Stephens 2012.

Study characteristics
Methods	Study design: RCT Study grouping: parallel group Unit of randomisation: individuals Power (power sample size calculation, level of power achieved): preferred sample size (n = 111) was predetermined by a power analysis using the G*Power 3.1 software (Faul 2007) with medium effect size (0.30), α = 0.05, power = 0.80, 2 groups, and 3 measurements; loss of statistical power by missing of 17% of 210 total measurements (measurements each for 70 participants) Imputation of missing data: for missing data points in items: missing value imputation methods using the expectation maximisation (EM) approach (according to authors, imputed values rounded to the nearest whole number) and computation of maximum likelihood estimation (as though there were no missing data)
Participants	Country: USA Setting: state‐supported universities Age: mean = 20.9 (SD = 0.95); range = 19 ‐ 23 years Sample size (randomised): 70 Sex: 62 women, 8 men Comorbidity (mean (SD) of respective measures in indicated, if available) at baseline: perceived stress (PSS): IG: 20.23 (6.37), CG: 20.37 (5.89) Population description: baccalaureate nursing students enrolled full‐time and in clinical nursing course at 2 university colleges of nursing Inclusion criteria: 1) full‐time status at 1 of the 2 universities; 2) enrolled in a clinical course; 3) between the ages of 19 ‐ 23; 4) currently have an active mobile phone account; 5) currently have the ability to send/receive text messages; and 6) have Twitter account or be willing to establish one prior to beginning of the study Exclusion criteria: not specified Attrition (withdrawals and exclusions): post‐intervention: 6 did not complete data collection (IG: 3/35 (8.6%); CG: 3/35 (8.6%)); 1‐month follow‐up: 8 did not complete data collection (IG: 3/35 (8.6%); CG: 5/35 (14.3%)); 1 student not completing time 2 (T2) completed time 3 (T3) assessment Reasons for missing data: dropped out of nursing programme (n = 2); never took time to set up Twitter account (n = 1); reasons for other missing data not specified
Interventions	Intervention: educational intervention by Twitter to enhance resilience (n = 35) delivery: online; by Twitter providers: researcher (Teresa M Stephens, PhD student) duration of treatment period and timing: 6 weeks; tweets sent on varying days of week and at varying times to avoid predictable schedule, 4 tweets a week description: educational intervention to increase resilience and social support and decrease perceived stress each week, 4 tweets are sent to participants by twitter account; participants may choose to respond or not intervention focused on enhancing protective factors found to be important in development/enhancement of resilience: 1) social support; 2) positive emotions; 3) humour; 4) knowledge of health behaviours; 5) self‐knowledge; and 6) effective coping TWITTER SCRIPT: WEEK 1 – SOCIAL SUPPORT: a) Monday: Call or visit someone each day this week who gives you support. Tell us about it; b) Wednesday: Who helps you the most with the stress of being a nursing student? How do they help you?; c) Friday: Who loves you “no matter what”? Do you rely on them when feeling stressed?; and d) Saturday: Who helps you stay on track or do what is best for you to remain positive and healthy? WEEK 2‐ POSITIVE EMOTIONS: a) Tuesday: Make your thoughts and words this week be positive. Encourage others to do the same; b) Wednesday: What have you learned from past mistakes or failures?; c) Friday: Who is the most positive influence in your life? What can you learn from him/her?; and d) Saturday: What are you thankful for? WEEK 3 – HUMOUR: a) Monday: Laugh out loud at least once a day. Try smiling at everyone you meet; b) Wednesday: Laughter is a great stress‐buster! Who/what makes you laugh?; c) Friday: Don’t forget to laugh at yourself. Humour can be found in almost every situation; and d) Sunday: Spend some time with someone who enjoys life and knows how to laugh. Learn from them WEEK 4 – KNOWLEDGE OF HEALTH BEHAVIOURS: a) Tuesday: Do something everyday this week to improve your health (diet, exercise, sleep). Tell us about it; b) Thursday: Sleep, healthy diet, and exercise are great stress‐busters! Try using them in your own life; c) Friday: What did you do this week to be healthier? How did it make you feel?; and d) Saturday: How do you plan to improve or maintain good health? Who supports you in these efforts? WEEK 5 – SELF‐KNOWLEDGE: a) Monday: Believing in your ability to make decisions and take actions helps you succeed in the challenge you are facing; b) Wednesday: What is your greatest strength? How does this help you?; c) Friday: Look at mistakes as learning opportunities. Make a plan for the next time you face a similar situation; and d) Sunday: Who/What are your top 3 priorities? Does the way you spend your time reflect your priorities? WEEK 6 – EFFECTIVE COPING: a) Tuesday: Physical coping methods include getting enough sleep, being physically active everyday, and eating healthily. Try them!; b) Wednesday: What creates stress in your life? What helps you cope with stress?; c) Friday: Emotional coping methods include talking to someone you trust, writing in a journal, or receiving counselling. Try them!; and d) Saturday: What can you do to improve your coping skills? Did you try anything new this week? compliance: 5 engaged in twitter dialogue; similar activity throughout the intervention between IG and CG; for both groups, highest participation in the first week, with 9 responses within each group; steady decline thereafter, with the least participation noted in the last 2 weeks of the study (beginning of new semester) integrity of delivery: not specified economic information: gift card at study conclusion theoretical basis: adolescent resilience model (Haase 2004); youth resilience framework (Rew 2003); Ahern’s model of adolescent resilience: adolescent resilience as outcome of triadic influences of risk, protection and interventions; intervention loosely based on National Center for Victims of Crime (NCVC 2005) Virginia resilience project (Reach In. Reach Out. Finding Your Resilience) Control: attention control (same number of tweets received) (n = 35) CONTROL delivery: online; by Twitter providers: researcher (Teresa M Stephens, PhD student) duration of treatment period and timing: 6 weeks; tweets sent on varying days of week and at varying times to avoid predictable schedule, 4 tweets a week description : Tweets consisting of nursing trivia or questions related to basic nursing knowledge; same style as tweets in IG (questions and statements) TWITTER SCRIPT: WEEK 1 – a) Monday: Check out the CDC website: www.cdc.gov; b) Wednesday: How many bones are in the human body?; c) Friday: What is the bell of the stethoscope used for? and d) Sunday: What is a naevus? WEEK 2: a) Tuesday: Bruxism is teeth grinding during sleep; b) Wednesday: What is a bruit?; c) Friday: How do you determine the mean arterial pressure?; and d) Saturday: Where is the spleen? WEEK 3: a) Monday: A medication's half‐life is the time it takes for 1/2 of the drug to be eliminated from the body; b) Wednesday: What does a Holter monitor do?; c) Friday: Emboli come in may forms: blood clot, fat, air, or amniotic fluid; and d) Sunday: R bronchus is longer and straighter than the L increasing the risk of right lobe aspiration pneumonia WEEK 4: a) Tuesday: Antidiuretic hormone is stored in the posterior pituitary gland; b) Thursday: Plain D5W is rapidly metabolised in children, leaving free water which can result in cerebral oedema. c) Friday: What is a low‐residue diet?; and d) Saturday: What are S/S of an allergic reaction? WEEK 5: a) Monday: Weight gain is an early symptom of congestive heart failure due to accumulation of fluid; b) Wednesday: If amniocentesis fluid contains Barr bodies, what is the sex?; c) Friday: The therapeutic serum level for Dilantin is 10 ‐ 20 mcg/mL; and d) Sunday: Who was known as the “angel of the battlefield?” WEEK 6: a) Tuesday: Morphine sulphate can suppress respiration and respiratory reflexes, such as cough; b) Wednesday: What is Glucagon?; c) Friday: The parathyroid glands regulate the calcium level in the blood; and d) Saturday: What are Fluorescein drops used for? compliance: 4 engaged in twitter dialogue; similar activity throughout the intervention between IG and CG; for both groups, highest participation in the first week, with 9 responses within each group; steady decline thereafter, with the least participation noted in the last 2 weeks of the study (beginning of new semester) integrity of delivery: not specified economic information: gift card at study conclusion theoretical basis: not specified
Outcomes	Outcomes collected and reported: resilience ‐ CD‐RISC perceived stress ‐ PSS sense of support ‐ Sense of Support Scale Time points measured and reported: 1) pre‐intervention; 2) post‐intervention (within 1 week following last tweet; and 3) 1‐month follow‐up (1 month post‐intervention) Adverse events: not specified
Notes	Contact with authors: We contacted the authors for the number of participants analysed in each group at each time point (Stephens 2018 [pers comm]). Study start/end date: not specified Funding source: research grant from Gamma Chi Chapter of Sigma Theta Tau International Declaration of interest: not specified Ethical approval needed/obtained for study: verbal approval obtained from the appropriate administrative personnel at both universities early in the planning process; IRB approval was granted by both institutions prior to the recruitment of participants and any data collection Comments by authors: not relevant Miscellaneous outcomes by the review authors: dissertation Correspondence: Teresa Maggard Stephens; now: East Tennessee State University; now: Stephenstl@etsu.ed
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Quote: "A computerized random number generator (www.randomizer.org) was used to randomly select half the participating students at each institution as the experimental group and half as the attention placebo control group." Quote: "An independent samples t‐test was used to determine if there were statistically significant differences between the experimental and control groups on age. Results indicate there were no statistically significant differences, t(68) = .47, p = .49." Quote: "According to the chi‐square analysis, there were no statistically significant differences between control and experimental groups on race, χ²(1, N =70) = 1.01, p = .31; and there were no statistically significant differences between control and experimental groups on gender, χ²(1, N = 70) = .56, p = .45." Judgement comment: investigators describe a random component in the sequence‐generation process (computer random‐number generator); verified baseline comparability of groups for some sociodemographic characteristics (age, gender, race); baseline comparability for other sociodemographic characteristics (e.g. high school education, employment, health behaviours, sources of financial/emotional support) and outcome variables (see Table 24, statistical (non)significance) unclear
Allocation concealment (selection bias)	Unclear risk	Judgement comment: insufficient information about allocation concealment to permit judgement of ‘Low risk’ or ‘High risk’
Blinding of participants and personnel (performance bias) Subjective outcomes	High risk	Judgement comment: no blinding of study personnel (intervention is provided by Twitter by only 1 researcher who also performs outcome assessment); blinding of participants unclear; outcomes are likely to be influenced by lack of blinding
Blinding of outcome assessment (detection bias) Subjective outcomes	High risk	Judgement comment: no blinding of outcome assessment (outcome assessment by the same researcher who provides the intervention by Twitter) and the outcome measurement is likely to be influenced by lack of blinding
Incomplete outcome data (attrition bias) All outcomes	Low risk	Quote: "All 70 participants completed T1 data collection." Quote: "A total of six students did not participate in T2 data collection, three from the experimental group (8.6%) and three from the control group (8.6%)." Quote: "A total of eight students did not participate in T3 data collection, three from the experimental group (8.6%) and five from the control group (14.2%)." Quote: "Nine items (Time 2) and one item (Time 3) were determined to be missing at random and were replaced via missing value imputation methods using the expectation maximization (EM) approach." Quote: "The EM method was used to compute missing values for the appropriate scale at the specified time for the missing items. Imputed values were rounded to the nearest whole number and the maximum likelihood estimation was computed as though there were no missing data." Quote: "According to Krueger and Tian (2004), MLM can be used to describe nonlinear relationships across time in a longitudinal dataset with multiple missing data points. This method was chosen over the repeated measures analysis of variance (RM ANOVA) because the MLM can accommodate flexible time schedules, missing data points and because of its emphasis on patterns of change." Quote: "In this study, six participants did not complete the data collection at Time 2 (three from the control group and three from the experimental group), and eight participants did not complete the data collection at Time 3 (three from the experimental group and five from the control group). Two participants did not complete the data collection because they dropped out of the nursing program and another student stated she never took the time to set up her Twitter account. The other students did not give a reason for not completing the data collection." Judgement comment: information received from authors on number of participants analysed in each group and imputation methods: "Each group contained 35 participants (total n=70). For the missing data: I used the expectation maximization (EM) approach. The EM method was used to compute missing values for the appropriate scale at the specified time for the missing items. Imputed values were rounded to the nearest whole number and the maximum likelihood estimation was computed as though there were no missing data."; reasons for missing data unlikely to be related to true outcome (relative balance in missing data between groups: T1: none missing; T2: IG: n = 3, CG: n = 3; T3: IG: n = 3; CG: n = 5); for missing values in single items: missing value imputation methods using the expectation maximization (EM) approach and computation of maximum likelihood estimation
Selective reporting (reporting bias)	Low risk	Judgement comment: no study protocol available but it is clear that the published reports include all expected outcomes, including those that were prespecified