Patient transitions relevant to individuals requiring ongoing ventilatory assistance: a Delphi study

seven transitions. Transitions comprised: T1 – acute ventilation to PMV; T2 – PMV to LTMV; T3 – PMV or LTMV to acute ventilation (reverse transition); T4 – institutional to community care; T5 – no ventilation to requiring LTMV; T6 – pediatric to adult LTMV; and T7 – active treatment to end-of-life care. Subsequent Rs sought consensus. REsuLTs: Experts from intensive care (n=14), long-term care (n=14) and home ventilation (n=10), representing a variety of professional groups and geographical areas, completed all Rs. Consensus was reached on 14 of 20 statements defining T1 and 21 of 25 for T2. ‘Physiological stability’ had the highest consensus (97% and 100%, respectively). ‘Duration of ventila- tion’ did not achieve consensus. Consensus was achieved on 13 of 18 statements for T3 and 23 of 25 statements for T4. T4 statements reaching 100% consensus included: ‘informed choice’, ‘patient stability’, ‘informal caregiver support’, ‘caregiver knowledge’, ‘environment modification’, ‘supportive network’ and ‘access to interprofessional care’. Consensus was achieved for 15 of 17 T5, 16 of 20 T6 and 21 of 24 T7 items. CoNCLusioN: Criteria to consider during key care transitions for ventilator-assisted individuals were identified. Such information will assist in furthering the consistency of clinical care plans, research trials and health care resource allocation. et 21 des 24 points de T7. CoNCLusioN : Les chercheurs ont déterminé les critères à envisager pendant les principales transitions des soins aux personnes sous ventila- tion. L’information contribuera à améliorer la cohérence des plans de soins cliniques, des essais de recherche et de l’affectation des ressources en santé. critical care ventilation transition); T4 community T5 transition pediatric to adult Delphi of Delphi

While using duration of MV or other markers, such as tracheostomy placement or transfer of care to another centre, offer the advantages of simplicity, uniformity and relative ease of identification in the medical record or administrative databases (11), many clinicians will use other clinical markers to assist with indications for a change in clinical management, the overall plan of care or information provided to patients and their families. However, there is little information regarding the key transition points used by health care professionals that may reflect change across the spectrum of care. Moreover, such information needs to be broadly based to include many clinical specialties reflecting the continuum of care. Therefore, our aim was to identify defining features of seven key transition points across the care continuum for individuals requiring ongoing ventilator assistance based on expert-derived consensus. Our goal is that these definitional criteria will inform decision making by clinicians, administrators and policy makers.

METhoDs study design
A two-stage approach was used to develop consensus for defining features of care transitions. First, a national workshop was held that invited 34 key stakeholders including clinicians and administrators providing services in three distinct areas -intensive care, long-term care and ventilation in the community -as well as international faculty from the United States and Europe, to identify clinical, research and policy priorities for ventilator-assisted individuals (VAIs) (12). An outcome of the workshop was identification of seven important transition points across the care continuum, and the need for greater clarity surrounding clinically defining features. These transitions comprised: T1 -ventilation in the acute phase of illness to PMV; T2 -PMV to LTMV; T3 -PMV or LTMV to acute critical care ventilation (reverse transition); T4 -institutional to community care; T5 -being 'at-risk of but not currently receiving' to 'requiring' LTMV; T6 -transition from pediatric to adult LTMV; and T7 -active treatment to end-of-life care. In stage 2, a four-round (R) Delphi approach was used to establish consensus for the defining features of each transition. The Delphi approach is advocated as an effective tool for establishing consensus in health care (13).

Delphi participants
Invited participants comprised clinicians and administrators considered to have expertise in service provision for VAIs in an institutional (acute or long-term) or community setting. Participants were identified by the authors' national advisory group members. Additional participants required to meet purposeful diversity sampling targets were identified during development of a national inventory of service providers to VAIs (12). Invited participants also nominated potential participants either as a designate or as an additional participant. Purposeful diversity sampling was used to obtain professional (medicine, nursing, respiratory therapy and physiotherapy; adult and pediatric specialities) and geographical (across Canadian provinces) diversity representing the stakeholder groups. Although recommended minimum or maximum sample sizes for Delphi panels vary, key aspects include common sense and practical logistics (14). To achieve 10 participants representing each stakeholder group (acute, long-term institutional and community) the authors oversampled by 15 (45 participants in total) due to anticipated attrition. Only participants providing responses to the preceding R were invited to continue participation.

Delphi instrument and Rs
In R1, participants were provided with the title of the seven transitions listed above and requested to list, using free text, criteria they perceived should (Part A) and should not (Part B) define each. Participants were directed to consider potentially relevant timeframes, patient characteristics and physical locations. The R1 questionnaire weblink was provided by e-mail; three reminders were sent over a fourweek period. Raw data generated in R1 were then subjected to inductive content analysis to identify categories and to generate an appropriate statement representing each category to be included in R2 (see data analysis section). In R2, participants rated agreement with each statement on a five-point Likert scale. Three reminders were sent over a six-week period.
For R3, the questionnaire was modified by removing Part B and adding statements not already represented in Part A, ensuring neutral language. This was performed for two reasons: first, in R1, many criteria suggested to define a transition were also suggested as criteria that should not define the transition producing duplication. Second, participants reported that difficulty rating agreement with suggested criteria should not be considered as definitional because most were worded in a negative direction such as 'lack of expertise' or 'failure to consider patient wishes'.
In R3, participants re-assessed or confirmed R2 responses. The most common aggregate group response (mode) was provided in addition to the participant's own response. Some statements had two R2 modes; both were presented. Participants also rated level of agreement with the 'new' neutrally restated statements derived from Part B. In R4, participants re-assessed or confirmed responses to these new statements ( Figure 1).

Ethics considerations
Research Ethics Boards of the University of Toronto (#26199) and St Michael's Hospital (Toronto, Ontario) approved the study. Participation was voluntary and consent implied by questionnaire return.

Data analysis
Inductive content analysis of R1 data comprised, for each transition, independent reading and rereading, open coding, data grouping, category identification and sorting of items into categories (15) by three survey research unit members contracted to manage the Delphi process and five study investigators. Study investigators then identified an appropriate statement to represent each category for inclusion in R2.
Five-point Likert scale responses were grouped into three categories: agree, no opinion and disagree. Frequencies and percentages were calculated for each item on completion of R2, R3 and R4 to ascertain whether raw scores achieved >70% consensus. Although no standard threshold for consensus is recommended, >70% was selected because it has been recommended as a reasonable cut-off point (16). Failure to Figure 1) Delphi method. R Round gain consensus was defined as ≤70% consensus in either agree or disagree groupings. Stability of individual opinion was measured using change scores (mean ± SD) for each transition, with a score of 0 reflecting no change in responses across rounds.

REsuLTs
Of the 73 experts identified, 45 completed R1 and R2, 39 R3 and 38 R4 (84% retention). Fourteen participants represented intensive care, 14 long-term care and 10 home ventilation services; five specialized in pediatrics (Table 1). R1 generated 291 statements of criteria that should define the seven transitions and 221 of what should not. Inductive content analysis resulted in identification of 150 unique definitional criteria.

Transition from acute to PMV
Thirteen of 20 (65%) statements achieved consensus that they should define transition from ventilation during acute illness to PMV; one achieved consensus that it should not be included (Table 2). Six statements did not achieve consensus. These were as follows: 'possibility of future successful weaning' (68%); 'number of consecutive days on MV' (64%); 'patient transfer to an alternative care setting' (64%); 'tracheostomy in situ' (62%); 'number of consecutive hours of MV each day' (59%); and 'current care environment' (28%). Although not achieving consensus of >70%, ≥21 consecutive days (13 of 25 [52%]) and ≥6 consecutive h (21 of 23 [91%]) of MV were the most frequent responses in terms of duration from participants that perceived they should define transition.

Transition from PMV to LTMV
Eighteen of 25 (72%) statements achieved consensus that they should define transition from PMV to LTMV; three statements achieved consensus that they should not (Table 3). Four statements did not achieve consensus: 'minimum number of consecutive hours on MV' (69%), 'inadequate attempts at weaning' (66%), 'minimum number of consecutive days of MV' (59%) and 'family/informal caregiver capability/resources to assume care' (14%). Although not achieving consensus, ≥30 days (14 of 23 [61%]) and ≥6 h (24 of 27 [89%]) were the most frequent responses in terms of duration from participants perceiving they should define transition.
Transition from institutional care to care within the community (home/assisted living) Twenty-one of 25 (84%) statements achieved consensus that they should define transition to community care; two statements achieved consensus that they should not (Table 5). Two statements did not achieve consensus: 'patient's diagnosis' (62%) and 'location of patient's home' (16%).

Transition from 'no ventilation' to 'requiring' LTMV
Ten (59%) of 17 statements achieved consensus that they should define transition from absence of ventilation to requiring LTMV; five   statements achieved consensus that they should not (Table 6). Two statements did not achieve consensus: 'inadequate treatment of underlying disease' (69%) and 'willingness/ability of family/friends/caregivers to provide support' (68%).

Transition from pediatric to adult LTMV services
Thirteen of 20 (65%) statements achieved consensus that they should define transition from paediatric to adult LTMV; three statements achieved consensus that they should not (Table 7). Four statements did not achieve consensus: 'patient's cognitive maturity' (70%), 'financial resource availability' (58%), 'physical maturity' (55%) and 'increased role of patient as opposed to parents in directing care' (18%).

Transition from active treatment to end-of-life care for PMV or LTMV
Seventeen (71%) of 24 statements achieved consensus that they should define transition to end-of-life care; four statements reached consensus that they should not (Table 8). Three statements did not achieve consensus: 'resource availability for ongoing care' (63%), 'family/caregiver expectations in terms of prognosis' (55%) and 'loss of communication ability' (13%).

stability of change scores
The

DisCussioN
The present study was the first to use an anonymized Delphi-derived expert consensus of national scope, with representation across professions, regions and differing health provision care sectors (acute, longterm and community) as well as adult and pediatric specialists, to identify defining features of important transitions across the care continuum for VAIs. Consistent across the seven transition points were: safety, including physiological stability; consideration of patient/family wishes and motivation based on informed choice; prognosis and anticipated quality of life; establishment of a transition plan that includes redefinition of care goals; and timely availability of adequate resources. These features represent a global set of issues for consideration during health care transition for VAIs. Concern for patient safety and the need for physiological stability recognizes the potential reduction in care intensity due to different staffing models either in the same care location or in an alternative venue such as a long-term acute care hospital, specialized weaning centre, long-term care facility or assisted living unit (4,17). The need to consider patient and family wishes enabled through informed choice, including an understanding of prognosis and anticipated quality of life, was encouraging to note because it reflects acceptance of the importance of collaborative communication processes and shared decision making endorsed by professional societies (18,19).  Establishment of a transition plan, a care routine and redefinition of care goals were recognized as key in many transitions, consistent with an understanding that a health care transition is a process as opposed to an event (20). Transition is frequently required due to a change in health status, symptoms or functioning (21), regardless of whether it requires a change in physical location or service providers. Timely availability of adequate resources was highlighted as a defining feature for transition from institutional to community care, consistent with studies in both adult and pediatric populations that have identified issues associated with delays and gaps in service provision and potential out of pocket expenses of family caregivers (22)(23)(24)(25).
Contrary to previous definitions (1,2), the number of consecutive days of ventilation did not exceed the threshold of 70% used to identify consensus in defining PMV or LTMV. Although participants did not regard these arbitrary time points to be pivotal in changing the overall plan of care in the present survey, in some jurisdictions, timebased changes in ventilation may trigger resource availability, which is a key determinant of level of care (4,5). Moreover, the lack of consensus should not infer discontinuing these time-based criteria at present because they remain important bedside clinical markers that may continue to play a role in clinical and research trials. Repeated unsuccessful weaning attempts and an indefinite need for invasive or noninvasive ventilation were considered to be strong indicators of transition.
There are several reasons why consistency of definitions that demarcate patient transitions is important. First, clinicians may benefit from being aware of factors that may alert them to the need for reconsidering their care plan, the information they provide to patients and their families as well as the care locations that may encourage a more rehabilitative focus. Second, policy decisions that determine resource allocations are simplified if descriptions of incidence and prevalence rely on consistent criteria (1). Third, such consistency promotes accurate benchmarking of outcomes such as survival, hospital length of stay, functional dependence or health care utilization, as well as various care models across different jurisdictions (26,27).
Limitations of the present study include the study being restricted to one country. Although we received advice regarding the selection of transitions from international faculty, the extent to which the identified criteria are consistent with perspectives from other countries with different professional outlooks and health care systems remains to be explored. Such an understanding would be an essential next step if our   definitions would be of benefit to an international consensus panel interested in advancing guidelines in this area. Despite diversity in sampling, there is the potential that some defining features of transition were not identified. Those that were identified constitute a checklist that will require refinement and reduction before being of value for decision making at the bedside. Despite reaching consensus using a predefined threshold of >70% on several transition statements, few reached 100%, underscoring the subjectivity, range of opinion and potential differences in health care organization. Notwithstanding the above, our observations provide useful information for streamlining care plans and achieving increased clarity for the care of those who require ongoing ventilator support.