Table 1.

Post-operative physical activity and/or diet data employed in comparisons of bariatric surgery outcomes: issues, potential solutions, and proposed preferred reporting^a

Issue	Potential action/s	Preferred reportinga
		A	B	C	D	E	F	G	H
1. No data collected on post-op PA/ diet and/or included in analyses	Highlight this point for future research direction/s	✓
2. Case-matched analysis	Usually matching undertaken for preoperative characteristicsb
	Post-op PA/diet included in analysis of comparisons of effectiveness?		✓	✓	✓			✓
	Post-op PA/diet not included in the analysis of comparisons of effectiveness?	✓
3. Data collected specifically on post-op PA/diet	Are such post-op PA/diet data are employed as “process”c vs “outcome”d variables					✓
	If data is used as processc variables				✓		✓	✓
	If post-op PA and/or diet data collected as processc variables but not actually included further in the analyses of comparisons of outcomes	✓
4. Data collected not specifically on post-op PA/diet e.g., data on quality of life parameters (SF 36, Moorehead-Aldert II, others)	Although such quality of life measures usually employed as outcomesd, they include information on, e.g., physical functioning, PA, and eating behavior. Such data could represent a way forward if used as process variables c to suggest some indication of post-op PA/diet data that could, if required, be included in analyses of comparisons of outcomes, using appropriate statistical technique/s. Drawbacks:		✓	✓		✓
	(a) Same quality of life data cannot be used as processc and outcomesd simultaneously (b) Using some data (domain) out of a quality of life measure as processc could disturb the use of whole measure as an outcomed. Such tools usually generate composite score across their domains despite individual scores for each domain
	If, however, decisions are made to use such information as process variablesc rather than outcomesd and be included in analyses of comparisons of outcomes:				✓	✓			✓
5A. Post-op PA/diet data specifically collected and included in the analyses: is any a confounder?	If post-op PA and/or diet data is included in analyses Assess a potential confounder by: (a) Formal tests of hypothesis (b) Inspect data for practically important/clinically meaningful relationship between variable and risk factor, and between variable and outcome (regardless of whether relationship is significant). If yes, variable could be a confounder (c) Appraise the measure of association before and after adjusting for potential confounder (change of ≥ 10% in estimated measure of association could suggest confounding) [81].		✓
	If variable found to be potential confounder, confounding can be dealt with by controlling, matching, randomizing, statistical control [79]. If testing for confounding is done:		✓	✓	✓			✓
5B. Post-op PA/diet data specifically collected and included in analyses: is any an effect modifier?	If post-op PA and/or diet data is included in analyses. For observational studies, test effect modifier by: Outcomes of BS (e.g., WL) are assessed across subgroups of patients of different PA levels and/or diet adherence (e.g., no adherence, moderate adherence, good adherence) [82] to see if differs depending on the level of a third variable (post-op PA or diet). Multivariable methods can also assess effect modification [81].		✓
	If variable found to be effect modifier, stratify analyses by levels/ categories of effect modifier. If testing for effect modification is done:		✓	✓	✓			✓

A: Acknowledge in limitations section the lack of inclusion of and/or control for post-op PA and/or diet in the analyses of comparisons of effectiveness. B: Explicit mention in “Materials and Methods” section and in “Statistical and Analytical Considerations” section. C: Report the specific findings of such inclusion or utilization in the “Results” section. D: Acknowledge their inclusion in the strengths of the study section regardless of whether such effects were significant or otherwise. E: Explicit mention in methods section whether post-op PA and/or diet employed as process^c vs outcome^d variables. F: Ascertain in methods section that post-op PA and/or diet data collected was actually included in analyses of comparisons of outcomes. G: Highlight in statistical analysis subsection of the methods section the test/s employed and their appropriateness. H: Report in methods section how the use of such variables as process^c variables might have disabled their simultaneous use as outcome^d variables

^aPreferred REporting of post-operative PHYsical activity and Diet data in comparisons of BS effectiveness: PRE-PHYD Bariatric

^bBe cautious of over-matching

^cPA/diet data used to increase certainty that effectiveness of various BS procedures are actually due to the procedures themselves and not due to an artifact

^dPA/diet data used to compare effectiveness of various BS procedures