Results of the search

The searches identified 1289 reports after duplicates were removed (Figure 1). We examined 11 full‐text articles in detail and assessed them for eligibility. We subsequently included three studies (four reports) and excluded seven studies.

Included studies

We included three studies (four reports) with 246 children undergoing tonsillectomy (Rømsing 1998; Sutters 2004;Sutters 2010). Sutters 2005 provided additional information for the study by Sutters 2004. Children were aged five to 15 years in one study (Rømsing 1998), and aged six to 15 years in the remaining two studies (Sutters 2004; Sutters 2010). The Characteristics of included studies table provides details.

Rømsing 1998 was conducted in Denmark; Sutters 2004 and Sutters 2010 took place in the US. Neither Rømsing 1998 nor Sutters 2004 declared a funding source; Sutters 2010 was supported by a grant from the National Institute of Nursing Research, USA.

Boys and girls were evenly assigned to the randomisation groups for all three studies.

Rømsing 1998 examined the outpatient management of post‐tonsillectomy pain in 53 children randomly assigned to two groups. Children were given paracetamol (acetaminophen) in weight‐appropriate doses for the first three days following discharge from hospital. The study reported that in Denmark, recommended dose ranges for this population were 60 mg/kg per 24 hours orally or 90 mg/kg per 24 hours rectally divided into four to six daily doses. The children in both groups were able to choose the route of administration, either oral (tablet (125 or 500 mg) or elixir (120 mg/5 mL)) or rectal (suppository (125, 250, 500, or 1000 mg)). The study group followed an ATC analgesic administration. The control group used paracetamol PRN (i.e. when the child complained of pain or was thought to be in pain by the parent or carer).

Sutters 2004 randomised 88 children to one of three groups. All children received elixir of paracetamol with codeine in weight‐appropriate doses (paracetamol 120 mg/5 mL with codeine 12 mg/5 mL) for three days after surgery. The first group ('PRN group') received analgesics every four hours PRN, without nurse coaching; parents were given instructions on dose measurement. The second group ('ATC group') received analgesics every four hours ATC, without nurse coaching. The third group ('ATC + coaching group') received analgesics every four hours ATC with nurse coaching. Children in all three groups were provided with a teaching booklet. Nurse coaching involved reviewing the pain management guidance in the booklet, and providing further information about postoperative pain (e.g. rationale for combining opioids and non‐opioids, and strategies for improving the participants' adherence to analgesic consumption).

Sutters 2010 randomised 123 children to one of three study groups as described above in Sutters 2004. All children received elixir of paracetamol with hydrocodone in weight‐appropriate doses (maximum daily paracetamol dose approximately 73 mg/kg and hydrocodone approximately 0.2 mg/kg/dose) for three days after surgery.

Excluded studies

We excluded seven studies after reading the full reports (Bean‐Lijewski 2007; Higgins 1999;King 1967;Najafi 2010;Patterson 2002;Smyth 2004;Tubbs 2007). Four studies were not RCTs, two did not assess PRN versus ATC analgesic administration, and one included a different population. The Characteristics of excluded studies table provides details.

Allocation

All three included studies were randomised.

Two studies were at low risk of random sequence generation (selection) bias, as they used an SPSS software‐generated random selection process (Sutters 2004; Sutters 2010). Rømsing 1998 was at unclear risk of bias as the sequence was only described as 'randomly assigned'.

Allocation concealment (selection) bias was low for Sutters 2010, but it was unclear for the remaining two studies.

Blinding

It was not possible to assess the included studies for blinding of the investigators or participants. We assessed any methods used to blind the outcome assessors from knowledge of which intervention a child received. Parents were asked to record the child's pain scores in a home diary or medication log; the studies did not report how these documents were assessed or whether the assessors were blinded to the intervention.

Incomplete outcome data

All three included studies were at low risk of attrition bias: in Rømsing 1998, 13/53 children or parents 'did not complete the forms'; in Sutters 2004, 8/88 children withdrew due to intolerable adverse effects; in Sutters 2010, 10/123 children became ineligible or parents withdrew consent.

Size of study

We assessed the size of the study as a potential source of bias. All studies were at high risk of bias due to size: there were fewer than 50 children in total in Rømsing 1998, and fewer than 50 children per treatment arm in Sutters 2004. More than 123 children were randomised in Sutters 2010, but there were fewer than 50 children in one arm, and a combined total of 81 children randomised to the other two arms; therefore, we assumed the children were divided evenly between arms two and three, hence the high risk of bias assessment.

Selective reporting

All included studies were at low risk of reporting bias.

Pain intensity and pain relief

Rømsing 1998: pain intensity was measured by the children using the poker chip tool (PCT), where one chip was 'a little piece of hurt' and four chips were 'the most hurt a child could have'. The scores were given on a 5‐point scale of 0 to 4, with scores of 3 and 4 classified as severe pain. Pain intensity scores were recorded by the parents based on child self report twice a day for three days following surgery.

For both ATC and PRN groups, mean pain intensity scores tended to be higher in the morning than the evening, except on day one where the mean pain scores were lower in the morning. A higher pain score in the morning is expected due to the length of time between doses overnight. A lower pain score on the morning after surgery is also expected due to the effects of preoperative analgesia given on the day of surgery.

On day two, the study authors stated that the ATC group reported lower mean pain scores than the PRN group both in the morning and the evening, but this was not statistically significant. These data were derived from tables 2 and 3 on page 237.

Pain relief was not reported.

Sutters 2004: pain intensity, with and without swallowing, was measured by the children using a descriptive NRS from 0 (no pain) to 10 (worst pain imaginable), prior to bedtime and upon awakening, on the evening of the day of surgery and twice a day for the following three days. Pain relief was rated using a 0 (no change) to 10 (pain gone) NRS at the same time. Mean pain intensity scores ranged from approximately 5 to 8 on the evening of the day of surgery, and had reduced to a range of approximately 2 to 5 on the evening of day three. There were no differences in pain intensity or pain relief scores between groups. There were no differences between pain intensity with swallowing and without swallowing. These data were derived from figure 2 on page 54.

Sutters 2010: pain intensity, with and without swallowing, was measured by the children using a descriptive NRS from 0 (no pain) to 10 (worst pain imaginable), prior to bedtime and upon awakening, on the evening of the day of surgery and twice a day for the following three days. Pain intensity scores were higher for both ATC and PRN groups in the morning than in the evening for all three days. The mean pain intensity scores were numerically but not significantly higher in the PRN group than the ATC group, both with swallowing (3.9 with PRN versus 3.0 with ATC) and without swallowing (4.2 with PRN and 3.2 with ATC). Pain relief was not reported.

Both Sutters 2004 and Sutters 2010 also assessed the impact of nurse coaching of parents. Although not a critical part of this review, they showed that compliance with the ATC regimen was high and not improved by nurse coaching. These data were derived from table 2 on page 16.

No data were available for analysis.

Adverse events

Rømsing 1998: two children experience nausea and vomiting (one in each group). Two children reported otalgia (one in each group). One child reported a sore tongue (control group).

Sutters 2004 reported adverse events in the secondary paper Sutters 2005: assessed frequency of adverse events; believed that increased analgesics according to an ATC regimen, would mean increased adverse effects.

There were no differences between groups in the following symptoms: lightheadedness, feeling dizzy (approximately 4% to 32% of children across groups); nightmares (occurred infrequently); nausea (12% to 43%); vomiting (2% to 29%); doses of antiemetics (30%); and constipation (6% to 24%). There was no difference for sedation, and across groups there were significant decreases in sedation over the four postoperative assessments; 14% to 37% of children experienced daytime sedation. However, on day three, the ATC groups experienced significantly more sedation (7/52 with ATC versus 0/52 with PRN).

The study authors reported that nausea and vomiting may have been related to the study medication.

Sutters 2010: assessed sedation, lightheadedness, dizziness, nightmares, nausea, vomiting, and constipation: there were no significant between‐group differences, except that children in the ATC group did experience increased constipation. Adverse events were rated by the child before bedtime and upon awakening, using a 0 (did not have) to 4 (very severe) NRS.

Child and carer global impression

Not reported.

Medication use

Rømsing 1998: reported the mean cumulative dose in milligrams per kilogram of paracetamol (acetaminophen) administered orally and rectally during the first three days following discharge.

Oral administration: on day two, the mean cumulative dose (for days one and two) was reported as 120.5 mg/kg (standard deviation (SD) 20.9) with ATC (study group) and 48.6 mg/kg (SD 22.9) with PRN (control group). On day three, the mean cumulative dose (for days one to three) was reported as 180.7 mg/kg (SD 31.8) with ATC and 69.7 mg/kg (SD 24.6) with PRN.

For rectal administration: on day two, the mean cumulative dose was reported as 172.3 mg/kg (SD 22.8) with ATC and 69.8 mg/kg (SD 26.5) with PRN. On day three, the mean cumulative dose was reported as 253.8 mg/kg (SD 32.7) with ATC and 81.8 mg/kg (SD 37.4) with PRN.

The study authors reported that the dose administered orally to the PRN group represented 39% of the dose administered to the ATC group, and that the dose administered rectally to the control group represented 32% of the dose administered to the study group.

In Sutters 2004 and Sutters 2010, the parents or nurses recorded the volume in millilitres when administering medication. Sutters 2004 reported that medication use decreased over time for both groups. It was not clear to the study investigators what criteria were used by parents in the PRN group when decreasing the amount of analgesic given to their child over time. There were no differences in the total amount of analgesic administered over time for the ATC groups (with and without coaching). Sutters 2010 reported that a higher amount of analgesics was consumed in the ATC group compared with the PRN group.

Requirement for rescue analgesia

Not reported.

Length of postoperative stay

All children were discharged home on the day of surgery.

Sleep duration and quality

Only Sutters 2010 reported night‐time doses, sedation, and nightmares. There were no differences between groups for sedation and nightmares. Across the three postoperative days, children in the ATC group received a mean nine night‐time doses of the analgesic, whereas the PRN group received only four on average. Both groups reported pain on waking.

Acceptability of treatment

No study specifically reported acceptability of treatment.

In Rømsing 1998, parents reported that children experienced nausea and vomiting (two children, one from each group), sore tongue (one child from control group), and otalgia (two children, one from each group). The study authors did not comment on whether these symptoms may have been caused by the study medication.

In Sutters 2004, eight children developed intolerable adverse effects (i.e. persistent nausea and vomiting); these children were withdrawn from the study and given different medication. Nausea and vomiting were also measured each evening on a scale of 0 (none) to 4 (severe). There were no between‐group differences for severity of nausea and vomiting reported.