Silva 2018.

Methods	Study design: controlled clinical trial Location, number of centres: northern Portugal Country: Portugal Study period: November to July Methods of analysis: "For respiratory infection assessment, a symptom diary was assembled using previously developed questionnaires and indexes (Barrett et al., 2009; Jackson, Dowling, Spiesman, & Boand, 1958; Powell et al., 2008). The questionnaire was subdivided into five main symptom domains: general (shivering, muscle pain, tiredness, chest pain, dyspnoea), nasal (runny nose, obstructed nose, sneezing, throat (sore throat, itching throat, cough and hoarseness)) and ear (ear pain). Body temperature was registered if the absolute value was above 37.5°C. Allergic symptoms were also included (nose and eye itching). To assess the suitability of the questionnaire, a pilot application was administered to a population of older adults in an exercise programme. Symptoms were easily understood; however, the grading scale was simplified from a Likert scale of 0–7 to a scale grading the symptoms as mild, moderate or severe due to comprehensibility issues. All participants were given a symptom diary with a written and verbal individual explanation of how to document symptoms. Participants were contacted monthly to assure compliance and asked whether they had been sick and whether medical care had been administered. In the EG, 17 participants completed the exercise programme and RIs diary. Three individuals dropped out due to the inability to continue training, 1 at the beginning and 2 after 3 months. In the CG, 9 participants were included, 3 were not compliant with the respiratory symptoms diary and 6 were inappropriate in their answers and could not be used." Statistical analysis: "Analysis was conducted per‐protocol. The results are presented as the mean values [95% confidence interval (CI)], medians (interquartile range) or counts (n, %). The Mann–Whitney U‐test was used for comparison of continuous data and Fisher’s exact test for categorical data. Univariate and multiple logistic regression models were developed using independent variables as risk factors for having an RI; results are presented as odds ratio (OR) with 95% CI. Variables used in the model included gender, age, body mass index, history of respiratory tract disease, diabetes, smoking habits, influenza immunization and history of performing exercise training. The model was performed by data driven and adjusted by considering its goodness‐of‐fit (Hosmer–Lemeshow test) and predictive power (ROC curve analysis). Analyses were performed using SPSS 23.0 software (Chicago, IL)."
Participants	Recruitment means: public advertisement from senior universities and nursing homes in northern Portugal Target participants: physically independent, community‐dwelling individuals over 60 years of age N screened: 80 participants (38 included) N completed: 26 completed: 9 (control); 17 (exercise) Gender F = 19; M = 7 Age: 70 (65 to 79) IQR Inclusion criteria: individuals were included if they had a medical report describing that they were physically able for exercise and had not been involved in regular exercise training in the previous 6 months. Exclusion criteria: volunteers were excluded if they had: acute or terminal illnesses; severe or uncontrolled cardiovascular or respiratory disorders; neurological or musculoskeletal disorders that compromised their compliance; or had undergone treatments that could affect exercise training or increase susceptibility to RI. Baseline details: body composition, other diseases (hypertension, diabetes, etc.), practice of exercise, smoking. Comorbidities: many comorbidities were described here from respiratory disease to osteoarticular disease, with only osteoporosis significantly different between groups.
Interventions	Setting of intervention: not reported Description of intervention: the exercise programme was performed 3 times per week (non‐consecutive days) for 36 weeks in sports facility of University of Porto. Each session lasted 90 min (10 min of warm‐up, 30 min of aerobic training, and 45 min of resistance training, ending with a 5‐minute cool‐down). Aerobic training consisted of 30 min of walking at moderate intensity reaching 60% to 80% of the maximum heart rate. Heart rate monitors (Polar Team System, Finland) were used during the exercise sessions. Resistance training was performed on variable‐resistance machines (Nautilus Sports/Medical Industries, Independence, USA) aimed at 9 different muscular groups (leg press, chest press, leg extension, seated row, seated leg curl, abdominal flexion, low‐back extension, biceps curl, and triceps extension), and participants completed 2 sets of 10 to 12 repetitions at 60% to 80% of repetition maximum (1RM calculated based on Epley formula). Every month, the 1RM was measured to keep the training stimulus consistently at 80% of the 1RM. Training was standardised but adapted to each participant according to their limitations. Delivered by: physical education instructor specialised in training older adults Follow‐up period: 36 weeks Co‐interventions: none
Outcomes	Acute respiratory infection, duration and severity of ARI, serum metabolomic profiling Follow‐up period: 36 weeks
Notes	Study funding: this work was supported by the European Regional Development Fund (FEDER) through the Competitive Factors Thematic Operational Programme (COMPETE) and Foundation for Science and Technology (FCT), Portugal, under the projects PEst‐C/QUI/UI0062/2013 (Research Unit 62/94 QOPNA) and PTDC/QUI‐QUI/117803/2010 (Future asthma management helped by non‐invasive sampling: contributes for the definition of a rapid and non‐invasive diagnostic tool). Project NORTE‐01‐0145‐FEDER‐000010 – Health, Comfort and Energy in the Built Environment (HEBE), co‐financed by Programa Operacional Regional do Norte (NORTE2020) through FEDER.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	"Participants were assigned to the exercise group (EG) and the control group (CG) according to the availability to attend the sessions"
Allocation concealment (selection bias)	High risk	Researchers did not conceal the allocation of each participant. They chose according to the availability to attend the sessions
Blinding of participants and personnel (performance bias) All outcomes	High risk	The participants could not be blinded due to the characteristics of the intervention.
Blinding of outcome assessment (detection bias) All outcomes	Unclear risk	The study authors do not report who assessed the outcomes.
Incomplete outcome data (attrition bias) All outcomes	High risk	High losses to follow‐up and no proper statistical adjustments. A total of 20 participants were enrolled in the exercise group, with data provided for only 17. Of the 18 original control participants, complete data were provided for only 9.
Selective reporting (reporting bias)	Low risk	All the outcomes proposed are reported in the study and additional tables.
Other bias	Low risk	The study seems to be free of other sources of bias.