Completion of Patient-Reported Outcome Questionnaires among Older Adults with Advanced Cancer

Abstract

Context:

Systematic collection of patient-reported outcomes (PROs) reduces symptom burden and improves quality of life. The ability of older adults to complete PROs, however, has not been thoroughly studied.

Objectives:

To determine whether older adults with advanced cancer received assistance completing PROs, the nature of the assistance, the factors associated with receiving assistance, and how the prevalence of assistance changed over time.

Methods:

Data were obtained from a multisite cluster randomized controlled study of geriatric assessment (Clinicaltrials.gov: NCT02107443). Adults ≥70 years with advanced cancer completed multiple PROs at 4 time points (enrollment, 6 weeks, 3 months, 6 months). Factors associated with receipt of assistance were assessed with bivariate and multivariate analyses.

Results:

The study included 541 adults (range 70–96 years, 49% female, mixed incurable cancer diagnoses). Twenty-eight percent (153/541) received assistance completing PROs. Of these, 42% received assistance from caregivers, 37% from research staff, and 15% from both. Factors associated with receiving assistance included older age [Adjusted Odds Ratio (AOR) 3.71, 95% Confidence Interval (CI) 1.03–13.38], lower education level (3.92, 2.11–7.29), impaired cognition (1.90, 1.23–2.93), impaired functional status (2.16, 1.33–3.52), and impaired hearing (1.38, 1.05–1.80). Eighty percent of individuals who received assistance were identified at study initiation. Receiving assistance decreased over time from 28% to 18%, partially due to drop-outs.

Conclusion:

Over a quarter of older adults with advanced cancer in this study received assistance completing PROs. Completing PROs is a key aspect of many clinical programs and cancer trials; assistance in completing PROs should be offered and provided.

Introduction 3485/3500

The importance of patients’ voices in the provision of person-centered care and evaluation of new treatments is increasingly recognized in oncology (1, 2). Patient voices are most often conveyed through the use of patient-reported outcomes (PROs). PROs are data provided by patients, in contrast to data generated by clinicians or from laboratory and radiological studies. Studies have demonstrated that collection and monitoring of PROs [i.e., symptoms, quality of life (QoL), and performance scales] with subsequent feedback provided to oncology clinicians improves outcomes. These outcomes include QoL, symptoms, patient-provider communication, satisfaction with care, and survival (3–9). Although the importance of incorporating PROs as a standard of care in oncology (5, 7) is clear, collection of PROs in clinical practice remains variable and inconsistent. User’s Guides to PROs implementation provide strategies to address clinician- and system- related barriers (e.g., limited provider buy-in, poor integration with electronic medical records, and work flow burden) (7, 10–13). However, these guidelines do not address the specific barriers that older adults encounter in the completion of PROs.

Older adults comprise the majority of individuals diagnosed with and dying from cancer and are a population projected to increase (14–16). Practice guidelines from the American Society of Clinical Oncology (ASCO) [for Geriatric Assessment (GA) (2) and palliative care (17)] and the International Society of Geriatric Oncology (SIOG) [guidelines for routine monitoring of symptoms and QoL in the older adult cancer population (2, 18, 19)] recommend routine completion of PROs by older adults with cancer (20). Concerns about possible burdens associated with completion of PROs discourage acceptance of these guidelines, and implementation of guideline-based care requires knowledge of the resources required for older adults to complete PROs.

While studies have shown that collection of PROs is feasible, older adults have often been underrepresented in study populations (21–23). Research staff have reported that completion of PROs was more challenging for older and ill individuals with cancer (22). Although data from GA trials suggest that many older adults with cancer can complete PROs (24–27), they also indicate that a subset of older adults requires assistance. Ironically, older adults with advanced cancer may be a group which could benefit most from a focus on improving symptoms and promoting QoL (17).

Two juxtaposing threats may limit the ability of older adults to benefit from implementation of PRO-based initiatives. First, they may not be approached to participate because of concerns about potential burden or inability to complete PROs. Second, older adults may be asked to complete PROs but not offered the assistance which may be required because of aging-related needs (e.g. visual impairment). Potential disparities in care can be further exacerbated if older adults cannot report PROs independently.

To better understand the receipt of assistance for PRO completion among older adults with cancer, a secondary analysis of data from a large multi-site cluster randomized clinical trial that tested the efficacy of GA guided intervention was conducted. In this trial, data on PRO completion and receipt of assistance with PRO completion were collected from participants. The study protocol allowed for research staff and caregivers to provide assistance to patients completing PROs. The objective of this analysis was to determine whether older adults with advanced cancer received assistance with completion of PROs, what kind of assistance was provided, and what factors were associated with receiving assistance with PRO completion. We also examined the trajectory of receiving assistance with PRO completion over 6 months and the relationship of receiving assistance with drop-out.

Methods

Study Design, Setting, and Participants

This was a secondary analysis of data from a nationwide GA cluster randomized controlled study conducted within the University of Rochester Cancer Center National Cancer Institute Community Oncology Research Program (clinicaltrials.gov: NCT02107443). Participants were enrolled from 31 community oncology settings across the United States. The institutional review boards at the University of Rochester and at each participating site approved the study, and all patients provided informed consent.

The primary study aim, which was previously reported (27), was to examine the effects of a GA intervention on discussions of aging-related concerns and patient and caregiver satisfaction with care (28). Eligibility criteria included age ≥70 years old, diagnosis of an advanced solid tumor malignancy or lymphoma (stage IV cancer or stage III if cure is not possible or anticipated), considering or receiving any kind of cancer treatment, have planned visits with oncology for at least 3 months, and have impairment in at least one GA domain (e.g., functional status, cognition) other than polypharmacy (Supplementary Table 1).

All patients in the study received a GA at baseline, which included eight domains (e.g. function status, nutritional status); each domain was scored as impaired or not impaired according to pre-established cut points (28). The intervention included a tailored list of recommendations which were given to the oncologist immediately prior to a clinic visit for discussion with the patient. For example, an individual with impaired physical status would receive recommendations for fall prevention and a referral to a physical therapist. In the usual care arm, oncologists were notified only of impaired scores in depression or cognition. As part of the study, patients were asked if they received assistance with completion of self-report measures, which is the focus of this analysis.

Measures

Patients completed study questionnaires at baseline and at 4–6 weeks, 3 months, and 6 months. At each time point, participants were asked to complete 12–22 pages of questionnaires, with a median of 4 questions per page (range 1–16) which included standardized scales. PRO questionnaires were completed on paper using standardized TeleForms (scanable forms for data capture), which were formatted in Arial 12-point font. Examples of PROs collected during the trial included a symptom scale (MD Anderson Symptom Inventory), a QoL measure (Functional Assessment of Cancer Therapy- General), and multiple functional measures (Activities of Daily Living, Instrumental Activities of Daily Living). These measures have been described in prior reports (29). A complete listing of PROs is provided (Supplemental Table 2).

Dependent Variable – receiving assistance with PROs completion

Included was a survey completion questionnaire that asked a) where the participant completed the questionnaires (home, doctor’s office, home and doctor’s office, or other location) and b) if someone assisted them in completing the questionnaires (yes or no). If participants answered yes to receiving assistance, they were asked c) who provided assistance completing the questionnaires (caregiver, research staff, caregiver and research staff, or other), and d) how the person(s) assisted (read the questions, wrote the answers, answered the questions for the patient, or assisted in some other way).

Covariates

Demographic, clinical, and aging-related factors that might be associated with receiving assistance with PRO completion were assessed. These factors included patient-reported demographics [age (70–79 vs. 80–89 vs. >90 years), gender (female vs. male), race (white vs. other), education (some college or above vs. high school graduate vs. less than high school), living situation (lives alone vs. not), marital status (partnered vs. not partnered)], Karnofsky Performance Status (KPS; ≤60 vs. 70. vs. 80 vs. 90 vs. 100), and cancer type (collected by study staff from the medical record; categorized into gastrointestinal vs. lung vs. other based on frequencies and clinical relevance). All aging-related factors were assessed using GA [i.e., comorbidity, vision (poor vs. fair vs. good vs. excellent), hearing (poor vs. fair vs. good vs. excellent), polypharmacy, physical performance, functional status, psychological health, nutritional status, cognition, and social support; see Supplemental Table 1 for a complete list of all measures included in the GA]. GA measures were scored following pre-established cut points as impaired vs. not impaired (29). For covariates for which many categories of data were collected, some categories were combined based on clinical relevance specific to older adults with advanced cancer, in order to achieve fewer groups with larger sample sizes.

Statistical Analyses

Descriptive statistics were used to summarize findings. Bivariate analyses (Chi-Square test) were used to describe associations of patients’ demographic, disease, and aging-related factors with receiving assistance with PRO completion at baseline. Logistic regression analysis was used to evaluate the multivariate relationship. The pool of independent covariates was predefined based on clinical relevance (all candidate covariates are listed in Table 1). The final model was selected using a stepwise selection process (SAS: proc logistic with stepwise option) with a p value of 0.15 chosen for entry and elimination from the model. A p value of 0.15 was chosen to balance between reducing the number of parameters and model fit. P=0.15 is close to the critical level (0.157) for which a stepwise procedure is asymptotically equivalent to select the model covariates based on Akaike Information Criterion (AIC)(30, 31). Covariates in the final model were examined for collinearity using variance inflation factor (VIF). No serious collinearity was detected (all VIF<1.3, where VIF=3 is a threshold for concern). Interactions of age and education variables with aging related conditions and performance status were examined (p>0.05 for all).

Table 1.

Demographics and clinical characteristics at baseline: Bivariate analysis for group differences

	Individuals who did not receive assistance with PROs	Individuals who received assistance with PROs	% Received assistance	P valuea
	N=388	N=153
Demographic and clinical characteristics	N (%)	N (%)	%
Age b
70–79	298 (76.8)	103 (67.8)	25.7	0.017
80–89	85 (21.9)	42 (27.6)	33.1
90–96	5 (1.3)	7 (4.6)	58.3
Gender b
Female	192 (49.5)	84 (55.3)	30.4	0.227
Male	196 (50.5)	68 (44.7)	25.8
Non-Hispanic white b
No	36 (9.3)	22 (14.5)	37.9	0.080
Yes	352 (90.7)	130 (85.5)	27.0
Education b
<High school	31 (8.0)	35 (23.0)	53.0	<0.001
High school graduate	132 (34.0)	63 (41.4)	32.3
>Some college or higher	225 (58.0)	54 (35.5)	19.4
Living arrangements b
Lives with others	304 (78.4)	124 (81.6)	29.0	0.405
Lives alone	84 (21.6)	28 (18.4)	25.0
Marital status
Married/partnered	138 (35.6)	54 (35.3)	28.1	0.952
Not partnered	250 (64.4)	99 (64.7)	28.4
Karnofsky performance status c
≤60	27 (7.0)	32 (21.1)	54.2	<0.001
70	38 (9.8)	16 (10.5)	29.6
80	141 (36.4)	57 (37.5)	28.8
90	120 (31.0)	29 (19.1)	19.5
100	61 (15.8)	18 (11.8)	22.8
Cancer type
Gastrointestinal	104 (26.8)	35(22.9)	25.2	0.408
Lung	103 (26.5)	37 (24.2)	26.4
Other	181 (46.6)	81 (52.9)	30.9
Aging-related factors
Vision
Poor	9 (2.3)	8 (5.2)	47.1	0.005
Fair	55 (14.2)	36 (23.5)	39.6
Good	241 (62.1)	89 (58.2)	27
Excellent	83 (21.4)	20 (13.1)	19.4
Hearing
Totally deaf	0 (0)	1 (0.7)	100	0.001
Poor	15 (3.9)	17 (11.1)	53.1
Fair	90 (23.2)	43 (28.1)	32.3
Good	213 (54.9)	77 (50.3)	26.6
Excellent	70 (18)	15 (9.8)	17.6
Impaired polypharmacy
No	73 (18.8)	15 (9.8)	17.0	0.011
Yes	315 (81.2)	138 (90.2)	30.5
Impaired comorbidity
No	151 (38.9)	45 (29.4)	23.0	0.038
Yes	237 (61.1)	108 (70.6)	31.3
Impaired social support
No	266 (68.6)	119 (77.8)	30.9	0.033
Yes	122 (31.4)	34 (22.2)	21.8
Impaired cognition
No	284 (73.2)	77 (50.3)	21.3	<0.001
Yes	104 (26.8)	76 (49.7)	42.2
Impaired nutritional status
No	163 (42.0)	52 (34.0)	24.2	0.086
Yes	225 (58.0)	101 (66.0)	31.0
Impaired physical performance
No	27 (7.0)	7 (4.6)	20.6	0.304
Yes	361 (93)	146 (95.4)	28.8
Impaired functional status
No	186 (47.9)	36 (23.5)	16.2	<0.001
Yes	202 (52.1)	117 (76.5)	36.7
Impaired psychologic status
No	310 (79.9)	95 (62.1)	23.5	<0.001
Yes	78 (20.1)	58 (37.9)	42.6

^aDifferences assessed with Chi-Square Test

^b1 patients had missing data

^c2 patients had missing data

Patient Reported Outcomes (PROs)

To examine longitudinal aspects of receiving assistance, the data were visualized graphically using Sankey flow diagrams (constructed by Python software). Sankey was selected for visualization because it demonstrated the flow of patients from one assessment to the next assessment. Patients were categorized as receiving assistance (help) vs. not receiving assistance (no help) with PROs completion. The figure incorporates drop out (collection of PROs was discontinued). Association of receiving assistance at baseline with the timing of drop out was evaluated by the Kruskal-Wallis Test. Statistical analyses were conducted using the SAS software package (version 9.4).

Results

The parent study included 541 older patients [mixed cancer diagnosis (26% gastrointestinal, 26% lung cancer), all had advanced incurable cancer]. Mean age was 77 years [standard deviation (SD) 5.2, range 70–96], 49% were female, 89% were white, and 48% had a high school or less education. Individuals reported a mean of 4.5 (SD 1.5; range 1–8) impaired GA domains (Table 1). Over the six months of the study, 35% of patients withdrew, most commonly due to death or hospitalization/hospice. Other detailed participant characteristics have been previously reported (29, 32).

At study entry, 62% (337/541) of patients reported completing the surveys at home, 29% (155/541) at the doctor’s office, 4% (19/541) at both places, and 6% (30/541) at another location. Seventy-two percent (388/541) completed the questionnaires independently, with 28% (153/541) receiving assistance from someone at baseline screening. All PROs were completed with very minimal (<1%) missed items.

Of the 153 patients who reported receiving assistance at screening, 42% (64/153) received assistance from their caregiver, 37% (57/153) from research staff, 15% (23/153) from both, and 6% (9/153) from others (Figure 1a). Most of the assistance was provided in the form of reading questions or writing down answers for the patient (75%; 114/153): specifically, 44% (68/153) both reading questions and writing answers, 22% (33/153) reading questions only, and 8% (13/153) writing answers only. An additional 18% (27/153) reported receiving assistance in another way (e.g., clarification), and 8% (12/153) reported that the questions were answered for them (Figure 1b). On the free text box, some participants commented about other assistance received and noted that assistance was in the form of explanation or clarification of a question, and they also stated that assistance was received for only one or a few questions and not for all.

Figure 1a. Who provided assistance in completing PROs.

Figure 1b. Types of assistance provided for completing PROs.

On bivariate analyses, patients who were older, those with lower educational levels, lower KPS, or impairment in vision, hearing, polypharmacy, comorbidity, functional status, psychological status, and cognition were more likely to report receiving assistance completing questionnaires (see Table 1). Compared to those with impairment in social support, those without impairment were more likely to report receiving assistance completing questionnaires (31 vs. 22%, p=0.03).

On multivariate analysis (Table 2), factors independently associated with receiving assistance included older age [90–96 years vs. 70–79 years, Adjusted Odds Ratio (AOR) 3.71, 95% Confidence Interval (CI) 1.03–13.38], lower educational level (high school graduate vs. some college or higher, AOR 1.80, 95% CI 1.14–2.84; less than high school vs. some college or higher, AOR 3.92, 95% CI 2.11–7.29), impaired cognition (AOR 1.90, 95% CI 1.23–2.93), impaired functional status (AOR 2.16, 95% CI 1.33–3.52), and impaired hearing (AOR 1.38, 95% CI 1.05–1.80). In contrast, patients who had impaired social support were less likely to report receiving assistance (AOR 0.55, 95% CI 0.34–0.90).

Table 2.

Multivariate logistic regression analysis: Associations with receiving assistance completing Patient Reported Outcomes

Variable	Category	Odds ratio	95% CI	P-value
Age	80–89 vs 70–79	1.25	(0.76–2.04)	0.377
	90–96 vs 70–79	3.71	(1.03–13.38)	0.045
Education	<HS vs some college or higher	3.92	(2.11–7.29)	<.0001
	HS graduate vs some college or higher	1.80	(1.14–2.84)	0.011
Karnofsky performance status	Excellent to poor	1.02	(1–1.04)	0.082
Hearing	Excellent to poor	1.38	(1.05–1.80)	0.020
Medical social support	Impaired	0.55	(0.34–0.90)	0.017
Cognition	Impaired	1.90	(1.23–2.93)	0.004
Functional status	Impaired	2.16	(1.33–3.52)	0.002
Psychologic status	Impaired	1.57	(0.99–2.51)	0.056

Abbreviations: HS = high school, CI =Confidence Interval

C statistics=0.75

CI Confidence Interval

Longitudinal examination revealed that both the percent and actual number of individuals reporting receiving assistance with PROs completion decreased over time: 28% at baseline (153/541), 22% at 4–6 weeks (104/472), 20% at 3 months (80/409), and 18% at 6 months (62/349). As displayed in the Sankey flow diagram (Figure 2), the pattern of responses over the 6-month course of the study indicates that the majority of individuals who received assistance were identified at study initiation and that this group of individuals was more likely to withdraw from the study than those who did not receive assistance (p=0.03). We further examined this finding by comparing individuals at baseline and at the 4–6 week follow-up time point. Approximately half (52%, 79/153) of the individuals who received assistance at baseline continued to receive assistance at time 2, while 18% (27/153) did not provide data at the second time point (due to drop out), and the remainder (31%, 47/153) no longer received assistance. In contrast, only 24% (25/104) of individuals received assistance at the 4–6 week interval who had not received assistance at baseline.

Figure 2. Sankey flow diagram of receiving assistance completing PROs.

Note: Help is substituted for assistance due to space constraints

Green –did not receive assistance

Red- received assistance

Gray- dropped out of study

Note: The diagram visualizes within-subject flow (change) from one time point to the next time point. The height of the bars corresponds to the number of patients. To read the Sankey, focus on one color. Focusing on red at baseline (the patients who received help), the arc/flow of these patients to assessment time point 2 (4–6 weeks) demonstrates that the majority still needed help (largest red arc), while a smaller group no longer needed help (smaller red arc), and an even smaller group dropped out.

Discussion

Two threats to older adults from benefitting from PROs were outlined in the introduction, and our findings addressed both. First, the risk that potential burden or inability to complete PROs may exclude older adults from being asked to provide PRO data. Second, the failure to offer assistance to older adults to complete PROs may prevent them from full participation - assistance which may be required because of aging-related needs. While we found that 72% of older adults with advanced cancer were able to complete PROs independently, over a quarter of them required assistance. Our study indicates that assistance in completing PROs should be provided to older adults through implementation of strategies and work flow processes. We also found that when provided a choice, more than 60% of older adults chose to complete questionnaires at home, indicating that home completion may be a viable and preferred choice for many older individuals when the option and timing are appropriate for data collection. Caregivers frequently provide assistance to patients in completing PROs. In addition, the type of assistance provided (i.e., reading questions and writing responses) and the characteristics associated with receiving assistance were quantified [e.g. educational level, age, impairments in functional level and psychological status].

Both QoL scales and PROs used in GA (24) can generally be completed by patients independently without assistance (33, 34). Similarly, commonly used symptom scales are reported to be feasible (35–37), but specific details on whether assistance is required to complete them is often not provided. In prior studies, the proportion of older adults who required assistance with GA questionnaires has ranged from 8% (38) to 14–24% (39), consistent with our findings. For example, Hurria (27) and colleagues found that 78% of older adults were able to independently complete the PROs in GA (administered electronically), although the participants had a mix of curable and advanced cancers. Results from our study provide evidence that many older adults can independently complete multiple PROs and expand the findings to those with advanced cancer. Whereas past studies have reported on completion of GA questionnaires, symptom scales, or QoL instruments individually, our study included all of these.

We found that 28% of older adults received assistance with PROs. Of these, over half (52%) received assistance from research staff. Patients frequently received assistance from caregivers in this study, and while this did not place resource demands on the research team, it did reveal an under-recognized role of caregivers. In this study, we specifically asked what type of assistance was provided, and the most common form of assistance was reading and/or writing responses for the patient. When assistance was provided, it was rare that questions were answered for the patient, indicating completion by proxy was not a major occurrence. Having self-report measures completed by proxy is known to provide different results from when the patient reports independently (40, 41). Further research is indicated to examine in greater detail the role of caregivers who assist patients in completing PROs.

Several key factors associated with receiving assistance were identified. Consistent with prior research, we found that lower educational levels were associated with receiving assistance (22). Our findings that older adults were more likely to receive assistance is aligned with prior reports that these individuals have more difficulty with completing PRO-CTCAE (Common Terminology Criteria for Adverse Events) (22). Impairment in multiple GA-related factors (i.e., in hearing, cognition, and functional status) were associated with receiving assistance, indicating the utility of GA in identifying or screening at risk individuals. In contrast, patients who had good social support were more likely to receive assistance completing PROs. One possible interpretation for this finding is that individuals with impaired social support did not have a caregiver to ask for assistance and thus completed PROs independently. Social support is a critical and important factor related to QoL outcomes in older adults (42, 43). Shahronki and colleagues (44) noted differences in CTCAE toxicity based on social support in older individuals. Individuals with poor social support were less likely to have documented CTCAE non-hematologic toxicities, supporting the importance of monitoring older adults using PROs.

Many of the individuals who were most likely to receive assistance with PROs were identifiable at study initiation, indicating that up front screening measures can be implemented to identify at risk individuals. It is imperative that there is equitable participation of vulnerable high risk groups such as older adults with advanced cancer in both clinical trials and clinical initiatives. To achieve this, procedures need to be in place to screen for at risk individuals so that assistance for PRO completion can be provided in this population. Our study may serve as a guide to estimate resource and budgetary requirements for clinical initiatives or research studies involving PRO collection from older adults with cancer. It is unsurprising that individuals who received assistance were more likely to drop out of the study because of death, hospital admission, or enrollment in hospice, perhaps indicating that requiring assistance with PRO completion is associated with more complex issues. This group of individuals can often be identified at the first time they are asked to complete; therefore, we recommend addressing by planning to offer assistance. One straightforward approach is to ask all older adults, “Do you need any help with completing these forms?” Another option is to screen for risk factors with GA to identify individuals who may need assistance with PROs. Practical implications include creating questionnaires that are easy to read, with specific font types (serif) and size (16) recommended for older adults. Recommendations and guidance are available from several organizations including the National Institute on Aging (45), the US Department of Health and Human Services Centers for Medicare & Medicaid Services (46), and the Office of Disease Prevention and Health Promotion (47).

A limitation of this study is that receiving assistance in our sample may have been overestimated; anecdotally, several research staff mentioned that they volunteered to assist individuals with PRO completion to expedite the process and to meet study requirements and timeframes. Assistance by research staff may have been provided when it was not required; thus, the percentage may be an overestimation of individuals who actually required assistance. Second, results may only be generalizable to similar patient populations; participants were drawn from multiple community oncology settings, had advanced cancer, and agreed to participate in a study and did not represent a racially diverse population. Paper questionnaires were used rather than electronic or digital capture of information; the percentage receiving assistance may not transfer to electronically completed formats which have been discussed by others (48). However, results of those receiving assistance in our study are similar to those reported for electronic data capture of PROs by other investigators (22, 23, 49). Challenges (e.g., reading questions, clarification of a question) faced by older adults when completing paper-based assessments are also relevant in the context of electronic assessments. While the use of electronic reporting systems to gather PROs is increasing (50) it is likely that paper-based assessments will remain for some older adults. The reasons for this could be due to patient factors (no access to electronic devices/internet) or system issues (variation in information technology support, resources, and infrastructure).

The importance of PROs for the older adult, especially in the context of advanced cancer where QoL outcomes may be prioritized more highly than length of life (20, 51), is well established. In order to advance our understanding of the tolerability of cancer symptoms and treatments in older adults, this information is essential. Our findings indicate that an identifiable group of older adults with advanced cancer will require assistance completing PROs. Individuals are the only source of information on QoL, and their symptom experience and the value of collecting patient-reported data cannot be overestimated(52).

Key Message:

Over a quarter of older adults with advanced cancer in this study required assistance in completing PROs from staff or family caregivers. Assistance with PROs should be provided to older adults through implementation of strategies and work flow processes.