Impact of Nurse Navigation on Overall Survival and Timeliness to Care in Patients With Pancreatic Cancer in Advanced Stages

Orlando Health Cancer Institute, Orlando, FL

Purpose: Oncology nurse navigation programs have demonstrated improved patient outcomes affecting time to treatment, adherence with evidence-based therapy, and patient experience. We assessed the impact of nurse navigation on outcomes among patients with pancreatic adenocarcinoma in a real-world setting at a tertiary referral center.

Methods: Patients with pancreatic ductal adenocarcinoma (stage III or IV) treated in 2019 and 2020 were identified using cancer registry data. In this time frame, nurse navigation availability varied due to real-world circumstances, allowing for a quasi-experimental design comparing patients assigned or not assigned nurse navigation.

Using a Bayesian causal inference framework, we estimated the average treatment effect (ATE) of nurse navigation primarily on overall survival as well as time to first treatment, obtaining a nutrition consult, completing genetic and molecular testing, and tumor board consultation. Clinical significance was judged by the probability of the posterior of ATE being greater than zero (Pr[ATE]>0).

Results: Among 59 patients with stage III or IV pancreatic adenocarcinoma treated in 2019 and 2020, we found that nurse navigation had a positive impact on survival over 24 months. Specifically, patients with nurse navigation had a 104% higher probability of survival at 1 year compared with patients with no nurse navigation (Pr[ATE>0]=0.993).

We found weak evidence for a positive impact of nurse navigation on the time to nutrition consult at 1 month (Pr[ATE>0]=0.816) and the completion of genetic testing (Pr[ATE>0]=0.825) when compared with patients with no nurse navigation. We found lack of evidence for an impact of nurse navigation on the time to first treatment (Pr[ATE>0]=0.765), completion of molecular testing (Pr[ATE>0]=0.406), and discussion at tumor board (Pr[ATE>0]=0.585) when compared with patients with no nurse navigation.

Conclusion: Patients with pancreatic cancer benefited from the involvement of nurse navigation in their care in a real-world setting, which supports further health system investments in this space.

Future work should aim to understand how nurse navigation mediates increased survival and should evaluate the generalizability of beneficial effects within a meta-analysis framework.

Multidisciplinary teams are a key factor in oncology care, and coordination of such diverse teams is essential to their core functions.¹ Nurse navigators play an essential role in communication among multiple disciplines, coordination of services, continuity of care, and patient communication.² The specific roles and definitions of a nurse navigator are diverse and varying and can range from specific focuses on helping patients during the screening, diagnostic, or treatment processes, or all of those roles combined.^2,3

A growing body of evidence supports that nurse navigation improves patient outcomes by coordinating care, identifying and addressing barriers, and educating and advocating for the patient.^2-5 Cancer patients are often faced with multiple appointments with multiple providers, which can be confusing, and as such, they need guidance. Nurse navigators assist in coordinating the correct appointment with the appropriate provider at the right time across the cancer continuum. In turn, navigation has been shown to decrease the time to key treatment landmarks such as time to first treatment.^6-8 Patients may encounter barriers related to advanced age, minority race, and low socioeconomic status. Nurse navigators help overcome these barriers by coordinating additional services, such as offering transportation or covering transportation costs.⁹ The nurse navigator serves as an advocate for the patient by making sure their interests are incorporated into their plan of care.^10,11 As a consequence, stressors such as anxiety and fear are reduced, which can improve treatment adherence and outcomes.^11,12 Due to the clinical expertise, nurse navigators are well positioned to provide a high quality of care.

Although there are many benefits of nurse navigation, several gaps still remain in assessing its impact on oncology patient care. There is a lack of assessments of the impact of nurse navigation on survival, one of the most important outcomes from cancer treatment.¹³ However, current clinical trials are incorporating survival as an outcome.^14,15 Furthermore, much work on navigation has been focused primarily on breast, cervical, and lung cancers over other types.^16,17 Other cancer types warrant further investigation, especially those that impact survival,¹⁸ such as pancreatic cancers.

Our goal in this retrospective study within a single center was to determine the impact of nurse navigation on timeliness of care, completion of ancillary services, and overall survival of patients with late-stage pancreatic cancer. We used a novel Bayesian method to estimate the average treatment effect (ATE) of nurse navigation on survival in patients with pancreatic cancer, which accommodates lower sample sizes and produces more clinically meaningful interpretations.

Methods

Patient Population

We conducted a retrospective analysis within a single health system for patients with stage III or IV (American Joint Committee on Cancer v8) pancreatic adenocarcinoma diagnosed between 2019 and 2020 and treated at our institution based on cancer registry data. Between 2019 and 2020, nurse navigation services were unavailable February through April 2018 and February through April 2020, a 6-month period. This resulted in a quasi-experimental study that allowed us to determine the impact of nurse navigation on patient outcomes. Nurse navigation was not randomly assigned, and the treatment assignment can be thought of as haphazard. The main point of comparison are patients with and without nurse navigation. The main role of nurse navigation is to address barriers to care and improve adherence with evidence-based treatment and diagnostic testing. If assigned, a nurse navigator began at the time of the new patient visit for cancer diagnosis or within 2 business days of a referral being placed. At the time of this study, there was 1 nurse navigator specializing in pancreatic cancer. Responsibilities included assessing and addressing barriers to care; coordination of diagnostic images, labs, and appointments across multiple providers; and educating the patient and caregiver about the plan of care. We excluded patients who did not seek care at our center, died before treatment could be administered, or enrolled in hospice prior to any cancer treatment (thus not having any contact with nurse navigation).

We collected whether patients had a nurse navigator or not, along with demographic information that would be confounders in causal models, including age, sex, race, insurance, treatment location, and cancer stage. The outcomes collected were overall survival, time to first treatment, time to nutrition consult, completion of germline genetic testing, completion of molecular testing, and whether the patient had a tumor board review. Overall survival was measured as the time between date of diagnosis and either 2 years of follow-up or mortality date. Time of first treatment was measured as time from date of diagnosis to first treatment date. Time of nutrition consult was measured as the time from date of diagnosis to date of nutrition consult.^19,20 Germline genetic testing included referral to cancer genetic counselors who performed pretest and posttest counseling. Molecular testing included somatic tumor or blood-based testing. Tumor board included multidisciplinary case review with surgical oncology, medical oncology, radiology, radiation oncology, pathology, and a nurse navigator.

Statistical Analyses

We used a causal inference framework to estimate the causal effect, or ATE, of nurse navigation on 6 outcomes: overall survival, time to first treatment, time to nutrition consult, completion of genetic testing, completion of molecular testing, and whether the patient had a tumor board review. Overall survival, time to first treatment, and time to nutrition consult are time-to-event outcomes, while the other outcomes are binary. Within the causal inference framework,^21,22 the causal effect is described as a difference in potential outcomes: E[g(T¹)]-E[g(T⁰)], where g is some function (ie, link for logistic regression), and T¹ and T⁰ are outcomes under nurse navigation and no nurse navigation, respectively.²³ However, only 1 outcome is observed for a given patient, and the missing outcome is the counterfactual. Nonetheless, ATE can be estimated under the assumptions of conditional ignorability, treatment positivity, and consistency.²³ Conditional ignorability assumes that potential outcomes are unrelated to the treatment assignment, conditional on observed covariates. In our study, the observed covariates are baseline confounders. For the overall survival, the baseline confounders included were sex, age, race, insurance, stage, and treatment location. For other outcomes, the confounders were insurance and treatment location because it is expected that the other variables do not affect completion of services. Positivity assumes that all patients have the option to experience both treatments that are represented by any subgroups of the confounders, and that there is overlap between the treatment groups (Table 1). Lastly, consistency assumes that there is 1 version of the treatment, and that the observed outcome for a patient is equal to the potential outcome. There is an additional assumption that there is noninformative censoring, which is common in survival models. Given these assumptions, estimation of ATE is possible through the process of g-computation,^24,25 which involves contrasting the average treatment distributions of predicted outcomes from a fitted outcome model.

For time-to-event outcomes, we used causalBETA package²³ for fitting Bayesian survival models using g-computation. ATE is P(T^nurse>t) – P(T^{no nurse}>t), which is the mean difference in the distribution of marginal survival (P[T>t]) probabilities between nurse and no nurse navigation. Briefly, a hazard model is fitted, and the baseline hazard is modeled as a piecewise constant function over partitions of time. Survival probabilities are estimated with Monte Carlo integration of the hazard function over the confounder distribution under each treatment and time period (full details²³). Then, ATE is the contrast between the survival probabilities between treatment groups at a given time point. We set the number of Monte Carlo simulations to 1000, set the posterior draws to 10,000, set the warm-up in Markov Chain Monte Carlo to 1000, set the number of partitions to 100, set the prior process for the baseline hazard to an autoregressive prior, and used 3 chains. For the prior of the regression coefficients of the covariates, we set 2 standard deviations of the hazard ratio (HR) to be 0.33 or 3 around a mean HR of 1. Therefore, we set a weakly informed prior for the log HR to be normally distributed with a mean of 0 and standard sigma of 0.5605165 (N[0, 0.5605165]). ATE was evaluated from 1 to 24 months for overall survival, and at 1 month for time to first treatment and time to nutrition consult outcomes. To aid in interpretation, time to first treatment and time to nutrition outcomes were converted to cumulative incidence by subtracting the marginal survival curves by 1 (1-P[T>t]). For binary outcomes, we also used a g-computation approach and fitted an outcome logistic regression in the brms^26,27 R package. ATE can be estimated by contrasting the probability of the event happening between nurse navigation and no nurse navigation conditional on baseline covariates (E[y_i^nurse -y_i^{no nurse}]D_i),^25,27 where D_i represents the baseline covariate and i indexes the patient), which was calculated using the marginal effects²⁸ R package. All of the Bayesian models fitted converged as indicated by visual inspection of chain trace plots and high potential scale reduction factor (nearing 1²⁹).

Clinical significance of the posterior distribution of ATE was determined by both the mean point estimate as well as the distribution of 95% Bayesian credible interval (BCI). In addition, we quantified the strength of evidence for a beneficial impact of nurse navigation by calculating the probability that the ATE posterior distribution is greater than 0 (Pr[ATE>0]). We interpret probabilities of nurse navigation benefits to have strong evidence when above 0.95 and weak evidence for probabilities ranging from 0.8 to 0.95.

Results

In total, 59 patients were included in the study out of 140 cases screened. Patients with nurse navigation had a significant benefit in overall survival over 24 months compared with patients without nurse navigation (Figure). In fact, ATE or survival difference between nurse navigation and no nurse navigation peaked at 8 months and then gradually declined. At 1 year, patients with nurse navigation had a marginal survival probability of 0.431, and patients without navigation had a 0.211 marginal survival probability, which amounts to a 0.22 (95% BCI=[0.047-0.392]) difference in survival probability (Pr[ATE>0]=0.993), or a 104% higher survival probability in patients with nurse navigation than patients without nurse navigation (Table 2). We found that nurse navigation did not significantly increase the probability of being discussed by tumor board (Pr[ATE>0]=0.585) or the probability of completing molecular testing (Pr[ATE>0]=0.406), and did not impact time to first treatment (Pr[ATE>0]=0.765; Table 2). However, we found weak evidence for the benefit of nurse navigation on time to nutrition consult (Pr[ATE>0]=0.816) and completion of genetic testing (Pr[ATE>0]=0.825; Table 2). Specifically, patients with nurse navigation had a point estimate of 0.083 (95% BCI=–0.104-0.258) higher cumulative incidence of obtaining a nutrition consult at 1 month than patients without nurse navigation (Table 2), which represents a 17.84% increase from a cumulative incidence of 0.4652 in patients without nurse navigation. Patients with nurse navigation had a point estimate of 0.103 (-0.104-0.327) higher probability of completing germline genetic testing than patients without nurse navigation (Table 2), which represents a 26.21% increase from the 0.393 probability of completing germline genetic testing in patients without nurse navigation.

Discussion

We demonstrated an overall survival impact of nurse navigation among patients with stage III or IV pancreatic adenocarcinoma treated at our center. However, there was no difference in time to first treatment, completion of molecular testing, and tumor board consultation, while a trend toward early nutrition consults and germline genetic testing completion was noted. These findings build upon prior and ongoing work of the benefits of nurse navigation. This is one of the few studies showing a survival benefit of nurse navigation, which has been lacking in the broader literature.^13,16 Assessment of impact in real-world application of nurse navigation is essential to direct program growth.^13,16

The survival benefit of nurse navigation may be due to unmeasured mediators. For example, nurse navigation is known to enhance patient satisfaction and quality of life,^17,30 support for patient and family,³¹ and decreasing barriers to care.^9,32 Examples of decreasing barriers to care include providing transportation assistance,³³ lodging assistance,^34,35 help with financial toxicity,³⁶ and health literacy.³⁷ However, these metrics were not formally measured in our study. Addressing these barriers may allow patients access to and adherence with those evidence-based treatments known to impact overall survival in this population. In support of this is the peak effect of nurse navigation at 8 to 10 months after diagnosis, a time frame when the ability to stay on treatment longer or engage in ancillary services can make an impact, as opposed to the first 2 months where a rapidly progressing disease process may not be intervenable. We excluded patients who went to hospice prior to first treatment, which would select for a better prognostic group with less comorbidities.

At our institution, pancreatic cancer patients are cared for within a multidisciplinary team based on evidence-based medicine,¹ which streamlines all levels of care across the cancer continuum.² At the same time, pancreatic cancer is very aggressive, and when diagnosed in late stage can result in a poor prognosis.¹⁸ Nevertheless, there is thoughtful consideration on quantity and quality of life, such as offering palliative care services for the patient and the family members for end-of-life care.^30,38

In addition, the survival benefit of nurse navigation may occur due to not 1 but many interventions and vary based on the needs of the patient and caregiver(s). For example, we found trends toward a positive benefit of nurse navigation on time to nutrition consult and overall completion of genetic testing. The decreased time to nutrition consults via nurse navigation may have elevated the nutritional status of patients,¹⁹ which could have allowed for improved performance status and treatment tolerance. Germline genetic testing may result in additional targeted therapies for patients and engage caregivers to be proactive in their care. In contrast to our findings, nurse navigation has been demonstrated to improve timeliness of care.^7,30 This is achieved when the nurse navigator plays an active role in coordinating care.^2,3,5 We found that nurse navigation did not impact time to first treatment, completion of molecular testing, and tumor board discussion.

This study has limitations in that it was conducted within a single health system for a specific cancer population. However, this population of stage III and IV pancreatic cancer patients allowed for analysis of survival impact in a shorter time frame and evaluation in a population in which multidisciplinary care coordination is of keen importance. As with any quasi-experimental design, there are unmeasured variables that could not be included in the model, and the small sample size may impact positivity.²³ Other institutions may not have the resources to have nurse navigation dedicated to only gastrointestinal/pancreatic malignancy to allow development of specialized knowledge. On the other hand, some comprehensive cancer centers have patients complete molecular testing and nutrition consult prior to or at the time of initial visit. This would potentially account for some of the benefits associated with nurse navigation at our institution. Future work should address these limitations and evaluate the mediators of the effects seen.

In conclusion, this study highlights the real-world impact of nurse navigation on pancreatic cancer outcomes at a single health system. Based on our results, further resources are recommended to ensure all patients receive support from a disease-specific nurse navigator during their cancer journey.

References

1. LaVigne AW, Doss VL, Berizzi D, et al. The history and future of multidisciplinary cancer care. Semin Radiat Oncol. 2024;34:441-451.
2. Cantril CA, Blaseg KD, Bartelt S, et al. Oncology Nurse Navigation: Delivering Patient-Centered Care Across the Continuum. Oncology Nursing Society; 2014. Accessed December 4, 2024. https://archive.org/details/oncologynursenav0000unse/page/n3/mode/2up
3. Byrne AL, Hegney D, Harvey C, et al. Exploring the nurse navigator role: a thematic analysis. J Nurs Manag. 2020;28:814-821.
4. Braun KL, Thomas WL Jr, Domingo JLB, et al. Reducing cancer screening disparities in Medicare beneficiaries through cancer patient navigation. J Am Geriatr Soc. 2015;63:365-370.
5. Lubejko BG, Bellfield S, Kahn E, et al. Oncology nurse navigation: results of the 2016 role delineation study. Clin J Oncol Nurs. 2017;21:43-50.
6. Hunnibell LS. Using nurse navigation to improve timeliness of lung cancer care at a veterans hospital. Clin J Oncol Nurs. 2012;16:29-36.
7. Enomoto LM, Fenstermaker J, Desnoyers RJ, et al. Oncology navigation decreases time to treatment in patients with pancreatic malignancy. Ann Surg Oncol. 2019;26:1512-1518.
8. Rodrigues RL, Schneider F, Kalinke LP, et al. Clinical outcomes of patient navigation performed by nurses in the oncology setting: an integrative review. Rev Bras Enferm. 2021;74:e20190804.
9. Mandelblatt JS, Yabroff KR, Kerner JF. Equitable access to cancer services: a review of barriers to quality care. Cancer. 1999;86:2378-2390.
10. Hacking B, Wallace L, Scott S, et al. Testing the feasibility, acceptability and effectiveness of a ‘decision navigation’ intervention for early stage prostate cancer patients in Scotland – a randomised controlled trial. Psychooncology. 2013;22:1017-1024.
11. Ali-Faisal SF, Colella TJF, Medina-Jaudes N, Benz Scott L. The effectiveness of patient navigation to improve healthcare utilization outcomes: a meta-analysis of randomized controlled trials. Patient Educ Couns. 2017;100:436-448.
12. Bush ML, Kaufman MR, Shackleford T. Adherence in the cancer care setting: a systematic review of patient navigation to traverse barriers. J Cancer Educ. 2018;33:1222-1229.
13. Wells KJ, Battaglia TA, Dudley DJ, et al. Patient navigation: state of the art or is it science? Cancer. 2008;113:1999-2010.
14. Duzova US, Can G. The effect of navigation programme on the management of symptoms related to head and neck radiotherapy. Accessed January 9, 2025. Transpl Immunol. 2021;69:101488.
15. Dalton S. NAVIGATE - Improving Survival in Vulnerable Lung Cancer Patients Through Nurse Navigation, Symptom Monitoring and Exercise: Study Protocol for a Multicenter Randomized Controlled Trial. clinicaltrials.gov; 2024. Accessed January 9, 2025. https://clinicaltrials.gov/study/NCT05053997
16. Baik SH, Gallo LC, Wells KJ. Patient navigation in breast cancer treatment and survivorship: a systematic review. J Clin Oncol. 2016;34:3686-3696.
17. Chen M, Wu VS, Falk D, et al. Patient navigation in cancer treatment: a systematic review. Curr Oncol Rep. 2024;26:504-537.
18. Cai J, Chen H, Lu M, et al. Advances in the epidemiology of pancreatic cancer: trends, risk factors, screening, and prognosis. Cancer Lett. 2021;520:1-11.
19. Laviano A, Lazzaro LD, Koverech A. Nutrition support and clinical outcome in advanced cancer patients. Proc Nutr Soc. 2018;77:388-393.
20. Ravasco P. Nutrition in cancer patients. J Clin Med. 2019;8:1211.
21. Imbens GW, Rubin DB. Causal Inference in Statistics, Social, and Biomedical Sciences. Cambridge University Press; 2015.
22. Hernan MA, Robins JM. Causal Inference: What If. CRC Press; 2020. www.hsph.harvard.edu/miguel-hernan/wp-content/uploads/sites/1268/2024/04/hernanrobins_WhatIf_26apr24.pdf
23. Ji H, Oganisian A. causalBETA: an R Package for Bayesian Semiparametric Causal Inference with Event-Time Outcomes. Published online October 20, 2023. Accessed February 8, 2024. http://arxiv.org/abs/2310.12358
24. Snowden JM, Rose S, Mortimer KM. Implementation of G-computation on a simulated data set: demonstration of a causal inference technique. Am J Epidemiol. 2011;173:731-738.
25. Muller CJ, MacLehose RF. Estimating predicted probabilities from logistic regression: different methods correspond to different target populations. Int J Epidemiol. 2014;43:962-970.
26. Bürkner PC. brms: a R package for Bayesian multilevel models using Stan. Journal of Statistical Software. 2017;80:1-28.
27. Kurz AS. Causal inference with Bayesian models. 2023. Accessed July 1, 2024. https://solomonkurz.netlify.app/blog/2023-04-30-causal- inference-with-bayesian-models/
28. Arel-Bundock V, Greifer N, Heiss A. How to interpret statistical models using marginaleffects in R and Python. Journal of Statistical Software. Published online 2024. www.jstatsoft.org/article/view/v111i09
29. Brooks SP, Gelman A. General methods for monitoring convergence of iterative simulations. J Comput Graph Stat. 1998;7:434-455.
30. Kokorelias KM, Shiers-Hanley JE, Rios J, et al. Factors influencing the implementation of patient navigation programs for adults with complex needs: a scoping review of the literature. Health Serv Insights. 2021;14:11786329211033267.
31. Dionne-Odom JN, Azuero A, Taylor RA, et al. A lay navigator-led, early palliative care intervention for African American and rural family caregivers of individuals with advanced cancer (Project Cornerstone): results of a pilot randomized trial. Cancer. 2022;128:1321-1330.
32. Hendren S, Chin N, Fisher S, et al. Patients’ barriers to receipt of cancer care, and factors associated with needing more assistance from a patient navigator. J Natl Med Assoc. 2011;103:701-710.
33. Wercholuk AN, Parikh AA, Snyder RA. The road less traveled: transportation barriers to cancer care delivery in the rural patient population. JCO Oncol Pract. Published online September 2022. https://ascopubs.org/doi/10.1200/OP.22.00122
34. Gany F, Melnic I, Ramirez J, et al. The association between housing and food insecurity among medically underserved cancer patients. Support Care Cancer. 2021;29:7765-7774.
35. Fan Q, Keene DE, Banegas MP, et al. Housing insecurity among patients with cancer. J Natl Cancer Inst. 2022;114:1584-1592.
36. Lentz R, Benson AB 3rd, Kircher S. Financial toxicity in cancer care: prevalence, causes, consequences, and reduction strategies. J Surg Oncol. 2019;120:85-92.
37. Holden CE, Wheelwright S, Harle A, Wagland R. The role of health literacy in cancer care: a mixed studies systematic review. PLoS One. 2021;16:e0259815.
38. Hui D, Kim SH, Roquemore J, et al. Impact of timing and setting of palliative care referral on quality of end-of-life care in cancer patients. Cancer. 2014;120:1743-1749.