A Survey of Saskatchewan Physicians

Health Policy and Practice / Santé : politique et pratique médicale

Chest X-Ray Ordering Related to Varied Clinical Scenarios: A Survey of Saskatchewan Physicians Brent Burbridge, MD, FRCPC; David Douglas, BSc; Stefan Kriegler, MBChB, MMRadD Abstract Introduction: The chest X-ray (CXR) is one of the most commonly requested diagnostic imaging examinations. It is estimated that over 250,000 CXRs are ordered in Saskatchewan annually. Judicious use of the CXR is valuable in patient care, while unnecessary use increases costs, adds to patient irradiation, and may be in conflict with standards of patient care. In 1993,the Saskatchewan Health Services Utilization and Research Commission (HSURC) developed clinical practice guidelines (CPGs) for the CXR. These guidelines were based on a metaanalysis of validated literature. We were uncertain about whether Saskatchewan physicians were ordering CXRs based on the HSURC CXR CPGs. Materials and Methods: A survey, based on recommendations from the HSURC CXR CPGs, was developed and distributed to 363 physicians in Saskatchewan by mail (30/363) and by email (333/363). The survey asked physicians if they would order, or not order, a CXR for 5 basic clinical scenarios. The question of whether to order a CXR was then repeated when the basic scenario was altered one variable at a time, to determine if the physicians would change their CXR orders. Results: According to our assessment of physician responses to the initial clinical scenarios, the surveyed physicians correctly followed the HSURC CXR CPGs in the following frequencies: 100% scenario 1, 91.9% scenario 2, 35.4% scenario 3, 100% scenario 4, and 61.2% scenario 5. Alteration of the basic clinical scenarios resulted in very unpredictable ordering of CXRs by the survey participants.

Burbridge, Kriegler—Department of Medical Imaging, Royal University Hospital Douglas—College of Medicine, Health Sciences Building, University of Saskatchewan, Saskatoon, SK Address for correspondence: Dr B Burbridge, Medical Imaging, Royal University Hospital, 103 Hospital Dr Saskatoon, SK S7N 0W8; [email protected] Submitted Mar 28, 2005 Accepted June 30, 2005 ©2005 Canadian Association of Radiologists Can Assoc Radiol J 2005;56(4):219–224.

Conclusion: The Saskatchewan physicians we surveyed are not ordering CXRs on the basis of HSURC CXR CPGs. They order too many nonindicated CXRs. Further communication with, and education of, the physician population about the HSRUC CXR CPGs may be warranted.

Abrégé Introduction : La radiographie pulmonaire (RXP) est un des examens d’imagerie diagnostiques le plus souvent demandé. On estime que plus de 250 000 RXP sont prescrites en Saskatchewan, par année. L’emploi judicieux des RXP est valable dans le soin des patients, alors que l’usage inutile accroît les coûts, ajoute à l’irradiation des patients, et peut entrer en conflit avec les normes des soins du patient. En 1993, la commission sur la recherche et l’utilisation des services de santé de la Saskatchewan (HSURC) a élaboré des lignes directrices de la pratique clinique (LDPC) pour les RXP. Ces lignes directrices se fondent sur une méta-analyse de la documentation validée. Nous n’étions pas certains que les médecins de la Saskatchewan demandaient des RXP conformément aux LDPC sur les RXP de la HSURC. Matériel et méthodes : Un sondage fondé sur les recommandations des LDPC sur les RXP de la HSURC a été mis au point et distribué à 363 médecins de la Saskatchewan par la poste (30/363) et par courriel (333/363). Le sondage demandait aux médecins s’ils prescriraient ou non une RXP dans 5 scénarios cliniques de base. La question de demander une RXP ou non était ensuite répétée avec un scénario modifié, une variable à la fois, pour déterminer si les médecins changeraient d’avis. Résultats : Selon notre évaluation des réponses des médecins aux premiers scénarios cliniques, les médecins interrogés ont suivi adéquatement les LDPC sur les RXP de la

CARJ Vol 56, No 4, October 2005 219

HEALTH POLICY AND PRACTICE / SANTÉ : POLITIQUE ET PRATIQUE MÉDICALE

HSURC aux niveaux suivants : 100 % au scénario 1; 91,9 % au scénario 2; 35,4 % au scénario 3; 100 % au scénario 4; et 61,2 % au scénario 5. La modification des scénarios cliniques de base a entraîné des prescriptions de RXP très imprévisibles par les participants au sondage. Conclusion : Les médecins de la Saskatchewan que nous avons interrogés ne prescrivent pas les RXP conformément aux LDPC sur les RXP de la HSURC. Ils demandent trop de RXP qui ne sont pas indiquées. Il serait judicieux de communiquer davantage avec la population des médecins au sujet des LDPC sur les RXP de la HSURC, et de les former à ce propos.

chest X-ray (CXR) is one of the most commonly ordered Thediagnostic imaging examinations. It is estimated that $8 million is spent annually in Saskatchewan for more than 250 000 CXRs.1 CXRs are commonly ordered for preoperative screening, on admission to hospital, as part of periodic health examination and for investigation of clinical signs and symptoms. Judicious use of CXRs is valuable in patient care, while unnecessary use increases costs, adds to patient irradiation, and may be in conflict with standards of patient care.2 In 1993, the Saskatchewan Health Services Utilization and Research Commission (HSURC) developed clinical practice guidelines (CPGs) for the CXR. These guidelines were based on an extensive review of the medical literature and developed by an expert panel using the best evidence available.1 These CPGs were reviewed by an HSURC expert panel again in 2001, and were revalidated without change. The HSURC CXR CPGs were extensively communicated to Saskatchewan physicians via news releases, Web site publication, email, and ground mail in 1993 and again in 2001, when the guidelines were revalidated. CPGs should be expected to facilitate appropriate physician use of testing.3,4 We were uncertain whether physicians practicing in Saskatchewan were applying the HSURC CXR CPGs in their practice of medicine. The purpose of our investigation was to determine whether Saskatchewan physicians were making CXR requests on the basis of HSURC CXR CPGs. We developed a survey based on those CPGs to determine whether CXR ordering practices were in accordance with these CPGs.

Materials and Methods From the HSURC CXR CPGs, we ascertained the following parameters related to appropriate CXR requisition: 1) Age alone is not an indication for a CXR. 2) Smoking, in and of itself, is not an indication for a CXR. Only when diseases from smoking are symptomatic does a CXR alter patient management or outcome. 3) In most instances, suspected malignancy, tuberculosis, and other active chest diseases are indications for a CXR. 4) CXRs should also be ordered if there are clinical indications of acute asthma or asthma exacerbation. 5) Chronic airways limitation (CAL), or chronic obstructive pulmonary 220 JACR vol 56, no 4, octobre 2005

disease (COPD), is a reason to order films if it is suspected as a new diagnosis or if an exacerbation of symptoms is evident. Otherwise, for CAL, a CXR should be ordered only every 1 to 2 years to radiographically document the progression of the disease. 6) Patients with human immunodeficiency virus (HIV) and acquired immune deficiency syndrome (AIDS) or those who are otherwise immunocompromised are at risk for chest infections. The HSURC CPGs suggest that a CXR be ordered on admission to hospital if patients in these groups have a chronic cough (> 4 weeks), unexplained fever (> 1 week), or signs or symptoms of an infection unresponsive to treatment. 7) Planned cardiopulmonary surgery is the only indication for a preoperative CXR in the absence of clinical signs or symptoms. Routine preoperative chest radiography in asymptomatic individuals is not warranted for other patient groups. 8) An acute cough (< 4 weeks) is not an indication for a CXR, unless it is felt to represent an exacerbation of a known underlying chest disease. Chronic cough (> 4 weeks) is likely an indication if the cause is unknown. 9) Signs and symptoms of pneumonia require a CXR for diagnosis. 10) In the emergency room, the CXR is routinely indicated only for serious trauma patients; any other CXR requests should be based on clinical signs and symptoms.1 We created 5 basic clinical scenarios based on the HSURC CXR CPGs. We then asked the respondents whether a CXR should be ordered, based on their review of these clinical scenarios.1 The 5 basic clinical scenarios presented to the physicians are provided in Appendix 1 (see page 224). After responding to each initial clinical scenario, the survey participants were asked whether they would order a CXR if the following factors in each of the scenarios were altered, one at a time, while leaving the other parameters of the scenario unchanged. The respondents were expected to answer “yes” (order CXR) or “no” (do not order CXR). The altered variables consisted of patient age (16, 50, and 80 years); symptoms suggestive of asthma, pneumonia, or pneumothorax; history of smoking (5, 10, 20, or 40 packs yearly); signs and symptoms of chronic airflow limitation (CAL); immunocompromised state or HIV/AIDS; chronic cough (> 4 weeks’ duration); and acute cough (< 4 weeks’ duration). If a CXR was ordered on the basis of the original scenario, it was expected that this decision would not be changed when altering the variables as described. For scenario 2, 4, and 5, it was expected that the physician should order a CXR on the basis of the original case presentation. Thus to be congruent with this decision, the respondent should order a CXR regardless of whether the stated variables were altered. The expected appropriate responses to the original clinical scenarios and to the altered variables are presented in Table 1. The clinical scenarios were presented to participants in the form of a survey. The survey was distributed to 363 Saskatchewan physicians. Those asked to participate consisted of all

Chest X-Ray Ordering Related to Varied Clinical Scenarios: A Survey of Saskatchewan Physicians

Table 1 Original CXR order decision compared with CXR order to be changed based on altered case variable Case 1

Case 2

Case 3

Case 4

Case 5

Original Scenario

No

Yes

No

Yes

Yes

Age 16 years

NC

NC

NC

NC

NC

Age 50 years

NC

NC

NC

NC

NC

Age 80 years

NC

NC

NC

NC

NC

Symptoms of acute asthma

Yes

NC

Yes

NC

NC

Symptoms of acute pneumonia

Yes

NC

Yes

NC

NC

Symptoms of acute pneumothorax

Yes

NC

Yes

NC

NC

Smoker 5 packs yearly

NC

NC

NC

NC

NC


NC

NC

NC

NC

NC


NC

NC

NC

NC

NC


NC

NC

NC

NC

NC

Symptoms of CAL/COPD

NC

NC

NC

NC

NC

Immunocompromised State

NC

NC

NC

NC

NC

Acute cough < 4 weeks

NC

NC

NC

NC

NC

Chronic cough > 4 weeks

Yes

NC

NC

NC

NC

NC = no expected change, Yes = ordering CXR is the expected change.

those physicians with valid email addresses registered with the Saskatchewan College of Physicians and Surgeons. Email messages were sent inviting 333 physicians to participate in our survey. In addition, 30 physicians without email addresses, from our local hospital, were sent stamped return envelopes with a paper copy of the survey. Email recipients responded to the survey by accessing an online survey site and completing the online questionnaire. The online and ground mail surveys were identical in structure and content. Surveys returned by ground mail were tabulated and entered into the database by one of the authors. The evaluation of clinical decision making necessitated the use of bivariate tabular analysis (2 ´ 2 table) and the Pearson’s chi-square test (c2). The assignment of a p value of < 0.05 reflects a statistically significant difference for the c2 analysis.5 Statistical analysis was performed based on case study variables and some respondent-specific characteristics. Responses were evaluated to determine which case study variables prompted a clinical decision different from the one based on the original case study. The variables analyzed were patient age (16, 50, and 80 years); symptoms suggestive of asthma, pneumonia, or pneumothorax; history of smoking (5, 10, 20, or 40 packs yearly); the presence of CAL; HIV/AIDS or other

immunocompromised state; and chronic or acute cough (greater or less than 4 weeks, respectively). Responses to all sections of the survey completed were entered into a computerized database. The Statistical Package for Social Sciences (SPSS) (SPSS Inc. Chicago, IL) was used to perform all statistical analyses. Responses marked “n/a” and skipped questions were not included in the statistical analysis.

Results Of the 363 physicians surveyed, 70 (19.3%) responded. Responses were returned predominantly online (61/70), while 9 were submitted by conventional mail. Of the responding physicians, 15.7% were 25–34 years old, 24.3% were 35–44 years old, 40% were 45–54 years old, 12.9% were 55–64 years old, and 5.7% were 65 years and older. Respondents had the following number of years in practice: 32.9% with less than 10 years, 27.1% with 10–19 years, 25.7% with 20–29 years, and 12.9% with more than 30 years (1.4% did not respond). Of these respondents, 65.7% were male and 32.9% were female (1 respondent did not specify). CARJ Vol 56, No 4, October 2005 221


Table 2 Demographics of the surveyed physicians compared with the general physician population of Saskatchewan Category

Surveyed population (%)

General physician population (%)

Urban

75.5

61

Rural

24.3

39

Male

66.7

75

Female

33.3

25

General practitioner

31.4

63.7

Specialist

68.6

36.3

The practice locations of physicians surveyed were Saskatoon (48.6%), Regina (27.1%), Moose Jaw (4.3%), Prince Albert (5.7%), Swift Current (2.9%), Yorkton (1.4%), Lloydminster (1.4%), and other locations (8.6%). Locations were then grouped into urban (Saskatoon and Regina, 75.7%) or rural (all others, 24.3%) categories. Responding physician areas of medical expertise were surgery (21.4%), anesthesia (12.9%), family medicine (31.4%), internal medicine (8.6%), radiology (4.3%), emergency medicine (2.9%), obstetrics and gynecology (4.3%), and other (14.3%). Of those who responded, 62.1% reported ordering more than 20 CXRs yearly, while 37.9% ordered fewer than 20 CXRs yearly. For the first basic scenario, none of the responding physicians ordered a CXR. This was the expected clinical behaviour based on the given history and physical examination data and the HSURC CXR CPGs. When asked whether they would order a CXR when the variables were altered, inappropriate CXR ordering was noted for age 50 years (14.3%) and age 80 years (42.9%); smoking 5 (14.1%), 10 (71.9%), 20 (57.8%), and 40 (68.8%) packs yearly; CAL (82%); immunocompromised state (HIV/AIDS) (87.5%); and acute (69.8%) and chronic cough (82.8%). The p values for all of these variables were < 0.01. The second scenario saw most of the physicians order a CXR. This was the expected decision according to the HSURC CXR CPGs. Three variables caused respondents to order significantly fewer CXRs, specifically, age 16 and smoking 5 and 10 packs yearly. The statistical analysis revealed p values for these variables of < 0.01. For the third scenario, 74.1% of the physicians chose to order a CXR. These CXRs are unnecessary based on the HSURC CXR CPGs. There were no statistically significant alterations in this ordering trend for the individually altered variable. 222 JACR vol 56, no 4, octobre 2005

For scenario 4, 100% of those physicians who responded chose to order a CXR. This was the appropriate choice based on the HSURC CXR CPGs. There were no significant changes in ordering when the variables were altered. For scenario 5, only 61.2% of physicians ordered a CXR. Based on the HSURC CXR CPGs, a CXR was appropriate. The following variables caused a significantly greater number of CXRs to be ordered compared with the initial response to the basic clinical scenario: smoking 20 or 40 packs yearly, CAL, HIV/AIDS, and acute and chronic cough. The p values for these variables varied from 0.05 to 0.01 based on statistical analyses.

Discussion A search of the literature (Medline and Ovid) failed to reveal any previous research evaluating physician adherence to CXR CPGs related to clinical scenario analysis. We felt that a survey was appropriate. However, surveys have some inherent disadvantages. They are subject to content, coverage, sampling, nonresponse, and measurement bias.6 Attempts were made to minimize bias by having a medical student construct and administer the survey after seeking and receiving input from a wide variety of radiologists and clinicians in the local medical community about the survey content and context. The order and wording of survey questions is known to possibly alter respondent choices.7 We attempted to minimize this by having a uniform pattern for presentation of the original clinical scenarios, with the same sequence of variable alteration for each scenario. A further potential limitation of our study may lie in the artificial nature of responses to a survey versus real-life clinical situations. Certainly, physicians may behave differently when dealing with a real patient. However, we are not aware of a mechanism to adjust for this possible discrepancy.

Chest X-Ray Ordering Related to Varied Clinical Scenarios: A Survey of Saskatchewan Physicians

An analysis of the demographic distribution of our sample population revealed a greater proportion of urban dwellers, men, and physicians with specialty training, compared with the general physician population of Saskatchewan. No other disparities were noted. The demographic data of the surveyed group and the general population of Saskatchewan physicians are presented in Table 2.

Evaluating whether physicians have adopted CPGs is an excellent gauge of the influence these guidelines have on health care decision making. HSURC communicated the CXR CPGs via mail directly to physicians and email to the medical community, as well as via Web site publication of guidelines. Substantial time and effort was expended to develop and communicate these CXR CPGs.

Physicians were asked to participate in the survey if they had a current email address registered with the College of Physicians and Surgeons of Saskatchewan. It is possible that this sample method favoured physicians who have more computer savvy.6

Identifying lack of adherence to CPGs allows for the development of targeted physician education to highlight the discrepancies between CPGs and actual clinical practice and inspire behaviour modification among physicians. Discussion surrounding CPGs also allows for reassessment of the guidelines themselves to determine whether revision is necessary to adapt to new clinical paradigms.

Based on initial scenario assessment, surveyed physicians correctly followed the HSURC CXR CPGs in the following frequencies: scenario 1 (no CXR), 100%; scenario 2 (order CXR), 91.9%; scenario 3 (no CXR), 35.42%, scenario 4 (order CXR), 100%; and scenario 5 (order CXR), 61.2%. For the first scenario, most of the subsequently presented case variables resulted in a significant number of physicians inappropriately ordering CXRs when they were not indicated. In the second case scenario, a statistically significant inappropriate reduction in requests for CXRs was detected when the patient was 16 years old or had a history of smoking 5 to 10 packs yearly. The third scenario saw most (74.1%) physicians choosing to order a CXR when the HSURC CPGs suggest it is unnecessary. This scenario is unique as it featured the only emergency case, the youngest patient, and debatably, a somewhat borderline indication for ordering a CXR. This possibly suggests that a scenario with a borderline indication leads physicians to order more CXRs. The fourth scenario resulted in most physicians appropriately ordering the CXR. For the last scenario, only 61.2% of physicians ordered a CXR based on the original scenario of preoperative work-up of a patient with a significant cardiac history and physical findings of cardiac disease. The scenario obviously described a significant cardiac history and positive physical findings of cardiac disease. Subsequent presentation of the individual variables revealed inappropriate CXR ordering for several variables thereafter. The appropriate application of the HSURC CXR CPGs to the initial case scenarios varied considerably, from 35.42% to 100%. The general tendency was for physicians to order CXRs more frequently than the HSURC CPGs would suggest. It is too extreme to suggest that the “rightness” of a clinical decision depends on strict adherence to CPGs. But CPGs do provide a reference point from which clinical decisions can be evaluated.

Although certain surveyed physician groups did demonstrate some ordering trends, there was no apparent correlation to the HSURC CXR CPGs. Physicians surveyed also ordered more CXRs than suggested by the HSURC CXR CPGs. These discordances suggest that decisions about CXR ordering are not based on the HSURC CPGs and that the surveyed Saskatchewan physician population may not be aware of them. The large number of CXRs ordered in Saskatchewan, and the associated cost, is a substantial burden for the Saskatchewan health care system. A certain proportion of these films will be unnecessary, resulting in an avoidable financial burden to the health care system. Additionally, these patients are exposed to unnecessary radiation. Promotion of the HSURC CXR CPGs could result in more appropriate CXR use, as well as facilitate the adoption of evidence-based health care decision making in Saskatchewan.

References 1. Health Services Utilization and Research Commission (HSURC). HSURC research studies. Chest radiography. Clinical practice guidelines. October 1997. Recertified January 2001. Available at: www.hqc.sk.ca/portal.jsp?wV5A6AhxQOnp6+UFLmREsjBIzBf0QfLQ kUwK4QBZaJstu25CyXW5NL/MnXMt45mxAaqQvnmclkZ2oGJ9Vwx sE5yY6NobxLVB. Accessed 2005 June 1. 2. Canadian Association of Radiology. Practice guidelines for chest x-ray. 1994. Revised 2000. Available at: www.car.ca/ethics/standards/index.html. Accessed 2005 June 1. 3. Meiring PDV, Wells IP. The effect of radiology guidelines for general practitioners in Plymouth. Clin Radiol 1990;42:327–9. 4. Influence of Royal College of Radiologists’ Guidelines on Referral from General Practice. Royal College of Radiologists Working Party. BMJ 1993;306:110–1. 5. Rohlf FJ, Sokal RR. Biometry. 3rd ed. New York: W.H. Freeman and Company; 1995. 6. Grandcolas U, Rettie R, Marusenko K. Web survey bias: sample or mode effect? Journal of Marketing Management 2003;19:541–61. 7. Groves RM. Survey errors and survey costs. New York: John Wiley and Sons; 1989.

CARJ Vol 56, No 4, October 2005 223


Appendix 1 Case 1

Case 4

A 34-year-old auto mechanic was admitted to undergo an inguinal hernia repair. The patient had had no previous surgeries or hospitalizations. There was no significant family history. Physical examination was unremarkable. The patient weighed 60 kg, heart rate was 70 beats per minute, respirations were 15 per minute, and blood pressure was 120/80 mm Hg.

A 28-year-old printer was admitted to hospital severely ill. Eight days previously, he had developed a painful swelling in his left axilla, for which he had been prescribed ampicillin. This had continued to increase in size over 6 days, during which time, he had became progressively more unwell with fever, anorexia, and vomiting. On questioning, he thought he had passed less urine than normal and that it had appeared concentrated. Two days before admission, he had developed dyspnea on climbing stairs. There was no significant past medical history, although he had been feeling more easily fatigued in the previous 2 months.

The expected decision for this scenario was not to order a CXR, because the patient was asymptomatic and booked for noncardiac elective surgery.

Case 2 A 47-year-old housewife was being admitted to hospital for further assessment. She felt lethargic. She noted a diminished appetite and weight loss of 8 kg over the last 6 months. She also reported having a persistent dry mouth. She does not smoke, drinks alcohol occasionally, and is not taking any medication. She also complained of generalized pruritus for 1 year. This had started insidiously but had recently become troublesome, keeping her awake at night. No other members of her family had experienced itching. On examination, she looked well and was not anemic, jaundiced, or cyanosed. No evidence of clubbing was noted. There was an enlarged lymph node in her left axilla. There were scattered minor excoriations but no other skin lesions. Cardiovascular and respiratory systems were normal. Her liver was palpable 2 finger breadths below the costal margin. No other abdominal masses were palpable, and the spleen could not be felt. Urinalysis and routine blood tests were negative. The expected decision was that the physician would order a CXR on admission, based on the suspicion of underlying malignant disease.

Case 3 A 23-year-old actress was found semiconscious with 2 empty tablet bottles on her bedside table. That morning she was known to have bought 50 aspirins and had also obtained 60 Tylenol 3 tablets from her doctor for a strained back. She had been seen by a friend 3 hours earlier, leaving a coffee shop.

On examination he was lethargic and pale. His temperature was 38.4°C with warm peripheries. Respiratory rate was 20 per minute, blood pressure was 130/90 mm Hg, jugular venous pressure (JVP) was raised 5 cm, and pulse was 110 beats per minute and regular. The apex beat was palpable in the anterior axillary line, and there was a presystolic gallop rhythm. Examination of the fundi revealed vessel tortuosity. There was slight ankle edema. In the respiratory system, extensive bilateral crepitations were heard. The other systems were normal, apart from the presence of a 6 × 8 cm tender, fluctuant abscess in the left axilla. A CXR is definitely warranted in this scenario, related to respiratory system signs and symptoms.

Case 5 A 46-year-old man is scheduled for elective cholecystectomy. He complained of right upper quadrant abdominal pain, accompanied by nausea, and indigestion that had been present for the last 2 months. He does not smoke and drinks 2 to 3 beers on Saturday nights. He had a myocardial infarction 5 years ago and had arthritis in both knees for the previous 5 years. His only medications are over-the-counter analgesics for arthritis.

On admission, she was semicomatose; there was no evidence of external injury, and she was hyperventilating and diaphoretic. Her pulse was 110 per minute, blood pressure was 100/60 mm Hg, and chest was clear. There were no abnormal signs in the abdomen or in the central nervous system.

On examination, he looked well. He was not anemic, jaundiced, cyanosed, or clubbed, and there was no lymphadenopathy. Pulse was 84 beats per minute, regular with ectopic beats. JVP was raised 4 cm. Cardiac apex was in the anterior axillary line. He had a prominent third heart sound, but there were no murmurs. Chest was clinically clear. In the abdomen, the liver was palpable 3 cm below the costal margin, smooth and firm, and the spleen was barely palpable. He had crepitus in both knees. Urinalysis was negative.

The expected decision for this scenario was not to order a CXR because the physical exam of the chest was clear, and hyperventilation was most likely related to an analgesic overdose.

A preoperative CXR is warranted given the history of myocardial infarction and the physical exam findings suggestive of cardiomegaly and arrhythmia.

224 JACR vol 56, no 4, octobre 2005