Quality in Primary Care Open Access

family practice, health priorities, multi-morbidity, older people, patient-centred care.

How this fits in with primary care

What do we know?

There is evidence that GPs have difficulties dealing with the multiple problems of older patients at the same time as balancing priorities for treatment. Little is known about which problems are important to patients and which to their GPs, and whether patients and doctors agree on priorities.

What does this paper add?

Patients and their GPs have poor agreement on the priority of individual health problems. Patients find different types of problems more important than their GPs. Patients and their GPs share a different concept of importance: whereas unfavourable prognosis predicts a problem’s priority most strongly for doctors, for patients it is their emotional experience of a problem. In order to integrate the multiple health needs of older patients into a holistic care plan, priority setting will be essential. This will need to involve transparent communication between patients and doctors, sharing their different views on how problems are prioritised.

Introduction

The growing proportion of older people with multiple health conditions prompts the search for new consul-tation approaches. Consultations usually focus on one or two agenda items, each requiring its own treatment regimen. This can lead to overburdened medication regimens when patients repeatedly consult and treat-ments are added with each contact.[1] Recent healthcare advances, such as disease management programmes and guidelines, are tools that intensify the focus on separate disease managements thereby risking patient safety when recommendations are added together.[2,3]

Patients with multimorbidity require a consultation approach that is comprehensive, holistic and patient centred.[4] A comprehensive approach for older patients means that a doctor gains a health overview, e.g. by utilising a standard geriatric assessment.[5] Holistic care requires attention to the interconnectedness and com-plexity of patients’ problems and their management.[6] Person-centredness involves focusing on the patient, acting in partnership and setting priorities.[7] Priority setting is necessary, whenever a great number of health problems compete for attention.[4] If patients’ health priorities and doctors’ treatment priorities are to lead to a common treatment plan, agreement is essential. Lack of agreement is generally recognised as damaging to patient compliance, to health outcomes and to the doctor–patient relationship.[8,9]

Research on individual health and treatment pri-orities of older patients is limited.[10] Whereas there is increasing literature on general health goals[11] and on shared decision making for single diseases,[12] few studies have dealt with setting priorities for an array of a patient’s health problems. The Canadian WOW (What Older Women Want) health survey from 2003, for example, assessed health priorities of 2161 women on 26 common health conditions and related issues.[13] In a South Tyrolean study, 45 GPs and their 894 older patients prioritised individual health problems ac-cording to their supposed therapeutic benefit.[14] In another approach, ranking was used by doctors and their patients with diabetes to identify the most important health problems.[15]

The aim of this study was to gain insight into setting individual priorities with older patients using a pri-ority definition that was coherent to the patients’ life and doctors’ work context. The study design involved a two-step procedure: initially study patients received a geriatric assessment to generate a health overview, resulting in a problem list. Patients and their GPs then independently rated the priority and the severity for each problem on the list.

The patient and doctor ratings formed the basis for the following research questions:

•  What are health priorities for patients (importance of problems in terms of their life) and treatment priorities for their GPs (importance of problems in terms of relevance to care)?

• What is the degree of agreement between patients’ and doctors’ priorities?

The last question explored the concept of a patient’s health priority and a doctor’s treatment priority:

•  How well does the severity rating of a problem predict a priority problem for patients and for doctors?

Methods

Recruitment of GPs and patients

We intended to recruit a convenience sample of ten practices with 12 older patients each in the adjacent districts of Hannover and Braunschweig (North Germany). The practice staff were instructed to sys- tematically enrol patients into the study in a pre-defined week. This meant that, starting on a Monday, the first three older patients entering the practice after 9 am, regardless of their reason for attending, were approached. Inclusion criteria were: patient consent, age of at least 70 years and living at home. Exclusion criteria were: limited consent capabilities or major communication problems as assessed by the practice staff.

Health assessment of the patients

From October 2006 until March 2007, two research doctors (DS and PS) visited the participating patients in their homes within six weeks of recruitment and conducted the STEP assessment[16] to gain an overview of existing health problems. STEP contains a battery of 44 items (questionnaires and tests), allowing a stan-dard and broad appraisal of eight health domains. A category ‘further problems’ permits patients and doc-tors to add individual patient problems that have not been included in the standard assessment (Table 1).

STEP was developed by GPs in a European Concerted Action.[16] It has been tested and used in feasibility studies[17] and surveys of older patients in general prac-tices.[14]

The collection of data on health priorities and severity of health problems

Immediately after the assessment, patients received a list of their disclosed health problems. For each of the problems, they had to answer five questions developed by Okkes et al[18] on an ordinal rating scale (for ques-tions 1–3: not at all, a little, fairly, very; for questions 4–5: worse, unchanged, better):

1 How important is this health problem for you (in your life)?

2 To what extent are you emotionally affected by this health problem?

3 To what extent are you hindered in your daily activities by this problem?

4 What do you think is the course of your condition without treatment?

5 What do you think is the course of your condition with treatment?

The GPs received the same problem lists separately for each patient and were asked to give independent ratings for each problem. GPs rated the importance and the course of the patient’s condition from their own perspective. To clarify this perspective, question 1 was rephrased as question 6: ‘How relevant is this health problem for your health care?’ For questions 2 and 3 GPs assumed the patient’s perspective. The doctors were also asked to indicate any newly dis-covered problems. They had the opportunity to con-sult their patients’ records.

Operationalisation

All patient and doctor ratings were dichotomised in order to accentuate the divergent responses. Priority problems were defined as problems which were rated by the patient as fairly or very important, or by the doctor as fairly or very relevant (questions 1 and 6). The evaluation of severity (questions 2 to 5) was mod-ified from Okkes et al[18] and included four com-ponents: ‘emotional experience’, ‘hindrance in daily activities’ and the ‘course of a condition’ with or without treatment as an approximation of the perceived prog-nosis. Problems were labelled as severe if they were assessed to be fairly or very emotionally affecting/ hindering of daily activities, or if they were deemed to have an unfavourable prognosis.

To ascertain the degree of agreement between patients and doctors, we matched the dichotomised patient and doctor ratings for each individual health problem. We described the observed proportion of concordant ratings for importance and the three com-ponents of severity (emotionally affected, hindered in daily activities and worse without treatment). The fourth severity component ‘worse with treatment’ was excluded in any further analysis because of a pronounced floor effect. Cohen’s Kappa, a statistical measure of chance corrected interrater agreement,[19] was used and interpreted according to Landis and Koch.[20] Values between –1 and 0 indicate no agree-ment (other than by chance). Values of 0.0–0.2 indi-cate slight, 0.21–0.4 fair, 0.41–0.6 moderate, 0.61–0.8 substantial and 0.81–1 almost perfect agreement.[20] In addition, the mean Kappa and standard deviation were calculated to show the variance of agreement on the patient and doctor level.

Two multilevel logistic regression procedures were performed using Stata 11 in order to investigate the conditional association between severity and priority of a problem. The three severity variables plus patient age and gender, as well as the doctor’s previous know-ledge of the problem, were entered as fixed effects. Because of the nested data structure, doctors and patients were included as random effects. One model represents the patient perspective, in which only patient ratings on priority and severity are included; the other model represents the doctor perspective with only doctor ratings. The results of both regression models are presented as odds ratios with 95% confidence intervals (CIs).

Data analysis

Data were entered into SPSS-17 and checked for accuracy. Basic doctor and patient characteristics were reported.

On the patient and doctor level, the mean number of important and relevant problems were respectively described. For further analyses, the number of dis-closed health problems constituted the denominator. We determined the proportion of problems that were important and the proportion of problems that were severe (using the separate severity components) from the perspectives of both patients and doctors. Import-ance ratings were additionally categorised into the eight health domains provided by the STEP assess-ment (Table 1) to explore how the proportion of priority problems varied by health domain. Domain specific differences between patients and doctors for the proportion of priority problems were calculated using the sign test for dependent samples. This test determines the number of positive and negative differences of paired data and is equivalent to the Wilcoxon test, if binary variables are used.

Results

Participating GPs and patients

A convenience sample of five GPs from Hannover and six from Braunschweig participated – one more than planned – because ultimately not all practices pro-vided 12 study patients. The GPs had a median age of 49 years (interquartile (IQR) range 45–54) and a median experience of 11 years (IQR 10–16) in general practice. Two GPs were female.

The reception staff systematically invited 176 patients for an assessment in their homes. Of these, 53 patients (30%) did not take part because they either immedi-ately declined (n=39) or were unable to undergo the assessment in the following six weeks (n=14; seven had died, two were in hospital, two had changed their GP and three had moved to nursing homes). Thus 123 patients with an average age of 77.7 ( 5.4) years participated, of whom 66.6% were female. The age of the female participants (77.2 4.6 years) did not differ significantly from that of the male participants (78.7 6.4 years). Non-participants did not differ in age (78.0 5.6 years, t-test: p=0.738) and gender (64.3% female, chi[2]: p=0.066) from the participants.

additionally indicated whether a problem was pre-viously unknown. They were unaware of 9% (n=132) of all problems, which is considerably less than the 17% of new detections from a STEP assessment trial in 45 South Tyrolean general practices.[14]

Health problems disclosed by the assessment

DS and PS visited the patients in their homes to conduct the STEP assessment. It revealed on average 11.9 ( 5.4) health problems per patient, a mean of 8.8 ( 4.6) in men and 13.5 ( 5.1) in women (t-test, p<0.001). Patients had at least one and a maximum of 27 problems. The 123 study patients reported a total of 1464 health problems. Most of the problems emerged in the somatic domain as this domain provides the greatest set of items in STEP (Table 2).

Proportion of important and severe health problems

Immediately after the assessment patients received a list of their health problems to enable them to evaluate each condition according to its importance and severity. Depending on the question, patients rated between 1191 and 1464 out of their 1464 problems. Doctors evaluated between 1320 and 1414 problems. Patients found on average 8.2 ( 4.9) problems important, whereas the GPs rated 7.3 ( 4.8) patient problems as relevant. Table 3 shows the proportion of health problems rated as important or severe – from the perspective of the patients and their GPs. The doctors

Importance and severity ratings of doctors and patients

For the following analysis, only problems that were rated by both the patient and the doctor were con-sidered (n 1057–1380; Table 4). Importance and severity ratings were divided into ratings that were concordant and discordant between doctors and patients. The discordant ratings were of particular interest because they indicated the number of differences in each direc-tion between doctors and patients. In the case of ‘importance’, significantly more problems were con-sidered important by patients than relevant by doc-tors. However, doctors perceived significantly more problems to be emotionally affecting the patient, to be hindering in daily activities and to have an unfavour-able prognosis (sign tests, p<0.05; Table 4).

Agreements between doctors and patients

Sixty percent of all health conditions were rated con-cordantly important or unimportant by patients and doctors. After allowing for chance agreements, agree-ment was reduced to 11% (Kappa=0.11, indicating slight agreement). Patients’ and doctors’ agreements on the severity components were ‘slight’ to ‘fair’ according to Landis and Koch.[20] The mean Kappa on the doctor level showed a lower standard deviation than that on the patient level, possibly due to the larger variation of Kappa within doctor-specific agreements. The agreement on hindrance in daily activities on the doctor level varied more than the similar agreement on emotional effect, perhaps due to a greater difficulty in evaluating daily activities (Table 5).

Importance according to type of problem

Patients’ problems were categorised into the eight health domains provided by STEP, to examine whether the importance of problems differed amongst the health domains. Patients found at least 80% of their prob-lems important in the domains ‘social participation’, ‘mood’ and ‘function’. ‘Lifestyle’ problems were least important to them. Doctors significantly less often judged problems to be relevant in the domains ‘social participation’, ‘mood’ and ‘function’, and in the ‘somatic’ domain. However, they found ‘lifestyle’

problems significantly more often relevant (sign tests, p<0.05; Figure 1).

Determinants for a priority problem

In order to investigate whether the three severity components (emotional experience, hindrance in daily activities and prognosis without treatment) predicted a high priority problem, two multilevel logistic regression models were calculated. One model represented the patient ratings, the other the doctor ratings.

Patients gave a full set of ratings for 1186 problems, which formed the basis for the first model. Patients’ emotional experience for a given health problem was the strongest predictor of an important problem. In fact the odds of a problem being rated as important if it was also emotionally affecting the patient was 11 times

that of one which was not emotionally affecting them (OR 11.11, CI 6.73–18.33). Hindrance in daily activi-ties (odds ratio (OR) 4.34, CI 2.45–7.70) and the perceived unfavourable prognosis (OR 2.28, CI 1.63–3.21) predicted a priority problem to a lesser extent (Figure 2).

Doctors fully evaluated 1325 patient problems. The second model based on doctor ratings determined unfavourable prognosis as the strongest predictor for perceived problem relevance (OR 6.39, CI 4.61–8.87). The perceived effect on patient emotion (OR 4.1, CI 2.73–6.17) and the hindrance in daily activities (OR 2.01, CI 1.32–3.05) were weaker predictors (Figure 3).

The study patients were able to discriminate between important (69%) and less important problems (31%). The GPs were asked to evaluate health problems according to the relevance for patient care, integrating both the patient’s experience of illness and the medical aspects of disease.[21] GPs rated a slightly smaller proportion of patient problems as a priority (64%). However, patient and doctor evaluations differed sig-nificantly in some health domains. Doctors under-valued the relevance of problems in the domains of ‘somatic problems’ and also in ‘social participation’, ‘mood’ and ‘function’. The latter domains are par-ticularly relevant to older people[22] and are known to entail unmet needs.[23] In contrast patients paid less attention to ‘lifestyle problems’ than their doctors. In practice, doctors’ advice concerning changes in health behaviour often met with little success. Underlying reasons are manifold and include difference in patient awareness and reasoning.[24]

Our results showed that doctors and patients agreed little on the importance and severity of each health problem. Poor agreement between patients and doctors has been demonstrated in many studies and embraces a number of health and care aspects, such as diverging views on disease impact,[18,25,26] quality of life,[27] infor-mation concerning diagnosis and prognosis[28] and on treatment preferences.[29] Agreement, however, has been related to better participation, satisfaction and adher-ence,[30] and in turn to better outcomes of treatment.[31] The benefit also lies in the process itself, allowing an open exchange of beliefs which may also result in an agreement to differ.[32] However, it has also been said that agreement is a means to persuade patients to follow the advice of doctors,[32,33] and that it can produce conflict if patients are not able to adhere.[34]

Discussion

This in-depth assessment of 123 older general practice patients resulted in three main findings: first, patients and their GPs differentiated between more or less important health problems. Second, doctors’ health-care priorities were in poor agreement with patients’ health priorities. Third, a problem’s priority from the doctor’s perspective was strongly determined by an unfavourable prognosis, whereas for patients the strongest predictor was the emotional experience.

Patient and doctor views of disease severity and their impact on a priority problem

Our multilevel logistic regression models show that doctors and patients hold divergent views on the con-tribution of disease severity towards a problem’s priority. The current illness experience mattered most to our study patients. For the doctors it was the unfavourable prognosis that mattered most. This finding confirms the different perspectives in the disease–illness model. The perception of disease severity has practical impli-cations: it prompts patients to visit the doctor,[35] and doctors to induce medical interventions.[18]

Strengths and limitations of this study

With this study we have generated new insights into individual health priorities of older patients and the respective care priorities of their doctors. There are, however, some limitations. We enrolled a non-random sample of doctors. They were slightly younger than the average German GP (49.6 versus 51.2 years); and a higher percentage of male than female GPs partici-pated (82% versus 62%).[36] Our recruitment design for patients excluded immobile patients receiving home visits only.

Some issues with health problem evaluations are worth noting. First, the type of response options may have influenced the validity. A Likert scale with four response options and no middle value was chosen because we wanted the participants to make a clear choice. Second, the cross-sectional design of the study did not allow us to examine the reliability of the evaluations. Third, some problems were previously unknown to doctors, who may have had difficulties in further evaluating these (unverified) problems. Finally, we deliberately gave two different definitions of a priority problem (doctors: relevance for care; patients: importance in their lives). Thus poor agreement on a priority problem should not be interpreted as a GP’s poor understanding of his patient.

Implications for research and practice

In the past 20 years, the proportion of patients with four or more chronic diseases has tripled.[37] GPs increasingly face patients with complex health prob-lems. The traditional consultation style with a single disease approach is not sustainable. In Germany con-sultations are short (on average 7.6 min) but frequent (between 30 and 40 contacts with a primary care doctor per year for patients aged above 70).[38] Thus doctors spend up to five hours on each older patient per year. This time may be better spent.

The presence of multiple chronic conditions neces-sitates that GPs and their patients gain an overview of their health status, share their views on health and care priorities and mutually develop a healthcare plan. Our results show that priorities differ between older patients and their doctors. A common understanding on priorities is not intuitive. Consequently communi-cation is the key aspect for the decision on what to treat. This is challenging in terms of time, communi-cation and reconciliation of perspectives. However, it makes the current practice of implicit priority setting transparent[39] and may contribute towards successful health care for older patients. Future research should investigate ways to determine priorities explicitly with the patient and strategies to integrate other profes-sionals into this process.

Acknowledgement

We are grateful for the statistical advice from B Wiese and W Hofmann.

FUNDING

This study was not externally funded.

Ethical Approval

The study aim, its procedures, consent forms and data handling and protection measures were approved by the Ethical Committee of Hannover Medical School in 2006, No. 4164.

Peer Review

Not commissioned; externally peer reviewed.

Conflicts of Interest

None.

References

1. Anthierens M, Tansens A, Petrovic M and Christiaens T. Qualitative insights into general practitioners’ views on polypharmacy. BMC Family Practice 2010;11:65.
2. Boyd C, Darer J, Boult C, Fried L, Boult L and Wu A. Clinical practice guidelines and quality of care for older patients with multiple co-morbid diseases: implications for pay for performance.  Journal of the American Medical  Association 2005;294:716–24.
3. Dawes M. Co-morbidity: we need a guideline for each patient not a guideline for each disease.  Family Practice 2010;27:1–2.
4. Evans P (ed), EURACT Council Working Group. The European Definitions of General Practice/Family Medicine.
5. Short version, EURACT 2005.  www.woncaeurope.org/ Web%20documents/European%20Definition%200f%20 family%20medicine/Definition%20EURACTshort%20 version.pdf (accessed 25 October 2010).
6. Gray L and Newbury J. Health assessment of elderly patients.  Australian Family Physician 2004;33:795–7.
7. Stange K. A science of connectedness.  Annals of Family  Medicine 2009;7:387–95.
8. Van Royen P, Beyer M, Chevallier P et al. The research agenda for general practice/family medicine and primary health care in Europe. Part 3. Results: person centred care, comprehensive and holistic approach.  European  Journal of General Practice 2010;16:113–19.
9. Gabbay M, Shiels C, Bower P, Sibbals B, King M and Ward E. Patient–practitioner agreement: does it matter?  Psychological Medicine 2003;33:241–51.
10. Buetow S. Four strategies for negotiated care.  Journal of  the Royal Society of Medicine 1998;91:199–201.
11. Schwenk T. Competing priorities and comorbidities. So much to do and so little time.  Archives of Family  Medicine 1997;6:238–9.
12. Schwartz A, Hazen G, Leifer A and Heckerling P. Life goals and health decisions: what will people live (or die) for?  Medical Decision Making 2008;28:209–19.
13. Edwards A, Evans R and Elwyn G. Manufactured but not imported: new directions for research in shared decision making support and skills.  Patient Education and Coun- seling 2003;50:33–8.
14. Tannenbaum C, Mayo N and Ducharme F. Older women’s health priorities and perceptions of care delivery: results of the WOW health survey.  Canadian Medical  Association Journal 2005;173:153–9.
15. Piccoliori G, Gerolimon E and Abholz H. Geriatric assessment in general practice: a study of the South Tyrolean Academy of general practice. Zeitschrift fu¨r Allgemeinmedizin 2005;81:491–98 (German).
16. Zulman D, Kerr E, Hofer T, Heisler M and Zikmund-Fischer B. Patient–provider concordance in the prior-itization of health conditions among hypertensive diabetes patients.  Journal of General Internal Medicine 2010; 25:408–14.
17. Williams E, Fischer G, Junius U, Sandholzer H, Jones D and Vass M. An evidence-based approach to assessing older people in primary care. Occasional Paper of the Royal College of General Practice 2002;82:1–53.
18. Junius U, Schultz C, Fischer G, Breull A and Langner D. Putting the European Assessment into practice. Results from a German feasibility study. Zeitschrift fu¨r Allge-meinmedizin 2003;79:620–3 (German).
19. Okkes I, Veldhuis M and Lamberts H. Severity of episodes of care assessed by family physicians and patients: the DUSOI/WONCA as an extension of the International Classification of Primary Care (ICPC).  Family Practice 2002;19:350–6.
20. Rigby A. Statistical methods in epidemiology. Towards an understanding of the kappa coefficient.  Disability and  Rehabilitation 2000;22:339–44.
21. Landis J and Koch G. The measurement of observer agreement for categorical data.  Biometrics 1997;33:159– 74.
22. Silverman J, Kurtz S and Draper J. Skills for Communi-cating with Patients. Oxford: Radcliffe Publishing, 2005.
23. Berg A, Hassing L, McClearn G and Johansson B. What matters for life satisfaction in the oldest-old?  Aging and  Mental Health 2006;10:257–64.
24. Iliffe S, Lenihan P, Orrell M, Walters K, Drennan V and Tai S. The development of a short instrument to identify common unmet needs in older people in general prac-tice.  British Journal of General Practice 2004;54:914–18.
25. Buetow S, Kiata L, Liew T, Kenealy T, Dovey S and Elwyn G. Patient error: a preliminary taxonomy.  Annals of  Family Medicine 2009;7:223–31.
26. Lesho E, Foster L, Wang Z et al. The accuracy of physicians’ perceptions of patients’ suffering: findings from two teaching hospitals.  Academic Medicine 2009; 84:636–42.
27. Kroenke K, Wyrwich K, Tierney W, Babu A and Wolinsky F. Physician-estimated disease severity in patients with chronic heart or lung disease: a cross-sectional analysis. Health and Quality of Life Outcomes 2006;4:60.
28. Pearlman R and Uhlman R. Quality of life in chronic diseases: perceptions of elderly patients. Journal of Ger-ontology 1988;43:M25–30.
29. Jackson J. Communication about symptoms in primary care: impact on patient outcomes. Journal of Alternative and Complementary Medicine 2005;11:S51–6.
30. Montgomery A and Fahey T. How do patients’ treat-ment preferences compare with those of clinicians? Quality in Health Care 2001;10:i39–43.
31. Carr A and Donovan J. Why doctors and patients disagree.  British Journal of Rheumatology 1998;37:1–6.
32. Vermeire E, Hearnshaw H and van Royen P. Patient adherence to treatment: three decades of research. A comprehensive review.  Journal of Clinical Pharmacy and  Therapeutics 2001;26:331–42.
33. Stevenson F and Scambler G. The relationship between medicine and the public: the challenge of concordance. Health 2005;9:1–9.
34. Segal J. ‘Compliance’ to ‘concordance’: a critical view.  Journal of Medical Humanities 2007;28:81–96.
35. Say R and Thomson R. The importance of patient preferences in treatment decisions – challenges for doctors. British Medical Journal 2003;327:542–5.
36. Braunack-Mayer A and Avery J. Before the consultation: why people do (or do not) go to the doctor.  British  Journal of General Practice 2009;59:478–9.
37. Dezernat 4 – Vertra¨ge und Verordnungsmanagement. Gesamtdatei I. Arzte. Archivnummer 1003755564. daris.kbv.de/daris.asp (accessed 6 November 2010).
38. Uijen A and van den Lisdonk E. Multimorbidity in primary care: prevalence and trend over the last 20 years. European Journal of General Practice 2008;14:S28–32.
39. ISEG, Institut fu¨r Sozialmedizin, Epidemiologie und Gesundheitssystemforschung. Barmer GEK Arztreport. Auswertungen zu Daten bis 2008. St Augustin: Aasgard Verlag, 2010.
40. Junius-Walker U, Voigt I, Wrede J, Hummers-Pradier E, Lazic D and Dierks ML. Health and treatment priorities in patients with multimorbidity. Report on a workshop from the European General Practice Network meeting Research on multimorbidity in general practice.  Euro- pean Journal of General Practice 2010;16:51–4.