Short term effects of a weight loss and healthy lifestyle programme for overweight and obese men delivered by German football clubs

21 Numbers of obese and overweight people continue to grow in Germany as they 22 do worldwide. Men are affected more often but do less about it and few weight 23 loss services attract men in particular. To evaluate the effectiveness of a men- 24 only weight loss program, Football Fans in Training (FFIT), delivered by football 25 clubs in the German Bundesliga, we did a non-randomized trial with a waiting 26 list control group. Participants’ data were collected between January 2017 and 27 July 2018. FFIT is a 12-week, group-based, weight loss program and was 28 delivered in stadia and facilities of 15 professional German Bundesliga clubs. 29 Inclusion criteria were age 35-65 years, BMI ≥ 28 and waist circumference ≥100 30 cm.. Clubs recruited participants through Social Media, E-Mail and match day 31 advertisement. 477 German male football fans were allocated to the 32 intervention group by order of registration date at their respective clubs. 84 33 participants on waiting list were allocated to the control group. Primary outcome 34 was mean difference in weight loss with treatment condition over time as 35 independent variable. We performed a multilevel mixed-effects linear regression 36 analysis. Results were based on Intention-to-treat (ITT) analysis with Multiple 37 Imputation.. After 12 weeks, the mean weight loss of the intervention group 38 adjusted for club, course and participants' age was 6.24 kg (95 % CI 5.82 to 39 6.66) against 0.50 kg (-0.47 to 1.49) in the comparison group (p<0.001). The

Previous research showed the feasibility of recruiting clubs to deliver the 104 program and fans to attend the program. 24

Study Design and participants 141
We conducted a pragmatic non-randomized trial with a waiting list comparison activity. Therefore, men who answered 'Yes' to any PARQ question or who had 155 resting systolic blood pressure of 160 and higher or diastolic blood pressure of 156 100 and higher had to provide a letter of support from their physician or were 157 excluded from participating in physical activity during club sessions (although 158 they were still able to take part in the 'classroom' part of the session and the pedometer-based walking program). Most clubs opened recruitment to all male 160 supporters, but three restricted participation in FFIT to season ticket holders. To strengthen the validity of the study, a comparison group (N=84) was 170 recruited from waiting lists. 28 The lists included all men who had applied to a 171 course at their club but had not been selected by the methods described above. 172 These men, if any, were then invited to take part in objective comparison group 173 measurements in the time leading up to the following course which they were 174 considered to join. Attendance to those measurements was voluntary, without 175 further incentives and the same through all clubs. They were measured twice, 176 following the same protocols as the intervention group measurements, with 177 follow up occurring 13 weeks after baseline data collection. A flow chart of

Outcome Measures 180
All measurements and questionnaire administration were conducted by the 181 FFIT coaches who had been trained to a standard measurement protocol. In 182 addition, to quality assure data collection, all measurement sessions were 183 supervised by members of the research teams. Men who were not able to take 184 part in the official measurement session were asked to attend at a subsequent 185 time that was convenient to them. The primary outcomes were objectively-186 measured weight and waist circumference. Secondary outcomes were BMI, 187 body fat percentage, and systolic and diastolic blood pressure. Weight and body 188 fat percentage were recorded with an electronic scale (Omron BCM BF 511) 189 with men wearing light clothes and having removed their shoes and anything in 190 their pockets. Waist circumference was measured with an ordinary tape 191 measure about 5cm above the navel. Blood pressure assessments were 192 conducted in a separate room for a more relaxed atmosphere and nobody to 193 talk to. Men were asked to sit down and relax for at least one minute before 194 measurement. Height was measured without shoes. All self-reported data were 195 obtained using a short questionnaire that participants filled out in between the 196 objective measurements. 197 To assess sedentary time, men were asked to estimate the average number of 198 hours per day they had spent sitting during the last 7 days. A modified, German version of the DINE questionnaire 29 was used to assess fruit and vegetable 200 intake, fatty food intake, sugary food intake and the proportion of whole grain 201 intake among pasta, rice and bread over the last week. In the Fatty Food Score, 202 Sugary Food Score, Vegetable and Fruit Score as well as Whole Grain Score, a 203 higher score indicated a higher number of days during the last week on which 204 the respective food types were consumed. Additionally, the Warwick-Edinburgh 205 Mental Well-being Scale 30 was used to measure participants' psychological 206 well-being. 207

Statistical Analysis 208
All statistical analyses were conducted with Stata 15 (Stata Corp, College 209 Station, TX). To follow the Intention-to-treat principle, Multiple Imputation was 210 used to decrease bias due to missing data following the assumption that data 211 were missing at random (MAR). 31 Missing data were imputed using the MICE 212 technique (multivariate imputations by chained equations) with M = 10 213 imputations. 32 The pooling of the regression estimates followed Rubin's rule. 33 214 Baseline characteristics were analyzed with linear regression to check for 215 baseline differences between intervention and comparison group. Table 2  216 reports mean values and standard deviations, as well as mean differences 217 between groups. Multilevel mixed-effects linear regression analysis was applied 218 to evaluate effects of the intervention on primary and secondary outcomes.
Time of assessment (baseline vs follow-up), group (intervention vs comparison) 220 and the interaction term between time and group were included as fixed effects. 221 Additionally, participants' age was included as a fixed effect because of a 222 significant baseline difference between groups (Table 1) Table 1. Between-groups 239 analysis showed no significant differences except for the men's age. Therefore 240 age was added to all analysis as a confounding variable. 241 242 Place holder for Table 1  The FFIT study in Germany was not a fully powered randomized controlled trial 339 to replicate the original FFIT study. 22 . Several considerations led to this 340 decision. Observational studies have found that without a specific intervention 341 the weight of German men who met the inclusion criteria for this study is very 342 unlikely to decrease and likely to increase slightly. 38, 39 It is therefore very 343 unlikely that decreases in weight could be attributed to "spontaneous 344 remission". The focus of our study was easy and practicable implementation of 345 an evidence-based, successful weight loss programme for clubs under routine 346 "field-conditions" and thus we prioritized high external validity. We made these 347 decisions based on the knowledge that clubs did not want to exclude their fans from a programme which existing evidence suggests the participants are very 349 likely to benefit from. Further, our main aim was to evaluate the transfer of FFIT 350 into the German Bundesliga and whether German fans would also experience 351 similarly positive outcomes. We found that the programme could be transferred 352 and German fans could benefit. 353 Although an effort was made to recruit participants to a comparison group we 354 were not wholly successful and there are many fewer participants in that group 355 Because of this large equality between group numbers we simulated an even 365 number as part of our sensitivity analysis described in the results. The numbers 366 indicated that the effects are strong enough to maintain in this much smaller 367 sample.
In spite of the lack of randomization, baseline data were very similar between 369 intervention and comparison group, with the exception of participants age, 370 which was significantly different between groups. Thus, age was included in the 371 regression models as a fixed effect, alongside club, course and time. We were 372 not able to follow up any fans that did not participate in the end of course 373 measurements. Thus, all results were analyzed following the Intention-to-treat 374 principle with Multiple Imputation to deal with drop-outs and missing data. There 375 were no drop-outs on course level. Although every FFIT coach was trained to 376 standard measurement protocols, facility circumstances during measurements 377 differed between clubs and sometimes courses. As blood pressure is strongly 378 affected by the environment or discomfort during the measurement procedure 379 this might have resulted in confounding effects for the BP outcomes. Such 380 systematic influences on club or course level have been considered in our 381 statistical model with the addition of club and course as a random effect.. To 382 assure high quality, all data collection sessions at clubs were monitored by the 383 scientific project staff. Outcomes like sedentary time and diet-related 384 information were self-reported and limited to the last week. This week could 385 have been influenced by confounding events like illness, injuries or holidays.

Conclusion 387
The study suggests that "Football Fans in Training" is a very promising program 388 to help fill a gap in Germany's health care landscape as far fewer men than 389 women are attracted to take part in existing preventive courses and offers of 390 health systems, including weight reduction programs. To date,there have been 391 very few programs that are specifically designed to try and attract men in 392 Germany. The FFIT has previously been shown to be very effective in Scotland 393 in attracting overweight, middle-aged men and supporting them in weight loss 394 and lifestyle changes, building on its concept of using the socio-cultural 395 environment of the professional football clubs as a 'draw'. We have shown that 396 the idea and concept was transferrable to professional football in Germany 397 before 24