Article Text

Download PDFPDF

Validation of the Ottawa Knee Rule in Iran: a prospective study
  1. Mohammad Jalili1,
  2. Hadi Gharebaghi1,2
  1. 1Department of Emergency Medicine, Tehran University of Medical Sciences, Tehran, Iran
  2. 2Department of Emergency Medicine, Imam Reza Hospital, Kermanshah, Iran
  1. Correspondence to Mohammad Jalili, Keshavarz Blvd., Imam Hospital, Emergency Medicine Department, Tehran, Iran; mjalili{at}tums.ac.ir

Abstract

Background This study was designed to determine the accuracy of the Ottawa Knee Rule (OKR) when applied to patients with acute knee injury in the Iranian population of the Imam Hospital Emergency Department (ED) at.

Methods This prospective cohort validation study included a convenience sample of all patients with a blunt knee injury sustained in the preceding 7 days presenting to the ED of a tertiary care teaching hospital during the study period. Patients were assessed for the five variables comprising the OKR, and a standardised data form was completed for each patient. Standard knee radiographs were ordered on all patients irrespective of the determination of the rule. The rules were interpreted by the primary investigator on the basis of the data sheet and the final orthopaedist radiograph reading. Outcome measures of this study were: sensitivity, specificity, positive predictive value and negative predictive value of the OKR.

Results A total of 283 patients were enrolled in the study. 22 fractures (7.77%) were detected. The decision rule had a sensitivity of 0.95 (95% CI 0.77 to 0.99), and a specificity of 0.44 (95% CI 0.37 to 0.50). The potential reduction in use of radiography was estimated to be 41%. The OKR missed only one fracture.

Conclusion Prospective validation has shown that the OKR is a highly sensitive tool for detecting knee fractures and has the potential to reduce the number of radiographs in patients with acute knee injuries.

  • Decision rule
  • fractures and dislocations, imaging
  • musculoskeletal, Ottawa Knee Rule
  • radiographs
  • x-ray

Statistics from Altmetric.com

Introduction

Acute knee injury is a common presenting complaint in the emergency department (ED).1 2 Although less than 7% of patients presenting with acute knee injury have fractures,3 more than 90% of these patients are radiographically evaluated.4 However, most radiograph results are negative,2 with fractures actually identified in only 6–12%.5 6 Despite this low incidence of fractures, common clinical practice in Imam Hospital ED is to evaluate virtually all acute knee injuries radiographically.

The Ottawa Knee Rule (OKR) is a decision device to reduce the number of negative knee radiographs without adversely affecting patient care, thereby reducing associated costs in the ED.3 7–11

A meta-analysis of the results of six OKR validation studies showed a pooled sensitivity of 98.5%.17 The authors, however, contend that further research on the OKR is warranted and should include additional validation studies. They hypothesise that other studies ‘will show a larger variation of sensitivities and specificities’.17

Considering the high incidence of knee injuries as well as the inefficiency of the current trend of its evaluation in Imam Hospital ED, it was decided to validate this rule in an Iranian population. The main objective of this study was to prospectively determine the sensitivity and specificity of the OKR when applied by physicians not involved in its development on a new set of patients. A secondary objective was to determine whether use of the rule would reduce the number of knee radiographs ordered for patients presenting with acute knee injuries to the ED.

Methods

Study design

This prospective, observational validation study was conducted between June 2007 and November 2007 on consecutive patients with documented acute blunt knee injuries presenting to the ED. The hospital is a 700 bed tertiary-care teaching hospital and offers 24 h emergency care. The ED has an annual census of approximately 40 000 patients and is staffed 24 h a day by board-certified Emergency Physicians (EPs) and Emergency Medicine (EM) residents. The study protocol was approved by the Institutional Ethics Committee. The requirement for informed written consent was waived because no intervention was performed and the data were kept confidential.

Study setting and population

Patients were potentially eligible for enrolment in the study if they were 18 years of age or older and had sustained a knee injury from any mechanism in the preceding 7 days. Patients were excluded who were referred from outside the hospital with a diagnosed fracture, were younger than 18 years, were pregnant, were paraplegic, had an altered level of consciousness, had isolated skin injuries (abrasion or laceration), had multiple injuries, had a history of metabolic bone disease or an underlying disease with sensory abnormalities, or had returned for reassessment of the same injury. ‘Knee’ was broadly defined to include the patella, the head and neck of the fibula, the proximal 8 cm of the tibia, and the distal 8 cm of the femur. A blunt trauma mechanism was defined as any injury involving a direct blow or mechanical force applied to the knee.

Study protocol

All patients considered eligible for the study were visited by supervised EM residents or EPs. Physicians were required to note their examination results on a standardised data collection instrument before reviewing the radiographs. Patients were assessed for the five variables comprising the OKR: tenderness at the fibular head or patella, inability to flex to 90°, and inability to bear weight immediately after injury and in the ED. Information on whether the patient could walk immediately after injury was obtained from patient history. Additional information collected included patient age, sex, mechanism of injury (direct blow versus sprain), and time from injury. Although the EPs had had prior exposure to the OKR, they were not specifically instructed on the application of the rules before this study. However, they underwent a 15 min training programme on how to seek the parameters of interest provided by the investigator (HG) before the initiation of the study. In accordance with usual practice, radiographs were ordered for all patients. All study participants received at least two views, anterior-posterior and lateral, of the knee, and the tangential (skyline) view of the patella was added at the discretion of the treating physician. The radiographs were, as usual, reviewed by the treating physician for immediate treatment and were interpreted, within 24 h, by a board-certified orthopaedic surgeon who was blinded to the contents of the data collection sheet. The result of the radiographic examination was recorded on the data collection sheet. Any fracture identified by the orthopaedist was included in the study. The sheet was then sent for processing.

Measurements

A positive outcome was defined as any fracture, regardless of size. A negative outcome was defined as patients who did not have a fracture on radiograph.

Data analysis

All data were entered into a Microsoft Access 2003 database and transported into SPSS version 11.2 for Windows for statistical analysis. The performance of the decision rule for identifying patients with a clinically important fracture was examined in the study cohort by calculating sensitivity, specificity, positive predictive value, negative predictive value and likelihood ratios with 95% CI. The potential relative reduction in the rate of obtaining radiography was estimated by comparing the theoretic rate of referral in this study with the usual rate.

Results

Overall, during the 6 month study period a total of 298 patients were eligible for enrolment in the study. Of these, 15 met exclusion criteria, leaving a total of 283 (94.96%) patients who were evaluated. Of the 15 patients excluded, six were excluded because they were younger than 18 years old, three were excluded because their injuries were more than 7 days old, two for previous knee surgery or fractures, and four for isolated skin injuries.

The mean age in the study group was 37.3±14.2 years (age range, 18–78 years), and 184 (65.01%) were male patients. Most patients were young adults, 62% being between 20 and 40 years old. Direct blow was a more frequent mechanism of fracture than sprain (95% vs 5%). Only 10% of the patients had been referred by emergency medical services. All patients received knee radiographs.

Overall, 22 clinically important fractures (incidence 7.77%) were found in this study; patella and tibial plateau fractures were the most common. The types of fractures and their frequencies are noted in table 1.

Table 1

Types of knee fractures identified (N=22)

All except one clinically important fracture were identified by the decision rule; this was an undisplaced fracture of tibial plateau that was then treated conservatively. Overall, the OKR was 95.4% sensitive in detecting fractures (95% CI 77.1% to 99.8%) with a specificity of 44.0% (95% CI 37.9% to 50.1%). Negative predictive value of the OKR was 99.1% (95% CI 95.2 to 99.9%). The positive predictive value was 12.5% (95% CI 7.9 to 18.5%). The negative likelihood ratio was 0.09% (95% CI 0.01 to 0.66%). This information is illustrated in table 2.

Table 2

Performance of the Ottawa Knee Rule in the identification of knee fractures among the study patients

The secondary objective of this study was to determine whether use of the rule would reduce the number of knee radiographs ordered for patients presenting with knee injuries to the ED. Application of the OKR by physicians to the study patients would have led to a relative reduction in the use of knee radiography of 41%, compared to the current practice of obtaining radiographs in every case (from 100% to 59%).

Discussion

Clinical prediction rules reduce the uncertainty of decision-making in emergency patient care.5 Over the past decade, several clinical decision rules for utilisation of radiography in patients with knee injuries have been derived.8 13 14 From these, the OKR was selected for study because it has been developed according to methodologically accepted criteria. The rule was originally derived and validated by Stiell et al in EDs in Ottawa, but has since been implemented and prospectively assessed in a multitude of EDs in North America, and Europe.4 8–10 17 No studies were found that questioned the generalisablity of the OKR. However, clinical prediction rules may not perform as well when tested in a population other than that in which they were derived.1 12 Therefore, even well-defined decision-making rules are not suitable for application in all clinical settings without validation studies. To the best of the authors’ knowledge, however, no other study has prospectively validated this decision rule in Asian countries. Therefore, considering the differences in patient populations, validation of the OKR was considered in this study.

The present study confirms the good performance of the OKR in identifying knee fractures, as previously published by the researchers who developed it.9 The proportion of fractures in the present setting was comparable to that in other published studies1 4 9; a similar distribution of fracture sites has also been observed.1 3 5 It was ensured that no clinically important fracture was missed, by obtaining radiographic examination in all patients. The present finding of blunt trauma (rather than sprain) being a more frequent mechanism of injury in knee fractures is in accordance with that reported in other studies.17

The present findings are slightly less promising than most other studies. Most previous studies1 5 9 10 18 19 have reported sensitivities approaching 100%. Only one previous study20 reported a sensitivity of 84.6%, which is lower than that found in this study. In the present study, OKR proved to be as specific as in most other studies.9 10 18 20 Two studies 5 19 reported lower levels of specificity (27% and 31.6% respectively), and only one study1 reported higher levels (52%). It should be mentioned that the lower confidence limit of the sensitivity in the present study is well below the critical value for cost-effectiveness, as calculated by Nichol and coworkers.21

According to the present study, implementing the OKR would have resulted in an overall 41% reduction in radiograph use. However, the true reduction in radiography cannot be determined unless an implementation trial is performed. The potential reduction of the number of requested knee radiographs that could be achieved in the ED by applying the OKR is slightly lower than the reduction estimated in a study performed in Europe,4 and much higher than the rate of reduction in the original report of the rule,10 which may have been caused by the different practices on use of radiography.

Additionally, although this variable was not measured in the present study, using these rules may decrease time spent in the ED by not ordering a radiograph. Stiell et al found that adults who underwent radiography spent an average of 118.8 min in the ED, whereas those who did not stayed for only 85.7 min.10

Although radiographs are relatively low cost in Iran (based on the tariffs confirmed by the Ministry of Health for public hospital services), they are high-volume tests. If only 41% of radiographs could be avoided by using the OKR, savings would be US$1000 each year. One should add to this amount the indirect costs saved by reducing patients' length of stay in the ED. These savings seem to be of paramount importance in developing countries, where shortage of resources is a challenge.

Strict methodological standards for the development and testing of clinical decision rules have been established.15 16 In the present study, the outcome measure (fracture) was biological in nature, and was assessed blindly (without knowledge of the predictor variables). All knee fractures were considered clinically significant. The predictor variables were well defined, have been shown to have good interobserver agreement, and were performed before knowledge of the outcome.

Despite these strengths, the present study faced several limitations: the major limitation is that the relatively low fracture rate of 7.77% makes it difficult to generalise the results to other populations. Another limitation is that no attempt was made to assess the interobserver agreement for the predictor findings. This may have resulted in performance bias.

Furthermore, the radiographic interpretation of the orthopaedist was relied on for diagnosis of fractures and no follow-up was conducted in patients with knee injuries. So, in spite of the fact that the orthopaedist was a board-certified and experienced practitioner, it is still possible that some fractures were missed. Reliance on the interpretation of two independent musculoskeletal radiologists would make the methodology more robust.

Suggestions for further evaluation include assessing the validity of the OKR with more cases and in different populations, and studying physicians' attitude about and satisfaction with using these rules in the clinic setting. Further research should also be conducted in the implementation of the OKR to see whether a reduction in radiography utilisation can really be achieved without compromising patient care.

In summary, it was found that the OKR is valid in adults (≥18 years) with knee injuries in the present study centres. It has been shown to be sensitive in detecting knee fractures and can potentially decrease the use of radiography in these patients without reducing the quality of care, therefore improving the allocation of medical resources. Results of this study confirm previously published results.

What is already known on this subject

  • The Ottawa Knee Rule has been developed and validated in different settings and has been shown to be a sensitive screening test.

  • It can reduce the number of unnecessary radiographs.

What this study adds

  • This study adds to the bulk of evidence that supports the use of Ottawa Knee Rule.

  • It validates the rule in a new setting.

Acknowledgments

The authors would like to thank Dr Nader Toosi, for interpreting all knee radiographs.

References

Footnotes

  • The abstract was presented in the Fifth Mediterranean Emergency Medicine Congress (September 2009; Valencia, Spain).

  • Competing interests None.

  • Ethics approval This study was conducted with the approval of the Institutional Ethics Committee (TUMS).

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

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.

Linked Articles

  • Primary survey
    Steve Goodacre