This randomized, parallel–arm, double–blind, prospective study adheres to CONSORT guidelines. All methods were carried out in accordance with CONSORT 2010 guidelines. The subjects were selected from two-stage randomization process (stage 1 being selection of schools randomly from Kandy district (an administrative unit) in Sri Lanka and stage 2 being selection of subjects randomly within the selected schools). A formal informed written consent was taken from all participants who were 18 years of age or above and from legally acceptable representative/guardian/parents of the participants who were below 18 years. The inclusion criteria were similar as the previous study 4 and included patients in the age group of 16–19 years, diagnosed with migraine according to the International Headache Society (IHS) criteria 5. Patients with more than 3 migraine attacks but not more than 15 attacks per month, who have not taken an acute treatment during the testing period, were selected. Exclusion criteria were; history of intracranial lesion or tumor, recent nasal or sinus infection, acute or chronic sinusitis, evidence of another infection (i.e., acute otitis media or pneumonia), history of allergic rhinitis, asthma or an underlying immune deficiency, cystic fibrosis, immotile cilia syndrome, recent head and facial trauma, runny nose, severe vomiting during the migraine attack, smoking, alcohol or drug abuse, participants who were on hormonal therapy for any condition or illness, patients with psychiatric illness, patients on non–medical/non–nutritional treatment for migraine prevention such as acupuncture or psychotherapy, patients on fasting and had exercise or used any nasal drops or steam inhalation one hour before the procedure. This flow chart was described in the figure − 1. Patients who did not consent were also excluded. Since the key response variable was severity of headache, sample size was determined based on this variable. In order to detect headache pain drop difference of 30% between the two groups with the power of the test of 0.8 and type–I error rate of 0.05, required sample size was computed to be 38 for each group 6. To minimize the bias due to dropouts, 43 patients were recruited for each group. The study was carried out from 07 March 2018 to 13 June 2018.
The selected children were examined at their own school premises. When the selected participants presented with typical migraine headache for more than one hour, they were randomized into treatment (paranasal air suction) or control (placebo suction) group. All subjects were studied only once. The air suction devise tested in this study was a low – pressure portable air sucker (Fig. 2 with 12–15 mm Hg suction capacity. The device is battery operated (four 1.5 V size AA). It is easy to carry and can be used by patient himself/herself.
The suction process was carried out similar to our previous research methadology4. During this process, nasal and paranasal sinus air were sucked six consecutive times from each nostril. Each suction was for 10–second duration with a 10 second suction free period between two suctions. Therefore, each subject was subjected to 120–second suction altogether. Participants were instructed to hold the breath by closing both nostrils by their own hand and then open one nostril for the suction for 10 seconds. They were allowed for mouth breathing if they need during the suction period. After 10 seconds of suction free period, they closed the opened nostril and opened the other nostril for air suction for 10 seconds. During suction, the suction tube did not contact the nostrils but we kept it in the air space of the outer part of nostrils, very close to nasal orifice. The control group was tested by keeping the same type of a nasal suction tube close to the nostrils. The appearance of the nasal air suction tube end was similar in both groups. They were also given the same instructions and were asked to close and open the nostrils in a similar manner. However, they were not exposed to air suction procedure but they were made to hear the sound of the air sucker. All these measures were taken to provide similar perception to both groups to reduce psychological bias.
The key symptom or the response variable studied was severity of headache (pain). In addition, left and right air flow rate, left and right scalp and supraorbital tenderness, photophobia, phonophobia, numbness over the face and scalp, nausea and generalized tiredness/weakness of the body were also studied as response variables. The primary endpoint was the headache pain relief. All response variables, except the numbness, were measured in ordinal scale and the examiners were also blinded. Air flow rates were measured on a scale of 0 to 3 based on the way the assessor felt the flow rate (0 = no flow, 1 = low flow, 2 = moderate flow and 3 = normal).The severity of the headache was measured using a standard pain rating scale (0 being pain free and 10 being very severe pain) before and after the air suction procedure. Supraorbital and scalp tenderness was assessed by the same examiner applying pressure over the area until some blanching of their fingernail was discernible7. This was assessed on both right and left. The supraorbital notch (lying between the nasion and the trochlea) where the supraorbital branch of the ophthalmic nerve and the supratrochlear branch of the ophthalmic nerve were pressed to elicit supraorbital pain7.The severity of tenderness felt to the subjects was measured using the above pain rating scale Photophobia, phonophobia, nausea and generalized tiredness/weakness were measured on a scale of 0 to 3 (0 = no symptoms, 1 = mild symptoms, 2 = moderate symptoms and 3 = severe symptoms). In order to measure photophobia, we exposed all participants to see the same bright light (using a same screen and illuminated for 30 seconds) that could be tolerated by healthy people (20 healthy subjects were exposed to different light intensities using the same screen and the lowest intensity they could tolerate for 30 seconds was taken). Phonophobia was measured by exposing all participants to listen to a high pitch noise using a headphone for 30 seconds that could be tolerated by healthy people (20 healthy subjects were exposed to different high pitch noise amplitudes using the headphone and the lowest amplitude they could tolerate for 30 seconds was taken). Numbness was measured based on presence or absence of numbness (subjective view). All endpoints, except for air flow rates, were assessed before and 2 minutes after application of suction using the device and was monitored at fixed intervals of 2, 6 and 24 hours after air suction. The explanatory variables (factors) considered in the study were, gender, type of migraine (with aura or without aura), side of the headache (left, right, bilateral), and group (treated /sucked or not).
The side effects of air suction were also inquired at fixed intervals of 2, 6 and 24 hours after air suction. For this, each participant was given a questionnaire containing a list of side effects and check boxes and was asked to check if a side effect was present. The side effects included sneezing, intolerance to suction and sound of the device, bleeding, nasal irritation, palpitation /increase in heartbeat, fear, and space was left for any other side effect. Blood pressures and heart rate were also measured before and after the procedure.
Descriptive statistical analysis outcome was used to describe the characteristics of the study population. Statistical analysis was carried out by fitting models. Specifically, since the response variables were in ordinal scale and several effects were studied simultaneously, analysis was carried out by fitting cumulative logit models 8 9. The effect of each factor after adjusting for all other factors was examined using the likelihood ratio type3 (LR Type3). The differences between levels of factors were studied using maximum likelihood estimates of the fitted models. In the study, all response variables were measured before and after the intervention on each patient. Thus, the difference between score before the intervention and 2 minutes after the intervention was used for the analysis. Since the pain and tenderness was measured using a scale of 0 to 10, the difference between score before and after the intervention were grouped into ordinal categories (≤ 2.5, > 2.5 to ≤ 5.0, > 5.0 to ≤ 7.5 and >7.5 to ≤ 10). Since the response variables photophobia, phonophobia, nausea and generalized tiredness/ weakness were measured on a scale of 0 to 3, each value of the difference between scores before and 2 minutes after the intervention were taken as a separate ordinal category. The variable numbness was measured as present or absent (two categories) and thus the variable was modeled by fitting ordinary logit models. When there was no interaction between factors and effects of only one factor was found, the difference between levels of the factor was further illustrated using non–parametric methods. In such a situation, especially when the factor had only two levels, the specific non–parametric method, Wilcoxon Rank Sum Test 10 was used. First, the score difference (before– after) for each patient was calculated and then using those score differences, two levels were compared by performing Wilcoxon Rank Sum Test.