Main

Over 8.3 billion metric tonnes of plastics have been produced since the 1950s and proliferation in their use has been exponential1. The agricultural sector uses 6.5 million tons of plastic film mulch annually2; food and drink packaging alone accounts for around 16% of the total plastic production over the past 70 years3. Single-use plastics continue to expand across food systems globally—although circular economy policies seek to reduce their source, repurpose their waste and halt their pollution, such strategies are likely to have consequences across wider sectors with overlapping implications for health, the environment, food security and economic outcomes. Scientific evidence on these broader effects of food system plastics is rapidly growing, but it remains fragmented across disciplines and impact domains.

Plastics are low-cost, versatile, lightweight and durable4. Plastic food packaging acts as a barrier to contamination5, and supports household economics and food security by minimizing post-harvest losses, extending shelf-life and storage capacity5, and agricultural plastics can reduce chemical fumigant use in farming as well as transportation fuel consumption6,7. However, global food waste has increased alongside plastic packaging use3. Macro-, micro- and nanoplastics have been detected in Arctic snow, mountain air and the tissue of marine species8,9,10. The accumulation of microplastics in food chains and leaching of harmful substances, such as bisphenols and phthalates, into food and drink products is of increasing concern—although their potential health effects remain unclear11,12.

To support evidence-based policy actions and inform evaluations of potential trade-offs, the extent, range and nature of evidence from disciplines across food system sectors need to be systematically synthesized. Here we use a systematic scoping review of plastics in the food system to describe the research landscape, highlight gaps, generate new research questions and identify mature evidence. The approach taken here minimizes selection bias and ensures that evidence maps are comprehensive across different disciplines, which include materials sciences, public health, agricultural sciences, food technology, nutrition, economics and environmental sciences. We developed an interactive open-access evidence map that describes the extent (volume of research), range (variety of exposure–outcome relationships) and nature (study characteristics) of quantifiable evidence on the effects of food system plastics, in which exposure to any of the seven categories of common plastics used within any stage of the food system, in relation to theorized intermediate and final outcomes relevant to the domains of Human Health, household food security and/or economics (Food Security/Economics) and the natural environment (Environment) can be viewed (Fig. 1, Box 1 and Methods)13,14,15.

Results

We returned 49,850 unique results (excluding duplicates), of which 3,362 records were included for analysis in our review (Fig. 2 and the interactive evidence map available at https://anh-academy.org/foodplastics_EGM.html).

Fig. 1: The logic model describes potential relationships between food systems’ uses of plastics and the intermediate and final outcomes.
figure 1

Non-exhaustive examples of our considered intermediate and final outcomes appear under the impact domains of Human Health, Food Security/Economics and Environment. The food system is adapted from Turner et al.13 and the Food and Agriculture Organization of the United Nations14.

Full size image
Fig. 2
figure 2

Preferred reporting items for systematic reviews and meta-analyses (PRISMA) flow chart.

Full size image

Extent of the evidence

Publications per year more than quadrupled from 2000 to 2018, with a similar volume of studies in middle-income countries (MICs) (50.1%) and high-income countries (HICs) (48.3%), but little evidence from low-income countries (LICs) (1.6%) where plastic usage is rapidly growing. China, the United States of America, India, Italy and Spain contributed to almost half of the evidence in our review. Most studies explored plastics used in Agricultural Production, followed by Processing, Storage and Distribution, but only four considered those used in Waste Disposal (Fig. 3a). Plastic types were commonly unspecified or polyethylene (PE) (n = 1,137 and 1,209 studies, respectively), followed by polypropylene (PP) (n = 574), low-density polyethylene (LDPE) (n = 453), miscellaneous plastics (n = 423) (Box 1), polyethylene terephthalate (PET) (n = 319), high-density polyethylene (HDPE) (n = 214), polyvinyl chloride (PVC) (n = 192) and polystyrene, styrene or Styrofoam (PS) (n = 157 studies, <5%).

Fig. 3: Food system subsectors and outcome domains covered in eligible studies for the review.
figure 3

a,b, Venn diagrams of the volume of literature found according to food system subsectors (a) and by outcome domain (b) for all the included studies (n = 3,362).

Source data

Full size image

Figure 4 and the interactive evidence map show the number of studies by plastic use within each food system subsector and across three impact domains. Plastic mulch, polytunnels and greenhouses predominantly made of unspecified plastic or PE accounted for 89% (n = 1,666 studies) in Agricultural Production, whereas fishing gear, a key input into production activities of the food system, featured least frequently in this subsector (n = 57, 3%). In Processing, Storage and Distribution, plastic packaging primarily made of PE, PP and LDPE was most frequently explored (n = 930, 83.3%). Similarly, in Retail, plastic packaging predominated (n = 313, 90.5%). Grocery bags—a focus of policies around the world—were studied in just 11 of the included publications. Within Household Consumption, pre-packaged food items and domestic products, such as infant feeding bottles and storage containers, featured equally, whereas kitchen surfaces and equipment rarely featured. Refuse sacks and compost containers were the only uses of plastics explored in Waste Disposal.

Fig. 4: Plastic use (exposure) and outcome categories under the three impact domains.
figure 4

The heat map cross-references all the included studies (n = 3,362) to show the number of studies that investigated specific uses of plastic within the food system alongside specific outcome categories under the Human Health, Food Security/Economics and Environment impact domains. Impact domains are split by vertical dashed lines into intermediate (left) and final (right) outcomes.

Source data

Full size image

Over 75% of studies contained at least one outcome that related to individual or Food Security/Economics (n = 2,546), 48% to the Human Health (n = 1,602) and 8% to the Environment (n = 282) outcomes (Fig. 3b), although this category experienced the greatest relative increase in publications between 2000 and 2018 (5,100%). Across all the impact domains, the vast majority of studies presented data that falls under intermediate outcomes, with few demonstrating a full exposure-to-impact pathway, in particular where the Food Security/Economics category was concerned. Within the Human Health category, most studies considered the nutrient content of crops, food or beverages (n = 885, 55.2%), with plastic and chemical contamination of food and estimated human exposure to the same measured in 438 (27.3%) and 134 (8.4%) studies, respectively. We found 47 (2.9%) studies that measured the actual presence of plastic-related chemicals in the human body, using indicators such as urinary concentration of bisphenol A (BPA) and phthalates. Finally, 39 (2.4%) studies investigated final outcomes that related to health conditions, which included diabetes mellitus, autism spectrum disorder, Parkinson’s disease and broader measures of disability-adjusted life years.

Range of evidence

Plastic uses and impacts

We found a range of exposure–outcome relationships in the literature (Fig. 4 and the interactive evidence map). The most frequent relationships were among perhaps the ‘less visible’ food system subsectors, such as agricultural plastics used for crop production, and outcomes for on-farm productivity and efficiency, which included yields (n = 1,343), climate control (n = 903), pest and disease control (n = 371) and crop nutrient content (n = 354). In contrast, soil and crop contamination by plastics or related chemicals (n = 45), subsequent human exposure (n = 15) and Environment outcomes were less frequently studied.

In Processing, Storage and Distribution, there was a focus on intermediate Human Health and Food Security/Economics outcomes, which included plastic packaging that affected the nutrient, bacterial or chemical content of food and drink items (n = 705), and a range of quality-control and shelf-life related outcomes (n = 574). Nine studies considered the effect of packaging on final Human Health outcomes, but we found no evidence that linked plastic use at this food system subsector to final Food Security/Economics outcomes, such as changes in income or food security indicators. Over 100 studies in this subsector assessed Environment outcomes, but only three of these quantified the final impacts on individual wildlife and population injury, illness and deaths.

For both the Retail and Household Consumption subsectors, pre-packaged food or drink and domestic items, such as infant feeding bottles, were linked to intermediate outcomes, such as human exposure to plastics and chemicals (n = 97) and their presence inside the human body (n = 47), as well as to our final outcomes relating to health or disease state and physiological changes (n = 25). These items were less commonly linked to Food Security/Economics outcomes (n = 36 overall) but did feature more frequently at the Environment level, particularly concerning air pollution and greenhouse gas emissions (n = 82). Although very few studies investigated plastics used in Waste Disposal, at least one explored outcomes under each of the domains.

Specific plastic types and their impacts

Every type of plastic considered in our review has been studied the literature, with an extensive range of outcomes captured (Fig. 5). Some plastic types were more frequently linked to certain outcomes than others: PE and PP were frequently explored in relation to effects on the nutrient or bacterial content of crops and food (n = 445 and 274, respectively), but only one study considered the effects of PP (commonly used in infant feeding bottles) on the presence of plastics or associated chemicals in the human body. Among the studies that looked at plastics and chemicals in the human body, as well as human health states and/or physiological changes, the most commonly explored types were unspecified plastics, miscellaneous plastics and PET. Despite finding evidence of each major plastic type in our sample, a lack of chemical specification of the plastic exposures was very common.

Fig. 5: Number of publications by plastic type and impact domain.
figure 5

Source data

Full size image

Nature of the evidence

Food system plastics have been investigated using many study designs, which include experimental, non-experimental and mixed methods (Fig. 6). The majority (85%) of publications employed experimental methods (n = 2,865 studies), mostly to examine plastics used in Agricultural Production (n = 1,713). Methods such as observational studies, modelling and life-cycle assessments were used in 591 publications (18%). We found just eight meta-analyses (1.4%), all of which assessed plastic mulching, and three longitudinal cohort studies (<1%). In total, the review captured 14 (<1%) case studies or descriptive cross-sectional studies in which post-mortem or the diagnosis of injury, illness or entanglement were linked to food system plastics.

Fig. 6: Frequency of study designs used to investigate the effects of plastic use in different food system subsectors.
figure 6

Sunburst diagram of all the included studies (n = 3,362) showing the frequencies of study designs (experimental versus non-experimental) employed to investigate the effects of plastic used at different levels of the food system and the corresponding frequencies of investigated Human Health, Food Security/Economics or Environment outcomes.

Source data

Full size image

Discussion

We synthesized research across diverse disciplines to characterize the extent, range and nature of evidence on food system plastics relevant to human health, food security, economic and environmental outcomes. Diverse forms of evidence exist on these interlinkages, but there is a distinct lack of systematic meta-analyses in the literature that could support policy-making and decision-making on plastics in the food system. The vast majority of evidence since 2000 (and 98% of the environmental studies) focuses on MICs and HICs, with little evidence from LICs. Further work is needed to ascertain whether these research trends are consistent with a plastics-oriented environmental Kuznets curve, which indicates a rise and fall in plastic pollution relative to economic development and technological research investment16.

Despite a rapid increase in planetary health research and ongoing discussion of consumer plastics use (for example, straws and coffee cups, for which we found no specifically targeted studies), only 8% of the included publications captured environmental outcomes. This may be a result of environmental research on plastics using methods that did not fit our inclusion criteria (for example, pollution prevalence studies with no comparator) or that many studies did not disaggregate beyond terms such as ‘plastic debris’ or ‘fishing gear’, which presents challenges in ascertaining either the source or the material. Additionally, micro- and nanoplastic particles are often too small to be explicitly linked to a food system use or specific sources. These represent considerable problems for accountability as negative environmental externalities of food system activities and actors become untraceable and ungovernable17.

Intermediate Food Security/Economics outcomes represented the greatest number of studies in our assessment, which included many studies that measured the effects of plastics on agricultural yields and on-farm soil conditions, evidence that may be critical in addressing potential climate-related yield downturns among key staple crops18,19. We found evidence to suggest that plastics may provide mitigation and adaptation functions (for example, mulching, tunnels and greenhouses) to improve agricultural productivity and resource efficiency, reduce fertilizer and pesticide use, and mediate CO2 emissions20,21. Nevertheless, we observed an imbalance between research on agricultural productivity versus the quality and safety of crops—these topics require more research or synthesis, particularly in light of debates around possible declines of key food-group nutrients22. Few studies investigated the pollution or contamination of crops and soil by agricultural plastics and their associated intentionally and non-intentionally added substances—the one meta-analysis we found for the latter23 cautioned against short-term gains versus long-term sustainability should environmental measures not be implemented. These asymmetries may echo ‘quantity over quality’ trends, in which yields, proteins and calories are prioritized over micronutrient-dense crops, production diversity and healthy diets—with consequences for hidden hunger24.

A large number of studies explored human health, including bacteria control, pathogen barriers and nutrient quality, but few studies presented changes measured or modelled in humans. This gap needs to be filled in all regions—particularly LICs—in which different challenges may drive plastic-use and context-specific trade-offs. Research is needed to model the consequences of removing or substituting plastics in settings that suffer high postharvest losses, persistent foodborne disease and limited infrastructure to support hygienic food environments. Despite these contextual food-safety challenges, we found little research in LICs compared with that in MICs. Similar trends exist among studies that quantify human exposure to food system plastics and associated chemicals, and few explored workplace exposures—33 risk assessments in MIC or HICs and none in LICs. The studies found on plastic packaging, which included the leaching or migration of chemicals and non-intentionally added substances into food or drink, were probably catalysed by research and policy around the effects of plastic-related substances, such as BPA, shown to be dangerous at low doses in young children, with carcinogenic and endocrine disrupting potential in adults25,26,27. Infant feeding bottles that contain BPA are now banned in most industrialized countries, yet in many LICs there remain no such restrictions and little emerging evidence explores the consequences of this policy divergence.

The human health implications of living in the ‘Plasticene’28 appear to be acknowledged at certain supranational levels, as reflected in risk assessments such as the World Health Organization’s report on microplastics in drinking water12 and the European Food Safety Authority’s quantification of human exposure to BPA29. This review uncovered research that explores exposures to individual uses and groupings of plastics, but we found a lack of interlinking research across food system subsectors, plastic types, uses and outcomes. Thus, we echo wider concerns regarding cumulative exposures to chemicals that emanate from different sectors, which, although different, may have similar effects on the body30,31. With moves towards circular economies, in which postconsumer materials are reused or repurposed, potentially toxic substances—known and unknown—may be inadvertently introduced into foodstuffs. Plastics recycled between food system subsectors (for example, repurposing single-polymer agricultural mulch for food contact materials), may be accompanied by those imported from other economic sectors (for example, repurposing electronic waste for food contact materials). Although this is generally illegal, evidence for these practices exists in several settings and is of particular concern given the divergent sectoral regulatory environments32,33.

Implications for research and policy

This systematic scoping review includes research across disciplines, and reveals important evidence gaps and highlights datasets that may be suitable for targeted systematic reviews and meta-analyses in different exposure–outcome domains. The interactive map and framework enable broad communities of research, practice and policy to explore existing evidence, and plan work that is collaborative and/or complementary. We hope this will be used to enhance interdisciplinary approaches, inform funding priorities and guide cross-sectoral research agendas to generate the much-needed evidence base for policy decisions across key human health, economic well-being and environmental outcomes.

The four Rs, remove, reduce, reuse and recycle/repurpose, are central among policy responses to the plastic problem and transitions towards circular economies34. This review can be used to locate evidence that concerns how, where and with what effects different plastics are used and studied across food systems, and enable further enquiries around substitution or discontinuation. However, our review also suggests that the type of cross-cutting evidence and analysis needed to inform such policies and understand trade-offs across different subsectors and outcomes is either lacking or not yet synthesized adequately. Despite numerous initiatives around food-related plastics35 (for example, Bangladesh’s plastic bag ban or Canada’s pledge to remove single-use plastics), we found little analysis of their associated intentional and unintentional beneficial or harmful impacts—particularly around human health and food security.

The COVID-19 pandemic is driving large increases in single-use plastics, particularly in food retail and consumption, as producers and consumers turn to perceivably more hygienic, disposable options and to online deliveries36,37. Recycling activities are reduced as many governments struggle with operations or seek to protect workers from viral transmission38. Coupled with the recent oil price decline—and thus the cost of virgin plastics—these developments pose substantial new questions for researchers and policymakers, and also emphasize the bidirectional and interconnecting pathways between food system plastics, human health, food security and the environment. Debates around ‘building back better’39,40 in a post COVID-19 world must take into account how plastics are used across food systems and the range of outcomes they may contribute to—both beneficial and harmful.

Conclusion

This systematic scoping review reveals that plastics are widely used and researched across food systems around the world, with multiple outcomes considered in a rapidly expanding evidence landscape that is both diverse and imbalanced. Exploring these materials through a food system lens and across interdependent sustainability and impact domains illuminates how many trade-offs may exist when transitioning towards circular economies. The extent, range and nature of this evidence highlights a mismatch between the prevailing discourse around consumer-level plastics and research agendas more focused on production, processing and storage. The types of collaborative research needed to account for final outcomes—particularly among human populations – is urgently needed to address food systems’ ‘plastic problem’.

The findings from this systematic scoping review provide a common roadmap for the formulation of research strategies by diverse stakeholder groups, such as researchers, funders, international agencies and government bodies. It enables disparate research communities to identify gaps and generate the robust cross-disciplinary evidence needed to inform policies to remove, reduce, reuse, recycle or repurpose food system plastics. The interactive evidence gap map is a customizable, shared tool to foster this interdisciplinary collaboration, to augment new research questions, identify mature evidence for systematic reviews and avoid duplication—with implications for sustainable, safe, affordable and nutritious diets (Box 2).

Methods

Scoping reviews are valuable for systematically synthesizing broad-based evidence on topics of intersectoral and interdisciplinary relevance, building bridges and bringing coherence to a diverse evidence base, but they do not assess the direction, quality or bias of evidence41.

We followed the PRISMA extension for Scoping Reviews guidelines41 (see Supplementary Methods 1 for a full checklist). We developed a protocol in liaison with technical experts and the Campbell Collaboration15. Our logic model (Fig. 1) guided the review and describes possible exposure–outcome relationships we expected to find in the literature, and also provides illustrative examples. For the purpose of this study, we use the term ‘impact’ to cover the effects or outcomes of plastics, for which terminology may differ among disciplines.

The logic model breaks down these impact domains into intermediate and final outcomes. Intermediate outcomes constitute either plausible steps on a pathway to a final outcome, such as contamination of food on the pathway to human health, or a grouping of factors that may precede either irreversible or substantially harmful outcomes within that domain, such as greenhouse gas emissions under Environment.

Eligibility criteria

The primary inclusion criteria for this review was a combined exposure of (1) any food system subsector and (2) any specified type of plastic used within it. Outcomes described in the literature must have fallen within at least one of the specified outcome groupings of human health, the environment or food security/economics, referred to as ‘impact domains’. Box 1 outlines the key characteristics of our exposure and outcome eligibility criteria.

Any experimental or non-experimental study design was eligible for inclusion provided the study presented quantitative data in relation to a comparator group. Possible comparator group(s) included a control, exposure to other materials (including other types of plastic), a dose–response relationship (including time exposure) or different environmental conditions or population characteristics that may alter outcomes. Case studies or descriptive cross-sectional studies with a clear cause-of-death diagnosis, injury or entanglement due to plastic, modelling (including risk assessments) and life-cycle assessments were exempt from having a comparator group. We considered any population except for plastic itself. We accepted any geographical location found in studies published from 2000 onwards in the English language. Our date range is due to major evolutions in food systems, plastic use, diets and circular economy legislation, as detailed in our protocol.

Search strategy, data extraction and analysis

We devised the search strategy in consultation with a search specialist, technical experts and librarians at the London School of Hygiene and Tropical Medicine, applying a systematic search of over 200 relevant terms to nine scientific databases (Agris, CAB Abstracts, CAB Global, Campbell Library, Cochrane Database of Systematic Reviews, Epistemonikos, GreenFile, Web of Science and Scopus) as well as 15 grey literature sources between December 2018 and February 2019. A full list of search terms and sources are given in our protocol15 and in Supplementary Methods 2.

We used EPPI-Reviewer 4 for the reference management, screening, data extraction and analysis. Records were double screened at the title and abstract, with disagreements resolved by a third reviewer, and single screened at the full text stage after training and consistency checks were performed on 10% of the records. The data-extraction form, piloted extensively, included publication date, study location, study design, type of plastic, food system subsector and specific function of the plastic, outcome category and funding source (Supplementary Methods 3). Categories were not mutually exclusive, and a single study could have multiple codes per category. Data extraction was carried out by three researchers with regular training updates and thematic consistency checks on 10% of all the included records. Texts that were not located for a full text screening were coded according to information provided in the abstract so that bias could be highlighted.

We used EPPI-Reviewer 4 and Microsoft Excel for cross-tabulation, frequency distributions and mapping data against our logic model and heat map. This map is also available online as an interactive resource which provides a full list of references for all studies included in this review. We grouped countries according to the World Bank’s 2019 assessment of economic incomes for LICs, MICs and HICs)42.

Strengths and limitations

The breadth of this review posed challenges to accommodating diverse research standards and methods from different disciplines. Some studies did not fulfil our inclusion criteria—for example, observational studies of plastics in natural environments often failed to link plastic explicitly to the food system or lacked a suitable comparator. Although many new bio- or plant-based plastics are under development, given their still early development, we utilized the seven major Resin Identification Codes for our selection criteria as they are widely used within the food system. We included non-specified and composite plastics if shown to include a plastic from this list, as well as epoxy resins—a family of plastics commonly used in aluminium food tins; however, some plastics, such as melamine, were not included. Our search was limited to English publications only and, although we demonstrate a wide geographical spread, a broader language inclusion would have yielded additional results.

We utilized distinctions between intermediate and final outcomes as well as where plastic uses occur, which sometimes transcended the food system subsectors. We employed an adapted food system framework, for which we acknowledge many others exist. This approach is subjective and debatable, depending on the particular discipline or point of view of the reader. To address this, we presented data in the interactive map according to our conceptualization and in disaggregated forms so they may be reorganized to suit the requirements of different fields or researchers. Our screening and data extraction strategy sought high levels of accuracy alongside feasibility in managing a large body of research—a much-debated balancing act among systematic review researchers43. Although errors may occur, our detailed protocol15 demonstrates that systematic misclassification of results is unlikely. Finally, our approach does not lend to causal statements between food system plastic exposures and human health, the environment or food security/economics. This is in line with PRISMA scoping review guidelines, which recommend the presentation of current research landscapes to guide future systematic reviews and meta-analyses.