Assessing residents prescribing behavior in renal impairment

Transcription

1 International Journal for Quality in Health Care 2003; Volume 15, Number 3: pp /intqhc/mzg034 Assessing residents prescribing behavior in renal impairment L. SALOMON 1, S. LEVU 1, G. DERAY 2, V. LAUNAY-VACHER 2,G.BRÜCKER 1 AND P. RAVAUD 3,4 1 Public Health and 2 Nephrology Departments, Groupe Hospitalier Pitié Salpêtrière et Faculté de médecine Pitié Salpêtrière (Paris VI), Paris, 3 Epidemiology and Biostatistics Department, Hôpital Bichat et Faculté de Médecine Xavier Bichat (Paris VII), Paris and 4 Unité INSERM U444, France Abstract Objective. Although fitting orders to renal function avoids overdosage and therefore iatrogenic risk, dosage adjustment is rarely made. The objective of this study was to assess residents prescribing behavior in renal impairment, through a standardized simulated clinical setting. Method. This criterion-referenced study was carried out in a French teaching hospital. The hospital had 118 residents; 71 of them were asked to complete a questionnaire including four vignettes, simulating drug prescription in four patients with various degrees of renal impairment (16 orders). The patients had an order of gentamicin sulfate, diclofenac sodium, and amlodipine bensylate. For each drug, the resident could maintain the order, discontinue the order, or change the dosage. A fourth drug, enalapril maleate, was to be started, with three possible dosages and the possibility of not prescribing it. The reference chosen for assessment was the Vidal dictionary, which corresponds to the Physician s Desk Reference and is the French reference for prescription. Results. All the residents approached for the survey accepted the offer to complete the questionnaire. Among the 16 simulated orders, the median number of appropriate orders per resident was nine. Considering the renal function of their patients, 62% of residents wrote an inappropriate order for gentamicin, 42% wrote an inappropriate order for diclofenac, and 52% wrote an inappropriate order for enalapril. Although no adjustment to renal function was required, 28% of the residents decreased the dosage of amlodipine and ordered an underdose. Conclusion. Considering the iatrogenic risk related to the lack of dosage adjustment, attention should be drawn to increasing residents awareness of dosage adjustment in renal impairment and to providing them with better information on patients renal function. Keywords: dosage adjustment, renal impairment, residents prescribing, vignettes acute renal failure and the glomerular filtration rate falls from 100 to 10 ml/min, the serum creatinine level will not change within 24 hours, until it reflects its accumulation in the plasma. In addition, an apparently minor increase in serum creatinine can reflect a marked fall in glomerular filtration rate (Figure 1). For that reason, the estimation of glomerular filtration rate through the calculation of creatinine clearance is mandatory in every patient. Although the necessity of dosage adjustment is probably underestimated in clinical practice, few studies have con- sidered this issue in hospitalized patients and the means to improve the use of medications in renal impairment [2,3,5, 8]. Measuring the competence of physicians and the quality of their actual practice has proven to be difficult and problematic. Because they control for case mix, vignettes or written cases Dosage adjustment according to renal function decreases iatrogenic risk and irreversible renal impairment [1]. Although recommended for many drugs, these adjustments are rarely made [2,3]. According to the Leape classification, the lack of adaptation may result from lack of knowledge of the drug, including: (1) underestimation of the iatrogenic risk related to decreased drug elimination in mild renal impairment; and (2) ignorance of the medications that require a dosage adjustment in renal impairment [4 6]. It may also result from a poor estimation of the renal function of the patient. Changes in glomerular filtration rate should be estimated by the clearance of creatinine calculated using the Cockcroft and Gault formula [7]. In daily routine, glomerular filtration is determined from the measurement of serum creatinine, which has severe limitations. For example, if a patient develops Address reprint requests to L. Salomon, Unité d Evaluation, Hôpital Louis Mourier, 178 rue des Renouillers, Colombes Cedex, France. laurence.salomon@lmr.ap-hop-paris.fr International Journal for Quality in Health Care 15(3) International Society for Quality in Health Care and Oxford University Press 2003; all rights reserved 235

2 L. Salomon et al. Figure 1 Relation between serum creatinine and glomerular filtration rate (GFR). simulations allow assessment among different providers and between organizations that may care for different populations of patients under different systems of care. They have been described as a valid way of assessing quality of care [9 12]. The objective of this study was to assess residents prescribing behavior in patients with renal impairment, through a standardized simulated clinical setting using hypothetical vignettes. Methods Design and setting Questionnaire The reference chosen for assessment was the 1999 edition of the Vidal dictionary [13]. This dictionary pools the man- ufacturers summaries of product characteristics. Annually updated, it is officially distributed by the French Regulatory The questionnaire included two questions exploring the concerns of residents about dosage adjustment in renal impairment: (1) Do you consider that the need for dosage adjustment in renal impairment is a frequent problem in hospital prescription? ; and (2) From what level of serum creatinine should you check if the drugs prescribed may need dosage adjustment?. It also included four clinical vignettes (Table 1), simulating drug prescription in patients with various degrees of renal This prospective criterion-referenced survey described res- idents practice through clinical vignettes, a valid and com- prehensive method of assessing quality of care and one that is able to control adequately for case-mix variation [12]. A questionnaire was distributed in May 2000 to the residents who assumed new positions in the Hospital Pitié Salpêtrière (Assistance Publique, Hôpitaux de Paris), a 2070-bed, tertiary care teaching hospital. This hospital has all major surgical and medical activities, with 40 in-patient departments, in- cluding 114 units, and > patients hospitalized every year. The hospital receives residents in all the medical and surgical departments. impairment. These drugs had been chosen in collaboration with a nephrologist and a pharmacist as they are frequently prescribed in general practice, or are well known to need adjustment in renal impairment (such as gentamicin), such that residents from medical or surgical departments might prescribe them regularly. The data provided were patient s age, gender, weight, and serum creatinine level. Creatinine clearance, which is not assessed directly in daily practice, was not provided in the survey, but could be calculated. Patients 1 and 3 had severe renal impairment, with similar calculated creatinine clearances (25 and 28 ml/min, respectively) but markedly different creatinine levels (1.2 and 2.2 mg/dl, respectively). Patients 2 and 4 had a creatinine level of 1.0 mg/dl, with, respectively, mild renal impairment (calculated clearance of 57 ml/min) and slight renal impairment (calculated clearance of 64 ml/ min). The patients were supposed to have just arrived in the department with the following order: (1) gentamicin sulfate 1 mg/kg three times a day; (2) diclofenac sodium 50 mg twice a day; (3) amlodipine bensylate 10 mg per day. The resident was to ratify each patient s order or to change the dosage (with a choice between reducing the concentration or volume required per dose, or increasing the dosing interval). Another option was to discontinue the order. A fourth drug, enalapril maleate, was to be started, in the context of high blood pressure, with three possible dosages and the possibility of not prescribing it. We considered the initial prescription of enalapril before further adjustment, which could have been made according to enalapril titration. Gold standard 236

3 Prescribing in renal impairment Table 1 Vignettes simulating drug prescription in four patients with various degrees of renal impairment Patient 1: female Patient 2: male Patient 3: female Patient 4: female (age 79 years, (age 80 years, (age 36 years, (age 48 years, weight 40 kg) weight 70 kg) weight 50 kg) weight 60 kg) Order: Cr 1.2 mg/dl Order: Cr 1.0 mg/dl Order: Cr 2.2 mg/dl Order: Cr 1.0 mg/dl... Gentamicin: 1 mg/kg 3 40 mg per day 3 70 mg per day 3 50 mg per day 3 60 mg per day 1 three times a day 3 20 mg per day 3 40 mg per day mg per day 3 30 mg per day 3 10 mg per day mg per day 3 10 mg per day mg per day Discontinue the order Discontinue the order Discontinue the order Discontinue the order Diclofenac: 50 mg twice a day 50 mg twice a day 50 mg twice a day 1 50 mg twice a day 50 mg twice a day 1 50 mg once a day 50 mg once a day 50 mg once a day 50 mg once a day 25 mg twice a day 25 mg twice a day 25 mg twice a day 25 mg twice a day Discontinue the order 1 Discontinue the order Discontinue the order 1 Discontinue the order Amlodipine: 10 mg per day 10 mg per day 1 10 mg per day 1 10 mg per day 1 10 mg per day 1 5 mg once a day 5 mg once a day 5 mg once a day 5 mg once a day 10 mg 1 day out of 2 10 mg 1 day out of 2 10 mg 1 day out of 2 10 mg 1 day out of 2 Discontinue the order Discontinue the order Discontinue the order Discontinue the order Enalapril: Start a treatment mg per day mg per day mg per day mg per day (high blood pressure) 5 10 mg per day 5 10 mg per day mg per day 5 10 mg per day 1 20 mg per day 20 mg per day 20 mg per day 20 mg per day Discontinue the order Discontinue the order Discontinue the order Discontinue the order Cr, serum creatinine level. 1 Recommended order. Creatinine clearance was not provided in the clinical case. It had to be calculated by the residents: patients 1 and 3 have severe renal impairment, with a calculated creatinine clearance <30 ml/min; patient 2 has mild renal impairment, with a calculated creatinine clearance <60 ml/min; patient 4 has slight renal impairment, with a calculated creatinine clearance of 64 ml/min. Authorities. It corresponds to the Physician s Desk Reference and is the French reference for prescription. Commonly consulted by physicians, it may be used as formal evidence in legal procedures. Participants and data Analysis A descriptive analysis of data was made. According to the calculated clearance of creatinine, the appropriateness of the 16 orders was determined (four medications and four patients ). The number of appropriate orders per residents was determined (range 0 16). A chi-square test of association was used to compare the frequency of inappropriate orders, with respect to: (1) whether or not an adjustment was required; (2) patient 1 versus patient 3 similar clearances but different serum creatinine levels; and (3) patient 1 versus patient 2 similar levels of serum creatinine and different renal function. To take into account multiple testing, significance was set In French hospitals, residents change departments every 6 months during the 2 5 years of their residency. The study took place at the hospital administration office, where the residents come to register at the start of the semester. An investigator was present at the administration office to deliver the questionnaire to the residents. They completed this anonymous questionnaire alone and returned it immediately to the investigator. The investigator specified that the purpose at P = 0.01 instead of P = of the study was not to test the knowledge of the residents, but to comprehend their behavior when prescribing drugs in renal impairment. Residents were encouraged to act as they Results would for genuine patients, and to check any document they would use for usual prescribing, but not more. A Vidal dictionary was placed in an obvious position near the in- vestigator, and was available for consultation if the residents requested it. Residents were also told to spend the same time on the questionnaire as they would actually take to write orders. Among the 118 residents, 71 (60% of the residents) came to register at the hospital administration office at the beginning of the semester. All completed the questionnaire. Of these, 73% were medical residents, 3% were residents in anesthesia, and 24% had a surgical speciality. Fifty percent had already 237

4 L. Salomon et al. Table 2 Number (and percentage) of appropriate orders per resident Score Appropriate orders per resident n (%) Total performed at least six postgraduate semesters, with seniority ranging from postgraduate semester 1 to 10 (out of 10). The necessity of dosage adjustment in renal impairment was considered to be a frequent problem in hospital prescription by 85% of the residents (n = 58). Among the residents, 20% (n = 13) reported checking if the dosage needs adjustment when serum creatinine is >0.9 mg/dl, 35% (n = 23) when creatinine is >1.2 mg/dl, 35% (n = 23) when creatinine is >1.7 mg/dl, and 6% (n = 4) when creatinine is >2.3 mg/dl. Four percent of the residents (n = 3) reported checking if the dosage needs adjustment for any level of serum creatinine. Among the 16 simulated orders, the median number of appropriate orders per resident was nine (range 0 15) (Table 2). Prescription of gentamicin sulfate (Table 3) For gentamicin sulfate, 62% of the respondents prescribed a dosage that was not appropriate. The lowest percentage of inappropriate orders was observed for patient 4 (26% of inappropriate orders), for whom the dosage did not require adjustment. The highest percentages of inappropriate orders were observed for patients 1 and 3 (78% and 79% of inappropriate orders, respectively), for whom the dosage needed adjustment. When the order needed to be modified (patients 1, 2, and 3), inappropriate orders were written more often than when no adjustment was required (74% versus 26%; P <10 4 ). For similar clearances (patients 1 and 3), the prescription Table 3 Simulated prescribing behavior of the residents for four patients with various degrees of renal impairment Clinical cases Patient 1 Patient 2 Patient 3 Patient 4 Cr 1.2 mg/dl, CrCl 25 ml/min Cr 1.0 mg/dl, CrCl 57 ml/min Cr 2.2 mg/dl, CrCl 28 ml/min Cr 1.0 mg/dl, CrCl 64 ml/min G D A E G D A E G D A E G D A E... Appropriate order (%) Maintained Decreased Discontinued Inappropriate order (%) Maintained (overdose) Decreased (overdose) Underdose Discontinued Number of responders Cr, serum creatinine level; CrCl, creatinine clearance; G, gentamicin sulfate; D, diclofenac sodium; A, amlodipine bensylate; E, enalapril maleate. 238

5 Prescribing in renal impairment of gentamicin was stopped more frequently in the patient patients. Although they adjusted the dosage, 13% and 9% of with the highest levels of serum creatinine (40% for 2.2 mg/ the residents still ordered an overdose for patients 1 and 3, dl creatinine versus 20% for 1.2 mg/dl; P = 0.009). respectively. For patients 2 and 4, 57% and 36% of the residents For similar levels of serum creatinine (1.2 and 1.0 mg/dl), overadjusted the dosage, respectively, leading to an underdose. there was no difference in the proportion of residents who reduced the dosage of gentamicin (28% for patient 1 and 12% for patient 4, respectively; P > 0.01), although patient 1 Discussion had a severe renal impairment and patient 2 had a mild renal impairment. This dosage adjustment led to an overdose for Medication misuse has frequently been reported and may patient 1, and to an underdose for patient 2. cause iatrogenic complications, which is an important cause of morbidity and mortality in hospitalized patients [6,14 17]. This Prescription of diclofenac sodium (Table 3) study revealed a lack of dosage adjustment for medications in patients with renal impairment. Although most residents in our For diclofenac sodium, 42% of the responders prescribed a hospital were aware of the necessity to adjust dosage in renal dosage that was not adjusted to the renal function of their impairment, very few of them (5%) reported checking whether patients. The lowest percentage of inappropriate orders was their order needed adjustment without an increase in serum observed for patient 4 (16% of inappropriate orders), for whom creatinine. This result indicates that the residents do not calculate the dosage did not require adjustment. The highest percentage the clearance of creatinine for every patient and thus may of inappropriate orders was observed for patient 1 (65% of underestimate the depth of renal impairment, since an apinappropriate orders), whose order needed to be discontinued. parently minor increase in serum creatinine can reflect a marked When the order required adjustment (patients 1 and 3), fall in glomerular filtration rate (Figure 1). inappropriate orders were written more often than when no A limitation of the study is that although it is supposed adjustment was needed (55% versus 29%; P = 10 3 ). to examine residents performance in their daily routine, the Considering patients with similar clearances, 55% of the residents may have either increased awareness in a simulated residents discontinued the order of diclofenac for patient 3, prescribing situation or decreased their concern about ficwho had the highest level of serum creatinine (2.2 mg/dl), titious patients. However, as demonstrated by Peabody, using compared with 35% of residents discontinuing the order for patient 1 (creatinine 1.2 mg/dl). Taking into account the vignettes is a relevant and valid method to assess quality correction for multiple tests, this difference is not significant of care, compared with chart abstraction and standardized (P = 0.02). patients [12]. Using vignettes allows assessment of the quality For similar levels of serum creatinine (1.2 and 1.0 mg/dl), of care provided by a group of physicians, and controls for there was no difference in the proportions of residents who case mix. It appears to be a valid and comprehensive method reduced the dosage of diclofenac (30% for patient 1 and that focuses directly on the process of care provided in actual 27% for patient 2, respectively; P > 0.05), although patient 1 clinical practice. The conclusions of surveys using vignettes, had a severe renal impairment and patient 2 had a mild renal therefore, may more likely be generalizable than surveys impairment. This dosage adjustment led to an overdose for performed in a population of patients. patient 1, and to an underdose for patient 2. The study took place at the hospital administration office, where the residents are supposed to come and register at the Prescription of amlodipine bensylate (Table 3) beginning of the semester, except for those who had already worked as residents or as students in the hospital. Since the For amlodopine bensylate, 31% of the responders wrote an policy of the hospital concerning medication prescribing and order that was not appropriate for the renal function of their dosage adjustment is similar to that of the other teaching patients. The lowest percentage of inappropriate orders was hospitals where the residents could have worked, we assumed observed for patient 4 (21% of inappropriate orders), the that the non responders (mostly residents who had already highest percentage of inappropriate orders were observed for worked in the hospital) would not differ from the responders patient 3 (42% of inappropriate orders). with regards to their prescribing behavior. All the residents who Although no dosage adjustment was required, 3% of the came to the administration office completed the questionnaire. residents discontinued the medication and 28% decreased The residents were told that their knowledge of the drugs the dosage of amlodopine, which led to an underdose. was not being tested and were encouraged to act as they would The proportion of residents who inappropriately modified in their daily routine. Thus, the Vidal dictionary was set on the the orders was similar when comparing (1) patient 1 with table and the residents were free to consult it to complete the patient 3 (who have both severe renal impairment, but questionnaire. During data collection, investigators noted that different levels of creatinine), and (2) patient 1 with patient residents rarely checked their orders in the Vidal dictionary and 2 (who have severe and mild renal impairment, but similar rarely calculated creatinine clearance. This observation was levels of creatinine). confirmed by the results of the study: there was no difference Prescription of enalapril maleate (Table 3) in the percentage of residents who reduced the dosage of gentamicin for patients 1 and 2, who had similar levels of For enalapril maleate, 52% of the responders prescribed a serum creatinine (1.2 and 1.0 mg/dl, respectively). Yet this dosage that was not adjusted to the renal function of their adjustment led to an overdose for patient 1, who actually had 239

6 L. Salomon et al. severe renal impairment, and to an underdose for patient 2, 4. Drayer DE. Active drug metabolites and renal failure. Am J who had mild renal impairment. Similar results were observed Med 1977; 62: for the prescription of diclofenac and confirm that most 5. Rind DM, Safran C, Phillip RS et al. Effect of computer-based residents consider serum creatinine level as an indicator of renal alerts on the treatment and outcomes of hospitalised patients. function, instead of calculating creatinine clearance, neglecting Arch Intern Med 1994; 154: patients age and weight. 6. Leape LL, Bates DW, Cullen DJ et al. System analysis of adverse In prescribing amlodipine bensylate, which required no drug events. J Am Med Assoc 1995; 274: adjustment, some residents inappropriately reduced the dosfrom serum creatinine. Nephron 1976; 16: Cockcroft DW, Gault MH. Prediction of creatinine clearance age. Similarly, some residents prescribed the lowest dosage of enalapril maleate where a higher dose could have been 8. Preston SL, Briceland LL, Lomaestro BM, Lesar TS, Bailie ordered. (Since dose titration is commonly performed with GR, Drusano GL. Dosing adjustment of 10 antimicrobials for patients with renal impairment. Ann Pharmacother 1995; 29: this medication, residents were told that they had to start the treatment, i.e. to write the order before dose titration.) Thus, residents were particularly cautious in their orders, but 9. Yager J, Linn LS, Leake B, Gastaldo G, Palkowski C. Initial clinical judgment by internists, family physicians, and psyinappropriately adjusted the treatment, risking underdosage, chiatrists in response to patient vignettes. I. assessment of rather than consulting the Vidal dictionary. problems and diagnostic possibilities. Gen Hosp Psychiatry 1986; To improve prescriptions in renal impairment, it would be 8: necessary to implement actions to draw physicians attention 10. Tait RC, Chibnall JT. Physician judgments of chronic pain to the importance of creatinine clearance and to provide them, patients. Soc Sci Med 1997; 45: at the time of prescription, with reminders of prescription 11. Colenda CC, Rapp SR, Leist JC, Poses RM. Clinical variables guidelines in renal impairment. A systematic review of con- influencing treatment decisions for agitated dementia patients: trolled studies has shown that among 14 studies assessing survey of physician judgments. J Am Geriatr Soc 1996; 44: this issue, six revealed an actual impact of computerized prescription on quality of care [18]. Therefore, implementing 12. Peabody JW, Luck J, Glassman P, Dresselhaus TR, Lee M. computing systems with reminders at the time of prescription Comparison of vignettes, standardized patients, and chart ab- could be of great interest. The implementation of computing straction: a prospective validation study of 3 methods for systems that screen for inappropriate dosage and provide measuring quality. J Am Med Assoc 2000; 283: reminders of prescription guidelines in renal impairment has 13. Vidal Dictionary. The Vidal Editions: Paris, already been reported, with a positive impact on the quality 14. Bates DW, Cullen DJ, Laird N et al. Incidence of adverse drug of medical orders [5,19 21]. events and potential adverse drug events. Implications for prevention. ADE Prevention Study Group. J Am Med Assoc 1995; 274: Conclusions 15. Leape LL. Error in medicine. J Am Med Assoc 1994; 272: Although residents believe that dosage adjustment is a fre- 16. Lesar TS, Briceland L, Delcoure K, Parmalee JC, Masta-Gornic quent problem in hospitalized patients, most of those in our V, Pohl H. Medication prescribing errors in a teaching hospital. study did not properly determine the renal function of J Am Med Assoc 1990; 263: fictitious patients and did not seek information on adjusting 17. Brennan TA, Leape LL, Laird NM et al. Incidence of adverse dosages of medications that are metabolized or excreted by events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. N Engl J Med 1991; 324: the kidney, or of nephrotoxic medications Considering the iatrogenic risk and irreversible renal impairment related to the lack of dosage adjustment, attention 18. Hunt DL, Haynes RB, Hanna SE, Smith K. Effects of computer- based clinical decision support systems on physician pershould be drawn to the need to increase residents awareness formance and patient outcomes: a systematic review. JAmMed of dosage adjustment in renal impairment and to provide Assoc 1998; 280: them with better information on their patients renal function. 19. Peterson JP, Colucci VJ, Schiff SE. Using serum creatinine concentrations to screen for inappropriate dosage of renally eliminated drugs. Am J Hosp Pharm 1991; 48: References 20. Goldberg DE, Baardsgaard G, Johnson MT, Jolowsky CM, Shepherd M, Peterson CD. Computer-based program for iden- 1. Alkhunaizi AM, Schrier RW. Management of acute renal failure: tifying medication orders requiring dosage modification based new perspectives. Am J Kidney Dis 1996; 28: on renal function. Am J Hosp Pharm 1991; 48: Cantu TG, Ellerbeck EF, Yun SW, Castine SD, Kornhauser 21. McMullin ST, Reichley RM, Kahn MG, Dunagan WC, Bailey DM. Drug prescribing for patients with changing renal function. TC. Automated system for identifying potential dosage problems at a large university hospital. Am J Health Syst Pharm 1997; 54: Am J Hosp Pharm 1992; 49: Wong NA, Jones HW. An analysis of discharge drug prescribing among elderly patients with renal impairment. Postgrad Med J 1998; 74: Accepted for publication 5 February