Complex adaptive chronic care – typologies of patientjourney: a case studyjep_1670 1..5

Carmel M. Martin MBBS MSc PhD MRCGP FAFPHM FRACGP,1 Deirdre Grady BSc MSc,2

Susan Deaconking MBBS,4 Catherine McMahon RN,4 Atieh ZarabzadehPhD Post. Dip. Stats. Post. Dip. Health Inf. BSc Soft. Eng.3 and Brendan O’Shea FRCGP MICGP5

1Visiting Professor, National Digital Research Centre, Department of Public Health and Primary Care, Trinity College Dublin, Dublin, Ireland andAssociate Professor of Family Medicine, NOSM, Canada2Clinical Research Assistant, 3Health Informatics Software Engineer, National Digital Research Centre, Dublin, Ireland4Clinical Advisor, National Digital Research Centre, Dublin, Ireland5Lecturer in General Practice, Trinity College Dublin, Dublin, Ireland and Specialist in Occupational Medicine, General Practitioner and MedicalDirector, Kildare and County West Wicklow Doctors on Call, Kildare, Ireland

Keywords

case management, chronic illness, complexadaptive systems, diagnostic typologies,health services research, life courseanalysis, observations of daily living, patientjourney, primary care

Correspondence

Associate Professor Carmel M. MartinNational Digital Research Centre, CraneStreet, Dublin 8, IrelandE-mail: carmelmarymartin@gmail.com

Accepted for publication: 23 March 2011

doi:10.1111/j.1365-2753.2011.01670.x

AbstractRationale Complex adaptive chronic care (CACC) is a framework based upon complexadaptive systems’ theory developed to address different stages in the patient journey inchronic illness. Simple, complicated, complex and chaotic phases are proposed as diagnostictypes.Aims To categorize phases of the patient journey and evaluate their utility as diagnostictypologies.Methods A qualitative case study of two cohorts, identified as being at risk of avoidablehospitalization: 12 patients monitored to establish typologies, followed by 46 patients tovalidate the typologies. Patients were recruited from a general practitioner out-of-hoursservice. Self-rated health, medical and psychological health, social support, environmentalconcerns, medication adherence and health service use were monitored with phone callsmade 3–5 times per week for an average of 4 weeks. Analysis techniques includedfrequency distributions, coding and categorization of patients’ longitudinal data using aCACC framework.Findings Twelve and 46 patients, mean age 69 years, were monitored for average of 28days in cohorts 1 and 2 respectively. Cohorts 1 and 2 patient journeys were categorized asbeing: stable complex 66.66% vs. 67.4%, unstable complex 25% vs. 26.08% and unstablecomplex chaotic 8.3% vs. 6.52% respectively. An average of 0.48, 0.75 and 2 interventionsper person were provided in the stable, unstable and chaotic journeys. Instability wasrelated to complex interactions between illness, social support, environment, as well asmedication and medical care issues.Conclusion Longitudinal patient journeys encompass different phases with characteristicdynamics and are likely to require different interventions and strategies – thus being‘adaptive’ to the changing complex dynamics of the patient’s illness and care needs. CACCjourney types provide a clinical tool for health professionals to focus time and careinterventions in response to patterns of instability in multiple domains in chronic illness care.

The patient journey in the complexadaptive chronic care (CACC)theoretical frameworkA CACC framework aims to describe the interdependent elementsof the personal care experience and the complex dynamic interac-tions between a patient and his or her health care providers withinthe broader health system over time as a complex adaptive system[1].1 The CACC model was designed to address the complexsystems nature of the chronic care model [2,3].

There are multiple discernable phases or patterns across thedisease and illness journey over time, which are associated withconsiderable expenditure variation [4]. Stages of the patientjourney vary according to the dynamics and interconnected feed-back loops among the bio-psycho-social, health care and environ-mental domains as well as chronic disease progression [5,6].

1 The term ‘complex system’ formally refers to an interdependent systemof many parts that is coupled in a non-linear fashion. When there is much

non-linearity in a system (i.e. many components are interacting and inter-dependent such as in a primary health care environment), its behaviour canbe unpredictable, and interventions frequently lead to unintended conse-quences. Understanding and changing the behaviour of such a complexdynamic system requires an appreciation of its key patterns, leveragepoints and constraints.

Journal of Evaluation in Clinical Practice ISSN 1365-2753

© 2011 Blackwell Publishing Ltd, Journal of Evaluation in Clinical Practice 1

Based upon the Cynefin framework [7], patient journeys ascomplex adaptive systems were operationalized as simple, com-plicated, complex and chaotic phases in CACC.

Stable – simple or complicated care phases ofchronic conditions

Simple – people are well, functioning and stable; the aim of care isto slow the progress of risk factors, single disease or a diseasecluster and optimize quality of life and prevent complications andco-morbidities – for example, raised cholesterol, high blood pres-sure, pre-diabetes or diabetes.

At this stage, medical care is stable; that is, patient care andhealth states do not involve unstable dynamics and linear protocolsare generally appropriate. Conversely at another level, publichealth ‘care’ may involve dynamic complex individual and societalinterventions. For example, smoking cessation involves interven-tions in a diverse range of complex systems from economics andmarkets, legislation, media as well as in health care with theprovision of ‘simple’ quitting advice [8,9].

Complicated – multiple factors cause morbidity, which usuallyare chronic, and include bio-psycho-social environment compo-nents; the aim is to balance self-care, health and pharmaceuticalinterventions and health-related co-morbidity. Treatment andmonitoring become more frequent and there are an increasednumber of providers and care settings involved. Yet, health isstable or deteriorating imperceptibly.

Complex (unpredictable dynamics) or chaotic(out-of-control) phases of chronic conditions

Complex – acute or subacute-on-chronic exacerbations, flaresbecause of potential destabilization in bio-psycho-social environ-ment components including self-care, health and pharmaceuticalinterventions or health-related co-morbidity. Care may includepre-terminal phase, frailty, risk of falls, depression and/or diseaseflare-up stages.

Chaotic – destabilization of multiple dimensions: falls, loss ofdiabetic control, severe pain, shortness of breath, additional diag-nosis of cancer, mental health crisis and/or additional acute con-ditions such as pneumonia resulting in environmental ‘blowouts’.Appropriate and timely community-based primary/primary healthcare interventions can avoid these chaotic states. Chaotic states ofchronic illness have a high risk of leading to death (total stability),but also may revert back to a stable trajectory or to an ongoing butincreasing unstable health journey.

Patients in these abovementioned states generally incur thegreatest health care expenditures, resulting from expensivehospitalization and re-hospitalization with its associated high-technology treatments, compared to people with similar diagnoseswho are more stable.

AimsThe study aims to categorize phases of the patient journey andevaluate their utility as diagnostic typologies using a case study oftwo cohorts, identified as being at risk of avoidable hospitalization.The first cohort would be monitored to describe patient journey

typologies, and subsequently a second cohort would be monitoredwith support interventions. This aimed to validate the typologies ina larger group and evaluate their clinical usefulness in identifyingthe need for different frequency and intensity of community-basedcare interventions. Interventions were non-clinical and aimed toidentify early signs of instability and provide information, supportor refer back to the general practitioner (GP) or appropriate socialservices.

MethodsPatients identified as being at risk of avoidable hospitalizationattending Kildare and West Wicklow Doctors on Call (KDOC) GPcooperative out-of-hours service (OOH) were recruited. TheKDOC database (October 2010–February 2011) was screened. Allunplanned home visits, all encounters resulting in transfers byambulance, referrals to hospital or advised to attend Accident andEmergency were secondarily screened for the inclusion criteria:• one chronic condition (>6 months), presenting as subacute or

chronic flare-up, not acute surgical problem, not in long-termcare;

• 18 years or older;• have had a recent unplanned hospital admission to a medical

ward, or• had a recent attendance at an emergency department, and• are able to record their health status online with an electronic

diary or family or caregiver or take regular phone calls abouttheir health.Summary data and outcomes of adult encounters were pro-

vided to the team of two GPs, one nurse and one research assis-tant in de-identified format. Potential cases were identified bytwo team members. Full case notes were then reviewed to iden-tify a list of eligible cases, which, if confirmed as suitable bytheir GPs, were recruited. The research team conducted an initialassessment of consenting participants and caregivers in theirhome and began daily health monitoring during working days ofthe week.

Monitoring daily health

Patients were phoned at a time suitable to their needs. The dailyhealth survey questions included health-related questions includ-ing self-rated health status, if their health status had changedsince the last interview, and if they had any other concerns. Psy-chological questions included how often they felt very nervous,calm and peaceful, and happy. Social questions included if theyhad someone to take them out if needed, would there besomeone to help cooking and cleaning and also if there had beenany changes to the patient’s caregiver and family supportnetwork. There were open-text fields available for the appropri-ate questions where more information could be added. After eachdaily interview, a summary of the interview was compiled tocomplete the survey.

FindingsA total of 19 000 KDOC encounters (1/9/2010 to 7/11/2010 –cohort 1) and (17/12/10 to 17/2/2011 – cohort 2) were screened.

Complex adaptive care – patient journeys C.M. Martin et al.

© 2011 Blackwell Publishing Ltd2

Using a method of consecutive sampling 12 patients wererecruited to cohort 1 in October 2010 and 48 to cohort 2 inDecember 2010, providing 286 and 720 daily monitoring reportsrespectively. The profiles of cohort 1 and cohort 2 are described inFig. 1. Cohort 1 was a purely monitoring phase, while cohort 2involved active care management by the project team.

Key elements of the patient journey are reported in fivedimensions of daily living – the presence of daily concerns, fluc-tuations in self-rated health, fluctuations in caregiver and per-ceived social support availability, medication changes and healthcare changes.

Patterns of the patient journey were graphed and cate-gorized as stable, unstable being complex or chaotic. This wascarried out by C. M. and D. G. initially on an independentbasis and consensus was reached on a case by case basis forcohort 1.

These predominant patient journey patterns were identified inthe following proportions described in Fig. 2. Key types of patientnarratives from cohort 1 are described using pseudonyms. Figure 3describes the frequency of interventions and average length ofphone calls for different types of patient journey.

Stable complex

Patient 1 – Eileen

Eileen is 93 years old and lives with her daughter, Sharon, andher family in a very comfortable home. Her problems are chronicshortness of breath because of chronic obstructive pulmonarydisease, cardiac problems including coronary artery bypass graft-ing, back pain and early Alzheimer’s disease. She has moved inwith her family following hospitalization for chronic obstructivepulmonary disease. Throughout the monitoring phase, Eileenremains very well and her social support and medical conditionremains stable despite a complicated medical condition withmultiple morbidity.

Unstable complex

Complex and chaotic re-stabilizing patient

Patient 2 – Bill

Bill is 63 years old and lives on his own in a hostel with a landlady.He has type 2 diabetes, vertigo and dizziness of unclear aetiology.He suffered a fall and fractured several ribs, with recurrent chestpains and vertigo 1 month before entering the study. Brian struggleswith chronic pain and vertigo, despite taking a 2-week holiday. On

Figure 1 Cohort 1 and cohort 2 profiles.

Figure 2 Types of patient journey identified. ‘Stable patient’ demon-strates an absence of daily concerns, and stability in self-rated health,support, medication and health care. ‘Unstable patient’ demonstratesdaily concerns about pain which preceded a worsening of self-ratedhealth followed by a change in medication and eventually re-stabilizes,while social support does not change, as he lives alone. ‘Chaotic patient’demonstrates caregiver support issues as the root cause which are notresolved and trigger a chaotic phase of illness in her and her motherresulting in hospitalization and death. Support change 1 = yes, 2 = no;concerns 1 = yes; 2 = no; medication change 1 = yes; 2 = no; healthcare change 1 = yes; 2 = no and self-rated health was scored verypoor = 0; poor = 2; fair = 4; good = 6; very good = 8 and excellent = 10.RIP, rest in peace (deceased).

C.M. Martin et al. Complex adaptive care – patient journeys

© 2011 Blackwell Publishing Ltd 3

return from his holiday, he suffered an attack of dizziness on theplane and was admitted via a KDOC attendance. Subsequently, hemade three visits to Accident and Emergency and was admittedtwice, without going through his GP or KDOC.

Edge of chaos – stable complex–chronic, severely

impaired and remains at risk of destabilization

Patient 3 – Ann

Ann is 32 years old and she has poor quality of life for 13 yearssince she developed Crohn’s disease. Her quality of life deterio-rated when she developed abdominal sepsis and underwentunsuccessful surgery which involved incision and drainage.Since the birth of her daughter 12 years ago, she has been inchronic pain with recurrent infection. She lives with her daughterbut cannot leave the house as she has unpredictable and explo-sive bowel movements. She presents to Dr Jones daily for painrelief injections and also received a pain injection from herpublic health nurse on weekday mornings. She requires the assis-tance with pain relief of an OOH service at the weekends. Shesuffers from panic attacks and depression as a result of hercomplex physical state and social isolation. Since her worseninghealth state, she has lost her job and her friends. She frequentlytakes her anger out on her daughter. Her daughter is also at riskof social isolation and neglect which Ann is aware of. Ann haschronic poor self-rated health and very severe pain and consti-pation with frequent medication changes. She has been referredto hospital numerous times but refuses to be admitted because ofconcerns over the care of her daughter. She is trying to movehouse to be closer to her Mum which also would allow herdaughter to be closer to her friends. She is addicted to morphineand continually requires antibiotics and pain relief. Her pelvicabscesses continually flare and require draining with increasingfrequency. She is on the edge of chaos with frequent suicidalthoughts and is at risk of requiring emergency surgery. Annstates that she really benefits from the support of monitoring 5days a week because her life is so difficult and needs encour-agement and social support on a daily basis.

Unstable chaotic leading to death

Patient 4 – Mary

Mary is 88 years old, widowed for 15 years, has very earlydementia and has been treated for hypertension and high choles-

terol for some years. She lives with her daughter Margaret whowas widowed 8 months previously, and who works in her ownbusiness as well as caring for her mother. Mary has becomeincreasingly difficult to manage as she is not sleeping at night,and Margaret is becoming increasingly stressed and her bloodpressure which is normal has become elevated associated withher chronic exhaustion because of her mother’s insomnia.Margaret reported daily concerns and issues and was increas-ingly depressed and fatigued. The insomnia predated the stress,and Margaret required an emergency visit to the OOH whereher blood pressure was found to be exceedingly high. And hos-pital admission was suggested, despite medication for stressMargaret’s condition worsened. Mary became increasingly agi-tated and concerned that she was being rejected, and went intoan acute anxiety state when she was admitted for respite care.She was diagnosed as having acute heart failure (probably thecause for her insomnia at home) but was unable to recover andwas admitted to hospital and died. The admission diagnosis andcause of death was heart failure, but the root cause of theproblem was apparent 2 weeks earlier as daily concerns and car-egiver reporting represented a complex interplay of early demen-tia, incipient heart failure, caregiver bereavement and stress andexhaustion.

Unstable journeys reflected a dynamic interplay of physical,psycho-social, caregiver-related, medication and medical issues,rather than purely a disease flare-up. Greater instability reflectsthe need for more interventions. Phone calls varied in lengthdepending on the journey phase of the participant, as well as theoccurrence of any health or social concerns requiring an inter-vention. Phone calls to stable participants were typically 1–2minutes in duration if there were no reported concerns, withtopics of conversation varying from one patient to the next.Phone calls to participants with greater instability were longer induration as there were more issues to discuss and longer again inthe cases with problem identification and interventions.

Over 1 month – there were an average of 0.48 interventions perpatient in the complex stable group; 0.75 interventions per patientin the ‘unstable complex and chaotic re-stabilizing’ and 2 inter-ventions per person in the edge of chaos group. Interventionsincluded advice to visit/call GP/practice nurse in response tosymptoms that were new or worsening including pain, and mentaldistress; to contact the pharmacist in relation to problems withmedication adherence; to contact public health or social servicesfor social, environmental or housing needs including heating. Car-egiver issues were addressed through referral to local services oralerting the GP.

Journey type Stable complex Unstable complex Unstable complex on the

edge of chaos

Death

Cohort 1 66.6% 26.8% 6.6%

Cohort 2

Rates of intervention (case management)

Phone call duration

67.4%

31 patients, 15 interventions

1–2 minutes

26.8%

12 patients, 16 interventions

>2–5 minutes

5.8%

3 patients, 6 interventions

>5 minutes

0

1

Figure 3 Frequency of ‘types’ of patientjourney over 4-week monitoring in cohort 1and interventions and call for cohort 2 –according to category of participant.

Complex adaptive care – patient journeys C.M. Martin et al.

© 2011 Blackwell Publishing Ltd4

DiscussionPatterns of patient journey in patients at high risk of hospitaliza-tion were identified using a CACC model. The majority of patientswere classified as stable complex, with no patients being simple orcomplicated. About 30% were unstable complex or on the edge ofchaos. Both cohorts scored highly on the probability of repeatadmissions score [10], indicating that OOH service contact mayoffer a potential screening opportunity for avoidable hospitaliza-tions. The lives of people with unstable journeys were both diffi-cult for them and their caregivers and challenging for the healthcare providers to continually monitor and respond to their riskof destabilization, even with frequent visits to the GP. Recogniz-ing the typologies of dynamics of patient journeys is a new andinnovative method of ongoing risk evaluation that enables theimplementation of complex adaptive care in response to the char-acteristics and domains of the at-risk dynamics. It overcomesthe current episodic care perspective ‘after the event’. Reasonsfor de-compensation and avoidable hospitalization can thus beaddressed by prospective analysis encompassing the whole bio-psycho-social environmental and illness treatment rather thansolely focusing on disease and functional status. Instability canstabilize or homeostasis can break down leading to death. Tradi-tionally, the GP knew their patients and their journeys, but asprimary care has become more specialized with a loss of personalcontinuity and with the developing roles of care management andguides, there is a need to externalize and make explicit the natureof patient journeys in order to plan and deliver integrated servicesthat meet the patient’s needs in a timely fashion.

The importance of personal continuity and social aspects ofdaily monitoring is evident within the typologies of participantgroups. Monitoring daily health also contributed greatly to socialcare with patients reporting the value of the monitoring as a socialoutlet and connection as well as a health monitoring system. Inter-ventions carried out varied from lifestyle change suggestions,advice to consult a medical professional, arranging immediatemedical attention in the case of a medical emergency, to interven-tions to improve social and environmental circumstances.

This paper presents a practical application of a number of theo-retical streams which converge in pattern recognition approachwithin a complex adaptive systems framework. These streamsinclude life events analysis which has been found particularlyuseful in understanding stress and depression in older people [11],observations of daily living input into personal health records[12]and the original chronic illness narrative research [13]. Furtherdevelopments will build on mixed-methods analytics – rangingfrom qualitative approaches to mathematical modelling techniquesand machine learning [14,15].

ConclusionThis paper presents a ‘novel tool’ to apply pattern recognitiontechniques to health care-related life event analysis such that itmay help health professionals predict illness trajectories towardstimely intervention. Observing the longitudinal patient journeythrough health and illness is central to providing successfulCACC. Types of patient journey in an older at-risk group with

chronic conditions can be stable or unstable, simple, complex orchaotic. Each pattern can be identified early and responds well toproblem-specific care approaches, be it medical, social or carersupport. CCAC is operationalized as adaptively responding tophases and instability in the patient journey.

References1. Martin, C. & Sturmberg, J. (2009) Complex adaptive chronic care.

Journal of Evaluation in Clinical Practice, 15 (3), 571–577.2. Wagner, E. H., Austin, B. T. & Von Korff, M. (1996) Improving

outcomes in chronic illness. Managed Care Quarterly, 4 (2), 12–25.3. Martin, C. M., Biswas, R., Sturmberg, J. P., Topps, D., Ellaway, R. &

Smith, K. (2010) Patient Journey Record Systems (PaJR) for prevent-ing ambulatory care sensitive conditions: a developmental framework.In User-Driven Healthcare and Narrative Medicine: Utilizing Colla-borative Social Networks and Technologies (eds R. Biswas & C. M.Martin), PP. 93–112. Hershey, PA: IGI Global.

4. Bigelow, J. H. (2005) Introduction. In Analysis of Healthcare Inter-ventions That Change Patient Trajectories (eds J. H. Bigelow, K.Fonkych, C. Fung & J. Wang), pp. 41–42. Santa Monica, CA: RandCorporation.

5. Martin, C. M., Biswas, R., Joshi, A. & Sturmberg, J. P. (2010) PatientJourney Record Systems (PaJR): the development of a conceptualframework for a patient journey system. In User-Driven Healthcareand Narrative Medicine: Utilizing Collaborative Social Networks andTechnologies (eds R. Biswas & C. Martin), pp. 75–92. Hershey, PA:IGI Global.

6. Murray, S. A., Kendall, M., Boyd, K., Grant, L., Highet, G. & Sheikh,A. (2010) Archetypal trajectories of social, psychological, and spiri-tual wellbeing and distress in family care givers of patients with lungcancer: secondary analysis of serial qualitative interviews. BritishMedical Journal, 340, c2581.

7. Kurtz, C. & Snowden, D. (2003) The new dynamics of strategy:sense-making in a complex and complicated world. IBM SystemsJournal, 42 (3), 462–483.

8. McDaniel, P. A., Intinarelli, G. & Malone, R. E. (2008) Tobaccoindustry issues management organizations: creating a global corporatenetwork to undermine public health. Globalization and Health, 4 (2), 2.

9. Young, D., Borland, R. & Coghill, K. (2010) An actor-network theoryanalysis of policy innovation for smoke-free places: understandingchange in complex systems. American Journal of Public Health, 100(7), 1208–1217.

10. Sidorov, J. & Shull, R. (2002) ‘My patients are sicker’: using the Prarisk survey for case finding and examining primary care site utilizationpatterns in a medicare-risk MCO. American Journal of Manag Care, 8(6), 569–575.

11. Kraaij, V., Arensman, E. & Spinhoven, P. (2002) Negative life eventsand depression in elderly persons: a meta-analysis. The Journals ofGerontology. Series B, Psychological Sciences and Social Sciences, 57(1), P87–P94.

12. Brennan, P. F., Downs, S. & Casper, G. (2010) Project health design:rethinking the power and potential of personal health records. Journalof Biomedical Informatics, 43 (5 Suppl.), S3–S5.

13. Glaser, B. & Strauss, A. (1967) Discovery of Grounded Theory. Strat-egies for Qualitative Research. New York: Sociology Press.

14. Murray, M. A., Fiset, V., Young, S. & Kryworuchko, J. (2009) Wherethe dying live: a systematic review of determinants of place of end-of-life cancer care. Oncology Nursing Forum, 36 (1), 69–77.

15. Cooper, G. F., Abraham, V., Aliferis, C. F., et al. (2005) Predicting direoutcomes of patients with community acquired pneumonia. Journal ofBiomedical Informatics, 38 (5), 347–366.

C.M. Martin et al. Complex adaptive care – patient journeys

© 2011 Blackwell Publishing Ltd 5