This was an international multi-centre, single blind, two arm (1:1), superiority, randomised controlled trial (RCT) to test the effectiveness of usual light touch osteopathic treatment with advice and guidance (the Test group) compared with simple light touch without intent with advice and guidance (the Control group). The detailed protocol has been published elsewhere [23] as has the study plan, the simplified protocol for participating osteopaths, and the statistical analysis plan.[24] Input was sought from parents in the design and development of the methods and protocol.
Setting
The study was conducted in a ‘real world’ setting in private osteopathic clinics in the UK, Australia and Switzerland. Participant parents were offered free treatment for their infants, with a registered osteopath providing paediatric care in their usual practice.
Participants
Participants were infants under 10 weeks old who were excessively crying, distressed and difficult to console but healthy and thriving. ‘Excessively crying’ was defined using the Rome IV criteria for research: infants who cry and fuss and who are difficult to console for more than three hours per day, for three days or more, for one week or more.[2] Crying time tends to subside after 12 weeks,[3, 25] so only infants who were under 12 weeks at the 14-day follow-up were recruited.
Recruitment, consent and procedures
All parents and or guardians with infants under 10 weeks old attending participating clinics with symptoms of colic were invited to participate in the trial. If interested, they were sent a participant information leaflet, a copy of the consent form and a crying diary. All parents were given at least 24 hours or more to consider the information. Parents were asked to monitor their infants crying for 24 hours prior to the first consultation.
At the first consultation, if they agreed to be in the trial, valid informed consent was obtained. After this, the infant was examined. If the infant was systemically well and thriving, the parent completed the baseline questionnaire and the infant was then randomly allocated to a study group.
Infants had up to four consultations (as determined by the parent and osteopath and the infant’s needs) over a two-week period. Parents were asked to complete a 14-day crying diary, recording the minutes their baby cried every hour.
Fourteen days after randomisation, the parents completed a follow-up questionnaire.
Parents were informed, if they wanted to know, about their infant’s group allocation once their follow-up data were collected or, if follow-up data were not received, after 21 days.
Randomisation and allocation concealment
Infants were randomly allocated to trial arms at a 1:1 ratio, using a randomised block design, with block sizes of 4 and 6 via the electronic clinical trials platform (Castor EDC). Osteopaths accessed the trial software during the first consultation to randomise and allocate the infant: after consent, baseline data capture, case history and examination of the infant. Parents were blinded to their infant’s allocation. The osteopath delivered both the Test and Control interventions and therefore was not blinded. To test the effectiveness of the blinding, parents were asked in the follow-up questionnaire (day 14) whether they thought their infant was in the Test group, Control group or if they did not know. The trial statistician was blinded to the Test and Control group allocations.
Data collection
Baseline data about the parent and infant were collected via an email link to an online questionnaire sent to parents prior to their initial consultation. A follow-up questionnaire was sent automatically 14 days after randomisation. Parents photographed their infant’s crying diaries and these were sent directly to the study team. Diary data were entered electronically and independently by two members of the study team, who compared entries and corrected any discrepancies.
The osteopaths also recorded data about the parent and infant, treatment given, the advice and guidance given at each consultation and how the infant was responding according to the parent, including whether the infant had unwanted or unexpected reactions to treatment. At baseline parents were asked about the expected response to osteopathic care (5- point Likert scale ranging from ‘very well’ to ‘not well’).
Interventions
Both groups received best practice advice and guidance appropriate to the parent and infant based on NICE Guidance.[26]
The Test intervention was designed to be as close as possible to what osteopaths usually do when treating infants with colic. It consisted of usual light touch osteopathic treatment delivered with treatment intent for 10–20 minutes. High velocity joint manipulations were excluded as they are contraindicated in the treatment of infants.
Light touch osteopathic treatment is delivered using gentle movements of the hands on the infant’s body with the aim of reducing tissue tension and encouraging movement of fluids and fascia. This targeted light touch was administered to areas of the body as determined by the osteopath after palpation of the tissues. Treatment techniques included: articulation, tension release (to ligaments, articular strains, fontanelles/cranial sutures), counter-strain/facilitated positional release, indirect functional techniques, myofascial release, soft tissue massage and/or stretch and visceral movement.
The Control intervention was designed to mimic the Test intervention to ensure parent blinding and to ensure that all non-specific effects were present in both groups. The Control intervention consisted of simple light touch to the infant, administered with no treatment intent to the following pre-chosen areas: cranium, thorax, abdomen, sacrum/pelvis. Osteopaths did their normal observations, palpation and testing as they would for a treatment with intent. Osteopaths randomly chose the locations to which they were to place their hands for three to five minutes. To help osteopaths not deliver treatment to the infants in the Control group, they were asked to do a cognitive mental task. This consisted of counting backwards in 6s, 7s or 8s from 200 or to name animals or vegetables for each letter of the alphabet, in their head, while touching the infants in the pre-specified areas. These cognitive tasks have been shown to modify osteopaths’ brain touch processing [17] and are believed to prevent touch feedback to prevent therapeutic touch.[27]
The light touch protocols were administered for between 10 and 20 minutes in both groups during one to four visits lasting 30–45 minutes.
Both interventions were delivered by registered osteopaths, specifically trained for the trial. The training covered the trial procedures, delivering the interventions, best practice advice and guidance as per NICE [26] and good clinical practice in research.
Fidelity
Fidelity of the intervention delivery was planned to be assessed via observation, the patient records and feedback/reflective interviews with the participating osteopaths. However due to the COVID19 pandemic social distancing restrictions, between 2020 and 2022, we were unable to carry out the observational checking.
Primary outcome
The primary outcome was total daily crying time in minutes, recorded by parents as the number of minutes of crying time (to the nearest 5 minutes) for each hour, over each 24-hour period. This was recorded for 14 consecutive days. The primary outcome was the average between group difference in daily crying time over 12 days excluding the day prior to enrolment and the day participants received the first treatment.
The crying diary has been validated and is an accepted standard method to record crying time.[25]
Secondary outcomes
Secondary outcomes at follow-up were parenting confidence, global change, experience and satisfaction with care.
Changes in parenting confidence were measured using the Karitane Parent Confidence Score at baseline and at 14-day follow up. This score relies on 15 questions about parenting confidence with a Likert scale of four choices (total score: 0 to 45 points). Movement of 6 points or more was considered as meaningful.[28]
Global change in the infant’s symptoms were measured using a 7-point Likert scale from completely recovered to vastly worse. Experience was measured using a 5-point Likert scale from ‘very good’ to ‘very poor’. Satisfaction was measured using a 5-point Likert scale from ‘very satisfied’ to ‘very dissatisfied’.
Additional care, unexpected and/or adverse events
During the trial period parents were asked about additional care they gave to their infants, additional health care consultations during the trial period and any changes in symptoms that caused concern.
Sample size
To detect a 30-minutes between group difference (90 min vs. 120 min, SD 45 min), with 80% power and a two-sided 5% significance level, 72 participants were needed. Allowing for a 15% drop-out, 84 infants were to be recruited. The initial estimation from the protocol had set the target to 112 infants with a power of 90%. In January 2022, following recruitment difficulties due to COVID-19, the data monitoring committee approved the trial steering committee’s request to lower the study power to 80% and suggested improving the statistical power by using serial data and including predictive co-factors in the primary analysis. This new estimated sample size was still conservative as it did not account for increased precision due to multiple measures and adjustment for co-factors. The study was therefore capable of excluding a minimal 30-minute between group difference even for a slightly smaller sample size.
Statistical analysis
All analyses were intent-to-treat (ITT) except for a secondary per protocol analysis on the primary outcome. This meant that participants’ data were allocated to the group they were randomised to independently of the true intervention they received or the presence of a protocol deviation. All hypothesis tests were two-sided, and the significance level was set at 5%. Missing data were not replaced except for those diaries that reported presence of crying at specific hours, without providing the duration in minutes. For these missing values, we imputed median reported crying times specific to each hour slot.
The primary outcome of crying time (i.e. individual daily measures of crying time in minutes) was analysed using linear mixed-effects modelling. The response variable was daily crying time, explanatory variables were: group allocation, baseline crying time, expectations of treatment, age of infant at entry, and day of measurement (2–13 days after randomisation). The additional variables were used to account for eventual between-group imbalances and changes of crying duration over time that were independent of group allocation. Random effects were the infants, modelled as random intercepts. This latter effect was included to account for lack of independence induced by repeated measures. The predictors were determined a-priori. Residuals were analysed to check model assumptions. The same approach was used for secondary outcomes.
Secondary analyses included a per-protocol analysis, worse and best-case scenario to account for missing data and sensitivity analysis. This included comparing: excluding vs imputing crying times, adjusting vs not adjusting for baseline crying time, age of infant at entry, and expectations of treatment, recruitment sites or performance bias (number of treatment sessions, and use of adjunct advice and/or complementary treatments). Each group’s 25th and 75th percentile for daily crying time were used to replace missing data; once to evaluate the worst-case scenario (25th for the Treatment group and 75th for the Control group), and once for the best-case scenario (75th for the Treatment group and 25th for the Control group). Bang’s blinding index was used to test the effectiveness of parent blinding at the end of the follow-up period.[29] All analyses were conducted using StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC.
Protocol amendments
There were three protocol amendments during the trial. The first related to the sample size (see sample size section), the second related to fidelity review (see interventions section) and the third was finalising the set of explanatory and adjusted variables during the development of the statistical analysis plan (see statistic plan). All these changes were proposed and validated by the data monitoring committee, the research management group, and the trial steering committee prior to the release of data.