Discontinuing Cancer Screening for Older Adults: a Comparison of Clinician Decision-Making for Breast, Colorectal, and Prostate Cancer Screenings

Abstract

Background

While guidelines recommend against routine screening for breast, prostate, and colorectal cancers in older adults (65+ years) with <10-year life expectancy, many of these patients continue to be screened. How clinicians consider screening cessation across multiple cancer screening types is unknown.

Objective

To compare and contrast clinicians’ perspectives on discontinuing breast, prostate, and colorectal cancer screenings in older adults.

Design

Qualitative, semi-structured interviews.

Participants

Primary care clinicians in Maryland (N=30)

Approach

We conducted semi-structured interviews with individual clinicians. Interviews were recorded, transcribed, and analyzed using standard techniques of qualitative content analysis to identify major themes.

Key Results

Participants were mostly physicians (24/30) and women (16/30). Four major themes highlighted differences in decision-making across cancer screenings: (1) Clinicians reported more often screening beyond guideline-recommended ages for breast and prostate cancers than colorectal cancer; (2) clinicians had different priorities when considering the benefits/harms of each screening; for example, some prioritized continuing colorectal cancer screening due to the test’s high efficacy while others prioritized stopping colorectal cancer screening due to high procedural risk; some prioritized continuing prostate cancer screening due to poor outcomes from advanced prostate cancer while others prioritized stopping prostate cancer screening due to high false positive test rates and harms from downstream tests; (3) clinicians discussed harms of prostate and colorectal cancer screening more readily than for breast cancer screening; (4) clinicians perceived more involvement with gastroenterologists in colonoscopy decisions and less involvement from specialists for prostate and breast cancer screening.

Conclusions

Our results highlight the need for more explicit guidance on how to weigh competing considerations in cancer screening (such as test accuracy versus ease of cancer treatment after detection). Recognizing the complexity of the benefit/harms analysis as clinicians consider multiple cancer screenings, future decision support tools, and clinician education materials can specifically address the competing considerations.

INTRODUCTION

A substantial proportion of older US adults who meet guideline criteria for discontinuing cancer screening continue to receive screenings for breast^1–3, colorectal^1,2,4, and prostate^2,5,6 cancers. For older patients with limited life expectancies, cancer screening may provide little to no benefits while at the same time pose significant risks and burdens. One meta-analysis estimates that it takes 10.7 years for mammography and 10.3 years for fecal occult blood tests to prevent one death from breast and colorectal cancers respectively among 1000 screened individuals⁷. Similarly, the lag-time to benefit for prostate cancer screening with prostate-specific antigen (PSA) is estimated to be around 11.2 years⁸. In contrast to the long-delayed benefits from screening, harms involved in screening are often immediate; these include complications from screening tests, cancer overdiagnosis—detecting cancers that would have never become clinically significant in a patient’s life time—further downstream testing, adverse effects of cancer treatments, and cumulative psychological, emotional, and financial burden to patients^9,10.

Older patients consider stopping cancer screening to be a major decision^11,12. These patients report that recommendations from trusted clinicians are an important factor in making a decision to stop screening¹³. On the other hand, clinician recommendations in favor of screening, when compared to no clinician recommendation or a clinician recommendation against screening, have been associated with higher screening rates for both PSAs and mammograms for adults over 75 years^14–16.

In the primary care setting, clinicians are tasked with managing multiple preventative care decisions, including screening for more than one cancer. Current screening guidelines differ in their approaches as to when screening should cease in older adults. Some guidelines use the same threshold across cancer screening types. For example, the American College of Physicians (ACP) and the American Cancer Society (ACS) use the same life expectancy threshold of less than 10 years to guide cessation of routine breast, colorectal, and prostate cancer screenings^17–22. Others, such as from the United States Preventive Services Task Force (USPSTF), use different age thresholds to guide different types of cancer screenings (i.e., stop breast and colorectal cancer screening at age 75 and prostate cancer screening at age 70) ^23–25.

Previous work has broadly examined clinician attitudes towards cancer screening decisions^26–28 or have focused on only a single cancer type^29–31. However, little is known about clinicians’ decision-making across screenings for different types of cancers. This qualitative study seeks to compare and contrast primary care clinicians’ decision-making, communication, and interaction with specialists across breast, prostate, and colorectal cancer screenings when caring for older adults with limited life expectancies. Better understanding these factors, and how clinicians think uniquely or similarly about these three cancer screenings, will critically inform ongoing efforts to improve cancer screening in older adults.

METHODS

This was part of a larger qualitative study using semi-structured, in-depth interviews with primary care clinicians to examine clinicians’ cancer screening recommendations and views on over-screening in older adults²⁷. For this paper, we focused on comparing and contrasting clinician decision-making and communication across the three cancer screening types.

We used a combination of snowball sampling and maximum variation sampling to recruit clinicians—physicians, nurse practitioners, and physician assistants—who provided primary care for older adults at different clinics in Maryland. We aimed to recruit clinicians diverse in age, sex, clinician type, specialty, and practice type. We recruited via email from three geriatric clinical programs (an ambulatory clinic, a house-call program for homebound older adults, and a Program for All-inclusive Care of the Elderly), three academic primary care clinics, and a large community group practice with 20 primary care clinic sites, all of which were affiliated with the same academic health system. We also recruited from 16 private practices in Maryland not affiliated with the academic health system.

For the clinicians affiliated with the academic health system, we had access to medical records and reviewed each clinician’s medical records to identify 2–3 patients of at least 66 years of age and with a <10-year life expectancy (as calculated according to a prediction algorithm which uses a patient’s age, sex, and comorbidities³²). We identified patients with and without screening in the previous year for breast, colorectal, or prostate cancers. For clinicians for whom we did not have access to the medical records, we asked each clinician to think of specific older patients who had <10-year life expectancy and had cancer screening in the previous year or recently decided against screening. We asked each clinician about their cancer screening decision-making in these specific patients; then we also asked about their approaches to cancer screening in general in their older patients.

The interview guide (Appendix) was piloted with two general internal medicine faculty members within our institution to ensure clarity and appropriateness. One investigator (NS) with prior qualitative research experience, who was a physician in the same academic health system and was acquainted with some of the participants, conducted and audio-recorded face-to-face clinician interviews (October 2018–May 2019). Investigators continuously reviewed the transcripts to assess for new ideas and themes, and data collection ended when the study reached theme saturation. The interview audio-recordings were transcribed verbatim and standard techniques of conventional qualitative content analysis were used to code the transcripts using textual data analysis software (Atlas.ti)^33–35. Investigators (J. E. and N. S.) generated a preliminary coding scheme based on the interview guide and initial review of 3 transcripts. Investigators employed the constant comparative approach to iteratively review, reassess, and reapply the coding scheme to analyze the data as new themes were identified^35,36. Each transcript was coded independently by two of the investigators (J. E. and N. S.) and inter-coder differences were reconciled by consensus. This project was approved by a Johns Hopkins School of Medicine institutional review board.

RESULTS

Thirty clinicians participated in the study. They were mostly physicians (24/30) and split between women (16/30) and men (Table 1). They came from 21 different clinic sites in Maryland and specialties included internal medicine (17), family medicine (6), geriatrics (5), and medicine/pediatrics (2). Interviews averaged 43 min in length (ranging 24–64 min). We identified four major themes. These themes are presented below and illustrated using representative quotes.

Table 1.

Characteristics of Primary Care Clinician Participants

Characteristics	No. (%) or mean (SD)
Age, years	48.2 (10.0)
Female sex	16 (53%)
Race
White	18 (60%)
African American	6 (20%)
Other	6 (20%)
Degree
Physician (MD/DO)	24 (80%)
Certified registered nurse practitioner	5 (17%)
Physician's assistant	1 (3%)
Years since completing training	17.5 (10.2)
Specialty
Internal medicine	17 (57%)
Family medicine	6 (20%)
Medicine/pediatrics	2 (7%)
Geriatrics	5 (17%)
Clinic site
Urban	13 (43%)
Suburban	17 (57%)
Clinic type
Clinics affiliated with academic university	8 (27%)
Clinics within a large academically-affiliated group practice	14 (47%)
Private practice clinics	5 (17%)
House-call program for homebound patients	1 (3%)
Program for all-inclusive care of the elderly	2 (7%)
Proportion of patients ≥65 years old in patient panel
<25%	7 (23%)
25 to 49%	13 (43%)
50 to 74%	4 (13%)
>75%	6 (20%)

Theme 1: Clinicians Reported More Often Screening Beyond Guideline-Recommended Ages for Breast and Prostate Cancers Than Colorectal Cancer

Clinicians reported patient age to be an important factor when deciding about screening cessation but described different approaches on how they incorporated age to inform decisions when dealing with different types of cancer screenings. Some clinicians reported using the same age cutoff for stopping screening across all three screening types. As said by one clinician: “I’ll just do 75 for everything” (participant 6, age 35, female, D.O.). Other clinicians had different ages for ceasing screening for each cancer. One clinician said: “I usually stop screening for colon cancer at 80…I usually do mammograms until about 85” (participant 3, age 41, female, M.D.).

Specifically, clinicians reported more often screening beyond guideline-recommended ages for breast and prostate cancers than for colorectal cancer. For example, one clinician described that age was less important for breast cancer screening: “Mammography is a little different…I will do [mammograms] at any age because the risks are so minimal” (participant 16, age 59, male, M.D.). Another clinician mentioned continuing to screen for prostate cancer with no stopping age “because it’s just bloodwork” (participant 2, age 40, female, C.R.N.P.). In contrast, in the case of colonoscopies, clinicians were more likely to report using a strict stopping age; one clinician said: “I guess I have a tighter [age threshold] for colonoscopy…Generally I don’t think of it for anyone after the age of 70” (participant 29, age 36, female, D.O.). Another clinician said: “I think the recommended age to stop doing just routine screening colonoscopies is about 75 so I stick with that” (participant 1, age 40, female, M.D.).

Theme 2: Clinicians Had Different Priorities When Considering the Benefits and Harms of Each Screening

We found that clinicians weighed the benefits and harms of screening depending on how they considered the characteristics of the screening modality and the underlying cancer (Table 2).

Table 2.

Clinician Priorities When Weighing the Harms and Benefits of Different Types of Cancer Screenings

Priority	Specific decision mentioned by participants	Example quote
Efficacy of screening test	Continue colonoscopy; discontinue mammogram	“Mammography for screening is not a great technology versus the standard with colonoscopy, and colon cancer screening is much better, so I tend to push with that a little bit a little more” (participant 10, age 32, female, M.D.).
Direct harm of screening test	Continue mammogram; discontinue colonoscopy	“When you think about mammograms for cancer screening, that’s less invasive than … a colonoscopy. You don’t have the same type of surgical risk or other risks of pain or discomfort. With a mammogram, it’s uncomfortable but they’re just pulling on the skin and getting images it’s not a lot of radiation either, so it’s less invasive. I’m less picky about that” (participant 1, age 40, female, M.D.).
Downstream harm of screening	Discontinue PSA	“The PSA by itself has a low risk but … if you have an abnormal PSA then all the urologists around here are gonna recommend a biopsy and that’s not such a benign procedure anymore” (participant 16, age 59, male, M.D.).
Harm of unscreened, advanced cancer	Continue PSA	“You’d have to factor in if they had the most aggressive kind of prostate cancer versus the least aggressive kind: would the treatment be different? If they had 8 years to live and the person would die in 3 years from prostate cancer metastatic with pain and all that and you could cure them now, would you?” (participant 21, age 60, male, D.O.).
Ease of cancer treatment	Continue mammogram	“Breast is the easiest to treat because it’s a very superficial organ and you don’t have to muck with anything else essential especially when the tumor is small” (participant 14, age 55, male, M.D.).

Some clinicians focused on the accuracy and efficacy of the screening test and felt more strongly about continuing colonoscopy in their older patients, which they felt to be a more reliable way of finding and preventing advanced cancer, than PSA or mammogram. Others prioritized minimizing the direct harm of screening tests and therefore would forego colonoscopies in a patient while continuing PSA or mammogram. In addition to direct harms from the screening test, clinicians also reported considering downstream harms in their screening decisions even when the upfront harm is low such as for PSA test.

Other clinicians focused on the negative consequences of late-stage cancers when prioritizing different screenings. One clinician justified decision to continue colonoscopies: “In terms of colon cancer screening, [colon cancer] can be a rapidly moving cancer from what I understand so … I feel like he should still be checked for it because we can still do something” (participant 18, age 43, female, C.R.N.P.). Finally, some clinicians prioritized screening for cancers that are easier to treat. One clinician reflected that they would continue breast cancer screening in a patient while stopping other screenings because cancers detected by mammogram require treatments that are less risky and invasive than those used to treat other cancers.

Theme 3: Clinicians Discussed Harms of Prostate and Colorectal Cancer Screening More Readily with Their Patients than Harms of Breast Cancer Screening

Several clinicians used the same techniques across cancer screenings when counseling their patients whether to continue or discontinue screening. As said by one clinician: “I think it’s much more of a discussion with almost an equal thought process” (participant 13, age 58, male, M.D.). Another reflected that they engage in similar processes in providing patients with the pros and cons for each cancer screening type: “So I might say ‘Do you still have any interest in having a mammogram or being checked for colon or prostate cancer?’ and if they say yes, I’ll maybe tell ‘em the pros and cons. If they say no, I’ll say I think that’s reasonable” (participant 8, age 59, male, M.D.).

However, other clinicians reported counseling their patients differently for each type of screening. Some clinicians described spending more time counseling patients on screening harms for PSA compared with colonoscopies because they perceived that the harms of PSA are less obvious to their patients. One clinician reported “I feel like [for] people who have gone through [colonoscopy], it’s more obvious what the risks are because it’s more clearly a procedure versus PSA, if you’re gonna do blood work anyway, they don’t really feel different. I found that sometimes it took extra counseling on their actual harms because I don’t know that people perceive the harms very much” (participant 11, age 32, male, M.D.).

Some clinicians mentioned often discussing the harms of false positive and downstream testing for PSA testing but not when counseling about mammograms: “With the PSA, I feel like with that conversation I’m more able to tell men because … I’ve read the numbers…that the impotence and the incontinence is much more likely than a prevention … but I don’t have that at my fingertips for mammography” (participant 4, age 45, female, M.D.). Several clinicians commented on being less aware of screening harms, especially for mammograms, which then leads to less discussion of screening harms with patients. One clinician said: “I guess I do a poor job in breast cancer screening for discussing the potential risks …I feel like I like personally I understand prostate cancer and the potential risks better than for mammography and [its] potential risks” (participant 11, age 32, male, M.D.). Another clinician said: “I don’t know if I have a clear conversation about what the harms of a mammogram should be. I certainly lean towards the benefit side than harms” (participant 15, age 41, female, M.D.).

Theme 4: Clinicians Perceived More Involvement with Gastroenterologists in Colonoscopy Decisions and Less Involvement from Specialists for Prostate and Breast Cancer Screening

Clinicians differed in how they perceived the specialist’s roles in different types of cancer screening and these differences impacted how clinicians considered and valued their interactions with specialists.

Because gastroenterologists were responsible for performing the colonoscopy procedures, clinicians discussed that their interactions were different compared to interactions with urologists for prostate cancer screening or gynecologists for breast cancer screening. Specifically, gastroenterologists could disagree with the primary care clinicians’ referrals and decline to perform the colonoscopy. One clinician described this difference between specialists: “I’ve never had pushback from any other specialty on cancer screening except for GI” (participant 18, age 43, female, C.R.N.P.). Relatedly, clinicians reported pre-emptively predicting whether gastroenterologists would perform a screening colonoscopy on a referred patient, therefore essentially deferring the decision to gastroenterologists. One primary care clinician said: “Some GIs won’t do it past 75, some GIs won’t do it past 80, and actually it’s interesting ‘cause in some ways colonoscopies are an easier discussion ‘cause it’s not me who’s making the decision” (participant 16, age 59, male, M.D.).

In breast and prostate cancer screenings, clinicians were less likely to view specialists as being integrally involved. As such, many clinicians instead reported that local factors influenced their decision on whether or not to include specialists in screening decisions. Some clinicians commented on specialists who would frequently order screening: “I think OB/Gyn is another sort of big source of continued mammogram screening”(participant 11, age 32, male, M.D.). Another clinician, when discussing not ordering PSAs for their patients, said: “I try to keep people away from the urologist… if you don’t want something done don’t send the patient to a specialist ‘cause it’ll happen” (participant 25, age 65, male, M.D.). For other participants, specialists were not as involved in shared patient care. One conveyed this sentiment about gynecologists, “I would say gyn is not involved at all [in older women]” (participant 28, age 61, male, M.D.).

DISCUSSION

This study explored the similarities and differences in clinicians’ decision-making around screening for breast, colorectal, and prostate cancers in older adults. While primary care clinicians reported considering patient age, the benefits and harms of screening, and specialist input across all three screenings, the clinicians reported differences on how these factors influenced the screening decision depending on the cancer screening type. Our findings add to the growing body of research on clinician factors that influence cancer screening discontinuation by providing a better understanding of the different priorities that clinicians employ when considering the different screenings in older adults.

We found that some clinicians used different age cutoffs when discontinuing different types of cancer screenings. Clinicians were generally more willing to screen into older ages with mammography and PSA than colonoscopy. One reason that was mentioned was that mammography and PSA have lower direct risks. A similar tendency to continue less invasive screening tests has been previously reported; it was found that clinicians at times switched to less invasive screening options or used physical exam for screening, rather than stopping screening together.²⁶ Prior studies have described that another reason for clinicians screening beyond recommended age cutoffs is if patients are healthy and have long life expectancies.^37,38 We did not specifically explore if life expectancy consideration motivated screening beyond guideline-recommended age cutoffs in this study but would not expect life expectancy considerations to explain the different decisions by cancer type since cancer screening guidelines that use life expectancy mention a 10-year life expectancy threshold for all three types of screenings.^17–22

Although less invasive screening tests such as mammography and PSA have less direct risks, they can nonetheless lead to unnecessary downstream tests and treatment that are more invasive and result in net harm. This is especially important as we also found that clinicians were not always aware or well-equipped to discuss with patients the downstream harms of cancer screening, particularly for mammography. Given that the literature has consistently shown that patients often overestimate the benefit of cancer screening and under-estimate the harms³⁹, the lack of awareness or discussion about screening harms on the part of the clinicians can contribute to over-screening in older adults. It was interesting that some clinicians reported being aware and often making a point to discuss downstream harms of PSA testing but less so for mammography. The reason for this difference is not clear but it encouragingly demonstrated that clinicians did consider and weigh downstream harms of non-invasive screening tests and can incorporate the discussion of such in their clinical practice. Better training clinicians to recognize and discuss the downstream harms of cancer screening, such as false positives, biopsies, and overdiagnosis, can be an important next step to improve cancer screening practices.

We also found that clinicians had different priorities when weighing the benefits and harms of cancer screening. Although the individual factors or considerations, such as the direct harm of the screening test, have been previously reported as important in clinicians’ cancer screening decisions,^26,31 this is the first study to show how clinicians may prioritize one factor over another when considering multiple screenings and, more importantly, how these difference in priorities can lead to different decisions. Although some level of variability in clinical practice is expected, it is somewhat concerning that this difference in priorities when weighing the screening benefits and harms at times led clinicians to reach completely opposite conclusions despite having access to the same evidence and clinical practice guidelines. These differences in priorities when weighing the screening benefits and harms may contribute to the observed variation in cancer screening practices and to over-screening.

Our finding has several implications. First, empiric evidence is lacking on how to best consider competing factors such as high efficacy of a screening test versus high procedural risk versus the ease of treatment if a cancer were detected in these decisions. More explicit guidance on how to weigh the various aspects of the screening decision and/or more specific evidence (such as differences in patient quality of life between early-stage treatment versus late-stage treatment for a specific type of cancer) may be needed to help clinicians systematically approach these complex factors. Second, it is not known how patients prioritize these screening benefits and harms and explicitly examining how well patients’ and clinicians’ priorities match can shed lights on ways to improve shared decision-making in this area. Third, recognizing the complexity of the benefit/harms analysis for clinicians as they consider multiple cancer screenings, future cancer screening decision support tools, and clinician education materials can specifically address the competing considerations.

Currently, guidelines for breast, colorectal, and prostate cancer screenings measure benefits mainly in terms of cancer-specific mortality ^{17,19–22,40,41}. Previous literature has reported, from the patients’ perspective, other outcomes that are important in cancer screening decision-making such as reassurance and the value of information.^42,43 Our results add to this literature by describing from the clinicians’ perspective the outcomes that they consider important in cancer screening but are not routinely captured in screening trials or included in guidelines, such as the ease of treatment for cancers detected at earlier stages and the quality-of-life impact from treatment of later stage cancers. Better measuring and incorporating these outcomes in future cancer screening research are key to enhancing informed decision-making and evidence-based practices in cancer screening.

Lastly, adding to previous literature that specialists’ influences are important in cancer screening decisions^26,44,45, we found that primary care clinicians considered gastroenterologists more integrally involved than gynecologists or urologists since they relied on gastroenterologists to complete the colonoscopies. This finding, although not surprising, provides support for interventions aimed to reduce over-screening of colorectal cancer to target both primary care clinicians and gastroenterologists whereas this may be less critical for the other two types of cancer screenings.

This study has several limitations. Although we included participants who were diverse in age, specialty, and practice setting, the study included a limited number of participants who were all from Maryland. The views of these participants may not be representative of primary care clinicians who practice elsewhere in different settings. Also due to the limited sample size, we were not able to analyze the results in subgroups by participant characteristics. The study design relied on self-report which may be subject to recall bias and social desirability bias, especially as the investigator who conducted the interviews was acquainted with some of the participants. There was only one interviewer which may also have impacted data collection. Since the study focus was to explore whether clinicians’ cancer screening decision-making was similar or different across cancer types and to characterize these similarities/differences, we did not comprehensively explore the rationales behind all the reported differences, which could be a topic for future studies.

In conclusion, clinicians often considered screening for distinct types of cancers differently and their evaluation and communication of screening-related benefits and harms varied by cancer type. Interventions that aim to reduce over-screening and optimize cancer screening in older adults need to address these complexity and nuances in clinicians’ cancer screening considerations. Specifically, better training clinicians to recognize and discuss the downstream harms of cancer screening, especially mammography, is needed. In addition, empiric evidence and more explicit guidance on how to weigh competing considerations in cancer screening (such as test accuracy versus ease of cancer treatment after detection) may help clinicians to more systematically approach these complex factors.

Funding

This study was supported from the NIA#T35AG026758 Medical Student Training in Aging Research (MSTAR) Summer Program. Dr. Schoenborn was funded by the K76AG059984 grant from the National Institute on Aging. In addition, Dr. Boyd was supported by 1K24AG056578 from the National Institute on Aging.

Declarations

Conflict of Interest

The authors have no conflict. Dr. Pollack has stock ownership in Gilead Sciences, Inc. However, we do not believe this has resulted in any conflict with the design, methodology, or results presented in this manuscript.

Disclaimer

The funding sources had no role in the design, methods, subject recruitment, data collections, analysis, and preparation of paper.