articles
nature publishing group
Intervention and Prevention
12-Month Outcomes and Process Evaluation of the SHED-IT RCT: An Internet-Based Weight Loss Program Targeting Men Philip J. Morgan1, David R. Lubans1, Clare E. Collins2, Janet M. Warren3 and Robin Callister4 This article reports the 12-month follow-up results and process evaluation of the SHED-IT (Self-Help, Exercise, and Diet using Information Technology) trial, an Internet-based weight loss program exclusively for men. Sixty‑five overweight/obese male staff and students at the University of Newcastle (Callaghan, Australia) (mean (s.d.) age = 35.9 (11.1) years; BMI = 30.6 (2.8)) were randomly assigned to either (i) Internet group (n = 34) or (ii) Information only control group (n = 31). Both received one face-to-face information session and a program booklet. Internet group participants were instructed to use the study website for 3 months. Participants were assessed at baseline, 3-, 6-, and 12-month follow-up for weight, waist circumference, BMI, blood pressure, and resting heart rate. Retention at 3and 12-months was 85% and 71%, respectively. Intention-to-treat (ITT) analysis using linear mixed models revealed significant and sustained weight loss of −5.3 kg (95% confidence interval (CI): −7.5, −3.0) at 12 months for the Internet group and −3.1 kg (95% CI: −5.4, −0.7) for the control group with no group difference. A significant time effect was found for all outcomes (P < 0.001). Per-protocol analysis revealed a significant group-by-time interaction for weight, waist circumference, BMI, and systolic blood pressure. Internet group compliers (who self-monitored as instructed) maintained greater weight loss at 12 months (−8.8 kg; 95% CI −11.8, −5.9) than noncompliers (−1.9 kg; 95% CI −4.8, 1.0) and controls (−3.0 kg; 95% CI −5.2, −0.9). Qualitative analysis by questionnaire and interview highlighted the acceptability and satisfaction with SHED-IT. Low-dose approaches to weight loss are feasible, acceptable, and can achieve clinically important weight loss in men after 1-year follow-up. Obesity (2011) 19, 142–151. doi:10.1038/oby.2010.119
Introduction
Two-thirds of Australian men are considered to be overweight or obese (1) and therefore are at risk of a range of adverse physio logical and psychological consequences (2). Engaging men in weight loss programs is a public health and research priority as, compared to women, men are less likely to perceive themselves as overweight (3), attempt weight loss, or participate in weight loss programs (4) despite being more susceptible to cardiovascular disease (5). Although reasons for this lack of engagement are not well established, it appears men perceive too many barriers and/or currently available programs do not appeal to them (6). It has been reported that men desire weight loss programs that are convenient, provide individualized feedback, and include participants they can identify with (6). Although intensive group programs that include weekly visits are seen as an important component of effective treatment, these are not practical for people who are “busy” and may not appeal to men in particular (7). Men are generally not enthusiastic
about attending structured face-to-face weight loss programs (6,7). Consequently, alternative treatment approaches such as the Internet may be more appealing and afford greater accessibility, anonymity, and convenience. Notably, in Australia, men are more likely to use the Internet than women (8). A recent systematic review and meta-analysis of online weight loss randomized controlled trials (RCTs) (9) reported that weight loss programs can be effectively delivered over the Internet. Limitations of previous studies have been identified and include short length of follow-up, no intention-to-treat (ITT) analysis, no assessor blinding, insufficient follow-up beyond the intervention and studies undertaken mainly in females (9,10). Few weight loss studies have been conducted in men (11,12), and none of these have evaluated an online weight loss RCT. In addition, process data from online weight loss programs are seldom reported, yet such data are invaluable to inform future interventions (13); this is particularly important in
1 School of Education, Faculty of Education and Arts, University of Newcastle, Callaghan, New South Wales, Australia; 2School of Health Sciences, Faculty of Health, University of Newcastle, Callaghan, New South Wales, Australia; 3Danone Baby Nutrition, Wiltshire, UK; 4School of Biomedical Sciences and Pharmacy, Faculty of Health, University of Newcastle, Callaghan, New South Wales, Australia. Correspondence: Philip J. Morgan (
[email protected])
Received 1 February 2010; accepted 28 April 2010; published online 3 June 2010. doi:10.1038/oby.2010.119 142
VOLUME 19 NUMBER 1 | january 2011 | www.obesityjournal.org
articles Intervention and Prevention studies of men, given the limited research in this area (9). Process data are a valuable adjunct to outcome data and can provide information about the quality of an intervention, participant views on the intervention, effects of intervention components and insights into the mechanisms of change and treatment effects (13,14). Process data provide greater ability to explain and interpret findings and understand the potential generalizability of the intervention (13). We conducted the first RCT of an online weight loss program that targeted men only (15). The current article expands on the findings of this novel approach by addressing pertinent gaps in the literature, specifically long-term outcomes and process evaluation. The primary aims of the present article were to examine the 12-month outcomes of the SHED-IT (Self-Help, Exercise, and Diet using Information Technology) Internet-based program for overweight men to determine whether men were able to maintain weight loss 9-months postintervention and to comprehensively evaluate the SHED-IT program by collecting information regarding the experience of men taking part in the trial. Methods and Procedures The methods of the SHED-IT study adhered to the CONSORT statement (16). They have been published previously (15) and summarized briefly here. Participants Overweight or obese (BMI between 25 and 37 kg/m2) male staff (academic and nonacademic) and students aged 18–60 years were recruited from the University of Newcastle from advertisements placed on the University notice boards and website in late August 2007. Based on 80% power to detect a significant difference (P = 0.05, two-sided), a sample size of 18 participants for each group was needed to detect a 3 kg difference in weight loss among groups. Assuming a 20% attrition rate, a total sample of 44 subjects was required. Participants were screened for eligibility via telephone. Ineligibility criteria included a history of major medical problems that would be a barrier to physical activity, recent weight loss of ≥4.5 kg, or taking medications that might affect body weight. Participants completed a pre-exercise risk assessment screening questionnaire (17) and provided written informed consent. Ethics approval was obtained from the University of Newcastle Human Research Ethics Committee. Study design Participants were randomly and blindly allocated to one of two groups: the SHED-IT Internet group or an information only control group. The random allocation sequence was generated by a computer-based random number-producing algorithm in block lengths of six to ensure an equal chance of allocation to each group. Outcome measures were obtained from all participants at baseline (September, 2007), 3 months (December, 2007), 6 months (March, 2008), and 12 months (September, 2008) from the start of treatment. Originally the study was funded for only 6 months. Following additional funding and ethics approval, participants were contacted and invited to attend an additional assessment at 12 months. Importantly, participants had no prior warning of this assessment, and assessments took place within 2 weeks of being contacted. Measurements were taken at the University and participants were blind to group allocation at baseline assessment. Assessors were blinded to treatment allocation at all time points. An experienced anthropometrist and a trained assistant measured all participants at all time points. Information only control group The control group attended one face-to-face information session (60 min), which was identical to that of the Internet group (but without obesity | VOLUME 19 NUMBER 1 | january 2011
a 15-min online component description), and received a weight loss program booklet, but had no website access. Separate information sessions were conducted for Internet and control participants to avoid contamination. The SHED-IT (Internet) group The SHED-IT program involved one face-to-face information session on weight loss (75 min), a weight loss program booklet, use of the free study website (www.calorieking.com.au) plus 3 months of online support, and was based on Bandura’s Social Cognitive Theory (18). Participants self-monitored their weight, dietary intake and exercise, set goals and received social support; these are recognized as cornerstones of behavioral treatment (19). Participants were asked to enter their weight (in kg) once each week online and submit online daily eating and exercise diaries for the first 4 weeks, for 2 weeks in the second month and for 1 week in the third month; feedback based on diary entries was provided on seven occasions by research assistants. Participants were also able to submit questions on a website notice board which were answered weekly by the research group and accessible to all Internet participants, however, participants were not able to email the research team individually. Key components of the SHED-IT program were to provide individualized feedback, other participants with whom men could identify, and to be convenient; these are factors that have been shown to appeal specifically to men (6). The program booklet and individualized feedback included anecdotes and weight loss strategies that men could relate to such as examples of physical activities that men commonly participate in. Outcome measures The primary outcome measure was change in body weight (kg and percent change from baseline). Weight was measured without shoes on a digital scale to 0.1 kg (model CH-150 kp; A&D Mercury, Adelaide, Australia). Secondary outcomes included:
BMI. Height was measured to 0.1 cm using the stretch stature method and a wall mounted stadiometer (model KaWe 44440; Medizin Technik, Mentone Education Centre, Morrabin, Australia). BMI was calculated using the standard equation (weight (kg)/height (m)2). Waist circumference. Waist circumference was measured level with the umbilicus and each measurement was recorded with a nonextensible steel tape (KDSF10-02; KDS, Osaka, Japan). Blood pressure. Systolic and diastolic blood pressure and resting heart rate were measured using a NISSEI/DS-105E digital electronic blood pressure monitor (Nihon Seimitsu Sokki, Gunma, Japan) under standardized procedures. Sociodemographic measures included age and socioeconomic status index (20). Process evaluation of the Internet group A mixed-mode methodology was employed in line with recommendations for the process evaluation of RCTs (13). We used quantitative (website use/questionnaire) and qualitative (interviews) data, hypothesizing that the data source triangulation achieved would strengthen any findings. The process evaluation reported here is focused on the Internet group, given the paucity of research in this area and the importance of informing the refinement of future programs in this area. Adherence to self-monitoring (total number of daily diet entries, daily exercise entries and weekly weigh-ins) was calculated from website data. We also employed a 26-item process questionnaire to examine the men’s perceptions of the SHED-IT program, categorized as follows: use of website features, satisfaction with website, adherence to self-monitoring, perceived social support and levels of overall satisfaction. The questionnaire also included open-ended questions that required men to describe the strengths and weaknesses of the program along with their suggestions for improvement, and to indicate whether they would be willing to participate in a semistructured interview. 143
articles Intervention and Prevention For the qualitative component, a framework for a semistructured i nterview was developed. The major categories for questions related to satisfaction with specific intervention components, perceptions of social support and suggestions for improvement. The framework was used to guide the interview topics but specific prompts in the interview were tailored to participant responses to the written questionnaire and individual weight loss success. Interviews, which lasted for ~30–35 min, were conducted by either the researchers or a trained research assistant by telephone or face-to-face and digitally recorded. Verbatim transcripts of all interviews were generated. Analyses Analyses of the quantitative data were performed using Statistical Package for the Social Sciences (SPSS version 16.0 software; SPSS, Chicago, IL). Differences between groups at randomization, characteristics of completers vs. dropouts and differences in process questionnaire results between compliers and noncompliers were tested using independent t-tests for continuous variables and χ2-tests for categorical variables. Analysis of variance was used to determine differences between change in weight for staff (academic and nonacademic) and students. Bivariate correlations were examined for percent weight loss at 3- and 12-months. The significance level was set at P < 0.05. Two analyses were performed on the data using linear mixed models which were fitted with an unstructured covariance structure for all primary and secondary outcomes. Differences of means and 95% confidence intervals (CIs) were determined using the mixed models.
1. ITT analysis included all randomized participants. Linear mixed models were used to assess all outcomes for the impact of group (Internet and control), time (treated as categorical with levels baseline, 3 months, and 12 months) and the group-by-time interaction; these three terms formed the base model (21). Age and socioeconomic status were examined as covariates to see whether they contributed significantly to the models. If a covariate was significant, two-way interactions with time and treatment were also examined and all significant terms were added to the final model to adjust the results for these effects. 2. Per-protocol analysis: Internet participants who complied well (compliers) with the assigned treatment (defined as submission of requested daily eating and exercise diaries (n ≥ 50) over the 3-month period and weekly check-ins (n ≥ 12)) were compared with noncompliers in the Internet group and the control group. For the qualitative component, a thematic analysis was undertaken applying the constant comparison method (22). In the early stage of code development, a researcher and qualified research assistant independently coded the transcripts from one interview and reached agreement. Data were initially organized according to categories in the interview schedule, and inductively derived codes were formulated. When agreement was reached on the primary codes, a more detailed hierarchical coding scheme was developed on the basis of this initial analysis. This draft was revised after the coding of a second transcript, and a final coding scheme was developed. Coding of the remainder of the data was conducted. During the coding, detailed code descriptors were developed and continually revised after discussion with one of the research team (P.J.M.). Themes were independently generated after reading the coded data, then discussed and agreed upon. Recording and continually reflecting on this process ensured transparency in the process.
eligible for the study but seven men were not randomized as no consent was received. In total, 65 overweight or obese adult men attended baseline assessments and were randomized. The target sample was recruited in 0.05). Baseline data
Table 1 presents baseline characteristics of the sample. The mean (s.d.) age was 35.9 (11.1) years and comprised 43% students, 41.5% nonacademic staff, and 15.4% academic staff. Change in body weight
The 6-month follow-up results have been reported previously (15). Figure 2 shows the mean change in absolute body weight by treatment group from the ITT analysis at 3- and 12-months. Adjusting for covariates made little difference to our results, but the adjusted results for those variables where the covariate was significant are reported. Both groups lost weight during the interventions and maintained significant weight loss at 12-months compared to baseline (P < 0.001) (Table 2). Weight decreased significantly in the Internet group from baseline to 3 months (P < 0.001) and baseline to 12 months (P < 0.001) and also decreased significantly in the information only control group from baseline to 3 months (P < 0.001) and baseline to 12 months (P < 0.001). There were no differences between groups for weight loss (P = 0.408). Analysis of variance revealed there was no significant difference in change in weight between students, academic staff, and nonacademic staff at 3 or 12 months (P > 0.05). Percent weight loss
There was no significant difference in percent weight loss between groups (Internet group 6.1%, control group 3.4%, P > 0.05). At 12 months, 57.7% and 30.0% of Internet and control group participants, respectively had lost >5% of their initial weight but this difference was not statistically significant (χ2 = 3.49, df = 1, P = 0.062). There was a significant correlation between percent weight loss at 3- and 12-months (r = 0.81, P < 0.001). Per-protocol analysis
Results Participant flow
Figure 1 illustrates the flow of participants through the trial. A total of 136 men responded to the SHED-IT recruitment materials with most participants responding to advertisements placed on University notice boards. Seventy-two men were 144
Of the 34 participants assigned to the Internet group, 14 (41.2%) complied well with treatment, defined as 7 weeks of submission of daily eating and exercise diaries (i.e., ≥50 days of entries) and weekly check-ins (n ≥ 12) over the 3-month weight loss period. Compliers were more likely to be nonacademic staff members than academic staff members or VOLUME 19 NUMBER 1 | january 2011 | www.obesityjournal.org
articles Intervention and Prevention Assessed for eligibility (n = 136) Excluded (n = 64) Did not meet inclusion criteria (n = 61) Other reasons (n = 3) Eligible (n = 72)
Completed baseline assessment (n = 65)
7 Not consented 1 Workload 1 Medical condition 4 No contact 1 Personal reason
Randomized (n = 65)
Internet (n = 34) Received allocated intervention (n = 34)
Lost to 3-month follow-up (n = 6) 3 No contact 2 Personal reasons 1 Unavailable
Allocation
3-Month follow-up
Lost to 12-month follow-up (n = 8) 5 No contact 2 Personal reasons 1 Unavailable
12-Month follow-up
Analyzed for primary outcome (n = 34)
Analysis
Information and self-help (n = 31) Received allocated intervention (n = 31)
Lost to 3-month follow-up (n = 4) 4 Unavailable
Lost to 12-month follow-up (n = 11) 7 No contact 1 Moved interstate 3 Unavailable
Analyzed for primary outcome (n = 31)
Figure 1 Participant flow through the trial and analyzed for the primary outcome (change in weight (kg)).
students (χ2 (3) = 14.41, P = 0.002). There was also a significant difference between compliers and noncompliers for age with older participants more likely to comply with the online program than younger participants (t(32) = −2.8, P = 0.008). Using linear mixed models, a significant group-by-time interaction at 12 months was found with compliers reducing their weight, waist circumference, BMI, and systolic blood pressure more than noncompliers and the control group (Table 3). Change in secondary outcomes
Values for all secondary outcomes improved significantly from baseline to 12 months in both groups; the only significant obesity | VOLUME 19 NUMBER 1 | january 2011
difference between groups was for systolic blood pressure (Table 2). At 12 months, participants reduced their waist circumference, BMI, systolic and diastolic blood pressure, and resting heart rate (all P < 0.001). Website use and relationship to weight loss
Significant correlations were found between weight and waist circumference change at 12-months and the number of days of diet entries (weight: r = 0.69, P < 0.001; waist: r = 0.71, P < 0.001), number of daily exercise entries (weight: r = 0.54, P = 0.004; waist: r = 0.69, P < 0.001), and number of weekly weight entries (weight: r = 0.56, P = 0.004; waist: r = 0.57, P = 0.003). 145
articles Intervention and Prevention Table 1 Baseline characteristics of men randomized to the control and Internet groups Control (n = 31) Characteristics Age (years)
Internet (n = 34)
Total (N = 65)
Mean
(s.d.)
Mean
(s.d.)
Mean
(s.d.)
34.0
11.6
37.5
10.4
35.9
11.1
Occupation, n (%) Student
14
45.1
14
41.2
28
43.0
Nonacademic staff
13
41.9
14
41.2
27
41.5
4
12.9
6
17.6
10
15.4
0
0.0
1
4.2
1
1.9
Academic staff SES , n (%) a
1–2 (lowest) 3–4
5
17.9
7
29.2
12
23.1
5–6
9
32.1
3
12.5
12
23.1
7–8
11
39.3
11
45.8
22
42.3
3
10.7
2
8.3
5
9.6
99.2
13.7
99.1
12.2
99.1
12.8
1.8
0.1
1.8
0.1
1.8
0.1
30.5
3.0
30.6
2.7
30.6
2.8
9–10 (highest) Weight (kg) Height (m) BMI (kg/m2) BMI category Overweight, n (%) Obese, n (%)
15
48.4
16
47.1
31
47.7
16
51.6
18
52.9
34
52.3
Waist circumference (cm)
103.4
8.3
102.8
6.8
103.1
7.5
Systolic blood pressure (mm Hg)
135
14
134
14
134
14
Diastolic blood pressure (mm Hg)
85
7
84
10
84
9
Resting heart rate (BPM)
79
12
74
12
76
12
BPM, beats per minute; SES, socioeconomic status. a Postcodes were assigned an Index of Relative Socioeconomic Advantage and Disadvantage tertile, as an indicator of SES. There are 10 quantiles of equal distribution, whereby 1 equates to the most disadvantage and 10 the most advantage.
Process evaluation
Results for the Internet group process evaluation questionnaire overall and for compliers and noncompliers are presented in Table 4. Overall, men were highly satisfied with the SHED-IT program. Most believed the website was useful and they were provided with enough support to achieve their weight loss goals. After 3 months, men were continuing to monitor their weight but were no longer monitoring their dietary intake and exercise. All men were comfortable with their computer ability with no men perceiving themselves to be either computer novices or experts. Statistical analysis revealed that compliers scored significantly higher on items relating to: understanding of website, enjoyment accessing the website, perceived quality and quantity of support and had a preference for the Internet over face-to-face as a treatment approach. In addition, compliers were more likely to be active with their families than noncompliers. We report the findings from the interviews with men who were randomized to the Internet group. Due to budget constraints, we used a purposive sampling strategy to select 12 of the 24 men who had agreed to be interviewed based on their weight loss. This provided a range of outcomes from men who lost varying amounts of weight (n = 7, weight loss >5%; n = 5, weight loss
