When should electronic medical records reminders for cancer screening stop?—Results from a national physician survey

Abstract

Background:

Many older adults are screened for breast and colorectal cancers beyond guideline recommended thresholds. Electronic medical record (EMR) reminders are commonly used to prompt cancer screening. Behavioral economics theory suggests that changing the default settings for these reminders can be effective to reduce over-screening. We examined physician perspectives about acceptable thresholds for stopping EMR cancer screening reminders.

Methods:

In a national survey of 1200 primary care physicians (PCP) and 600 gynecologists randomly selected from the AMA Masterfile, we asked physicians to choose whether EMR reminders for cancer screening should stop based on a list of criteria that included age, life expectancy, specific serious illnesses, and functional limitations. Physicians could choose multiple responses. PCPs were randomized to questions about breast or colorectal cancer screening.

Results:

A total of 592 physicians participated (adjusted response rate 54.1%). 54.6% chose age and 71.8% chose life expectancy as criteria for stopping EMR reminders; only 30.6% chose functional limitations. Regarding age thresholds, 52.4% chose ages ≤75, 42.0% chose a threshold between 75 and 85, 5.6% would not stop reminders even at age 85. Regarding life expectancy thresholds, 32.0% chose ≥10 years, 53.1% chose a threshold between 5 and 9 years, 14.9% would not stop reminders even when life expectancy is <5 years.

Conclusions:

We found that many physicians would continue EMR reminders for cancer screening even in light of older age, limited life expectancy, and functional limitations. This may reflect reluctance to stop cancer screening and/or reluctance to stop EMR reminders so that physicians can retain control to decide for individual patients, for example, to assess patient preference and ability to tolerate treatment. There was consensus for stopping EMR reminders at ages 85+ and <5-year life expectancy. Interventions that seek to reduce over-screening by suppressing EMR reminders may be important for these groups but may have limited physician buy-in outside these thresholds.

INTRODUCTION

In older adults, the short-term harms and burdens of cancer screening may outweigh the long-term potential benefits, which typically lag by 10+ years for breast and colorectal cancers.^1,2 Some guidelines use an upper age threshold and recommend routine screening up to age 74 for breast cancer and age 75 for colorectal cancer.^3,4 Other guidelines have shifted to using limited life expectancy (i.e., <10 years) to guide when to stop screening.^5–9 However, national studies show that many older adults who meet guideline criteria for stopping screening continue to be screened.^10–15

Since screenings for breast and colorectal cancers are key quality measures for healthcare systems, physician practices, and insurance plans, the use of reminders or alerts within the EMR to enhance cancer screening rates is increasingly common.^16–20 Currently, quality measures follow age-based guidelines; correspondingly, EMR reminders typically prompt screening for patients up to age 75 years. Although physicians can override or customize EMR reminders, prior work has shown that the use of EMR reminders has been linked to less deliberate decision-making, where the physicians are less likely to weigh the risks and benefits of cancer screening for specific patients and more likely to simply follow the EMR’s recommendation.²¹ Such practice can lead to both over-screening of older adults younger than 75 years who are in poor health and under-screening of older adults 75+ who are healthy.

When decision-making is more automatic and less deliberate, behavioral economics theory suggest that changing the default can be an effective strategy to change behavior.²² Adjusting cancer screening related EMR reminders to be based on not only age, but also to incorporate other measures of patients’ health and function could be a promising intervention to optimize cancer screening in older adults. Establishing physician buy-in and support is a critical factor for the successful implementation of such interventions. Therefore, we aimed to examine, in a nationally representative sample, physician perspectives about acceptable thresholds for stopping cancer screening related EMR reminders.

METHODS

Study design

In a national cross-sectional survey, we assessed physicians’ perspectives about EMR reminders for breast and colorectal cancer screenings in adults 65 years or older. We used the American Medical Association (AMA) Masterfile, which contains information on all practicing physicians in the United States and are not only limited to AMA members.²³ We surveyed a random sample of 1200 physicians in internal medicine, family medicine, general practice, geriatric medicine (hereafter referred to as “primary care physicians” or PCPs) and 600 gynecologists. The PCPs were randomized to receive questions on breast or colorectal cancer screening. Gynecologists were surveyed to screen for breast cancer.

Physicians were ineligible if they did not care for older adults or did not practice in the outpatient setting, as cancer screening decisions almost always occur in the outpatient setting rather than acute or urgent care. This study was approved by the Johns Hopkins School of Medicine institutional review board.

Data collection

We mailed the surveys, with two follow up mailings to non-responders, between April and November 2021. We also made follow-up phone calls to non-responders. We included an unconditional $20 incentive in the first mailing and a $40 gift card upon the completion of the survey in the last mailing. Physicians can respond to paper-based or online surveys.

Survey instrument

The survey instrument was developed by the study team, pilot tested with 8 primary care physicians at our institution, and revised iteratively. The survey mentioned that many EMRs generate automated reminders about ordering cancer screening, and currently, these reminders often stop at age 75. We stated that physicians can still order screening tests, even in the absence of automated reminders. We then asked: “in your opinion, automated reminders for [breast/colorectal] cancer screening should stop based on which of the following criteria”? The response options included: “age,” “life expectancy (estimated using validated algorithms based on age, sex, health conditions, and functional status),” “functional limitations (e.g., difficulty in self-care)”, and “specific serious illnesses” including “dementia, advanced stages of cancer, advanced stages of congestive heart failure, oxygen-dependent lung disease, end-stage renal disease, and poorly controlled diabetes mellitus.” Physicians could choose more than one response. Physicians could also write in additional criterion. (see Supplement for full survey).

We also asked about specific age and life expectancy thresholds for stopping EMR reminders regarding cancer screening. We asked that “assuming an age-based criterion was used, in your opinion, automated reminders should stop after age _____.” Similarly, we asked “assuming a life-expectancy-based criterion was used, in your opinion, automated reminders should stop when estimated life expectancy is <_____ years.” Physicians were asked to fill in the blank for these two questions.

The AMA Masterfile provided information on physician age, sex, year of graduation, primary specialty, and mailing address. We collected information including race, ethnicity, practice setting, practice size, number of hours per week seeing patients in the clinic, patient panel characteristics, whether the practice used an EMR, and whether the physicians received EMR reminders about cancer screening.

Statistical analysis

Responses were summarized descriptively. In exploratory analyses, we examined whether choosing each criterion for stopping EMR reminders was associated with how often the physicians cared for older adults, whether the physicians currently received EMR cancer screening reminders, and cancer screening type and physician specialty. All statistical analyses were performed using STATA version 17.

RESULTS

Of the 1800 physicians to whom surveys were mailed, 394 did not receive the surveys, including 298 surveys that were undeliverable by postal service, 74 for whom we had incorrect address as confirmed in follow up phone calls, and 22 physicians who were deceased, on medical leave, or retired (Figure S1). Among the remaining 1406 physicians, 761 responded (response rate 54.1%). Compared to responders, non-responders were younger, but did not differ by sex, specialty, or geographic region. (Table S1).

Among the 761 responders, we excluded 114 who did not care for older adults, 30 who did not practice in the outpatient setting, and 25 who did not respond to the primary outcome, leaving 592 in the analytical sample. Slightly over half of the patients were female (50.4%) and two-thirds were white (67.4%), with a mean age of 51.6 years. The most common specialty represented was family medicine/general practice (34.8%), followed by gynecology (32.4%), and internal medicine (31.3%) (Table 1).

TABLE 1.

Participant characteristics (N = 592).

	#(%)
Age, years—mean (range)	51.6 (27–91)
Sex
Female	298 (50.4%)
Male	293 (49.6%)
Race
White	386 (67.4%)
Black	36 (6.3%)
Asian	101 (17.6%)
Other	50 (8.7%)
Geographic regiona
Northeast	127 (21.5%)
Midwest	136 (23.0%)
South	188 (31.8%)
West	140 (23.7%)
Self-reported % of patient panel who are 65+
≤25%	232 (42.5%)
>25% to 50%	183 (33.5%)
>50%	131 (24.0%)
# hours in clinic/week—mean (SD)	31.0 (12.3)
Specialty
Family medicine/general practice	206 (34.8%)
Internal medicine	185 (31.3%)
Geriatric medicine	9 (1.5%)
Gynecology	192 (32.4%)
Practice typeb
Physician-owned practice	184 (32.3%)
Health maintenance organization (HMO)	44 (7.7%)
Medical school or university	74 (13.0%)
Non-government health system	200 (35.2%)
Government	46 (8.1%)
Other	57 (10.0%)
Number of physicians in practice
1	72 (12.7%)
2–10	272 (48.0%)
11–49	132 (23.3%)
50+	91 (16.1%)
Practice uses electronic medical record (EMR)
Yes	542 (95.3%)
No	27 (4.8%)
Practice has EMR-generated alerts or reminders related to cancer screening
Yes	391 (67.3%)
No	190 (32.7%)

^aaGeographic regions were defined based on participant’s mailing address according to the U.S. Census Bureau Region.³⁰

^bbParticipants could choose more than one practice type; therefore, the percentages do not sum to 100%.

Slightly over half (54.6%) of physicians chose patient’s age and 71.8% chose patient’s life expectancy as criteria for stopping EMR cancer screening reminders. In contrast, only 30.6% chose patient’s functional limitations (e.g. difficulty in self-care) as a criterion. Regarding specific serious illnesses, advanced cancer was most often chosen as a criterion (78.0%), followed by advanced congestive heart failure (69.1%). Less than two-thirds (62.5%) chose dementia and about half (50.7%) chose oxygen-dependent lung disease as criteria for stopping EMR cancer screening reminders. (Table 2). Free text responses on other criterion for stopping EMR cancer screening reminders included patient preference and patient’s ability to tolerate test or subsequent cancer treatment.

TABLE 2.

Proportion of physicians who believed that automated reminders for breast/colorectal cancer screening should stop based on each of the following criteria (n = 592).

Criteria	% Choosing each criterion
Age	323 (54.6%)
Life expectancy	425 (71.8%)
Specific serious illnesses
Dementia	370 (62.5%)
Advanced stages of cancer	462 (78.0%)
Advanced stages of congestive heart failure	409 (69.1%)
Oxygen-dependent lung disease	300 (50.7%)
End-stage renal disease	320 (54.1%)
Poorly controlled diabetes mellitus	98 (16.6%)
Functional limitations (e.g., difficulty in self-care)	181 (30.6%)

Assuming age-based criterion was used, fill in the blank responses on acceptable age thresholds ranged from 50 to “never” or “death.” A total of 52.4% of physicians chose an age threshold of 75 or younger for stopping EMR reminders, 42.0% of physicians chose an age threshold between 75 and 85. In total, 94.4% would stop reminders after age 85, 5.6% would not stop EMR reminders even after age 85 (Figure 1).

FIGURE 1.

The age and life expectancy thresholds at which physicians would stop automated EMR reminders about breast or colorectal cancer screening.

Assuming a life expectancy-based criterion was used, fill in the blank responses on acceptable life expectancy thresholds ranged from 0.1 to 15 years. Only 32.0% of physicians chose a threshold of 10 or more years for stopping EMR reminders and 53.1% of physicians chose a life expectancy threshold between 5 and 9 years. Overall, 85.1% would stop EMR reminders when patients have <5-year life expectancy, and 14.9% would not stop EMR reminders even when life expectancy is less than 5 years (Figure 1).

Physicians who had higher proportion of older adults in their patient panels were more likely to choose dementia, advanced cancer, advanced congestive heart failure, oxygen-dependent lung disease, end-stage renal disease, and functional limitations as criteria for stopping EMR reminders (Table S2). Responses did not differ by whether physicians currently receive EMR reminders (Table S3). Compared to PCPs, gynecologists were less likely to choose any of the criteria for stopping EMR reminders, except poorly controlled diabetes mellitus (Table S4). PCPs were more likely to continue EMR reminders for breast cancer than for colorectal cancer screening when patient’s life expectancy was <10 years; there were otherwise no differences by cancer screening type (Table S4).

DISCUSSION

This is the first study to explore, in a large national survey, physician perspectives about acceptable thresholds for stopping cancer screening related EMR reminders. Prior literature have not focused on tailoring EMR reminders to de-implement inappropriate services.^17–20 We only found one pilot study where the intervention involved turning off colorectal cancer screening reminders when patient’s life expectancy was limited but the study did not assess physicians’ perspectives on whether this approach was acceptable.²⁴

We found surprisingly high reluctance to stop EMR reminders, even in scenarios of specific serious illnesses and functional limitations. This may reflect a combination of reluctance to stop cancer screening in general and reluctance to stop the EMR reminders, even if the physician may ultimately decide to stop cancer screening. Factors that contribute to clinician’s reluctance to stop cancer screening in older adults include conflicting/changing guidelines, malpractice concerns, and discomfort with discussing stopping screening.^25,26 Another contributor may be considerations around the advances in cancer treatments that have less toxicities and can enhance older patients’ ability to undergo treatment. As physicians have the option to turn off or opt out of EMR reminders for a patient, they may prefer to retain the ability to make that decision for individual patients. This may be especially true since certain factors that physicians reported as important to the decision, such as patient preference and patient’s ability to tolerate test or subsequent cancer treatment, are not routinely captured in the EMR and need to be assessed and discussed individually.

We found that physicians often chose older age and shorter life expectancy thresholds for stopping EMR cancer screening reminders compared to what is currently recommended in the guidelines for stopping cancer screening (i.e., screening up to age 75 or screening when life expectancy is at least 10 years).^3–9 A prior study showed most physicians considered stopping cancer screening on the basis of <10-year life expectancy to be reasonable.²⁷ Our results suggest that even though physicians may consider it “reasonable” to stop cancer screening when life expectancy is <10 years, they may not prefer to automatically stop EMR reminders based on <10-year life expectancy. This may stem from concerns about the predictive accuracy of life expectancy and the preference for more control to individualize screening decisions together with patients. Tailoring EMR reminders to better support individualized decision-making, rather than simply suppressing EMR reminders, may be a viable strategy to promote appropriate cancer screening in older adults and reduce overuse.

There appeared to be broad consensus for stopping EMR reminders for screening in older adults ages 85 and older and for those with <5-year life expectancy. Reducing over-screening in these groups is important given that national studies showed that 37.5% of women ages 85+ and 34–36% of women with life expectancy <5 years had breast cancer screening in the previous 2 years, suggesting over-screening in these groups.^10,11 EMR reminders usually already stop after age 75; to our knowledge, there is no EMR reminder suppression based on life expectancy. Stopping EMR reminders may be one important tool to help reduce over-screening among patients with <5-year life expectancy.

We found that physicians with a higher proportion of older adults in their patient panels were more likely to discontinue EMR reminders in the context of several serious illnesses and functional limitations, possibly because they may be more familiar with the diminishing benefits and increased risk of screening complications in these patients. We found that gynecologists were less likely than PCPs to choose almost all of the criteria for stopping EMR reminders. This is consistent with prior findings that gynecologists were more likely to recommend cancer screening than PCPs.^28,29

Limitations of the study include non-response bias which may limit the generalizability of our findings. However, we used multiple strategies to improve response rate and achieved a response rate of 54.1%. Stratifying recruitment by physician specialty (PCP vs. gynecology) led to oversampling of gynecologists. Therefore, we conducted subgroup analyses according to the physician’s specialty. Second, relying on self-reporting may lead to social desirability bias. Last, the questions focused on physicians’ opinions on when EMR reminders should stop but did not directly assess the acceptability of specific thresholds. We also did not specify or collect information about colorectal cancer screening modalities.

In summary, we found that many physicians would not want to stop EMR reminders for breast and colorectal cancer screening based on age (75 years) or life expectancy (10 years) thresholds used in current clinical practice guidelines. Rather, there was consensus for stopping EMR reminders only at ages 85+ and <5-year life expectancy. Interventions that seek to reduce over-screening by suppressing EMR reminders may have limited physician buy-in outside these thresholds. Alternative strategies, such as active decision-support tools that recommend physicians to consider stopping screening based on patient health characteristics, may need to be considered.

Supplementary Material

Supplemental Info

Figure S1. Study flow diagram.

Table S1. Responder and non-responder characteristics.

Table S2. Responses stratified by proportion of older patients.

Table S3. Responses stratified by current receipt of EMR cancer screening reminders.

Table S4. Responses stratified by cancer screening type and physician specialty Survey instrument..

Key points

• In a national mailed survey of 1200 primary care physicians and 600 gynecologists, we found that 47.6% would continue electronic medical record (EMR) reminders for breast or colorectal cancer screening beyond age 75.

• We found that 68.0% of physicians would continue EMR reminders for breast or colorectal cancer screening even when patient’s life expectancy is <10 years.

Why does this paper matter?

There was consensus for stopping EMR reminders only at ages 85+ and <5-year life expectancy. Interventions that seek to reduce over-screening by suppressing EMR reminders may have limited physician buy-in outside these thresholds.