If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Reprint requests: Esther H. Chung, M.D., Stanford Fertility and Reproductive Health Services, Department of Obstetrics and Gynecology, Stanford University, Sunnyvale, 1195 W. Fremont Ave, Sunnyvale, California 94087.
To assess recall bias by evaluating how well female cancer survivors remember details regarding their cancer diagnosis, treatment, and fertility preservation (FP) counseling.
Oncofertility literature cites recall bias as a pitfall of retrospective surveys, but limited data exist to quantify this bias.
Design
Retrospective secondary analysis of cross-sectional survey data.
Setting
Single academic medical center.
Patient(s)
Female oncology patients of reproductive age, 18–44 years old, at least 6 months past their last chemotherapy treatment.
Intervention(s)
Not applicable.
Main Outcome Measure(s)
Recall of details surrounding cancer diagnosis and chemotherapy regimens, recall of FP counseling and ovarian reserve testing, and rates of chart-documented FP counseling.
Result(s)
In total, 117 patients completed the survey, with 112 verified via chart review. When asked to report the chemotherapy regimen, 57% (64 of the 112) marked “I don’t know/prefer not to say.” Regarding FP, 80% (90 of the 112) denied being offered counseling. Of the 37 (33%) who had documented FP conversations, 13 (35%) did not recall mention of fertility. Only 2 of 8 patients with ovarian reserve testing recalled this being performed at their initial visit. Multivariable logistic regression revealed older age was significantly associated with not being offered FP (odds ratio [OR] 0.87).
Conclusion(s)
Our results confirm that the accuracy of oncology patients’ reporting is limited by a poor recall, particularly regarding their specific chemotherapy regimen. More than 1 in 3 patients documented to have been offered FP counseling do not recall this discussion. Importantly, only one-third of cancer survivors had chart-documented FP counseling. Increased efforts are needed to ensure adequate follow-up beyond the initial visit.
Ethics committee of the American Society for Reproductive Medicine Fertility preservation and reproduction in patients facing gonadotoxic therapies: an Ethics Committee opinion.
). Despite guidelines by the American Society of Clinical Oncology, the American Society for Reproductive Medicine, and the Oncofertility Consortium, multiple studies continue to demonstrate inconsistent referrals to reproductive endocrinology and infertility specialists (REIs) by oncologists (
), even at large academic institutions. This indicates that many patients are initiating gonadotoxic treatment without discussing fertility preservation (FP) options first (or without recalling this conversation). Much of the existing oncofertility literature has thus focused on surveying patients and oncologists on their respective experiences with FP counseling at the time of diagnosis (
). Overt discrepancies exist within the data, with studies showing 95% of oncologists at major US cancer centers reporting routine discussions of a treatment’s impact on fertility, but a disparate and wide-ranging 9–72% of patients reporting having received any FP counseling at the time of diagnosis (
Partridge AH, Gelber S, Peppercorn J, Sampson E, Knudsen K, Laufer M, et al. Web-based survey of fertility issues in young women with breast cancer. https://doi.org/10.1200/JCO200401159 2016;22:4174–4183.
). The majority of these were retrospective recollections by patients and clinicians, thus with the inevitable limitation of recall bias.
Recall bias refers to differential responses or inaccuracies of recall that limit conclusions drawn from retrospective studies involving self-reporting about the past (
). Study participants are systematically more or less likely to recall/relay information depending on the characteristics of the exposure of interest (ie, the degree of detail, significance to the patient, social standards, time duration spanned) and subsequent events (
Partridge AH, Gelber S, Peppercorn J, Sampson E, Knudsen K, Laufer M, et al. Web-based survey of fertility issues in young women with breast cancer. https://doi.org/10.1200/JCO200401159 2016;22:4174–4183.
A Delphi consensus study among patients and clinicians in the Netherlands on the procedure of informing young breast cancer patients about fertility preservation.
), a frequently cited limitation or concern is the effect of recall bias, with concerns about patients’ abilities to accurately report events that took place several years prior, especially in the setting of the stress and anxiety associated with a new cancer diagnosis (
Our primary objective was to assess recall bias by evaluating how well female cancer survivors remember details regarding their cancer diagnoses, oncology treatments, infertility risk factors, and FP counseling. Secondary outcomes included the rate of chart-documented FP counseling and the identification of any variables associated with that.
Materials and methods
This study was a secondary analysis of existing data collected from an internal review board-approved cross-sectional survey of female cancer patients at our academic institution (see Supplemental File 1, available online), followed by a systematic review of the electronic medical records of all enrolled patients. Female oncology patients 18–44 years old, at least 6 months past their last chemotherapy treatment, were contacted, given this was the earliest time point at which the literature defines primary ovarian insufficiency (POI) after chemotherapy. Patients >45 years old at the time frame examined in the survey were excluded, given the difficulty separating the effects of natural menopause versus gonadotoxic therapy on the ovaries. Additional exclusion criteria were prior hysterectomy or bilateral oophorectomy and patients <16 years old at the time of chemotherapy.
A Maestro Care or Epic build, the clinical research management system that can screen the electronic medical record at our institution, was used to identify patients at our academic institution who met these criteria above (
). Of those identified and recruited 117 consented and completed the survey. Participants were compensated with entry into a $100 gift card lottery for their participation.
The survey instrument included demographic questions regarding age, ethnicity, race, and body mass index, as well as other variables relevant to fertility risk in the setting of their oncology treatment. Patients self-reported information about their cancer diagnoses (type and stage), treatment received (chemotherapy and radiation), menstrual history before and after chemotherapy, baseline antimüllerian hormone (AMH) level if collected, among other fertility-impacting factors, as well as whether they were offered any FP counseling at the time of diagnosis.
Chart reviews were then completed to verify all self-reported responses, and patient-reported survey data versus chart-documented data were compared. Three investigators (E.C., S.M., and B.H.) independently reviewed each patient chart, and if there were discrepancies, what was identified or determined by 2 investigators versus 1 was deemed accurate. If critical details of the patient’s oncology treatment plan could not be confirmed, the chart was considered nonverifiable and excluded from the analysis. Chart-documented FP counseling was uniformly determined based on any discussion of “fertility” or “fertility preservation” options in either the oncology notes (primarily in the first oncology visit, although each patient’s entire record was searched) or by the presence of an oncofertility consult with an REI. Primary outcomes included accuracy of patient recall of the following: cancer type, cancer stage, chemotherapy regimen, number of cycles of chemotherapy received, exposure to radiation, concurrent GnRHa therapy received, baseline AMH being drawn, and FP counseling. The secondary outcome focused on the actual rate of chart-documented FP counseling found in the chart review.
Summary statistics were performed, and frequencies were reported as n (%) (Table 1). For the primary outcome of recall accuracy, a univariate regression analysis was performed to identify patient factors associated with the accurate recall of discussion regarding FP. For the secondary outcome of whether FP was chart-documented or discussed, a univariate analysis was performed to identify patient variables associated with a chart-documented discussion regarding FP services; further, a multivariable logistic regression model for this relationship was created while adjusting for the following covariates: age, race, ethnicity, cancer type, stage, number of chemotherapy cycles, radiation, and GnRHa use. All statistical analyses were performed in R (version 4.0.2).
Table 1Summary statistics of the study population
Characteristic
N = 112, n (%)
Mean age (y) [range]
31 [19–42]
Mean duration from last chemotherapy to survey [range]
In total, 117 patients out of 400 completed the survey (response rate of 29.2%), and 112 were able to be fully verified via chart review. Five patients with missing data surrounding their oncology treatment were removed from the analysis. The demographics of the patients are shown in Table 1. The mean age of the participants was 31 years (range 19–42). On average, the time from their last chemotherapy to the time of taking the survey was 3 +/- 0.19 years (SE), with a range of 0.5 to 7 years, with 98% of the participants at 5 years or fewer from the last treatment. Most patients were white, and they were fairly evenly divided into patients with breast, hematologic and other cancer types. Twenty-one percent of survey respondents had received a GnRH analog (typically agonist, GnRHa) with their chemotherapy. Approximately one-third had regimens with a cyclophosphamide equivalent dose of >4000 mg/m2, and less than 10% had exposure to total body irradiation or pelvic radiation. Seven percent of respondents reported being aware that they started with diminished ovarian reserve (DOR), whereas 42% of respondents reported amenorrhea or POI after completion of their cancer treatment.
Regarding the accuracy of recall, 95.5% (107/112 participants) correctly identified their cancer type, and 81.3% (91/112) reported their cancer stage accurately (Fig. 1). When asked to report their chemotherapy regimen, the majority, or 57.1% (64/112), reported “I don’t know or do not prefer to say.” Of the 48 participants who did recall their treatment, their recall accuracy was high at 93.8%. Eighty-seven percent (97/112) reported remembering an approximate number of cycles of chemotherapy received. However, of those 97 participants who reported the number of cycles they thought they had received; over 25% were inaccurate in their recall. On the other hand, 96.4% (108/112) accurately recalled their exposure to pelvic radiation or total body irradiation. Additionally, in our dataset, chart review confirmed that 21% of patients (23/112) received concurrent GnRHa therapy during chemotherapy, and 95% (22/23) accurately recalled this exposure.
Figure 1Accuracy of patient recall regarding chemotherapy regimen.
Regarding patient-reported FP counseling, only 19.6% of all participants reported being offered counseling regarding the effect of their chemotherapy on future fertility. However, from our chart review, we found that a discrepant 1 in 3 oncologists (37/112, or 33%) had documented that this conversation around FP had occurred. And of these 37 chart-documented FP conversations, 13 (or 35%) of these patients did not recall that this counseling took place (Fig. 2B). Interestingly, of the 24 patients who reported receiving GnRHa therapy with chemotherapy, 9 denied FP being discussed. Our survey, however, did not elucidate whether this was because of the patients being offered GnRHa therapy for menstrual suppression versus FP.
Figure 2Patient recall of fertility preservation counseling.
Finally, fewer than 10% of patients appeared to have a consult with an REI, with most patients receiving their FP counseling from their primary oncologist at diagnosis. Only a very small number of patients (N = 8) had their baseline AMH drawn prechemotherapy after a consultation with an REI. Of those who did, only 1 in 4 (or 25%) recalled ovarian reserve testing being performed. Of these patients, 3 patients had AMH values <1 ng/mL but specifically responded “no” when asked if they had prechemotherapy DOR.
Univariate and multivariable logistic regression revealed that patient factors (including age, race, ethnicity, cancer type, chemotherapy, and radiation) had no effect on the accuracy of recall (all P values were not significant). Multivariable logistic regression revealed that older age was significantly associated with not being offered FP (OR 0.87 [0.77–0.97], P = .02). All other variables were nonsignificant regarding the effect of receiving FP counseling.
Discussion
Our results confirm that the accuracy of patients’ reporting surrounding the details of their oncology treatment is limited by a poor recall, particularly regarding specific chemotherapy regimens and the number of cycles. Only 1/5 patients recall and report receiving FP counseling at the time of diagnosis. More than 1 in 3 patients documented as being offered FP counseling does not recall this conversation, and 3 in 4 patients do not recall baseline AMH testing. Importantly, only about one-third of cancer survivors had chart-documented FP counseling.
A barrier to accurate recall from the patient’s perspective may be that recall accuracy is poor during the initial cancer diagnosis and initial visits, given that this is an emotionally challenging time. Interestingly, we hypothesized that younger age at diagnosis and treatment would be associated with poor recall accuracy, but this was not found to be the case. Given that our initial survey contacted patients at least 6 months from their last chemotherapy, without an upper limit on time since chemotherapy, it had been several years since some of the patients received counseling and treatment; the longer time since cancer treatment may have added to worse recall over time. It has been noted in the literature that up to 50% of details of a critical event are not retrievable beyond 5 years from the incident(
), and in our study, 98% of participants were 5 years or fewer from the time of their last treatment. This may limit our study’s generalizability of the recall rate to patients with more recent cancer diagnoses or treatments.
There are multiple barriers to providing oncofertility referral and counseling from an oncologist’s perspective. First, many of these patients were seen and treated years ago, before an official pathway for oncofertility referral (and patient care navigator) existed at our university medical center. With more recent visits, as more notes become automated in electronic medical records, providers may be using prepopulated documentation that states that FP options were discussed, regardless of whether counseling was meaningfully offered. In addition, providers may be uncomfortable counseling the patient about their risks and options. Finally, the first oncology visit is often one in which a large amount of information needs to be relayed, and the oncology treatment plan and options discussed; given this large amount of ground to cover in that first visit, oncology providers may feel limited by time constraints to complete FP counseling.
Better systemic solutions are needed to improve FP counseling rates and increase oncologist comfort in discussing the risk to future fertility. A personalized tool, such as a patient-specific fertility risk calculator, has been developed to enable easy access for all parties involved and invested in FP counseling (
). This fertility risk calculator could function similarly to other web-based clinical predictive tools, such as the Gail model for breast cancer risk assessment tool or the Maternal-Fetal Medicine Units vaginal birth after cesarian calculator. Patients, oncologists, and REIs could easily access this tool, with a sample featured here (Fig. 3). Similar user-friendly egg freezing calculators used in FP counseling are already widely used (
), not only to help physicians counsel their patients consistently but also to serve as an easily-accessible resource for patients to make informed decisions or to refer to after their consultation to help recall the counseling provided. Furthermore, the importance of adequate follow-up beyond the initial mention of fertility at the oncology visit, or even the initial REI visit, was highlighted in our study. For example, even when an AMH was drawn for FP counseling purposes, most patients did not recall that their ovarian reserve had been tested, whereas ideally, the ovarian reserve would be followed up after treatment (particularly for those patients who had before treatment DOR and were at greatest risk of POI). Possible interventions to improve follow-up would include the establishment of REI care before treatment (to facilitate ease of follow-up after treatment), patient attendance of survivorship-oriented clinics after treatment, and establishment of an official FP program with a patient navigator that can assist in pretherapy and posttherapy consultations and coordination of care with continued REI follow-up. An FP program with standardized referral pathways and regular stakeholder meetings in the network’s oncology departments would facilitate this increase in referrals and follow-up care (
), as has been seen at the investigators’ institution (notably after most patients in this study had been initially diagnosed with cancer, which helps to explain the low number of referrals to REI).
Figure 3Fertility risk calculator tool for oncofertility patients.
Limitations of this study include its small sample size, with 112 patients at a single institution. This limits the generalizability of the study as there appear to be growing wide variations in FP programs across different institutions. Methodologically, as a retrospective survey study, a clear disadvantage was response bias. To address this, we enrolled patients via MyChart messages, so we are unable to comment on rates of fertility discussions in the subset who did not respond. Of note, there was a low number of patients with before treatment ovarian reserve testing (eight patients), which we attributed to the fact that (
Ethics committee of the American Society for Reproductive Medicine Fertility preservation and reproduction in patients facing gonadotoxic therapies: an Ethics Committee opinion.
) patients who did not plan on ovarian stimulation for FP often did not have ovarian reserve testing performed in past years. Further, we used chart documentation to determine whether the FP counseling occurred; it is possible that chart documentation differed from the conversation that actually took place. Although we demonstrate that patient recall of FP consultation can be inaccurate, this finding could alternatively be interpreted to suggest that chart reviews may not accurately capture these discussions.
The key strength of our study is that it uniquely addresses a very common limitation cited in the oncofertility literature that has not been explored much. Additionally, the fact that we were able to link the surveys to clinical documentation allowed for a detailed comparison and verification of each survey response. Finally, this is the largest study to date evaluating recall bias in the oncofertility population.
Conclusions
In summary, caution must be used in interpreting retrospective oncofertility studies without verifiable chart-reviewed data. Patients particularly struggle to recall their chemotherapy regimen, the details of which are critical for gonadotoxic risk counseling. With 35% of patients unable to recall discussions about FP, with an even poorer recall of ovarian reserve testing and its implications, increased efforts (including better counseling tools and streamlining the process of referral to make uptake easier for oncologists) are needed to ensure adequate follow-up and counseling beyond the initial visit.
Partridge AH, Gelber S, Peppercorn J, Sampson E, Knudsen K, Laufer M, et al. Web-based survey of fertility issues in young women with breast cancer. https://doi.org/10.1200/JCO200401159 2016;22:4174–4183.
A Delphi consensus study among patients and clinicians in the Netherlands on the procedure of informing young breast cancer patients about fertility preservation.
This paper in abstract form was accepted for oral presentation at the American Society for Reproductive Medicine in October 2021.
E.H.C. reports funding from Turtle Health outside the submitted work. K.S.A. reports funding from Charles Hammond Research Fund for the submitted work. S.M. has nothing to disclose. B.S.H. has nothing to disclose. E.W. has nothing to disclose.
The primary study on which this secondary analysis was based was generously supported by the Charles Hammond Research Fund, Duke University School of Medicine, Durham, North Carolina.