Safety and efficacy of routine postoperative ibuprofen for pain and disability related to ectopic bone formation after hip replacement surgery (HIPAID): randomised controlled trial

BMJ 2006; 333 doi: (Published 07 September 2006) Cite this as: BMJ 2006;333:519
  1. Marlene Fransen, senior research fellow (mfransen{at},
  2. Craig Anderson, director, neurological and mental health division1,
  3. Jan Douglas, project manager1,
  4. Stephen MacMahon, principal director1,
  5. Bruce Neal, director, cardiac and renal division1,
  6. Robyn Norton, principal director1,
  7. Mark Woodward, director, epidemiology and biostatistics division1,
  8. Ian D Cameron, director and professor of rehabilitation medicine2,
  9. Ross Crawford, professor of orthopaedic research3,
  10. Sing Kai Lo, associate professor of biostatistics4,
  11. Garnet Tregonning, orthopaedic surgeon5,
  12. Margaret Windolf, research nurse HIPAID Collaborative Group6
  1. 1 The George Institute for International Health, University of Sydney, Australia
  2. 2 Faulty of Medicine, University of Sydney, Australia
  3. 3 Queensland University of Technology, Australia
  4. 4 Deakin University, Melbourne, Australia
  5. 5 Middlemore Hospital, Auckland, New Zealand
  6. 6 Royal Prince Alfred Hospital, NSW, Australia
  1. Correspondence to: M Fransen
  • Accepted 14 July 2006


Objectives To determine the benefits and risks of a non-steroidal anti-inflammatory drug (NSAID) as prophylaxis for ectopic bone formation in patients undergoing total hip replacement (or revision) surgery.

Design Double blind randomised placebo controlled clinical trial, stratified by treatment site and surgery (primary or revision).

Setting 20 orthopaedic surgery centres in Australia and New Zealand.

Participants 902 patients undergoing elective primary or revision total hip replacement surgery.

Intervention 14 days' treatment with ibuprofen (1200 mg daily) or matching placebo started within 24 hours of surgery.

Main outcome measures Changes in self reported hip pain and physical function 6 to 12 months after surgery (Western Ontario and McMaster University Arthritis index).

Results There were no significant differences between the groups for improvements in hip pain (mean difference −0.1, 95% confidence interval −0.4 to 0.2, P = 0.6) or physical function (−0.1, −0.4 to 0.2, P = 0.5), despite a decreased risk of ectopic bone formation (relative risk 0.69, 0.56 to 0.83) associated with ibuprofen. There was a significantly increased risk of major bleeding complications in the ibuprofen group during the admission period (2.09, 1.00 to 4.39).

Conclusions These data do not support the use of routine prophylaxis with NSAIDs in patients undergoing total hip replacement surgery.

Trial registration NCT00145730 [].


Chronic symptomatic osteoarthritis of the hip is common in those aged ≥ 50 and total hip replacement is a well established and highly effective treatment.1 While surgery reduces pain and improves physical function in most people, residual symptoms are common.24 One determinant of the risk of long term pain and disability after hip replacement is ectopic bone—abnormal bone that forms postoperatively in the soft tissues around the operated hip.5 Some ectopic bone occurs in more than one third of all patients who undergo hip replacement.6 Both the risk of occurrence and the severity, as judged from radiographic measurements, can be greatly reduced by a short course of postoperative non-steroidal anti-inflammatory drugs (NSAIDs).7

Routine prophylaxis with a short course of postoperative NSAIDs has been advocated for all patients undergoing total hip arthroplasty89 because it is not possible preoperatively to identify patients at risk of developing ectopic bone. All recent randomised clinical trials are uniformly characterised by the lack of a placebo group. Before the widespread introduction of such a preventive strategy, more evidence is required about the balance of benefits and risks, particularly in light of recent concerns about the safety of some NSAIDs.10 Specifically we need to determine whether improvements in radiographic abnormalities produced with NSAIDs result in worthwhile benefits for long term pain and physical function.7 We also need to determine whether there are any safety issues.

We established the effects of a routine short course of postoperative ibuprofen1112 on pain and physical function 6 to 12 months after total hip replacement surgery. We also evaluated the effect of treatment on other measures of physical function, radiographic ectopic bone formation, and bleeding complications.


We carried out this double blind, randomised, placebo controlled trial in patients undergoing elective total hip replacement surgery at 20 hospitals in Australia and New Zealand between February 2002 and May 2004. All patients provided written informed consent, and an independent safety and monitoring committee reviewed the data during recruitment. Details of the study methods and implementation strategy have been published elsewhere.13

Participants—Patients identified within 24 hours of completed elective total hip replacement surgery (primary or revision) were eligible for inclusion, irrespective of age, reason for surgery, or procedure performed. Patients were ineligible if there was, in the opinion of the responsible physician, a definite indication for treatment with an NSAID during the 14 day study treatment period (for example, patients in whom no other analgesic agent was deemed suitable) or a definite contraindication for treatment with an NSAID (for example, previous serious adverse reaction to an NSAID, previous major gastrointestinal bleed, serious renal impairment, or known bleeding disorder). In addition, patients who had taken an NSAID (other than low dose aspirin) in the 48 hours before the operation were not eligible nor were patients with a postoperative spinal catheter in situ unless the catheter had been removed at least two hours before randomisation.14

Randomisation—Randomisation was performed centrally by using a computer based system accessible 24 hours a day via a toll-free telephone call within 24 hours after surgery. On confirmation of eligibility, researchers used a minimisation algorithm15 to provide a unique randomisation code corresponding to a treatment pack held at the centre. We used a minimisation program to stratify treatment by study centre and type of surgery performed (primary or revision). Treatment allocation was blinded and concealed from patients and study staff until the database was locked.

Treatment and control—Participants were randomised to receive 14 days of treatment with either ibuprofen (2x200 mg tablets three times daily) or matching placebo tablets. All study tablets were packaged in identical blister packs. Treatment was scheduled to start within 24 hours after surgery and patients could not take other NSAIDs (with the exception of low dose aspirin) during the study period. The protocol required no other changes to usual preoperative or postoperative care.

Data collection and follow-up—Demographic variables and baseline clinical data were obtained during the usual pre-admission clinic visit, typically one to two weeks before surgery. Information about the anaesthetic and surgical techniques, postoperative care, and early indicators of bleeding complications was collected during hospital admission. All other outcomes were recorded at a routine follow-up visit scheduled, depending on the operating centre, for between 6 months and 12 months after surgery. Serious adverse events in hospital were documented as they occurred and were specifically inquired about during the 14 day monitoring telephone call and at the follow-up visit. All assessments were standardised and performed blind to randomised treatment allocation.

Outcomes—Our primary study outcomes were the changes from baseline to follow-up in self reported hip pain and physical function measured by the Western Ontario and McMaster Universities arthritis index (WOMAC) questionnaire (Likert version).16 We standardised scores to a range of 0-10, with 0 indicating no hip pain or no difficulty with daily activities and 10 indicating severe hip pain or severe difficulty with daily activities. Secondary outcomes were general health status (summary scale scores on physical and mental components17) of the medical outcomes study short form 36 (SF-36)18); patients' global assessment of effectiveness of treatment (hip status compared with before surgery; hip status today) with five response levels; frequency of use of analgesia for hip pain during the past week; ability to get “about the house” and ability to get “out of the house” with five response levels ranging from “not at all” to “no difficulty”; time spent participating in physical activity during the past week; objective measures of physical performance (hip flexion,19 time to walk 50 feet (about 15 metres), and timed “up and go”20); radiographic evidence of ectopic bone formation according to the Brooker classification21; and major bleeding complications during hospital admission (bleeding from the wound for more than three days, evacuation of a wound haematoma, haematemesis, melaena, other serious bleeding event). Red cell transfusions (or re-infusions), suction drainage volumes, and postoperative haemoglobin concentrations (measured ≥ 48 hours after surgery) were also recorded.

Analysis—We planned to recruit 1000 patients to provide 90% power (α = 0.05) to detect a difference of ≥ 10% between the randomised groups for each of the two primary outcomes.13 The actual sample size of 902 achieved at the end of the recruitment period provided 87% power to detect these effects. We used t tests to evaluate changes in the primary outcomes of pain and physical function (WOMAC) and health related quality of life (SF-36) and to compare differences in other continuous outcome measures at follow-up. In each case we calculated the estimated difference between randomised groups, the 95% confidence interval of the difference, and the corresponding P value. We compared categorical outcomes for the proportions of patients with events with χ2 tests to obtain a P value and calculated the relative risk and 95% confidence intervals. For ordinal outcomes we evaluated the effects of randomised treatment by fitting a proportional odds model and calculating the odds ratio (and 95% confidence intervals) of an improved outcome with ibuprofen.15 In each case we tested the assumption of proportionality and found it was not violated. All analyses were done according to the principle of intention to treat. We carried forward baseline assessments when follow-up data were missing.


We were able to randomise only 902 patients because of slow recruitment and funding limitations (452 to ibuprofen and 450 to placebo, figure). Baseline data were missing for four (0.4%) and primary outcome data were missing for 27 (6%) allocated to ibuprofen and 22 (5%) allocated to placebo (figure). Of those who completed the primary outcomes assessment, 823 (96%) assessments were conducted during an outpatient clinic visit and the remainder by telephone. The median (range) period of follow-up was 7.6 months (5-18 months) and 7.9 months (5-20 months) for the ibuprofen and placebo groups, respectively. Only 16 (2%) outcomes assessments (6 and 10 participants respectively) occurred 14 or more months after surgery. Standard anteroposterior radiographs, scheduled to be taken 6 to 12 months after surgery, were obtained for 798 (88%) participants. There was no significant difference between the two allocation groups in follow-up rates for any of the outcome measures.

Embedded Image

Details of flow of participants through study

Baseline characteristics

The mean age of participants was 66, and 54% were men. Most had a diagnosis of osteoarthritis and were undergoing a primary hip replacement. The groups were well balanced regarding demographics, clinical history, surgical technique, and anaesthesia (table 1).

Table 1

Baseline characteristics and features of surgery in patients undergoing hip replacement according to postoperative treatment. Figures are numbers (percentages) of patients unless stated otherwise

View this table:

Adherence to randomised treatment

In total 875 (97%) patients started the randomised treatment, and 188 (21%) stopped prematurely (106 (24%) in the ibuprofen group and 82 (19%) in the placebo group, P = 0.06, figure). Treatment was stopped, usually on medical advice and mainly because of suspected side effects or intolerance (11% v 8%, P = 0.13) or other unspecified medical reasons (8% v 7%, P = 0.34).

Effects of randomised treatment

Pain and physical function—There was no significant differences between the groups for improvements in hip pain (mean difference −0.1, 95% confidence interval −0.4 to 0.2, P = 0.59) or physical function (−0.1, −0.4 to 0.2, P = 0.48) 6 to 12 months after surgery (table 2).

Table 2

Mean (SD) score for pain, physical function, and general health status before surgery (baseline) and at follow-up (6 to 12 months after surgery) in patients undergoing hip replacement according to postoperative treatment with change and difference in change with 95% confidence intervals

View this table:

Secondary clinical outcome measures—There were no significant differences between the groups on the secondary clinical outcomes of general health status (table 2), global assessments (table 3), participation in physical activity, or objective measures of physical performance (table 4). Furthermore, the odds of having a better outcome in terms of global assessment of effectiveness of treatment (hip status today, mobility “out of the house”) or use of analgesia for hip pain were not significantly increased among patients allocated to ibuprofen (table 3).

Table 3

Numbers (percentages) of patients with global assessments 6 to 12 months after hip replacement according to postoperative treatment

View this table:
Table 4

Mean (SD) measures of physical activity and physical performance, 6 to 12 months after hip replacement according to postoperative treatment

View this table:

Bleeding complications during admission—There was a significantly increased risk of major bleeding complications among those in the ibuprofen group (risk ratio 2.09, 1.00 to 4.39, P = 0.046) (table 5). There were no significant differences between groups in the proportion of patients requiring red cell transfusion (ibuprofen 37% v placebo 34%, P = 0.35), suction drainage volumes (415 ml v 424 ml, P = 0.71), or postoperative haemoglobin concentrations measured ≥ 48 hours after surgery (105 g/l v 105 g/l, P = 0.80). The latter result was materially altered when we excluded transfused (or re-infused) patients from the analyses (102 g/l v 100 g/l, P = 0.26).

Table 5

Adverse events during admission and follow-up after hip replacement according to postoperative treatment

View this table:

Serious adverse events during follow-up—Eight participants died. All deaths occurred between 6 days and 180 days after the end of the study treatment (median 78 days) (table 5). The difference in the numbers of serious adverse events between the allocation groups was not significant.

Ectopic bone formation—There was a highly significant decrease in the risk of developing ectopic bone of any grade (0.69, 0.57 to 0.83) and in the risk of developing severe ectopic bone (Brooker grade 3 or 4) (0.44, 0.22 to 0.88) among patients in the ibuprofen group (table 6). The odds of developing a more severe grade of ectopic bone with ibuprofen was 0.55 (0.41 to 0.73). Patients with Brooker grade 3 and 4 had higher pain and disability scores than those with less severe grades of ectopic bone formation, though this trend was not significant.

Table 6

Ectopic bone formation and associated clinical outcomes (WOMAC pain and function), 6 to 12 months after hip replacement surgery according to postoperative treatment. Figures are numbers (percentages) of patients with mean (SD) pain and function

View this table:


Ibuprofen routinely administered after total hip replacement surgery does not result in better long term clinical outcomes, despite significantly decreasing the risk of ectopic bone formation. Postoperative ibuprofen also increases the risk of serious bleeding complications.

We carried out this study because observational studies had reported adverse effects of ectopic bone on clinical outcomes5 and earlier trials provided clear evidence that NSAIDs decreased the occurrence of radiographic ectopic bone.7 We observed the anticipated beneficial effect of ibuprofen on radiographic outcomes but found no corresponding improvement in clinical outcomes. The most plausible explanation for this finding is that minor or moderate ectopic bone has little effect on clinical outcomes after hip arthroplasty. This explanation is supported by the WOMAC pain and function subscales reported in table 6. While severe ectopic bone (Brooker grades 3 and 4) can impair outcome, this forms in only a small proportion of patients. Hence, although a much larger trial might detect some beneficial effect of ibuprofen on clinical outcomes, any such clinical benefit would be small in absolute terms and probably inconsequential in the context of the large improvement in clinical outcomes achieved with joint replacement surgery itself.

The study treatment was generally well tolerated, with no significant difference in rates of discontinuation between groups. There was a borderline significant increase in major bleeds among patients in the ibuprofen group, which might reflect the antiplatelet effects of cyclo-oxygenase I inhibition.22 While there was no clear effect of study treatment on other measures of bleeding in the postoperative period, an increase in risk of bleeding is consistent with the established effects of other NSAIDs.23

Our results provide no evidence of clinical benefit 6 to 12 months postoperatively and raise concerns about the safety of ibuprofen for the prevention of ectopic bone formation after hip arthroplasty. There is no particular reason to believe that other conventional NSAIDs would have produced materially different results.2324 These findings, therefore, suggest that recommendations promoting routine prophylaxis with a short course of postoperative NSAIDs for all patients undergoing hip arthroplasty are not justified. While some patients at high risk of ectopic bone formation (such as those with a history of it) may derive clinical benefits from prophylaxis with NSAIDs that outweigh any risks, randomised trials are required to substantiate this. Our results provide further evidence that guidelines for routine clinical care in surgery, as in other specialties, must be based on clinically important outcomes rather than surrogates such as radiographic ectopic bone formation.25 Without such evidence, the widespread use of routine prophylaxis with NSAIDs on the basis of radiographic changes may well have resulted in net harm rather than benefit.

What is already known on this topic

Ectopic bone, or abnormal bone that forms in local soft tissues, is common after hip replacement surgery

A short course of postoperative NSAIDs greatly reduces the risk of this abnormal radiographic outcome

As it is not possible to identify patients at risk, routine prophylaxis has been recommended

What this study adds

Despite a significantly reduced rate of ectopic bone formation among patients who took NSAIDs postoperatively, there were no significant clinical benefits 6 to 12 months after surgery

Postoperative NSAIDs are also associated with an increased risk of bleeding events among these patients, and routine prophylaxis is not recommended


  • Embedded ImageMembers of the HIPAID Collaborative Group are on

  • We thank Ken Faulder, radiologist, Royal North Shore Hospital, NSW, Australia, who kindly assessed and scored all trial radiographs.

  • Contributors All members of the writing committee contributed to the study or protocol design, or both, overall study management, interpretation of the data, and writing of the paper. MF and RN directed the project. JD was project manager. MF and RN are guarantors.

  • Funding National Health and Medical Research Council of Australia and the Medical Benefits Fund of Australia provided funding through competitive peer reviewed processes. Competing interests: None declared.

  • Ethical approval University of Sydney human research ethics committee and ethics committees at each collaborating hospital approved the study.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
View Abstract