Primary conceptual analysis of grounded theory data

The primary definitions and conceptual analysis of stress and crisis presented in the first part of this paper were developed from a grounded theory study whose goal was to develop a substantive theory of how caregivers navigate autism-related care from the time they first perceive developmental concerns in their child [35]. Importantly, that study’s aim was not to develop a full theory of stress, and thus our conceptual analysis does not provide a comprehensive mechanistic account. The grounded theory methods and rigor used for this study [36–39] have been described previously [7, 40]. The primary data comprised 45 verbatim transcripts of 90-minute intensive interviews with 32 caregivers (four completing two interviews) and 9 professionals selected for extensive experience supporting caregivers—representing a large participant and interview data sample for grounded theory [41].

As described previously [7, 40], caregiver backgrounds varied widely in terms of rural (22%) and urban (78%) status, child ages (range 2.5–18 years), child’s autism severity level, number of autistic children (13% with multiple, maximum five children), ethnocultural backgrounds (34% non-Canadian), education (from high school diploma [9%] to graduate degree [16%]), and time spent navigating care at time of interview (<1 year to 8+ years). Of all aspects of caregivers’ backgrounds, time navigating care was the most relevant metric from life course and process-oriented perspectives for analyzing differences in stress as it related to the evolving process of navigating care. Importantly, the concept and terminology for reporting children’s “autism severity level” (mild, moderate and severe) below is outdated due to the version of the diagnostic manual, the DSM-IV-TR, that was in use when most children in this study received their autism diagnosis; it provides only an imperfect estimate of child support needs.

Using a flexible interview guide, iteratively revised to address emerging theoretical foci according to theoretical sampling, participants were asked about experiences and actions navigating autism-related care over the child’s life course. Participants all described a prominent role for stress, regardless of background, and the caregiver and professional interview guides were adjusted early in the study to include questions about stress and crisis (see S1 Appendix). Interviews thus produced a substantial body of qualitative data on the topic, including antecedents and consequences of stress, which enabled the current conceptual analysis.

All interview data were open-coded and incorporated into the final analysis. Analysis included constant comparison throughout coding and category development, analytic memo writing, developing abstract conceptual diagrams, and final integrative writing of the theory. Firsthand person-centered perspectives were prioritized both by the grounded theory methods, and the social theory of symbolic interactionism [42] used as an explicit analytic framework. Verification of the evolving categories proceeded with successive rounds of data collection informed by the ongoing analysis, per theoretical sampling. Notably, the typology of stress elaborated below was tested in a dedicated internal verification step by selectively coding roughly 50 previously incompletely-coded but relevant interview segments (open coded early in the grounded theory analysis prior to analytic development of the typology). Analysis and data management were supported by software (NVivo; QSR).

Role of a priori knowledge

Given the existing theory on stress, an important methodological consideration for this study is when and how possibly relevant prior theory was used in the grounded theory analyses regarding stress and crisis. Grounded theory methods require that the substantive theory produced be developed firstly from the empiric-level study data. Furthermore, while any pre-existing knowledge the analyst arrives with is accepted as an inevitable source of theoretical sensitivity that informs the resulting analysis and theory construction, researchers are not expected and even discouraged from seeking additional knowledge of relevant theory until analysis is complete [36, 37, 43]. Thus, the lead analyst did not expand their knowledge of stress theory until study completion. It remains important for grounded theory analysts to report reflexively on their prior theoretical knowledge, and how it influenced their analysis [44]. In this case, prior knowledge of formal stress theory was minimal and limited to Seyle’s General Adaptation Syndrome [45]. This physiological stress theory was applied as a metaphor for organizing the time-evolving aspects of caregivers’ stress process. Its coherence (consistency) with the empirical data was verified throughout the analysis per grounded theory requirements. The constructs presented in this paper were not explicitly informed by other well-known stress theories, although they have been related to other such theories in the discussion.

Crisis definition

The definition for crisis, meanwhile, was influenced by prior knowledge of the definition used in developing the BFDS (described above for measuring proximity to crisis among caregivers of autistic children) [30]:

“An acute disruption of psychological homeostasis in which one’s usual coping mechanisms fail and there exists evidence of distress and functional impairment. The subjective reaction to a stressful life experience that comprises the individual’s stability and ability to cope or function. The main cause of a crisis is an intensely stressful, traumatic, or hazardous event…” [46]

This definition was adapted and revised through constant comparative analysis to fit caregivers’ responses to questions about what crisis meant to them, and to iterative interpretation of coded examples of crisis in the data.

Conceptual verification and extension using survey data

In the subsequent analysis exercise to verify and extend the primary conceptual analysis on stress, we coded stress-relevant qualitative data from a separate survey study, whose methods were described previously [19]. The survey’s purpose was to explore the impacts of local autism policy, and barriers and facilitators to accessing private and publicly funded services. It featured 179 closed and open-ended questions, completed online by 654 Ontario caregivers of autistic children three years after the grounded theory data collection—representing a geographically similar population in which both samples shared the same autism policy and service landscape. Nine open-ended questions containing stress-relevant data were coded (see S2 Appendix). Unlike questions in the grounded theory interview guide, these asked less directly about stress or crisis, and several questions addressed service-related aspects expected to be associated with stress. Codable responses ranged from several words to several lines of participant-typed text.

Framework analysis approach

The framework analysis approach followed to code survey responses has been used to analyze primary data in qualitative research in similar situations [32–34]. Framework analysis has been further adapted as an analytic method for synthesizing published qualitative study findings [47, 48], resulting in the related method named best-fit framework synthesis [49–52]. In both methods, new qualitative data are analyzed using a pre-existing theoretical framework, chosen for its relevance to the topic under study, and used as an initial coding framework. Themes, categories, or concepts from the pre-existing framework—in this case, the stress-related typology and categories from the grounded theory study—are thus used as an a priori coding structure. This framework functioned to organize the data to be coded, but was left open to refinement and extension as data from the new source (the open survey responses) were compared but found to represent novel ideas unaccounted for by the framework, allowing inductive generation of new open codes. Open codes from the survey data were considered new if they did not have a similar counterpart within the grounded theory coding structure.

Results of the survey data analysis do not feature example quotes because of ethics requirements. Thus, results of the survey data coding and framework analysis exercise is presented in aggregate summary form only. We present these results as answers to two questions derived from the aims of framework analysis: 1) What can be said about how the survey data replicates the stress-related coding structure from the grounded theory study? and, 2) How does the survey data extend the stress-related coding structure of the grounded theory study? In the discussion, we consider the methodological learnings and utility of the framework analysis exercise using survey data.

Ethics statement

The grounded theory study described in this report received ethics approval from the Hamilton Integrated Research Ethics Board, and all participants provided prior written and ongoing verbal consent for interviews (recruitment: 2011-October-24 to 2012-November-26). The survey study described in this report received ethics approval from the Wilfrid Laurier University Research Ethics Board, and all participants provided electronic written consent (recruitment: 2018-May-03 to 2018-July-22).

Definitions of stress and crisis

Stress was defined as a caregiver’s subjective emotional and physiological experience (which caregivers described in emotional and physical terms) and response to a triggering thing or event that they perceive as a threat to personal well-being (including that of a child) or continued ability to function. The physiological idea of stress from Seyle’s General Adaptation Syndrome [45]—where stress is understood to mobilize increased cognitive and physical energy for responding to threatening situations in both the short term (alarm, or fight-or-flight, reaction) and the medium term (resistance), but leads to negative consequences when stress is overly prolonged (exhaustion)—was a useful, empirically coherent analogy for interpreting and explaining caregiver accounts of stress in its varying forms. Specifically, it helped account for the earlier empowering consequences of stress in terms of caregivers’ motivation and capacity for action, the destructive consequences in terms of its effects on caregivers’ emotional and even physical well-being, and the evolving nature, or process, of stress. This analogy fit empirically with the timeline of the different consequences of stress manifested in most caregivers’ trajectories of navigating care—with effects on motivation and capacity for action happening earlier in the trajectory, and negative consequences happening throughout but changing in nature, with exhaustion manifesting over the longer term. The rough timing of these differing sources and consequences of stress is depicted in the model below (Fig 1), with time in caregivers’ care navigation trajectory represented from left to right. The model of four main categories (causes) of stress derived from the grounded theory is elaborated (with examples from the data) after introducing the definition for crisis, below.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Model of four main categories (causes) of caregiver stress, and their consequences for caregiver action and emotional and physical well-being over time, derived from the grounded theory analysis.

https://doi.org/10.1371/journal.pone.0312391.g001

Crisis was defined as an event occurring when caregiver stress (due to parenting, navigating care, etc.) reaches or passes a threshold level that disrupts homeostasis of a caregiver-related functioning system (listed below), resulting in a sudden or progressive loss of function, and sometimes the sense that a catastrophic failure is imminent. Crisis compels the caregiver to respond by seeking either resolution to the problem, or some other change from the status quo to restore balance. Failure of the system can occur but is generally rare. Two important qualifications captured caregivers’ varying experiences and accounts of crisis:

1. Different functioning systems can be in crisis including the caregiver’s self (comprising emotional/psychological, physical, or financial aspects), the family, spousal unit, child, or caregiver-child unit. Thus, for any example of crisis it is important to identify the system that crisis centers around.
2. The triggering cause of crisis can include not only a single acute stressful event, but also (more commonly) a progressive buildup of stress where the final threshold-breaking event can seem relatively benign (like the straw that breaks the camel’s back). One caregiver described crisis this way:

“I would say the crisis then for us was several peaks like that where they were just too close…I don’t think there’s ever been a time where it’s been one specific [trigger of] crisis. It’s always been too many things. When it rains, it pours.” (mother from non-metropolitan city, one 17-year-old autistic son, originally diagnosed with what the parent recalled as ‘severe autism’)

In most cases in the data, caregivers clearly and explicitly attributed crisis to a buildup of stress-related precursors. In the following caregiver’s example, the buildup of stressors led to multiple systems threatened by crisis, and ultimately her response to restore balance: Dilin (pseudonym; divorced mother of one 6.5-year-old autistic son, originally diagnosed as ‘severe’ per parent’s recollection, and two non-autistic siblings) described stress from exhaustion due to prolonged expendable resource loss (i.e., of time, finances, and physical and emotional reserves for coping), which became compounded by a series of additional stressors. She suffered from both generalized anxiety disorder triggered by navigating intervention, and depression due to overextending herself that led her to shut the world out. Additionally, she described how a combination of disagreement with her husband about approaches to intervention and her neglect of their relationship after dedicating most of her time to navigating, led to marital crisis ending in divorce. Her ex-husband’s subsequent unwillingness to provide any financial support, combined with the fact they had previously re-mortgaged the house to finance intervention, led Dilin to financial crisis in the form of personal bankruptcy, where putting food on the table for her three children became a greater concern than her son’s autism. The stress of these events, combined with the underlying autism-related stresses and the associated self-neglect, finally took a toll on Dilin’s physical health. She developed hypertension and was even hospitalized for a “mild heart attack” while home alone caring for her children. This experience of physical crisis, combined with her doctor’s warnings that she “had to find ways to focus on [her]self,” forced her to recognize the catastrophic possibility that her children could potentially be left with nothing if she continued at the same pace without tending to her own health. This was the breaking point that pushed her to come to peace with her limits and seek balance both physically and emotionally with a new psychological strategy. “I gave myself permission basically to take some time off,” she said, and further reflected that, “one of the things I have learned since then is that you need to change your expectation level when you’re dealing with autism.” Other caregivers also described a desperate need for “change in direction” that they achieved through psychological strategies or by changing their level of functioning.

While less common in the data, some crises were caused by single acute events. In one example, a caregiver described a single mother she knew who had an autistic young adult daughter. She was running an errand, with her daughter briefly alone at home, when she had an episode of atrial fibrillation complicated by a seriously low heart rate for which she later required a pacemaker. When the ambulance arrived, she explained she could not go to hospital because her daughter was at home and could not care for herself for long periods. After signing a waiver with the ambulance attendant, she drove home despite the strain of her low heart rate, where she called the local autism service agency to arrange for emergency respite before finally admitting herself to the emergency department. In this case, respite relieved the parent’s potentially fatal medical crisis.

Categories of stress

The model of four main categories (i.e., causes or sources) of caregiver stress, and their consequences for caregiver action over time is depicted in Fig 1. Causes (or sources) of stress could be distinguished from consequences of stress only by analytic reference to the specific empirical instance of stress being described. Thus, common causes of stress (e.g., sleep deprivation, marital problems), could often alternatively be defined as consequences depending on the specific case of stress emphasized in the caregiver’s narrative. There were many stories where stress had a feed-forward influence—consequences subsequently causing more stress. The model was developed and verified by tracing numerous paths from causes to consequences, and relating each empirical instance to the stage the caregiver was at in their journey (or process) navigating autism intervention. Each category of stressor is defined and linked to general types of consequences for action or emotional well-being, below. Table 1 shows example quotes and other data from the source grounded theory study that highlight each category.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Four categories of stressors with example data from the grounded theory study.

https://doi.org/10.1371/journal.pone.0312391.t001

1. Sources of urgency to take action included the following: a) feelings of anxiety about possible future outcomes of autism (e.g., while coming to understand the child has autism [40], or when encountering significant obstacles to intervention), b) the experience of seeing one’s child struggle, c) knowing the importance of intervening early, d) learning it’s up to caregivers to take action, and e) guilt for not doing enough.
   While still considered distressing and worrisome by caregivers and linked to varying levels of strain on their mental health, these sources of urgency generally increased their motivation and capacity to engage in navigating intervention. Such stressors also had the potential to drive caregivers to use personal expendable resources at a rate that was not sustainable in the long term. While the depleted expendable resources this caused down the line represented another stressor (see the fourth category), most participants reported not regretting their losses.
   The timeline and intensity of this category of stress can be mapped onto the trajectory of changes in readiness or motivation for involvement in autism-related care, characterized in detail previously [7]. Namely, it corresponds to the early stage of going into high gear, the rapid (steep) increase in motivation for involvement, which occurred for most parents over a period of months starting soon after first noticing developmental concerns in their child. It contrasts, meanwhile, with the subsequent period of easing off, defined by a slow (gradual) decline in motivation for involvement generally stretching over several years.
2. Buildup of multiple interacting sources refers to situations in which stress attributable to the often-overwhelming demands of navigating intervention coincided in time with other stressors such as problematic child behaviors, demands of another newborn, needs of typical siblings, death in the family, or physical aspects of caregivers’ daily living environment.
   It was important to consider how such contextual and potentially difficult-to-identify sources could combine to exacerbate the negative consequences of caregivers’ navigation-related stress.
3. Obstacles causing feelings of helplessness refers to caregivers’ encounters with setbacks to their efforts to obtain intervention, which they described as causing a sense of helplessness (loss of control), despair (emotional loss), or reduced capacity to cope emotionally. Obstacles include system-imposed limitations such as inadequate supply of funded intervention resulting in waitlists and qualifying criteria that excluded certain children; and individual actors within the system (professionals, agencies, schools) blocking access to such solutions through discretionary decisions, and sometimes involving interpersonal confrontation, conflict, and power imbalance—contributing stress attributable to the process of negotiating obstacles. This resulted in delaying, diminishing or fully denying autism-related care or accommodations sought by the caregiver, which in turn threatened the child’s current or future well-being—causing further stress. Additionally, obstacles often worsened caregivers’ burden in terms of the scarce expendable resources already invested in pursuing the solution being denied, the extra work of pursuing subsequent alternatives and workarounds, or of otherwise coping with the fallout of unmet needs.
   These obstacles to obtaining care or supports were arguably the most needless and unhelpful sources of stress, especially given the mental anguish they caused. Consequently, some caregivers and professionals highlighted them as priority targets for policy change.
4. Depleted resources reducing capacity to cope refers to the progressing deficits in caregivers’ expendable resources as they were used to meet the work-related demands of autism (parenting and navigating intervention). These sources of stress were analogous to exhaustion because they compromised caregivers’ capacity to handle further challenges. Caregivers often responded by shutting down further resource use, or seeking balance.
   This form of stress could have potentially serious consequences by feeding forward to fuel new forms of stress, with one or more possible breakdowns in function (crisis), and further reduced capacity to cope or function.
   Collectively, the harmful consequences of stress included becoming overwhelmed (manifesting as denial, avoidance), self-neglect (disrupted sleep, eating, self-care), onset or aggravation of depression, anxiety, obsessiveness, neglect of family members, guilt, family and marital dysfunction (anger, blame, conflict), divorce or separation, and financial distress.

Survey data coding and framework analysis exercise

Results of the survey data coding and framework analysis exercise reported here include observations regarding replicability of the grounded theory coding structure, how this coding structure was extended by new open codes from the survey data, and methodological limitations and learnings from applying this approach.

Replicability of grounded theory coding

S3 Appendix shows the stress-relevant codes that the survey data were coded to—organized within (i.e., as smaller sub-categories of) the four major stress categories described above, and the two categories, consequences and crisis. They include both original codes from the grounded theory study, and new open codes generated to account for novel ideas recognized after engaging with the survey data. Table 2 summarizes the number of codes (or sub-categories) applied to the survey data within each of these six major categories, and number of references from the survey data to those codes—providing a rough indication of the extent to which the major grounded theory categories were replicated by the survey data. (Note, the more extensive grounded theory study included numerous additional codes not used to code the survey data). Table 2 also indicates the number or new open codes generated from the survey data, and the number of references to them—providing an indication of which categories were extended by the new survey data.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Summary of survey data coding to six major stress-related categories from the grounded theory study.

https://doi.org/10.1371/journal.pone.0312391.t002

Each major category had some coverage (indicated by number of codes) and level of coding (indicated by number references coded) from the survey data. The first stress category, sources of urgency to take action (the largest in the original grounded theory study), had the lowest coverage and level of coding from the survey data. Being action-oriented, this category was more amenable to development in the grounded theory study, which was heavily focused analytically on action and process [7], and therefore dependent on accessing participants’ unfolding stories to a deeper level than is practical in surveys with brief open-ended responses.

The third stress category, obstacles causing feelings of helplessness, had the highest coverage (number of codes) and level of coding (references). Survey questions about stressors affecting mental health, frustrations with the education system, and reasons why their child had not received adequate support, were the greatest contributors to this category (S2 Appendix). The available data yielded high levels of coding to the specific subcategories (codes) of feeling loss of control, complexity as a setback to access, waitlists, experiencing conflict and lacking expertise within the education system, and a lack of educator capacity to keep the child in school (see S3 Appendix).

The category, crisis, was relatively sparsely represented in the survey data, but nevertheless had references coded to all seven of the functioning systems described in the grounded theory-derived definition described above (S3 Appendix). Consequences of stress had broad coverage in the sense that a wide variety of effects were represented in the survey data (S3 Appendix), especially related to physical and mental health (for which there were specific survey questions), and self-care.

Survey data coded to the above categories—i.e., excepting the second and fourth stress categories—generally contributed few or no new codes or references to new codes. Results for these two categories are described next.

Extending the grounded theory coding structure with new survey data-derived codes

The second stress category, buildup of multiple interacting sources, was noteworthy for being the most extended by the survey data, with seven new highly referenced codes (S3 Appendix)—the highest number of new codes. All seven reflected the same novel analytic insight: that any family members with special conditions requiring the caregiver’s attention (including the caregiver’s children, parent(s), spouse, or self) represented a condition contributing to the buildup of multiple interacting sources because they reduced expendable caregiver resources and capacity to respond to stress. This insight was also supported by the grounded theory data. For example, one of the new codes, multiple autistic children, was also true for four (13%) caregivers interviewed in the grounded theory study. The survey questions that were the greatest source of coding for this category asked about the effects of stressors on caregiver mental health, and about the contributions of family mental and physical health problems to stress (S2 Appendix).

The fourth stress category, depleted resources reducing capacity to cope, was only slightly extended by survey data coding. The survey question producing most data coded to this category asked how stressors impacted caregiver physical health (S2 Appendix). New codes related to time burdens attributable to being a single parent, the lack of capacity for seeing a personal doctor for self-care needs, and effects of physical health problems on mental health and psychological capacity. Additionally, there were extensive references to existing codes of financial and emotional burdens, and to needing a break due to exhaustion.

Crisis definition

We developed the caregiver-centered definition of crisis to expand on and update the only prior published definition of crisis we were aware of that is autism-specific [46, as cited in Weiss & Lunsky, 2011]. Our definition features two important new elements with implications for evaluating crisis in this population. First, it explicitly acknowledges that different systems can be in crisis from the caregiver’s perspective (the self, marital unit, family, child, or caregiver-child unit). Second, it explicitly recognizes that crisis can be triggered by progressive buildups of stress, and not just acute (major) triggering events. We are unaware of measurement tools that currently consider these caregiver-relevant aspects. Addressing this measurement gap could improve research understanding and facilitate selecting appropriate targets for intervention when supporting caregivers currently in, or at risk of, crisis, with potential downstream benefits for caregiver and family well-being.

Comparison to prior stress theory

There are several reasons to compare the conceptual analysis of stress developed from grounded theory data with prior formal stress theories. We followed grounded theory methods [36, 37, 43] by not expanding our knowledge of prior theory until after the study to promote analytic openness to new concepts from the empirical setting. Except for the use of Seyle’s General Adaptation Syndrome [45, 53], knowledge of prior formal stress theory was low during the analysis, and did not explicitly influence the resulting definitions and typology. While this theoretically agnostic approach is a strength for developing grounded theory, it may be considered a limitation from other perspectives if the results do not fit with accepted propositions from formal theory. Comparison to prior formal stress theory is therefore important to appraising our analysis and informing decisions about its use. The comparison we provide here includes a consideration of 1) elements from prior theory that are unaccounted for by our analysis, to highlight key aspects from prior stress theory that are useful to consider in addition to what we have presented, 2) consistencies with elements of prior theory, to show the elements from our analysis that reinforce prior theory and vice versa, and 3) novel differences of our analysis, to suggest possible new insights that could extend prior theory.

Novel ideas our analysis adds

A significant new contribution of our analysis involves its conceptualization of social-psychological stress as an evolving process analogous to the physiological conception of stress contained in Seyle’s General Adaptation Syndrome model [45, 53]. This metaphor was adopted because it accounted for observed progressive changes in caregivers’ responses to, and effects of, stress over time. Namely, it fit the observations of stress motivating and increasing caregivers’ capacity for action early on, and later causing exhaustion and breakdown. This time-varying characteristic is missing from other models discussed, and to our knowledge has not previously been applied to social-psychological conceptualizations of stress. Meanwhile, this novel process-oriented understanding is compatible with at least two prior conceptions of stress as an evolving process. Patterson’s FAAR model [54] represents the process of stress by outlining how families respond to demands with available capabilities differentially within two smaller processes, pre-crisis adjustment and post-crisis adaptation. And as discussed, Pearlin’s [55] Stress Process model recognizes the longitudinally varying nature of stress both by emphasizing the interaction of two types of stressors defined by duration, discrete events and enduring life strains, and by its mechanistic path from stressful event to stress outcome along which mediators can intervene. While both models resemble our analysis by having early and late phases whose durations are flexible, the different representations of process cannot be assumed to line up across models. Finally, we suggest that our representation of process has potential implications for the measuring stress. Namely, it provides a structure whereby varying measures can be understood (or designed) to differentially measure the intensity of different phases of the stress process—targeting increased motivation and capacity for action in the short- and medium-term, and exhaustion in the long-term. This would fit with Pearlin’s suggestion to pair the measure of stress to the known source of stress by relying variously on situational (short-term) or global (long-term) assessments [55].

Learnings and limitations of the framework analysis exercise

The framework analysis exercise yielded several methodological observations with implications for future use of survey data to extend prior empirical theory. Importantly, while some of the primary grounded theory analysis was reinforced by the survey data coding, it was unreasonable to expect extensive reproduction of its coding structure for at least two reasons. First, the number of stress-relevant codes that could be applied to the survey data was much smaller than for the original grounded theory. This was for two reasons: a larger grounded theory interview dataset relevant to stress (notwithstanding its smaller participant sample) compared to the survey dataset (with only 1,226 mostly brief codable responses); and the relative inability to access depth, complexity, and process with static survey questioning compared to the live, adaptive, and process-focussed interviewing method used in the grounded theory study. Second, the two studies produced subtly different areas of stress-relevant content, both due to different data-generation methods, and to differences in their areas of inquiry.

One survey question—about stress effects on caregivers’ physical (medical) health—stood out for producing relevant data on a topic that had little coverage in the grounded theory study. Analysis of this data led to a new insight that extended the grounded theory: that sometimes-serious physical health problems caused by caregiving-related stress could reduce caregivers’ capacity for meeting autism-related care demands, both by directly affecting physical capacity, and through effects on mental health and psychological capacity. This demonstrates the potential value of open survey data for extending conceptual analyses from prior qualitative studies. It also highlights the possibility of increasing utilization of survey data, which commonly go under-used. To promote more widespread availability of qualitative survey data for secondary analysis, we suggest survey researchers consider how to embed ethical processes that allow survey participants to consent to such re-use.

While survey data can be useful to extend conceptual analyses, we note two limitations of this data type. First, open participant-typed responses were noticeably less articulate than for transcribed interviews—due to typos, abbreviations, inattentive writing, varying literacy levels, etc. Unlike interviews, surveys do not allow returning to clarify intended meanings. Second, due to their brevity and typical focus on single timepoints, open survey questions are poorly suited to gathering information needed to understand action as a process that evolves over time or about relative timing of events needed to understand unfolding narratives. Reflecting this limitation, one respondent articulated frustration with not being able to share more within the survey format about how certain stressors had changed over the course of her journey. Grounded theory interviewing, by contrast, is suited to reflect process and narrative timelines [36, 37]—in this case, the evolving nature of and caregiver responses to stress over their trajectory navigating autism-related care.