Nutrition and listeriosis during pregnancy: a systematic review

Listeriosis is a rare but severe foodborne illness which is more common in populations such as pregnant women, and can result in serious complications including miscarriage, prematurity, maternal and neonatal sepsis, and death in the newborn. Population recommendations exist for specific foods and food preparation practices to reduce listeriosis risk during pregnancy. The aim of the present systematic review was to assess the association between listeriosis and these practices during pregnancy to confirm appropriateness of these recommendations. We searched MEDLINE, Embase, CINAHL Plus, Web of Science Core Collection, included articles’ references, and contacted clinical experts. All databases were searched until July 2017. Case–control and cohort studies were included which assessed pregnant women or their newborn offspring with known listeriosis status and a nutritional exposure consistent with international population recommendations for minimising listeriosis. Outcomes included listeriosis with or without pregnancy outcomes. Risk of bias was assessed through the Newcastle–Ottawa Scale. Results were described narratively due to clinical heterogeneity in differences in nutritional exposures. Eleven articles comprising case–control or cross-sectional studies met the inclusion criteria. Cases of maternal, fetal or neonate listeriosis were more likely to have consumed high-risk dairy products, meat products or some fruits during pregnancy in comparison with women without listeriosis. Cases of listeriosis were more likely to have consumed foods that are highlighted in population guidelines to avoid to minimise listeriosis in comparison with those without listeriosis during pregnancy. Further research is warranted assessing means of improving the reach, uptake and generalisability of population guidelines for reducing listeriosis during pregnancy.

Listeria monocytogenes is a foodborne pathogen which causes listeriosis, a systemic illness that can cause symptoms ranging from gastroenteritis to meningitis and severe sepsis. L. monocytogenes is environmentally widespread and able to contaminate a range of foods (1) , accounting for 19 and 17 % of the known causes of foodborne disease-related deaths in the USA and France, respectively (2,3) . Overall, listeriosis was estimated to have caused 23 150 illnesses, 5463 deaths and 172 823 disability-adjusted life years worldwide in 2010 (4) . L. monocytogenes is relatively resistant to diverse environments, which enables it to survive food processing and grow in refrigerated or ready-to-eat chilled foods (5) . It is therefore difficult to control in the food production environment, and consequently listeriosis may arise from food contamination outbreaks or from sporadic consumption of contaminated food (6,7) . Recommendations to avoid listeriosis differ slightly between countries but generally include avoiding high-risk foods which are susceptible to L. monocytogenes contamination, such as pre-prepared or pre-packaged salads, pre-prepared foods, soft, semi-soft and surface-ripened cheeses, processed meats and unpasteurised dairy products, and practising appropriate food hygiene practices (8,9) .
Groups at higher risk of contracting listeriosis include the immunocompromised, the elderly, pregnant women and neonates. Pregnant women appear 10-20 times more likely to contract listeriosis (1) compared with the general population, possibly owing to a down-regulation of cellular immunity (1) . Listeriosis in pregnancy is generally defined as a clinical illness in a mother and/or child in conjunction with isolation of L. monocytogenes from the mother, neonate, fetus or placental surface (10) .
Listeriosis can frequently affect the fetus and neonate by transplacental transmission (1) . Pregnancy-related cases account for 20·7 % of listeriosis globally, with an overall case fatality of 14·9 %, including neonatal deaths (4) . Mothers with listeriosis may be asymptomatic or have influenza-like symptoms such as fever, malaise, myalgia or headache (1,7,11) . Listeriosis can result in more serious consequences in the fetus or neonate, including miscarriage, prematurity, central nervous system infections, septicaemia and death (1,4,12) due to immune system insufficiency.
The incidence of listeriosis during pregnancy has decreased recently in a number of countries, including France, Belgium and the USA (12)(13)(14) , possibly due to regulatory and industry efforts, or increased awareness following active prevention campaigns targeting pregnant women (12,13) . However, it may be under-reported due to variable or asymptomatic maternal presentation or under-recognition in spontaneous miscarriages or stillbirths. It is thus still important to minimise the risk of development of listeriosis during pregnancy due to the severity of the implications for fetal and neonatal health.
Recommendations for reducing nutritional exposure to high-risk or contaminated foods are the best ways to reduce the risk of listeriosis. The objective of the present systematic review was therefore to assess the association between foods and food preparation practices and listeriosis during pregnancy to guide the refinement of population-specific recommendations.

Eligibility criteria
We included case-control and cohort studies. Case reports, case series and reviews were excluded. Eligible studies included pregnant women or their newborn offspring with known listeriosis status. Exclusion criteria were reports on men or non-pregnant women and animal studies. Only articles published in English were included. The protocol was registered in the international prospective register of systematic reviews PROSPERO (CRD 42017056134). No ethics or institutional review board approval was required for this work.
The aetiological exposure was defined as nutritional exposure during pregnancy consistent with international population recommendations (such as Food Standards Australia New Zealand or the Centers for Disease Control and Prevention) for minimising listeriosis, including delicatessen-style meat, cold meats, dairy products made from unpasteurised milk, soft, semi-soft or surface-ripened cheeses, ready-to-eat foods, and appropriate food preparation and storage techniques (8,9) .
The primary outcome was a diagnosis of listeriosis during pregnancy. Secondary outcomes were maternal (pregnancy complications including miscarriage, preterm delivery, maternal sepsis, chorioamnionitis or hospitalisation), fetal (stillbirth) or neonatal (death, preterm birth, admission to hospital or infection) outcomes.  Table S1 and was modified for the other databases using relevant subject headings. Additional eligible articles were identified by hand-searching the reference lists of all included articles, or by contacting clinical experts. Where abstracts were identified, these were included if a corresponding full-text article could be identified.

Study selection
One independent reviewer (L. J. M.) who was not blinded to the names of investigators or sources of publication, identified and selected the articles that met the inclusion criteria. A second independent reviewer (Y. V.) performed article selection on a subset (10 %) of articles. Disagreements between these two authors were discussed and resolved by consensus or arbitration.

Data extraction
Relevant data from included studies were extracted independently by one reviewer (L. J. M.), with a second reviewer (M. B. K.) independently checking data extraction for all of the eligible studies. Disagreements between these two authors were discussed and resolved by consensus or arbitration. The data extracted included information on author(s), year of publication, study design, study location, participant characteristics, nutritional exposures (retrospective or prospective) and outcomes.

Assessment of risk of bias
All included studies were evaluated by one independent reviewer (L. J. M.), with a second reviewer (M. B. K.) independently evaluating all of the eligible studies, with neither of the two authors blinded to the names of investigators or sources of publication. The quality of the included studies was assessed using criteria based on the Newcastle-Ottawa Scale for non-randomised studies, with a maximum score of 9 (15) . These criteria assess the selection of case and control groups, comparability of case and control groups, and the quality of outcome measurement. Studies were classified as good quality if they scored 3 or 4 in the selection domain, 1 or 2 in the comparability domain, and 2 or 3 in the outcome/exposure domain; fair quality if they scored 2 in the selection domain, 1 or 2 in the comparability domain, and 2 or 3 in the outcome/exposure domain; and poor quality if they scored 0 or 1 in the selection domain, or 0 in the comparability domain, or 0 or 1 in the outcome/exposure domain.

Data synthesis
Data were presented as subgroups of (1) studies where data were available for pregnant women and neonates as a separate group and (2) studies where data were not available for pregnant women and infants as a separate group but instead were presented as combined data. Due to clinical heterogeneity relating to differences in nutritional exposures, it was not possible to perform a meta-analysis, and thus results are described narratively. It was not possible to assess publication bias through funnel plots as no statistical data synthesis could be performed.

Risk of bias of included studies
The quality assessment of the included studies is presented in Table 1 and Supplementary Table S4. Overall, four studies were classified as good quality (21,22,25,27) , five studies were classified as fair quality (6,19,20,24,26) and two studies were classified as poor quality (10,23) . The case definition was adequate and was independently validated for all studies. The representativeness of the cases was adequate with the exception of one study for which it was not stated if the cases comprised all eligible cases in a certain catchment area or sample of these cases (23) , and three studies for which the surveillance population was representative of the broader population but it was not stated how the cases were selected for the case-control study from this population (6,19,24) . The selection of the controls was adequate for all studies, with the exception of three studies where hospital controls were used (20,23,26) . None of the studies stated that the controls had no history of listeriosis. For the comparability of the cases and controls on the basis of design or  Continued analysis, four studies were comparable on one factor (19,(24)(25)(26) , and six studies (6,10,(20)(21)(22)27) were comparable on two or more factors. The cases and controls were matched for characteristics including time of birth (24) , birth weight (24) , gestational age (19,21,22,26,27) , maternal age (6,10,(20)(21)(22)25,27) , geographical area (6,20) , hospital (19,21) , ethnicity (21,22) and health-care provider (27) . Pourkaveh et al. (23) did not perform matching for any study characteristics and noted that cases were younger and were less likely to have a tertiary qualification or Internet access. The ascertainment of exposure was performed by a structured interview, where the outcome assessor was blinded to case or control status for three studies (24,25,27) , an interviewer where the outcome assessor was not blinded to case or control status for five studies (10,19,20,22,23,26) , and no description for two studies (6,21) . The exposure was ascertained in the same way for cases and controls for all studies. The nonresponse rate was the same for cases and controls, with the exception of one study for which n 5 were missing from nonpregnancy cases (10) , and one study where non-respondents were described (23) .

Discussion
We report here a systematic review assessing the contribution of specific foods to listeriosis in pregnancy. Both outbreak and non-outbreak studies included in the present systematic review report that mothers with listeriosis during pregnancy were more likely to have consumed either pasteurised or unpasteurised dairy products, cooked, semi-cooked, smoked or processed meat products, or ready-to-eat foods including fruits and vegetables, although we note the OR for an increased consumption of these foods ranged from 1·6 to 20·5. With regards to secondary maternal, fetal and neonatal outcomes, these ranged from acute febrile illness and gastrointestinal discomfort to premature birth, meningitis, septicaemia and neonatal mortality. Only one study reported secondary maternal, fetal and neonatal outcomes separately for both cases and controls, and reported a lower gestational age of the current and first spontaneous abortion for cases compared with controls (23) . These findings support guidelines for preventing listeriosis during pregnancy. The results showed an increased risk for listeriosis during pregnancy due to the consumption of a number of dairy food products, including unpasteurised products such as raw milk, soft cheeses or Hispanic-style cheeses. These are high-risk food products for pregnant women as they either do not undergo a treatment process to remove L. monocytogenes, or L. monocytogenes can survive the production process (28) . Furthermore, two studies comprised outbreaks associated with the consumption of Mexican-style cheese. These styles of cheeses mainly infect pregnant women with a Hispanic ethnicity in US studies (21,22) and are high-risk foods for listeriosis due to the fact that they are manufactured from raw milk, undergoing processing by thermoplastification or ripening which is insufficient to kill foodborne pathogens or contamination at the post-processing stage (29) . Population recommendations should include culturally relevant information on specific higher-risk foods to avoid and safer food alternatives, in addition to safe food preparation methods (30) . This is particularly important as there may be a lower uptake or delivery of food safety messages to ethnic minorities (31) and higher pregnancy-associated listeriosis in ethnic minorities (32,33) .
Outbreaks of listeriosis related to the consumption of pasteurised products can occur due to contamination after the heat-treatment process (34) . However, these result in fewer hospitalisations than outbreaks caused by unpasteurised products (34) . An association between listeriosis and pasteurised products was observed here in two studies (6,20) , although we note that these were comprised predominantly of outbreak cases and that the association reported by Jensen et al. (20) was of borderline statistical significance. This suggests that guidelines for reducing listeriosis risk need not be modified to incorporate pasteurised products. Our results also showed an association between consumption of ready-to-eat products or products prepared in commercial establishments and a range of meat products and a higher risk of listeriosis in pregnancy including uncooked, undercooked, processed or smoked meats or pâté or smoked fish (19,20,(22)(23)(24)(25)(26)(27) . These products are high risk as L. monocytogenes can grow and multiply during cold storage (35) . As these foods are often consumed without further heating, this will increase the chance of a consumer being affected by a contaminated product. This is consistent with ready-to-eat products such as delicatessen meats being classified in the very high-risk category for potential risk of listeriosis (36) . This highlights the importance of physical facility, equipment design and cross-contamination controls to avoid contamination of products, and of outbreak investigations in identifying suspected food vehicles for foodborne illness.
While dietary guidelines for preventing listeriosis differ from country to country, in general they consistently cover the themes of specific high-risk foods and food handling practices to avoid (8,9) . The dietary exposures associated with listeriosis in the present systematic review were in keeping with these dietary guidelines in that unpasteurised dairy products, soft cheeses, undercooked, uncooked, smoked or processed meat, and ready-to-eat foods were associated with a higher risk of listeriosis. This highlights the utility of the current guidelines. However, there is some concern that these guidelines may overly focus on exclusion of certain foods and handling practices. It is possible that the lack of a focus on the inclusion of safer alternatives and practices may overall lead to poorer diet quality (8,9) . The uptake of the current recommendations in different populations is also unknown given variations in health literacy with demographic characteristics such as age and education (37,38) . This is consistent both with our finding of younger age and lower education and Internet access for cases compared with controls (23) . It is also consistent with previous reports of an association of listeriosis with lower age (39) or lower socio-economic status (40) , which is also of concern given potential delays in appropriate specimen collection to confirm illness for individuals living in more deprived areas (41) .
While we did not include literature reporting on listeriosis outbreaks more broadly, these were beyond the scope of the present review inclusion criteria. These focused solely on case-control and cohort studies to allow a comparison of nutritional exposures in cases compared with controls. We note important findings from the broader listeriosis outbreak literature including confirmation of high-risk foods reported here such as soft cheeses (42) , smoked fish (43) , processed meats (44) , ready-to-eat vegetables (45,46) and pâté (47) . We also highlight that this broader outbreak literature aids in identification of high-risk foods that were not significantly associated with listeriosis in the limited identified literature on case-control and cohort studies in pregnancy, such as rockmelon or cantaloupe (48) . These foods are well documented as being associated with outbreaks in the past which have included pregnant women (48) . Where foods identified as high risk from population outbreaks such as vegetables (45,46) were not significantly associated with listeriosis here (10) , the authors also comment on potential limited power to detect significant associations. This supports the inclusion of these foods in current population recommendations for minimising listeriosis (8) .
We identified a relatively limited literature base of casecontrol studies examining the association between specific foods and listeriosis, and an even smaller number of studies that reported data for the cases during pregnancy separately (10,(21)(22)(23) . Furthermore, the majority of studies reported on maternal, fetal and neonatal complications associated with listeriosis just for the cases, with only one study reporting on differences in perinatal complications between cases and controls (23) . This made it impossible for us to quantify the effects of consumption of various foods in pregnancy on adverse pregnancy outcomes mediated by listeriosis. We note that many exposures are required to cause one case of pregnancy-associated listeriosis (estimated at about 1 per 10 000 exposed pregnant women) (49) , which may limit the utility of case-control studies in assessing the relationship between outcome and exposures. Additional limitations relating to the individual studies include the possibility of recall bias due to the studies being case-control and also as dietary information was retrospectively collected 1-3 months prior to disease confirmation in all studies. The shorter time-frame may also have contributed to bias as the incubation period in pregnancy-associated listeriosis cases ranges from 17 to 67 d (50) . Dietary information was collected by a blinded investigator in only a limited number of studies (24,25,27) , which increases the possibility of detection bias. We also note variability in the source and matching of controls, and heterogeneity in the studies in that they comprised both outbreak and nonoutbreak cases of listeriosis, which makes it difficult to interpret data more broadly for general background risk of listeriosis. The bulk of the literature (seven out of eleven studies) also focused on studies from the USA; thus research from a broader range of countries is warranted to maximise generalisability of these results. The studies also generally examined only specific food consumption, with a more limited number examining different means of food preparation (27) , and no studies examining food hygiene practices. As all of these aspects are highlighted in recommendations for reducing listeriosis (8,9) , they also warrant assessment in future studies. Additional biases may also have occurred through identifying articles only in English.
Listeriosis is associated with severe maternal, fetal and neonatal complications and warrants consumer and health professional guidelines to minimise risk. In the present systematic review, cases of listeriosis during pregnancy were more likely to have consumed certain high-risk dairy and meat products and ready-to-eat foods in comparison with controls, with avoidance of these specific high-risk foods generally being recommended in population guidelines. Further research is warranted assessing the reach and uptake of population guidelines for preventing listeriosis in pregnant women from a range of countries, reproductive life stages and demographic backgrounds. This will aid in increasing the uptake of guidelines to reduce cases of listeriosis during pregnancy and maternofetal morbidity and mortality.

Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/jns.2018. 16