Background: Globally, an estimated 15·6 million people inject drugs. We aimed to investigate global variation in the age profile of people who inject drugs (PWID), identify country-level factors associated with age of PWID, and assess the association between injecting drug use (IDU) in young people and rates of injecting and sexual risk behaviours at the country level.

Methods: We derived data from a previously published global systematic review done in April, 2016 (and updated in June, 2017) on the percentage of young PWID, duration of IDU, average age of PWID, average age at IDU initiation, and the percentage of PWID reporting sexual and injecting risk behaviours. We also derived national development indicators from World Bank data. We estimated the percentage of young PWID for each country, using a random-effects meta-analysis (DerSimonian-Laird methodology) and generated pooled regional and global estimates for all indicators of  IDU  in  young  people.  We  used  univariable  and  multivariable  generalised  linear  models  to  test  for   associations between the age indicators and country urban population growth, youth unemployment percentage, the  percentage  of  PWID  who  are  female,  the  percentage  of  the  general  population  aged  15–24  years,  Gini  coefficient,  opioid substitution therapy coverage (per PWID per year), gross domestic product (GDP) per capita (US$1000), and  sexual and injecting risk behaviours.

Findings: In the original systematic review, data on age of PWID was reported in 741 studies across 93 countries.  Globally, 25·3% (95%  uncertainty  interval  [UI]  19·6–31·8)  of  PWID  were  aged  25  years  or  younger.  The highest percentage of young PWID resided in eastern Europe (43·4%, 95% UI 39·4–47·4), and the lowest percentage resided in the Middle East and north Africa (6·9%, 5·1–8·8). At the country level, in multivariable analysis higher GDP was associated with longer median injecting duration (0·11 years per $1000 GDP increase, 95% CI 0·04–0·18; p=0·002), and older median age of PWID (0·13 years per $1000 increase, 0·06–0·20; p<0·0001). Urban population growth was associated with higher age at IDU initiation (1·40 years per annual percentage change, 0·41–2·40). No associations were identified between indicators of IDU in young people  and  youth  unemployment,  Gini  coefficient,  or  opioid  substitution therapy coverage provision  at  the  country  level.  No associations were identified between injecting and sexual risk behaviours and age of PWID.

Interpretation: Variation in the age profile of PWID was associated with GDP and urbanisation.  Regions with the highest prevalence of young PWID (aged ≤25 years) had low coverage of interventions to prevent the spread of blood-borne viruses. Data quality highlights the need for improvements in monitoring of PWID populations.