Introduction
A recent analysis of studies on induced prescription (IP) reveals differences in how this phenomenon is conceptualized. Most studies consider IP to originate in specialized health care, private practice, the patients themselves, or other agents in the health care process.1-3 Other studies have looked at IP occurring in specialized care within public systems,4 and some authors have considered IP only in patients who are followed up by specialists.5 A notable feature of these studies is the different methodological approaches they have used: some have estimated prevalence,1-3,5-8 others incidence;9,10 some have focused on specific therapeutic subgroups,11 others on certain types of prescriptions such as those for chronic medication.12 As a result, reported percentage rates of IP range from 9% to 77%.1-12
Few studies have measured the degree of agreement between the primary care physician and the physician who requests the IP.1-3 We consider this aspect to be fundamental, because of the potential to create conflict for family physicians who dissent from the decision to request an IP.
Compounding the problem is the fact that primary care physicians are subject to a number of prescribing targets.13 which, if disagreement arises with the referring physician, may be difficult to achieve because of the patients' faith in the specialist´s opinion (the so-called preacher effect)2 over the family physician's point of view.
We are unaware of any studies in Spain that have analyzed prescriptions induced by reference hospital colleagues but written by primary care physicians, the information supplied by the hospital, and the quality of prescribing practices evaluated according to criteria used in primary care. Those studies that have examined the quality of IP measure the intrinsic value of the drug.2-4,7,10,12,14
With the goal of developing activities to counter this problem, we chose as our overall objective the characterization of induced prescribing by physicians at a reference hospital (Hospital Universitario Vall d'Hebron), in physicians at primary care centers (PCC) located in the health care administration district served by the reference hospital.
Specific Objectives
1. To quantify the volume of patients with medication indicated by the reference hospital physician and referred to their PCC physician for prescriptions and follow-up.
2. To describe the characteristics of patients with IP.
3. To identify the health problems that lead to IP.
4. To determine the percentage of patients for whom information provided by the reference hospital is insufficient for adequate follow-up.
5. To describe the information supplied by the hospital regarding IP.
6. To describe the degree of agreement of primary care physicians with IP.
7. To analyze IP according to prescription quality criteria used for primary care by the Catalonian Institute of Health.
Material and Methods
In this descriptive, cross-sectional study we invited all family physicians at six reformed urban PCC to participate. The centers were located within the area served by the reference hospital. The population served by the participating PCC was 107 826 inhabitants, 16.7% of whom were older than 64 years.
Study Population
The study population consisted of patients on medication indicated by the reference hospital physician, and who were referred back to their PCC for follow-up and to obtain prescriptions on the days the study was carried out. Patients younger than 14 years were excluded.
Sample
To estimate the proportion of patients with IP we estimated that a minimum of 384 patients were needed (assuming 50% prevalence, 0.05% precision and 95% confidence intervals). Because it was possible to study a larger number of patients we decided to include all patients referred to their PCC and all medications indicated by reference hospital physicians and for which prescriptions were written at the PCC on 10 consecutive working days during the months of April, May and June 2000.
We used a specially designed data sheet and instruction booklet for completing the sheet. All data were recorded by participating physicians.
Variables
1. Information about the center: name of the PCC, family physician, date the sheet was completed.
2. Information about the patient: age, sex, medical record number.
3. Information about the IP: origin (emergency room, out-patient clinic, discharge from hospital ward), specialty of the physician who induced the prescription, diagnosis according to the International Primary Care Classification,15 pharmaceutical specialty according to the ACT classification of drugs,16 dosage, duration of treatment, agreement of the family physician with the IP, and reason for dissenting with the IP. We included an item that asked family physicians whether they considered the information supplied by the hospital to be sufficient for follow-up of the patient.
Evaluation of Prescription Quality
We used a prescription quality standard which describes the pattern of use of medications in primary care.13 This document consists of indicators grouped into two sections. The first set of indicators is designed to analyze prescribing practices in general terms of intrinsic pharmacological value, prescription of generic pharmaceutical specialties, consumption of new medications, and compliance with the recommendations of the Comité de Evaluación de Nuevos Medicamentos (New Drug Evaluation Committee).17 The second set of indicators deals with compliance with the recommendations for the use of drugs from certain therapeutic subgroups such as antihypertensives, antacids, antiinflammatory drugs, antibiotics, lipid reducing agents, antiasthmatics, antidepressants, anxiolytics/hypnotics and sedatives, and oral antidiabetics.
The information on each data sheet was entered in a database on a personal computer, and a descriptive analysis was obtained with the SPSS, v. 10.
Results
Of the 52 physicians eligible to participate in the study, 36 (69.2%) recorded data for 323 patients with IP from the reference hospital. Hospital visits or stays by these patients led to 768 prescriptions for a total of 844 active principles. The mean number of active principles per patient was 2.65 (SD, 1.8, range 1 to 9).
The total number of patients seen by participating physicians was 5997. Patients with IP made up approximately 5.4% (95% CI, 4.8%-5.9%) of all patients seen.
Of the 323 patients, 168 (52%, 95% CI, 47%-58%) were referred from the emergency room, 105 (33%, 95% CI, 27%-38%) from out-patient clinics, and 50 (15%, 95% CI, 12%-19%) after discharge from the hospital ward. Slightly more than half (174 patients, 53.9%) were women. The proportion of women was greater among patients referred from out-patient departments (61%), and smaller among patients who were discharged from the hospital ward with requests for prescriptions (40%). Mean age was 56.8 years (SD, 20.8; range 16 to 100 years). The greatest numbers of patients were from the traumatology (59 patients, 18.3%) and internal medicine departments (46, 14,2%).
Of the total of 844 medications, 399 (47%, 95% CI, 44%-51%) were indicated by emergency room physicians, 226 (27%, 95% CI, 24%-30%) by out-patient department physicians, and 219 (26%, 95% CI, 23%-29%) by inpatient ward physicians. Medications were prescribed for a total of 129 health problems (mean number of diagnoses per patient: 2.7; SD, 1.8). Almost one-fourth (196 medications, 23.2%) were prescribed for musculoskeletal problems, 166 (19.7%) for respiratory problems, 112 (13.3%) for cardiovascular problems, 91 (10.8%) for alimentary tract problems, and 55 (6.5%) for genitourinary tract problems.
Table 1 shows the distribution of the IP (number of patients and type of medication) by indication. Table 2 shows the IP in 26 patients diagnosed as having chronic obstructive pulmonary disease (COPD) and respiratory tract superinfection.
In 64 patients (20%, 95% CI, 16%-25%) the primary care physician considered the information from the hospital to be insufficient to ensure adequate follow-up. Of these patients, 46 (71.9%) were referred from outpatient departments (Figure 1). The traumatology department was identified as the source of the greatest number of patients for whom information was inadequate (11 patients, 17.2%).
Figure 1. Family physician's opinion regarding information from the hospital on patient follow-up.
Notably, no diagnosis was indicated for 93 medications (11.2%), 77 (82.8%) of which were requested by physicians at outpatient departments. For 46 medications (5.4%) no dosage was indicated, and for 376 (44.6%) the duration of treatment was not specified. The diagnosis, dosage and duration of treatment were indicated for 306 medications (44.5%).
For 534 (63%, 95% CI, 60%-67%) medications the family physician concurred with the referring physician who requested the IP; for 114 (14%, 95% CI, 11%-16%) there was partial agreement, whereas for 44 (5%, 95% CI, 4%-7%) there was complete disagreement (Table 3). The groups of medical conditions associated most frequently with disagreement over the IP were musculoskeletal (42 medications, 26.6%) and respiratory disorders (23 medications, 14.6%). For 61 medications (7.2%) the physicians disagreed on the choice of drug, and for 48 (5.6%) the family physician felt that no prescription medication was needed.
The distribution of IP (number of patients and type of medication) by therapeutic subgroup is shown in Table 4. The most frequently indicated medications were paracetamol (acetaminophen) (57 patients, 6.8%), diclofenac (51, 6%), metamizole (50, 5.9%), omeprazole (32, 3.8%) and ranitidine (31, 3.7%).
Quality of the different IP according to prescribing standards used in primary care is shown in Table 5.
Discussion
We found IP in approximately 5% of the patients referred by the reference hospital for follow-up to their primary care physician. Although most studies measure IP per se rather than the proportion of patients with at least one IP, we felt it was important to obtain information on the latter, which is potentially useful in managing patient care as opposed to controlling pharmaceutical costs. One study9 found IP in 3.8% of the patients. Two studies1,2 found IP in 36% and 58% of the patients, but because they estimated the prevalence of patients with IP of any origin, their figures cannot be compared to ours, as we recorded only those patients with IP originating from the reference hospital.
The percentage of patients with IP who were referred from the emergency department was higher in our study than in an earlier report that analyzed public specialized care.9 This difference may have resulted from structural and geographical characteristics (continued care not provided by some PCC, and proximity of the reference hospital) found by two studies in our region1,2 to be associated with a greater prevalence of IP.
Our study revealed possible discrepancies between prescribing practices for 11 different antiinflammatory drugs (e.g., rofecoxib and dexketoprofen) and the unavailability of these drugs in the hospital formulary. This suggests that some physicians may use a limited number of medications for inpatients, while prescribing different medications for outpatients and patients seen in the emergency department.18 Of note was the finding that systemic corticosteroids were prescribed for 50% of the patients diagnosed as having COPD, and that antibiotics not considered to be the first choice in primary care (ciprofloxacin, cefuroxime-axetil and clarithromycin) were prescribed for 19.2% of the patients diagnosed as having respiratory superinfection.
Our analysis of the patient characteristics and the appropriateness of prescriptions is limited, as more sophisticated studies are needed to compile detailed data for other clinical variables in the patients. Nevertheless, our study does make it possible to identify the source of referrals, the health problems that led patients to seek medical care (the distribution of medical problems was similar to that in an earlier study10), and the treatments that were recommended. We thus consider our information to be useful for setting priorities for future research aimed at enhancing treatment for the most prevalent health problems in our setting.
The proportion of patients for whom little or no information was provided about the diagnosis or treatment was similar to that in other studies, which reported figures of 13% and 17% for IP originating from hospital physicians.4,9 We agree with those who consider that all requests for prescriptions from hospital physicians should be accompanied by a minimum amount of information about the patient´s medical problem and course of treatment, and with those who feel that the patients should not be relied on as a means of communication.19
In the present study 30% of the physicians who were invited to participate declined to take part, therefore we cannot assure that our sample was completely representative of all primary care physicians. We believe this may have affected mainly the data for physicians´ opinions on the compliance of the IP with good prescribing criteria. Nevertheless, the rate of lack of agreement we observed was similar to that of other studies, which reported figures of 13.7% and 24.9%.1,2 In our study the main areas of disagreement were the choice of drug and the need for prescription medication. An earlier study found a similar percentage rate of disagreement over the choice of medication (8.4%), and a smaller percentage of primary care physicians who felt no prescription was warranted (2.6%); however, this study included IP from other sources not considered in the present study.3 The lack of consensus is a challenge for family physicians who may wish to manage and change induced prescribing practices they consider inappropriate or undesirable.20,21
Analyses of IP according to standards for prescription quality are limited, as thus far IP has been studied only for brief periods of time, and the indicators available were designed specifically to analyze prescribing practices in primary care. Discrepancies between IP and quality standards for primary care may reflect a lack of knowledge, on the part of hospital physicians, of the quality criteria for prescribing practices in primary care. In any case the gap is evidence of the need for greater coordination and communication between different levels of health care.
Joint efforts between primary care and hospital physicians to develop clinical practice guidelines and formularies, and to plan shared continuing education activities, may help close the gap between prescribing practices at different levels of health care.19,22,23
Researchers of the Grupo de estudio de la Prescripción Inducida
SAP Nou Barris: Eladio Fernández, Francesc Orfila and Núria Torres. Fundació Institut Català de Farmacologia: Dolores Rodríguez and Eduard Diogène. SAP Gràcia-Horta-Guinardó, Institut Català de la Salut: Núria Sabaté.
EAP Canteres: M. Teresa Adell, Pilar Algueró, Ana Bonillo, Anna M. Callejero, Josep Despuig, Inmaculada Grau, María Llevadot and Mercedes Martínez.
EAP Chafarinas: Joan Atmetlla, Montse Delgado, Ana M. Farran, Xavier Flor, Lucía Gallego, Montse Mas and Carmen Munar.
EAP Ciutat Meridiana: Gemma Baulies, Francisco José Luque, Araceli Martín and M. Carmen Troyano.
EAP Horta 7D: Gemma Amorós, Gemma Badell, Edelmira Barraquer, M. Luisa Galan, Elena Jordi, M. Luisa Rodríguez, Carmen Rodrigo and M. Eulalia Teixidor.
EAP Prosperitat: José M. Cots, Encarnación Fernández, Cristina Ros and Pere Subirana.
EAP Verdum: Montserrat Callado, Francisco Javier Clar, Antonio Dedeu, Silvia Güell, Álvaro Mendez, Jorge Muela, Consuelo Palacios and Josep Màrius Salabert.
