Do Navigators’ Estimates of Navigation Intensity Predict Navigation Time for Cancer Care?

Abstract

Purpose

Patient navigation requires patient load be equitably distributed. We examined whether navigators could predict the relative amount of time needed by different patients for navigation.

Methods

Analysis of 139 breast and colorectal cancer patients randomized to the navigation arm of a trial evaluating the effectiveness of navigation. Navigators completed a one-item scale estimating how much navigation time patients were likely to require.

Results

Participants were mostly females (89.2%) with breast cancer (83.4%); barriers to cancer care were insurance difficulties (26.6%), social support (18.0%) and transportation (14.4%). Navigator baseline estimates of navigation intensity predicted total navigation time, independent of patient characteristics. The total number of barriers, rather than any specific type of barrier, predicted increased navigator time, with a 16% increase for each barrier.

Conclusion

Navigators’ estimate of intensity independently predicts navigation time for cancer patients. Findings have implications for assigning navigator case loads.

Background

In the case of cancer treatment in the United States, rising tides have not lifted all boats equally. Despite major technological advances in cancer treatment, poor, uninsured and/or minority patients often receive less optimal treatment than affluent, insured non-Hispanic Caucasian patients[1–4].

Patient navigation (PN) – the process of assessing and alleviating barriers to adequate health care by a specifically-trained person[5–7]– helps patients complete recommended treatment and reduce socio-economic, racial and ethnic disparities in cancer care[8; 9]. PN programs were originally developed from community health worker programs over 40 years ago in the United States[8; 9]. Early studies of community health workers and lay/peer navigators focused on screening and initial work-up, showing that navigation improved access to care and completion of cancer related screening and diagnostic evaluation for underserved populations[5–7]. Since then PN programs are being widely implemented in the US to help patients complete the initial diagnostic workup in a timely fashion[10], assist newly diagnosed patients begin treatment[11], and provide assistance with palliative care issues[12]. However, very little is known about the role of navigation during cancer treatment.

PN programs vary widely in how they are implemented in different settings. PN programs are generally not reimbursed by insurers, so they are usually funded through local resources, institutional funds, or foundation support[13]. Training, salary, and local resources available to support navigation work vary considerably. Recent work has shown that PN is effective for improving patient satisfaction and decreasing barriers to care[14]. Because navigation can be time-and resource intensive working with the complex needs of underserved patient populations, approaches to determine how to best allocate navigation services are needed, especially for cancer treatment.

Patients differ widely in the number and types of needs related to cancer care[15; 16], and thus in the amount of navigation time they will need. Whether navigators can assess patients’ time needed is not known. In fact, very limited information exists about how well health care professionals or paraprofessionals are able to assess time needs or service complexity in order to plan caseloads and work tasks efficiently. A study of public health nurses[17] found that nurses were able to accurately predict patient needs and level of nursing time among a sample of 1,352 patients receiving community nursing services. In that study, the elderly and children required the most time. In a study of a patient navigation program in Pennsylvania[18], the two most time consuming tasks for navigators were dealing with financial and transportation problems. In order to effectively allocate staff time and caseloads-especially for patients with multiple barriers to care-a simple tool to predict which patients will be the most time consuming would be beneficial. Specifically, the tool could be beneficial to assess the number and type of navigators needed and assign workloads, based on the cancer patient population’s needs.

The overall goal of this paper is to explore whether the use of a new tool, a simple one-item measure completed by navigators prior to navigation, would be beneficial in assessing caseload mix. The specific objective of this paper is to assess whether navigators are able to accurately estimate cancer patients’ navigation intensity (i.e., relative time needed) prior to navigation. We hypothesized that navigators’ estimates of navigation intensity would predict actual time needed for navigation independent of patient characteristics.

Methods

This study was conducted as part of a larger study, “Randomized Controlled Trial (RCT) of Patient Navigation-Activation,” designed to evaluate the effectiveness of patient navigators on cancer-related healthcare quality and outcomes. In the larger study, patients with newly diagnosed breast or colorectal cancer were randomized to receive either navigation or usual care until completion of their cancer treatment. All patients completed baseline assessments. The study was approved by the University of Rochester Research Subjects Review Board.

Patient Inclusion Criteria

Patient inclusion criteria have been detailed elsewhere [19]. Briefly, newly diagnosed breast or colorectal cancer patients were recruited from oncology and primary care practices across the Rochester, NY community. Once a patient was enrolled in the program, s/he was navigated through the completion of active cancer treatment for a maximum of one calendar year. In the larger study, 166 patients were randomized to receive navigation services for either cancer screening or navigation; in the present report, we had n=139 participants randomized to receive navigation for cancer care. Reasons for the 27 exclusions were unrelated to navigator training and were due to the measure not being completed or incorrectly completed (n=19), or not having cancer (positive screen only, n=8).

Brief Summary of Navigator Intervention and Training

Navigator recruitment and inclusion criteria have been reported, including details of their intensive training and supervision.[2; 20] Consistent with the original model of PN[5], the Rochester Patient Navigation Research Project (PNRP) navigators were lay community health workers (without medical training) that provided mostly face-to-face and sometimes telephone navigation. Navigators (n=5) were male and female, had at least a high school education, relevant knowledge of the community, and previous experience or training in case management and health care. Two of the navigators were proficient in Spanish and English.[2] All navigators were newly trained in their role.

Description of assessments completed by patients and navigators

Patients completed a preliminary assessment of their navigation needs at baseline with their navigators. The assessment consisted of an evaluation of patient barriers to receiving cancer care. Navigators also completed a simple one-item measure (“Navigator Perceived Time of Patient Navigation”) after reviewing patients’ needs assessment but prior to beginning navigation services. Navigators were all new and the assessments were consistently completed according to a structured assessment procedure, for the available sample of participants (n=139).

Navigators completed a simple one-item measure of navigation intensity developed as part of the Rochester PNRP. Navigators estimated how time intensive their patient would be for navigation. This face-valid one-item measure was developed based on expert opinion and team consensus. Establishing that this measure independently predicts navigator time represents a first step in its validation.

Documentation of navigator time spent with patients (encounter logs)

Each time a navigator provided a service, interacted with a patient, or worked on behalf of them in any way, they completed encounter logs documenting the type of activity conducted on behalf of the patient as well as the time it took. The encounter logs also listed twenty-two potential barriers to care and the navigators circled all that applied. The logs also contained a “navigator action” section for the navigator to record the action taken for the specific barrier and the associated amount of time required. All navigated patients, regardless of barriers at baseline, received proactive navigation services. Examples of proactive navigation include coordination of care, transportation assistance, and coaching. All patients randomized to navigation accepted navigational services.

Data Analysis

The dependent variable, time spent in navigation, had a lognormal distribution. Thus, a generalized linear model was used to assess the data. Whereas normal ordinary least squares regression results in arithmetic means, a lognormal regression results in expected geometric means due to the exponentiation of a log-transformed variable to obtain the original units, in this case, to obtain time in minutes from the log of time in minutes. The exponentiated betas resulting from a lognormal regression are a ratio of geometric means and can be interpreted in terms of percent change.[2; 21]

We performed a single and multiple regression with the navigator time intensity rating as the predictor for total time spent from baseline to three months. Our initial model included variables listed in Table 1 as well as patient barriers (both individual barriers and a count of these barriers). The barrier count remained in the model after stepwise selection while the individual barriers did not. Navigator was included in the model as a random effect to account for the variation between the five navigators and we found there were no significant differences due to the navigator. Our final model with navigation time as the outcome (dependent variable) included predicted time intensity (10-point scale), sex (male; female), race/ethnicity (non-Hispanic Black; non-Hispanic White; Hispanic/Other), days of treatment (number of days at baseline into primary cancer treatment – positive if treatment is started and negative if not), barrier count (the number of unique barriers identified at baseline assessment), and distance from clinic (miles between center of patients home zip code and zip code of treatment center).

Table 1.

Elements of the 5A Model and Definition of 5As Used in the Coding of Audio-Recorded Office Visits

Counseling Activity	Description
Ask/assess	Inquire about health behaviors, self-efficacy Did the clinician ask about the patients exercise or physical activity habit in any way? Was there any discussion about initiating, maintaining, or changing physical activity levels in any way?
Advise	Discuss health risks; benefits of change; review appropriate amount, intensity, and frequency of behavior A clear statement that the clinician makes to the patient recommending the patient do regular physical activity. The clinician gives clear, personalized, and specific advice to change physical activity/exercise habits, which may or may not include information about personal benefits to health.
Agree	Collaboratively set physical activity goals based on patient’s interest and confidence The clinician and patient collaboratively select appropriate treatment goals and tasks based on the patient’s interest and willingness to change the behavior
Assist	Identify personal barriers and problem-solving techniques; community opportunities for physical activity and social support The clinician assists the patient to aid them in changing their physical activity/exercise plans by addressing any challenges or barriers that the patient may face. This category also refers to the clinician helping the patient strategize or come up with a plan on how the patient might actually change their exercise level to meet their goals. The clinician might also mention available community resources, programs, or referral options for physical activity/exercise programs in this step.
Arrange	Help the patient complete the plan by providing referrals, reminders, access to resources; specify future arrangements (follow-up visit, call, reminder) to follow-up on progress The clinician and patient explicitly discuss a follow up plan. This means that the doctor schedules a follow up appointment to provide ongoing assistance and support to the patient for helping them change their exercise level. This step may also involve a referral to a program or specialist to help the patient with the exercise program (in this sense may overlap somewhat with the Assist step)

Results

Sample characteristics

Table 1 shows demographic and health characteristics of the participants (e.g., navigated breast and colorectal cancer patients), along with the mean total minutes actually spent in navigation. Most (83.4%) participants had breast cancer and were female (89%). Age was variably distributed with 69% age 50 or greater. Participants were non-Hispanic Caucasian (61%), African American (24%), Hispanic (10%), and other (4%). Educational attainment, insurance status, income, and other socio-demographic variables varied. [Table 1 here].

Baseline barriers to cancer care

Table 2 shows barriers to cancer care reported by participants at their baseline assessment. The most common barriers were insurance difficulties (27%), needing social and practical support (18%), transportation (14%), and financial problems (13%). [Table 2 here].

Table 2.

Participant Sociodemographic and Health Status Information

Patient Participants (n=12)
Age group (years)
18–29	12 (6)
30–39	12 (6)
40–49	27 (13)
50–59	2 (10)
60–69	18 (9)
70+	8 (4)
Not reported	2 (1)
Gender
Female	71 (35)
Male	29 (14)
Race/ethnicity
African-American	51 (25)
White	37 (18)
Other	12 (6)
Employed
Yes	33 (16)
No	35 (17)
Not reported	33 (16)
Annual income
<$12,000	27 (13)
$12,000–$20,000	20 (10)
$21,000–$39,000	10 (5)
$40,000+	8 (4)
Not reported	35 (17)
Education
<12 years	29 (14)
High school diploma	37 (18)
Partial college	18 (9)
College degree	14 (7)
Not reported	2 (1)
Number of Comorbidities
0	31 (15)
1–2	31 (15)
3–4	24 (12)
5–8	14 (7)
Common Comorbidities
Diabetes	27 (13)
Hypertension	47 (23)
Congestive heart failure	8 (4)
Body mass index
18.5 ≥ 24.9 (normal)	6 (3)
25.0 ≥29.9 (overweight)	22 (11)
≥30.0 (obese)	53 (26)
Not reported	18 (9)
Achieving recommended level of physical activity*
Yes	22 (11)
No	78 (38)
Clinician Participants (n=12)
Gender
Female	83 (10)
Male	17 (2)
Clinician specialty
Physician	67 (8)
Physician assistant or nurse practioner	33 (4)
Race/ethnicity
African-American	25 (3)
Asian	8 (1)
White	67 (8)
Years in practice
≤5	33 (4)
≥5 to ≤10	33 (4)
≥10 to ≤15	17 (2)
≥15	17 (2)
Years at health center
≤5	66 (8)
≥5 to ≤10	0 (0)
≥10 to≤15	17 (2)
≥15	17 (2)

Number of barriers to cancer care with time spent in navigation

Table 3 shows the number of barriers for participants at baseline (before navigator intensity rating was done) and the corresponding mean time in minutes spent in navigation. No baseline barriers were identified for 37% of participants. The majority of the remainder had either one (26%) or two (21%) barriers at baseline. As the number of barriers increased, so did the time spent in navigation (134 minutes for a patient with no barriers to a max of 1222 minutes for a patient with seven barriers). [Table 3 here].

Table 3.

Patterns of Observed Use of 5As Among Clinicians

Clinician	Ask/Assess	Advise	Agree	Assist	Arrange
MD-1	[1]			[12]
MD-2	[2] [13] [19]	[11] [13]	[11] [13]
MD-3	[9]
MD-4	[17]	[17]	[17]	[17]
MD-5
MD-6	[14] [15]	[14] [15]		[14]
MD-7	[3]
MD-8	[6]	[6]
NP/PA-1	[18]			[18]
NP/PA-2	[4] [7] [16]	[5] [7] [16]		[16]
NP/PA-3	[8] [10]	[8]	[10]
NP/PA-4

[n] = visit in which an A term was used among all 19 visits in which physical activity was discussed among the 12 participating clinicians. Visit number is shown to distinguish between multiple visits per clinician with A term used. For example, MD-2 used Ask statements on 3 separate visits, coded as visits [2],[13], and [19].

MD, physician participants; NP, nurse practitioner participant; PA, physician assistant participant.

Multivariable models of prediction of time intensity for navigation

Navigators spent 62% more time with females, 34% more time with African Americans, and 26% more time with Others (mainly Hispanics in this sample) than with Caucasians after adjusting for covariates. The total number of barriers, rather than type of barrier, was a better predictor of total navigator time; navigation time spent went up 16% for each additional barrier identified by the navigator. Navigator prediction of time intensity remained a significant predictor in both crude and adjusted models; there was a 29% increase in navigation time for each additional predicted intensity unit on the 10-point scale (p=0.0001).

Discussion

To our knowledge, this is the first published study to examine the ability of lay navigators to predict navigation time for cancer patients. Our main finding was that navigators’ estimate of navigation intensity was an independent predictor of time spent with patients based on an initial needs assessment. We also add to the literature by quantifying the time needed for managing many of the common barriers to cancer treatment that patients face. Other work[22–25] has shown that psychosocial barriers are related to adverse outcomes for cancer care such as delays in follow-up or more advanced disease at time of diagnosis.

We do not know with certainty how navigators estimated time requirements. Discussion with navigators indicates that they based their estimates in part on the number and types of barriers and on their gut sense of patient needs. Navigators’ estimation of time intensity remained independently predictive following adjustment for patient barriers. This finding suggests that navigators’ qualitative assessment of patient barriers capture important information beyond quantitative barrier measures.

We found that in addition to the total number of barriers predicting time spent in navigation, being female, African American, and/or Other racial/ethnic group (mostly Hispanics in this population) were also predictive of time spent in navigation. If females, African Americans, and others such as Hispanics have greater numbers of barriers that are more time intensive to address, this might contribute to disparities in cancer treatment and outcomes, particularly in the absence of intensive navigation.

As noted previously, our navigators’ ratings of time intensity were subjective and, by definition, not blinded. Therefore, it is possible that navigators’ subsequent behaviors were influenced by their own ratings, introducing a conscious decision or unconscious bias to spent more time with a patient they had rated likely to be time intensive. However, it should be noted that the navigators in the course of their day-to-day work completed a large volume of various assessments, tracking forms, logs, and other administrative tasks and documentation. Given this, recall of a single one-item measure may not have figured prominently to them in their overall scope of work, though we cannot be certain of this. Additionally, navigation is a bidirectional process whereby navigators can contact patients, and vice versa. It seems less likely– given that all navigated patients were equally encouraged to contact their navigator, and a navigator would not have direct control over a patient’s decision to call them– that a single navigator intensity rating would account for bias in time spent.

Understanding the time and personnel effort needed to address barriers to improve cancer care are critically important. Yet, such reports are nearly absent from the literature on patient navigation. At an individual (patient) level, it is important for navigators and navigator programs to be able to estimate intensity in order to adjust case mix to best identify patients with the most challenging, time-consuming psychosocial barriers. At a systems or organizational level, estimating time intensity is important in order to identify populations at risk, and establish priorities for targeted systemic intervention. In a 2007 report from the Institute of Medicine on the delivery of psychosocial services to cancer patients and their families, a standard to guide provision of appropriate psychosocial health services for cancer was established. Key among the recommendations was the need for research on tools and strategies to correct maldistributions in the health care workforce and payment systems that impede delivery of psychosocial services. Understanding of time needed to address barriers with patients in cancer navigation is one initial step towards systematically improving the delivery of psychosocial aspects of cancer care.

Limitations

Study limitations should be kept in mind when interpreting these results. Our study population was a single cohort limited to a relatively small sample of breast and colorectal cancer patients enrolled in a PN study who may be different from cancer patients in other settings. We had a relatively small pool of navigators which may have influenced our results. Our face-valid measure of navigator time intensity prediction has not been previously validated. The measure was intended to be brief and focused on time intensity; it did not address other issues such asking navigators to rate the relative importance or salience of barriers to accessing cancer care. Therefore, our findings should be interpreted as preliminary with further research and confirmation needed.

Relevance

Given constrained resources for PN, we need simple strategies and tools to assist navigator programs in assigning caseloads. These findings are relevant because they suggest that navigators may be able to help allocate resources more efficiently for cost effective, high quality cancer care.

Future directions

Future studies should test the effectiveness of using time predictions to more efficiently allocate navigation services resources to patients who would benefit the most. Coordinated approaches to address psychosocial barriers to care should be developed for the most time intensive barriers and evaluated in terms of their effect on cancer outcomes.

Conclusion

Navigators accurately estimated the time intensity of navigation for cancer patients, even when adjusting for multiple patient characteristics and barriers. Findings suggest that navigator estimates of patients’ needs for navigation time could assist in allocation of patients to receive navigation.

Table 4.

Examples of Ask, Advise, Agree and Assist Statements

Statement Type	Examples
Ask	Clinician: “Exercising three times a week … that’s what you had set as a goal last year in March. How is that going? … How much are you thinking you are exercising … once a week? Once a month?” Clinician: “Do you do any sort of exercise? … Anything in particular?”
Advise	Clinician: “You have to do at least half an hour of some kind of motion before it is considered beneficial to you … It does help to control your blood pressure … it helps control your diabetes … helps strengthen your bones, makes you less likely to fall … prevents osteoporosis.” Clinician: “Try to exercise for three or four times a week. What is walking, what is riding a bike for a good half an hour at a time … cause that will also help.” Clinician: “What else is good for you is … getting out and getting to do things that you like to do, getting some exercise …. getting your body going will help you a little bit as well. (in relation to depression) If you do this power walk 5 days a week for a half an hour where you’re sweating, eat a little healthier, more fruits and vegetables and water, less of the bad stuff, you will lose that weight.” Patient: “I’m working on it.”
Agree	Clinician: “You did the water flexibility fitness _class_? I’d love to see you go back.” Patient: “Is it still going on?” Clinician: “Yep, and it’ll continue.” Patient: “What day is it? Mondays and Wednesdays?” Clinician: “Yeah, Mondays and Wednesdays.” Patient: “I’m going back. I don’t care if my blood pressure’s high or not. I just want to go, cuz see Dr. X stopped me from going. I was kinda tired; I was kinda glad I had an excuse not to go. But, you know, I was going – getting up and going, and it was in my schedule. I was going.” Clinician: “I know.”
Assist	Clinician: “I think you would be a good candidate because you are doing well with [local exercise program] and I know when you get excited when you do this. … this weight loss. So I think you’d be a good candidate for this. So can I refer you there?”