Article Text

Download PDFPDF

Pain and behaviour changes in children following surgery
  1. Nina Mary Power1,
  2. Richard F Howard1,2,
  3. Angie M Wade1,
  4. Linda S Franck1,3
  1. 1UCL Institute of Child Health, London, UK
  2. 2Department of Anaesthesia, Great Ormond Street Hospital for Children NHS Trust, London, UK
  3. 3Family Health Care Nursing, School of Nursing, University of California, San Francisco, California, USA
  1. Correspondence to Professor Linda S Franck, Department of Family Health Care Nursing, School of Nursing, University of California, San Francisco, 2 Koret Way, N411F, Box 0606, San Francisco, CA 94143, USA; Linda.Franck{at}


Objectives To quantify postoperative pain and problematic behaviour (PB) in children at home following day-case (same day admission and discharge) or inpatient (≥1 night in hospital) surgery, to identify factors associated with PB at 2 and 4 weeks after discharge and to determine whether pain is associated with PB after adjustment for other factors.

Patients and methods Children scheduled for elective surgery were recruited to a descriptive study involving direct observation and self-report questionnaires. The principal outcomes were pain and PB on the 2nd post-discharge day and after the 1st, 2nd and 4th weeks.

Results 131 parents and their children (aged 2–12years) participated in the study. 93% of children had pain and 73% exhibited PB on day 2 after discharge. The incidence of pain and PB decreased over time, but 25% of children still had pain and 32% PB at week 4. Factors associated with PB were child's previous pain experience, parent and child anxiety and parent's level of education.

Conclusions There was a high incidence of pain and PB persisting for several weeks after surgery in this cohort of children. Previous painful medical experiences and anxiety were important modifiable factors that require further attention from healthcare providers and researchers to potentially improve health and social outcomes for children after surgery.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

What is already known on this topic

  • Non-UK studies suggest that following surgery a significant proportion of children exhibit pain and behaviour adjustment problems for 2 weeks or more following discharge.

  • Research also indicates that post-operative pain and problematic behaviour often co-occur and that child demographic and situational factors may increase the risk for these problems.

What this study adds

  • A high percentage of the British children studied experienced pain and problematic behaviour (PB) at home for at least 4 weeks after discharge following surgery.

  • The study also shows a strong association between a child's previous painful medical experience and post-hospital PB.


In 2010–2011, over 600 000 surgical procedures or interventions were performed on children in England.1 Hospitalisation for surgery is a stressful healthcare event in a child's life.2–6 Studies in the USA, Finland and Australia report an incidence of post-hospital problematic behaviour (PB) ranging from 24% to 67% over the first 3 days7–9 and from 23% to 54% at 2 weeks after discharge.7 ,10–12 Typical PB described includes attention seeking, temper tantrums, waking up at night, eating problems, generalized anxiety and regression. Possible risk factors identified for PB include younger age, difficult temperament or pre-operative behaviour problems, higher pre-operative anxiety, overnight hospital stay and greater pain intensity.7 ,9–13 Parent risk factors for child PB include higher pre-operative anxiety, trait anxiety and maternal over-indulgence.9 ,11 ,13 Postoperative pain at home is reported to be worse in the first 3 days and to decrease significantly thereafter.14–16 Nevertheless, severe pain at home on the day of surgery was a significant predictor of post-hospital PB 4 weeks after discharge in one study.8

The relationship between postoperative symptoms and post-hospital PB in British children following surgery had not been investigated prior to the current study. Therefore, the aims of this research were to comprehensively describe children's pain and post-hospital PB following day-case or inpatient surgery, to identify factors associated with PB at 2 and 4 weeks after discharge and to determine whether pain is associated with PB after adjustment for other factors.


This descriptive study was conducted at three London hospitals with specialised children's surgical units between November 2006 and March 2008. The study protocol was approved by the National Research Ethics Service.

Parents of all children aged 2–12 years who were American Society of Anesthesiology (ASA) physical status17 1 and 2 (normal, healthy patients and those with mild systemic disease) consecutively admitted to undergo general, urology, and ear, nose and throat surgery as day cases (children admitted and discharged on the same day) or inpatients (children who spent ≥1 night in hospital) were invited to participate. Children with an ASA status ≥3 were excluded as their underlying health status would add potentially confounding variables to the study. Other exclusion criteria were non-English speaking families, emergency surgery or children with pre-existing behavioural/psychiatric conditions and/or significant neurological disability.

Procedures and measures

Families were recruited 1–2 weeks prior to surgery and followed until 4 weeks after surgery. Eligible children were identified from the surgical admission registers. Information about the study was sent to families with the appointment letters for surgery and a researcher contacted them 1 week later to ascertain if they were interested in participating. The principal outcomes were children's post-discharge pain intensity and PB. Pain was measured using a 0–10 numerical rating scale (NRS) where 0 indicates no pain, 4–6 indicates moderate pain and ≥7 indicates severe pain.18 PB was measured using the Post-Hospital Behaviour Questionnaire (PHBQ), a validated 27-item self-report questionnaire for assessing children's behavioural adjustment after hospitalisation.4–5 7–13 For each item, parents rate the extent to which behaviour changes on a 1–5 scale compared with before surgery. Results are reported as more PB (scores 4 or 5), less PB (scores 1 or 2) or no change in PB (score 3). Subscale scores within six domains can also be derived: general anxiety (GA), separation anxiety (SA), sleep anxiety (SL), eating disturbances (EA), aggression towards authority (AA) and apathy-withdrawal (AW). Table 1 provides details of all measures and measurement intervals used to explore the influence of potential modulating factors such as anxiety, coping style and previous experiences of pain and surgery.

Table 1

Study measures and details of measurement

Statistical methods

Data analyses were performed using V.15.0 of the Statistical Package for the Social Sciences (SPSS Inc, Chicago, Illinois, USA). The χ2 test for trend was used to compare changes in percentages with pain or PB over time. Mann-Whitney U tests were used to compare categorical variables with PB (scores 1–3 (less PB than before surgery or same as before surgery) were given a score of 0 and scores 4–5 (more PB than before surgery) were given a score of 1, giving a total score range of 0–27).4–5 ,30

Stepwise multiple binary logistic regression analyses were performed to explore the relationships between children's PB and potential predictor variables at weeks 2 and 4. PB was dichotomised as PB (≥1 of the 27 PB items on the PHBQ) versus no PB (score of 0).11 Table 1 shows the large number of variables, many of which are related, that were collected to assess background factors. Univariable models were initially fitted. To provide some structure to the exploratory stages, the anxiety, coping, temperament, emotional and other questionnaire data were grouped into (i) parent factors, (ii) child factors, (iii) in-hospital factors and (iv) home experience factors.

For each of these four subgroups, a stepwise model was used to identify independent predictors and the remaining variables removed from subsequent analysis. A final model was constructed using the variables thus identified. Entry and removal of variables into the stepwise regression models were set at p=0.05 and p=0.1, respectively. Pain was then investigated for its independent association with PB.

Missing data were dealt with via listwise deletion of variables on a model-by-model basis. This meant that models were based on differing numbers as the listwise approach omits variables with missing values. The Hosmer-Lemeshow goodness-of-fit test and the −2log-likelihood/deviance were used to assess the adequacy of the models.30 ,31

Sample size

An assumption was made that at least two or three variables per factor group could be predictive of PB. Therefore, a total of 8–12 (two or three multiplied by four factor groups) variables was considered as a likely number of possible predictors. Newton and Rudestam32 propose a rule of thumb for testing a multiple correlation, applying the formula n≥50+8k, were k is the number of independent variables. Therefore with 12 likely predictor variables, n≥50+8(12)=146. This formula assumes an α of 0.05, a power of 0.8 and a medium effect size. After possible withdrawals and loss to follow-up (up to 15%) were taken into account, a sample size of 170 was chosen. If one in three (or less) have pain or PB, then this sample size would also be sufficient to estimate the percentages of any pain and PB to within ±7.7% (or better) with 95% confidence.


Recruitment, withdrawal and follow-up

Of the 1252 families who received information about the study, 522 (42%) were ineligible. In 227 (18%) cases, the parent or researcher was unavailable and in 203 cases (16%) parents declined participation, with the most common reasons being feeling too stressed, lack of time or disinterest. Therefore, total of 300 families consented to take part in the study. However, 169 (56%) exited prior to study completion because they became ineligible (21%), withdrew (25%) or were lost to follow-up (11%), leaving 131 who completed the study (figure 1).

Figure 1

Flow chart of families included in the study. #Reasons for exclusion: children not aged 2–12 years (n=148); children who had previous surgery (n=144)—the study protocol was amended after the initial 3–4 months of data collection to include children with a surgical history; pre-morbid mental health problems (n=24); non-English speaking families (n=62); too little time prior to surgery to complete baseline measures (n=82); and children no longer having surgery (n=62). *Reasons for withdrawal: lack of time (n=26); no longer interested (n=36); and families too stressed (n=10). †Reasons for ineligibility: children had turned 13 since recruitment due to surgery being postponed by 1–2 months (n=2); children with mild neurological disability, for example, cerebral palsy, that only became clear when meeting the families and discussing the post-discharge questionnaires with them as the child's behaviour could not be assessed against the Post-Hospital Behaviour Questionnaire (n=9); children admitted to intensive care after surgery (n=2); surgery cancelled: surgery cancelled/postponed/not under general anaesthetic (n=39); and families who, on meeting them, it became clear would not be able to understand/complete all questionnaires in English (n=10).

Study participants

The final sample included more boys than girls (69% male, 31% female), with a median age of 5 years (IQR 3.7); see table 2 for more demographic details. Parents were mostly mothers (91%) with the following characteristics: married/cohabiting with a partner (78%), of British ethnicity (77%), educated to graduate/post-graduate level (58%) and employed (63%). Baseline psychological measures were all within population normal ranges for both parents and children. Overall, 46% of children attended a pre-admission clinic to assess physical readiness for surgery, but only 34% (42) received any formal pre-operative preparation (eg, hospital tour, preparatory information from play specialist or nurse, etc).

Table 2

Child demographic and clinical characteristics

Pain and other symptoms

On day 2 after discharge, 93% (120/129; 95% CI 87% to 96%) of children had pain, with 26% (33) experiencing moderate pain (NRS 4–6) and 33% (43) severe pain (NRS ≥7). On day 7, 89% (102/115; 95% CI 82% to 93%) of children had pain, with 16% (18) having moderate pain and 11% (13) severe pain. At 2 weeks after discharge, 56% (67/119; 95% CI 47% to 65%) of children had pain with 7% (9) having moderate pain and 5% (6) severe pain. By the end of week 4, 25% (28/113; 95% CI 18% to 33%) of children were still experiencing some pain, with 1% (1) having moderate pain, but none severe pain (figure 2). Pain decreased significantly over the follow-up period (p<0.001, χ2). Parents reported their children had few other symptoms, such as nausea (7% at day 2, 0% at week 4) or vomiting (7% at day 2, 1% at week 4).

Figure 2

Children in pain based on intensity.

Problematic behaviours

Table 3 shows the frequency of every PB with the PHBQ subscale indicated for each. Overall, 73% (85/117; 95% CI 64% to 80%) of children exhibited PB on day 2 after discharge from hospital. The incidence of PB decreased significantly over time to 59% (68/116; 95% CI 50% to 67%) at the end of week 1, 43% (51/119; 95% CI 34% to 52%) at the end of week 2 and 32% (35/110; 95% CI 24% to 41%) by the end of week 4 (p<0.001). The most frequently reported types of PB were SA and EA. Apathy-withdrawal and GA were common on day 2 and decreased thereafter, while SL and AA peaked at week 1 and then decreased. (figure 3).

Table 3

Child behaviour change items (percentage who scored 4 or 5 indicating more problems than before surgery)

Figure 3

Problematic behaviours exhibited after discharge based on the subscales of the Post-Hospital Behaviour Questionnaire.

Associations with PB at weeks 2 and 4

Factors associated with increased PB following univariable analyses at week 2 were children 2–4 years of age (vs 5–12 years) (median PB (possible range 0–27): 1 vs 0, p=0.023, n=118), children who had experienced pain during their last medical procedure (1 vs 0, p=0.03, n=98), more educated parents (1 vs 0, p=0.037, n=103), overnight hospital stay (1 vs 0, p=0.009, n=118) and parents who took additional time off work (2 vs 0, p=0.009, n=99).

Stepwise logistic regressions for PB at weeks 2 and 4 yielded slightly different potential predictors at these time points. At weeks 2 and 4, parent baseline state anxiety and baseline trait anxiety, respectively, were independently significantly associated (tables 4 and 5). At 2 weeks, child pre-operative anxiety and child temperament sociability also independently accounted for variation in outcome, whereas at 4 weeks parental education was an independent factor. The ORs and CIs for a unit increase in each factor are given in tables 4 and 5. Thus, for every unit increase in child pre-operative anxiety, the odds of the child exhibiting PB at week 2 increased by 11% and for every unit increase in parent baseline state and trait anxiety the odds increased by 10% and 14% for weeks 2 and 4, respectively, that is, the more anxious the children (and their parents), the greater risk of children exhibiting PB. However, for every unit increase in child temperament sociability scores, the odds of children exhibiting PB at week 2 decreased by 37%, that is, the more sociable children are, the lower the odds of exhibiting PB.

Table 4

Overall predictors for problematic behaviour at the end of week 2

Table 5

Overall predictors for problematic behaviour at the end of week 4

After all other factors were accounted for, the child's pain during his/her last medical procedure was independently significantly associated with PB at both 2 and 4 weeks. The odds of PB were 8.2 (95% CI 1.9 to 35.6) times higher at 2 weeks when the child had experienced pain during his/her last medical procedure, and although this was less at 4 weeks (OR 4.8; 95% CI 1.1 to 21.3) it was still highly significant, indicating that pain during prior medical encounters is a very strong predictor of PB following a subsequent surgery.


The current results indicate that children's postoperative pain and PB can persist for several weeks. This is the first study to report such findings in British children, but our results are broadly in agreement with those from studies in children in other parts of the developed world.14–16

We found that 73% of children exhibited PB on day 2 after discharge from hospital. This incidence fell significantly but remained substantial at the end of week 4, with almost one-third (32%) continuing to exhibit PB. Overall, the factors most associated with increased PB were child temperament, children who experienced pain during their last medical procedure, child pre-operative anxiety, parent baseline state and trait anxiety, overnight hospital stay, younger child age and more educated parents. Although the incidence of PB in this study is towards the higher end of the previously reported range,7–9 this result is broadly in accordance with findings in non-British children that higher child and parent anxiety,7 ,9 ,11 ,13 younger child age9 ,10 ,12 ,14 and overnight hospital stay9 are factors associated with PB within the first 2 weeks following surgery. The child's pain during his/her last medical procedure has not been previously identified as a risk factor, but our results imply that previous pain experienced during a medical procedure can influence responses to subsequent procedures. The association between parents taking additional time off work and their child's PB has also not been reported previously. Of note, we found no effect of child attendance at a pre-assessment clinic or having received pre-operative preparation. Although there is strong research evidence that pre-operative preparation reduces peri-operative and postoperative pain and anxiety,33 our results suggest that the ad hoc manner in which pre-operative preparation is often delivered in clinical practice may not be effective.

Strengths of the current study were the multi-site sample identified from consecutive admissions for planned day-case and inpatient surgery, and its prospective design. Limitations include the relatively small sample size due to the large withdrawal rate, and that the sample was recruited from three London hospitals with dedicated children's surgical units, which may not be representative of the general paediatric population, but which represents an accelerating trend over the past 15 years.34

The most often cited reasons for withdrawal were lack of time and stress and both the lengthy questionnaires and longitudinal design may have influenced parents’ decisions to withdraw. Higher baseline anxiety scores noted in the parents who were lost to follow-up and the possibility that parents whose children had pain and/or PB were more inclined to complete the study, increases the potential for bias and may reduce generalisability. However, even if all of those who withdrew (n=72) and were lost to follow up (n=35) were free of pain and PB, a clinically significant 12% (95% CI 8% to 16%) and 15% (95% CI 11% to 20%) of children, respectively, would still be estimated to have pain and PB at 4 weeks.

In summary, there was a high incidence of postoperative pain and PB at home persisting for several weeks in this cohort, a previously unreported finding in British children that coincides with the results of similar studies from other countries with well-developed healthcare systems. PB was associated with children having had a previously painful medical experience, the presence of pre-operative anxiety and a number of other modifiable factors. We suggest that children's pain management be improved and that investigations focus on interventions targeted at the factors known to most strongly influence these outcomes.


This study would not have been possible without the valued participation of all the young children and their parents who took part in this study. We would also like to thank Susie Aldiss and Annette McQuillan for their help with participant screening and data collection, and Ru-Xin Foong for her help with data entry.



  • Contributors NMP, RFH, LSF: substantial contributions to conception and design, and interpretation of data; NMP: acquisition and analysis of data; LSF: analysis of data; AMW: substantial contributions to interpretation of data; NMP, LSF: drafting the article; NMP, RFH, AMW, LSF: revising the article critically for important intellectual content and final approval of the version to be published.

  • Funding NMP was supported by a PhD Research Fellowship from the WellChild Trust.

  • Competing interests None.

  • Provenance and peer review Not commissioned; externally peer reviewed.