Introduction
Haematological cancers originate in blood forming tissue, such as bone marrow or immune cells, and utilise chemotherapy as first line treatment followed by haematopoietic stem cell transplant (HSCT). HSCTs involve infusing stem cells from the bone marrow of the individual, taken prior to treatment (autologous) or from a matched donor (allogeneic). HSCT also involves a high dose of chemotherapy to destroy cancerous blood cells and suppress the immune system to allow the body to accept the stem cell transplant.(Reference Galgano, Hutt, Kenyon and Babic1,Reference Sahdev and Lanzkowsky2) This treatment may cause other healthy cells to be destroyed including neutrophils, which are integral in preventing infections.(Reference Miller and Campbell3) Due to the high dose of chemotherapy received, these individuals generally become neutropenic characterised by neutrophil counts below 1.5 × 109 neutrophils per litre of blood, and reduction below 0.5 × 109 neutrophils/l classified as severe neutropenia.(Reference Donadieu, Beaupain and Fenneteau4) The decline in neutrophil cell counts leads to an increased risk of infection, prolonged bleeding time due to low platelet count, increased pain, and reduced nutritional intake due to mucositis and tiredness.(Reference Galgano, Hutt, Kenyon and Babic1)
Due to the increased infection risk, individuals receiving cancer treatment in hospital in the 1960s would often be placed in sterile environments including laminar flow rooms and receive gut decontamination with antibiotics in addition to a ‘sterile’ diet.(Reference Bodey, Hart and Freireich5) However, an early literature review in 1984 determined that protective environments did not reduce infection rates, but had many negative effects including poor psychological impact, increased costs, and increased staff labour associated with their use.(Reference Armstrong6) Although complete protective environments are no longer employed, many healthcare institutions continue to provide patients with a neutropenic diet (ND),(Reference Wolfe, Sadeghi and Agrawal7) which have changed from very restrictive ‘sterile’ diets to one that limits ‘high-risk’ foods: raw fruits and vegetables, raw/undercooked meat, fish and eggs and unpasteurised dairy. Restriction of these foods is widely considered the basis of the ND; however, no set guidelines exist.(Reference Wolfe, Sadeghi and Agrawal7)
In recent years the validity of the ND has been questioned. Numerous systematic reviews and meta-analyses(Reference Sonbol, Jain and Firwana8–Reference Matteucci, De Pasquale and Pastore14) have shown that the ND does not significantly reduce the risk of infection or mortality within this population. Additionally, The European Society for Clinical Nutrition and Metabolism (ESPEN), a major society in nutrition, do not recommend an ND.(Reference Arends, Bachmann and Baracos15,Reference Muscaritoli, Arends and Bachmann16) Despite this, clinicians(Reference Carr and Halliday17,Reference Braun, Chen and Frangoul18) and hospitals that perform HSCTs(Reference Baumgartner, Bargetzi and Bargetzi19–Reference Toenges, Greinix and Lawitschka22) continue to provide patients with NDs in Switzerland,(Reference Baumgartner, Bargetzi and Bargetzi19) China,(Reference Fang, Liu and Zhang20) Italy,(Reference Peric, Botti and Stringer21) Germany,(Reference Toenges, Greinix and Lawitschka22) Austria,(Reference Toenges, Greinix and Lawitschka22) UK,(Reference Carr and Halliday17) and the US.(Reference Braun, Chen and Frangoul18) All studies reported between 50 and 80% usage of the ND.(Reference Carr and Halliday17–Reference Toenges, Greinix and Lawitschka22) Interestingly, the studies with >80% usage are the most recent studies within this area, published between 2018 and 2021.(Reference Fang, Liu and Zhang20–Reference Toenges, Greinix and Lawitschka22)
Although there are numerous studies that have assessed the ND and its impacts on infection and mortality rates, other outcomes aside from infection rates that are of importance in relation to diet during cancer have not been considered. To our knowledge, many of these additional areas have not been systematically reviewed and need to be considered as part of the overall impact of NDs in the health and lifestyle of people with cancer. Therefore, the aim of this research is to systematically scope the current evidence-base to identify studies on the ND and any outcomes for people with cancer that relate to their medical, nutritional, social, psychological, or physical health as well as costs associated with treatment. Additionally, gaps in the evidence and opportunities for future research will be identified.
Methods
Protocol
The methods for this scoping review were prospectively designed and registered with Open Science Framework on 19 July 2022 and can be accessed at https://osf.io/gan2p.
Selection criteria
Included studies were required to be (i) in English, (ii) human studies assessing people with cancer of any type, (iii) consuming a ND as defined by the paper authors, (iv) any age group, date of study, or setting. There were minimal restrictions in the included studies in attempt to capture as many studies as possible. The eligible study types included randomised control trials (RCTs), retrospective, prospective, cohort, observational, comparative, systematic review, and meta-analyses.
Studies were excluded if (i) in a language other than English, (ii) non-human, (iii) assessed the wrong diet or (iv) outcomes were not related to the clinical health or lifestyle outcomes of the scoping review. Excluded study types included letters, conference proceedings, books, book chapters, and guidelines.
Search strategy
The following databases were searched CINAHL Complete (EBSCO Publishing, Inc), The Cochrane Library (John Wiley & Sons, Ltd), Embase (Ovid), Emcare (Ovid), MEDLINE (Ovid), Scopus (Elsevier Science Publishers), and Web of Science (Clarivate Analytics).
The following search terms with Boolean operators were used in all databases with no other filters applied: “neutropenic diet*” or “low bacteria* diet*” or “low-bacteria* diet” or “low microbial diet*” or “low-microbial diet*” or “germ free diet*” or “germ-free diet*” or “sterile diet*”. Databases were searched from inception to 19th July 2022 with additional papers added from search alerts of the above searches in all databases between 20 July 2022 and 23 August 2023.
Screening sources and data extraction
Screening was undertaken in Covidence (Veritas Health Innovation Ltd). Title and abstract screening were completed independently by two reviewers (TG + one of MH, PK, LT, EM) with any conflicts being resolved by discussion (TG + EM). Full text screening was completed independently by two authors (TG + one of MH, PK, LT, EM) with any conflicts being resolved by discussion (TG + EM) or by a third author if needed. Reasons for exclusion were given for each study at the full-text screening stage.
Data from included articles was extracted into Excel (Microsoft Corporation, Washington, US) using standardised tools formulated by one reviewer (TG) and checked by one reviewer (EM). Extraction was completed by one reviewer (TG) and checked by a second reviewer for accuracy (one of MH, PK, LT, or EM). The outcomes of interest that were extracted were infection rate, mortality rate, fever, antibiotic use, side effects (including diarrhoea, nausea, vomiting), comorbidities (including neutropenic enterocolitis, graft vs host disease, mucositis), hospitalisation, quality of life, diet acceptability, nutritional status, and costs. Data was extracted and used as it was presented in the corresponding paper, and the results are presented in a narrative summary.
Critical appraisal
Critical appraisal was completed for all included studies using the most suitable JBI checklist for study type (https://jbi.global/critical-appraisal-tools). Appraisal was completed by TG and checked by EM. JBI critical appraisal checklist for descriptive/case series was used for all retrospective studies included in the review. This checklist is no longer listed on the JBI website but is however considered a key tool for critical appraisal of descriptive studies.(Reference Ma, Wang and Yang23)
Results
The initial search in July 2022 retrieved 1037 citations and after the removal of 481 duplicates, a total of 550 articles were available for title and abstract screening. Following initial screening, 264 articles were eligible for full-text screening. Following full-text screening 19 articles were deemed eligible for inclusion. Two articles were added from email alerts from the databases for a total of 21 included articles. Figure 1 shows the PRISMA Flow Diagram.
PRISMA Flow Diagram.
Nine studies were conducted in the United States,(Reference Sonbol, Jain and Firwana8,Reference Ball, Brown and Das9,Reference Moody, Finlay and Mancuso24–Reference Taggart, Neumann and Alonso30) three in India,(Reference Ramamoorthy, Rubens and Appunni11,Reference Radhakrishnan, Lagudu and Gangopadhyay31,Reference Gupta, Gupta and Meena32) two in The Netherlands(Reference van Dalen, Mank and Leclercq10,Reference van Tiel, Harbers and Terporten33) and Iran(Reference Jalali, Hosseini and Ajami34,Reference Hosseini, Jalali and Ajami35) and one in Australia,(Reference Heng, Barbon Gauro and Yaxley36) China,(Reference Ma, Lu and Liu12) Germany,(Reference Jakob, Classen and Stecher37) The Philippines(Reference Ng and Dujua13) and Italy.(Reference Matteucci, De Pasquale and Pastore14) Four studies(Reference Moody, Finlay and Mancuso24,Reference Taggart, Neumann and Alonso30,Reference Gupta, Gupta and Meena32,Reference Hosseini, Jalali and Ajami35) included only paediatric participants and nine studies included only adult participants.(Reference Matteucci, De Pasquale and Pastore14,Reference DeMille, Deming and Lupinacci25–Reference Lassiter and Schneider28,Reference van Tiel, Harbers and Terporten33,Reference Jalali, Hosseini and Ajami34,Reference Heng, Barbon Gauro and Yaxley36,Reference Jakob, Classen and Stecher37) Eight studies,(Reference Sonbol, Jain and Firwana8–Reference Ng and Dujua13,Reference Moody, Baker and Santizo29,Reference Radhakrishnan, Lagudu and Gangopadhyay31) including the seven systematic reviews/meta-analysis, included all age groups. Eleven studies were conducted in an inpatient hospital setting, and one in an outpatient setting.(Reference DeMille, Deming and Lupinacci25) Two studies included both inpatients and outpatients(Reference Moody, Finlay and Mancuso24,Reference Moody, Baker and Santizo29) and six of seven systematic reviews/meta-analyses(Reference Sonbol, Jain and Firwana8–Reference Ng and Dujua13) included one or more of the outpatient studies whereas one(Reference Matteucci, De Pasquale and Pastore14) only included studies with an inpatient population. All studies included participants with haematological cancers and four also included participants with oncological cancers(Reference Moody, Finlay and Mancuso24,Reference DeMille, Deming and Lupinacci25,Reference Moody, Baker and Santizo29,Reference Gupta, Gupta and Meena32) however, none analysed results by cancer type. A summary of the characteristics of the included studies are shown in Table 1.
Characteristics of Included Studies
ND, neutropenic diet; CD, comparator diet; RCT, randomised control trial; CDN, comparator diet not randomly assigned, ages presented as mean ages or age range.
There are no guidelines for the use of the ND and the diet can vary between hospitals in the foods restricted. This is reflected in the differences between the NDs in the included studies. All included studies restricted most raw fruits and vegetables – some allowed fruits and vegetables with thick skin or that could be hand-peeled.(Reference Moody, Finlay and Mancuso24,Reference Lassiter and Schneider28,Reference Moody, Baker and Santizo29,Reference Hosseini, Jalali and Ajami35) Other restricted foods included raw grains,(Reference Trifilio, Helenowski and Giel27,Reference Taggart, Neumann and Alonso30,Reference Jakob, Classen and Stecher37) raw nuts/nut butters,(Reference Moody, Baker and Santizo29,Reference Jakob, Classen and Stecher37) raw seeds,(Reference Trifilio, Helenowski and Giel27) raw miso,(Reference Trifilio, Helenowski and Giel27) yoghurt,(Reference Moody, Baker and Santizo29,Reference Heng, Barbon Gauro and Yaxley36) dairy from bulk machines (i.e. soft serve ice cream, frozen yoghurt), raw honey,(Reference Taggart, Neumann and Alonso30) dried/raw herbs and spices,(Reference DeMille, Deming and Lupinacci25,Reference Trifilio, Helenowski and Giel27,Reference van Tiel, Harbers and Terporten33,Reference Heng, Barbon Gauro and Yaxley36) yeast,(Reference Trifilio, Helenowski and Giel27,Reference Taggart, Neumann and Alonso30) cold desserts(Reference Heng, Barbon Gauro and Yaxley36) tap/un-boiled water,(Reference van Tiel, Harbers and Terporten33–Reference Hosseini, Jalali and Ajami35,Reference Jakob, Classen and Stecher37) takeaway/fast food,(Reference Moody, Finlay and Mancuso24,Reference Moody, Baker and Santizo29,Reference Taggart, Neumann and Alonso30) buffet/street food,(Reference DeMille, Deming and Lupinacci25,Reference Taggart, Neumann and Alonso30) and sharing food.(Reference Taggart, Neumann and Alonso30)
The major comparator diet in the included studies was the food safety diet, used by nine studies.(Reference Moody, Finlay and Mancuso24,Reference Trifilio, Helenowski and Giel27–Reference Gupta, Gupta and Meena32,Reference Heng, Barbon Gauro and Yaxley36,Reference Jakob, Classen and Stecher37) This diet restricts unpasteurised eggs, dairy, raw/undercooked meat and fish and has regulations on food handling, washing, preparation, and storage.(38) One study termed the comparator diet ‘Modified Bone Marrow Transplant Diet’ however restrictions were similar to the food safety diet.(Reference Taggart, Neumann and Alonso30) Two studies(Reference Trifilio, Helenowski and Giel27,Reference van Tiel, Harbers and Terporten33) used a standard hospital diet which followed the food safety diet with some additional restrictions: raw tomatoes, cold smoked fish, raw miso, raw grains/seeds, and brewer’s yeast. One study did not impose any restrictions for the comparator diet,(Reference Gardner, Mattiuzzi and Faderl26) and another included participants in the comparator group if they were non-compliant with the ND.(Reference DeMille, Deming and Lupinacci25)
The incidence of infections was a major outcome in all but two of the 21 included studies. However, the type of infections, grouping of infection types, and the method of reporting varied amongst studies. As shown in Table 2, we report these results as they have been reported in the included studies.
Effect of the Neutropenic Diet on Rates of Infection
NA, Not Assessed.
a During neutropenia.
b Overall and when neutropenia had resolved.
c Significant differences not reported however of the six studies 4/6 found no significant difference 1/6 did not include a comparator and 1/6 ND increased infection risk for HSCT transplant recipients only.
d HSCT recipients and overall.
e Non-HSCT recipients.
f Cases of febrile neutropenia.
g Number of febrile days.
h HSCT recipients.
i Non-HSCT recipients and overall.
j At baseline.
k Day 15 of study.
− No significant difference between Neutropenic Diet and comparator, ↑Neutropenic Diet increases risk, ↓Neutropenic Diet decreases risk.
Total infection, reported in 13 studies(Reference Ball, Brown and Das9,Reference Ramamoorthy, Rubens and Appunni11,Reference Ma, Lu and Liu12,Reference Matteucci, De Pasquale and Pastore14,Reference Moody, Finlay and Mancuso24–Reference Trifilio, Helenowski and Giel27,Reference Moody, Baker and Santizo29,Reference Radhakrishnan, Lagudu and Gangopadhyay31,Reference Gupta, Gupta and Meena32,Reference Heng, Barbon Gauro and Yaxley36,Reference Jakob, Classen and Stecher37) included infection of any body-site with any bacteria, fungi, or virus. One RCT(Reference Gupta, Gupta and Meena32) reported a significantly higher rate of infection in the ND group with 12 infections (n = 21) compared to nine in the comparator group (n = 21) (P = 0.049).(Reference Gupta, Gupta and Meena32) A retrospective study(Reference Trifilio, Helenowski and Giel27) found a significant difference between the diets for total diagnostically confirmed infections – 135 infections in the ND group compared to 106 in the comparator group (P = 0.03). Infections present when neutropenia had resolved was also significantly higher in the ND group compared to the comparator group (P = 0.01). However, infection during neutropenia was not significantly different (P = 0.22).(Reference Trifilio, Helenowski and Giel27) Ten of the 13 studies which reported on total infection did not find any significant difference between diet groups. Overall, from the included studies the ND either results in no significant difference in infection rates compared to the comparator diet (n = 12 studies) or increases the infection rate (n = 2 studies).
A further three studies reported combined rates of total infection and/or fever.(Reference Sonbol, Jain and Firwana8,Reference van Dalen, Mank and Leclercq10,Reference Gardner, Mattiuzzi and Faderl26) A systematic review/meta-analysis(Reference Sonbol, Jain and Firwana8) found higher rates of infection in the ND group for the total population (RR 1.17, 95% CI (1.04–1.32)). When haematopoietic stem cell transplant (HSCT) recipients following the ND were compared to those following the comparator diet, significantly higher rates of infection in the ND group were demonstrated (RR 1.25, 95% CI (1.02–1.54)). However, no significant difference was seen between diet groups for participants who were not recipients of HSCT.(Reference Sonbol, Jain and Firwana8)
Major infection was assessed in three studies;(Reference Sonbol, Jain and Firwana8,Reference Gardner, Mattiuzzi and Faderl26,Reference Radhakrishnan, Lagudu and Gangopadhyay31) however, as there is no clinical definition of what constitutes a major infection, there were differences in the definition between studies. All three studies included pneumonia, bacteraemia, or fungaemia as major infections,(Reference Sonbol, Jain and Firwana8,Reference Gardner, Mattiuzzi and Faderl26,Reference Radhakrishnan, Lagudu and Gangopadhyay31) whilst one study also included urinary tract infection, meningitis, cellulitis, or diarrhoea.(Reference Radhakrishnan, Lagudu and Gangopadhyay31) A systematic review/meta-analysis reported a significantly higher rate of major infection in the ND group when only assessing participants who were HSCT recipients (RR 1.25, 95% CI (1.02–1.54)). No significant difference was seen in participants who were not transplant recipients or when all participants were assessed together.(Reference Sonbol, Jain and Firwana8)
Minor infection was assessed by two studies;(Reference Gardner, Mattiuzzi and Faderl26,Reference Radhakrishnan, Lagudu and Gangopadhyay31) however, as there is no clinical definition of what constitutes a minor infection, both studies included all other infections not defined as a major infection. Both studies found no significant difference between diet groups.(Reference Gardner, Mattiuzzi and Faderl26,Reference Radhakrishnan, Lagudu and Gangopadhyay31)
Gut colonisation by pathogenic bacteria or yeasts was assessed in two RCTs by faecal analysis.(Reference Radhakrishnan, Lagudu and Gangopadhyay31,Reference van Tiel, Harbers and Terporten33) Radhakrishnan, Lagudu(Reference Radhakrishnan, Lagudu and Gangopadhyay31) reported significantly higher rates of colonisation in the comparator group (n = 40/96) compared to the ND group (n = 29/102) at baseline (P = 0.05). Bacteria isolated from stool included multidrug resistant (MDR) Escherichia coli, MDR E. faecalis, MDR Klebsiella pneumoniae, MDR E. faecium, Vancomycin resistant E. faecium. A positive stool culture at baseline is considered more likely to be reflective of the diet consumed while participants are in the community. At the end of the study (day 15), there were no differences, indicating that the diet did not significantly affect gut colonisation.(Reference Radhakrishnan, Lagudu and Gangopadhyay31) No significant differences were reported in the study by van Tiel, Harbers.(Reference van Tiel, Harbers and Terporten33)
Pneumonia is the inflammation of the lungs caused by bacteria, fungi, or viruses(Reference Moeinafshar, Rezaei and Rodriguez-Morales39) and while there is no evidence linking it to diet, it was assessed in three studies.(Reference Gardner, Mattiuzzi and Faderl26,Reference Radhakrishnan, Lagudu and Gangopadhyay31,Reference Gupta, Gupta and Meena32) Two studies found no significant difference between diet groups(Reference Gardner, Mattiuzzi and Faderl26,Reference Radhakrishnan, Lagudu and Gangopadhyay31) and the third did not analyse statistical significance, however they reported two pneumonia cases in the ND group (n = 21) and one in the comparator group (n = 21).(Reference Gupta, Gupta and Meena32)
Bacteraemia, presence of bacteria in the bloodstream was assessed in seven studies, none of which found a significant difference between diet groups.(Reference Ng and Dujua13,Reference Lassiter and Schneider28,Reference Taggart, Neumann and Alonso30,Reference Radhakrishnan, Lagudu and Gangopadhyay31,Reference van Tiel, Harbers and Terporten33,Reference Heng, Barbon Gauro and Yaxley36,Reference Jakob, Classen and Stecher37) Fungaemia, the presence of fungi in the bloodstream was assessed by one RCT which found no significant difference between diet groups.(Reference van Tiel, Harbers and Terporten33) Combined bacteraemia or fungaemia was assessed in two studies,(Reference Sonbol, Jain and Firwana8,Reference Gardner, Mattiuzzi and Faderl26) a quasi-experimental study found significantly higher rates in the comparator group (n = 17) than the ND (n = 7) (P = 0.03);(Reference Gardner, Mattiuzzi and Faderl26) however, a systematic review/meta-analysis found no significant differences between diet groups.(Reference Sonbol, Jain and Firwana8)
Norovirus, which causes gastrointestinal infection(40) was assessed in an observational study which reported no significant differences between diet groups.(Reference Taggart, Neumann and Alonso30)
Neutropenic fever or febrile neutropenia is characterised as a high temperature (>38.3°C or >38°C on two occasions).(Reference Lingaratnam, Slavin and Koczwara41) Eight studies assessed fever with five reporting on neutropenic fever,(Reference Ng and Dujua13,Reference Moody, Finlay and Mancuso24,Reference Radhakrishnan, Lagudu and Gangopadhyay31,Reference Gupta, Gupta and Meena32,Reference Heng, Barbon Gauro and Yaxley36) four reported on fever of unknown origin (>38.3°C on multiple occasions for three weeks, with diagnosis unclear after one week(Reference Brown and Finnigan42)),(Reference Gardner, Mattiuzzi and Faderl26,Reference Radhakrishnan, Lagudu and Gangopadhyay31,Reference Heng, Barbon Gauro and Yaxley36,Reference Jakob, Classen and Stecher37) one assessed persistent fever(Reference Jakob, Classen and Stecher37) and one assessed high or low temperature,(Reference van Tiel, Harbers and Terporten33) with none of the eight studies reporting significant differences between diet groups.
Antibiotic use was assessed in two studies with different metrics; length of antibiotic use (the consecutive number of days which an antibiotic is used)(Reference Gupta, Gupta and Meena32,Reference van Tiel, Harbers and Terporten33) and antibiotic duration (the total amount of time in which antibiotics are used).(Reference Gupta, Gupta and Meena32) Neither study found a significant difference between diet groups for use of antibiotics (see Table 3).
Effect of the Neutropenic Diet on Clinical Factors
NA, not assessed; NE, neutropenic enterocolitis.
a Significant difference not reported 3/3 studies included had no significant differences between groups.
b Only included results from Jakob, Classen.(Reference Jakob, Classen and Stecher37)
− No significant difference between Neutropenic Diet and comparator, ↑Neutropenic Diet increases risk.
Gastrointestinal side effects such as diarrhoea, nausea, and vomiting during cancer treatment may be caused by chemotherapy,(Reference Krishnamurthi and Macaron43) use of antibiotics which causes healthy bacteria of the gut to be destroyed and/or, use of other medications such as opioids.(Reference Krishnamurthi and Macaron43) These symptoms are also commonly attributed to Clostridium difficile infection.(Reference Mullish and Williams44) Diarrhoea was assessed in five studies(Reference Matteucci, De Pasquale and Pastore14,Reference Moody, Finlay and Mancuso24,Reference Trifilio, Helenowski and Giel27,Reference Heng, Barbon Gauro and Yaxley36,Reference Jakob, Classen and Stecher37) and only one study found a significantly higher rate in the ND group (P < 0.001). The remaining studies saw no significant differences. Two studies reported on Clostridium difficile – with both reporting no significant difference between groups – however, no link was made between the incidence of diarrhoea and Clostridium difficile infection.(Reference Trifilio, Helenowski and Giel27,Reference Heng, Barbon Gauro and Yaxley36) Nausea was assessed in two studies, a retrospective study reported a significantly higher rate of nausea in the ND group(Reference Jakob, Classen and Stecher37) and a systematic review included only findings from the retrospective study.(Reference Matteucci, De Pasquale and Pastore14) Vomiting was assessed in one study which reported two of nine participants in the ND group and two of ten in the comparator diet group had instances of vomiting, however, significance was not reported(Reference Moody, Finlay and Mancuso24) (see Table 3).
Neutropenic enterocolitis is the inflammation of the gastrointestinal tract occurring in a neutropenic individual. Two studies(Reference Gupta, Gupta and Meena32,Reference Heng, Barbon Gauro and Yaxley36) assessed neutropenic enterocolitis, one study reported higher rates in the ND group (P = 0.044) and the other reported no significant difference between groups.(Reference Heng, Barbon Gauro and Yaxley36)
Graft vs Host Disease (GvHD) can occur post-stem cell transplant when donor T-cells attack healthy cells of the recipient. Two observational studies assessed GvHD;(Reference Trifilio, Helenowski and Giel27,Reference Taggart, Neumann and Alonso30) however, neither saw a significant difference between diet groups.
Mucositis, defined as, inflammation of the mouth and/or gut, was assessed by one RCT(Reference Moody, Baker and Santizo29) which reported four cases in the ND group and two in the comparator group, however this was not significant.
Mortality was assessed in ten studies;(Reference Sonbol, Jain and Firwana8,Reference Ramamoorthy, Rubens and Appunni11,Reference Ma, Lu and Liu12,Reference Matteucci, De Pasquale and Pastore14,Reference Gardner, Mattiuzzi and Faderl26,Reference Trifilio, Helenowski and Giel27,Reference Radhakrishnan, Lagudu and Gangopadhyay31,Reference Gupta, Gupta and Meena32,Reference Heng, Barbon Gauro and Yaxley36,Reference Jakob, Classen and Stecher37) however, no studies reported any significant differences between groups (see Table 3).
Serum albumin, whilst no longer used as a sole indicator of nutritional status, recent literature suggests decreased levels may be associated with gut dysbiosis.(Reference Maslennikov, Ivashkin and Efremova45) Two RCTs(Reference Jalali, Hosseini and Ajami34,Reference Hosseini, Jalali and Ajami35) assessed serum albumin levels and both found significantly lower levels in ND groups compared to the comparator diets post-intervention (see Table 4). This may indicate a decreased dietary intake or gut dysbiosis. Neither study reported on fluid status, inflammation, microbiome, or other GI related side effects associated with decreased serum albumin levels.
Effect of the Neutropenic Diet on Biochemical Factors, Hospitalisation, Nutritional Status, and Well-being
NA, Not Assessed; SNR, Significance Not Reported.
a Neutropenic Diet significantly decreases serum albumin when compared to diet with additional Vitamin C or olive oil and when compared pre and post intervention.
b Neutropenic Diet is not significantly different from diet with additional Vitamin C or olive oil pre intervention and diet with additional Vitamin C olive oil is not significantly different pre and post intervention.
c Neutropenic diet significantly increases risk of weight loss of between 1 and 3 kg.
d No significant difference between diets for weight loss of greater than 3 kg.
e Neutropenic diet significantly lower QoL score for PEDS core module.
f No significant difference between diet groups for PEDS cancer module.
g only included results from Jakob, Classen.(Reference Jakob, Classen and Stecher37)
↑Neutropenic Diet increases risk, − No significant difference between Neutropenic Diet and comparator.
Four studies assessed hospitalisation by length of stay(Reference Trifilio, Helenowski and Giel27,Reference Gupta, Gupta and Meena32,Reference Heng, Barbon Gauro and Yaxley36) or admissions.(Reference DeMille, Deming and Lupinacci25,Reference Gupta, Gupta and Meena32,Reference Heng, Barbon Gauro and Yaxley36) A retrospective study(Reference Trifilio, Helenowski and Giel27) reported that those in the ND group (n = 363) who underwent HSCT spent, on average, one day longer in hospital than those following the comparator diet (n = 363), but level of significance was not reported.(Reference Trifilio, Helenowski and Giel27) No significant difference was seen in the remaining two studies.(Reference Gupta, Gupta and Meena32,Reference Heng, Barbon Gauro and Yaxley36) None of the three studies which assessed hospital admissions found a significant difference between diet groups(Reference DeMille, Deming and Lupinacci25,Reference Gupta, Gupta and Meena32,Reference Heng, Barbon Gauro and Yaxley36) (see Table 4).
Patient Generated Subjective Global Assessment (PG-SGA) is a common, validated tool used in cancer patients to evaluate nutritional status.(Reference Ottery46) The PG-SGA categorises individuals into three groups: ‘appropriate nutrition’, ‘prone to malnutrition’ or ‘severe malnutrition’. Two RCTs(Reference Jalali, Hosseini and Ajami34,Reference Hosseini, Jalali and Ajami35) reported a significantly higher proportion of the ND group being placed into the ‘prone to malnutrition’ or ‘severe malnutrition’ categories.
Weight change was assessed in four studies.(Reference Matteucci, De Pasquale and Pastore14,Reference Jalali, Hosseini and Ajami34,Reference Hosseini, Jalali and Ajami35,Reference Jakob, Classen and Stecher37) A retrospective study reported a significantly greater proportion of those in the ND group losing between 1 and 3 kg (P = 0.05), however further sub-analysis showed no significance for weight loss greater than 3 kg.(Reference Jakob, Classen and Stecher37) This study only reported absolute weight loss and did not report percentage weight loss, hence making it difficult to determine if this weight loss is clinically significant. Weight loss was an outcome in the systematic review by Matteucci, De Pasquale(Reference Matteucci, De Pasquale and Pastore14) and the only included study for this outcome was Jakob, Classen.(Reference Jakob, Classen and Stecher37) As no further analysis was conducted in the systematic review, results were the same.(Reference Matteucci, De Pasquale and Pastore14)
One study published in 2007(Reference van Tiel, Harbers and Terporten33) assessed the total financial costs as a secondary outcome of the ND contrasted with the comparator diet at different stages of care – including hospital costs, other healthcare costs, and inability to work. The areas in which costs are associated were identified, however the reason for cost differences were not determined. Whilst they reported higher costs for the comparator diet during hospitalisation, in contrast during follow-up and in total, the ND had higher costs, however significance of these results was not reported.(Reference van Tiel, Harbers and Terporten33)
Quality of Life (QoL) is defined by the World Health Organisation as ‘an individual’s perception of their position in life in the context of the culture and value systems in which they live and in relation to their goals, expectations, standards, and concerns’.(47) Two RCTs from the same research group(Reference Moody, Finlay and Mancuso24,Reference Moody, Baker and Santizo29) assessed QoL in paediatric cancer patients (Table 4) using the Paediatric Quality of Life Inventory. These two studies were similar in design and in the outcomes assessed, however the 2017 study(Reference Moody, Baker and Santizo29) had more participants (ND = 77, CD = 73) compared to the 2006 study (24) (ND = 9, CD = 10). In the 2006 study, they found the Core QoL of the ND group was significantly lower (indicating lower overall QoL) than the comparator group (P < 0.05), however, this was not significant for cancer specific QoL.(Reference Moody, Finlay and Mancuso24) In the 2017 study, they did not find a significant difference between diet groups.(Reference Moody, Baker and Santizo29)
Diet acceptability was assessed by three different metrics across three studies(Reference van Dalen, Mank and Leclercq10,Reference Moody, Finlay and Mancuso24,Reference Taggart, Neumann and Alonso30) ; ease of following assigned diet, food not tasting how participants remembered, and inability to consume desired foods. All three RCTs found no significant difference between ND and comparator diet groups for diet acceptability.(Reference van Dalen, Mank and Leclercq10,Reference Moody, Finlay and Mancuso24,Reference Taggart, Neumann and Alonso30)
Diet adherence was measured by two different metrics in two RCTs; number of meals for which participants (n = 19) were following their assigned diet(Reference Moody, Finlay and Mancuso24) and number of weeks participants (n = 42) were following the assigned diet.(Reference Gupta, Gupta and Meena32) Both studies found significantly better adherence in the comparator groups. Participants were reported to have adhered to the ND (n = 9) 94.10% of the time and to the comparator diet (n = 10) 99.99% of the time.(Reference Moody, Finlay and Mancuso24) Participants followed the ND for 93 of the 98 weeks (n = 21) and the comparator 94 of the 98 weeks (n = 21), however, this was not significantly different.(Reference Gupta, Gupta and Meena32)
Discussion
The aim of the present scoping review was to systematically search the literature and identify articles which assessed the use of the ND for those undergoing cancer treatment on any outcomes relating to medical, nutritional, social, psychological, physical health, and all associated costs. We identified 21 relevant articles which covered outcomes including participants’ clinical health with most assessing infection and mortality rates. Ten of sixteen outcomes included in the present study have not previously been included in systematic reviews.
Clinical outcomes identified in this scoping review were related to risk of infection, mortality, and fever in 19 of the 21 studies. The majority of studies observed no significant difference between groups, which aligns with the rationale for recommendations of the clinical guidelines for nutrition in cancer patients from The European Society for Clinical Nutrition and Metabolism (ESPEN) published in 2016(Reference Arends, Bachmann and Baracos15) and updated in 2021.(Reference Muscaritoli, Arends and Bachmann16) Based on the evidence base at the time, ESPEN recommended that “There are insufficient consistent clinical data to recommend a low bacterial diet for patients more than 30 days after allogeneic transplant”.(Reference Arends, Bachmann and Baracos15,Reference Muscaritoli, Arends and Bachmann16) As such the ESPEN guideline for hospital nutrition.(Reference Thibault, Abbasoglu and Ioannou48) was “Neutropenic diets (also called ‘germ-free’, ‘no microbial’ or ‘sterilised’ diets) shall not be used (e.g. in neutropenic patients with cancer including haematopoietic stem cell transplant patients)”,(Reference Thibault, Abbasoglu and Ioannou48) it received an A grade recommendation as it was supported by a strong evidence base, including a meta-analysis(Reference Sonbol, Jain and Firwana8) and a Cochrane review.(Reference van Dalen, Mank and Leclercq10) The recommendation received strong consensus from ESPEN members, and it was recommended that this population follow food safety guidelines.(Reference Thibault, Abbasoglu and Ioannou48)
The ND generally includes well-cooked meat and excludes raw fruit and juices, however, within a population of Haematopoietic Stem Cell Transplant (HSCT) recipients the main food aversions were shown to be meat, specifically beef and chicken, due to the association with dysphagia.(Reference Hinkelmann, Possa and de Oliveira49) Preferred foods included fruit, fruit juices, and soup due to the association with improved gastrointestinal symptoms i.e. nausea.(Reference Hinkelmann, Possa and de Oliveira49) The ND has been shown to have reduced Vitamin C(Reference Maia, da Cruz and Gregianin50,Reference Galati, Lataro and Souza51) and fibre(Reference Maia, da Cruz and Gregianin50) due to restrictions on raw fruits and vegetables, which increases the risk of nutrient deficiencies in this population.(Reference Maia, da Cruz and Gregianin50,Reference Galati, Lataro and Souza51) These nutrient deficiencies coupled with a decreased overall energy intake could lead to malnutrition. Malnutrition associated with cancer is common, with rates of 30–40% in Australia and was an outcome in two included studies.(Reference Jalali, Hosseini and Ajami34,Reference Hosseini, Jalali and Ajami35) The combination of loss of taste and gastrointestinal symptoms, such as nausea could lead to decreased intake of food which may result in malnutrition.(Reference Hinkelmann, Possa and de Oliveira49)
Weight loss during cancer treatment may be caused by a decreased intake of food due to loss of taste and other common side effects from treatment such as nausea and vomiting;(Reference Hinkelmann, Possa and de Oliveira49) however, this scoping review determined that weight loss was seen in people with cancer following an ND.(Reference Jalali, Hosseini and Ajami34,Reference Jakob, Classen and Stecher37) Low weight in people with cancer is concerning as it has been shown to decrease overall survivorship in allogeneic HSCT recipients(Reference Yang, Xue and Zhang52) and may be a sign of cancer cachexia, characterised by lower skeletal muscle mass. Moreover, cachexia requires medication, nutrition therapy, exercise, and psychosocial interventions and if untreated can reduce positive chemotherapy outcomes, increase side effects, and decrease survivorship.(Reference Nishikawa, Goto and Fukunishi53)
The impact on the microbiome and associated health problems have been identified in this scoping review as one of the major gaps in ND research. The understanding of the importance of the microbiome is developing, and it can affect cancer development, prevention, and treatment efficacy, conversely, treatment can also affect the microbiome.(Reference Álvarez-Mercado, del Valle Cano and Fernández54) Problematically, chemotherapy can cause gut dysbiosis – loss of diversity or changes to the gut microbiota, which affects the immune system and increases infection risk.(Reference Helmink, Khan and Hermann55) Antibiotics, commonly prescribed during cancer treatment, alter the balance of bacteria in the gut and negatively impact immunotherapy.(Reference Helmink, Khan and Hermann55) One study found that multi-drug resistant bacteria were detected in the faeces of ND patients in greater quantities than those following the comparator diet at baseline.(Reference Radhakrishnan, Lagudu and Gangopadhyay31) Interestingly, the participants included only those receiving induction chemotherapy, therefore, they would not have begun consuming their study diet, and it was not specified whether they had any prior use of antibiotics. Participants’ diets prior to the study were not assessed. No studies have been conducted to demonstrate the impact of the ND on the gut microbiome, however the ND has been shown to have reduced fibre when compared to a standard hospital diet.(Reference Maia, da Cruz and Gregianin50) As raw fruits and vegetables are limited in the ND, this may limit the number of sources of fibre and probiotics. This may have an impact on the microbiome but as no research has been conducted this remains unknown.(Reference Maia, da Cruz and Gregianin50)
Evaluation of differences in financial costs between the ND and the comparator diet was only assessed in one study. Overall, the ND had higher costs (hospital costs, other healthcare costs and, inability to work) of EU€1,760 more compared to a standard hospital diet for the duration of treatment (EU€41,769 vs EU€40,009 in 2007). These costs were determined from hospital records, questionnaires and estimated from expert opinion.(Reference van Tiel, Harbers and Terporten33) Additionally, haematological cancers have been identified as some of the most expensive cancers to treat.(Reference Goldsbury, Yap and Weber56) Further research is needed to determine financial costs associated with use of the ND in a broader context including the financial impact on people following the ND.
Understanding the impact of dietary quality on the quality of life (QoL) among individuals with cancer is paramount. QoL when consuming an ND has only been assessed in populations of children and young people with cancer, which found following an ND was associated with a decreased QoL compared to the comparator diets.(Reference Moody, Finlay and Mancuso24,Reference Moody, Baker and Santizo29) More research is needed as QoL is shown to be improved by eating with other people(Reference Yagasaki, Komatsu and Hamamoto57) and this social connection remains critical – potentially more so – in people with cancer.(Reference Williams, McMahon and Tapsell58) Currently there is limited data on how the ND affects the way people eat with others. A qualitative study of older people with cancer(Reference Yagasaki, Komatsu and Hamamoto57) found they experienced taste alterations and decreased appetite due to treatment and decreased social interactions around food but had increased family connection irrespective of food.(Reference Yagasaki, Komatsu and Hamamoto57) As QoL is lower in people with cancer(Reference Quinn, Gonçalves and Sehovic59,Reference Quinten, Coens and Ghislain60) it is important that future studies consider the impact of ND on QoL in all populations.
A key finding was of the small number of studies that looked at diet acceptability and diet adherence there was no significant difference between the ND and comparator for acceptability(Reference van Dalen, Mank and Leclercq10,Reference Moody, Finlay and Mancuso24,Reference Taggart, Neumann and Alonso30) however, adherence was significantly greater for the comparator diets.(Reference Moody, Finlay and Mancuso24,Reference Gupta, Gupta and Meena32) As these studies were conducted in 2006, 2012, and 2019, more recent studies may be needed to confirm these findings with contemporary menu designs.
Another major gap identified was how the ND impacts (i.e. procurement, cooking and safe handing, and storage) the normal routine of those with cancer, as well as their friends and family. None of the included studies reported on this area. While studies have considered what impact cancer treatment has had on families and lifestyle,(Reference Buchbinder, Longhofer and McCue61,Reference Santos, Crespo and Canavarro62) none have assessed the impact of the ND.
A major strength of the present research is that, to our knowledge, this is the first scoping review of the ND for those with cancer, to scope the literature for all health-related outcomes using a systematic search process. The present scoping review is not without limitations. Samples sizes of the included studies – mainly of RCTs – were relatively small. Few studies assessed each included outcome making it challenging to draw conclusions from this. Due to variability of the ND, each study had a different definition in addition to the comparator diet used, however, most were a form of the food safety diet. Additionally, each study had implemented prophylactic measures in their study population in addition to the ND to reduce infection making it difficult to draw conclusions across the data. Included studies did not have consistent units or measures across outcomes particularly for the outcome of weight loss. It was difficult to determine whether there was a true difference in costs between the ND and the comparator due to ambiguity in the included study.
Conclusion
Despite the need for further research into several areas related to cancer and the administration of the ND, the current evidence suggests that the ND does not serve its original purpose: to reduce the risk of infection in this population. Additionally, the ND may lead to malnutrition due to it lacking variety and providing an unpleasant experience at mealtimes, may be costing us more in the long-term, and has been shown to decrease paediatric patient quality of life. This is of particular importance as the ND is used for patients with cancer by more than 50% of hospitals in Europe and China.(Reference Carr and Halliday17–Reference Toenges, Greinix and Lawitschka22) Further research is needed regarding how the ND affects the microbiome and quality of life of its consumers as well as associated costs, but in the interim, it is important for hospitals and other institutions providing an ND to their patients to liberalise the diet wherever possible.
Abbreviations
ND: Neutropenic Diet; QoL: Quality of Life; RCT: Randomised Control Trial; HSCT: Haematopoietic Stem Cell Transplant; ESPEN: European Society for Clinical Nutrition and Metabolism; MDR: Multidrug Resistant; GvHD: Graft vs Host Disease.
Acknowledgements
The authors would like to acknowledge Owen Kuhr for reviewing the manuscript.
Author contributions
TG: Conceptualisation, methodology, article screening, formal analysis, investigation, data curation, writing: original draft, writing: review and editing, visualisation, and project administration. MH: Screening, writing: review and editing. PK: article screening, writing: review and editing. LT: article screening, writing: review and editing. EM: Conceptualisation, methodology, validation, article screening, and writing: review and editing.
Financial support
This research received no specific grant from any funding agency, commercial or not-for-profit sectors.
Conflict of interest
None.
Open access
