Université Paris-Descartes, Faculté de Médecine, INSERM U729, AP-HP, Hôpital Européen Georges Pompidou, Department of Hospital Informatics, Evaluation and Public Health, 20 rue Leblanc, 75015 Paris, France,¹ Emergency Network Department (Medicine and Traumatology), AP-HP, Hôpital Européen Georges Pompidou, 20 rue Leblanc, 75015 Paris, France,² SPIM, Paris 6 University, Paris, France,³ AP-HP, Hôpital Lariboisière, 75010 Paris, France,⁴ Centre Hospitalier de Versailles, 78000 Versailles, France⁵

Received 7 April 2005/ Returned for modification 17 May 2005/ Accepted 10 June 2005

ABSTRACT

Top Abstract Introduction Materials and Methods Results Discussion References

INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

In 2000, almost 1,320,000 patients were referred to the 550 French hospital emergency departments for wounds, accounting for 12% of the 11 million referrals (7). According to guidelines issued in 1992 by the French Ministry of Health, measures to prevent tetanus in these patients (essentially, appropriate wound cleaning and the injection of immunoglobulin and/or vaccine) should depend on wound characteristics and the patient's immunization status (10). However, this algorithm is expected to change because it is difficult to apply for two reasons. First, the severity of the wound does not accurately predict the risk of tetanus; for example, a small cut while gardening (contact with soil) may entail a greater risk than a larger wound made with a clean kitchen knife. Second, enzyme-linked immunosorbent assay (ELISA) is currently the only methodology available for evaluation of serum tetanus antitoxin levels.

The objective of this study was to evaluate the feasibility and the accuracy of a rapid blood test, Tetanos Quick Stick (TQS), used in emergency departments for the diagnosis of tetanus immunization status. The ELISA test was considered as the criterion standard. The feasibility of the test must be evaluated prior to its cost-effectiveness. A secondary objective was to compare the performance of the TQS to the performance of a reported structured clinical history for the diagnosis of vaccination status.

MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Study design and setting. We carried out a prospective multicenter study to evaluate the feasibility and diagnostic accuracy of the Tetanos Quick Stick performed in the emergency departments of 30 French hospitals.

Eighty-two of the 550 emergency departments contacted by the French Emergency Sentinel Network agreed to participate. We selected participating centers from these 82 centers based on the following criteria: availability of a participating laboratory at or near the center and written consent of the emergency department and the laboratory for participation in the study. We selected centers so that the ratio of university to general hospitals was 1 to 3 and all types of emergency departments were represented (e.g., urban and rural).

Selection of participants. All patients over 18 years old referred to the emergency room for open wounds within 24 h of injury were prospectively included in the study after giving written informed consent and receiving appropriate treatment. Pregnant women, patients with severe injuries requiring immediate care or surgery, and patients with acute psychiatric problems requiring a special hospitalization procedure were excluded.

We wanted to recruit at least 30 centers, each including 30 patients. These numbers were considered sufficient to ensure an appropriate trade-off between representativeness of patients (largest number of patients per site) and operator multiplicity (which increases with the number of sites). Emergency physicians and biologists attended a national meeting, and they also were trained on site in the use of the TQS.

Diagnostic tests: TQS and ELISA. The Tetanos Quick Stick test was performed with blood obtained by finger prick by an attending physician or nurse from the emergency department (blood/emergency-TQS), and another blood sample was simultaneously sent to the hospital laboratory. The staffs of the emergency departments were trained in the use of the TQS.

The TQS method uses a combination of a solid phase coated with tetanus toxoid and tetanus toxoid-dye conjugate (colloidal gold). The operator takes a whole-blood sample from the patient's fingertip by means of a small pipette provided in the test kit. This sample is placed in the test sample well, to which diluent is added. The diluent flows through the absorbent pad, carrying the toxoid-dye conjugate along the chromatographic strip. A complex with antitetanus antibodies forms if such antibodies are present in the blood sample. These complexes react with the immobilized toxoid to form a pink line in the "T" (test) window. In the absence of anti-tetanus antibodies, no line appears in the test window. The excess gold conjugate binds to a control reagent immobilized in the "C" (control) window, forming a pink line, indicating that the test has been carried out correctly. The test should be read 20 min after diluent is added to the blood sample in the device.

Within 6 hours of the blood/emergency-TQS, the participating center laboratory carried out a second TQS test on a total blood sample (blood/lab-TQS) and a serum sample (serum/lab-TQS). The reference laboratory (Laboratory of Bacteriology and Virology, Lariboisière Hospital, Paris, France) performed ELISA (quantitative detection of immunoglobulin G after binding on a microplate coated with tetanus toxoid and calibration with the World Health Organization standard [Genzyme-Virotech GmbH, Russelsheim, Germany; distributed by InGen, Rungis, France]), according to the manufacturer's recommendations, for all included patients. ELISA, blood/lab-TQS, and serum/lab-TQS tests were performed blind to the results of the blood/emergency-TQS.

Data collection and processing. After inclusion, a structured clinical history was taken with (i) the main sociodemographic characteristics, (ii) a description of the wound and its risk for tetanus contamination (particularly contact with soil), and (iii) vaccination status for tetanus (items on the questionnaire are detailed in Table 1). Physicians were asked to perform the TQS test after appropriate care had been given and any prescription made, to ensure that they did not take the result of the test into account when deciding on their tetanus prevention strategy. They were encouraged to define the strategy used according to the algorithm recommended by the French Ministry of Health: (i) vaccine and immunoglobulin, in cases of no vaccination or doubts about vaccination history or severe wounds with incomplete or out-of-date (>10-year-old) vaccination; (ii) vaccine alone for minor wounds with incomplete or out-of-date (>10-year-old) vaccination or severe wounds with vaccination 5 to 10 years before; and (iii) neither vaccine nor immunoglobulin in cases of more recent vaccination (less than 5 years before for severe wounds and less than 10 years before for minor wounds) (10). The prescriptions made by each physician were recorded on the standardized form used to record the clinical history.

View this table: [in this window] [in a new window]

TABLE 1. Main characteristics of patients and description of wounds

The feasibility of the TQS test was documented on two forms. The first concerned the procedure and result of the test (required for addition of the sample to the device, use of the pipette provided, time until reading, presence of a control line, and interpretation time) as performed at the emergency department and the laboratory. The form was filled out by the operator who performed the test for each included patient. The operator reported the TQS as positive if both the "C" and "T" lines were clearly apparent, as negative if the "C" line appeared but not the "T" line, and as inconclusive if the "C" line was not apparent or if the operator was unsure whether there was a line in either the "T" or the "C" window.

The second form (feasibility form) was filled in by each operator of the emergency department at the end of the study. Operators scored the feasibility of the test in an emergency context (with respect to time delays, simplicity, and interpretation) on a scale of 1 to 8 at the end of the study.

The results of the TQS tests were reported by the operators (a physician or nurse in the emergency department and a biologist or technician at the laboratory) on standardized forms, including details concerning the performance of the test.

Statistical analyses. The likelihood that the agreement between emergency department and laboratory TQS tests was not due to chance alone was assessed by means of the kappa test.

The accuracy of vaccination status diagnosis was assessed by determining the sensitivity, specificity, and positive likelihood ratio, with 95% confidence intervals (CI), for blood/emergency-TQS, blood/lab-TQS, and serum/lab-TQS and for status reported in the clinical history. We defined sensitivity as the proportion of subjects protected against tetanus according to the ELISA test with positive TQS results. We defined specificity as the proportion of unprotected subjects (on the basis of ELISA) with negative or inconclusive TQS tests. The positive likelihood ratio was calculated for any possible result i of the test (positive, inconclusive, or negative) as the proportion of patients with result i in protected patients divided by the proportion of result i in unprotected patients.

ELISA was considered to be the gold standard. We used a threshold of 0.1 IU/ml of serum to define positive results in ELISA and serum/lab-TQS, which implied the use of a threshold of 0.2 IU/ml of blood in the blood/emergency-TQS and blood/lab-TQS tests.

We used McNemar's tests (1 degree of freedom) to compare the sensitivities and specificities of blood/emergency-TQS and vaccine status reported in the clinical history.

Ethical aspects. The patient was required to give informed consent for inclusion. The therapeutic strategy was not modified by the result of the test, but the patient was provided with information and advice concerning his or her vaccination status based on the results of the various tests. The study was approved by the local ethics committee and was considered to correspond to research with direct individual benefit.

Two of the authors (G.C. and I.C.) were responsible for data input and statistical analysis.

RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

View larger version (27K): [in this window] [in a new window]

FIG. 1. Flowchart of the study.

Main results. The blood/emergency-TQS test was inconclusive in 97 (9.8%) patients, whereas the blood/lab-TQS test was inconclusive in only 38 patients (3.8%). We compared the conditions under which the test was performed in the emergency department and the laboratory (Table 2). The main differences were the grade of the operator (physician versus nurse at the emergency department or biologist versus technician at the laboratory), use of the disposable pipette provided with the test, and time between performance and reading. The mean scores (and standard deviations) awarded by operators in the satisfaction survey (scale of 1 to 8) were 5.7 (±1.7) for ease of blood sampling from the fingertip, 4.9 (±1.8) for ease of use of the disposable pipette, 6.7 (±1.3) for the overall simplicity of the test, and 7.3 (±1.1) for the simplicity of the test in an emergency context.

View this table: [in this window] [in a new window]

TABLE 2. Conditions of reading and interpretation of TQS in the emergency room and in the laboratory

Agreement between the results obtained was excellent for the blood/lab-TQS and serum/lab-TQS (0.912; 95% CI, 0.910 to 0.914) and moderate for blood/lab-TQS and blood/emergency-TQS (0.510; 95% CI, 0.508 to 0.511). Agreement did not depend on the qualification of the operator (0.513; 95% CI, 0.512 to 0.515 for physicians and 0.526; 95% CI, 0.519 to 0.533 for nurses) or the time at which the test was performed (0.497; 95% CI, 0.479 to 0.515 for nights and weekends and 0.513; 95% CI, 0.512 to 0.515, for weekdays between 8 a.m. and 6 p.m.).

Results and likelihood ratios for TQS tests performed in the emergency department and in the laboratory are presented in Table 3, together with the vaccine status reported in the patient's clinical history. If we compared positive results with pooled inconclusive and negative results, diagnostic accuracy was higher for the blood/emergency-TQS (69% [95% CI, 66 to 72] and 98% [95% CI, 90 to 99], respectively, for sensitivity and specificity) than for clinical history (62% [95% CI, 59 to 65] and 79% [95% CI, 67 to 88]). This difference was highly significant (P < 0.001). The sensitivity and specificity of the blood/lab-TQS test were 84% (95% CI, 82 to 86) and 99% (95% CI, 92 to 100), respectively. The sensitivity and specificity of the serum/lab-TQS test were 86% (95% CI, 83 to 88) and 99% (95% CI, 92 to 100), respectively. The median (interquartile range) ELISA antibody titers for ELISA-positive and blood/emergency-TQS-negative and ELISA-positive and blood/emergency-TQS-positive patients were 0.7 (1.3) and 3.0 (3.3), respectively.

View this table: [in this window] [in a new window]

TABLE 3. Diagnostic accuracy of vaccination history and TQS test in the emergency department and in the laboratory, with ELISA as reference

Prescriptions were recorded for 969 patients, three of whom could not have the ELISA test (Fig. 1): 914 (95%) were protected according to ELISA, and 52 were not. Immunoglobulin was unnecessarily given to 120 (13.1%) protected patients. If the blood/emergency-TQS had been used for decision making, this prescription could have been avoided in 23 patients for whom positive results were obtained in this test. Eight unprotected patients (15%) received no treatment. If TQS had been used for decision making, this error would have been avoided in all these patients (all tested negative in blood/emergency-TQS).

DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

Our study was subject to certain limitations. The reference test was not performed for three patients examined at one of the centers in the study. The other 27 patients from this center were retained in the final analysis, and their exclusion from analysis had no effect on the performance results for the blood/emergency-TQS. Four centers included less than 15 patients, and at one of these centers, the physician was less compliant with study procedures than staff elsewhere. This may account for the false-positive blood/emergency-TQS result obtained at that center.

Moderate agreement was observed between the blood/emergency-TQS and blood/lab-TQS results, probably due to the proportion of inconclusive tests, which was higher for TQS tests performed in the emergency department (9.8%) than for those performed in the laboratory (3.9%). Moreover, in a few cases, the emergency department operator reported the TQS test to be positive, despite being unsure or not seeing a pink strip in the test zone. Laboratory operators gave no such incoherent interpretations. This finding identifies several technical characteristics of the test, which may limit its feasibility in the emergency department: (i) the use of the disposable pipette may have resulted in false-negative results due to the presence of insufficient blood in the device; (ii) difficulties reading the control line were also reported by a few emergency department operators, representing 23 tests; (iii) compliance with the 20-minute interval until reading was less stringent in the emergency department than in the laboratory; and (iv) 76% of tests were performed by emergency department physicians, whereas routine tests in these departments (i.e., urinary or glucose tests) are performed by nurses well trained for these tasks. This suggests simple improvements that could be made to the test presentation: improvement of devices, such as the pipette or the lancet, and automation of reading. A quality assurance process should be defined for implementation of the TQS test in practice, taking into account all these difficulties. This should help to decrease the number of inconclusive test results and to improve diagnostic accuracy. Close collaboration between clinicians and biologists is required for the selection of tests and for evaluation of test performance, for the training of staff and evaluation of their technical skills, for following results, and for setup and implementation of quality assurance procedures.

Almost 88.5% of the patients selected in this study reported a vaccination status that was considered by the physician to be reliable. This proportion may be lower in the real population in emergency departments. The selection criteria may also account for the high prevalence of people with protection against tetanus (as assessed by ELISA). Although this should not affect the performance of the test in terms of sensitivity, specificity, and likelihood ratios, it may result in underestimation of the impact of the blood/emergency-TQS test in preventing the inappropriate prescription of immunoglobulin.

The concordance between blood/emergency-TQS and blood/lab-TQS results was moderate, illustrating the feasibility and limitations of decentralized biological tests. The different thresholds used to interpret blood/emergency-TQS and ELISA tests may also partly account for the number of false negatives obtained with the TQS. However, the diagnostic accuracy of blood/emergency-TQS is much higher than that of the structured clinical history taken by the attending physician. Subject to a medicoeconomic study, we therefore recommend the use in France of a new strategy for evaluating serum tetanus antitoxin levels in patients referred to emergency departments with open wounds, based on use of the Tetanos Quick Stick.

Two factors determine the risk of tetanus in patients with open wounds: immunization status and exposure risk, assessed on the basis of the wound description and the circumstances in which the wound occurred. Various strategies are therefore possible for improving the efficiency of tetanus prevention in emergency departments. The first involves investing resources in a fully implemented, regular 10-year booster program for all adults in the general population. This vaccination policy is currently recommended by the Centers for Disease Control and Prevention. The cost-effectiveness of this strategy has been analyzed in the United States (2) and criticized in the United Kingdom (4, 11). The second approach involves appropriate prevention in the case of a wound. This system is based on cleaning of the wound and on the diagnosis of immunization status. This diagnosis is often limited and can be improved by the use of rapid diagnostic tests in the emergency department. Such decentralized biological testing is of particular value if a large proportion of patients with wounds are seen during the weekend, in the evening, or during the night (8.6% in our study). Another diagnostic test has been compared with ELISA for use in emergency practice, but it was not retained for routine use because of its poor feasibility (3). Once a test is demonstrated to be valid in this setting, it is necessary to consider the impact of its use on medical prescriptions, its cost, and its acceptance by both physicians and biologists before it can be implemented in practice.

In summary, the high performance of the TQS test performed in the laboratory obtained in this multicenter study confirms the interest of this test over the usual clinical-history taking. Its use in emergency departments should not be considered without a quality assurance process and close collaboration between biologists and clinicians. We plan to complete our study with a cost-effectiveness study to compare the different strategies for tetanus prophylaxis treatment for wounds: test all, or test only those who are not protected. Since the French algorithm is not "appropriate" to identify tetanus-prone wounds, it will be necessary to change it.

ACKNOWLEDGMENTS

We thank the InGen Laboratory, which provided a grant for data management and for the remuneration of a clinical research assistant.

We thank the reference laboratory (Laboratory of Bacteriology and Virology, Lariboisière University Hospital), particularly Nicole Cantaloube, who performed the reference tests. We thank all the investigators of the French Emergency Sentinel Network (Réseau Sentinelle Urgence), one physician in the emergency department and a biologist in the laboratory of each of the 37 participating centers: Bernard Clere and Antoine Defrance (CH Montdidier), Michel Maignan and Michèle Weber (CHU, Nancy), Charles Jeleff and Bruno Darchis (CH Compiègne), Corinne Bergeron and Alain Dublanchet (CH Villeneuve St. George), Cécile Bracq and Alain Savage (CH Dunkerque), Patrick Werner and Fredéric Bert (CHU Clichy), François Lecomte and Matthieu Bernier (CHU Cochin), David Elkharrat and Nicole Cantaloube (CHU Lariboisière), Christian Alba and Catherine Alba-Sauviat (CH Chaumont), Elisabeth Pezé and Catherine Bouquigny (CH Soissons), Hassan Awada and Jean-Philippe Rouimi (CH Denain), Pierre Denonce and Marie-Pierre Jaby (CH St. Denis), Marc Alazzia and Catherine Farnarier (CHU Marseille), Frederic Pochet and Pierre Rambaux (CH Roubaix), Cécile Douchet and François Marcolli (CH Arras), Christoph Kaletka and Pierre Moritz (CH Lure), Marc Prevel and Xavier Anglaret (CH Rueil Malmaison), Christophe Perrier and Mireille Romaszosko (CHU Hôtel Dieu, Clermont Ferrand), Xavier Roy and Laure Rebeyrotte (CH Chateauroux), Emmanuel Perreira and Martine Harzic (CHG Le Chesnay), Daniel Juvin and Guy Chambreuil (CH La Roche/Yon), Laurent Boidin and Michel Chelle (CH Fontenay Le Comte), Sophie Gonnet and Marie-France Marchal (CH Annemasse), Dominique Sabot and Jean-Louis Stach (CH Gueret), Daniel Le Bras and Hélène Senechal (CH Quimperle), Christophe Savio and Edouard Bichier (CH Saumur), Caroline Fritsch and Jacques Capdeville (CH Neufchateau), Habib Hachimi and Marc Rieffel (CH Selestat), Anne Marie Zix Minni and Michel Minery (CH Saverne), Jean Rottner and Andre Trevoux (CH Mulhouse), Herve Riviere and Claude Grasmick (CH Cahors), Azouz Dbedioui and Abel Mars (CH Decazeville), Jean Bouthier, Jean Bonneau, and Alain Capbern (CH Libourne), Martine Bisset and Michel Brun (CHU Montpellier), Jean-Philippe Brettes and Dominique Mas (CH Auch), Laurence Abadie and Nicole Constantin (CH Lourdes), and Raffik Masmoudi, Anthony Scemana, and Serge Houssaye (CHU HEGP). The InGen laboratory provided all resources for functional needs (TQS test material, blood sampling material, and material for the transport of samples).

FOOTNOTES

REFERENCES

Top Abstract Introduction Materials and Methods Results Discussion References