| Study |
Bias |
| Randomisation process |
Deviations from intended interventions |
Missing outcome data |
Measurement of the outcome |
Selection of the reported results |
Overall |
| Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
Authors' judgement |
Support for judgement |
| Bazzano 2014 |
Some concerns |
"We used a computer‐generated, blocked randomization, stratified by sex, to allocate participants to 1 of the 2 diet groups." No information on allocation concealment. "Demographic characteristics and cardiovascular risk factors were similar between groups." No formal hypothesis tests were reported, but baseline values in Table 1 showed no apparent imbalance. |
Low risk of bias |
"...dietitians were not blinded to the study hypothesis" No information about participant blinding, however, blinding of participants for these dietary interventions is highly unlikely. No reasons or details about deviations from intended interventions reported. "All analyses were conducted using the intention‐to‐treat principle" n=148 randomised and n=148 analysed. |
Low risk of bias |
16/75 (21.3%) in the intervention and 13/73 (17.8%) in the control group did not have data. "...performed sensitivity analyses to assess the robustness of our conclusions and departures from the MAR assumption." "Results of sensitivity analyses using multiple imputation techniques to impute missing values were consistent with those presented in our primary analyses." |
Low risk of bias |
"Body weight ..... were measured to the nearest 0.1 kg and 0.1 cm, respectively, using a single calibrated scale" No evidence that a different method was used across intervention groups to measure the outcome. "....all outcome assessors were blinded to the diet group assignment." |
Some concerns |
"Study protocol and data set: Not available." No mention of pre‐specified data analysis plan. Trial registry outcomes amended after study publication; but trial registry entry at date prior to study publication mentions the outcome and timepoints. All time points mentioned in trial registry at date prior to study publication, were reported on in publication. Analysis intentions are not reported in sufficient detail to enable an assessment. |
Some concerns |
No specific information on allocation concealment. Lack of information in the pre‐registered methods. |
| Benassi‐Evans 2009 |
Some concerns |
"Subjects were randomised to one of two dietary interventions…" No further information about allocation sequence or concealment. "Subject characteristics were not significantly different between groups at baseline." |
Low risk of bias |
No information, however, blinding of participants and people delivering these dietary interventions is highly unlikely. No reasons or details about deviations from intended interventions reported. No details about analysis reported, but looks to be ITT (number randomised per group = number analysed per group). |
High risk of bias |
No details about attrition reported, but some attrition is likely after 52 weeks and implied: "In the subjects who completed the 12 weeks intensive weight loss and 9 months weight maintenance..." |
Low risk of bias |
"Volunteers fasted overnight and were weighed (model AMZ14; Mercury Digital Scales, Tokyo, Japan) wearing light clothing and no shoes at weeks 0, 12 and 52." No evidence that a different method was used across intervention groups to measure weight. Weight is not an observer‐reported outcome involving judgement. |
Some concerns |
No mention of pre‐specified data analysis plan. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
No information on how randomisation was done or whether, and how, allocation concealment was ensured. Attrition is implied, but no numbers are reported; some missingness could have and likely did depend on the true value of the outcome. No pre‐registered document available to assess analysis intentions. |
| Ebbeling 2007 |
Low risk of bias |
"Enrolled participants were entered sequentially onto a list of random group assignments prepared in advance by the study statistician, with stratification by sex and ethnicity/race..." "To avoid any bias in assigning participants to diet groups, staff conducting recruitment and enrollment were masked to sequence. The study director assigned participants to groups." No problematic differences in baseline characteristics between the two groups |
Low risk of bias |
Masking: none (open‐label). No reasons or details about deviations from intended interventions reported. ITT with imputation for missing values. |
High risk of bias |
8/36 (22%) participants in the intervention and 14/37 (38%) in the control group did not have data for this outcome. "For body weight after dropout, an increase of 1 kg per year was imputed starting from the last available measurement." Differential attrition and high attrition in both groups, with no reporting of reasons, but it is likely that with such high attrition, missingness is possibly and likely to depend on the true value. |
Low risk of bias |
Dieticians measured at each visit using an electronic scale. No evidence that a different method was used across intervention groups to measure weight. "Data were collected by personnel who were masked to group assignment". |
Some concerns |
No mention of a pre‐specified analysis plan; trial registry does not mention the outcome and timepoint in any versions. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
Differential attrition and high attrition in both trial arms, with no reporting of reasons; therefore missingness could and is likely to be dependent on the true value of the outcome. Lack of information in the pre‐registered methods. |
| Foraker 2014 |
Some concerns |
"Women were randomized using stratified randomization, based on BMI (< 30 kg/m2 vs. ≥ 30 kg/m2) to ensure equal distribution of overweight and obese participants into each intervention group." No information on allocation concealment. No problematic imbalance in baseline characteristics between groups |
Low risk of bias |
Double‐masked (investigator and outcome assessor). Deviations reported are expected to arise in usual care e.g. personal/family medical issues, time constraints. Modified ITT (completers analysed "numbers still under follow‐up"). |
High risk of bias |
31/38 (82%) participants in the intervention group and 27/41 (66%) in the control group did not have data for this outcome. Very high attrition in both groups, with limited reporting of reasons, but it is likely that with such high attrition, missingness is possibly and likely to depend on its true value. |
Low risk of bias |
Anthropometric tests at each visit included measurement of body weight. No evidence that a different method was used across intervention groups to measure weight. Double‐masked (investigator and outcome assessor). |
Some concerns |
No mention of pre‐specified data analysis plan; trial registry at date prior to study publication does not mention the outcome and timepoint (mentions only number of women who lose weight as outcome). Analysis intentions are not reported in sufficient detail to enable an assessment, but weight findings are emphasised as part of the primary aim in the paper and trial registry, with limited reporting of these findings in the paper and none reported in the abstract. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
No specific information on allocation concealment. Differential and very high attrition in both groups, with limited reporting of reasons; therefore missingness could and is likely to be dependent on the true value of the outcome. Lack of information in the pre‐registered methods. |
| Frisch 2009 |
Some concerns |
"Participants (n = 200) were randomly assigned by computer‐generated random number lists in two equal groups…" No information on allocation concealment. No problematic imbalance in baseline characteristics between groups suspected |
Low risk of bias |
Masking none (open‐label). Deviations reported are expected to arise in usual care e.g. non‐compliance. ITT analysis conducted. |
High risk of bias |
15/100 (15%) participants in the intervention group and 20/100 (20%) in the control group did not have data for this outcome. "Data were evaluated following the intention‐to‐treat and the per‐protocol method. Missing data were replaced with baseline data in the intention‐to‐treat analysis." "Results did not differ when the per protocol method was used (data not shown)." 13/100 in intervention group and 16/100 in control group withdrew due to non‐compliance, which may be related to change in weight (missingness possibly influenced by true value). |
Low risk of bias |
Calibrated electronic scale. No evidence that a different method was used across intervention groups to measure weight. Masking: None (open label) from trial registry. Weight is not an observer‐reported outcome involving judgement. |
Some concerns |
No mention of pre‐specified analysis plan; trial registry at date prior to study publication mentions the outcome and the timepoint. The time point mentioned in trial registry at date prior to publication, is reported on in publication. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
No specific information on allocation concealment. Attrition reported in both groups, with high attrition in the control group and the majority of loss to follow up in both groups due to withdrawals for non‐compliance; therefore missingness is likely to be related to the true value of the outcome. Analyses were done per protocol and using ITT with baseline values carried forward, neither of which are considered appropriate to address bias due to missing data. Lack of information in the pre‐registered methods. |
| Gardner 2007 |
Low risk of bias |
"Randomization was conducted in blocks of 24 (6 per treatment group) and occurred by having a blinded research technician select folded pieces of paper with group assignments from an opaque envelope." No problematic imbalance in baseline characteristics between groups suspected |
Low risk of bias |
Masking: single ("Clinic and laboratorystaff members were blinded to treatment assignment.") Deviations reported are expected to arise in usual care e.g. moved away, disliked diet, pregnancy. ITT analysis conducted. |
High risk of bias |
9/77 (12%) participants in the intervention group and 18/79 (23%) in the control group did not have data for this outcome. ITT with baseline values carried forward. "Also for exploratory purposes, all analyses of weight and secondary outcome measures were tested using only available data, without using baseline values carried forward for missing data or other imputation methods. There were no substantive differences in any of these findings compared with the analyses with baseline values carried forward and, therefore, only the primary analyses are presented." 5/9 participants in the intervention group and 5/18 in the control group withdrew due to health or "other" reasons; resulting in differential attrition and high attrition in the control group. With such high and differential attrition as well as vague reasons missingness could and is likely to depend on the true value of the outcome. |
Low risk of bias |
"Body weight was measured to the nearest 0.1 kg on a calibrated clinical scale." No evidence that a different method was used across intervention groups to measure weight. "Clinic and laboratory staff members were blinded to treatment assignment." |
Some concerns |
No mention of pre‐specified analysis plan, but trial registry at date prior to study publication me ntions the outcome and timepoint. The time point mentioned in trial registry at date prior to publication is reported on in publication. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
Differential attrition and high attrition reported in the control group, with the some attrition in both groups due to vague reasons such as 'health' or 'other'; therefore missingness could and is likely to be dependent on the true value of the outcome. Analyses were done per protocol and using ITT with baseline values carried forward, neither of which are considered appropriate to address bias due to missing data. Lack of information in the pre‐registered methods. |
| Griffin 2013 |
Low risk of bias |
"The study consisted of a random block design per every four participants. Random allocation was prepared by a researcher not involved in participant enrolment or assignment." "random number table"; "sealed opaque envelopes" (trial registry). No problematic differences observed between the two groups for baseline characteristics |
Low risk of bias |
"Participants were blind to diet macronutrient composition and were informed eating plans differed only in the type of meat (i.e. red or white). "....an inability to double‐blind the intervention were limitations". Deviations reported are expected to arise in usual care e.g. did not like diet. ITT analyses conducted. |
High risk of bias |
15/36 (42%) participants in the intervention group and 20/35 (57%) in the control group did not have data. ITT with last observation carried forward, and sensitivity analysis comparing completers with ITT with completers. "HP completers demonstrated significantly greater weight loss than HC at 6 (p=0.034) but not 12 months (p=0.07)". "In all participants (intention‐to‐treat), no significant between diet differences were observed (Table 2; figure S2)". 5/7 participants in the intervention group and 5/11 in the control group discontinued due to illness, too difficult and not losing weight, resulting in differential and very high attrition. With such high and differential attrition as well as vague reasons missingness could and is likely to depend on the true value of the outcome. |
Low risk of bias |
Weight was measured to the nearest 0.1 kg on a digital platform scale at every visit, with participants wearing light clothing and no shoes. No evidence that a different method was used across intervention groups to measure weight. Single blind (participants only). Weight is not an observer‐reported outcome involving judgement. |
Some concerns |
No mention of pre‐specified analysis plan, but trial registry at date prior to study publication mentions the outcome and timepoint. The time point mentioned in trial registry at date prior to publication is reported on in publication. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
Differential and very high attrition in both groups, for reasons such as 'illness', the diets being 'too difficult' and for 'not losing weight'; therefore missingness could and is likely to be dependent on the true value of the outcome. Completers analyses and ITT with last observation carried forward conducted, neither of which are considered appropriate to address bias due to missing data. Lack of information in the pre‐registered methods. |
| Jesudason 2013 |
Low risk of bias |
From registry: "Random allocation using computer software "Clinstat"" and "Allocation was concealed when it was done by central randomisation by computer." No significant differences between groups at baseline for four tested variables |
Low risk of bias |
"Although the investigators were blinded until the end of the study, subjects and dietitians were unblinded." Deviations reported were expected to arise in usual care, e.g. unwilling/unable to follow protocol. Appeared to be an modified ITT; completers analysed; however n = 1 participant excluded from control group analysis for some outcomes with no explanation provided |
High risk of bias |
95/164 (58%) participants in the intervention and 92/159 (58%) in the control group did not have data available for this outcome. "As shown in Table 7, baseline characteristics of subjects who dropped out before the end of the trial were matched between HP and HNP groups and were similar to the whole cohort, which suggested dropouts did not bias the results. The last recorded measurements of participants who did not complete the 24‐mo trial..." Very high attrition in both groups with missing data due to non‐compliance/unwilling or unable to follow trial protocol and diet being unpalatable; 68/95 (72%) in the intervention and 59/91 (65%) in the control group, which may be related to change in weight thus missingness possibly and likely influenced by true value of the outcome |
Low risk of bias |
Weight measured at baseline, 3, 6, 12, 24 months (trial registry). No evidence that a different method was used across intervention groups to measure weight. "Although the investigators were blinded until the end of the study, subjects and dietitians were unblinded." |
Some concerns |
No mention of prespecified analysis plan, but trial registry at date prior to study publication mentioned the outcome and time point. The time point mentioned in the trial registry at the date prior to publication is reported on in the publication. Analysis intentions were not reported in sufficient detail to enable an assessment. |
High risk of bias |
Very high attrition reported in both groups for reasons such as 'non‐compliance', 'unwilling or unable to follow protocol' and diets 'being unpalatable'; therefore missingness could and was likely to be related to the true value of the outcome. Lack of information in the pre‐registered methods |
| Klemsdal 2010 |
Some concerns |
No information about randomisation or allocation concealment. Footnote to Table 1: "There were no significant differences between groups." |
Low risk of bias |
Masking none (open‐label). Deviations reported are vague, but those reported are expected to arise in usual care e.g. non‐attendance of study visits. ITT analysis: "Analyses were done on an intention to treat basis, with the last value carried forward for non‐completers". |
High risk of bias |
22/100 (22%) participants in the intervention group and 16/102 (16%) in the control group did not have data. "Analyses were done on an intention to treat basis, with the last value carried forward for non‐completers". No mention of sensitivity analyses. No reasons for loss to follow‐up or withdrawal provided, so cannot judge if related to participants’ health status. High attrition in the intervention group. With such high attrition, missingness could and is likely to depend on the true value of the outcome. |
Low risk of bias |
Body weight ...were measured with standard methods. No evidence that a different method was used across intervention groups to measure weight. Open label. Weight is not an observer‐reported outcome involving judgement. |
Some concerns |
No mention of pre‐specified analysis plan, but trial registry at date prior to study publication mentions the outcome and timepoint. The time point mentioned in trial registry at date prior to publication is reported on in publication. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
No information on how randomisation was done or whether, and how, allocation concealment was ensured. Attrition reported in both groups, with high attrition in the intervention group, and no reasons provided; therefore missingness could and is likely to be dependent on the true value of the outcome. Lack of information in the pre‐registered methods. |
| Layman 2009 |
Some concerns |
No information on how the randomisation sequence was generated. No information about allocation concealment. "Randomization after stratifying participants for gender, age, BMI, and TC was successful with no significant differences between treatment groups in baseline characteristics." |
Low risk of bias |
"Due to study design of the diet treatment, it was not possible to blind subjects and research staff to group assignment." Deviations reported are expected to arise in usual care. "Between the 4‐ and 12‐mo data collection points, 21 participants withdrew from the CHO group (16 stated personal reasons and 5 stated unable to comply with diet protocol) and 11 participants withdrew from the PRO group (8 stated personal reasons and 1 stated unable to comply with diet protocol). Modified ITT conducted. |
High risk of bias |
23/64 (36%) participants in the intervetion group and 31/61 (52%) in the control group did not have data for this outcome. Modified ITT with last observation carried forward and sensitivity analysis including only completers, comparative data not shown. "Assessing long‐term weight loss using ITT or completers, weight loss did not differ between treatments at 12 mo. However, the PRO group had a greater number of participants (64%) complete the 12‐mo study than the CHO group (45%; P0.05). For completers, the mean weight loss was 23% greater for the PRO group, but individual weight loss ranged from 0.6 to 30.8 kg within the PRO group and 1.7 to 23.2 kg for the CHO group..." Very high attrition in the control group and differential attrition, and differential attrition by inability to comply with diet: 11 participants withdrew from the PRO group (8 stated personal reasons and 1 stated unable to comply with diet protocol); 21 participants withdrew from the CHO group (16 stated personal reasons and 5 stated unable to comply with diet protocol). Inability to comply may be related to change in weight (missingness possibly influenced by true value). |
Low risk of bias |
Body weight was measured using an electronic scale (Tanita). No evidence that a different method was used across intervention groups to measure weight. "Due to study design of the diet treatment, it was not possible to blind subjects and research staff to group assignment." Weight is not an observer‐reported outcome involving judgement. |
Some concerns |
No mention of pre‐specified data analysis plan. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
No specific information on allocation concealment. Attrition reported in both groups, with differential attrition and very high attrition in the control groups. Differential attrition also detected for inability to comply with diet. As a result, missingness is considered likely to be dependent on the true value of the outcome. Completers analyses and ITT with last observation carried forward conducted, neither of which are considered appropriate to address bias due to missing data. No pre‐registered document available to assess analysis intentions. |
| Lim 2010 |
Some concerns |
No information about randomisation or allocation concealment. Baseline characteristics table showed small differences, consistent with differences resulting by chance with small sample sizes. |
Low risk of bias |
No information, however, blinding of participants and people delivering these dietary interventions is highly unlikely. Deviations reported are expected to arise in usual care e.g. travel, work commitment, did not like diet. Modified ITT, included all partial data contributed by non‐completers. |
High risk of bias |
13/30 (43%) participants in the intevention group and 15/30 (50%) in the control group did not have data. ITT with all partial data contributed by non‐completers included in linear mixed effects model. Results of ITT not presented and no sensitivity analyses performed (author correspondence). Very high attrition in both groups, with limited reporting of reasons ("unspecified"), but it is likely that with such high attrition, missingness is possibly and likely to depend on the true value. |
Low risk of bias |
Body weights recorded in light clothing (Mercury digital scales). No evidence that a different method was used across intervention groups to measure weight. Weight is not an observer‐reported outcome involving judgement. |
Some concerns |
No mention of pre‐specified data analysis plan. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
No information on how randomisation was done or whether, and how, allocation concealment was ensured. Very high attrition reported in both groups, with limited reporting of reasons; therefore missingness could and is likely to be related to the true value of the outcome. ITT with partial data from non‐completers using linear mixed effects model was conducted, but results are not available. No pre‐registered document available to assess analysis intentions. |
| Sacks 2009 |
Low risk of bias |
"Random assignments to one of four diet groups were generated by the data manager at the coordinating center on request of a study dietitian, after eligibility of a participant was confirmed." Independent trial unit, fully funded by public funds and large trial; it was likely the sequence was random. " Baseline characteristics were similar among participants assigned to the four diets..." |
Low risk of bias |
"Blinding was maintained by the use of similar foods for each diet. Staff and participants were taught that each diet adhered to principles of a healthful diet and that each had been recommended for long‐term weight loss, thereby establishing equipoise". True blinding is known to be a challenge in these diet trials. Deviations reported are expected to arise in usual care e.g. moved away, did not appear at scheduled appointments. ITT analysis conducted: "We performed an intention‐to‐treat analysis in which long‐term weight loss for persons who withdrew from the study early (after at least 6 months of participation) was imputed on the basis of a rate of 0.3 kg per month of regained weight..." |
High risk of bias |
33/201 (16%) participants in the intervention group and 35/204 (17%) did not have data for this outcome. Single imputation "intention‐to‐treat analysis in which long‐term weight loss for persons who withdrew from the study early (after at least 6 months of participation) was imputed on the basis of a rate of 0.3 kg per month of regained weight." No sensitivity analysis conducted (author correspondence). No reasons for loss to follow‐up or withdrawal provided, so not able to judge if missingness is possibly and likely to depend on the true value. |
Low risk of bias |
Calibrated hospital scale. No evidence that a different method was used across intervention groups to measure weight. "Investigators and staff who measured outcomes were unaware of the diet assignment of the participants". |
Some concerns |
No mention of pre‐specified analysis plan, but trial registry at date prior to study publication mentions the outcome and timepoint. The time point mentioned in trial registry at date prior to publication is reported on in publication. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
Attrition reported in both groups, with no reasons provided; therefore it is not possible to judge how likely it is that missingness was dependent on the true value of the outcome. Lack of information in the pre‐registered methods. |
| Tay 2008 |
High risk of bias |
From registry: "random allocation using computer software clinstat" "allocation is not concealed". "Baseline characteristics between diet groups were not different". |
Low risk of bias |
From registry: "Open (masking not used)". Most deviations reported are expected to arise in usual care, e.g. work commitments, personal reasons. Maximum likelihood mixed‐model analysis, including all participants who started in the study. |
Low risk of bias |
22/55 (40%) in the intervention and 16/52 (31%) in the control group did not have data available for this outcome. Results of the maximum likelihood mixed‐model analysis, including all participants who started the study, correspondend with that of the completers analysis (no statistically significant difference between groups). Independent t tests conducted on baseline characteristics of study dropouts and completers showed that data were missing at random. |
Low risk of bias |
Weight was measured using calibrated electronic digital scales (AMZ14; Mercury, Tokyo, Japan). No evidence that a different method was used across intervention groups to measure weight. From registry: "Open (masking not used)". Weight is not an observer‐reported outcome involving judgment. |
Some concerns |
No mention of pre‐specified analysis plan; trial registry at date prior to study publication mentions outcome. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
Allocation was not concealed. Lack of information in the pre‐registered methods. |
| Wycherley 2012 |
High risk of bias |
"...participants were blocked, matched for age and body mass index, then randomised by the trial coordinators using computer‐generated random number allocation..." Allocation concealment: No (trial registry). From Table 3, only triglycerides were significantly different between groups at baseline (P < 0.05), of 17 characteristics measured. |
Low risk of bias |
Open (masking not used). Deviations reported are expected to arise in usual care e.g. time, unable to commit to study protocol. Modified ITT conducted. |
Low risk of bias |
26/59 (44%) participants in the intervention group and 29/64 (45%) in the control group did not have data. "...intention to treat evaluation was conducted using maximal likelihood mixed model analysis with fixed and random effects to analyse expected mean changes over time. The secondary analysis was based on the 120 participants that had outcomes assessed at week 0 and commenced the dietary programme." "The secondary maximal likelihood mixed model analysis showed a similar pattern of results to the primary completer’s analysis". Results of a sensitivity analysis not reported, but MD reported at 52 weeks is the same with completers analysis (Table 3 in Wycherley et al 2012), as with ITT analysis (section 3.2 in Lutze et al 2013), but variances differ. |
Low risk of bias |
"…body weight was measured using calibrated electronic digital scales..." No evidence that a different method was used across intervention groups to measure weight. Open (masking not used). Weight is not an observer‐reported outcome involving judgement. |
Some concerns |
No mention of pre‐specified analysis plan, but trial registry at date prior to study publication mentions the outcome and timepoint. The time point mentioned in trial registry at date prior to publication is reported on in publication. Analysis intentions are not reported in sufficient detail to enable an assessment. |
High risk of bias |
Allocation was not concealed. Lack of information in the pre‐registered methods. |
