% CI, 95% confidence interval.

^*p<0.05.

^**p<0.01.

^***p<0.001.

P-values are not adjusted for multiple models using same dataset. Post-estimation goodness of fit tests all p>0.05

Discussion

The primary objective of this study was to understand the relationship between family circumstances prior to street migration, reasons for street migration, and street exposures, and the use of substances, and more specifically, non-inhalants and inhalants, among a group of street-involved children and youth in Kenya.

Patterns for any substance use and for inhalant use were noticeably different from patterns for non-inhalant use among street-involved children and youth, though use of one form of substance was predictive of the other. A significantly higher odds of reporting substance use in general was associated with length of time on the street, abuse experienced on the street, and migrating to the street for peer-self reasons. Longer time on the street, experiencing abuse on the streets, street-migration motivations for connecting with other children on the street, behaviors common among street-involved children such as begging or rummaging through garbage for resources, and interest in socializing with other SICY were significantly predictive of a higher odds of reporting inhalant use. Non-inhalant use was not significantly predicted by any of the model variables except number of forms of abuse experienced on the street.

We were surprised age was not associated with substance use; however, there was little variation in age overall, with almost 90% of children between 10 and 15 years of age. Increased time on the streets was associated with a significantly higher odds of reporting substance use, and inhalant use specifically, which is consistent with other literature reporting that children who spend more time on the streets are more likely to report substance use (Embleton, 2013a). Our finding that experiencing abuse on the street was significantly predictive of substance use is consistent with other research. Multiple studies have confirmed that a history of abuse is predictive of adolescent substance use (Tommyr et al., 2010), with longitudinal studies confirming development of substance abuse disorders later in life (Halpern et al., 2018). Our variable measured abuse while on the streets (versus measuring abuse prior to migration) and suggested that recent victimization could play a role in substance use. Whether these effects are additive or whether the damage of earlier child abuse has the greatest effect cannot be determined by our data (Tyler and Melander, 2015). Family substance use was not associated with substance use among SICY in our sample. This is different than other research with homeless youth, which has found that parental drug use is associated with adolescent and young adults’ use of substances while on the street (Tyler and Melander, 2015) as well as research among youth not on the streets (Rusby et al., 2018). This could be due to differences in types of substances used among adults and SICY in Kenya (e.g. Jaguga et al., 2022; Okoyo et al., 2022).

Limitations

A major limitation of the study is that study data were collected as part of an ongoing program to support the reintegration of street-involved children and youth. As secondary data, data collection and measures were not designed to support the aims of this investigation. For example, the option to select “other” for substances use makes meaningful interpretation of this response difficult. Additionally, we treated data gaps as “0s”, because the original paper intake forms had a place to write information that was available but no way to indicate whether information was missing because it was a “no” response or because the question had not been asked or the respondent had refused. We considered this approach to be the most conservative approach, leading to the interpretation of any substance use as “reported” vs “unreported.”

This study contributes to the literature on substance use among SICY by investigating how home environments prior to street migration and lived experiences on the street are related to substance use in general, and in particular inhalant use. This is the first study, to our knowledge, to examine how street exposure, street experiences, street location, and family origin characteristics predict inhalant use in Kenya. A further strength of this study is that the applied setting that generated these data provides measurement validity beyond what is possible with a survey-based approach where rapport between data collectors and street-involved children and youth is minimal. The traumatic histories experienced by many SICY often lead them to be distrusting of adults outside of their environment. The applied setting also facilitated establishing relationships with families and communities of origin, providing an additional data check where possible.

Conclusion

Globally, there are potentially tens of millions of children living on urban streets without adult care and protection. Substance use among this population is consistently high and may be associated with the process of social identification. This study provides new insights connecting forms of motivations for street migration, abuse suffered on streets, duration of street exposure, street activities, and social motivations to substance use, and inhalant and non-inhalant use, specifically. Children should be supported off the streets as rapidly as possible to limit pernicious street exposures; however, insights from this study can inform interventional approaches. Based on the different risk profiles of substance users, interventions may be able to promote safer streets for children without other options. Further research should explore the role of peer-self dynamics in substance use among SICY and how it can inform approaches to preventing migration, reintegrating children from street situations, and sustaining their development in non-street environments.