Appendix B. Quality assessment* of randomized trials, non-randomized trials, and single group repeated measures studies.

Selection Bias / Performance Bias / Attrition Bias / Detection Bias / Other bias and comments
Study, Year / A1. Randomization / A2. Allocation concealment / A3. Comparable at baseline / B1. Same care / B2. Blinded participants / B3. Blinded healthcare workers / C1. Equal follow-up time / C2. Comparable for treatment completion / C3. Comparable for outcome data / D1. Appropriate length of follow-up / D2. Precise definition of outcome / D3. Valid and reliable outcome / D4. Investigators blind to exposure / Key
Yes
Unclear
No
Auditory
Arnon, 2006 / Did not use appropriate statistical methods (repeated measures) to analyze data
Cassidy, 2009
Coleman, 1997 / Unconventional study design; statistical analysis not well-defined
Farhat, 2010 / Unclear how data were combined in statistical analysis; incomplete reporting of outcomes
Garunkstiene, 2014
Keidar, 2014
Krueger, 2010
Loewy, 2013
Loewy, 2015 / Crossover study with unclear carryover effects, as no washout time specified
Picciolini, 2014 / No repeated measures analysis for physiologic and behavioral outcomes. Data collected across the 21 days was combined, eliminating the ability to see changes over time.
Webb, 2015 / Length of follow-up until discharge; dose of intervention likely varied among subjects. No measure of total dose was reported.
Zimmerman, 2013 / Use of a historical control population. Length of follow-up until discharge; dose of intervention likely varied among subjects. No measure of total dose was reported.
Gustatory
Lee, 2015
Kinesthetic
Vignochi, 2012 / Length of follow-up until discharge; dose of intervention likely varied among subjects. No measure of total dose was reported.
Olfactory
Ӧzdemir, 2013 / Length of follow-up until discharge; dose of intervention likely varied among subjects. No measure of total dose was reported.
Yildiz, 2011 / Length of follow-up until transition to oral feeds; dose of intervention likely varied among subjects. No measure of total dose was reported.
Tactile
Azevedo, 2012 / Single group repeated measure study, unable to address questions on selection bias; post hoc power calculation
Bahman Bijari, 2012
Bauer, 1998 / Single group repeated measure study, unable to address questions on selection bias; possible use of inappropriate statistical methods for type of data
Bier, 1996 / No intent-to-treat analysis, as control patients that crossed over into the treatment group were excluded from analysis; dose of intervention likely varied among subjects
Bohnhorst, 2001 / Single group repeated measure study, unable to address questions on selection bias
Chen, 2008
de Macedo, 2007 / Minimal description of the intervention
Feldman, 2002a Feldman, 2002b Feldman, 2014 / Studies assessed same population
Ferber, 2002
Ferber, 2005 / Ferber 2002: Did not use appropriate statistical methods (repeated measure)
Fӧhe, 2000 / Single group repeated measure study, unable to address questions on selection bias
Harrison, 2000
Im, 2009
Im & Kim, 2009 / Studies assessed same population (addition of control group in Im & Kim 2009); unclear calculation of outcome (Im 2009), did not account for repeated measures (Im & Kim 2009)
Lee, 2011 / Mothers permitted to speak, sing, and whisper to their infants, which may have varied among subjects.
Legault, 1995 / Did not use appropriate statistical methods (repeated measures); crossover study with unclear carryover effect, wash-out period varied among subjects
Maastrup, 2011 / Single group repeated measure study, unable to adequately address question on selection bias
Messmer, 1997 / Single group repeated measure study, unable to adequately address question on selection bias; did not use appropriate statistical methods (non-parametric)
Miles, 2006 / Large variation in skin-to-skin care between groups; statistical methods not well-defined
Ramamathan, 2001 / Length of follow-up until discharge, dose of intervention likely varied among subjects; did not use appropriate statistical methods (repeated measures); unclear calculation of outcomes
Roberts, 2000 / Dose of intervention likely varied among subjects; did not use appropriate statistical methods (repeated measures)
Rojas, 2003 / Dose of intervention varied significantly between groups
Samra, 2013 / Possible inappropriate statistical methods for weight data (repeated measures); appears to have performed a post hoc power calculation
Scher, 2009 / Did not use appropriate statistical methods (repeated measures)
Schneider, 2012 / Subjects received varying levels of exposure to the intervention
Smith, 2001
Smith, 2003 / Studies assessed same population.
Tallandini, 2006 / Length of follow-up until discharge, dose of intervention likely varied among subjects
Tӧrnhage, 1998 / Length of follow-up unclear, dose of intervention likely varied among subjects
Vision
Guyer, 2015
Vásquez-Ruiz, 2014
Multimodal
Aly, 2004
Ang, 2012 / Length of follow-up until discharge, dose of intervention likely varied among subjects
Arnon, 2014 / Crossover study with unclear carry over effects, wash out period unspecified
Cameron, 2005 / Length of follow-up dependent on status at discharge, dose likely varied among subjects
Choi, 2015 / No reporting of total dose or adherence to the intervention
Diego, 2005
Diego, 2007
Diego, 2008
Diego, 2014
Dieter, 2003
Field, 2006
Field, 2008
Gonzalez, 2009
Haley, 2012
Hane, 2015
Welch, 2013 / Dose of interventionvaried among subjects; outcomes from Hane 2015 study were blinded, but not those reported in Welch 2013.
Hernandez-Reif, 2007
Holditch-Davis, 2014 / Intervention dose not well-defined; dose of intervention likely varied among subjects
Kanagasabai, 2013
Massaro, 2009 / Dose of intervention likely varied among subjects; unclear statistical methods
Mathai, 2001
Matricardi, 2012 / Intervention dose not well-defined, likely varied among subjects
Mendes, 2008
Procianoy, 2010 / Both studies assessed the same population; unclear baseline comparison in Procianoy 2010 (did not adjust for differences in statistical analysis as in Mendes 2008)
Moyer-Mileur, 2013
Schlez, 2011
Smith, 2013
Standley, 1998 / Unclear statistical analysis; intervention dose not well-defined, likely varied among subjects
Teckenberg-Jansson, 2011 / Unclear statistical analysis; Crossover study, unclear carryover effects, washout period unspecified
Valizadeh, 2012 / Patients and caregivers were blind to type of oil, but not to massage vs. no massage
White-Traut, 1997
White-Traut, 1999 / Length of follow-up until discharge, dose of intervention likely varied among subjects
White-Traut, 2002 White-Traut, 2004 / Both studies assessed same population; blinding not specified in White-Traut 2004; possible inappropriate statistical methods for type of data
White-Traut, 2013 White-Traut, 2014
White-Traut, 2015 / White-Traut 2014: did not specify randomization methods, and unclear if comparable for outcome data (sample sizes in flow chart did not match numbers in tables)
White-Traut 2015: indicated random assignment was centrally controlled, which may have provided adequate allocation concealment
Note: Cells with two colors in a single cell indicates that some study characteristics met one condition, while other characteristic met another condition.
*Assessed for quality using a tool adapted from: National Institute for Health and Care Excellence (NICE). Appendix C: Methodology checklist: randomized controlled trials. The Guidelines Manual 2012:

Quality assessment† of systematic reviews.

Study, Year / 1. Appropriate research question development / 2. Described search methods / 3a. Searched at least two databases / 3b. Chose most appropriate databases/dates / 3c. Search likely to capture all relevant studies / 4a. Provide adequate inclusion criteria / 4b. Selection criteria applied by >1 person / 4c. Stated how study selection conflicts resolved / 4d. Summary of included/excluded studies / 4e. Included all appropriate study designs / 5. Characteristics of included studies provided / 6. Statement/assessed for publication bias / 7a. Quality assessment well-specified / 7b. Quality assessment performed by >1 person / 7c. State how QA disagreements resolved / 8a. Assessed for confounding / 8b. Assessed for sufficient sample size / 8c. Assessed for outcome reporting bias / 8d. Assessed for completeness of follow-up / 8e. Assessed for randomization / 8f. Assessed for allocation concealment / 8g. Assessed for blinding/masking / 9a. Used standard forms for data extraction / 9b. Counted >1 publications from 1 study once / 9c. Data extraction performed by >1 person / 10. Assessed and accounted for heterogeneity / 11. Appropriate methods for synthesizing results / 12. Sensitivity analyses performed / 13. Author conclusions support data quality / 14. Fully accounted for conflicts of interest / Key
Yes
Partial or unclear
No
Kinesthetic
Schulzke, 2014
Vision
Morag, 2013
Abbreviations:QA, quality assessment.
†Assessed for quality using the tool developed by: Diekemper RL, Ireland BK, Merz LR. Development of the Documentation and Appraisal Review Tool for systematic reviews. World J Meta-Anal. June 26 2015;3(3):142-150.

Appendix B |Page 1