Article Text
PDF
Abstract
Background Older adults with complex care needs face trade-offs in determining the right course of treatment. The Centers for Medicare and Medicaid Services identified ‘Care is personalized and aligned with patient’s goals’ as a key meaningful measures category, yet existing quality measures typically assess disease-specific care and may not effectively evaluate what is most important to older adults and family members. Measures based on individualised goals and goal-based outcomes have been proposed as an alternative but are not routinely assessed or implemented.
Objectives We tested two approaches to assessing goal-based outcomes that allow individuals to set goals based on their own priorities and measure progress—(1) goal attainment scaling and (2) existing, validated patient-reported outcome measures (PROM).
Methods A prospective cohort study of feasibility in seven sites (33 clinicians) of the two approaches with 229 individuals. We calculated performance on a measure of achievement of individually identified goals.
Results Both approaches were successfully implemented in a non-randomly selected population, and a goal-based outcome could be calculated for 189 (82%) of participants. Most individuals met their goal-based outcome (73%) with no statistical difference between the goal attainment scaling approach (74%) and the patient-reported outcomes approach (70%). Goals were heterogeneous ranging from participating in activities, health management, independence and physical health. Clinicians chose to use goal attainment scaling (n=184, 80%) more often than PROMs (n=49, 20%) and rated the goal attainment scaling approach as useful for providing patient care.
Conclusion Goal-based outcomes have the potential to both improve the way healthcare is provided and fill a critical gap in value-based payment.
- performance measures
- shared decision making
- patient-centred care
- quality measurement
Statistics from Altmetric.com
Background
There is broad agreement that individuals’ priorities and goals should guide their healthcare, particularly for adults with complex conditions (eg, multiple chronic conditions and functional limitations).1 These patients often face trade-offs when choosing treatments and frequently need services and supports beyond traditional medical care.1–3 There is a growing movement to provide complex patients goal-based care in which clinicians elicit personal goals and preferences and engage patients in shared decision-making to develop a plan of care that will help them achieve those goals.4–9 Assessing ‘what matters most’ to patients is a key pillar in the John A Hartford Foundation and Institute for Healthcare Improvement Age-Friendly Health System initiative10 and the Health Resources and Services Administration Geriatric Workforce Education Program.11 However, implementing goal-based care will require change at all levels of healthcare delivery, particularly quality measurement.12 Existing clinical quality measures used in value-based payment arrangements target disease-specific clinical services and outcomes that are often not meaningful to patients with complex care needs, and clinicians have reported concerns about the relevance of quality measures.13 Assessing achievement of individualised and prioritised goals of care has the potential to be a more meaningful measure of quality that could fill the Centers for Medicare and Medicaid Services (CMS)-identified desire for measures where ‘care is personalized and aligned with patient’s goals’.14 If implemented in value-based purchasing, goal-based outcomes, or other quality measures that encourage clinicians to provide goal-based care, could drive significant change in the way care is delivered and improve how patient-centred care is measured.
A central challenge to measuring patient goal attainment is the lack of adequate processes to elicit, document and monitor progress towards patient goals. Goals—when they are discussed and documented—are frequently documented in multiple places in the electronic record (progress notes, scanned documents and problem lists) and may conflict with one another.15 16 When clinicians document goals of care, the identified goals often focus on end-of-life care or the clinician’s goals for disease management, resulting in disease-specific biomarker goals (eg, blood pressure) or referral for specific medical care (eg, get preventive screenings)16 17 rather than on quality-of-life outcomes, such as participating in social activities.15 Furthermore, clinicians and patients may disagree about documented goals of care.18 19 Even when documented, these goals are rarely communicated across care teams or followed systematically.20 Movement towards patient-centred, goal-based care requires a more structured approach to eliciting, documenting and monitoring goals from the patient’s perspective. Recent studies have explored more structured approaches to eliciting patient-centred goals.5–7 9 This work demonstrates the feasibility and value of assessing patient-centred goals in routine clinical care but has not explored how patient-centred goals could be used in quality measurement, specifically the documentation and measurement of goal achievement.
Goal attainment scaling (GAS) presents a structured approach to setting and documenting achievement of goals (see figure 1). Originally developed for use in mental health, GAS is a reliable, valid and sensitive measurement approach often used for evaluating complex interventions.21 22 Achievement of goals is associated with increased patient engagement in and satisfaction with their treatment23 24 and improved health outcomes.25 However, a frequently cited limitation of GAS is the inability to control the scope of goals identified by patients.24 26 Clinicians may fear that patients may identify goals that seem unrealistic, outside the scope of the clinician to influence, or not relevant to their health or healthcare. The approach can also be time consuming and requires additional training for clinicians.23 27 Despite these limitations, recent studies have begun to evaluate the feasibility of this approach for older adults in the clinical setting to facilitate goal-oriented care.5 9 28–30 None of these studies have evaluated the potential of using GAS results in clinical quality measurement.
Description of goal-based outcome measurement approaches piloted in the study. The patient-reported outcome measures used in the pilot were drawn from validated survey tools.51–58 Source: author’s own description and examples.
An alternative approach to goal-based care involves the use of patient-reported outcome measures (PROMs) selected by the patient. PROMs can address outcomes from the patient’s perspective and can be used to measure different aspects of health-related quality of life. PROMs add value by bringing attention to feelings, functioning and experiences that matter to the patient.31 32 These measures can help both patients and clinicians track the impact of lifestyle changes and treatments on patient symptoms and inform when additional treatment may be necessary.33 34 There is increased interest in using PROM results in quality metrics as part of value-based purchasing.35 36 However, the goals expressed by older adults and their caregivers are heterogeneous,18 37–39 and a single PROM tool, such as a standardised quality-of-life questionnaire, may not address the goals and priorities relevant to a specific individual. Some individuals may prioritise their physical functioning, while others may prioritise their mental health. To address this limitation, some experts recommend clinicians use multiple PROMs to measure the condition or symptom most relevant to a patient’s priorities.40 We call this individualised selection of PROMs, prioritised PROMs (see figure 1). No studies have evaluated the feasibility of implementing a prioritised PROM approach.
Both GAS and prioritised PROMs hold promise for helping clinicians implement goal-based care and provide data that can be used to develop a valid and reliable clinical quality measure of goal-based outcomes. However, questions remain about the feasibility of these approaches and the interpretability and validity of clinical quality measures based on data generated from these approaches. This study piloted both approaches to assess their feasibility for implementation in diverse clinical settings and whether either method provides meaningful information for quality measurement.
Study data and methods
Study design and setting
This was a cohort study of implementation feasibility of goal-based outcome measurement. The Standards for Quality Improvement Reporting Excellence 2.0 guidelines were followed in reporting this manuscript.41 Using our existing relationships with health plans and organisations, we recruited a convenience sample of seven clinical practices and case management organisations that served an older adult, functionally disabled population and had an existing process for eliciting patient-centred goals. Between July 2016 and July 2017, 33 clinicians from the seven sites participated in the pilot. Each site individually identified clinical and support staff providers (referred to here as clinicians) to participate in the project and there was a wide range in clinician type (table 1).
Goal-based outcome pilot sites
Clinician training and learning collaborative
Study sites participated in a learning collaborative to support the pilot and document the process changes and tools needed for implementation (using the Plan, Do, Study, Act quality improvement approach42). Clinicians participated in two, in-person, 8-hour training sessions held in Washington, DC led by a trained quality improvement coach. During the trainings, sites received instruction in eliciting person-centred specific, measurable, attainable, relevant, time-bound (SMART) goals,43 using an optional goal inventory adapted from Jennings et al 9 and setting goals using both the GAS and prioritised PROM approaches. The training included a mix of didactic method and role-playing in clinical scenarios. During the second training, study investigators (DBR and LAJ) reviewed clinicians’ implementation of GAS using the SMART goal criteria and provided feedback on whether the scaling of goals was precisely described at each level. Throughout the pilot, clinicians participated in monthly coaching calls (60–90 min each month) where they shared feedback on their process changes.
Patient and caregiver participants
Each site was asked to identify 20 potential participants from their existing case load. Target participants were adults age 50 and over with either activity of daily living (ADL) or instrumental ADL impairments and their family or friend caregivers. Building on approach used by Jennings et al,9 clinicians identified the best participant for goal setting among patient and caregiver dyads. For individuals with cognitive impairment, clinicians were encouraged to set goals with the patient if possible and engage the caregiver in goal setting when necessary. For individuals without cognitive impairment, we encouraged clinicians to set goals with the patient, but set goals with the caregiver (alone or in tandem with the patient) if they were normally part of the clinical encounter.
Goal-based measurement intervention
Participants and clinicians used their existing approaches for eliciting patient-centred goals and then used either the GAS or prioritised PROM approach (figure 1) to measure goal achievement. The use of an inventory of possible goals adapted from Jennings et al 9 was optional. We allowed practice sites to decide which of the two approaches to pilot first, but requested they implement each approach with at least 10 participants. Given concerns about time added to a clinical encounter, clinicians were only asked to address one goal per participant. The clinician and participant independently rated the difficulty of the goal (not at all difficult, somewhat difficult, very difficult, extremely difficult) and independently assessed achievement of the goal at follow-up (defined as approximately 6 months after setting the goal).
Evaluation of clinician experiences of implementation
To understand clinicians’ experience of the goal-setting intervention, we asked clinicians to rate the usability of both approaches on a scale from 1 (not at all useful) to 10 (most useful) in three areas: (1) helping patients track their progress; (2) determining which services and supports to provide to patients; and (3) helping patients achieve their goals.
Analysis
We summarised pilot participant characteristics (table 2). We compared characteristics between those who participated in GAS and those who participated in the prioritised PROM approach using an independent samples two-tailed t-test for linear variables and χ2 test for binary variables. We calculated a clinical quality metric of whether each participant achieved at least one goal over the 6-month follow-up period. For the GAS method, goal achievement was defined as either the participant or clinician rating the goal as achieving a ‘0—expected outcome’ or better within the 6-month performance period. For the PROM method, goal achievement was defined as any improvement on the PROM by one or more points (if the goal was to improve) or maintaining the same score on the PROM (if the goal was to maintain the current state). We evaluated whether patient characteristics (ie, ADL disability, number of chronic diseases, dementia) and the presence of a caregiver in goal setting were associated with goal achievement using independent samples two-tailed t-test for linear variables and χ2 test for binary variables.
Characteristics of study participants
To understand the heterogeneity in goals, we reviewed and categorised free-text goals. One author (CAC) coded the free-text goals recorded in patient encounters and created initial goal categories using an existing goal taxonomy44 as a reference. A second researcher (ERG) independently reviewed the goals using the same goal categories. Discrepancies were discussed and reconciled among the study investigators.
Results
Seven sites implemented the goal-setting intervention with 193 patients alone, 30 patient and caregiver dyads, and 6 caregivers alone for a total of 229 participants, exceeding our target of 20 participants per site (table 2). Of the 36 caregivers who participated, 20 cared for patients with dementia. On average, patients were older (mean age 75.3) with multiple chronic conditions (mean 3.9), multiple ADL limitations (mean 2.6) and had experienced at least one emergency department visit and hospitalisation in the year prior to participating in the study (table 2). Twenty-one per cent had dementia. The majority of patients were white (79%) and non-Hispanic (96%). On average, patients who used GAS were younger than patients who used prioritised PROM (p<0.01) but had more ADL limitations (p<0.01).
Implementation
There was variation in the implementation of the goal-based outcome approaches across sites. The number of participants ranged from 21 to 64 per site. While all seven sites used GAS, only five of the seven sites implemented the prioritised PROM approach (one telephone-based site thought collecting PROMs over the phone would be too burdensome and one attempted prioritised PROMs but did not provide complete baseline data). The goal inventory was used with 40% of participants (range 0%–100% across sites). The GAS approach was used with more participants (n=184, 80%) than the prioritised PROM approach (n=49, 20%); four participants participated in both GAS and prioritised PROM approaches (table 2).
Of the 184 GAS participants, 19% (n=35) were lost to follow-up (participant dropped out (either refused to participate or could not be contacted), died or left provider) (table 2). Twelve per cent (n=22) revised their goal, or expected outcome, over the 6-month period. Sometimes goal revision occurred when participants achieved a goal and set new ones (n=9); in other cases, participants decided to revise their goal because they were not making progress and wanted to set a more realistic goal or they preferred to work on a different goal (n=13). Among GAS participants with follow-up data (n=149), 74% met at least one goal over 6 months (see figure 2). In most cases (n=169, 92%), the individual and clinicians’ rating of goal achievement matched; however, there was a minority of cases where individuals rated their goal attainment higher than the clinicians’ ratings (nine cases) and vice versa (six cases).
Study participant flow diagram. Consolidated Standards of Reporting Trials (CONSORT) diagram of study participants in pilot study of goal-based outcomes. Source: author’s analysis of pilot data for 229 participants. PROM, patient-reported outcome measure.
Of the 49 prioritised PROM participants, follow-up data were not available for one-third (n=16) primarily because clinicians did not collect follow-up PROM data (n=7) or dropped out of the project (n=6) (see table 2). Each PROM tool was used by at least one participant, with the PROMIS Physical Activity questionnaire being the most commonly selected PROM tool (n=16) (figure 1). Twenty-nine per cent (n=14) of participants set a goal to maintain a current state, and all others set a goal to improve their current state. Five participants revised their goal over the study period. Among the prioritised PROM participants with follow-up data (n=33), 23 (70%) achieved their PROM-specific goal. The rate of goal achievement was not significantly different between PROM and GAS approaches (p>0.05).
Across both approaches, there were no significant differences (p>0.05) in patient characteristics between patients who chose not to complete the goal-based outcome process (ie, dropped out) and those who completed the process. In addition, there were no significant differences (p>0.05) in goal attainment by patient characteristics (ie, number of ADL dependencies, number of chronic conditions, dementia, or patient-rated difficulty of the goals) or caregiver participation.
Content of goals
All study participants set at least one goal and 31 participants had more than one goal for a total of 263 goals (see table 3).
Content of goals set by participants (n=263 goals)
Goals frequently covered multiple categories and subcategories (average 1.25 subcategories per goal).
The most common goals addressed participation in activities (29%; eg, social activity, being able to leave the house), health management (29%; eg, pain control, mental health), physical activities (25%; eg, walking, avoiding falls) and independence (21%; eg, self-care, living arrangement). The majority of goals did not mention management of a specific health condition. Participants rated goals as low to moderate difficulty on average (mean difficulty rating 1.47 (SD 0.75) on 0–3 scale). Clinicians also independently assessed difficulty of the goal (mean difficulty 1.44 (SD 0.67)), and clinician ratings differed from participant ratings 21% of the time (clinicians mostly rating the goal less difficult than patients).
Clinician experience
Seventeen of the 33 participating clinicians responded to the usability questions (response rate 51%). All 17 clinicians rated the GAS approach, and 9 of 11 clinicians who participated in the prioritised PROM approach rated this approach. On a 10-point scale (1–10), clinicians rated usability of GAS as high on all three domains on average: determining which services and supports to provide (M 7.83 (SD=2.06)), helping patients achieve their goals (M 8.0 (SD=1.30)) and helping patients track their progress (GAS M 7.75 (SD=1.36)). The usability of the PROM approach was rated lower but the difference was not statistically significant in this small sample (p>0.05): determining which services and supports to provide (M 6.22 (SD=2.91)), helping patients achieve their goals (M 5.5 (SD=2.10)) and helping patients track their progress (M 7.21 (SD=1.46)).
Discussion
Seven sites successfully piloted approaches to GAS, and five sites successfully piloted prioritised PROMs. Despite setting expectations for equal implementation of both approaches, clinicians chose to implement the GAS approach more often than the prioritised PROM approach, and we observed fewer missing data with the GAS approach. Clinicians reported high usability for GAS for helping to determine services and supports and helping patients achieve their goals. Given the familiarity of PROMs in clinical practice (eg, frequent use of PROMs to screen for conditions such as depression45) and the unfamiliarity of GAS, we found this result surprising. It is possible that GAS may provide a more individualised measure of goal achievement than PROMs. Other studies have demonstrated GAS is more sensitive to change than PROMs24 which could impact clinicians’ perspectives on the relative usability of these approaches. These results suggest that GAS could be a feasible approach for assessing goal-based outcomes, which are valuable for both care planning and clinical quality measurement.
Implementing goal-based outcomes into routine care has the potential to promote care that is person centred instead of disease oriented,4 12 and GAS may be a more acceptable approach to achieving this shift than the use of PROMs alone. Goals in this pilot varied widely in topic from health issues (eg, ‘Improve sleep patterns’) to broader topics such as ‘Socializing with friends and family’ and ‘Practicing faith’. This is consistent with other studies which have shown heterogeneity in goals expressed by older adults, particularly studies using GAS.9 27 29 46 While the domains expressed by older adults were similar to those found in generic quality of life PROMs (ie, physical functioning, pain, mental health), the specificity of the goals provided rich details from which a care plan could be built. For example, goals centred on physical functioning spoke to specific activities such as ‘to keep fishing as much as possible’. The goal categories from this pilot study (eg, Independence, Activities) align with existing goal categorisations and typologies in the literature, suggesting that goals and priorities moving beyond traditional medical care are commonly voiced.6 9 46
The implementation of both approaches to goal-based outcomes also provided structured data that could be used to calculate a clinical quality measure. A commonly cited concern about using GAS results in clinical quality measurement is that patients will set goals that are not realistic or unobtainable. Our results show that patients were able to set goals that were achieved 74% of the time, which is consistent with other studies of GAS.9 29 Patients and clinicians usually agreed about how difficult the goals were and when they disagreed, clinicians reported the goal as less difficult than the patient’s rating of the goal’s difficulty. Some studies have suggested that older adults with more disability and cognitive impairment are less likely to achieve their goals,47 48 which could require risk adjustment for these individual characteristics in a clinical quality measure. While this pilot did not show significant difference in goal achievement by difficulty of the goal and patient level of complexity, additional research is needed.
This pilot highlighted several implementation challenges that need to be addressed if either GAS or prioritised PROMs are to become a more widespread part of clinical care. Although not measured routinely in this pilot, other studies have suggested the time commitment to elicit, document and measure the goal is not insubstantial (20–40 min9). Most project sites used registered nurses or social workers to engage patients in goal setting instead of the primary care provider; it is unclear whether the types of goals developed or perceptions of difficulty may vary when a physician versus other clinicians set goals with a patient. Additional research is necessary to understand how clinical teams can collaboratively ensure that goals are realistic given a patient’s prognosis and diagnoses and ensure that all members of the care team are aware of the patient’s goals when making recommendations for treatment. An additional implementation challenge is how to negotiate goals when both a patient and caregiver are involved in the discussion. An estimated 38% of adults attend routine medical visits with another person,49 so any approach must consider how these companions are included in the discussion. Implementation of these approaches could be further strengthened by recent work on structured approaches to eliciting patient priorities, such as Patient Priorities Care.5–7 This multistep process for eliciting patient priorities and incorporating them into care plans has shown improved outcomes for older adults with multiple-chronic conditions.7 The incorporation of goal-based outcomes into value-based payment could drive further adoption of such patient-centred care delivery models.
The pilot also highlighted several quality measurement challenges that would need to be addressed if results from GAS or prioritised PROMs are to be tied to value-based purchasing. If achievement of goals is tied to payment, there could be an incentive to set easily achievable goals with patients. One way to avoid this unintended consequence would be to measure the process of goal setting rather than the outcome of meeting goals but require healthcare organisations to use the outcome for their own quality improvement efforts. Another challenge is ensuring goals reflect the patient’s priorities, as opposed to the goals of the clinician or goals of the health system. To avoid this, goals could be limited to a set of commonly expressed patient priorities of care,6 44 46 50 ensuring fidelity to a patient-centred approach. A third challenge is whether a health system has the knowledge or resources to help individuals achieve goals that require both a medical and community-based solution (eg, ‘To move to an assisted living facility in the next three months’). However, there is increasing recognition on the part of health systems that community-based solutions are needed to improve population health and address social determinants of health.
There are several limitations of this pilot study. First, the pilot was only conducted with a convenience sample of organisations who volunteered. Second, despite providing equal training and laying out expectations for equal implementation, clinicians implemented the prioritised PROM approach less frequently than the GAS approach. Sites and clinicians were also allowed to decide for themselves which approach was used first and which patients to pilot each approach. Therefore, clinician views of the usefulness of the respective approaches could be biased by greater experience with GAS. Third, patients were not randomly selected to participate and therefore there is likely selection bias in the study sample. Clinicians chose to use the GAS approach with a younger but more disabled population compared with the PROM approach, suggesting there was some selection bias. Fourth, goal achievement defined by the prioritised PROM approach did not account for a clinically meaningful difference in score. Future research on the prioritised PROM approach would benefit from using tools in a single measurement system, such as PROMIS, which would allow for more in-depth comparison of the change in score across different PROMs. We also only included English-speaking individuals and did not collect information about socioeconomic status. Additional research is necessary in a broader sample to fully understand the impact of patient and clinician factors on goal achievement, especially if goal achievement is used in value-based purchasing and for risk adjustment for patient factors.
Conclusion
Although understanding ‘what matters most’ to older adults is a key component of providing age-friendly care10 11 and a meaningful measure category defined by CMS,14 methods for systematically eliciting, documenting and tracking goal-based outcomes are relatively unstudied. This study suggests that using structured approaches to goal-based outcomes for older adults with complex care needs could be feasible in clinical care and a promising approach for quality measurement that could lead to improvements in person-centred care delivery for this population. GAS may be a more useful tool to implement than PROM-based measurement for this complex ageing population, but additional research is necessary. Future implementation of goal-based outcomes is needed to test additional quality measures and better understand how the approaches can be more uniformly implemented across organisations.
Acknowledgments
We would like to acknowledge the contribution of the sites that participated in the research: iCare; Priority Health; University of California, Los Angeles; University of California, San Francisco; US Medical Management; and Kaiser Permanente.
References
Footnotes
Contributors ERG was the study PI and led the analysis and writing of the manuscript. CAC was responsible for qualitative analysis and contributed to writing of the manuscript. LAJ, DBR and SFS were coinvestigators and contributed to the design of the study and writing of the manuscript. SHS was an advisor on the project and provided substantial feedback on the manuscript.
Funding This work was funded by The John A. Hartford Foundation and The SCAN Foundation.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval The study intervention was implemented at each site as a quality improvement project. Secondary analysis of the quality improvement data was reviewed and approved by the Chesapeake Institutional Review Board (Pro00017844).
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement No data are available. Due to the terms of agreement with the participating organisations no data are available at this time.