Continuity and Coordination of Care for Childhood Cancer Survivors with Multiple Chronic Conditions: Results from the Childhood Cancer Survivor Study

Abstract

Introduction:

Continuity and coordination-of-care for childhood cancer survivors with multiple chronic conditions are understudied but critical for appropriate follow-up care.

Methods:

From April-June 2022, 800 Childhood Cancer Survivor Study participants with ≥2 chronic conditions (≥1 severe/life-threatening/disabling) were emailed the “Patient Perceived Continuity-of-Care from Multiple Clinicians” (CC-MC) survey. The CC-MC asked about survivors’ main (takes care of most healthcare) and coordinating (ensures follow-up) provider, produced three care-coordination summary scores (main provider, across-multiple-providers, patient-provider partnership), and included six discontinuity indicators (e.g., having to organize own care). We defined discontinuity (yes/no) as poor care on ≥1 discontinuity item. Chi-square tests assessed associations between discontinuity and socio-demographics. Modified Poisson regression models estimated prevalence ratios (PRs) for discontinuity risk associated with the specialty and number of years seeing the main and coordinating provider, and PRs associated with better scores on the 3 care-coordination summary measures. Inverse probability weights adjusted for survey non-participation.

Results:

377 (47%) survivors responded (mean age 48, 68% female, 89% non-Hispanic White, 78% privately-insured, 74% ≥college graduate); 147/373 (39%) reported discontinuity. Younger survivors were more likely to report discontinuity (chi-square p=0.02). Seeing the main provider ≤3 years was associated with more prevalent discontinuity (PR; 95%CI) (1.17; 1.02–1.34 vs ≥10 years). Cancer specialist main providers were associated with less prevalent discontinuity (0.81; 0.66–0.99 vs. primary care). Better scores on all 3 care-coordination summary measures were associated with less prevalent discontinuity: main provider (0.73; 0.64–0.83), across-multiple-providers (0.81; 0.78–0.83), patient-provider partnership (0.85; 0.80–0.89).

Conclusion:

Care discontinuity among childhood cancer survivors is prevalent and requires intervention.

Précis:

Childhood cancer survivors with multiple chronic conditions require care from a variety of providers, but the coordination and continuity of their care is understudied. Among Childhood Cancer Survivor Study participants with 2+ chronic conditions (with at least 1 severe/life-threatening/disabling), nearly 40% reported discontinuity in their care.

INTRODUCTION

As a result of treatment advances, better risk stratification, and improved supportive care, there are approximately 500,000 adult survivors of childhood cancer in the US.^1–5 Unfortunately, due to their cancer and its treatment, these survivors experience a high burden of chronic conditions such as cardiovascular disease, endocrine dysfunction, and subsequent neoplasms.^6–10 The Childhood Cancer Survivor Study (CCSS), the largest cohort of childhood cancer survivors in North America,¹¹ has quantified this burden using the Common Terminology Criteria for Adverse Events.⁸ They found that the cumulative incidence of a severe, disabling, life-threatening, or fatal health condition by age 50 is 54% in survivors vs. 20% in their siblings.⁸ In a clinically assessed cohort, by age 50, childhood cancer survivors had an average of 17 chronic health conditions versus 9 in community controls.⁷ Previous research with the CCSS cohort demonstrated that most survivors (81%) were seen by a primary care provider (PCP) but that survivors with the highest level of chronic health conditions were more likely to seek and receive additional specialty care.¹² However, the continuity and coordination of survivors’ care has been understudied.

Continuity and coordination of care involve having “a longitudinal relationship with a single identifiable provider and cooperation between providers and between venues of care.”¹³ In cancer survivors and other populations, better continuity and coordination lead to improved outcomes such as better symptom control, fewer medical errors, decreased hospitalizations and emergency department (ED) encounters, and lower costs.^14–15 Continuity and coordination of care are particularly important for patients with multimorbidity¹³ and can be measured regardless of the specific chronic conditions being treated. Further, multimorbidity among adult survivors of childhood cancer poses unique challenges related to continuity and coordination of care due to the need for multidisciplinary care (PCP, oncology, subspecialty)¹⁵ for which the relative roles and responsibilities of individual providers may not be clear. Further complicating matters are disparities in care, which are common for healthcare in general, as well as for childhood cancer survivors.¹⁶

To improve care and outcomes based on the previous three decades of childhood cancer survivorship research, it is essential to evaluate the continuity and coordination of care for childhood cancer survivors and identify gaps that can be targeted for intervention. In this study, we explored the continuity and coordination of care received by adult survivors of childhood cancer with multiple chronic conditions.

MATERIALS & METHODS

Study Design and Procedures

This study was conducted in the CCSS cohort, a multi-institutional, multi-disciplinary collaborative research resource comprised of >24,000 survivors of childhood cancer diagnosed between 1970–1999 (U24CA55727; PI: Dr. Greg Armstrong). CCSS participants were diagnosed with cancer prior to age 21 years old and survived at least 5 years. All participants are now adults.

From April 18-June 8, 2022, randomly selected CCSS participants with multiple chronic conditions were invited for one-time completion of the “Patient-Perceived Continuity of Care from Multiple Clinicians” (CC-MC) survey.^17–18 Email invitations were sent with a survey link and included contact information (email and tollfree telephone number) for questions about the survey, which was only available online. Up to 6 reminders (2 text messages [for those with known textable numbers] and 4 emails) were sent to invited participants who had not yet responded. No monetary (or other) incentive was offered for completion. This study was reviewed and approved by the St. Jude Institutional Review Board.

Participants

A random sample of 800 childhood cancer survivor participants in the CCSS with two or more chronic conditions (≥1 severe or life-threatening/disabling) were invited to participate from a total eligible sample of approximately 3300. Our target sample size was 400 respondents, which was considered sufficient for the planned exploratory analyses, and we assumed a 50% response rate. Eligible participants were identified based on the chronic conditions reported in their most recent CCSS survey available at the time of invitation (completed between 2014 and 2016). To classify the severity of chronic conditions, we used Oeffinger et al’s methods¹⁹ to apply the CTCAE²⁰ grades to the comorbidities reported through the CCSS: 1=mild, 2=moderate, 3=severe, 4=life-threatening or disabling, 5= fatal. To be eligible for this study, the CCSS participant must have had ≥2 chronic conditions Grade 2–4, with ≥1 condition Grade 3 or 4. Example grade 3–4 conditions are shown in Table 1. Updated information on the sample’s chronic condition burden was collected between 2019 and 2021 and available for analysis in 2023. Those data are reported to update the chronic condition burden experienced by the sample nearer the time of this study’s survey. Other eligibility criteria included being alive, residing in the US, and ability to complete the data collection in English.

Table 1.

Demographic and Clinical Characteristics of Survivors of Childhood Cancer by Completion Status

		Survey Completers (N=377)		Survey Non-Completers (N=423)
		N	%	N	%	P-value
Sex	Female	256	67.9	229	54.1	<.0001
	Male	121	32.1	194	45.9
Race and ethnicity	Non-Hispanic White	337	89.4	371	87.7	0.37
	Other Race/Ethnicity	37	9.8	50	11.8
	Unknown	3	0.8	2	0.5
Current age	25–39	80	21.2	103	24.3	0.03
	40–49	123	32.6	163	38.5
	50+	174	46.2	157	37.1
Year of cancer diagnosis	1970s	131	34.7	132	31.2	0.52
	1980s	153	40.6	176	41.6
	1990s	93	24.7	115	27.2
Primary cancer diagnosis	Central Nervous System	52	13.8	84	19.9	0.06
	Leukemia/Lymphoma	194	51.5	195	46.1
	Solid tumors	131	34.7	144	34.0
Education	<College graduate	100	26.5	193	45.6	<.0001
	College graduate	277	73.5	230	54.4
Household income	<20,000	36	9.5	67	15.8	0.0002
	20,000–39,999	33	8.8	61	14.4
	40,000–79,999	86	22.8	94	22.2
	≥80,000	190	50.4	155	36.6
	Unknown	32	8.5	46	10.9
Residence 1	Metropolitan	311	82.5	327	77.3	0.05
	Nonmetropolitan	52	13.8	80	18.9
	Unknown	14	3.7	16	3.8
Private insurance 2	Yes	293	77.7	78	18.4	n/a
	No	79	21.0	24	5.7
	Unknown	5	1.3	321	75.9
Grade 3–4 chronic conditions used to evaluate study eligibility (collected 2014–2016)	Examples
Subsequent malignant neoplasm	Invasive malignancy	100	26.5	83	19.6	0.02
Hearing	Loss of hearing	41	10.9	83	19.6	0.0006
Vision	Blindness in 1–2 eyes	43	11.4	65	15.4	0.10
Endocrine	Diabetes requiring insulin	142	37.7	142	33.6	0.23
Pulmonary	Emphysema requiring medication	14	3.7	15	3.5	0.90
Cardiac	Myocardial infarction	118	31.3	126	29.8	0.64
Gastrointestinal	Cirrhosis	35	9.3	44	10.4	0.60
Renal	Dialysis	13	3.4	16	3.8	0.80
Musculoskeletal	Amputation	62	16.4	52	12.3	0.09
Neurological	Paralysis	39	10.3	57	13.5	0.17
Grade 3–4 chronic conditions updated using 2019–2021 data	Examples
Subsequent malignant neoplasm	Invasive malignancy	111	29.4	97	22.9	0.04
Hearing	Loss of hearing	47	12.5	94	22.2	0.0003
Vision	Blindness in 1–2 eyes	54	14.3	69	16.3	0.44
Endocrine	Diabetes requiring insulin	163	43.2	153	36.2	0.04
Pulmonary	Emphysema requiring medication	19	5.0	15	3.5	0.30
Cardiac	Myocardial infarction	141	37.4	138	32.6	0.16
Gastrointestinal	Cirrhosis	40	10.6	45	10.6	0.99
Renal	Dialysis	17	4.5	20	4.7	0.88
Musculoskeletal	Amputation	72	19.1	56	13.2	0.02
Neurological	Paralysis	48	12.7	65	15.4	0.29

P-values based on chi-square comparison among those with known values.

¹Based on Rural-Urban Commuting Area (RUCA) codes of metropolitan vs. micropolitan/small town/rural.

²Insurance only known for a subsample (n=112) of non-completers so p-value not calculated.

Measures and Variables

CCSS data on-file were used to summarize the age, race, sex, and chronic conditions of both survey completers and non-completers. CC-MC survey responses were used to summarize education, income, and insurance for survey completers. For survey non-completers, CCSS survey data from 2017–2019 were used to summarize education, income, and insurance. Last known residence was classified as metropolitan vs. micropolitan/small town/rural using U.S. Census Rural-Urban Commuting Area (RUCA) coding.²¹

The CC-MC asked respondents whether they had a main provider (doctor, physician assistant, nurse practitioner) “who takes care of most of your healthcare” and a coordinating provider (doctor, nurse, other) “who ensures the follow-up of your healthcare.” The main provider and coordinating provider could be the same person. For this study, we added questions regarding how long participants had been seeing the provider, the provider’s specialty (PCP, cancer specialist, other), and whether the provider practiced at a specialized cancer survivorship clinic.

The CC-MC has 9 subscales, summarized in Table 2.^17–18 Three subscales address the main provider: coordinator role, comprehensive knowledge of the patient, confidence and partnership. Four subscales address care across multiple providers: confidence in team (2 items), role clarity and coordination within clinic (3 items), role clarity and coordination between clinics (3 items), and information gap between clinicians (6 items). Two subscales cover patient as a care partner: care plan (7 items) and self-management information provided (4 items). Notably, “care plan” in the CC-MC is not the same as a “survivorship care plan” used in the cancer context but more generally refers to providers explaining health conditions and their treatments, tests and follow-up, and consideration of patient goals. When applicable, a 6-month recall period is used. All scales were rescaled to 0–100 by dividing the maximum possible value and multiplying by 100. So that higher scores would consistently reflect better continuity, items were reverse-scored if needed.

Table 2.

Description of “Patient-Perceived Continuity of Care from Multiple Clinicians” (CC-MC) Subscale Content¹⁷

Subscale	Number of Items	Description of Content
Subscales Addressing the Main Provider
Coordinator Role1	5	How well coordinating provider knows patient’s health care needs, keeps in touch with the patient, coordinates with other providers, and helps patient get care from other providers
Comprehensive Knowledge of Patient2	4	How well provider takes into account the patient’s whole medical history, what worries the patient most about health care, the patient’s responsibilities at work/home, and the patient’s personal values
Confidence and Partnership2	3	Whether the patient thinks the provider considers the patient’s ideas important, is comfortable discussing personal problems related to health with the provider, and is confident that provider will take care of the patient regardless
Subscales Addressing Care Across Multiple Providers
Confidence in Team	2	Whether patient feels “known” by care team and can count on all care team members for help
Role Clarity and Coordination Within Clinic	3	Frequency patient was told different things, providers did not work well together, and providers did not know who was doing what
Role Clarity and Coordination Between Clinics	3	Frequency patient was told different things, providers did not work well together, and providers did not know who was doing what
Information Gap Between Clinicians	6	Frequency provider did not know medical history, have access to test or exam results, repeated tests, was unaware of recommended treatment changes, relied on patient to convey medical information
Subscales Addressing Patient as Care Partner
Care Plan	7	Whether providers explained the impact of health conditions, why and how to take recommended treatments, tests and exams needed, and why and how to follow up with other clinicians; whether providers considered patient goals and discussed how to achieve them
Self-Management Information	4	Whether providers informed patients about what to do to stay healthy/improve health, how to take recommended treatments, and how to cope with complications

¹Only asked to those reporting ‘Yes’ to ‘Thinking about ALL the persons you saw in ALL different places you went for your care in the last 6 months, is there one who ensures the follow-up of your healthcare (doctor, nurse, other)?

²Only among those reporting ‘Yes’ to ‘Do you have a provider (doctor, physician assistant, nurse practitioner) who takes care of most of your healthcare’

The CC-MC also includes six stand-alone indicators of discontinuity. We developed a dichotomous variable based on the six stand-alone items to serve as the dependent variable. Specifically, respondents were considered to have experienced discontinuity if they reported that their health care was “hardly organized at all” or “somewhat” organized; or feeling that they have to organize their care “more than they would like” or “too much”; or answered “yes, often” to feeling that no one was in charge of their healthcare, or feeling abandoned by the healthcare system, or going to ED for care, or physical/emotional health suffering due to poorly organized care. For 4 of the 6 items, respondents are asked to specify the reason(s) for the discontinuity by checking boxes from a list of options, selecting all that apply.

The CC-MC is applicable to a broad range of health conditions, including multimorbidity, and is appropriate for use in an ambulatory setting.¹⁷ Internal consistency reliability for the 9 subscales exceeds 0.80 in all cases, except for reliability for the 3-item role clarity and coordination within clinic (alpha=0.66).¹⁷ The CC-MC’s validity has been demonstrated in 376 adults with diverse health conditions seeing clinicians in various settings.¹⁷

Statistical Analyses

Using chi-square tests, we compared the characteristics of invited participants who did and did not complete the questionnaire. We calculated the proportion of participants who experienced discontinuity. Chi-square tests assessed univariable associations between discontinuity and socio-demographics (White non-Hispanic vs. other race/ethnicity; private vs. non-private insurance; age 25–39, 40–49, 50+ years; college graduate or more vs. not college graduate; and male vs. female).

We calculated the frequencies for the presence and specialty of main and coordinating providers. Using age adjusted modified Poisson regression models with robust errors to estimate prevalence ratios (PRs), we examined the relationship between discontinuity and provider specialty (primary care, cancer specialist, non-cancer specialist), practice at a specialized survivorship clinic (yes/no), and length of relationship (0–3 years, 4–9 years, ≥10 years) for the main provider, and separately for the coordinating provider. Two observations were excluded from models examining length of relationship due to outlier responses (−1 and 103 years, respectively). Similar age adjusted regression models estimated PRs for risk of discontinuity associated with 10% improvements on the subscales and summary scores for the three levels of care coordination (main provider, across multiple providers, and patient-provider partnership). For both sets of models, inverse probability weights were applied to adjust for survey non-participation, with weights determined from logistic models of participation among all invited survivors, including sex, age, era of diagnosis, race, education, income, and cranial radiation as predictors. No correction for multiple testing was included in these descriptive analyses; two-sided p<.05 was considered statistically significant. Analyses were conducted using SAS v9.4 and R v.4.3.1.

RESULTS

Of the 800 invited participants, 377 (47%) completed the questionnaire. Among the respondents, the mean age was 48 years (SD=9.5), mean time since cancer diagnosis was 38 years (SD=8.0), 337 (89%) were non-Hispanic White, 293 (78%) had private insurance, and 311 (83%) resided in a metropolitan area (Table 1). Based on the chronic condition data from 2014–2016 used to evaluate eligibility, nearly half of respondents (46%) had ≥2 severe/disabling conditions (Grade 3 or 4), with 34% reporting ≥1 disabling condition (Grade 4). Based on the updated 2019–2021 chronic condition data, the chronic condition burden had increased: 60% had ≥2 severe/disabling conditions (Grade 3 or 4), with 42% reporting ≥1 disabling condition (Grade 4).

Respondents were more commonly female (68% vs. 54%; p<.0001), older (46% vs. 37% ≥50 years of age; p=.03), better educated (74% vs. 54% college graduates; p<.0001), and higher income (50% vs. 37% with incomes ≥$80,000; p=.0002). Based on the 2019–2021 chronic condition data, survey completers were more likely to have been diagnosed with a subsequent malignant neoplasm (29% vs 23%; p=.04), endocrine condition (43% vs 36%; p=.04), or musculoskeletal problem (19% vs 13%; p=.02) but less likely to have hearing loss (13% vs. 22%; p=.0003).

Almost all respondents reported having a main provider (350/377; 93%), the majority of whom were PCPs (286/350; 82%) followed by other (non-cancer) specialists (40/350; 11%) and cancer specialists (23/350; 7%) (Table 3). Only 28 respondents (8%) had a main provider who practiced at a specialized survivorship clinic. Respondents less frequently reported having a coordinating provider (204/377; 54%). For those reporting having a coordinating provider, 57% of the time (116/204) it was the same as the main provider. Most coordinating providers were PCPs (119/204; 58%) followed by other (non-cancer) specialists (48/204; 24%) and cancer specialists (32/204; 16%). As with main providers, participants’ coordinating providers were infrequently practicing at a specialized survivorship clinic (28/204; 14%).

Table 3.

Providers Involved in Care for Adult Survivors of Childhood Cancer

		N	%
Has a main provider1	Yes	350	92.8
	No	27	7.2
Main provider specialty	Primary care provider	286	81.7
	Cancer specialist	23	6.6
	Other (non-cancer) specialist	40	11.4
	Missing	1	0.3
Main provider practices at survivorship clinic	Yes	28	8.0
	No	321	91.7
	Missing	1	0.3
Has a coordinating provider2	Yes	204	54.1
	No	171	45.4
	Missing	2	0.5
Coordinating provider specialty	Primary care provider	119	58.3
	Cancer specialist	32	15.7
	Other (non-cancer) specialist	48	23.5
	Missing	5	2.5
Coordinating provider practices at survivorship clinic	Yes	28	13.7
	No	174	85.3
	Missing	2	1.0
		Mean	SD
Main provider (n=343)	Length of relationship	8.8	9.2
Coordinating provider (n=197)	Length of relationship	8.8	7.4

¹Main provider defined as doctor, physician assistant, nurse practitioner “who takes care of most of your healthcare”

²Coordinating provider defined as doctor, nurse, other provider “who ensures the follow-up of your healthcare”

Of the respondents, 147/373 (39%) reported discontinuity in their health care. The most frequently endorsed reasons for discontinuity on the four items that asked this question included regular provider not being available, too difficult/long to be seen by specialist, provider not knowing personal health situation, no one in charge of care, provider not knowing what other providers had done or said, and not knowing what to expect or next steps (Table 4). The least common reasons for discontinuity were not having a regular provider or clinic, the provider not knowing who was in charge of care, and not having the information needed to cope between appointments. The only socio-demographic characteristic significantly associated with discontinuity was younger age (p=0.02): survivors aged 25–39 years were most likely to report discontinuity (42/80; 53%) vs. survivors aged 40–49 years (45/120; 38%) and survivors aged 50+ years (60/173; 35%) (Figure 1).

Table 4.

Reasons for Discontinuity of Care Selected Among Survivors Who Reported Discontinuity¹

	Were there times when it felt like no one in the health care system was really in charge of your health care? (n=163)		Were there times during or between health care visits when you felt abandoned by the health care system or left too much to your own resources? (n=133)		Did you go to a hospital emergency room for health care? (n=103)		Were there times when your physical or emotional health suffered because your health care was poorly organized? (n=95)
Reason	N	%	N	%	N	%	N	%
Don’t have a regular provider or clinic	29	17.8	19	14.3	7	6.8	13	13.7
Regular provider not available	26	16.0	22	16.5	46	44.7	20	21.1
Too difficult/long to be seen at regular clinic	34	20.9	24	18.0	16	15.5	19	20.0
Too difficult/too long to see specialist or another person I had been referred to	46	28.2	47	35.3	13	12.6	31	32.6
Person I saw didn’t really know my personal health situation	56	34.4	46	34.6	8	7.8	33	34.7
No one seemed to be in charge of my health care	65	39.9	54	40.6	5	4.9	32	33.7
Provider didn’t seem to know who was in charge of my health care	17	10.4	19	14.3	1	1.0	11	11.6
Provider didn’t know what others had done or told me	58	35.6	43	32.3	1	1.0	26	27.4
Didn’t know what to expect about my health condition or next steps in my care	35	21.5	36	27.1	12	11.7	29	30.5
When things went wrong or changed, I could not get answers or advice quickly	25	15.3	24	18.0	15	14.6	26	27.4
Providers gave me different information	32	19.6	24	18.0	5	4.9	19	20.0
Didn’t have the information needed to cope with my health between appointments	10	6.1	13	9.8	5	4.9	17	17.9

Shading indicates ≥20.0% reporting

¹For 4 of the 6 discontinuity items, respondents were provided a list of reasons for why the discontinuity occurred and asked to check all that apply

Figure 1. Percentage of Participants Reporting Discontinuity by Socio-Demographic Characteristics.

Horizontal bars indicate the percentage of participants reporting discontinuity overall and by socio-demographic characteristics. P-values are based on chi-square comparisons among those with known values.

In age-adjusted models, having a cancer specialist (vs. a PCP) as a main provider was associated with lower prevalence of discontinuity (PR=0.81; 95% CI: 0.66–0.99), although the provider’s association with a specialized survivorship clinic was not associated with discontinuity prevalence (Figure 2). A shorter relationship with the main provider (0–3 years vs ≥10 years) was associated with higher discontinuity prevalence (PR=1.17; 95% CI: 1.02–1.34). There were no statistically significant associations between the specialty and length of relationship with the coordinating provider and discontinuity prevalence. We compared the results of the inverse probability weighted models to models with no weighting. There was only one difference in the statistical significance of the results. In the unweighted models, having a non-cancer-specialist coordinating provider was associated with a higher discontinuity prevalence (1.23; 1.05–1.44; p=0.01), but the difference was no longer statistically significant in the weighted model (1.16; 0.99–1.37; p=0.07).

Figure 2. Prevalence Ratios for Risk of Discontinuity Associated with the Specialty of and Length of Relationship with the Main and Coordinating Providers.

Prevalence ratio (circle) with 95% confidence intervals (whiskers) based on age-adjusted Poisson models of discontinuity with inverse probability weighting. Referent for specialty of provider is Primary care; referent for seen at specialized cancer survivorship clinic is not seen at one; referent for Years seeing provider is 10 or more

Average subscale and summary scores on the CC-MC were generally high (indicating better care coordination), with the main provider summary score average (SD) of 83.2 (14.41), care across multiple providers of 71.7 (23.49), and survivor-provider partnership of 72.2 (26.17) (Figure 3). Among subscales, the lowest average scores were for care plan (55.6 [37.91]), information gap between clinicians (61.4 [34.01]), and confidence in team (62.5 [23.49]). Better scores on subscale and summary measures of coordination at all 3 levels were associated with lower discontinuity prevalence (PR; 95% CI): care from main provider (0.73; 0.64–0.83), care across multiple providers (0.81; 0.78–0.83), patient-provider partnership (0.85; 0.80–0.89) (all p<0.0001) (Figure 4). There were no differences in the statistical significance of the results in unweighted models. Patients with better scores on each of the care coordination subscale and summary measures were less likely to experience discontinuity.

Figure 3. Subscale and Summary Scores on the “Patient-Perceived Continuity of Care from Multiple Clinicians” (CC-MC)¹.

Figure displays mean score with whiskers at +/− 1 standard deviation.

¹ All were rescaled to 0–100 by dividing by max possible value and multiplying by 100. For all scales a higher score indicates better continuity (items for Information Gap and Role Clarity initially reflected higher value=lower continuity, so they were reverse scored before summarizing).

² Only among those reporting ‘Yes’ to ‘Thinking about ALL the persons you saw in ALL different places you went for your care in the last 6 months, is there one who ensures the follow-up of your healthcare (doctor, nurse, other)?

³ Only among those reporting ‘Yes’ to ‘Do you have a provider (doctor, physician assistant, nurse practitioner) who takes care of most of your healthcare’

⁴ Excludes those who reported ‘I don’t have a regular clinic’

⁵ Excludes those with at least two items indicating information not needed

Figure 4. Prevalence Ratios for Risk of Discontinuity Associated with the Three Levels of Care Coordination.

Prevalence ratio (circle) with 95% confidence intervals (whiskers) for discontinuity associated with 10% improvement in care coordination. Based on age adjusted Poisson models with inverse probability weighting.

¹ Only among those reporting ‘Yes’ to ‘Thinking about ALL the persons you saw in ALL different places you went for your care in the last 6 months, is there one who ensures the follow-up of your healthcare (doctor, nurse, other)?

² Only among those reporting ‘Yes’ to ‘Do you have a provider (doctor, physician assistant, nurse practitioner) who takes care of most of your healthcare’

³ Excludes those who reported ‘I don’t have a regular clinic’

⁴ Excludes those with at least two items indicating information not needed

We conducted post hoc analyses comparing discontinuity prevalence and CC-MC scores between the 116 respondents who had the same main and coordinating provider and 84 respondents who had both a main and coordinating provider – but not the same provider. Respondents with different main and coordinating providers were more likely to experience discontinuity (33.7% vs. 20.0%; age-adjusted PR=1.68, 1.05–2.69; p=0.03) and significantly worse scores on the main provider summary measure and subscales (all age-adjusted p<.01). There were no differences on the other summary measures and subscales.

DISCUSSION

Adult survivors of childhood cancer are a unique population given that their cancer and its treatment frequently lead to a significant burden of chronic conditions. Previous studies have shown that survivors experience multimorbidity at five times the rate of siblings, representing a morbidity burden comparable to siblings twice their age.⁸ However, there is a paucity of research investigating the continuity and coordination of survivors’ complex care needs. This study provides important initial findings regarding the prevalence of discontinuity experienced by adult survivors of childhood cancer, the reasons for this discontinuity, and the risk factors for experiencing discontinuity.

We found that approximately 4 in 10 participants reported experiencing discontinuity, thereby identifying significant gaps in care for a substantial minority of childhood cancer survivors. Reasons endorsed as causing the discontinuity included coordination issues (e.g., not knowing what other providers had done, not knowing the patient’s health situation). However, not having a regular provider or clinic was not reported as an important cause of discontinuity. In fact, almost all respondents (93%) reported having a main provider and over half (54%) reported having a coordinating provider. In comparison, among the CC-MC survey validation population, which included patients recruited from primary care clinics receiving care for the same health condition at more than one place, the rate of having a main provider was very similar (92%), but the rate of having a coordinating provider was much higher (76%).¹⁷ In an age-stratified sample of the CCSS cohort (without a specific focus on those with chronic conditions), 81% had visited a PCP in previous year,¹² but the specific concepts of “main” and “coordinating” providers are unique to the CC-MC survey and not available for comparison.

Notably, only 43 of the 377 respondents (11%) had a main or coordinating provider who practiced in a specialized survivorship clinic (15 had a survivorship clinic main provider only, 15 had a survivorship clinic coordinating provider only, and 13 had survivorship clinic main and coordinating providers, of whom 9 were the same individual). This finding highlights the fact that specialized survivorship clinics are not delivering care to most survivors and underscores the need to support educational programs and resources for providers in the community.

The only socio-demographic characteristic associated with discontinuity prevalence was younger age. Overall, respondents aged 25–39 years reported the greatest discontinuity prevalence at 53%. Previous research has shown that adolescent and young adult cancer survivors are more likely to be uninsured.²² Our study’s results also demonstrate that the physician specialty and length of the relationship are associated with discontinuity prevalence. Having a newer relationship with the main provider was associated with a greater discontinuity prevalence, while having a cancer specialist as the main provider was associated with a lower discontinuity prevalence. These findings emphasize the importance of rapport and familiarity between survivors and their providers. Better care, as demonstrated by better scores on the CC-MC subscale and summary measures of continuity, was associated with lower discontinuity prevalence. These results provide valuable information for intervention development to improve care. That is, interventions that lead to improved coordination of care at the main provider level, across multiple providers, and between patients and providers would be expected to lead to decreased discontinuity prevalence.

Given that the worst scores on the CC-MC continuity subscales addressed care plans and information gaps between clinicians, these areas might be prioritized for intervention. As a reminder, the care plan subscale evaluates whether providers explain health conditions and their treatments, tests and follow-up, and consider patient goals. This care plan domain is not specific to the survivorship care plans recommended by the Institute of Medicine,²³ but future research should investigate whether patients who report having a survivorship care plan had higher scores on this subscale – adding further support to their value for survivors of childhood cancer. The subscale addressing information gaps between clinicians assesses the frequency the provider, for example, did not know the patient’s medical history, had access to test or exam results, and relied on the patient to convey medical information. Future research should investigate whether the availability of and linkage among electronic health records could address these information gaps.

The findings from this study should be interpreted in the context of its design. The chronic condition information used to determine eligibility for this study was collected in 2014–2016. We were able to provide updated chronic condition information based on 2019–2021 data, and as expected, the sample’s chronic condition burden had increased. While our response rate was slightly less than 50%, we achieved this response rate with less than two months for completion, limited reminders, and no incentive for completion. We were limited to two months to field the survey so that it would not conflict with the 8^th Follow-Up Survey of the entire CCSS cohort. Due to limited resources, we were not able to offer incentives. Notably, our target sample size was 400, and we came very close to reaching that goal. We used inverse probability weighted models to adjust for survey non-participation, but the weighted and unweighted model results were largely similar. The only difference in statistical significance was that having a non-cancer-specialist coordinating provider was associated with greater discontinuity in the unweighted model but not in the weighted model. However, we could only weight the analyses based on observed characteristics. Respondents and non-respondents might vary on unobserved characteristics, which would limit the generalizability of our findings. Finally, the survey was only available for online completion and may not be generalizable to childhood cancer survivors who are unable to complete online surveys.

Compared to non-responders, responders were more likely to be female, older, more educated, and with higher incomes. Because discontinuity prevalence was higher among younger survivors, these results may underestimate the true prevalence of discontinuity. While there were not statistically significant differences in the racial distribution of responders and non-responders, the overall population was heavily non-Hispanic White (89%), and additional research with greater minority representation is needed. In addition, the composite measure of discontinuity was developed for this study, and future research should explore its association with important processes of care (e.g., receipt of recommended screening and surveillance) and outcomes (e.g., health status, mortality). This study provides only cross-sectional information on continuity and coordination of care. Future research should explore longitudinal relationships between survivors’ reported continuity and coordination of care with patterns of health service use (e.g., receipt of appropriate screening for long-term effects) and outcomes (e.g., health-related quality of life and survival). Notably, continuity and coordination of care are survivor-centered measures of care quality that are agnostic to the specific disease.

CONCLUSION

These findings demonstrate that a substantial minority of childhood cancer survivors experience discontinuity and provide insights on the domains most in need of improvement. These findings and additional research should be used to develop interventions to improve continuity and coordination for the unique needs of this vulnerable population.

Data Availability:

The Childhood Cancer Survivor Study (CCSS) is a US National Cancer Institute funded resource (U24 CA55727) to promote and facilitate research among long-term survivors of cancer diagnosed during childhood and adolescence. CCSS data are publicly available on dbGaP at https://www.ncbi.nlm.nih.gov/gap/ through its accession number phs001327.v2.p1 and on the St Jude Survivorship Portal within the St. Jude Cloud at https://survivorship.stjude.cloud/. In addition, utilization of the CCSS data that leverages the expertise of CCSS Statistical and Survivorship research and resources will be considered on a case-by case basis. For this utilization, a research Application of Intent followed by an Analysis Concept Proposal must be submitted for evaluation by the CCSS Publications Committee. Users interested in utilizing this resource are encouraged to visit http://ccss.stjude.org. Full analytical data sets associated with CCSS publications since January of 2023 are also available on the St. Jude Survivorship Portal at https://viz.stjude.cloud/community/cancer-survivorship-community~4/publications.