We collected data on STEC O157 and non–O157 outbreaks that were linked to leafy greens during 2009–2018 from the following sources: Centers for Disease Control and Prevention (CDC) Foodborne Disease Outbreak Surveillance System (FDOSS; 2009–2017 only); internal CDC and PHAC databases used to manage multistate outbreak investigations; and PulseNet, the national molecular subtyping network for foodborne disease surveillance (16). We defined an outbreak as >2 similar illnesses in persons with a common exposure. Outbreaks for which STEC was listed as the single causative pathogen, with >2 culture-confirmed cases of infection, and for which leafy greens were listed as a suspected or confirmed source, were included in this report. HUS was identified by physician diagnosis.

Local, state or provincial, and federal health officials assessed 3 types of evidence (epidemiologic, traceback, and microbiologic) to determine outbreak sources during an investigation. For epidemiologic evidence, health officials interviewed ill persons to gather detailed information on foods they ate, determine whether any foods were reported more frequently than expected (compared with the FoodNet population survey [17]), and determine whether persons ate food from the same point of sale (e.g., grocery stores, restaurants) or event (collectively defined as a subcluster). For traceback evidence, officials collected and evaluated records documenting the movement of foods to and from all points in a distribution chain (e.g., receipts, grocery store shopper cards, restaurant rewards numbers, invoices, bills of lading) to determine whether there was a common point of contamination from at least two distinct points of sale. The Canada Food Inspection Agency (CFIA) conducted traceback to points of importation, based on methodology employed in Canada (18). For microbiologic evidence, officials sampled foods and environments of restaurants, production facilities, or growing areas suspected to be the source of outbreaks and conducted microbiologic testing for the outbreak strain. Food sources were classified as suspected or confirmed outbreak vehicles based on the evidence collected during the investigation. For outbreaks linked to a single event or meal, only 1 type of evidence (epidemiologic, traceback, or microbiologic) was needed to be considered confirmed. For multistate outbreaks, or outbreaks during which ill persons reported exposures in multiple venues, vehicles were classified as suspected if only 1 type of evidence was identified and confirmed if >2 types of evidence were identified.

Outbreaks that occurred in both the United States and in Canada were counted as a single outbreak if they occurred at the same time and had the same outbreak strain. Outbreaks were classified by the state or province where ill persons were exposed to leafy greens.

We calculated the outbreak duration as the number of days between the first and last illness onset dates. We defined the outbreak investigation lag as the number of days between the first illness onset date and the date the coordinating agency began its investigation. We compared median outbreak size by vehicle status using the Kruskal-Wallis test and leafy green type by vehicle status using the Fisher exact test. We defined seasonality using the date of first illness onset for each outbreak and divided the year into 4 periods: spring (March–May), summer (June–August), fall (September–November), and winter (December–February). Outbreak vehicles were categorized as leafy greens according to the Interagency Food Safety Analytics Collaboration categorization schema (19). Outbreak strains were characterized using 2-enzyme pulsed-field gel electrophoresis.

CDC, FDA, PHAC, US state and local, and Canadian provincial health departments described outbreaks via press releases, Internet, Facebook, and Twitter posts to inform the public of measures they could take to protect themselves. Data on outbreak announcements were collected from CDC, FDA, and PHAC. Additional announcements may have been posted by state or provincial and local health departments but were not captured in this report.

Over the past decade, multiple STEC outbreaks linked to leafy greens occurred in the United States and Canada, causing illness that was widespread and often severe. Most STEC outbreaks linked to leafy greens were caused by STEC O157, even though non-O157 STEC cause more sporadic US illnesses and are more frequently isolated from cattle (10); reasons for this discrepancy are unclear.

Despite year-round US leafy green production, 73% of STEC outbreaks linked to leafy greens began during the spring or fall. This seasonality was noted in a previous study of STEC O157 outbreaks (5); however, reasons for this seasonal pattern are unclear. Seasonal differences in consumption and production are one possible explanation. However, US data from 2007 (CDC FoodNet, unpub. data) and Canadian data from 2014–2015 (23) indicate that leafy green consumption changed little by month and did not show increases during the spring and fall. Data for leafy greens produced in 2009 showed some variation in domestic shipment volume by month but did not show an apparent increase in shipments in the spring and fall (14). Notably, the peak outbreak months in our report (October and April) coincided with the period when growers have historically used the short-term California Central Valley growing region to fill the gap in leafy green production between the California Central Coast and the desert regions of California, Arizona, and Mexico (24). We could not further assess any potential link between outbreak timing and harvest location because the movement of growing and harvesting operations varies by year and company, and there was a limited number of outbreaks during 2009–2018 for which growing locations were identified. Additional data on growing and harvesting practices, intrinsic factors of leafy greens that might make them susceptible to contamination, and the effect of climate or specific weather events during the growing seasons are needed to further assess seasonality of outbreaks.

Environmental assessments rarely occurred during investigations of leafy green outbreaks. Environmental assessments can occur only after epidemiologic and traceback investigations identify where implicated product is grown, processed, or distributed. Although they are resource intensive, these assessments can provide valuable insight into outbreak causes and identify possible areas to target for prevention efforts. Environmental assessments conducted during multiple leafy green investigations have suggested a possible link between product contamination and STEC contamination of nearby soil or water caused by cattle or wild pigs (25,26), dairy farms (27), or CAFOs (28). These findings build on 2 studies conducted in leafy green growing regions in California; one identified a higher prevalence of STEC O157 in a watershed near cattle (29), and another detected STEC O157 in beef cattle feces, mostly during dates in the spring and fall when leafy greens are typically grown, although the number of samples that yielded STEC O157 was small (13). Furthermore, some leafy green growing regions are home to large numbers of cattle: the Central Valley growing region (encompassing Fresno, Tulare, and Kings Counties) had >1.8 million head of cattle in 2016–2017, comprising nearly one third of all cattle in California (30). Conducting additional environmental assessments to better understand the relationship between cattle and leafy green growing locations, water, and the occurrence and timing of outbreaks may be beneficial.

More STEC outbreaks were linked to romaine than to any other type of leafy green, similar to an analysis of leafy greens–related incidents linked to California (31). More iceberg was harvested and available for purchase than romaine each year during 2009–2017, although romaine harvest and availability increased over time (32,33). The share of category dollars spent on iceberg and romaine was the same during 2012, higher for iceberg in 2013–2014, and higher for romaine in 2016–2017 (34). Together, these data suggest that even though romaine increased in popularity, it is unlikely that this alone explains why more STEC outbreaks in the past decade were linked to romaine than to any other leafy green. Because standard investigation questionnaires include questions about multiple leafy green types, such as spinach, iceberg, kale, and romaine (35), investigational bias toward romaine is unlikely. Romaine may have some characteristics that may make it more vulnerable to STEC contamination, including its shape and physiology (romaine is tall with loosely clumped leaves, open at the top; iceberg is smaller with compact leaves). Additional studies comparing the likelihood of STEC contamination and bacterial survival dynamics by leafy green type are warranted.

More than half of STEC outbreak investigations identified leafy greens as a suspected, rather than confirmed, source. Several characteristics of leafy green outbreaks make them inherently difficult to solve, and therefore challenging to implement timely interventions to reduce illness. The short shelf life of leafy greens (12–16 days) (36), the lag in identifying outbreaks (22 days), and the short duration of most outbreaks (21 days) all limit opportunities for investigators to interview ill persons in a timely fashion. This limitation can hamper hypothesis generation and limit opportunities to test leafy greens for contamination. Finally, because leafy greens, especially iceberg and romaine, are commonly consumed in the United States (17) and Canada (23), it can be difficult to show that they were eaten more often than expected by ill persons who are part of outbreaks. Establishing an epidemiologic link between cases and contaminated leafy greens often requires other corroborating pieces of evidence (e.g., brand or variety) to implicate a specific leafy green type. To help solve outbreaks, investigators have used successful strategies such as subcluster and purchase record analyses (37–39).

Traceback investigations for leafy green outbreaks are complex. First, product information from packaging is rarely available when an investigation begins. Therefore, ill persons are asked to remember crucial information needed to identify and trace leafy greens (e.g., purchase location and date, type/brand) instead of simply referring to an open package. Second, associating leafy greens at a point of sale location with a particular distribution lot can be challenging. Even though lot information may be available on the packaging for prepackaged leafy greens, points of sale may not record and track it after the packages are received. Investigators rely solely on records collected at each point in the distribution chain to determine the lots and source of a product, but they often lack the data elements needed to link lots of incoming shipments of products with lots of outgoing shipments. Finally, commingling of leafy greens from different farms throughout the distribution chain further complicates efforts to identify a single lot or source.

Complete, detailed records of transactions at each point along the fork-to-farm continuum are critical to accurately and quickly trace leafy greens during an outbreak investigation. Several strategies could increase the likelihood of success. Industry could assist by maintaining records that are consistently available, accurate, and complete. Retailers that sell leafy greens could consider developing systems to track lot and source information for leafy greens after they are received. Retailers may also wish to require producers be able to trace leafy greens and components of packaged mixes back to the farms from which they were harvested.

Several policies and recommendations were put into place before and after the study to improve the safety of leafy greens and prevent future outbreaks. In 2011, the US FDA Food Safety Modernization Act (FSMA) was signed into law (40). Under that law, in 2016, the Final Rule for Produce Safety went into effect, which established science-based minimum standards for the safe growing, harvesting, packing, and holding of US produce, including leafy greens. The first major compliance date for produce other than sprouts was in January 2018 (41). Routine regulatory inspections were set to begin in spring 2019, and compliance with agricultural water requirements were extended to become effective in 2022 (42). In 2007, the California and Arizona leafy greens industries each formed their own leafy greens products handler marketing agreement and enacted food safety recommendations (43) after a large 2006 STEC O157 outbreak linked to spinach (25). In response to the 2 large 2018 outbreaks, in 2019, California and Arizona modified their recommendations for leafy green growers, including increasing buffer zones between CAFOs and leafy green fields; requiring environmental assessments after severe weather events; requiring that all lot data be identified for products entering the marketplace; limiting or prohibiting the use of surface water for overhead irrigation of leafy greens before harvest; and requiring farmers to categorize sources of water, consider how it is applied to leafy greens, and test and sanitize it if needed (44–47). Future analyses should be conducted to assess the effect of these policies, recommendations, and any other implemented changes.

STEC outbreaks linked to leafy greens have continued to occur over the past decade. The combination of challenges investigators face during epidemiologic and traceback investigations of leafy greens make timely communication of actionable advice for consumers difficult. Despite challenges, results from leafy green outbreak investigations have led to changes in industry recommendations. However, knowledge gaps remain, including the drivers of the seasonality of leafy green outbreaks, and knowledge of why outbreaks are disproportionately linked to romaine lettuce. Investigators should work with federal and state health partners, the research community, the leafy green industry, and retailers to fill these knowledge gaps and collect additional information. Additional efforts should include identifying data points that would improve traceability of leafy greens during outbreaks. Collectively, these efforts can help inform prevention strategies to avoid or mitigate future outbreaks and lead to further changes in the way food is grown and processed, which could make leafy greens safer for the public to consume.