Data collection

The survey instrument (S3 Appendix) was designed and hosted on Typeform [12], an online survey platform. Participation in this survey was voluntary and respondents were informed they would not receive a reward for completion (for full statements, see S4 Appendix). Respondents self-selected their preferred language (from 10 options) and completed three demographics questions to check eligibility and city location. The survey questions included: 5 ’closed’ multiple choice questions to identify the ‘best’ and ‘worst’ parts of their cities (with the option to submit their own answer via an ‘other’ choice); to rank their city’s air quality on a Likert scale [13] from zero (‘awful & toxic’) to ten (‘amazing & fresh’); to suggest the main sources of AP in their city; and to indicate if they felt their city was becoming a better or worse place to live. Six ’open-ended’ questions collected long-form comments about the CYPP’s cities in general and the quality of the air. CYPP were asked to elaborate on previous answers and how they would improve their cities generally for young people and specifically regarding air quality.

Informed consent

Informed consent was gained from all survey respondents. This was through completion of an online form as approved by the London School of Hygiene and Tropical Medicine (LSHTM) Ethics Committee (see S4 Appendix for consent questions). Given the data collected by the survey was exclusively non-identifiable and would be instantly anonymous, the Ethics Committee agreed this research low risk. In agreement with the Ethics Committee, we did not seek to inform or obtain consent from parents or guardians of child respondents (who were all 13 years old or older).

Missing data

Only respondents with full demographic data who met the eligibility criteria, provided informed consent via an online form (approved by LSHTM Ethics committee—S4 Appendix), and successfully completed the survey were included in the full survey sample. An available case analysis method was used for each survey question, ensuring each question sub-population consisted of complete responses only. The demographics of subpopulation respondents were compared to the full sample’s demographics to identify any patterns of item non-response.

Where possible, imputation methods were used to limit non-response bias for age and location using the following methods. Where age was not reported or reported as an unrealistic value in excess of 100 or less than 13 years old, age was recoded as ‘missing’ and survey respondents were analysed together. Where location was not reported or only reported by region or country, probable location was imputed based on the city featured in the recruitment advert clicked on.

Data analysis

The data analysis plan (available on Data Compass [14]) was developed in advance of data collection. In short, descriptive statistics were completed for the full sample and subpopulations by city and for each survey question. The quantitative and qualitative survey responses were analysed for the full sample and results were stratified by age, whether the respondent was a young person or parent/expectant parent, and by the PM2.5 pollution in their city.

Where there were fewer than five responses from an individual city, these were clustered based on location linked to the advertisement they responded to. Countries were categorised as high or middle and low income as identified by the Organisation for Economic Co-operation and Development [15]. Parents (n = 509) and expectant parents (n = 283) were combined in analyses. Where relevant, ‘other’ responses to closed questions were coded against existing options, and when appropriate options didn’t exist in the survey instrument, answers were analysed thematically with the other qualitative answers.

Analysis of the quantitative data was completed using write Stata Statistical Software: Release 17 (College Station, TX: StataCorp LLC). The demographics (age and target group) of the full sample and each subpopulation by question were described using averages and frequencies. The sample was stratified using these demographics prior analysis of the quantitative questions. Responses to multiple choice questions were summarised using frequencies while the Likert scale responses were reported using the mean and standard deviation.

Reponses to the qualitative questions were analysed together using an interpretive grounded theory approach. Free text responses were translated into English by RJ using Google Translate and were not edited for grammar or punctuation. Quotations that included identifiable information were redacted before analysis. These responses were coded (by RJ) in NVivo 12 [16] using initial open coding methods to identify and group important words or groups of words, with a focus on the underlying meaning of responses. Coding was regularly discussed and reviewed with RH, who also reviewed open-ended questions. Where possible, in-vivo codes were used to capture the participant’s words as representative of a broader concept. Intermediate codes were created to identify core categories drawing on initial coding. Analysis was undertaken simultaneously with data coding to look for emerging concepts and groups of themes through a series of meetings between RJ and RH, and the wider team. RJ and RH are both public health and climate change researchers. RH’s particular expertise is in AP and early childhood development, while RJ focuses on food systems and planetary health. The expertise of the wider team comprises pediatric medicine, epidemiology, mathematical modelling, maternal health, and research communications. Where data saturation became apparent and no new codes or groupings of codes were identifiable, coding was ceased. Finally, a conceptual framework was developed to synthesise the themes and sub-themes which emerged.

Patient and public involvement

This project involved local collaborators in Kenya and Dar Es Salaam. Our partner in Kenya and Dar Es Salaam, Shujaaz Inc., contributed to survey method pretesting and translation into the Sheng local dialect of Swahili. They additionally contributed to recruitment to our online survey, through neutral pushes to their viewer-base. Finally, they assisted in dissemination of research findings throughout East Africa and at a variety of events hosted by LSHTM. Additional collaborations involved the Zimbabwe LSHTM Research Partnership, and a group of young people throughout the world. These collaborations similarly occurred during the research dissemination, with partners contributing to several LSHTM panel-events both in-person in the United Kingdom at the COP26 and Conference of Youth (COY) 16 climate change conferences and online.

Ethical approval

Ethical approval was granted by the LSHTM Research Ethics Committee (LSHTM Climate change, cities, child and young people’s health: children and young people survey and engagement, reference 26313). In addition, municipal authorities in all 16 focal cities were informed about the planned survey and were invited to provide feedback and offered a briefing on the results.

Overall perceptions about their cities

Initially, respondents were asked about their cities in general, followed by specific questions on air quality. The most common “best” aspects reported were proximity to family (reported by 30%); the diversity of activities (27%); and proximity to school or work (27%); the most common “worst” things were traffic congestion (reported by 49%); pollution (39%); and a shortage of work opportunities (23%) (Fig 3). No trends were observed when the best and worst aspects of cities were stratified by age, PM_2.5 quartile, or respondent group (S10 and S11 Appendices). 43% of the sample reported that their cities were “improving”, 23% said there was no change and 34% reported their cities were getting worse (S12 Appendix).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. The percentage of n = 2,939 and n = 3,051 respondents that reported each item within the top 3 ’best’ and ’worst’ aspects of their cities.

https://doi.org/10.1371/journal.pgph.0000963.g003

Perceptions on air quality and major causes of AP

The mean air quality rating (on a scale of 0 (‘awful & toxic’) to 10 (‘amazing & fresh’) reported by 3,054 respondents was 5.1 (sd 2.9). There was no obvious association between self-rated air quality and quartile of air quality of cities; the mean score in the least polluted cities was 4.71 (SD 2.9) and in the most polluted was 6.01. No trends were identified when stratifying by age or respondent group (S13 Appendix).

The most commonly reported sources of AP were motor transport, factories, rubbish burning and construction/building work. Agriculture/farming, household cooking and household heating were all reported by smaller proportions of respondents (2%, 5% and 5% respectively); all considerably lower than AP blown in from outside the city (8%). ’Other’ responses included natural sources such as dirt and dust, but were mostly based on human actions (burning of waste/sewage) and specific man-made objects (aeroplanes, plastic shopping bags). No clear patterns were identified when stratifying the major sources of AP by PM_2.5 quartile, major city, eligibility criteria, or age of respondent (S14 Appendix).

Thematic analysis of open-ended questions

Three sets of themes emerged from the analysis of qualitative survey responses. An overall summary of the three themes and their respective subthemes is illustrated in Fig 4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Summary of the thematic analysis findings.

https://doi.org/10.1371/journal.pgph.0000963.g004

The first theme that emerged was a set of specific ideas or asks representing distinct changes or policy priorities to curb AP and improve cities more generally for young people. Sub-themes identified included ‘city design and space’, ‘urban mobility’, ‘health’, ‘education’, ‘skills and jobs’, and ‘other basic services’. The breakdown of these ideas into sectors with exemplary quotes is provided in S15 Appendix.

A second emerging theme concerned structural barriers that need to be addressed before these specific ideas and asks can be achieved. These subthemes were felt to represent broader, more fundamental issues faced by cities and were ‘inequality’, ‘corruption and bad governance’, ‘lack/absence of young people in decision making’, and ‘a lack of consciousness for the environment’, listed with illustrative quotes in S16 Appendix.

The third set of themes described Guiding principles for how to act to address the structural barriers and to achieve the specific ideas and asks of urban young people. These principles were to ‘be ambitious and creative’, ‘engage young people in meaningful ways’, and to ‘understand and tackle inequalities’. Illustrative quotes for each of these are included in Box 1.

Strengths of this study

The use of social media recruitment and online surveying garnered a large number of responses, from a broad variety of cities. A combination of closed and open-ended questions allowed both quantitative and qualitative insights to be drawn, enabling us to explore where AP sits amongst a broader set of concerns and a rich set of underlying themes and ideas.

Recruitment through social media may have allowed for the inclusion of people who are not easily recruited through ‘analogue’ methods or are often systematically excluded from research panels (e.g. under 18, lower income, lower consumption groups). Additionally, the online survey instrument could be directly completed by young people, rather than by their parent/carer, or their school which limited misclassification of ideas which may occur when provided by a proxy adult.

The survey instrument was available for completion in 10 languages and was optimised using iterative pre-testing and translation checking by native speakers. The selected languages enabled global participation, as they are spoken as a first language by 40% of the global population, and are all (excluding Swahili) in the top 22 most commonly spoken languages worldwide [17].

Limitations

Selection bias is likely to have occurred due to a combination of factors. Firstly, using online recruitment data collection systematically excludes those who do not have access to the internet, a computer or mobile device, and those who do not use social media platforms, meaning that our sample may have been biased towards more affluent people who can afford to be online and utilise such devices, and those who spend more time on social media.

Secondly, the costs per click for online recruitment advertisements varied by location, and those areas with lower cost per click were disproportionately advertised to due to the algorithms employed by the social media networks (despite attempts to mitigate this). The sample is therefore biased towards low- and middle-income countries, more polluted cities and especially those in South Asia (48% of the sample was from South Asia and over half the South Asian population from Dhaka, Bangladesh). Although the recruitment adverts were deliberately ‘generic’, aiming to appeal to all CYPP living in the city and the survey was short to reduce the risk of drop out, the topic of the survey is likely to have appealed more to certain groups of people (those who are interested in the climate/environment/urban living/child health).

Finally, though young people were involved in the development and pre-testing of the survey questions and discussion and dissemination of the results, they were not involved in the formal analysis or write up of the survey results.

Putting this study into context

This survey was one of the largest, most geographically diverse of its kind, seeking to highlight the perspectives of global CYPP on urban living and AP. Several surveys were identified that more generally sought out the priorities of urban youth for improving how they live, work and play in cities [8–11], though none of these surveys explicitly sought CYPP perspectives on the air they breathe, or their recommendations for improving their city’s air quality. Additionally, though several of these surveys were comparable in size [8–10] (n = 1,000, n = 2,002, and n = 9,000 respectively), only two included the voice of CYPP in low- and middle-income countries (LMICs) [10, 11], and neither provide a rich narrative of the challenges faced by CYPP in LMICs nor specific and discernable recommendations.

One survey of 1000 young people aged 12–19 in London identified some similar priorities for city improvements, including creating “car free streets” and “more beautiful and green places” [8] both of which emerged from our study also. In another survey of 2,002 young people aged 16–24 in London, 30% of respondents ranked AP in the top three most pressing issues to be addressed by their local authority [9]. AP was the fourth most prioritized issue out of all social, environmental, and economic issues. Within the environmental issues, improving air quality was the second-most prioritised issue, following only improvements in recycling and waste reduction. Although these young people prioritised environmental issues and independently demonstrated a level of environmental consciousness, they also noted the need for increased environmental education, corroborating a systemic lack of environmental consciousness remains a barrier that needs to be addressed in cities.

A global survey of 9,000 young people aged 15–29 highlighted some themes consistent with our survey, most notably the importance of promoting environmental awareness alongside delivery of specific policies [10]. Another study that included both LMICs and high-income countries (HICs), identified similar barriers young people aged 12–24 face to living sustainably to those reported in our survey. These included growing inequality and poor availability of public services and safe, efficient public transport [10].

Implications of this research for policy makers

There is obvious need for acceleration of efforts to address urban air pollution: the importance which young people place on this issue is clear. This is not surprising in the context of school closures due to extreme pollution events in South Asia [18, 19]. These extreme events, in combination with the known health consequences of AP, led UNICEF to declare climate change and AP to be a child’s rights crisis (3)

Secondly, the rich insights and ideas shared by participants in this study suggest that policy makers should make greater effort to engage CYPP in urban and air pollution policy discussions. Both Rights-based and pragmatic arguments for this can be convincingly made given the United Nations Convention of the Rights of the Child (which asserts their Right to be involved in developing solutions to the challenges they face [20] and the creative ideas they report in this study).

Unanswered questions and future research

Future research into this issue could employ better methods to improve sample representativeness (e.g. combining online and offline surveying with particular efforts to include marginalised communities (e.g. homeless and migrant youth and those in informal settlements)) as air pollution levels and youth’s awareness of air pollution will differ depending on the social infrastructure, and other determinants. Further research on the experience of urban youth is needed across all socio-economic levels, races, and must include cities of all economic statuses and levels of urbanization.

Additionally, richer perspectives could be gained from collecting narrative via focus groups or urban co-design workshops with CYPP in cities. Co-development and leadership of these focus groups with CYPP would allow for deeper and more genuine insights to be collected regarding the lived experience of urban children.

Finally, in this time of rapidly changing climate, future research is needed to understand the perspectives of urban CYPP on other environmental health and climate-related issues, including, for example, exposure to urban heat, different models of urban mobility and shifting urban food systems. The health and non-health effects of these shifts must also be understood, given the impact they are likely to have on the health of the next generation.

Data storage and sharing

All responses were stored on the Typeform.com servers (hosted in a Virtual Private Cloud by Amazon Web Services) and processed and analysed on password-protected encrypted devices. All data collection tools, anonymised raw data, and data analysis code are available on the LSHTM Data Compass Site [14] mirrored on Figshare [21]. The dataset has been assigned a unique DataCite DOI [22]. Confidentiality of survey responses was maintained throughout; no personally identifiable data was collected. Age in years was collected, and the location was recorded at the city level. In addition, no IP address or other locator/identifiers (like cookies) were captured.