Satomura 2005.
| Methods | Randomised controlled trial, randomisation was achieved by simple computer‐generated random digit. Allocation was concealed using sealed opaque envelopes. Not clear if there was a central randomisation centre. Post hoc exchange of envelopes was prevented by writing both the name of each subject and the number on the envelope he/she drew before breaking the seal. Participants were not blinded to the intervention, however, disease incidence was determined by 1 study physician who was not informed of the results of assignment. Analysis was done based on the intention‐to‐treat principle. The study targeted community healthcare all over Japan and was conducted between December 2002 and March 2003 for a follow‐up period of 60 days | |
| Participants | Three hundred and eighty‐seven participants at 18 sites were recruited. Included in the analysis 384, follow up was completed on 338 participants. Attrition was fully explained for URTI analysis, however, 2 subjects were not accounted for in the ILI analysis. Forty‐six participants did not complete the follow up due to either discontinuation of diary use (n = 9) or contracting influenza‐like illness (ILI) (n = 37) Of the 37 participants with ILI, 11 were in povidone‐iodine group, 12 in water group and 14 in control. Analysis was performed on 35 participants (Kitamura 2007) | |
| Interventions | Participants were randomised to 1 of the following: water gargling, n = 122 (20 mL of water for about 15 seconds 3 times consecutively, at least 3 times a day); povidone‐iodine gargling, n = 133 (20 mL of 15 to 30 times diluted 7% povidone‐iodine (as indicated by the manufacturer) in the same way as water gargling); and control, n = 132 (retain their previous gargling habits) All groups were asked to fill a daily gargling diary (standardised form to record: gargling habits, handwashing and influenza complaints) The frequency of gargling in the water group was higher (3.6), frequency of handwashing was similar between the 3 groups URTI symptom was classified according to Jackson methods. Diary recording was continued throughout the follow‐up period and for 1 week after the onset of URTI. ILI were reported separately | |
| Outcomes | Laboratory: none Effectiveness: primary outcome: incidence of first URTI. Index cases were defined as all of the following conditions: (1) both nasal and pharyngeal symptoms, (2) severity of at least 1 symptom increased by 2 grades or more, and (3) worsening of a symptom of 1 increment or more for > 3 days Secondary outcome: severity of URTI of the incident cases was assessed by grading each symptom during the initial 7 days after the onset of URTI in numeric scores: none = 0, mild = 1, moderate = 2 and severe = 3 ILI was defined as both developing a fever of 38 °C or higher, and worsening arthralgia in addition to some respiratory symptoms (Kitamura 2007) Safety: no harm was reported. However, 2 patients in the povidone group switched to water gargling (analysed in their assignment group) | |
| Notes | The authors conclude that simple water gargling is effective to prevent URTIs among healthy people. However, no significant difference was observed against ILIs Study was well‐conducted, blinding would have added to the validity of the results. In addition, the study was not powered enough to detect significant preventative effect against ILI The study demonstrated that in addition to handwashing, simple gargling even with simple water can reduce URTI but not ILI. However, during periods of endemic influenza, multiple inexpensive and simple modalities (handwashing, masks, gargling) can be utilised together to reduce infection and transmission. Overall, the reporting of the 2 combined studies together is highly confusing. In the first study (Satomura 2005) the main outcome is URTI defined as fever and arthralgia. The second study (which is a presentation of further data from the 2005 publication in the guise of a short report) introduces the outcome ILI with a definition similar to that of URTI in the first study but referring to the earlier outcome as common cold. Also of note is reporting of significance without confidence intervals. Overall this potentially important study should be repeated with a larger denominator Medium risk of bias because of confused reporting and absence of double‐blinding | |
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | "Group assignment was based on simple computer‐generated random digits..." |
| Allocation concealment (selection bias) | Low risk | "By an individual drawing of sealed opaque envelopes, subjects were randomly assigned to the following three groups" "allocation was completely concealed from study administrators" |
| Blinding (performance bias and detection bias) All outcomes | High risk | Not blinded |
| Incomplete outcome data (attrition bias) All outcomes | Low risk | 338 of 385 randomised followed up; reasons reported |
| Selective reporting (reporting bias) | Unclear risk | Confusing reporting |