Growing numbers of older adults are living with and managing multiple chronic conditions, with concomitant complex treatments.1–4 A person who also reports high treatment burden can be described as experiencing the aggregate influence of the direct consequences of those diseases in combination with the prescribed or recommended care plans. Treatment burden is defined here as a patient’s perception of the aggregate weight of the actions and resources they devote to their healthcare, including difficulty, time, and out-of-pocket costs dedicated to the healthcare tasks such as adhering to medications, dietary recommendations, and self-monitoring.
The importance of treatment burden for decision-making is acknowledged by systems designed to develop clinical practice guidelines.5 It is generally recognized that patients with multimorbidity face increased healthcare demands associated with adherence to treatment regimens.6 However, even for many single diseases, how patients view treatment burden is poorly understood.7 Treatment burden has been previously defined and measured for patients with terminal conditions and for patients with specific conditions based on the characteristics of treatment (such as hospitalization), complications that arise, and self-reported burden of a specific aspect of treatment.8–12 Despite the inherent relevance of treatment burden to decision-making for older patients with multimorbidity, comprehensive measures of treatment burden for this population do not exist.7 Healthcare task difficulty (HCTD) (eg, difficulty planning medication schedule) is one aspect of treatment burden. HCTD may be addressed in part through targeting patient activation, defined as patients’ knowledge, skills, and confidence for self-management.13,14 As described in the Chronic Care Model, patient-centered chronic illness care is particularly important when considering complex treatment plans in older adults with multimorbidity.15
What remain unclear are the frequency and magnitude of HCTD among diverse populations of older adults with multimorbidity, and the association between HCTD and quality of chronic illness care and health-related quality of life (HRQoL).
Therefore, we sought to describe and validate a measure of HCTD in a sample of older adults with multimorbidity. Cross-sectionally, we examined whether lower patient activation was associated with greater HCTD and whether greater HCTD was associated with lower HRQoL and lower quality of chronic illness care. To further assess predictive validity of the HCTD scale, we analyzed the longitudinal associations between change in patient activation, change in the HCTD score, and changes in mental and physical HRQoL as well as changes in patient-reported quality of chronic illness care. We examined whether increases in patient activation over time were associated with declining HCTD and whether increasing HCTD was associated with declining patient-reported quality of chronic illness care and HRQoL.
We examined baseline data collected in a cluster-randomized controlled trial of “Guided Care.”16 The Institutional Review Boards of Johns Hopkins University, Kaiser-Permanente Mid-Atlantic States, and MedStar Physician Partners approved the study.
Between 2006 and 2009, 904 older patients with multimorbidity and 308 of their informal caregivers were enrolled.16 Community-dwelling, English-speaking patients aged 65 years or older were eligible if they had been seen by a participating physician within the previous year and were predicted to be in the highest 25% of Medicare utilization for the next year using the hierarchical condition category predictive model.16,17 Potential participants were excluded if they had cognitive impairment (unable to verify age, address, and spelling of name) and no legal representative, or were unable to participate in the baseline or follow-up interviews.16 Among eligible patients, 38% consented to participate (n=904).
For longitudinal analyses, we restricted our sample to 419 of the 904 patients who were randomized to receive usual care. We restricted longitudinal analyses to usual care because Guided Care was originally hypothesized to affect patient-reported quality of chronic illness care and HRQoL. Patients were administered a baseline interview in-person and a phone interview at 6 and 18 months. Among the 419 patients receiving usual care who survived and received care from the same physician at follow-up, 94.2% (n=370) had at least 1 follow-up interview and were thus eligible to be in this analysis. Patients in the longitudinal sample for this analysis (n=370) did not differ from the other participants who are not in the longitudinal sample (n=534) in terms of age, sex, or race, patient activation, or HRQoL, but the longitudinal sample did have slightly worse baseline Patient Assessment of Chronic Illness Care scores (2.55 vs. 2.65, P<0.05).
Analyses of patient activation were restricted to study participants who self-reported responses at baseline and at least 1 follow-up (n=855 and 351, respectively). For all other measures, approximately 5% of the baseline sample (n=49) and longitudinal sample (n=19) were proxy responses.
Data Collection and Measures
In-person baseline interviews were conducted with patients and caregivers. Patient measures included socio-demographics; number and type of self-reported chronic conditions; physical, sensory, and cognitive functioning; number of medications; social support; and income sufficiency. Caregiver measures included relationship to patient, co-residence with patient, and hours of care provided. The Patient Activation Measure (PAM), assesses patient knowledge, skill, and confidence for self-management and provides a continuous score (0–100).14 Quality of chronic illness care was assessed using the Patient Assessment of Chronic Illness Care (PACIC), a 20-item validated scale that measures the patient report of receiving aspects of care congruent with the Chronic Care Model.18 For example, questions assess whether the patient was “asked to talk about any problems with my medicines or their effects,” “asked to talk about my goals in caring for my condition,” etc.18 HRQoL is measured with the Physical and Mental Component Scores from the SF-36.19
Although treatment burden is a multidimensional construct, in this study we focus on a single dimension—perceived difficulty in performing healthcare management tasks. HCTD was measured using 11 items that ask patients to identify the amount of difficulty they have performing healthcare tasks (Table 1). These items were generated through review of existing instruments, input from clinical experts in geriatric medicine, and testing in a pilot study of Guided Care.20 A summary of HCTD scale was created from 8 out of 11 of the questions that all participants answered where 0=no difficulty, 1=some difficulty, and 2=a lot of difficulty (range, 0–16). Three questions were excluded from the summary scale because they included an additional response category of “does not do” task. Participants who did not know or refused were recorded as having no difficulty (0).
Internal Construct Validity of the Summary HCTD Scale
Exploratory factor analysis was used to assess the dimensionality of the summary HCTD scale (0–16, described above). Polychoric correlations were used to account for the ordinal response categories in the 8 tasks in the exploratory factor analysis. The “eigenvalue-greater-than-1” rule, the scree test, and the parallel test were used to select the number of factors to extract. The Comparative Fit Index, the Tucker-Lewis Index, and the root mean square error of approximation were used to assess the model goodness-of-fit. The unidimensionality of the scale was further verified using a confirmatory factor analytic model that allows the residual errors of the 8 items to be correlated, and a likelihood-based χ2 difference test was used to compare the models with and without the correlated residual errors.21 For a scale to be unidimensional, the correlations among the items must be fully explained by a single factor.22 Therefore, nonsignificant difference between the 2 models will support the unidimensionality of the scale.
Predictive Validity of the Summary HCTD Scale
Cross-sectional analyses: Patient and caregiver characteristics were compared across categories of HCTD for the combined usual care and intervention group participants. Categories were generated by dividing summary scale scores ≥1 into tertiles resulting in the following 4 categories: no difficulty (score=0), low level of difficulty (score <2; first tertile), medium level of difficulty (score=2–4.5; second tertile), and high level of difficulty (score=4.6 or more; third tertile). Trends for continuous and binary characteristics were tested using the Cuzick test and the Royston Ptrend test, respectively; and for multilevel characteristics, multinomial (insurer) or ordinal (number of medications) logistic regression was used treating HCTD category as a continuous variable (Table 2). Reports of HCTD for each question were compared between respondents with high levels of mental and physical HRQoL (SF-36 score in the highest tertile) to those with low levels of mental and physical HRQoL (SF-36 score in the lowest tertile).
We analyzed the independent association of HCTD categories with the PACIC, mental, and physical HRQoL after adjusting for age, sex, education, and race using multiple linear regression analyses.
Longitudinal analyses: To further assess predictive validity of the HCTD scale, we analyzed the longitudinal associations between change in the summary HCTD score and changes in the mental and physical HRQoL and the PACIC using random effects models, to account for the different follow-up time intervals and within-person correlation among the repeated measurements of the outcome variables over time.
The model is specified as follows:
where xit and yit, respectively, are subject i’s continuous HCTD score and outcome score (eg, mental HRQoL) at time t, t=1…T; sit is the time gap between visit 1 and visit t; and ε is the error term. The coefficient bc summarizes the population-average “cross-sectional” effect of HCTD score recorded on mental HRQoL at baseline; whereas bL summarizes the population-average “longitudinal” effect of within-person change in HCTD score and change in mental HRQoL, so named because, for example, subtracting the first equation from the second in (A) results in an expression for yi2−yi1 in terms of xi2−xi1 and biL. To account for baseline between-subject heterogeneity in yi1 and within-subject correlation over time, we treat bi0 and αi as random effects. Results of fitting model (A) before and after adjustment for age, sex, education, and race are presented.
We analyzed the effect of baseline patient activation (PAM) and change in PAM on change in the HCTD summary score using a random effects model similar to model (A), with the exception that the HCTD score was modeled as a negative binomial outcome with a log link to account for the excess variability (ie, over-dispersion) in the HCTD score. The coefficients bc and bL are presented as the percentage change in the HCTD score for 1 SD unit change in PAM.
Missing data were uniformly low; ranging from none on most items to 2.4% for difficulty knowing when to change medication dosages. Sensitivity analyses were conducted by (1) dropping those with missing data for difficulty items included in the HCTD score and (2) reassigning missing items a “3” (a lot of difficulty) as opposed to “0.” The results of both methods were similar to the original analysis that assumes missing is having no difficulty with the given task.
We generated and validated a scale (0–16) of HCTD based on 8 questions. The 1-common factor model explains 50% of the total variance of all 8 difficulty items; and the goodness-of-fit statistics all suggested a reasonably good fit (Comparative Fit Index=0.91, Tucker-Lewis Index=0.96, root mean square error of approximation=0.05). Although both the “eigenvalue-greater-than-1” rule and the scree test favored 1-common factor model, the parallel test identified 2 factors. The confirmatory factor analytic model also revealed significant partial corrections among 3 out of 8 items (planning medication schedule, taking medications, and making decisions about changing medications) after removing the effect of the “difficulty” factor (likelihood-based χ2 different test statistics =34.476 with 3 df; P<0.01), suggesting the possibility of a second factor underlying these 3 items. However, given that (i) all 3 items had “very significant” loadings (≥0.7) on the “difficulty” factor in the 1-factor model,23 (ii) each item was similarly affected by the “difficulty” factor with the percentage of variance of each item explained by the factor ranging from 40% to 53% with the exception of scheduling medical appointments (71%), and (iii) the ordinal coefficient α was 0.89 indicating strong interrelatedness among the 8 items, the proposed summary HCTD scale may still be warranted for its simplicity and ease of interpretation.24,25
The prevalence of heart failure and stroke was significantly greater with increasing HCTD (Table 2). Hypertension and dementia showed a similar, but marginal, association. The number of diseases was significantly associated with an increased HCTD. Conditions adding to complexity of clinical decision-making and understanding and implementing treatment regimens (falls, visual impairment, hearing impairment, and sleep disturbances) were associated with an increased HCTD. Insufficient funds at the end of the month was associated with an increased HCTD. Among participants with a caregiver (n=308), greater caregiving hours was associated with an increased HCTD. Among all participants, activation was strongly associated with HCTD; scores decreased in a stepwise manner by nearly 10 points from 58.8 among people with no HCTD to 49.3 among people with high HCTD. Greater HCTD was associated with lower PACIC scores, indicating worse quality of chronic illness care. Greater HCTD was associated with worse mental and physical HRQoL.
To explore how difficulty with specific healthcare tasks was associated with HRQoL, we examined tertiles of mental and physical HRQoL in relation to responses to individual questions about different aspects of a treatment regimen. Worse physical and mental HRQoL were significantly associated with greater HCTD for all but 1 question, where the relationship was in the expected direction but nonsignificant at the 0.05 level (physical HRQoL and difficulty planning a medication schedule, P=0.076).
We examined the associations of HCTD score with mental and physical HRQoL, and PACIC scores in 3 separate models adjusting for age, sex, education, and race (Table 3). Compared with individuals with no HCTD, those with low, medium, and high HCTD had significantly worse physical and mental HRQoL (P<0.05). Individuals with high HCTD scored 11.86 points lower on the SF-36 mental health score (P<0.001) and 6.22 points lower on the SF-36 physical health score (P<0.001) compared with individuals with no HCTD. Individuals with HCTD (low, medium, and high) also reported lower quality of chronic illness care on an average (P<0.05); those with high HCTD rated the PACIC 0.24 points lower compared with individuals with no HCTD (P<0.001).
Patients with higher activation at baseline reported lower healthcare task difficulty at follow-up (Table 4). An adjusted random effects negative binomial model with random intercept and random time slope showed that a 1 SD unit (ie, 12.8) higher baseline PAM score was associated with a 35% lower HCTD at follow-up (P<0.01). We also found that for every SD unit (ie, 14.6) increase in PAM score over time, HCTD decreases by 12% (although these results were not statistically significant, P=0.08). In other words, for a person whose PAM score increases by 14.6 points over any 2 points of time, his/her HCTD score is expected to decrease by 12% during the same interval after adjusting for all other covariates including baseline PAM score. All results are adjusted for age, sex, race, education, and follow-up time.
Physical and Mental HRQoL
Patients with higher HCTD at baseline reported lower mental HRQoL at follow-up (Table 5). The SF-36 mental health score was lower by 2.42 points for every 1 point higher baseline HCTD (P<0.001), a difference in magnitude that approaches clinical significance (3 points).26 Similar, albeit weaker results were found for physical health. The SF-36 physical health score was lower by 1.09 points at follow-up for every 1 point higher baseline HCTD (P<0.001). We also found that patients who reported increasing HCTD over time reported decreased mental and physical HRQoL over time. Adjusted longitudinal analysis showed that for every 1 point increase in HCTD score over time, SF-36 mental health score decreased by 1.32 points (P<0.001) and SF-36 physical health score decreased by 0.59 points (P=0.001). All results are adjusted for sex, age, education, race, and follow-up time.
Quality of Chronic Illness Care
Although we found a statistically significant relationship between HCTD and patient-rated quality of chronic illness care (PACIC) at baseline, and between change in HCTD and change in PACIC at follow-up, the effect size was small (Table 5). The PACIC score was lower by 0.04 points for every 1 point higher baseline HCTD score (P<0.05). Similarly, the longitudinal analysis showed that for every 1 point increase in HCTD score over time, the PACIC score decreased by 0.04 points (P<0.05).
This study describes and validates a measure of HCTD in a sample of older adults with multimorbidity. We find HCTD to be strongly associated with greater disease burden, and conditions that add to complexity of decision-making and self-management. HCTD was also greater among older adults who were less activated and was associated with worse mental and physical HRQoL and worse quality of chronic illness care. From the perspective that HCTD was highly correlated with worse health status, lower education, and more limited financial resources; factors that are strongly associated with self-management challenges, study results lend support to construct validity of treatment burden. The association of HCTD over time with mental HRQoL approaches clinical significance (3 points).26 Future work is needed to understand whether approaches to reduce difficulty with healthcare tasks lead to improvements in quality of life and quality of chronic illness care. A careful examination of a patient’s treatment plan and the associated HCTD may help the physician to reduce needless complexity, enhance patient activation, and prioritize which healthcare tasks are most important. Although HCTD represents just 1 aspect of treatment burden, it is an important dimension that is broadly relevant to a large segment of the population, irrespective of specific illnesses and settings of care.13,27
The concept of treatment burden has been largely used for, and developed from, end-of-life care, whereas treatment burden and treatment outcomes have been defined as key components of patients’ medical decision-making.10,28,29 Studies of patients at the end-of-life have found that patients vary in their willingness to accept treatment with high degrees of associated treatment burden.10,30 Treatment burden has been explored in the context of single diseases or screening tests.8,9,31,32 Adult veterans with diabetes report significant burden associated with glycemic control therapies, and this burden is the primary predictor of adherence and willingness to accept insulin.9 A qualitative study of middle-aged adults with multimorbidity suggests that significant burden may result from medication-related treatments and that this in turn influences adherence, patient priorities, and self-management.11 Thus, building on the concepts developed largely in the end-of-life and single disease literatures, treatment burden associated with treatment regimens for chronic conditions was defined in this study for a broader population who were not immediately confronting end-of-life care and encompassed both nonpharmacologic and pharmacologic treatments across all of their conditions.8–10,12,27,30–32
Well-informed patients vary in the burden they assign to treatments they are receiving and in their willingness to accept burdensome treatments. Some patients may intentionally prefer to depart from “best practice” treatments.9,11,30–35 Treatment burden related to just 1 disease may affect adherence, self-management, and prioritization of therapies.9–11 Currently, how to combine disease-specific evidence-based recommendations and patient preferences to create feasible, integrated, and appropriately prioritized treatment regimens for older patients with multimorbidity is unknown. Unlike comorbidity scales, an understanding of treatment burden has the advantage of being potentially actionable in a clinical setting. Assessment of treatment burden should directly affect decision-making because treatment burden is inherently part of how patients operationalize their priorities concerning their healthcare. Assessing and addressing treatment burden in clinical practice may serve to improve patient and caregiver empowerment, help identify situations where patients would benefit from additional assistance, and guide prioritization and decision-making in a patient-centered manner. The relationship between treatment burden and quality of chronic illness care observed here emphasizes the importance of this type of patient-centered care. Future work is needed to determine the best methods to assess and address treatment burden in a busy clinical practice, particularly among populations with heterogenous patterns of multiple chronic conditions.2,36 Such approaches may include screening for treatment burden ahead of time or during the visit, and providing prompts and tools for providers to discuss treatment burden with older patients, guiding patients, caregivers, and providers through strategies to improve activation, secure additional assistance with healthcare tasks, and inform prioritization of aspects of care plans to best align with patient goals.
The measurement of HCTD does not address all potential dimensions of treatment burden.27 However, this assessment is a worthwhile foundation from which the broader concept of treatment burden in medical decision-making might be further explored. Our study sample was selected on the basis of their predicted future use of health services. Although they represent a vulnerable population, only a subset of all eligible people agreed to participate. This may introduce bias if nonparticipants differ from participants, but we are not able to assess the characteristics of nonparticipants. Although this study relies on a relatively small sample, study participants are from a diverse population of older adults with multimorbidity with different types of medical coverage. This is an observational study, and therefore we cannot draw conclusions about causality. We are using longitudinal data to examine concurrent relationship of changes in HCTD with changes in HRQoL and quality of chronic illness care. Available measures in this study are robust, though assessment of adherence, health literacy, and each component of treatment regimens and how they change over time would have been ideal to explore additional hypotheses. We do not have data on disease-specific quality standards. Given the debate about the optimal measurement of quality of care for older patients with multimorbidity, a strength of this work is the use of a patient assessment of quality of chronic illness care.37
Understanding complexity of decision-making and self-management among older adults with multimorbidity, their families, and clinicians is important. Currently, clinicians are without explicit guidance or evidence on how to approach decisions for such patients. Older patients with greater difficulty with prescribed treatment regimens experience lower mental and physical HRQoL, worse quality of chronic illness care, and low levels of activation. Results from this study suggest that identifying HCTD may be feasible using broad-based approaches that are not disease-specific and that may be more widely applicable to older adults with multiple chronic conditions. Patient-centered approaches might help to improve patient activation, minimize interactions and complexity, and unify treatment recommendations coming from all providers and could ultimately affect patient-important outcomes like HRQoL and quality of chronic illness care.
The authors thank the invaluable contributions to this study made by Johns Hopkins Community Physicians, MedStar Physician Partners, Kaiser-Permanente Mid-Atlantic States, the United States Family Health Plan, Battelle Centers for Public Health Research, the Centers for Medicare & Medicaid Services, Accumen, ResDAC, the University of Minnesota Survey Research Center, and all of the participating patients, caregivers, physicians, and Guided Care nurses.
1. Tinetti ME, Bogardus ST Jr, Agostini JV.Potential pitfalls of disease-specific guidelines for patients with multiple conditions.N Engl J Med.2004;351:2870–2874.
2. .Patient-centered care for older adults with multiple chronic conditions: a stepwise approach from the american geriatrics society: american geriatrics society expert panel on the care of older adults with multimorbidity.J Am Geriatr Soc.2012;60:1957–1968.
4. Boyd CM, Darer J, Boult C, et al..Clinical practice guidelines and quality of care for older patients with multiple comorbid diseases: implications for pay for performance.JAMA.2005;294:716–724.
5. Guyatt GH, Oxman AD, Vist GE, et al..GRADE: an emerging consensus on rating quality of evidence and strength of recommendations.BMJ.2008;336:924–926.
6. Bayliss EA, Ellis JL, Steiner JF.Barriers to self-management and quality-of-life outcomes in seniors with multimorbidities.Ann Fam Med.2007;5:395–402.
7. May C, Montori VM, Mair FS.We need minimally disruptive medicine.BMJ.2009;339:b2803.
8. Walter LC, Eng C, Covinsky KE.Screening mammography for frail older women: what are the burdens?J Gen Intern Med.2001;16:779–784.
9. Vijan S, Hayward RA, Ronis DL, et al..Brief report: the burden of diabetes therapy: implications for the design of effective patient-centered treatment regimens.J Gen Intern Med.2005;20:479–482.
10. Fried TR, Bradley EH, Towle VR.Assessment of patient preferences: integrating treatments and outcomes.J Gerontol B Psychol Sci Soc Sci.2002;57:S348–S354.
11. Townsend A, Hunt K, Wyke S.Managing multiple morbidity in mid-life: a qualitative study of attitudes to drug use.BMJ.2003;327:837.
12. Gallacher K, May CR, Montori VM, et al..Understanding patients’ experiences of treatment burden in chronic heart failure using normalization process theory.Ann Fam Med.2011;9:235–243.
13. Giovannetti ER, Wolff JL, Xue QL, et al..Difficulty assisting with health care tasks among caregivers of multimorbid older adults.J Gen Intern Med.2012;27:37–44.
14. Hibbard JH, Stockard J, Mahoney ER, et al..Development of the Patient Activation Measure (PAM): conceptualizing and measuring activation in patients and consumers.Health Serv Res.2004;39pt 11005–1026.
15. Wagner EH.Chronic disease management: what will it take to improve care for chronic illness?Eff Clin Pract.1998;1:2–4.
16. Boyd CM, Boult C, Shadmi E, et al..Guided care for multimorbid older adults.Gerontologist.2007;47:697–704.
17. Ash AS, Ellis RP, Pope GC, et al..Using diagnoses to describe populations and predict costs.Health Care Financ Rev.2000;21:7–28.
18. Glasgow RE, Wagner EH, Schaefer J, et al..Development and validation of the Patient Assessment of Chronic Illness Care (PACIC).Med Care.2005;43:436–444.
19. Elliott TE, Renier CM, Palcher JA.Chronic pain, depression, and quality of life: correlations and predictive value of the SF-36.Pain Med.2003;4:331–339.
20. Boyd CM, Shadmi E, Conwell LJ, et al..A pilot test of the effect of guided care on the quality of primary care experiences for multimorbid older adults.J Gen Intern Med.2008;23:536–542.
21. Satorra A, Bentler PM.A scaled difference chi-square test statistic for moment structure analysis, 2009 Psychometrika
22. Netemeyer RG, Bearden WO, Subhash S.Scaling Procedures: Issues and Applications.2003.Thousand Oaks, CA:Sage.
23. Hair JF, Anderson RE, Tatham RL, et al..Multivariate Data Analysis.1998.Upper Saddle River, NJ:Prentice Hall.
24. Carver CS.How should multifaceted personality constructs be tested? Issues illustrated by self-monitoring, attributional style, and hardiness.J Pers Soc Psychol.1989;56:577–585.
25. Richins ML, Dawson S.A consumer values orientation for materialism and its asurement: scale development and validation.J Consumer Res.1992;19:303–316.
26. Ware JE, Kosinski M, Bjorner JB, et al..Determining important differences in scores.User’s Manual for the SF-36v2TM Health Survey.2007:2nd ed.Lincoln, RI:QualityMetric Incorporated.
27. Eton DT, Ramalho de Oliveira D, Egginton JS, et al..Building a measurement framework of burden of treatment in complex patients with chronic conditions: a qualitative study.Patient Relat Outcome Meas.2012;3:39–49.
28. Danis M, Mutran E, Garrett JM, et al..A prospective study of the impact of patient preferences on life-sustaining treatment and hospital cost.Crit Care Med.1996;24:1811–1817.
29. Jani B, Blane D, Browne S, et al..Identifying treatment burden as an important concept for end of life care in those with advanced heart failure.Curr Opin Support Palliat Care.2013;7:3–7.
30. Fried TR, Bradley EH, Towle VR, et al..Understanding the treatment preferences of seriously ill patients.N Engl J Med.2002;346:1061–1066.
31. Huang ES, Brown SE, Ewigman BG, et al..Patient perceptions of quality of life with diabetes-relatedomplications and treatments.Diabetes Care.2007;30:2478–2483.
32. Kerr EA, Heisler M, Krein SL, et al..Beyond comorbidity counts: how do comorbidity type and severity influence diabetes patients’ treatment priorities and self-management?J Gen Intern Med.2007;22:1635–1640.
33. Deyo RA.A key medical decision maker: the patient.BMJ.2001;323:466–467.
34. Protheroe J, Fahey T, Montgomery AA, et al..The impact of patients’ preferences on the treatment of atrial fibrillation: observational study of patient based decision analysis.BMJ.2000;320:1380–1384.
35. Fried TR, Bradley EH, Towle VR.Valuing the outcomes of treatment: do patients and their caregivers agree?Arch Intern Med.2003;163:2073–2078.
36. Boyd CM, McNabney MK, Brandt N, et al..Guiding principles for the care of older adults with multimorbidity: an approach for clinicians: American Geriatrics Society Expert Panel on the Care of Older Adults with Multimorbidity.J Am Geriatr Soc.2012;60:E1–E25.
37. Werner RM, Greenfield S, Fung C, et al..Measuring quality of care in patients with multiple clinical conditions: summary of a conference conducted by the Society of General Internal Medicine.J Gen Intern Med.2007;22:1206–1211.