European Journal of Epidemiology 15: 815±819, 1999. Ó 1999 Kluwer Academic Publishers. Printed in the Netherlands.
Evidence based general practice M.M. Morales SuaÂrez-Varela1;2 , A. Llopis-GonzaÂlez1 , J. Bell3 , M. TalloÂn-Guerola1 , A. PeÂrez-Benajas4 & C. CarrioÂn-CarrioÂn1 1
Unit of Public Health, Hygiene and Environmental Health, Valencia University; 2 Unit of Clinical-Epidemiological Research, Dr Peset University Hospital, Valencia, Spain; 3Thames Cancer Registry, London, UK; 4Chemist's Montanejos, Montanejos, CastelloÂn, Spain Accepted in revised form 11 July 1999
Abstract. Objectives: to estimate the proportion of interventions in general practice that are based on evidence. Design: a one-year cross-sectional study involving all consultations by patients over age 15 years seen in 34 national primary health care centers. Setting: the rural CastelloÂn provincial district within the Valencian Community in eastern Spain, with a total population of 21,155 inhabitants. Subjects: of 1990 case histories registered in the course of one year, 4800 consultations were identi®ed; of these, 2341 (49%) distinct diagnosis±intervention pairs were identi®ed and coded. Main results: the evidence basis for the diagnosis±intervention pairs in the study was derived from a computerized search of the scienti®c literature published in 1992±1996. The quality of the evidence was classi®ed according to the method of
Ellis et al. Within the 2341 diagnosis±intervention pairs, there was positive evidence in support of the intervention used in 55%. The evidence basis was sound for 42%, with 38% being based on Type I (clinical trials) evidence and 4% on Type II evidence. The most frequently presenting diseases involved the circulatory (18.7%), respiratory (14.9%), nervous (14.2%), musculo-skeletal (12.5%) and nutrition and metabolism and digestive systems, with 12.1% each. Conclusions: clinical practice was clearly supported by positive evidence of all Types (I±III) in a total of 55% of interventions, and by good positive evidence of Type I or II in 42% of interventions. The percentage of evidence-based interventions in general practice serving a substantial population in rural Spain was lower than had been reported by some authors.
Key words: Evidence-based medicine, General practice, Primary care Introduction As has been aptly pointed out by Rose [1], `all decisions and therapies should be based upon precise measurements and strict research studies'. To this eect, health professionals need to continuously modify protocols to re¯ect both patient problems and new scienti®c advances. In recent years there has been growing interest in the development of evidence-based medicine [2], as re¯ected by the Cochrane Collaboration Study [3], the publication of a new journal (Evidence Based Medicine) [4], and the numerous papers published on this subject [5±13]. A number of studies, including our own, have attempted to more accurately quantify the need for and the extent of evidence-based medicine [5±8]. The literature suggests a great variability in clinical practice, with a large proportion of procedures being carried out without a thorough evaluation of their true clinical ecacy [14, 15]. This in turn aects the cost/ecacy ratio. Following a study by Dubinsky and Ferguson [16] in 1991, in which only 21% of scienti®c evidence was
found to support therapeutic decisions in a series of 126 patients in the USA, it has been claimed that only 10±20% of clinical practice is based on sound scienti®c evidence [17, 18] (implying that 80±90% of medical procedures lack valid evidence), although the percentage has varied with the setting in subsequent literature (increasing the percentage of clinical practice based on evidence) [14, 15]. As in other medical areas, evidence-based practice is beginning to have important repercussions in primary care [19]. The present study extends the work of Ellis et al. [17] and Gill et al. [20] on evidence-based general practice (GP) to primary care in a rural setting. The Gill study found 81% of interventions to be evidence based in one English multi-GP practice. The purpose of this study was to estimate the proportion of evidence-based interventions in primary care in a Spanish rural area served by a large number of general practitioners, according to the methodology proposed by Ellis et al. [17], who classi®ed the Type of Evidence as follows: I. Therapeutic interventions based on evidence from randomized controlled trials.
816 II. Interventions with convincing non-experimental evidence. III. Interventions with no substantial scienti®c evidence, meeting neither criterion I nor II. Methods A cross-sectional study was conducted based on 20% of the patients over the age of 15 years seen in 34 public primary care centers in the course of one calendar year, 1996. The primary care centers were all those serving rural populations within a de®ned region. Rural communities are de®ned by the Encuesta de Salud de la Comunidad Valenciana (ESCV) ± Health Questionnaire of the Valencian Community ± as municipalities with fewer than 5000 inhabitants [21]. The total population of the study area was 21,155 inhabitants (10,574 males and 10,571 females), as recorded by the municipal census statistics of 1986. This rural population comprised 0.5% of the 3,856,234 inhabitants of the administrative area of Valencia (Spain). Patients were selected by systematic sampling performed to this eect, selecting one history out of every ®ve. Our aim was to avoid sample bias generated as a result of the alphabetical order in which the histories were ®led. The patients' case-records were used to extract data on consultations (de®ned as any patient visit to a Primary Care Center). The researchers assigned a single diagnosis and an intervention to each consultation. Intervention was de®ned as both advice (counseling, weight supervision, referral to the specialist) and medical treatment. The study unit was distinct diagnosis±interventions pairs derived from consultations in 1996. Multiple occurrences of the same diagnostic-intervention were not counted. The diagnoses were classi®ed according to the detailed list of the Ninth Revision of the International Classi®cation of Diseases (ICD-9) [22]. Coding was carried out by the same researchers. The evidence basis for the interventions was derived from a computerized search of ®ve years of literature on the Medline SilverPlatter (1992±1996), using all the diagnosis and intervention terms as search keys. We identi®ed all scienti®c studies addressing interventions for the diagnoses within the study and all studies on the use of the interventions within the study. Then, for each diagnosis±intervention pair, we assessed the quality of the evidence in support of the use of the particular treatment for the diagnosis. The scienti®c evidence in favor of a given intervention was classi®ed according to the methodology proposed by Ellis et al. [17]: evidence Type I, II or III. Where several studies were found relating to a given diagnosis±intervention pair, the intervention was as-
signed to a single category (I±III), based on the highest quality of evidence found. The `Ellis categories' imply that some supporting evidence exists. However for some diagnosis±interventions pairs the Medline search yielded no literature at all on either the diagnosis or the intervention in that diagnosis. The evidence was then coded as zero. In some instances the evidence found contraindicated rather than supported the use of a speci®c intervention, and this evidence was coded as `C'. Results A total of 1990 patients seen in the 12 months of 1996 were selected. They represented a total of 4800 consultations, corresponding to 2.4 consultations per patient within one year. A diagnostic code was assigned to all 4800 consultations, and these are shown by broad disease group in Table 1. The most frequently presenting diagnoses were diseases of the circulatory (18.7%), respiratory (14.9%), nervous (14.2%), nutritional, metabolic and immune systems (12.1%), and osteomuscular diseases (12.5%). A series of consultations was often needed before the physician decided on the appropriate intervention. Of the 4800 consultations, both a diagnosis and an intervention were present in only 2341 consultations (49%). The 2341 consultations yielded 2156 (92.1%) unique diagnosis±intervention pairs and 185 (7.9%) non-unique pairs. Table 2 shows the evidence found by Ellis category. Type I evidence was found for 817 (38%) of the diagnosis±interventions pairs, Type II evidence for 76 (4%), and Type III evidence for 292 (14%). No evidence was found for 940 (43%), and the intervention was contraindicated in 18 (0.8%) pairs. Clinical practice was clearly supported by positive evidence in all Types (I±III) in a total of 55% of interventions, and by good positive evidence of Type I or II in 42% of interventions, as re¯ected in Table 2. Discussion Since its formation in 1986, the Spanish National Health System has provided health care (which is free at the point of delivery) for common diseases, and occupational disease and accidents. State health care is provided by the Social Security system, ®nanced by means of employer and employee contributions on one hand, and State subsidies on the other [23, 24]. The National Health System provides service through two health care levels: Primary Care and Specialized Care. The present study focuses on the primary care setting, which is the habitual access route to the public health system. Each area, has one
817 Table 1. Frequency of consultations by broad diagnostic group (ICD-9) ICD-IX codes
Pathology
No. of consultations (n = 4800)
Percentage (%)
95% CI UL±LL
001±139 140±208 210±239 250 280±289 240±279 290±389 393±459 460±519 531±599 600±608 610±629 630±676 680±709 710±739 767±779 780±799 810±999
Infectious diseases Malignant tumors Benign and nonspeci®ed tumors Diabetes Diseases of the blood and hematopoietic organs Nutritional, metabolic and immune disorders Diseases of the nervous system Diseases of the heart and blood vessels Respiratory tract disease Digestive pathology Diseases of the male reproductive apparatus Diseases of the female reproductive apparatus Pathology of delivery and puerperium Diseases of the skin and subcutaneous tissue Osteomuscular disorders Diseases of birth and congenital disorders Poorly de®ned diseases External causes and accidents
107 37 12 175 48 582 682 897 715 579 31 53 2 197 600 3 74 6
2.23 0.77 0.25 3.65 1.00 12.13 14.21 18.69 14.90 12.06 0.65 1.10 0.04 4.10 12.50 0.06 1.54 0.13
2.65±1.81 1.02±0.52 0.39±0.11 4.18±3.12 1.28±0.72 13.05±11.21 15.20±13.22 19.79±17.59 13.98±16.82 12.98±11.14 0.88±0.42 1.40±0.80 0.10±0.00 4.66±3.24 13.44±11.56 0.13±0.00 1.89±1.19 0.23±0.03
UL ± Upper limit; LL ± Lower limit.
or more public Primary Care Centers which service most patient health problems, with referral to more complicated and specialized services where required [25]. The primary care professionals comprise both health care and other personnel (general practitioners, pediatricians, nurses, and nursing and administrative auxiliary personnel, etc.), and their services are dispensed through the network of rural and urban Health Care Centers and Clinics. These services comprise advice and prescriptions, the performance of basic diagnostic tests, health education activities, vaccinations, minor surgical interventions and other measures for the prevention of disease, the promotion of health, and rehabilitation [23]. The present study area was selected because a substantial proportion of the population of Spain
lives in the rural setting, and because the rural population has a well-de®ned pathway through primary care. In contrast, city dwellers frequently present initially through hospital accident and emergency departments. The Spanish rural population is older than the urban population, and consequently the spectrum of diseases shifts towards degenerative disease ± particularly diabetes, metabolic disorders and circulatory disease. The Dubinsky study referred to patients in the hospital setting. Such studies are much easier to perform: the case histories tend to be more complete (particularly as regards diagnostic investigations), and normally establish a clear diagnosis. The use of the ICD-9 classi®cation in this setting posed some diculties, due to the wide variety of clinical terms used in general practice, including terms referring to
Table 2. Quality of the evidence found for all diagnosis±intervention pairs Ellis category evidence type
Recommendation from evidence
Diagnosis±intervention pairs
I II III
Positive Positive Positive Total positive
817 76 292 1185
37.9 3.5 13.5 54.9
C Type I C Type II
Contraindicated Contraindicated
9 9
0.4 0.4
940
43.6
No evidence found Unclear Type III Unclear Type I Total
No. cases
12 1 2156
Total (%)
0.6 0.04 100
818 uncertain diagnoses and early stage disorders ± many of which do not ®t well into the ICD-9 classi®cation. The diculties encountered in using the ICD-9 are not surprising, as this classi®cation has been developed to classify causes of death and hospital admission. A classi®cation system better adapted to the conditions presenting in Primary Care is needed. Emphasis is placed on the fact that disease prevalence was not addressed as such in the present study, but rather the frequency of consultations in the course of a calendar year. The frequency of diseases may be speci®c to the geographical area (eastern Spain) and the rural extraction of our population. Our study population presented mainly with chronic illnesses, particularly of the circulatory system. Metabolic disorders were also common, particularly diabetes and lipid metabolic diseases. It may be that the rural population attaches less signi®cance to lipid disorders, whereas physicians are active in diagnosing such disorders due to the availability of laboratory tests. The evidence basis that we have used (Medline Silver Platter 1992±1996) was limited to ®ve years of literature; as a result, our study may have underestimated the proportion of interventions that are evidence based. Our results, with 55% of interventions being evidence based, are somewhat lower than the 81% found by Ellis [17] in English primary cases but certainly substantially greater than the 21% reported by Dubinsky et al. [16] ± though the latter study was not within primary care. The primary care setting has characteristics and problems of its own, and it is not easy for health care professionals in this setting to stay abreast of the continuous scienti®c advances. These problems are even more apparent in the case of general practitioners who work in the rural setting, where access to scienti®c libraries, electronic databases and computerized support is even more restricted [16]. Because general practice is so broad, general practitioners need better access to a uni®ed evidence base covering medicine, psychiatry, Primary Care and all common diseases. Professional access to scienti®c evidence would increase the use of appropriate treatment by hard pressed general practitioners. Conclusions ± The study sample of 1990 patients represented 20% of the patients seen in the calendar year 1996 by 34 primary care practices in rural CastelloÂn (Valencia), Spain. ± The most frequent presenting conditions were circulatory and respiratory. ± The distribution of disease groups seen in primary care did not re¯ect disease prevalence in the population.
± A mean of 2.4 consultations per patient was required to establish both a diagnosis and an intervention. 49% of consultations resulted in both a diagnosis and an intervention. ± Over 90% of the 2341 diagnosis±intervention pairs were unique, re¯ecting the diversity of primary care. ± For 940 (43%) of diagnosis±intervention pairs, there were no published studies in the literature 1992±1996 referring to the diagnosis. For the remaining 57%, some studies were found in the literature. ± Positive evidence was found in support of 1185 (55%) of diagnosis±intervention pairs, negative (contraindication) evidence for 18 (0.8%) and unclear evidence for 13 (0.64%). ± The majority (38%) of positive evidence was strong (Type I), only 3.5% was Type II and 13% Type III (weak). ± Excluding diagnoses not covered in the ®ve years of literature used as the evidence source 97.4% of interventions were supported by positive evidence, and 67.2% by Type I (strong) positive evidence. ± These results support ®ndings in the UK that the majority of treatment in primary care is evidence based. References 1. Rose G. Environmental Health: Problems and prospects. J R Coll Physicians Lond 1991; 25(1): 48±52. 2. Smith R. The scienti®c basis of health services. Br Med J 1995; 311: 961±962. 3. Cochrane's legacy. Lancet 1992; 340: 1131±1132. 4. Davido F, Haynes B, Sackett D, Smith R. Evidence based medicine. Br Med J 1995; 310: 1085±1086. 5. Sackett DL, Rosenberg WM. The need for evidencebased medicine. J R Soc Med 1995; 88(11): 620±624. 6. Rosenberg WM, Sackett DL. On the need for evidencebased medicine. Therapie 1996; 51(3): 212±217. 7. Michaud GC, McGowan JL, van der Jagt RH, Dugan AK, Tugwell P. The introduction of evidence-based medicine as a component of daily practice. Bull Med Libr Assoc 1996; 84(4): 478±481. 8. Michaud GC, McGowan JL, van der Jagt RH, Wells G, Tugwell P. Are therapeutic decisions supported by evidence from health care research? Arch Intern Med 1998; 158(15): 1665±1668. 9. Fugelli P. Clinical Practice: Between Aristotle and Cochrane. Schweiz Med Wochenschr 1998; 128(6): 184± 188. 10. Saad ED, Grunspun H. Evidence-based medicine [Article in Portuguese]. Rev Hosp Clin Fac Med Sao Paulo 1996; 51(1): 34±36. 11. Galloway MJ, Reid MM. Is the practice of haematology evidence based? I. Evidence based clinical practice. J Clin Pathol 1998; 51(5): 345±346. 12. Reid MM, Galloway MJ. Is the practice of haematology evidence based? II Evidence based morphological diagnosis. J Clin Pathol 1998; 51(6): 417±419.
819 13. James NT. Evidence based medicine. Scienti®c method and raw data should be considered. Br Med J 1996; 313(7050): 169±170; discussion 170-171. 14. Pozo F, Ricoy JR, LaÂzaro P. Una estrategia de investigacioÂn en el Sistema Nacional de Salud: La EpidemiologõÂ a ClõÂ nica. Med Clin (Barcelona) 1994; 102: 664±669. 15. Wennberg JE. Dealing with medical practice variations: a proposal for actions. Health Aairs 1984; 2: 6± 31. 16. Dubinsky M, Ferguson JH. Analysis of the National Institutes of Health Medicare Coverage Assessment. Int J Technol Assess Health Care 1990; 6: 480±488. 17. Ellis J, Mulligan I, Rowe J, Sackett D. Inpatient general medicine is evidence based. The Lancet 1995; 346: 407±410. 18. Eddy DM, Billings J. The quality of medical evidence: Implications for quality of care. Health Aairs 1988; 7(1): 19±32. 19. Benech I, Wilson AE, Dowell AC. Evidence-based practice in primary care: Past, present and future. J Eval Clin Pract 1996; 2(4): 249±263. 20. Gill P, Dowel AC, Neal RD, Smith N, Heywood P, Wilson AE. Evidence based general practice: A retrospective study of intervention in one training practice. Br Med J 1996; 312: 819±821.
21. Generalitat Valenciana. ConsellerõÂ a de Sanitat i Consum. Encuesta de Salud de la Comunidad Valenciana 1990±1991. Valencia: Generalitat Valenciana. ConsellerõÂ a de Sanitat i Consum 1993. 22. Instituto Nacional de Salud. Ministerio de Sanidad y Consumo. Clasi®cacioÂn Internacional de Enfermedades IX revisioÂn. Modi®cacioÂn ClõÂ nica. Madrid. Instituto Nacional de la Salud. 1989. 23. Plan de Salud. Madrid: Ministerio de Sanidad y Consumo, 1995: 34±37. ISBN: 84-7670-442-9. 24. LimoÂn MendizaÂbal, C. Prestaciones sanitarias en los paõÂ ses de la C.E.E.E. Madrid: Ministerio de Sanidad y Consumo, 1994: 14±15, 50±51. ISBN: 84-7670-379-1. 25. Almenar Roig FD. UtilizacioÂn de Servicios de Asistencia Primaria en la Seguridad Social en la Comunidad Valenciana. Valencia: DireccioÂn Provincial del Instituto Nacional de la Salud, 1986: 13±15. ISBN: 847579-109-3. Address for correspondence: Dr M.M. Morales SuaÂrez-Varela, Unit of Public Health, Hygiene and Environmental Health, Valencia University, Avda, Vicent AndreÂs EstelleÂs s/n 46100 Burjassot, Valencia, Spain Phone/Fax: +34 6 3864951 E-mail:
[email protected]
