Objective: The Chronic Care Model (CCM) and the 5A approach have achieved widespread acceptance and reflect the core elements of patient-centred care in chronic diseases, including osteoarthritis (OA). The aim was to assess to what extent current care for patients with osteoarthritis accords with the CCM in Germany and to reveal possible predictors to assess whether certain patients are more likely to receive care complying with the CCM than others.
Methods: Cross-sectional observational study, addressing 1250 patients from 75 primary care practices in Germany. 1021 (81.7%) of the administered 1250 questionnaires were returned. The Patient Assessment of Chronic Illness Care (PACIC-5A) was used to assess accordance to the CCM and the 5A-approach. The impact of OA was assessed by means of the Arthritis Impact Measurement Scale (AIMS2-SF); the Patient Health Questionnaire (PHQ-9) was used to assess depression symptoms. Two stepwise multiple linear regression models with the PACIC sum score and the 5A score as dependents were performed to reveal predictors of a high accordance to the CCM and to the 5A approach, respectively.
Results: With a mean (SD) of 2.79 (0.83) in men and 2.67 (0.89) in women (p for difference = 0.89), the PACIC sum score was notably lower than in previous studies conducted in health maintenance organisation settings in the US. The PACIC score was associated with a higher educational level (β = 0.421; p = 0.008) and younger age (β = −0.319; p = 0.016); the 5A score was predicted by educational level (β = 0.344; p = 0.002), age (β = −0.386; p = 0.004) and the PHQ-9 score (β = −0.288; p = 0.005).
Conclusions: Younger and better educated patients achieve higher scores on the PACIC score, indicating that their care accords to a higher degree with the CCM. Whether this reflects differences in physician behaviour toward different patient groups or rather different demands of these patient groups cannot be concluded from the presented data. Further research is needed to confirm our results and assess possible implications for implementing the Chronic Care Model in primary care.
Statistics from Altmetric.com
Chronic diseases like diabetes, hypertension, arthritis and asthma are expected to increase tremendously in the years to come.1 They are associated with high economical costs but also with a high burden on individuals’ quality of life. Multiple interventional studies have been conducted to increase the quality of life of patients with various chronic diseases. Based on the evidence of these interventions, Wagner et al have developed the Chronic Care Model (CCM) as a conceptual framework.2–4 Current care for the chronically ill is often event-driven, despite solid evidence that a structured, planned and proactive approach to chronic diseases helps to reduce the burden of many chronic diseases.5 The CCM contains six key dimensions of care: organisation of healthcare, clinical information systems, delivery system design, decision support, self-management support and community resources. A recent review of the literature reiterates that successful improvement strategies concerning chronic diseases are consistent with the concept and components of the CCM.6 The CCM has achieved widespread acceptance, and recently an intense discussion has started among German physicians as to if and how the CCM or its components can be implemented in healthcare.78 At the annual meeting of the German Association of Family Medicine (DEGAM) in 2006, a statement was made reflecting the intention to implement the CCM or at least some elements thereof in primary care. Nevertheless, no data are available to date showing to what extent current primary care for the chronically ill is congruent to the CCM. To assess the congruency of provided healthcare to the CCM, Glasgow et al developed the “Patient Assessment of Chronic Illness care” (PACIC).9 It is organised according to the key elements of the CCM and assesses the behaviour of professionals and practice teams from a patient’s perspective. The PACIC contains 20 items assessing five scale constructs: patient activation, delivery system design/decision support, goal setting/tailoring, problem solving/contextual, follow-up/coordination. “Patient activation” assesses to what extent the patient was motivated and supported by the physician to initiate changes, and “decision support” assesses if the patient was supported, for example, by booklets and how satisfied they were with the organisation of their care. “Tailoring” assesses to what extent general instructions and suggestions were adapted to his personal situation. “Problem solving” addresses how the physician dealt with problems which interfered with achieving predefined goals. Finally, “follow-up” addresses how frequently and consequently the whole process was followed up. Recently, a German version of the PACIC 5A has been validated in a sample of osteoarthritis patients.10 Its psychometric properties have been reported in detail elsewhere:10 The Cronbach alpha ranged from 0.78 to 0.90, and the test–retest validity, estimated by the intracluster correlation coefficient (ICC), was at least 0.77.
The “5A” model represents an evidence-based approach to induce a behavioural change.11 Originally developed for smoking cessation, it represents the recommended counselling approach for behavioural changes according to the recommendations of the US Preventive Services Task Force (USPSTF). The key elements are assessment of present behaviour (Assess), patient counselling (Advise), collaborative agreement with the patient about realistic goals (Agree), assisting the patient during his lifestyle changes (Assist) and frequent follow-ups (Arrange).12 Glasgow et al expanded the PACIC by including six items assessing to what extent physicians’ counselling reflects the 5A approach. They validated the PACIC-5A in a sample of diabetes patients.12
The aim of this study was to assess the congruency between care in general practice in Germany and the CCM as well as between general practitioners’ (GPs) counselling style and the 5A approach. Since we assumed that care and counselling may differ according to sociodemographic or disease-related characteristics, we aimed at revealing possible predictors of the PACIC-5A score.
Recruitment of patients and data collection
The PraxArt-project, performed by the University Hospital Heidelberg and financed by the German Ministry for Education and Research (BMBF), intends to improve the quality of life of patients suffering from osteoarthritis.13 A total of 1021 OA patients from 75 primary care practices in Germany are currently enrolled in this project. The 75 practices are located across the federal state of Baden-Württemberg and are a representative sample of German primary care practices—for example, most of them are single-handed. The anonymous questionnaires are linked within the project via a code list to the medical file of the practices. Thus, detailed data about sociodemographic variables, duration of OA as well as information about comorbidities, medication and health services utilisation were available and considered in the analyses. Since GPs prepared a list of all addressed patients, these data were available for respondents as well as non-respondents. Inclusion criteria were aged over 18 years, meeting the criteria of OA to the hip or knee according to the American College of Rheumatology (ACR),1415 and with sufficient German-language skills for understanding and answering the questionnaire. In all 75 practices, GPs consecutively asked patients enrolled in the PraxArt-project to answer the PACIC-5A. After giving their written informed consent, patients received the questionnaire and a return envelope to the university. Patients were informed that there was no possibility of their GP knowing about their answers and were asked to complete the questionnaire on their own. The inclusion of patients did not start unless there was a written and unrestricted positive vote of the ethics committee of the University of Heidelberg which was received in March 2005.
Depression was measured using the depression module of the German version of the Patient Health Questionnaire (PHQ-9).1617 The PHQ-9 is a completely self-administered questionnaire that enables screening for depression and assessment of depression severity. The PHQ-9 score ranges from 0 to 27, where higher scores indicate a worse health status. The impact of OA on patients’ health was assessed by the AIMS2-SF, which provides a comprehensive assessment of patients’ health status including the dimensions physical limitation, symptom (reflecting perceived pain), social (reflecting social contacts), affect (reflecting mood) and work (reflecting the ability to work). It has recently been validated in German using a sample of OA patients and proved to be comparable with the original version regarding reliability and validity.1819 The AIMS2-SF dimensions score between 0 and 10, with 0 representing the best and 10 representing the worst health status. Based on clinical relevance in primary care, the following comorbid conditions were collected from the medical record: high blood pressure (defined as >140/90 mm Hg), diabetes, chronic heart failure, coronary vessel disease (CVD), elevated cholesterol level (total cholesterol >200 mg/dl), ulcer or stomach disease, asthma/chronic obstructive pulmonary disease (COPD), renal insufficiency, (prior) cancer and (prior) stroke. Educational level was assessed on five (nearly equidistant) stages: (1) no school degree, (2) basic degree (⩽7 years of education), (3) ⩽10 years of education, (4) college degree, (5) university degree.
Data were transferred into the SPSS program (version 14.0) after questionnaires were scanned. Scores were calculated according to the scoring instructions for the PACIC-5A questionnaire, yielding values between 1 and 5 for each scale. Higher scores represent a higher congruency to the CCM. Descriptive analysis included mean and standard deviation. The intracluster correlation coefficients (ICCs) for each PACIC-5A scale were calculated to estimate the variation within the clusters and to choose the appropriate regression model.20 After calculating bivariate correlations, stepwise regression analyses were performed to reveal predictors of a high PACIC sum score and a high 5A score.
A total of 1021 (81.7%) of the administered 1250 questionnaires were returned. An analysis of the non-respondents revealed no significant differences with respect to sociodemographic data, OA duration and comorbidities in comparison with patients who returned the questionnaires. Table 1 displays characteristics of the study sample. Women were over-represented (66.0% of respondents), an effect most likely related to the prevalence of OA being nearly twice as high among women as among men. The mean duration of osteoarthritis was 14.9 (SD 14.3) years.
Table 2 shows the descriptive statistics of the individual scales of the PACIC-5A scores separated for gender. The average overall score of the PACIC was 2.79 (SD 0.83) in men and 2.67 (0.89) in women. There was adequate variability in the overall scale and all subscales, as indicated by the standard deviation. Significant differences between men and women occurred in the “problem solving” scale of the PACIC and the “agree” scale of the 5A.
Correlations of the PACIC sum score and the 5A score to patient characteristics, PHQ-9 and AIMS2-SF scores are displayed in table 3. Notable correlations were found for age, educational level and the PHQ-9 score. Interestingly, the correlations for the AIMS2-SF scales were all statistically significant but, with the exception of the affect scale, quite low. All factors achieving significance were entered in the regression model.
To decide which regression model would be appropriate, we first calculated the ICCs of the PACIC-5A to estimate the variation between the different practices. Interestingly, the ICC was below 0.01, and so we decided to choose a linear regression model without considering the cluster effect. Table 4 displays the results of the regression analysis with the PACIC sum score as dependent variable. As can be seen, only two factors remained in the final model, which explained 29.5% of the variation in the PACIC sum score. The educational level was the strongest predictor with a β of 0.421 (p<0.008), reflecting that a higher educational level predicted higher scores on the PACIC score. The relationship to age was inverse: a higher age predicted lower PACIC sum scores, reflected in a β of −0.319 (p = 0.016). The PHQ-9 score was eliminated in the last step of the regression model while slightly surpassing the demanded significance level (β = 0.107; p = 0.057)
Interestingly, the regression model revealed similar predictors for the 5A sum score as for the PACIC score: age as a negative predictor of high 5A scores and educational level as a positive predictor (table 5). Additionally, the PHQ-9 score remained a significant predictor in the final model. Higher PHQ-9 scores were associated with lower 5A scores.
The CCM has been promoted as a template of care for the chronically ill, aiming at substantially improving quality of life.421 Our study showed that certain patients rated those aspects of their care which were consistent with the CCM more favourably. Being younger, better educated and less depressed increased the chance of achieving higher scores on the PACIC-5A. Assuming that the PACIC-5A reflects the accordance of physicians’ behaviour with the CCM, these patients are more likely to receive care that contains the core elements of chronic care as activation, support, goal setting, assistance and a frequent follow-up.
Tsai et al showed in their meta-analysis that interventions containing at least one CCM element could improve clinical outcomes as well as patient-relevant outcomes.22 They included 112 studies, assessing diabetes, asthma, chronic heart failure and depression. Furthermore, a number of studies are available to date that focused on implementing at least a few elements of the CCM. The results emphasise that the CCM is a theoretical framework that improves process parameters as well as patient-relevant outcomes. In a recently published study, Vargas et al also showed that the CCM approach can reduce risk factors for a heart disease in diabetes patients.23 Regarding chronic obstructive pulmonary disease, Adams et al reported in a recent review that patients who received interventions with two or more CCM components had lower rates of hospitalisations and emergency/unscheduled visits, and a shorter length of stay in a hospital compared with control groups.24 It should be considered, however, that most of these data were retrieved in hospital settings or HMOs and can not easily be transferred to primary care settings. So far, only one study is available, showing that CCM elements can be implemented in small independent practices and result in improved care for diabetics.25
Our paper has three main findings. First of all, the comparison of the scores of our study sample with previously collected data suggests that current care for patients with OA only poorly reflects the key elements of the CCM. Glasgow et al, for instance, found notably higher values among diabetes patients as we did: their reported means in the “tailoring/goal-setting” scale (3.1 in men and 3.0 in women), in the “follow-up” scale (2.9 in men and 3.0 in women) and in the “problem-solving” scale (3.4 for both gender) do not even fall within the 95% CIs of our outcomes.12 The same result applies to the sum score (3.2 in men and 3.2 in women). The results regarding the 5A approach are quite similar, with notably lower scores regarding all scales of the 5A part of the PACIC-5A.
In our opinion, there are several reasons that account for the significantly lower scores in our study. First of all, Glasgow collected his data in an HMO and not, as we did, in a primary care setting with a large number of single-handed practices. Furthermore, and related to the first reason, care according to the CCM is proactive, focused on activating, involving and accompanying the patient. This kind of care frequently requires the involvement of qualified practice nurses or physicians’ assistants. However, these medical professions are currently not available in Germany.26 Another important reason may be the observed disease itself. It could be assumed that physicians’ engagement in diseases such as diabetes, heart insufficiency or depression is higher than in OA, which is regarded as less threatening to patients’ health or at least associated with a lower burden of disease.27
The second important finding is that the PACIC as well as the 5A scores are not correlated with severity of OA. None of the AIMS2-SF scales, reflecting different aspects of quality of life of OA patients, was significantly correlated with the PACIC or the 5A scales, and none of them remained in the regression model. This suggests that GPs’ counselling efforts are not dominated by the disease itself.
The most important finding is related to the predictors of high PACIC and 5A scores. The finding that younger, better educated patients with lower PHQ-9 scores are more likely to achieve high scores on the PACIC-5A could reflect differences in physician behaviour towards different patient groups as well as the fact that these patients are more actively seeking care that complies with the CCM. Still, this association cannot be concluded from our data and remains speculative. Nevertheless, this information is valuable, since it may suggest that in implementing the CCM or its elements, it will be of great importance to ensure that all patient groups are able to benefit to the same extent from this advance in chronic illness care. It is well known that most chronic diseases such as diabetes and high blood pressure are associated with social status;2829 a recent German health survey confirmed the association of chronic conditions and social status once more.30 It is quite obvious that these patients would benefit the most from advanced approaches like the CCM. Our data suggest that efforts in implementing CCM elements should consider ensuring that these patients will eventually benefit from the CCM as well as other patients.
Since Glasgow et al could not reveal significant differences in the PACIC scores regarding gender, ethnicity, income and comorbidities, it will also be of great importance to consider the setting in which the data were collected:912 their patients were enrolled in an HMO, ours in a primary care setting.
Strengths and weaknesses
There are several weaknesses of our study which should be acknowledged: first of all, the assumed linkage between different PACIC scores and differences in received care has not empirically been proved. However, preliminary data from a nationwide research project using the PACIC-5A in evaluating diabetes disease-management programmes confirm this assumption. Second, the data were derived from a cross-sectional observational study within a sample of OA patients, and it remains unclear whether the findings can be transferred to other diseases and patient groups. Furthermore, the social situation of participants could not be assessed properly, since it is still problematic in Germany to ask for the annual income. Since everyone has a health insurance in Germany, and the healthcare system can be freely accessed by everyone, this weakness may be limited. The strength of this study is its reasonably large and representative sample of primary care practices and patients. In contrast to previous studies, which assessed the congruency to the CCM by a health professionals’ perspective (by means of the Assessment of Chronic Illness Care (ACIC)31), we assessed it from a patient’s perspective.
Without a doubt, the framework of the CCM represents an important step towards an improved care for the chronically ill. Our findings suggest that currently, this structured approach is only rarely implemented in the care for patients with OA in a primary care setting. The finding that younger and better educated patients are more likely to receive care which complies with the CCM suggests that the implementation of CCM or CCM elements in primary care will be challenging to assure that all patients benefit equally. Further research is needed to confirm our results and assess possible implications for implementing the Chronic Care Model in primary care.
The authors wish to thank all participating doctors and patients.
Competing interests: None.
Funding: The study was financed by the German Ministry of Education and Research (BMBF), grant number 01GK0301.
Ethics approval: Ethics approval was obtained from the ethics committee of the University of Heidelberg.
Patient consent: Obtained.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.