Simple Segmentation Tool to Understand Elderly Health Status and Needs

Background: In the context of a young population, Singapore has established a superb system of care, focusing largely on acute hospitals and specialty services. This has addressed the needs of a dominantly young population, which are usually acute and self-limited, associated with asymptomatic risks, and occasionally relate to single disabling or life-limiting conditions.

As a population ages, the needs for health and related social services change as more and more individuals shift into more complex stages, with multiple conditions, loss of resilience, and late decline; addressing the needs associated with these stages is less amenable to an acute care, disease-centric approach. The needs of an older population are in some ways similar to those of a younger population (though generally of a greater likelihood and degree); however, the elderly require different strategies for supporting them through the later stages of their lives. For example:

• For chronic asymptomatic risk factors, the elderly may be more or less benefited by screening, definitive testing, and treatment;
• Elders may be less likely to have their risk factors identified and successfully managed;
• Elders may have unidentified symptomatic, treatable conditions (e.g., cataract);
• Social issues, such as family support and social networks may have a dominant effect on the introduction, use and success of medical interventions;
• Elders experience a loss of resilience which may occur without anyone being aware, and through self-reinforcement can lead to frank frailty (e.g., both as a phenotype and as a physiological state of loss of reserve in multiple systems);
• Though end of life issues are not unique to elders, supporting older people through the late stages of life-limiting conditions presents distinctive features, which can require a different set of services;
• The high and increasing use of acute services by elderly; both in volume and in proportion to their numbers) may point to opportunities to care for elders at complex stages of health more effectively, humanely, and sustainably.

The primary aim of this study is to validate the Simple Segmentation Tool (SST) as a method of patient population segmentation in terms of inter-rater reliability, as well as convergent validity and predictive validity.

Design of the SST

1. To address the expanding needs of the elderly, it is important to take a person-centric approach , tailoring services to the individual. From a health system perspective, this presents a challenge for at least two reasons. First, health system services have focused on sites of care and thus one may miss opportunities to alleviate progression to complex stages, with high morbidity and cost. Second, implementing excessively tailored programs can be extremely expensive from a health system perspective, especially when they are hospital-based. To address the above challenges, the investigators propose a population segmentation approach. This entails assigning categories of health and associated social features that characterize "packages" of interventions that aim to alleviate progression to more symptomatic/higher risk categories, and improve quality of life. This approach has been advocated internationally (Hewner \& Seo, 2014; Van Der Laan, Van Offenbeek, Broekhuis, \& J., 2014; Zhou, Wong, \& Li, 2014), following foundational work by Lynne et al (Lynn, Straube, Bell, Jencks, \& Kambic, 2007). Population segmentation recognizes that while there is wide diversity among people and individualized approaches to care, understanding prototypical needs can be a powerful tool for purposes of planning and evaluation of health and social services.
2. The Simple Segmentation Tool was designed to facilitate the segmentation of an ageing Singapore population so as to facilitate the provision of cost-effective person-centric care. Specifically, the tool facilitates the capture of meso-level medical needs and complicating factors of a patient before classifying said patient into one of 7 medical categories and one of 3 care complexity levels. A healthcare need in the context of the SST is defined as a potential to benefit from healthcare whereupon benefit can be understood as a reduction in the probability of a patient transitioning into a worse clinical state or experiencing an adverse clinical outcome. Meanwhile, the type of needs that the SST captures are normative needs which are needs of a patient as assessed by a clinician. The 7 medical global impressions of patient of the SST were adapted from the original 8 based on the "Bridges to Health" model by Joanne Lynn et al (Lynn et al., 2007) in order to better suit our evaluation of an elderly population. Patients are classified into one of the 7 health categories namely 1) Healthy, 2) Acutely ill but curable, 3) Chronic conditions, asymptomatic, 4) Chronic conditions, stable, 5) Long course of decline, 6) Limited reserve and serious exacerbation and 7) Short period of decline before dying; that best characterizes their most salient clinical needs in the medium to long term (months to years). With the exception of the category "Acutely ill but curable" all patients should only be assigned into one category at any point in time. The complicating factors section of the SST is designed to measure the degree of need in 8 different healthcare relevant characteristics. These characteristics were selected based on their significance towards patient management at transitional points of care. They are as follows: 1) functional assessment, 2) social support in case of need, 3) hospital admissions in last 6 months, 4) disruptive behavioral issues, 5) polypharmacy, 6) organization of care, 7) activation in own care, and 8) skilled nursing type task needs. Decline in ADL function (Buurman, Parlevliet, Van Deelen, De Haan, \& De Rooij, 2010) and recent unplanned hospital readmissions (L. W. \& C. K. W., 1999) are known to precede unplanned hospital readmission. On the other hand, social support enhances an individual's ability to cope with stressors, promotes health-seeking behavior, and also provides basic healthcare services (Glanz, Rimer, \& Viswanath, 2008). Disruptive behavioral issues are important to characterize as they can cause significant caregiver stress (Ahn \& Horgas, 2013) . Complex medication management plans due to polypharmacy (Cameli et al., 2012) or multiple healthcare provider confusion (Services, 2010) are important to recognize so as to prevent iatrogenic harm to patient. Finally, the degree of patient activation (Mosen et al., 2007) as well as successful management of geriatric syndromes (Van Craen et al., 2010) if present are also important determinants of elderly health. Depending on the highest level of social healthcare need identified in the patient, the following complexity category can then be assigned. For example, if the highest level of social healthcare need identified is a 1 for Polypharmacy, the patient is thus deemed as moderately complex. Complexity can be understood as an "interference with standard care"(Peek, Baird, \& Coleman, 2009). Increased complexity in a patient increases the probability of failure of a medical intervention and hence needs to be recognized and mitigated in order to improve clinical outcomes. The SST is designed for use in an outpatient setting. Clinicians trained in its use are expected to first assess a patient as part of their routine clinical assessment before using the SST to categorize the patient. By keeping the number of inputs required from the clinician brief and succinct, it is hoped that utilization of the SST integrates well with usual clinic operational flow.
3. Validation and Improvement of the SST

As the SST is a novel tool, it needs to be psychometrically validated before it can be used to segment patient populations in various sites of provision of healthcare.There are various benefits associated with successful validation of the tool in the proposed manner:

Inter-rater reliability: Validation would allow us to assess the interchangeability of potential users of the tool in a heterogenous patient population (Gwet, 2010). If the investigators are able to establish interchangeability of clinicians of different backgrounds and experience such as senior doctors, nurses and even medical students, the investigators may then be able to more easily operationalize the tool since more medical personnel will be able to reliably administer it.

Convergent validity: Validation would allow us to verify that two measures of constructs that theoretically should be related, are in fact related (Trochim, 2006). Demonstration of convergent validity when the SST is paired with widely accepted and validated pre-existing tools will allow us to gain reasonable confidence in the ability of the tool to capture similar constructs.

Predictive validity: Validation that the SST can be used to predict future health status or healthcare utilization patterns such as unplanned readmissions after discharge would increase the SST's utility for shaping interventions that target high readmission risk individuals in a bid to lower their risk of unplanned hospital readmissions.

Based on the results of the validation studies, the SST can be further modified to make it optimally robust before subsequent patient segmentation studies in the various sites of care. For example, specific sections of the SST with low inter-rater reliability or convergent validity can be modified to improve the tool's psychometric properties.

SST Benefit 1 - Identification of leverage points for intervention. Through cross-sectional and longitudinal studies of patient profiles in terms of their medical and social healthcare needs using the SST, areas that are characterized by high healthcare needs can be detected in the various sites of care thus allowing for implementation of health service interventions in order to better meet these healthcare needs. Furthermore, longitudinal studies would allow for the determination of risk factors associated with increased transition rates from lower to higher severity global impression of patient category thus also facilitating targeted follow up intervention studies.

Health service reorganization to better meet areas of high healthcare need would lead to improved clinical outcomes through reduction in the probability of clinical complications. In the context of an ageing population this would help keep more elderly healthy and functional for a longer period of time thus reducing healthcare costs associated with exacerbation of chronic diseases.

SST Benefit 2 - Greater accountability. Better identification of areas of healthcare need allows for better comparison between needs and clinical infrastructure or services provided so as to allow for greater accountability in the delivery of healthcare. Services which do not meet identified needs can be scaled down so as to reduce healthcare costs. Resources both financial and manpower alike can then be diverted towards areas of greater need.

SST Benefit 3 - Packages of care. Understanding the common combinations of population medical needs and complicating factors would allow for packages of care to be developed which are designed for common prototypical patients and can be easily modified to suit individual patients. Such an approach would allow for more effective resource planning and capitation bundling purposes, as well as integrate the delivery of healthcare interventions for the patient.