Van Eys 1980.

Methods	RCT Parallel design Single‐centre USA
Participants	AGE Age range not reported ‐ patients were 21 years of age or younger SEX Not reported DISEASE STATUS Patients with metastatic disease to or from bone PN: Neuroblastoma (3) Osteosarcoma (4) Giant cell tumour (1) Wilms' tumour (1) Undifferentiated ectodermal neoplasm (1) EN (usual food intake): Neuroblastoma (2) Ewing's sarcoma (2) Osteosarcoma (3) Lymphoid malignancy (2) Unknown primary (1) Only well‐nourished patients randomised CHEMOTHERAPY/RADIOTHERAPY All patients received chemotherapy
Interventions	20 patients randomly assigned: PN ‐ n = 10, EN (usual food intake) ‐ n = 10 PN ‐ consisted of a standard solution, which was adjusted for sodium or potassium content as needed. All PN was administered through a subclavian vein, except during periods when infections or other complications required temporary peripheral PN administration. Patients were allowed to continue on oral intake. EN (usual food intake) ‐ consisted of dietary advice and monitoring
Outcomes	Primary outcomes Adverse events ‐ Number of infections requiring antibiotics and infections due to sepsis Secondary outcomes ‐ Number of deaths
Notes	PN ‐ mean duration = 106.7 days, SD ± 70.93 days EN ‐ mean duration = 145.1 days, SD ± 116.98 days During an episode of chemotherapy, PN was administered for a minimum of 10 days, starting 3 days before chemotherapy was administered. Each chemotherapy episode could have had nutritional sub‐episodes in which the patient crossed over from PN to control, or vice versa, as nutritional status mandated.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote (correspondence from author): "randomization was assigned by computer generated random numbers by the Department of Biostatistics at the University of Texas MD Anderson Cancer Center"
Allocation concealment (selection bias)	Low risk	Quote (correspondence from author): "When a new eligible patient was admitted, the assignment was obtained from the Statistician by telephone. The physicians had no advanced knowledge what the assignment was going to be."
Blinding of participants and personnel (performance bias) All outcomes	High risk	Comment: Clinician/person delivering treatment: not discussed (not possible as PN obvious) Participants: not discussed (not possible as PN obvious)
Blinding of outcome assessment (detection bias) All outcomes	Low risk	Comment: not discussed. All outcomes objective, so lack of blinding unlikely to influence outcome measurements.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Comment: Patients randomised: PN n = 10 EN n = 10 Withdrawals described: EN n = 3 One randomised EN patient was lost to follow up. Two EN patients deteriorated and were put onto PN. Patients analysed: PN n = 10 EN n = 9 From tables, all PN patients are included in analyses (n = 10) and 9 EN patients are included in analyses (n = 9). Intention‐to‐treat analysis was therefore not undertaken as drop‐outs we not imputed.
Selective reporting (reporting bias)	Low risk	Comment: Outcomes described in trial as being measured but results not reported: Biochemical values, but this is justified ‐ the study was not designed to, nor could it be expected to answer questions regarding the effect of PN on white cell count, lymphocyte count, drug toxicity, and biochemical parameters, and these are routine investigations in clinical care.
Other bias	Low risk	Comment: None noted