Physicians’ Knowledge of and Compliance With Guidelines
An Exploratory Study in Cardiovascular Diseases
; ; ; ; ;
Background: Guidelines are one of the means by which health care organizations try to improve health care and lower its cost. Studies have shown, however, that guidelines are still not being adequately implemented. In this exploratory study, we examine the link between physicians’ knowledge of and compliance with guidelines: specifically, guidelines for the treatment of three cardiovascular diseases (arterial hypertension, heart failure and chronic coronary heart disease [CHD]) in primary care.
Methods: We assessed primary care physicians’ knowledge of the guidelines with a representative postal survey, using a questionnaire about the treatment of cardiovascular diseases (2500 questionnaires sent). We assessed the responding physicians’ compliance with the guidelines by analyzing patient data from a sample of 30 of them for various indicators of compliance. Of these 30 physicians, 15 met our operational criteria for adequate knowledge of the guidelines, and 15 did not.
Results: 437 (40%) of the physicians knew the guidelines adequately. Physicians answered questions about chronic CHD in accordance with the guidelines more often than they did questions about arterial hypertension (74% versus 11%). Our exploratory analysis of guideline compliance revealed that physicians who knew the guidelines adequately performed no differently than physicians who did not with respect to 12 of the 16 compliance indicators. As for the remaining 4 compliance indicators, it turned out, surprisingly, that physicians who did not know the guidelines adequately performed significantly better than those who did.
Conclusion: These preliminary findings imply that physicians’ knowledge of guidelines does not in itself lead to better guideline implementation. Further studies are needed to address this important issue.
The introduction of guidelines for medical care is one of many strategies adopted by various health care organizations to deal with quality-related and economic deficits in health care provision. To date, however, the goal—patient care conforming to guidelines—has not been attained. Numerous national and international studies have shown that guidelines are still not being adequately implemented in practice (1–3), despite physicians’ proven acceptance of guideline-oriented, evidence-based medicine (4–8).
With regard to cardiovascular diseases, investigations show a wide variation in physicians’ knowledge of the recommendations embodied in guidelines (9–11). The German follow-up of the Hypertension Evaluation Project clearly demonstrated inadequate knowledge of the diagnosis and treatment of arterial hypertension (9, 12). There is also evidence of deficiencies in treatment quality that appear to be due, among other factors, to inadequate implementation of existing cardiological treatment recommendations (13–15).
The translation of guideline recommendations into concrete medical practice is a complex process in which physician-related, patient-related, guideline-related, and educative factors all play a role (16, 17). Little research has yet been conducted to ascertain the extent to which knowledge of prevailing guidelines for diagnosis and treatment affects how physicians manage their patients (18).
The aim of this exploratory study was to investigate the link between physicians’ knowledge of and compliance with guidelines in the primary care of three cardiovascular diseases—arterial hypertension, heart failure, and chronic coronary heart disease (CHD). Primary care physicians’ knowledge of guidelines was assessed by means of a representative postal survey. In a group of 30 responders we evaluated adherence to guidelines on the basis of data from patients’ records (Figure 1 gif ppt).
Assessment of physicians’ knowledge of guidelines
We cooperated with an interdisciplinary expert advisory group to develop a questionnaire to assess primary care physicians’ knowledge of guidelines on the basis of the existing published recommendations for the diagnosis, treatment, and follow-up of the three selected diseases (e1–e6). (Following publication of Guideline No. 9 on heart failure by the German College of General Practitioners and Family Physicians [DEGAM] in 2006 and Version 1.5 of the German National Disease Management Guidelines on CHD in 2007, the questionnaire was inspected for discrepancies. None were revealed.)
The questionnaire on treatment of cardiovascular diseases contained a total of 15 multiple-choice questions, five on each of the three diseases. The questions were primarily patient-oriented and related to the primary care physicians’ treatment strategies. Only one of the possible answers to each question corresponded to the guidelines’ recommendations.
The scoring system for analysis of the responses comprised a quantitative aspect (number of answers complying with the guidelines) and a qualitative aspect (three cardinal questions). A physician was judged to know the guidelines adequately if his/her answers to 10 of the 15 questions, including three questions identified as particularly important by the expert advisory group, corresponded to the guidelines.
After preliminary testing, the questionnaire was sent to 1250 members of the Association of Statutory Health Insurance Physicians (ASHIP) in each of two parts of Germany, the region of North Rhine and the state of Saxony. These physicians were randomly selected from the registers of the two associations. The data were pseudonymized in accordance with the regulations on data protection, and the study was approved by the ethics committee of the Faculty of Medicine, University of Cologne.
The data were statistically analyzed with the aid of the program SPSS 18.0. Relative and absolute frequencies were calculated and a logistical regression model was computed. A detailed account of the methods used can be found online (eSupplement).
Assessment of physicians’ compliance with guidelines
The fundamental requirements for determining how closely physicians were adhering to guidelines included the physicians’ willingness to participate, a sufficient number of patient data sets, and identification of suitable indicators.
► Recruitment of participating physicians—All physicians who responded to the postal survey on treatment of cardiovascular diseases (n = 1152) were asked in writing whether they would be prepared to take part in a further survey to be carried out at their offices. The aim was to recruit a total of 30 physicians’ offices (ASHIP North Rhine: 15 offices—50% urban, 50% rural; ASHIP Saxony: 15 offices—50% urban, 50% rural). The office survey comprised an office visit by a physician from the research team and, on data protection grounds, structured collection of patient data by specially schooled members of the office staff. The participating primary care physicians received 15 euros per patient included in the survey. A maximum of 50 patients per physician could be documented. The office survey began in November 2007.
► Acquisition of patient data—In each primary care physician’s office, the patient data were collected by a member of staff who had been instructed by a physician from the research team. Altogether, data on 1500 patients were expected. The patients eligible for the survey were those aged 40 years or over who attended the physician’s office for any reason on one of three predefined days. The pseudonymized questionnaire on knowledge of guidelines enabled each patient to be classified according to the treating physician’s familiarity with guidelines. The data were evaluated with the aid of the statistical software SPSS 18.0; analysis included calculation of relative and absolute frequencies (Table 1 gif ppt).
► Development of an indicator set—The technique used to develop a catalog of indicators for evaluation of physicians’ adherence to guidelines on the basis of the sampled patient data was oriented on the RAND method (19). The factors taken into consideration were the treatment principles deemed relevant by the expert advisory group together with central guideline recommendations and the evidence on which these were founded. The indicators that were considered related to the medications prescribed, diagnostic procedures, and clinical parameters of treatment success. Final selection of indicators for the catalog was based on repeated anonymous written opinions from the experts on the basis of relevant criteria (20–22). Evaluation of the indicators according to the quality criteria of the QUALIFY instrument (23) was not yet possible at the time our study was carried out. However, the methods overlap. A total of 16 agreed indicators were employed.
► Indicator-guided analysis of patient data— Indicator-related evaluation of the patient data to determine physicians’ adherence to guidelines was performed with the aid of the statistical program SPSS 18.0. Absolute and relative frequencies were calculated to reveal, for each indicator, what proportion of the patients had been treated according to guideline recommendations. We then carried out chi-square-based linkage analysis to identify any significant differences in distribution (Did the patient data fulfill the requirements of the given indicator more frequently when they came from the office of a physician with adequate knowledge of guidelines as defined by the operational criteria?). Furthermore, the data were considered descriptively with regard to individual risk constellations and their consequences for treatment (for example: Was medication prescribed for patients with documented hypertension?).
Data from patients aged 80 years or over were excluded from indicator-guided analysis. On the one hand the evidence for guideline recommendations regarding this age group is not always adequate, and on the other hand their doctors may deviate from standard recommendations because of multimorbidity in elderly patients.
Representative survey of physicians’ knowledge of guidelines
The survey was sent to 2500 primary care physicians, 1152 of whom returned questionnaires suitable for analysis. Excluding 31 surveys sent to incorrect addresses, the effective response rate was therefore 47%. The responders were analyzed by ASHIP membership, sex, specialization, and experience (Table 2 gif ppt). With regard to sex distribution and specialization, our sample of physicians did not differ essentially from the overall ASHIP membership in North Rhine and Saxony.
On average, 11 of the 15 questions on treatment of cardiovascular diseases were answered in accordance with guidelines. Three physicians answered only five questions in accordance with guidelines; 28 (3%) answered all 15 questions in agreement with guideline recommendations. The questions on hypertension were answered in accordance with guidelines by 11% of the responders. A far higher proportion (74%) gave answers to the questions on chronic CHD that conformed adequately to guideline recommendations. According to the operational criteria for this study, 437 (40%) of responding physicians demonstrated adequate knowledge of the guidelines, while 695 did not (Table 2). Logistic regression analysis revealed an insignificant impact of sociodemographic characteristics on physicians’ knowledge of guidelines (McFadden’s pseudo-R2: 0.027).
Exploratory survey of physicians’ compliance with guidelines
A total of 76 physicians’ offices (7%) offered to participate in the office survey. On grounds of resource availability and practicability, 15 offices in and around Cologne and Bonn (North Rhine) and 15 offices in and around either Dresden or Leipzig (Saxony) were selected. After exclusion of the three pre-test offices, data from 27 offices were acquired for analysis.
Comparison of the physicians recruited for the office survey with those who completed the postal questionnaire showed higher proportions of women (52% vs 43%), physicians with their own office for less than 2 years (15% vs 7%), and physicians with their own office for 10 to 14 years (30% vs 17%). The physicians who took part in the office survey answered an average 11 out of the 15 questions in the postal survey in accordance with the guidelines and thus did not differ from the total sample. One participant in the office survey answered seven questions (minimum) and one answered 14 questions (maximum) in accordance with guideline recommendations. Adequate knowledge of the guidelines according to the operational criteria was displayed by 48% (n = 13) of those who took part in the office survey, a slightly higher proportion than in the total sample (Table 3 gif ppt).
The 27 physicians’ offices collected data on the treatment of 1318 patients. Data on 21 patients were excluded because their age was below 40 years or not stated. The effective response rate was therefore 96% (n = 1297). After exclusion of those over 79 years of age, treatment data for 1107 patients were eligible for analysis.
More than half of the patients were women (n = 592), and 64% (n = 706) of them were aged 60 years or over. The prevalence of the three target diseases varied considerably among the offices surveyed. Overall, 68% of the patients (n = 749) had arterial hypertension, 24% (n = 269) had CHD, and 7% (n = 79) had heart failure. Forty-two percent of the patients (n = 464) had two or more of the following diagnoses: hypertension, CHD, heart failure, kidney failure, diabetes mellitus. Forty-eight percent of the patients (n = 536) were treated by a physician with adequate knowledge of the guidelines as defined here. The physicians whose knowledge of the guidelines was inadequate according to the operational criteria tended to have documented more patients aged 50 to 59 years (145 vs 93 patients) and fewer patients in the 70 to 79 years age group (148 vs 204 patients). The prevalence of the three target diseases showed no essential differences between physicians with and those without adequate knowledge of guidelines. The number of patients with at least two of the five diseases mentioned above was higher, though not significantly so, for the physicians who had adequate knowledge of the guidelines (243 vs 221 patients) (Table 4 gif ppt).
Indicator-guided analysis shows the proportion of patients whose data reveal that they were treated in adherence to the guidelines. This proportion varied from 46% to 94% for the five hypertension-related indicators, from 30% to 87% for the seven heart failure-related indicators, and from 64% to 100% for the four CHD-related indicators. Chi-square-based linkage analysis showed no significant differences in distribution (p ≤ 0.05) for 12 of 16 indicators. In these cases the distribution of the individual indicators (patient data show treatment adhering to guidelines: yes/no) did not differ between the physicians with and those without adequate knowledge of guidelines. For the remaining four indicators (nos. 4, 5, 7, and 11) the patient data showed significant differences between the two groups of physicians: the proportion of patient data from which treatment complying with guidelines could be concluded was higher among physicians who did not know the guidelines adequately (Table 5 gif ppt).
Exploratory analysis of the treatment data with regard to individual risk constellations and their consequences for treatment was severely limited by the low number of patients, but it does point to insufficient treatment intensity. In 52% (n = 367) of hypertension patients receiving drug treatment, for instance, blood pressure had not been restored to normal. Of these 367 patients, 23% (n = 84) were receiving monotherapy and 33% (n = 120) were being treated with two drugs in combination. Patients with diabetes and hypertension benefit from treatment with ACE inhibitors, but 55 (19%) of the 289 diabetics with high blood pressure in this study were not being prescribed ACE inhibitors. Fourteen patients also had documented CHD; six of these were receiving neither ACE inhibitors nor a beta-blocker.
Compared with other surveys of primary care physicians, the response rate to our questionnaire was high at 47% (24, 25, e7, e8). With regard to previous similar studies (12), our representative findings indicate increasing awareness of the content of guideline recommendations for the treatment of cardiovascular diseases. Nevertheless, there is considerable potential for improvement in 60% of doctors providing primary care.
Contrary to what is generally assumed, exploratory analysis of the selected indicators showed no essential differences in treatment between physicians with adequate knowledge of the guidelines and those who were less familiar with current guideline recommendations. This was confirmed in descriptive comparisons of the treatment data in extreme groups (comparison of the three “best” and the three “worst” physicians; comparison of patient data from physicians who answered the question regarding the definition of hypertension in accordance with the guidelines and those who did not). It therefore seems that physicians’ familiarity with guidelines does not, as previously thought, translate into better implementation of the guidelines in daily practice. In this light, purely cognitive strategies to improve the quality of health care need to be reconsidered.
The treatment of patients with particular risk constellations was not a primary consideration of this study. For this reason the case numbers were low and do not permit the conclusion of generalized failure to make proper allowance for risks in patient treatment. Taken together with the results of other investigations (e9–e11), however, the signals generated here can be interpreted as showing that treatment decisions are based less on medical data than on other practice-relevant factors. These factors may include the internal organizational routines of the physician’s office, financial parameters, or patient-related aspects. Differentiation of these facultative relationships was not a goal of this project, neither is it feasible.
The interpretation of the findings is limited by the following factors:
- The results are based on a representative survey of physicians’ knowledge of guidelines and on an exploratory investigation of physicians’ adherence to guidelines. Their validity is limited by the cross-sectional design and by the sometimes low numbers of cases in the exploratory part of the study. Longitudinal studies are therefore essential.
- Knowledge of the guidelines was not tested directly; rather, theoretical treatment strategies were selected with the aid of case-oriented questions. An agreed scoring system was then used to divide the physicians into two groups—“adequate” and “inadequate” knowledge of guidelines—on the basis of their responses.
- The physicians’ adherence to the guidelines was ascertained by means of indicator-guided analysis of patient data. On data protection grounds, these data were collected by a specially schooled member of staff in each doctor’s office. Errors in data acquisition cannot be completely ruled out.
- The results of indicator evaluation and the findings on adaptation of treatment to allow for individual risk constellations are based on an exploratory investigation with sometimes very low numbers of cases.
- Financial restrictions did not permit discussions with the physicians to find out why they deviated from the guidelines in concrete individual cases (patient’s preferences, quality of life, prioritization in the case of multimorbidity).
Although the above-mentioned practical constraints do not permit any conclusions regarding the causal link between physicians’ knowledge of and compliance with guidelines, our results point the way for future research. Increased efforts should be made—as has already happened in some integrated care projects—to implement guideline recommendations as in-process control variables in standardized software. This is particularly important because it cannot be excluded that feedback from targeted, controlled action leads to a secondary gain in knowledge. This could be the reason why the responses to the questions on CHD, for which there is a Disease Management Program in Germany, were in accordance with the guidelines much more frequently than was the case for arterial hypertension. These incompletely resolved questions should form the stepping-off point for future investigations.
The authors thank the members of the expert advisory group—Dr. Sigrid Eufinger, Dr. Joachim Feßler, Prof. Markus Flesch, Dr. Ady Osterspey, Prof. Christoph Pohl, Prof. Gernot Waßmer, Dr. Karl-Gustav Werner—for their assistance and are grateful to all those who took part in the survey. Special thanks are due to all participating primary care physicians and their office staff who willingly gave their time; without their commitment this project could not have been completed.
The authors are also grateful to the German Medical Association for its support (project no. 06–39) and to their mentor for this project, Prof. H.-K. Selbmann.
Conflict of interest statement
The Primary Health Care Research Group (PMV forschungsgruppe) receives project support from statutory health insurers (Federal Association of the AOK, AOK Hesse, AOK Plus, AOK Baden-Württemberg, LKK Baden-Württemberg), ministries (Federal Ministry of Education and Research, Federal Ministry of Health, Hessian Ministry of Social Affairs), foundations (Boll Foundation, Lesmüller Foundation), and pharmaceutical companies (Sanofi-Aventis, Sanofi Pasteur, MSD, Bayer-Schering, Novo Nordisk, Abbott, Janssen-Cilag, Merz). I. Schubert has received no personal honoraria. The remaining authors declare that no conflict of interest exists according to the guidelines of the International Committee of Medical Journal Editors.
Manuscript received on 28 January 2010, revised version accepted on
9 November 2010.
Translated from the original German by David Roseveare.
Dr. rer. pol. Ute Karbach
50933 Köln, Germany
@For eReferences please refer to:
eSupplement available at:
diagnosis and treatment of dyslipidemia in the primary care setting in Germany. Atherosclerosis 2007; 190(2): 397–407.
evidence-based drug treatment in patients with heart failure is only partially explained by dissimilarity to patients enrolled in landmark trials: a report from the Euro Heart Survey on Heart Failure.
European Heart Journal 2005; 26(24): 2706–13.
Sage Publications 1998.
ZUMA-Arbeitsbericht 1998; 02: 1–45.
Kölner Patientenfragebogens durch kognitive Pretest-Techniken. In: Pfaff H, Freise DC, Mager G, Schrappe M (Hg). Der Kölner Patientenfragebogen (KPF): Entwicklung
und Validierung eines Fragebogens zur Erfassung der Einbindung des Patienten als Kotherapeuten. Sankt Augustin: Asgard-Verlag 2003; 91–100.
ZUMA-Nachrichten 1996; 39: 95–116.
Herzzentrum der Universität zu Köln, Klinik III für Innere Medizin: Dr. med. Hagemeister, Prof. Dr. med. Höpp
PMV forschungsgruppe, Klinik und Poliklinik für Psychiatrie und Psychotherapie des Kindes und Jugendalters, Universität zu Köln: Dr. rer. soc. Schubert