Survivorship Care Plan: Correlation between Practitioners and a Digital Tool (LOG-After)

Abstract

Objective

To compare survivorship care plans (SCPs) generated by the digital tool LOG-after with those proposed by oncologists, pediatric oncologists, and radiation oncologists for childhood, adolescent, and young adult cancer survivors. LOG-after was developed through a participatory approach within the Grand Ouest Cancer de l'Enfant network and subsequently disseminated to other French comprehensive cancer centers to support the implementation of individualized, evidence-based survivorship care. This study was conducted as part of the Institut National du Cancer-funded Toxicity and Health after Irradiation on Neurocognition in Kids project.

Study design

A questionnaire comprising 3 clinical cases (acute lymphoblastic leukemia, Hodgkin's lymphoma, and nephroblastoma) was circulated among 3 national networks (Société Française des Cancers de l'Enfant, Adolescents and Young Adults Group, and Pediatric Radiotherapy). Responses were compared with SCPs generated by LOG-after, and attitudes were discussed in workshops.

Results

Thirty-six practitioners participated, including 5 radiation oncologists. Thirteen respondents considered themselves experts in survivorship care. SCP varied significantly between practitioners, with no complete agreement. However, there was greater alignment between the software and practitioners for high-grade recommendations, particularly among experts (eg, 100% of experts recommended postanthracycline cardiac monitoring).

Conclusions

Given the observed heterogeneity, the LOG-after tool, which integrates the latest recommendations, is expected to improve SCP by enhancing consistency and reducing unnecessary screenings. During the workshops, participants considered the algorithm to be generally aligned with published guidelines, confirming its potential to facilitate their broader dissemination. This digital tool can help standardize evidence-based survivorship care, potentially increase consistency in monitoring, and potentially lead to better long-term health outcomes for cancer survivors.

Thanks to advancements in therapeutic approaches, the survival rates of survivors treated for cancer before the age of 25 have been steadily increasing. Currently, the 5-year survival rate for this population, across all cancer types, is between 80% and 85%.^1,2 However, childhood cancer and its treatments are associated with significant long-term effects, with 60%-65% of survivors experiencing medical and/or psychosocial sequelae within 20 years of treatment.^2,3 Moreover, the cumulative incidence of serious adverse events reaches 40% by 30 years postdiagnosis.⁴ Consequently, the need for long-term follow-up (LTFU) for these survivors is undeniable. This was emphasized in both the 2006 Erice Declaration⁵ and the 2006 report by the Institute of Medicine.⁶ These documents highlight the importance of providing survivors with a written summary at the end of their active treatment, including details on diagnosis, tumor characteristics, treatment timeline, and dosages and a survivorship care plan (SCP) with information on potential side effects, recommended follow-up assessments, referral, and psychosocial prevention measures.

Despite the recognized importance of LTFU, particularly for individuals treated before the age of 25 years,7, 8, 9, 10 few institutions have effectively implemented LTFU programs or provided the recommended end-of-treatment summaries. A study by the Children's Oncology Group showed that 68% of institutions routinely provided a treatment summary and SCP to survivors,⁸ yet only one-third of survivors currently have access to comprehensive LTFU services.^11,12

To help practitioners in creating these documents, the inter-regional Grand Ouest Cancer de l'Enfant group developed a software called LOG-after. This tool is based on a prospective cohort of all patients aged less than 25 years treated since 2012 in the 11 centers of the inter-region. Data were collected in 5 modules, including the patient's clinical history and treatment summaries. Since 2021, patients can be included retrospectively during follow-up consultations.¹³ The algorithm takes into account the diagnosis, treatments received, and associated comorbidities (family history, constitutional frailty, etc.) and integrates the latest international recommendations for childhood cancer survivors, including those from the International Guideline Harmonization Group,^14,15 pediatric normal tissue effects in the clinic group,¹⁶ PanCare,¹⁴ and the Childhood Cancer Predisposition Workshop Series,¹⁷ alongside general population surveillance recommendations (eg, gynecologic examinations, dental check-ups, and vaccinations). LOG-after has since been implemented in other French centers to facilitate the long-term care of survivors. This tool differs in a number of ways from those developed previously (notably Passport for Care¹⁸ and Survivorship Passport¹⁹), despite the shared aim of standardizing post-treatment follow-up. LOG-after is based on international follow-up guidelines and is continually updated through collaborative work with survivors. It also creates a genuine patient-health-care-provider interface, with the ability to share informational content in addition to the treatment summary and personalized postcancer plan.

Despite its widespread use, no study has yet evaluated the consistency between the SCPs generated by the LOG-after algorithm and those created by practitioners. This study aims to assess the agreement between personalized SCP generated by the algorithm and practitioners' recommendations, understand the challenges in interpreting care plans, and suggest improvements to the LOG-after software.

Methods

Three theoretical clinical cases were developed by 2 LTFU specialists involved in the Toxicity and Health after Irradiation on Neurocognition in Kids project to cover different therapeutic situations (Supplemental Table 1; available at www.jpeds.com). These clinical cases were submitted to the LOG-after software, using its algorithm to automatically generate SCPs. The recommendations were classified according to their level of evidence.¹⁵

The clinical cases were then emailed to practitioners with a link to a totally anonymous online questionnaire. It included 2 initial questions concerning the clinician's involvement in LTFU (“As part of your professional practice, do you receive survivors for LTFU? [Yes/No]” and “How do you perceive your expertise in the field of LTFU? [Expert/Intermediate/Low]”). Specialists were then asked to review the 3 clinical cases and provide their recommended monitoring types and frequencies in blank fields. The questionnaire was first distributed in February 2024, with a reminder sent 15 days later.

After analyzing the responses, 2 videoconference meetings were organized to present and discuss the results. Invitations to these meetings were sent via the same email list used for the questionnaire, with additional solicitations extended to former patients through the national association “Les Aguerris.”

Invitation of Participating Practitioners

The questionnaire was distributed via 3 mailing lists: the French Society of Pediatric Oncology (Société Française des Cancers de l'Enfant), the Groupe Onco-hématologie Adolescents et Jeunes Adultes, and the French pediatric radiotherapy group (French Pediatric Radiotherapy Group). Since practitioners may have been included in multiple mailing lists, the exact number of invited participants cannot be precisely determined. All clinicians were encouraged to participate regardless of their perceived LTFU knowledge.

Statistical Methodology

Free-text responses from the questionnaires were transformed into qualitative variables (“type of examinations” and “frequency of performance”). The degree of agreement between the 2 SCPs was classified as total agreement if no differences were observed, intermediate agreement if there were no more than 2 differences, and disagreement if there were 3 or more differences. Examinations not included in the algorithm were considered.

Qualitative variables were expressed as counts of occurrences and percentages. The correlation between SCPs generated by the algorithm and those created by practitioners, as well as concordance between SCPs from different clinicians, was described in terms of the number of occurrences and percentage for each level of agreement. Statistical analyses were performed using R software (version 4.1.0; R Core Team, R Foundation for Statistical Computing, Vienna, Austria) (2021-05-18).

Ethics

In accordance with French regulations, ethical approval was not required for this study. Since data collection was completely anonymous, no declaration was necessary to the Commission Nationale Informatique et Libertés.

Results

Characteristics of Respondents

Thirty-nine practitioners completed the questionnaire after the reminder. Of these submissions, 4 could not be analyzed for all cases because of missing data or responses such as “refer to a doctor for LTFU” and 3 could not be evaluated for clinical case 1 because of incomplete information.

Among the 35 analyzable participants, 11 (31.4%) identified themselves as experts in managing patients undergoing LTFU (all of whom reported seeing patients for LTFU). Eighteen respondents (51.4%) described their expertise as intermediate, with 13 (72.2%) of them also managing patients for LTFU. Six respondents (17.1%) rated their knowledge as low, including 1 (16.7%) who reported seeing patients for LTFU.

Agreement between Survivorship Care Plans Generated by the Algorithm and Those Generated by Practitioners

The SCPs automatically produced by the LOG-after algorithm are summarized in Table I. None of the clinicians achieved total or intermediate agreement with the SCP generated by the algorithm for any of the clinical case (Figure 1 and Supplemental Table 2; available at www.jpeds.com).

Table I.

SCPs generated by the LOG-after algorithm

	Level of evidence
	Grade A	Grade B	Grade C	For general population
	Type of examination
Clinical case 1	Blood pressure Echocardiography Serum ferritin	Eye test Liver function tests	Bone mineral density test Metabolic assessment	Dental examination
Clinical case 2	Blood pressure Breast MRI Dermatologic examination Echocardiography Meningococcal vaccine Respiratory functional examination Stool blood testing Thyroid function Thyroid ultrasound	Metabolic assessment	None	Dental examination Gynecologic consultation
Clinical case 3	Blood pressure Dermatologic examination Echocardiography FSH Meningococcal vaccine Serum creatinine Stool blood testing	Liver function tests Renal function tests	Audiogram Metabolic assessment	Dental examination Gynecologic consultation

FSH, follicle-stimulating hormone; MRI, magnetic resonance imaging.

Type of the examinations and their level of evidence automatically produced by the software.

Figure 1.

Agreement between practitioners and LOG-after algorithm by expertise level and recommendation category (proportion of clinicians in each agreement category [intermediate or total disagreement] across 3 clinical cases and 4 recommendation subsets. Total agreement = 0% in all conditions). CC = clinical case. Categories: All recs = all recommendations (type + frequency); High-ev. recs = recommendations with high level of evidence only (type + frequency); Type only = type of examination only regardless of frequency; High-ev. type = type of examination with high level of evidence only regardless of frequency.

Focusing solely on the type of tests prescribed, regardless of their frequency, reveals that many SCPs provided by practitioners are incomplete. For instance, in clinical case 1, less than one-third of clinicians prescribed 7 of 8 recommended tests; in clinical case 2, 7 of 12; and in clinical case 3, 11 of 13 (Table I). For example, appropriate examinations and correct frequency of screening tests for radiation-induced neoplasms were recommended by 9 of 35 practitioners (25.7%) for breast monitoring and 5 of 35 clinicians (14.8%) for thyroid monitoring in clinical case 2. In clinical case 3, only 1 of 35 specialists (2.8%) prescribed stool blood testing and 3 of 35 practitioners (14.3%) included dermatologic monitoring (Figure 2, A and Supplemental Table 3; available at www.jpeds.com).

Figure 2.

Evaluation of examination recommendations by practitioners. A, Proportion of clinicians who recommended the correct examination for each clinical case, as outlined by LOG-after, based on their self-reported knowledge. B, Proportion of appropriate examinations and accurate test frequencies for each clinical case. “Prescriber” refers to the percentage of clinicians who recommended the correct examination, while “correct prescription” indicates the percentage of those who prescribed both the appropriate examination and its correct frequency.

Assessing the types of examinations prescribed according to high-level evidence recommendations (Grade A), and considering both the suggested frequency and the practitioner's self-reported levels of knowledge (Figure 2, A), cardiac ultrasound is the most commonly prescribed examination across all expertise levels. However, the prescription of other examinations varies considerably based on the clinician's self-reported competence, with higher adherence to recommendations among expert specialists (Figure 2, B).

Correlation between Survivorship Care Plans Prescribed by Practitioners

There was no consensus among clinicians regarding the SCPs they proposed, including among those who identified themselves as experts in LTFU.

Examinations Prescribed by Practitioners Not Found in the Algorithm

On average, clinicians prescribed 1.4 (SD, 1.2) investigations for clinical case 1, 1.3 (SD, 1.5) for clinical case 2, and 0.8 (SD, 0.9) for clinical case 3 that were not advised by the prescribing algorithm. The most frequently prescribed procedures not included in the algorithm were (1) fertility tests and blood cell counts (17/32 responders, 53.1%, and 10/32 responders, 31.2%, respectively) for clinical case 1, (2) fertility tests (17/35 responders, 48.6%) for clinical case 2, and (3) radiological investigations other than magnetic resonance imaging for breast monitoring (6/35 responders, 17.1%) for clinical case 3 (Table II).

Table II.

Examinations prescribed by practitioners and not included in the SCPs proposed by the algorithm

	Prescriptions	n	%
Case 1	Fertility assessment	17	53.1
	Cell blood count	10	31.2
	Renal assessment	7	21.9
	Neurologic assessment	3	9.4
	Cardiac MRI	1	3.1
Case 2	Fertility assessment	17	48.6
	Breast ultrasound	7	20.0
	Cell blood count	6	17.1
	Mammography	4	11.4
	Renal assessment	3	8.6
	Osteodensitometry	3	8.6
	Exercise test	2	5.7
	Gastroduodenal fibroscopy	1	2.8
Case 3	Breast ultrasound/mammography	6	17.1
	Pelvic ultrasound	5	14.3
	Cell blood count	5	14.3
	Fertility assessment	3	8.6
	Renal ultrasound	2	5.7
	Colonoscopy	1	2.8

MRI, magnetic resonance imaging.

Workshop Discussion

Two workshops were held in May 2024, bringing together clinicians and former survivors to discuss the results of the study. In the first workshop, 9 practitioners (2 radiation oncologists, 3 pediatric oncologists, 1 hematologist, 1 endocrinologist, 1 adult oncologist, and 1 clinical research manager) participated. In the second one, 1 survivor and 10 professionals (1 radiation oncologists, 5 pediatric oncologists, 1 hematologist, 1 general practitioner, 1 psychologist, and 1 sociologist) were connected. One pediatric oncologist and the radiation oncologists, who led the study, attended both sessions. Several key points emerged:

• 1.Accuracy of the algorithm's recommendations: There was general agreement that the algorithm's suggestions were accurate, without errors, and aligned with established guidelines.
• 2.Frequency of fertility tests: Clinicians observed frequent queries about fertility test during LTFU consultations, even for low-risk patients. It was decided to explicitly notify the risk of treatment-related fertility issues for all patients (low, intermediate, and high) in the LOG-after, enabling practitioner to provide appropriate guidance on the necessity of these tests.
• 3.Managing follow-up frequency: Both clinicians and survivors emphasized the challenges of implementing excessive follow-up as outlined in current guidelines. It was proposed to harmonize follow-up frequencies when possible and to group multiple examinations to simplify the care pathway and improve adherence.
• 4.Incorporating general population recommendations with treatment-related follow-up: Practitioners reported difficulties in integrating general population guidelines with treatment-related follow-up. While LOG-after provides helpful reminders for various recommendations, the lack of a clear distinction between general population and treatment-related follow-ups led to some confusion. As a result, it was decided to prioritize treatment-related monitoring in the next software update to ensure clarity and simplify the management of both types of follow-up for clinicians.

Discussion

Our results indicate that although one-third of practitioners considered themselves to have expert knowledge, significant omissions were observed in the creation of SCPs. Indeed, no clinician fully aligned with the algorithmic recommendations or with other practitioners' plans, which raises concerns about the feasibility of realizing a “perfect” follow-up modeled strictly on current guidelines. The International Guideline Harmonization Group has made notable progress in standardizing follow-up care protocols for childhood cancer survivors across Europe. However, despite these efforts, a considerable gap remains in the dissemination and implementation of these procedures among health care practitioners. This limited uptake contributes to persistent inconsistencies in LTFU management, highlighting the need for enhanced education and better integration of these guidelines into routine clinical practice. Several factors may explain these challenges: (1) cancers diagnosed before the age of 25 are relatively rare, resulting in limited experience and expertise among clinicians7, 8, 9, 10; (2) the delayed onset of long-term sequelae creates a transition period between pediatric and adult medicine, leaving some practitioners with limited insight into long-term outcomes as they only follow patients for a specific period²⁰; and (3) the complexity of the pathologies, the treatments administered, and the recommendations themselves lead to a high volume of information for specialists to manage, often resulting in prioritization of their knowledge from the therapeutic phase over follow-up guidelines.⁷

Ongoing training for physicians on LTFU is crucial to addressing the information needs of cancer survivors. Numerous studies have shown a direct link between the level of information provided to survivors and their quality of life.^21,22 Survivors consistently ask for information about potential long-term side effects,22, 23, 24, 25, 26, 27, 28, 29 often requesting this information multiple times^25,30 and in formats that are clear and accessible.^22,31 Furthermore, better communication of LTFU information has been associated with higher adherence to care plans,^22,24,32,33 which is essential for effective LTFU and, ultimately, for reducing morbidity and mortality among former patients.^33,34

It is, however, important to acknowledge a methodological limitation of this study: the comparison between the SCP generated by physicians and that generated by the algorithm simply measures agreement. Neither of the 2 approaches has been independently validated; therefore, neither can be considered the gold standard. The discrepancy between practitioners and the algorithm cannot, therefore, be interpreted as the algorithm being superior or as physicians failing to implement the recommendations adequately or having insufficient knowledge. The observed higher concordance with the algorithm grade A recommendations, such as cardiac monitoring, supports this interpretation. Indeed, real-world clinical practice may reflect the physician's clinical judgment. For example, physicians' more frequent prescribing of fertility assessments responds to a patient need that goes beyond the recommendations. Similarly, the algorithm itself may be flawed, particularly when the guidelines are not based on a high level of evidence, and their application may vary from case to case. The working groups enabled the algorithm to be compared with real-world clinical practice. These considerations highlight the importance of physicians adopting digital tools, so as to guide their management while maintaining their clinical judgment.

One of the major advantages of the LOG-after tool is that it provides a digital interface to facilitate interactions between survivors and their physicians.³⁵ In the postcancer phase, empowering patients to take an active role in their follow-up care is essential, as it allows them to take control of their health management and organize their care closer to home.³⁶ This tool effectively addresses the mobility needs of this population,³⁷ while maintaining a connection with the LTFU care team. Additionally, LOG-after also provides access for various health care providers, including general practitioners and organ specialists involved in patients' care.¹³

However, during the 2 workshops where we presented the study's results, the survivors expressed concerns about the complexity of the monitoring proposed by the algorithm. On average, LOG-after suggested an average of 12.3 follow-up tasks per patient (including general population recommendations), with frequencies ranging from one-time follow-up at the start of LTFU to annual, biennial, or 5-year interval. It has been noticed that some biological tests were spread across multiple lines (eg, different types of endocrine follow-up or metabolic monitoring), and other tests were contingent on the presence of symptoms (eg, pulmonary, auditory, or ophthalmologic follow-ups). Such detailed follow-up protocols pose practical challenges, both for physicians to prescribe and for survivors to adhere to. Moreover, prior research has demonstrated that the accumulation of excessive information can have counterproductive effects.³⁸ As a result, survivors suggested the possibility of grouping follow-up tests, even if this means that some examinations might not align perfectly with current recommendations in terms of timing.

Some adjustments have been proposed for the adaptation of the LOG-after tool:

• 1.Given the importance of postcancer fertility for survivors39, 40, 41 and the frequent demand for information on this subject,⁴² a reminder of the expected fertility risk (low, intermediate, and high) has been added to LOG-after. This feature allows practitioners to provide appropriate information and avoid unnecessary fertility evaluations, which can cause undue anxiety for patients at a low risk of infertility.
• 2.Certain follow-ups, such as dental and gynecologic check-ups, were previously included in the tool as routine recommendations for the general population. To simplify the personalized SCPs, these general population check-ups will now be clearly distinguished from postcancer-specific follow-ups in the updated version of LOG-after.
• 3.The updated version organizes the personalized follow-up plan into 4 distinct sections: (1) information: this section includes links to relevant resources, such as websites (eg, Société Française des Cancers de l'Enfant, Les Aguerris) and explanatory videos, and some links are universal, while others are tailored to the patient's specific needs; (2) follow-up examinations: this section covers postcancer surveillance tailored to the patient's treatment history; (3) general population check-ups: this section includes routine health screenings (eg, dental and gynecologic) recommended for all individuals; and (4) other: this section provides additional guidance, such as tests recommended only if symptoms are present or optional follow-ups for patients participating in clinical studies.

One of the limitations of this study concerns the definition of the degree of agreement. Indeed, the degree of agreement was defined pragmatically in the absence of a gold standard for this type of comparison. Although this approach has the advantage of being simple and reproducible, it is true that it remains arbitrary and must be interpreted with caution. Furthermore, this degree of agreement was not weighted according to the grade of the recommendations, although this grade has an impact in clinical practice. Indeed, quite intuitively, it is more reasonable to expect strong agreement on grade A recommendations. The analysis presented in Figure 1 highlights greater agreement for grade A recommendations, thereby allowing for a more clinical perspective on the degree of agreement defined previously. This study has some more limitations, including a low number of respondents, workshops with few participants, and the establishment of concordance on only 3 clinical cases, which represent highly diverse post-cancer scenarios. The low response rate and the overlap between mailing lists have likely led to selection bias. Indeed, it is possible that the practitioners who responded to the questionnaire were already the most aware of the issues surrounding LTFU, which may have overestimated their level of knowledge. It is, therefore, possible that the degree of heterogeneity is even greater in real life.

Despite these limitations, our findings highlight both the strengths and weaknesses of the LOG-after tool. The strengths include the completeness of the recommendations, standardization of follow-up care between clinicians, and avoidance of unnecessary and/or unvalidated follow-ups. However, the main weaknesses noted were the complexity and cumbersome nature of the recommended follow-up schedule. Improvements were suggested during the 2 feedback sessions, particularly by the survivors present. The tool can serve as a practical aid in prescribing follow-up care and as a bridge between the patient, the LTFU doctor, and other involved health care providers.

Conclusion

This study demonstrates that, despite the availability of harmonized international recommendations and dedicated platforms like PanCare, a significant gap persists between clinicians' prescriptions and current guidelines. This discrepancy mainly concerns the type of surveillance and the frequency of examinations, with some unnecessary tests being prescribed while other necessary tests are overlooked. Notably, no clinician achieved full or intermediate agreement with the algorithm, even among those who considered themselves experts, reflecting an issue in the dissemination and implementation of these recommendations. Contributing factors may include the rarity of cancers in individuals aged less than 25 years, the transition from pediatric to adult care, and the complexity of treatment-related sequelae.

During the workshops, participants considered the algorithm to be generally aligned with published guidelines, indicating that such digital tools could be valuable in promoting guideline dissemination. However, survivors emphasized the need for a simplified SCP that is not strictly reliant on the guidelines to optimize adherence, particularly when certain examination frequencies are supported by limited scientific evidence. Thus, these findings suggest that while digital tools offer promise in standardizing care, further refinements are necessary to improve their practical application and ensure that SCPs are both comprehensive and manageable in day-to-day clinical practice.

Ethics Approval and Consent to Participate Declarations

In accordance with French regulations, ethical approval was not required for this study. Since data collection was completely anonymous, no declaration was necessary to the Commission Nationale Informatique et Libertés (CNIL).

Data Statement

Data sharing statement available at www.jpeds.com.

CRediT authorship contribution statement

Amandine Bertrand: Writing – original draft, Formal analysis, Data curation. Isabelle Pellier: Writing – review & editing, Funding acquisition, Conceptualization. Marie-Dominique Tabone: Writing – review & editing, Data curation. Véronique Christophe: Writing – review & editing. Isabelle Ray-Coquard: Writing – review & editing. Christine Rousset-Jablonski: Writing – review & editing. Les Aguerris: Writing – review & editing. Pascal Veillon: Writing – review & editing, Data curation, Conceptualization. Charlotte Demoor-Goldschmidt: Writing – review & editing, Formal analysis, Conceptualization.

Declaration of Competing Interest

The authors have no relevant financial or nonfinancial interests to disclose.

This study has been funded as part of the THINK project (Toxicity and Health after Irradiation on Neurocognition in Kids) by the French National Cancer Institute (INCa) in 2023 (convention n 2023-014) and by GoAJA association. Dr A. Bertrand's doctorate was partially funded by the ARC Foundation for Cancer Research(n ARCMD-DOC22021020003083).