Electronic Patient Reported Outcome Measures and quality of life in cancer (E-PROMISE): systematic review of the evidence and meta-analysis

Abstract

Objective

The objective of this systematic review and meta-analysis (SRMA) was to evaluate the impact of electronic patient-reported outcomes (ePROs) on health-related quality of life (HRQoL) in patients with cancer.

Design

We performed SRMA of randomised controlled trials (RCTs) comparing ePRO interventions with usual care in patients with cancer. The primary outcome was HRQoL. We used a random effects model a priori due to the anticipated clinical heterogeneity. Subgroup analyses and meta-regressions were performed to explore sources of heterogeneity. After assessing the risk of bias using risk-of-bias tool (RoB V.2), we rated the evidence certainty using the Grading of Recommendations, Assessment, Development and Evaluations framework.

Eligibility criteria

We included studies meeting the following criteria: (1) RCTs; (2) patients diagnosed with any type of cancer, undergoing or having completed treatment; (3) comparing ePROs with usual care without ePRO interventions; (4) assessing the effect on HRQoL.

Information sources

We systematically searched PubMed, Embase, and the Cochrane Central Register of Controlled Trials up to April 2024.

Results

We screened 7706 records to include 36 RCTs with 9608 patients. ePRO interventions showed a standardised mean difference (SMD) of 0.35; 95% CI 0.18 to 0.51 compared with usual care. Patients receiving ongoing therapy had an SMD of 0.39 (95% CI 0.21 to 0.58), while those who had completed therapy had an SMD of 0.12 (95% CI 0.01 to 0.22), with a significant subgroup difference (p=0.01). No statistically significant differences were observed across the method of ePRO assessment, cancer site, metastasis status, therapy status, average age or duration of ePRO use. The results remained consistent with Bayesian and other sensitivity analyses.

Conclusions

ePRO interventions improve HRQoL more than usual care in patients with cancer, with greater effect in those currently undergoing therapy. This improvement is independent of cancer type, duration of ePRO use or patient age. Future research should address sources of heterogeneity, explore long-term impacts and develop strategies to increase patient engagement and adherence to ePRO systems.

PROSPERO registration number

CRD42024531708.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Electronic patient-reported outcomes (ePROs) are increasingly used in cancer care to monitor health-related quality of life (HRQoL); however, previous studies have shown mixed results, and there is limited comprehensive evidence evaluating the overall impact of ePRO on HRQoL.

WHAT THIS STUDY ADDS

• This meta-analysis synthesises data from 36 randomised controlled trials and demonstrates that ePRO interventions significantly improve HRQoL in patients with cancer compared with usual care, providing robust evidence of their effectiveness across various cancer types and treatment stages.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• These findings support the integration of ePROs into routine cancer care as a strategy to enhance patient outcomes.

• They offer valuable insights for clinical practice and policy development, promoting patient-centred care.

Introduction

Electronic healthcare interventions (EHIs) represent the intersection of digital technologies and healthcare, encompassing a wide range of tools, including mobile health apps, wearable devices, electronic health records and telemedicine. These interventions have gained momentum due to their potential to improve healthcare access, reduce costs and deliver personalised medicine.¹ WHO advocates for the use of EHIs, recognising their potential to enhance health outcomes when integrated with standard care.²

Among EHIs, patient-reported outcomes (PROs) capture patient perspective³ and are increasingly used alongside clinical outcomes to offer a more comprehensive assessment of health status.⁴ Electronic PROs (ePROs) leverage digital tools such as computers, smartphones and telephone systems for real-time data collection, enabling dynamic patient monitoring.⁵ The impact of ePRO on the health-related quality of life (HRQoL) in patients with cancer has been explored in several studies, yielding mixed results. Some studies suggest that ePROs improve HRQoL,^{6 7} while others report no significant difference compared with usual care.^8–14 These varying outcomes have led to uncertainty regarding the utility of ePROs in cancer care.

Given this variability in findings, there is a need for robust evidence to guide healthcare stakeholders in adopting ePROs. Very few meta-analyses (MAs) have been conducted to date on the effect of ePRO on the HRQoL of patients with cancer.^15–18 For instance, the MA by Perry et al included only 13 studies and did not address heterogeneity in the HRQoL measurement scales. It reported outcomes using only the Functional Assessment of Cancer Therapy-General (FACT-G) scale, thereby overlooking other widely used tools such as the European Organisation for the Research and Treatment of Cancer Quality of Life Questionnaire (EORTC QLQ-C30).

ePRO interventions enhance quality of life (QoL) by improving patient engagement, facilitating better symptom management and enabling timely clinical interventions—making them an essential component of modern cancer care. This systematic review and meta-analysis (SRMA) aims to comprehensively synthesise the evidence on the impact of ePROs on HRQoL in patients with cancer.

Methods

Study design and reporting

This SRMA was conducted and reported in accordance with the Cochrane Collaboration Handbook for Systematic Review of Interventions and adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement.^{19 20} The study protocol was registered in the International Prospective Register of Systematic Reviews (PROSPERO: CRD42024531708) (registration details blinded for confidentiality but shared with the editors).

Eligibility criteria

We only included studies that met the following eligibility criteria: (1) randomised controlled trials (RCTs); (2) patients diagnosed with any type of cancer, either currently undergoing treatment or having completed treatment; (3) comparison between an ePROs and usual care without ePRO enhancement and (4) assessment of HRQoL as an outcome measure.

In this review, ‘usual care’ refers to standard oncology symptom monitoring without structured ePRO integration. This typically involves periodic symptom assessment during clinical visits, with symptom-related decisions made at the discretion of treating clinicians. In contrast, ePRO interventions involve structured, real-time digital patient-reported data collection, which may facilitate early symptom recognition and management.

Studies that did not provide overall HRQoL scores were excluded. Additionally, duplicate studies, including those presented at multiple conferences or published under different citations, were carefully identified and removed.

Search strategy and data extraction

We systematically searched PubMed, Embase and the Cochrane Central Register of Controlled Trials databases for studies in English published from inception to April 2024. The search strategy included terms related to cancer, patient-reported outcomes, electronic reporting and QoL (online supplemental file). We further performed a backward snowballing search using references from included studies and previous systematic reviews. Notably, we also incorporated recent research up until the final preparation of our manuscript, including the latest reviews by Perry et al and Urretavizcaya et al.^{15 17} This approach ensured the most up-to-date evidence was included.

A standardised template was developed for data extraction. We extracted the details on the study settings (author, year of publication, time of follow-up, study design and sample size), participants (population characteristics, age, sex, cancer type and stage), intervention (type of ePRO and its specific characteristics) and outcomes of interest. Data extraction was performed independently by two reviewers, with discrepancies resolved by consensus.

Outcomes and subgroup analysis

Definitions and details for the HRQoL measurements specified by each individual trial can be found in online supplemental tables S1 and S2.

HRQoL was measured by various validated tools, including the EORTC QLQ-C30, FACT-G and others. One instrument per study was selected to prevent duplication, with a preference for the EORTC QLQ-C30 when available.^{21 22} We standardised scores from these scales to a standardised mean difference (SMD) to facilitate meaningful comparisons and pooled estimates, covering all the included studies.²³

In this study, a multifaceted approach was employed to provide different views of both individual and overall effectiveness of the impact of ePROs. First, HRQoL at the final time point (end point) was evaluated to assess the overall effectiveness of interventions. Second, HRQoL changes from baseline to the end point were analysed to systematically assess progress over time. Additionally, the proportion of patients experiencing at least a 5-point improvement in HRQoL was analysed, which indicates clinically meaningful changes. Finally, patients were categorised based on whether their HRQoL scores improved or did not deteriorate compared with baseline. The ‘not deteriorated’ category quantified the proportion of patients who either showed improvement or maintained their HRQoL.

Prespecified subgroup analyses were conducted based on factors such as the method of PRO assessment, cancer site, metastasis status, therapy status and type of tool used to measure HRQoL.

Quality assessment and evidence certainty assessment

Risk of bias of individual studies was assessed independently using the revised Cochrane risk-of-bias tool (RoB V.2). Studies were categorised as low risk, high risk or having some concerns across five domains: randomisation, deviations from intended intervention, missing outcome data, measurement of the outcome and selection of the reported result. The RoB for two cluster trials and a crossover study was conducted with the aid of the revised Cochrane risk of bias tool with additional considerations. Disagreements were resolved through consensus. Grading of Recommendations, Assessment, Development and Evaluations (GRADE) was used to evaluate the certainty of evidence (CoE), considering risk of bias, precision, consistency, directness and publication bias. CoE was rated as high, moderate, low or very low (online supplemental file).

Data analysis

Binary outcomes were assessed using relative risks (RRs) with 95% CIs, coupled with the number needed to treat (NNT) for clinical decision-making. The NNT was determined using standard methodology with baseline risks derived from the included studies and RRs obtained from our MA. Continuous outcomes were expressed as SMDs with 95% CIs. The pooled SMD was back calculated to the EORTC scale for standardisation and ease of interpretation.

Forest plots were employed to visualise the findings, which also included details on interventions, risk of bias assessments and the CoE. Heterogeneity was assessed using a 95% prediction interval (95% PI), Cochran’s Q test and the I² statistic. It refers to variability across studies in patient characteristics (eg, cancer type, therapy status), methodological factors (eg, study design, ePRO delivery modes, follow-up duration) and outcome measurements (eg, different HRQoL scales). Due to the subjective nature of PROs and HRQoL assessments, a random effects model was used from the outset. We used PIs to capture the potential range of true effects of ePRO interventions, accounting for variability across different study settings. This approach offers a more comprehensive understanding of the uncertainty surrounding treatment outcomes, extending beyond the information provided by CIs alone.²⁴

We also explored the influence of the duration of ePRO interventions and average age on HRQoL through meta-regression, visualised with bubble plots, focusing on outcomes with six or more studies. Publication bias was assessed using funnel plots and Egger’s test to examine asymmetry in effect sizes. For outcomes with 10 or more studies, we employed trim-and-fill contour-enhanced funnel plots, while Doi plots were used for outcomes with fewer studies.^25–27

To ensure the robustness of our findings, we performed several sensitivity analyses: Bayesian sensitivity analysis, excluding all studies with a high risk of bias and leave-one-out analysis. All statistical analyses were conducted using available coding templates for R V.4.3.2.

Patient and public involvement statement

This MA provides valuable real-world evidence on the impact of ePRO on HRQoL in patients with cancer, benefiting clinicians, patients and other stakeholders. Due to funding constraints, patients and the public were not directly involved in the study. Since the research relied on precollected data, direct participation was not possible. However, we plan to share the findings through medical conferences, social media and publication to ensure they reach patients and the public. Future research could involve patients in the design of ePROs to make these tools even more effective.

Results

Study selection and characteristics

The initial database search identified 7706 records, as detailed in figure 1. After removing duplicates and title/abstract screening, 83 records were selected for full-text review. An additional 12 studies were identified through backward snowballing, resulting in 95 reports assessed for eligibility. Finally, 36 RCTs were included.

Figure 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses flow chart. QoL, quality of life.

A total of 9608 patients were included, of whom 49.99% received ePRO interventions. The ePRO interventions included mobile applications, web-based platforms, telephone-based systems and tablet-based systems. The studies were conducted across various geographical regions, with representation from Europe, Asia, the USA and Australia. Follow-up durations ranged from 3 weeks²⁸ to 2 years,²⁹ with patient ages ranging from 28 to 87 years table 1[]. Approximately 65.6% of participants were female, nearly double the proportion of male participants (34.4%). Demographic and intervention details are provided in online supplemental file.

Table 1.

Baseline characteristics of included studies

Study	Setting	Population n (age (years): mean/median), % female (F)	Cancer type/treatment	Intervention	Comparator	QoL tool	Follow-up	Funding
Banks et al 30	USA, single centre	60, (70), 2% F	LC (stage I–IV)/C+I+R	UC+TMI	UC	FACT-G	9 mo	Industry
Pradhan et al 28	India, tertiary centre	110, (53), 47% F	BC, LC, GUC, OC/chemo	MAI+PRO-CTCAE	UC	EORTC QLQ- C30	3 wk	None
Holch et al 6	UK, cancer centres	80 (22–84), 80% F	PC, GIC, GC/C+R	ePRO (e-RAPID)	UC	FACT-G, EORTC QLQ- C30	6 wk	Government-funded
Lee et al 8	Korea, single centre	213, (56), 62% F	BC, LC, HNC, GC/C+R	ePRO-CTCAE (MAI)	UC	EORTC QLQ- C30	8 wk	Government-funded
Singleton et al 43	Australia,tertiary+university	156, (54.8), 100% F	BC (survivors)/S+R+C+H+T	LMP (EMPOWER-SMS)	UC	EORTC QLQ- C30	6 mo	Government-funded
Sprave et al 44	Germany, tertiary+university	100, (60), 36% F	HNC/R	ePRO (MAI-myoncare)	SCM	EORTC QLQ- C30	RT completion	Government-funded
Verweij et al 31	The Netherlands, 12 hospitals	108, (18–65), 50% F	CML/TKI	WBI (CMYLife)	Offline questionnaire	EORTC QLQ- C30	6 mo	Government-funded
Basch et al 10	USA, multicentre	1191 (63), 48% F	Metastatic cancer/chemo	ePRO (Web/Phone)	UC	EORTC QLQ-C30	1 yr	PCORI
Mir O et al 45	France, single centre	559 (20–92), 52% F	OC/C+T	ePRO (App/Web)	In-hospital FU	EORTC QLQ-C30	6 mo	Government-funded, Industry
O’Hea et al 46	USA, oncology institutions	200 (60), 100% F	BC/C+S+R	CBP (POST)	UC	QoL-BC	6 mo	NCI
Prasongsook et al 47	Thailand, tertiary	33 (62), 36% F	LC/C+T+I	ePRO (MAI-self-report)	Paper-based	FACT-L	3 mo	Government-funded
Shiroiwa et al 11	Japan, multicentre	102 (60), 50% F	LC, BC/C+I	ePRO (tablet)	Paper-based	EORTC QLQ-C30	4 wk	CSPOR-PHRF
Tolstrup et al 7	Denmark, university hospital	146 (32–87), 52% F	MEL/Immuno	ePRO (immune AE)	SC	EQ-5D, FACT-M	48 wk	Danish Cancer Society
Wang et al 48	China, tertiary	103 (28–67), 100% F	BC (post-op)/S+C+R+I	LMP (iMBCR-MAI)	SC+education	FACT-B	1 mo	NSFC
Warsame et al 12	USA, Mayo Clinic	383 (32–87), 45% F	HM, HNC, GC/not specified	ePRO (PROQOL-iPad)	UC	LASA-QOL	12 mo	Mayo Clinic
Absolom et al 49	UK, cancer centre	508 (56), 60% F	CRC, BC, GC/C+T	ePRO (eRAPID)	UC	EORTC QLQ-C30, FACT-G	18 wk	NIHR
Maguire et al 50	Multicentre, EU	829 (52.4), 60% F	BC, CRC, HD, NHL, (0–IV)/C	ePRO (ASyMS)	SC	FACT-G	6 cycles	EU funding
Seven et al 51	Turkey, nursing school	40 (52–56), 100% F	BC (II–III)/C	MAI (MSY)	UC	EORTC QLQ-C30	12 wk	TUBITAK
Vos et al 29	The Netherlands, 8 hospitals	303 (63–75), 45% F	GIC (I–III)/S+C	WBI (Oncokompas)	GP/Surgeon-led	EORTC QLQ-C30	60 mo	Dutch Cancer Society
Fjell et al 13	Sweden, university hospitals	149 (48–50), 100% F	BC/C	MAI (Interaktor)	SC	EORTC QLQ-C30	2 wk	Karolinska Institute
Greer et al 52	USA, general hospital	181 (21–88), 51% F	HM, LC, BC, GUC, MEL, sarcoma/C+T	MAI	SC	FACT-G	12 wk	PCORI
Hou et al 53	Taiwan, medical centres	112 (20–64), 100% F	BC (0–III)/S+C+R+I	MAI (BCSMS)	UC	EORTC QLQ-C30	3 mo	Government-funded
Zhang et al 38	China, 28 centres	278 (57.6), 35% F	GC, LC, CRC, BC, HNC/I+C	ePRO (App)	UC	EORTC QLQ-C30	6 mo	Government-funded
Denis et al 37	France, 5 centres	121 (35.7–88.1), 45% F	LC (II–IV) C+R+T	WBI (e-FAP)	UC	FACT-L	12 mo	Industry
Ruland et al 54	Norway, USA	325 (56.7), 60% F	BC, PC/S+C+H+R	WBI (WebChoice)	Public URLs	15D HRQoL	12 mo	Government-funded
Velikova et al 55	UK, cancer centre	216 (54.9), 60% F	BC, GC, GUC, SC/C+H	ePRO (clinic—touchscreen)	UC	FACT-G	6 mo	Government-funded

Cancer types: BC, LC, HNC, CRC, GC, PC, MEL, HM, GIC, GUC, SC, OC, TC.

Interventions: ePRO, MAI, WBI, TMI, LMP.

Control groups: SC, PBR, WLC, SCP, SCM.

Therapies: C, R, I, T, H, S, M.

QoL tools: FACT-G, FACT-L, FACT-M, FACT-B, EORTC QLQ-C30, PCORI.

BC, breast; C, chemo; CRC, colorectal; EORTC QLQ-C30, European Organisation for Research and Treatment of Cancer Quality of Life Questionnaire; ePRO, electronic pros; EU, European Union; FACT-B, breast; FACT-G, functional assessment of cancer therapy-general; FACT-L, lung; FACT-M, melanoma; GC, gynaecological; GIC, gastrointestinal; GUC, genitourinary; H, hormonal; HM, haematological; HNC, head & neck; I, immuno; LC, lung; LMP, lifestyle/psychosocial; M, multimodal; MAI, mobile app; MEL, melanoma; mo, month; OC, oral; PBR, paper-based; PC, prostate; PCORI, Patient-Cantered Outcomes Research Institute; R, radio; S, surgery; SC, sarcoma; SC, standard care; SCM, standard monitoring; SCP, survivorship care; T, targeted; TC, thyroid; TMI, telemedicine; WBI, web-based; wk, week; WLC, wait-list/no intervention; yr, year.

Quality and evidence assessment

All studies were open-label and assessed QoL, a subjective outcome. This led to moderate to high risk of bias, commonly due to lack of blinding and outcome ascertainment. Each forest plot includes GRADE and RoB assessments to facilitate interpretation.

Primary outcome: HRQoL at end point

Pooled estimate

ePRO interventions showed improved QoL compared with usual care (SMD 0.35; 95% CI 0.18 to 0.51; 30 RCTs with 7204 participants; low CoE; table 2). This corresponds to a difference of 6.29 units (95% CI 3.24 to 9.16) on the EORTC QLQ-C30 scale.

Table 2.

GRADE table for the effect of ePRO interventions

ePRO compared with standard care for better quality of life
Patient or population: better quality of life Intervention: ePRO Comparison: standard care
Outcomes	No. of participants (studies)	Certainty of the evidence (GRADE)	Anticipated absolute effects with ePRO
QoL (end point)	6587 (30 RCTs)	⨁⨁◯◯ Low*†	SMD 0.35 SD higher (0.18 higher to 0.51 higher) EORTC scale: 6.29 units higher (3.24 higher to 9.16 higher)
Improvement in QoL (continuous outcome)	2049 (8 RCTs)	⨁◯◯◯ Very low*‡§	SMD 1.17 SD higher (0.38 lower to 2.72 higher)
Improvement in QoL (binary outcome)	1765 (4 RCTs)	⨁◯◯◯ Very low*‡	75 more per 1000 (14 less to 198 more) RR (95% CI): 1.33 (0.94 to 1.87) Risk with standard care: 228 per 1000
No deterioration in QoL (binary outcome)	1612 (4 RCTs)	⨁⨁◯◯ Low*§	92 more per 1000 (46 more to 144 more) RR (95% CI): 1.14 (1.07 to 1.22) Risk with standard care: 655 per 1000

GRADE Working Group grades of evidence: high certainty: we are very confident that the true effect lies close to that of the estimate of the effect. Moderate certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different. Low certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect. Very low certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect.

The risk in the intervention group (and its 95% CI) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).

*The studies were open-labelled and assessed a subjective outcome, that is, quality of life. Due to this nature of the review question, most studies demonstrated moderate to high risk of bias, with common issues including lack of blinding and outcome ascertainment.

†The point estimates of all or most of the studies show a benefit, and the CIs may cross the line because of smaller/pilot studies in this emerging area.

‡Downrated two levels for imprecision: the CIs include the possibility of harm.

§Downrated one level due to potential publication bias.

EORTC, European Organisation for the Research and Treatment of Cancer; ePRO, electronic patient-reported outcome; GRADE, Grading of Recommendations, Assessment, Development and Evaluations; QoL, quality of life; RCT, randomised controlled trial; RR, relative risk; SMD, standardised mean difference.

Heterogeneity and publication bias

The pooled estimates showed substantial variability between studies, with a PI of −0.53 to 1.22 and an I² of 81%. Subgroup analysis indicated that therapy status contributed to this variability (p=0.01) (online supplemental file). ePRO interventions resulted in a greater improvement in QoL with ongoing therapy (SMD 0.39; 95% CI 0.21 to 0.58) compared with those who had completed therapy (SMD 0.12; 95% CI 0.01 to 0.22). No significant differences were observed based on the method of PRO assessment (p=0.48; figure 2A), cancer site (p=0.11), metastasis status (p=0.60) or treatment experience (p=0.06). Meta-regression found no association between results and average age (p=0.65) or ePRO duration (p=0.15; figure 3B). The contour-enhanced funnel plot showed slight asymmetry confirmed by Egger’s regression test (p=0.06) (online supplemental file).

Figure 2.

Forest plot showing the effect, risk of bias assessment and certainty in evidence with electronic patient-reported outcomes (ePROs) intervention compared with standard care for (A) health-related quality of life (HRQoL) (at end point), (B) change in HRQoL, (C) improvement in HRQoL and (D) no deterioration in HRQoL.

Figure 3.

(A) Bayesian sensitivity analysis for difference in health-related quality of life (HRQoL) with electronic patient-reported outcomes (ePROs) intervention compared with standard care. (B) Bubble plot to visualise meta-regression for difference in HRQoL with ePRO intervention compared with standard care based on the duration of intervention. (C) Contour-enhanced trim-and-fill funnel plot for quality of life (at end point).

Sensitivity analyses

Findings remained consistent across sensitivity analyses, including exclusion of high-risk bias studies (SMD 0.30; 95% CI 0.15 to 0.46; online supplemental file), Bayesian sensitivity analysis (SMD 0.34; 95% credible interval (CrI) 0.18 to 0.51; figure 3A) and leave-one-out MA (online supplemental file).

Secondary outcomes

HRQoL (change from baseline)

The pooled data showed an SMD of 1.17 (95% CI −0.38 to 2.72; eight RCTs with 2611 participants; very low CoE), for differences in QoL improvement with ePRO compared with usual care (figure 2B). Subgroup differences were not detected based on metastasis status (p=0.57) and treatment experience (p=0.08). Significant subgroup effects were found for method of PRO assessment (p<0.01), cancer site (p<0.01) and type of tool used (p<0.01). No association detected between change in QoL and average age (p=0.62) or duration of ePRO (p=0.77). The Doi plot was asymmetric, with an LFK index of 2.81 indicating a potential publication bias.

Results remained consistent after excluding studies with a high risk of bias (SMD 0.50; 95% CI 0.04 to 0.96), in the Bayesian sensitivity analysis (SMD 0.71; 95% CrI −0.57 to 1.95) and majorly in the leave-one-out MA (online supplemental file).

Improvement in HRQoL (binary measure)

The pooled data showed an RR of 1.33 (95% CI 0.94 to 1.87; four RCTs with 841 participants; very low CoE; figure 2C) for QoL improvement with ePRO compared with standard care. Subgroup analysis revealed a difference based on the type of PRO assessment (p=0.01). Bayesian sensitivity analysis (RR 1.30; 95% CrI 0.85 to 1.95) and leave-one-out MA confirmed consistent results. Excluding high-risk bias studies left only one study (RR 1.06; 95% CI 0.72 to 1.57) (online supplemental file).

No deterioration in HRQoL

ePRO was associated with prevention of QoL deterioration compared with usual care (RR 1.14; 95% CI 1.07 to 1.22; four RCTs with 1612 participants; low CoE; figure 2D). The NNT was calculated as 11 (95% CI 7 to 22). Subgroup analysis found no significant difference for the type of PRO assessment (p=0.39). Bayesian sensitivity analysis (RR 1.16; 95% CrI 0.97 to 1.38) and leave-one-out MA showed consistent results (online supplemental file).

Discussion

Our MA of 36 RCTs involving 9608 patients with cancer found that ePRO interventions significantly improved HRQoL by 7.83 points on the EORTC QLQ-C30 scale. Given that the minimal clinically important difference (MCID) typically ranges from 5 to 10 units, this improvement is both statistically significant and clinically meaningful.²² Musoro et al ²² synthesised data from 21 trials across nine cancer types, supporting this MCID threshold for the EORTC QLQ-C30. Additionally, Basch et al ¹⁰ demonstrated that ePRO interventions often lead to HRQoL improvements within this range.

We also found an NNT of 11, emphasising the clinical relevance of ePRO interventions. However, we observed no significant change in HRQoL from baseline or in the proportion of patients improving, likely due to the small number of studies and low event rates, which reduced statistical power.

Our findings align with four recent MAs supporting the role of ePROs in improving HRQoL. Perry et al ¹⁵ limited their analysis to 13 studies due to heterogeneity in HRQoL measures, whereas we assessed HRQoL at multiple time points. This comprehensive approach allowed us to include 36 RCTs—>2.5 times the number included in the meta-analysis by Perry et al. Similarly, Li et al ¹⁶ found shorter interventions or phone calls were less effective when compared with application or web-based tools. The authors categorised HRQoL data based on the intervention’s duration, with therapies lasting >3 or <3 months showing different results. Larson et al ¹⁸ found that telehealth therapies improved HRQoL, while research by Urretavizcaya et al,¹⁷ which included only eight studies, noted scale variability and study population heterogeneity made clinical outcome comparisons challenging. These characteristics complicate the interpretation of findings difficult and make it challenging to draw consistent conclusions about the effectiveness of ePRO interventions.

Our meta-regression analysis included interventions ranging from 3 weeks to 2 years. We found no significant association between HRQoL improvement and intervention length, contradicting previous MAs suggesting longer interventions are more effective. Instead, our results indicate that patient engagement and intervention design may be more influential, as we observed improvements in both short-term (6 weeks)²⁸ and long-term (2 years)²⁹ interventions.

Our subgroup analysis revealed that patients undergoing active therapy (SMD 0.39) benefited more from ePRO interventions than those who had completed treatment (SMD 0.12). This suggests that ePRO interventions may be particularly effective during periods of active treatment. However, beyond the active treatment phase, ePRO interventions continue to play a role in survivorship care. It supports long-term symptom monitoring, mental health tracking and QoL assessments. It allows oncologists, general practitioners and allied healthcare professionals to evaluate patient-reported data, enabling timely follow-ups and personalised care strategies. Studies such as those by Banks et al ³⁰ and Vos et al ²⁹ illustrate how structured symptom assessments via ePRO remain an integral part of survivorship management, ensuring continued engagement and proactive health monitoring.

ePRO serves as an enhanced symptom-monitoring tool rather than a therapeutic intervention. Our findings suggest that structured patient-reported symptom tracking via ePRO improves HRQoL by facilitating timely clinical interventions, which is distinct from passive symptom assessment in usual care. ePRO also enhances communication with healthcare providers, making care more personalised and responsive. Continuous monitoring and real-time feedback, as shown by Shiroiwa et al,¹¹ help detect issues early, allowing for timely management of symptoms and improvements in HRQoL, especially between chemotherapy sessions. Structured symptom tracking, as noted by Banks et al,³⁰ has also been effective in reducing the need for acute care by facilitating systematic assessments. Other studies, such as those by Verweij et al ³¹ and Basch et al,¹⁰ highlight how patient engagement through digital platforms supports proactive symptom management.

Despite these positive outcomes, some studies reported neutral or negative effects of ePRO on HRQoL. Lee et al ⁸ suggested that small sample sizes and recall bias could have limited the findings, while Vos et al ²⁹ found no significant differences, possibly due to the study’s broad focus and high baseline HRQoL. Similarly, Ryhänen et al ³² noted no significant impact from the Breast Cancer Patient Pathway programme, suggesting standard care was adequate in these cases.

The Oncokompas eHealth tool, though well-validated, produced contrasting results. van der Hout et al ³³ reported significant improvements in HRQoL, likely due to personalised feedback and high levels of patient engagement. In contrast, Vos et al ²⁹ found no significant changes, which may be explained by the participants’ already high baseline HRQoL and the comprehensive standard care they received. These differences suggest that the effectiveness of eHealth tools like Oncokompas may depend on initial health status and the level of patient engagement.

Although our MA did not include a direct economic evaluation, previous studies have indicated that ePRO interventions may provide significant cost savings. Pantiora et al ³⁴ demonstrated that ePROs reduced administrative costs and healthcare utilisation without affecting HRQoL, while Riis et al ¹⁴ reported similar findings. These results suggest that ePRO interventions are both clinically beneficial and cost-effective, making them valuable tools in cancer care.

Despite positive effect sizes (SMD), large PIs were observed, indicating uncertainty about the effects of ePRO interventions in future studies. While the overall impact appears beneficial, the efficacy may vary. IntHout et al ^{24 35} recommended including PIs in all reviews, as they provide valuable context for clinicians. Wide PIs, often caused by small sample sizes and continuous outcomes, were observed in our SRMA, highlighting the need to account for patient diversity and study design when assessing the effectiveness of ePRO interventions.

We analysed HRQoL in four different ways to capture patient experiences. Converting SMD to EORTC QLQ-C30 units provided clearer clinical context, enhancing practicability. Using NNT helps direct clinical interpretation of the benefit. Our review included research from Europe, Asia, the USA and Australia showing global interest in ePRO’s impact in cancer care. Our SRMA synthesised report from 12 multicentric RCTs,^{10 36–41} including cluster^{10 41} and crossover trials.⁴² It provided broader insights than previous SRMA.

Our MA has some limitations. None of the included studies employed blinding, which can introduce bias. High heterogeneity across studies was expected due to the diverse conditions under which they were conducted. It was largely due to differences in therapy status, as ongoing therapy patients showed greater HRQoL improvements than those who had completed therapy. The use of multiple HRQoL scales further contributed to statistical heterogeneity. We accounted for this using PIs, which provide a more comprehensive representation of variability across different clinical settings. We addressed this through subgroup analysis and meta-regression based on clinically relevant variables. Another limitation is the lack of standardisation in HRQoL measurements across studies. To mitigate this, we used SMDs and back-transformed the pooled estimates to the EORTC QLQ-C30 scale, allowing for a more clinically relevant interpretation of the results. Future research should focus on addressing sources of heterogeneity, exploring long-term impacts and enhancing patient engagement with ePRO systems.

Conclusion

Our study suggests that ePRO interventions provide a structured approach to symptom monitoring and are associated with improved HRQoL in patients with cancer. Their integration into routine oncology care may facilitate timely symptom management, potentially leading to better patient outcomes. However, given that ePRO is a symptom-monitoring tool rather than a direct therapeutic intervention, these results should be interpreted as associative rather than causative. Future research should explore whether these improvements translate to long-term clinical benefits.

Data availability statement

All data relevant to the study are included in the article or uploaded as supplementary information.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

Not applicable.