Abstract
Highlights
Are screening mammography referrals, given to women younger than 50 y of age, adherent to current guidelines?
In this cross-sectional study of a randomly selected sample of 1,287 referral letters, given to women aged 18 to 50 y, only 10.9% were adherent with the guidelines when examined with a strict approach and 30.6% with a forgiving approach. Adherence significantly correlated with the field of the referring physician.
Despite known risks of screening mammography, women younger than 50 y are commonly referred to such screening in a deviation from current guidelines.
Background
Breast cancer screening by mammography for younger patients is the subject of much controversy. 1 Guidelines from professional associations differ in the age and criteria for initiating screening. For instance, at the time of our study (2019–2020), the United States Preventive Services Task Force (USPSTF) recommended biennial screening for American women at average risk for breast cancer aged 50 to 74 y. 2 However, they have recently updated their guidelines to begin screening at age 40 y. 3 The American College of Obstetricians and Gynecologists (ACOG) recommends annual or biennial screening starting at the age of 40 y, 4 and the National Comprehensive Cancer Network (NCCN), in their recently updated guidelines, state that women should undergo a breast cancer risk assessment starting at age 25 y and annual mammography screening should begin at age 40 y for those at an average risk. 5
At the heart of this controversy is the complex balance of benefits and harms when screening for breast cancer. On one hand, multiple wide-ranging studies have demonstrated a reduction in breast cancer–related morbidity and mortality with increased screening.6,7 On the other hand, there are multiple well-established screening-related risks.8,9 Perhaps the most relevant risk of screening mammography is the possibility of overdiagnosis, the diagnosis of breast cancer that would never have become clinically relevant in the lifespan of a woman.10–12 This can result in unnecessary surgery, exposure to radiation therapy and complementary medication (overtreatment), as well as the ensuing side effects and complications. 1 In a study examining women’s perception of the harms and benefits of screening mammography, only 40% of the participants were aware of the possibility of overdiagnosis, yet after being informed, 60% perceived it as at least slightly important. 12
Other risks include false-positive results, followed by additional testing and the accompanying negative emotional consequences.13–15 In 2 cohort studies, Danish and Norwegian, the women studied following positive screening mammography results who were eventually diagnosed as a false positive had similar levels of anxiety and dejection as women with an actual diagnosis of cancer.16,17
Despite the risks, referral of young women to screening mammography is not uncommon.18–22 In a 2017 study from Johns Hopkins University,
19
a national survey of primary care physicians revealed that most physicians recommended screening initiation at the age of 40 y. This was in line with the ACOG guidelines, which were perceived as the most reliable by physicians surveyed. In a Minnesota-based study, 66% of primary care providers surveyed reported
2019 Guidelines for Screening Mammography Referral for Women Younger than 50 y of Age 18
In an effort to examine the common practice among Israeli physicians, our study focused on actual referral letters for screening mammography given by physicians working under the largest health care provider in Israel, in 2 separate geographical and socioeconomical areas. Our aim was to explore the adherence of these referrals to the recommended 2019 guidelines in relation to different aspects of the screened women and the referring physicians.
Methods
Study Design and Setting
Following approval by the ethics in research committee, we retrieved from the Clalit Health Maintenance Organization (CHMO) database details from all referral letters for screening mammography for patients aged 18 to 49 y in the districts Haifa–West Galilee and Dan–Petach Tikva between March 1, 2019, and February 29, 2020 (prior to the spread of the COVID-19 pandemic in Israel). The data received were retrieved in the form of an Excel file that included information regarding the patient’s age, relevant medical history including comorbidities, free text regarding reason for referral, chief complaint, physical exam and discussion, and specialty of the referring doctor. Further demographic details for either patients or physicians were not included in the database. Identifying details of patients or physicians were not included in the study. Further information retrieved from the database was the number of women insured at CHMO per district and age group at the time of the study.
Referrals were excluded if information contained within or found within the medical file indicated that the referral was given for investigation of symptoms, for the surveillance of known breast findings, and for patients with personal history of breast cancer or known BRCA gene mutations. Referrals given to male patients were also excluded from the study (Figure 1).
Of 9,960 mammography referral letters for women aged 18 to 49 y, 1,287 were randomly selected. After 288 were excluded, 999 remained. Of these, 138 patients were investigated through their medical files.
Participants
Between March 1, 2019, and February 29, 2020, there were approximately 252,000 women aged 18 to 49 y insured at our 2 selected districts. During that period, 9,960 mammography referrals were given (∼4% of women). We analyzed the number of referrals per each age group (Figure 2). Of these, 1,287 referrals were randomly selected using a random selection algorithm. Twenty-nine referrals (2.2%) were excluded because of a personal history of breast cancer, 6 referrals (0.4%) were excluded because of a personal history of BRCA mutation, and 253 referrals (19.6%) were excluded because the referrals were for symptomatic investigation and not screening referrals. A total of 999 referrals (77.6%) were analyzed in the first stage of the research, which involved analyzing only the referrals themselves.
(A) Number of insured women in the examined districts, mammography referrals given to women in the examined districts during the study period, and number of referrals in the random sample, by age group. (B) Accumulative number of referrals in the study, by age of the patient.
Our study relied on information extracted from the referral letters, which may be lacking and an inadequate representation of the full spectrum of considerations of the referring physician. To address this confounder, we further investigated, at the second stage of the study, 13% (138) of referrals classified as nonadherent by examining patients’ medical files to extract additional information. We used this information to compare whether the referral letter adequately represented the medical file. Six of these referrals (4.3%) were excluded because of a personal history of breast cancer, and two referrals (1.4%) were excluded because of a personal history of BRCA mutation revealed in the medical file. A total of 130 referrals (94.2%) were included in the statistical analysis of the second stage of the research that involved the exploration of medical files.
Determining a referral letter, or a conclusion of medical file information, to be adherent or not was done according to the specific criteria presented in the 2019 guidelines (Table 1). If a referral letter met 1 or more of the criteria, it was classified as adherent with the Israeli guidelines for screening referral.
We assumed that a number of the referrals would lack key information that would show a clear adherence to the 2019 guidelines (for example, a referral that states a history of hormone replacement therapy but omits the treatment duration or a referral due to a family history of breast cancer without specifying the degree of relative or age at diagnosis). Therefore, we classified adherence using 2 distinct methods. First, a “forgiving approach,” assigning all referrals with general referral criteria appearing in the 2019 guidelines (Table 1, “Criteria” column) as adherent to 2019 guidelines, even when lacking specific information (Table 1, “Details” column) to match the guidelines. Second, a “strict approach,” by which only referrals that fully demonstrated clear adherence to the Israeli guidelines (Table 1, “Details” column) were deemed adherent to the 2019 guidelines.
When examining the medical file for further details, we analyzed 1) documented visits with the primary care physician in the month prior to the referral for mammography; 2) visits with physicians of the same specialty as the referring physician, gynecologists, and general surgeons 3 mo prior to the referral for screening mammography; 3) prior mammograms performed; 4) current and past diagnoses listed; and 5) current medications listed.
Additional information retrieved from the medical file was used for further assessment of the referred patients’ characteristics of morbidity and medication use. In accordance with Israel’s Ministry of Health definitions, we defined multimorbidity as having 2 or more active chronic diseases in the list of diagnoses. We defined polypharmacy as having 5 or more active medications in the medication list. 23 We classified the type of visit as in-person, telemedicine, or absent the patient (i.e., a referral provided by the doctor without an actual patient visit, usually as a response to an online asynchronous request by the patient). When assigning specialties to referring doctors, we did not differentiate between residents and fully licensed specialists.
We address the potential selection bias by widening our pool of referrals to 2 different districts in Israel. In doing so, we aimed to increase the validity of our results.
For our statistical analysis, we used IBM SPSS version 28.0.0.1. 15 We analyzed the correlation between independent factors (e.g., age, comorbidities, polypharmacy, specialty of the referring physician, etc.) and adherence with the 2019 guidelines as well as a comparison of referral letters to the medical files using t tests and chi-squared tests. Any missing information was excluded from our analysis. P values lower than 0.05 were considered statistically significant.
Results
Age of Referral
Our study investigated referrals of patients aged 18 to 49 y and found referrals for women of every age within our range. When analyzing the data from all referrals (prior to selection of referrals for adherence analysis), we observed that the referral percentage for women increased with age, with only 0.2% referred between ages 18 and 29 y and 18% referred between ages 45 and 49 y (see Figure 2).
Reasons for Referral
Of all referrals that were coded as “screening referrals,” 48% detailed their reason for referral as a screening test, 20% detailed their reason for referral as investigation of symptoms, 8% detailed their reason as follow-up to previous imaging results, and 24% of referrals contained no information.
Of all detailed screening mammography referrals, 67% were given because of family history, 23% because of history of hormone replacement therapy, 9% because of breast density, and 1% were given after a discussion with the patient (Figure 3).
Pie chart indications for screening mammography. (A) Type of referral for mammography. (B) Proportion of each screening indication for referrals with stated reason.
Referring Physician
Of the referring physicians, 45% of referrals were given by general surgeons, 32% were given by family physicians, 10% were given by nonspecialists, 3% were given by internists, 7% were given by gynecologists, 1% were given by oncologists, and 2% were given by other specialists (Figure 4).
Proportion of adherent versus nonadherent referrals for each specialty when examined with the forgiving versus strict approaches.
Adherence to Guidelines
We found that most referrals did not correlate with the 2019 guidelines for screening mammography. Based on our data, only 30.6% of referrals in the forgiving approach and 10.9% in the strict approach were adherent to the guidelines.
We examined the correlation between the specialty of the referring physician and adherence to the 2019 guidelines. We found that the specialty of the referring physician predicted the level of adherence (P = 0.014 forgiving, P = 0.002 strict). It is important to note that we had only 2 referrals given by oncologists in our sample.
In the forgiving approach, the specialists that gave referrals that correlated the least with 2019 guidelines were gynecologists (14.9%). In the strict approach, it was found to be internists (0%) followed by nonspecialists (5.4%).
Age of the Patient and Adherence to Guidelines
In the forgiving approach, we found that a younger patient age predicted greater adherence of the referring doctor to the 2019 guidelines. This correlation was found to be statistically significant with a mean of 42.49 y of age in the group that adhered to the guidelines and 43.2 y of age in the group that did not adhere (P = 0.033). In the strict approach, there was no correlation between the patient age and adherence to the guidelines. In the random sample, 166 of referrals were for women under the age of 40 y, and 839 referrals were for women ages 40 to 49 y. When examining each of these age groups separately, there was no correlation between patients’ age and adherence of referral.
Difference between Districts
To examine the effect of differing demographics on adherence to the guidelines, we compared referrals from the districts Haifa–West Galilee and Dan–Petach Tikva. The former being a more rural region in the periphery of Israel and the latter a mostly urban region in central Israel. Both regions have a heterogenous ethnic population. We found a significant difference in the strict approach between the districts: 39.5% of the referrals in Dan–Petach Tikva adhered to 2019 guidelines versus 20.4% in Haifa–West Galilee District (P < 0.001). No difference was found between the districts in the forgiving approach.
Documentation in the Referral versus the Medical file
In the study’s second stage, we compared the information documented in the referrals to the documentation in the patients’ file to analyze whether or not the referrals appropriately represent the patients’ medical information.
Our analysis found that, based on the strict approach, there was a difference between the 2 information sources. We found that 9% of referrals contained details that showed adherence to the guidelines versus 22% of medical files examined that contained information that justified a screening exam (P < 0.001). However, based on the forgiving approach, no difference was found (34% of referral letters adhered to the guidelines versus 33% of the medical files).
When accessing the medical files, we also looked at additional factors present that could help predict adherence to the 2019 guidelines. We found no correlation between adherence to the guidelines and comorbidities (P value of 0.43 for the strict approach and 0.76 for the forgiving approach), polypharmacy (P value of 0.28 for the strict approach and 0.48 for the forgiving approach), or type of visit (in-person v. telemedicine v. absent the patient) (P value of 0.17 for the strict approach and 0.33 for the forgiving approach).
Discussion
In this study, we aimed to examine the scope of screening mammography referrals among women younger than 50 y and the adherence of these referrals with the 2019 Israeli guidelines. Our results demonstrated that a substantial percentage of the female population insured in the 2 examined districts were given screening mammography referrals. Notably, 14.5% of the women aged 40 to 45 y and 18% aged 45 to 49 y were referred. Our samples of 1,000 referral letters and 132 medical files revealed upon thorough examination that a minority of referrals adhered to the 2019 guidelines.
Our research found referrals for women of every age within our examined range, including women as young as 18 y. It is concerning that women at these young ages were referred for screening, as even the more inclusive guidelines, such as the ACOG and NCCN, do not recommend screening before the age of 40 y.4,5 Moreover, mammography is not the recommended modality for breast assessment at these ages.24,25 It is important to note that the number of these referrals is most likely underestimated in our study, as most women in Israel between the ages of 18 and 20 y are cared for by the medical services in the Israel Defense Forces and are not represented in our data.
On the other end of our age spectrum, we found that the number of referrals increased with patient age, and most referrals were given to women older than 40 y. This result is expected as the risk of breast malignancy increases with age. While statistically significant, the difference in the average age of the patient referred based on the guidelines versus not according to the guidelines was minor (42.5 v. 43.2 y).
Other patient characteristics (e.g., comorbidities, polypharmacy, and type of visit) were also examined to determine if these factors could predict the adherence of their referrals. We found that these factors did not predict adherence, which suggests that neither a patient’s exposure to the health care system nor health care–seeking behavior affect adherence.
To minimize selection bias and increase the validity of results, we chose to examine 2 distinct socioeconomic and geographic districts. Our data suggest a significant difference in the strict approach between the 2 districts. In Dan–Petach Tikva, a mostly urban district located in the center of Israel, the referrals were more adherent than in the more rural and peripheral Haifa–West Galilee District. However, this difference disappears when using the forgiving approach. These findings could point to less informed physicians in the latter district, yet an equally valid option is a difference in the level of documentation.
Such incomplete documentation in the referral letter is, in itself, an issue worthy of examination and in particular when referring to a test with a delicate harm–benefit balance such as screening mammography of young women. The issue is further stressed when considering the fact that 25% of all referral letters were left blank and could prove a worrisome finding if it reflects the perceived importance of the referral for the referring physician.
Another noteworthy finding is the low number of referrals stating patient–physician joint decision as the reason for referral (1% of all referrals). The 2019 guidelines make clear that, for women 45 y and older with no specific indication for early screening mammography, a discussion of the risks and benefits is necessary prior to issuing the referral letter. If such discussions were regularly conducted, they were not documented in the referral letters nor in the medical file. This finding is particularly worrisome when considering that many women would prefer the decision to undergo mammography screening be a shared one between woman and physician. 26
When examining the specialty of the referring physician, the specialists with a greater proportion of adherent referrals were general surgeons and family physicians, yet nonadherent referrals were the rule for all disciplines.
These results are in line with other studies performed internationally, such as the 2017 study by Radhakrishnan et al. 19 that surveyed physicians’ recommendations for screening mammography. More than 80% of primary care physicians surveyed reported recommending screening mammography to women younger than 50 y. Several explanations were suggested for these findings, including the possible influence of the United States’ payment system rewarding referrals for testing regardless of their appropriateness. 22 Such influence was not applicable in our study, as the local health care system does not reward physicians for referrals. Other explanations, revolving around the patient–physician relationship, physicians’ perceptions, and influence of the media might very well apply. Several studies examining physicians’ perceptions found that the benefit from an intervention is often exaggerated relative to the harm, sometimes owing to a confusion between relative risk and absolute risk or overestimation of the test’s positive predictive value for an asymptomatic patient.25,27
It is important to note that although at the time our study was conducted, the USPSTF recommended screening initiation at the age of 50 y, 2 a policy that undoubtedly influenced the 2019 Israeli guidelines, a new recommendation was recently published with the lower starting age of 40 y. 3 Such a change sheds new light on physicians’ apparent preference for screening initiation at this age, as demonstrated in our study. Yet the gap between national, multidisciplinary guidelines and common practice remains a significant issue, raising questions about how guidelines are developed and disseminated.
Our study has several noteworthy limitations. First, our information was extracted from medical records using specific codes used only for screening mammography and perhaps missed relevant referrals if they were inappropriately coded. Moreover, there were cases of referrals using screening mammogram codes that were not for screening mammograms, and it is possible that some of the blank referrals included in our study were given for indications that may have met the exclusion criteria. In addition, our study did not examine a crucial element of guideline adherence: the proportion of women younger than 50 y who met the criteria for breast cancer screening but were not referred.
Despite including 2 of the largest districts in Israel, our study was still limited to these specific populations, which may not be generalizable to the whole of Israel. Similarly, the health care system used in Israel may not be generalizable to other health care systems globally.
Finally, as mentioned above, there is an inherent bias in using referral letters and medical files as representations of the actual medical encounter and the process of physicians’ decision making. Patients’ requests and worries, as well as physicians’ experience and values, are without a doubt underrepresented in these documents. We hope to further explore this issue in our ongoing qualitative study, interviewing physicians with different levels of referrals’ adherence to guidelines, identified through our quantitative study, about their perspectives and attitudes.
In conclusion, our study indicates that despite the Israeli Ministry of Health’s efforts to establish unified breast cancer screening guidelines based on the latest evidence of benefits and harms, many physicians from various disciplines still deviate from these guidelines for reasons that remain to be investigated.
Supplemental Material
sj-docx-1-mpp-10.1177_23814683251317524 – Supplemental material for Screening Mammography for Young Women in Israel: Between Guidelines and Common Practice
Supplemental material, sj-docx-1-mpp-10.1177_23814683251317524 for Screening Mammography for Young Women in Israel: Between Guidelines and Common Practice by Neta Essar Schvartz, Michal Rotem-Green, Dikla Kruger, Anat Gaver, Inbar Safra, Danielle Mira Harari, Nadav Niego and Mordechai Alperin in MDM Policy & Practice
Footnotes
Acknowledgements
We thank Dr Linoy Segal for statistical analysis consultation and Dr Ben Renkosinski for proofreading.
Authors’ Note
The work was presented at the annual “Shabtai” conference for family physicians in Israel (July 2022).
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article. The authors received no financial support for the research, authorship, and/or publication of this article.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
