Three-, 6-, and 12-month screening mammography adherence rates among the intervention group (n = 408) (same-day screening mammography availability) and the control group (n = 512) (usual scheduling) women for phases 1 and 2 combined.
Dolan NC, McGrae McDermott M, Morrow M, Venta L, Martin GJ. Impact of Same-Day Screening Mammography AvailabilityResults of a Controlled Clinical Trial. Arch Intern Med. 1999;159(4):393-398. doi:10.1001/archinte.159.4.393
We conducted a prospective controlled clinical trial in an urban academic general medicine practice to test the effect of same-day mammography availability on adherence to physicians' screening mammography recommendations.
Patients and Methods
Participants were a consecutive sample of 920 female patients aged 50 years or older who had received a physician's recommendation for screening mammography at an office visit and had no active breast symptoms, history of breast cancer, or a mammogram within the previous 12 months. Women were assigned to same-day screening mammography availability (intervention group) or usual screening mammography scheduling (control group).
Main Outcome Measures
Three-, 6-, and 12-month rates of adherence to physicians' recommendations for screening mammography.
Twenty-six percent of women in the intervention group obtained a same-day screening mammogram. At 3 months, 58% of the women in the intervention group underwent the recommended screening mammography compared with 43% of the women in the control group (P<.001), increasing to 61% and 49% at 6 months (P<.001), and 268 (66%) of 408 vs 287 (56%) of 512 at 12 months (P = .003). The difference between the intervention and control groups 3-month adherence rates was most marked among women aged 65 years or older (58% vs 34%; P<.001), women who were not employed (54% vs 36%; P<.001), and women with a history of having had either no mammograms (39% vs 20%; P = .02) or only 1 to 2 mammograms (57% vs 38%; P<.001) within the last 5 years.
Same-day mammography availability increased 3-, 6-, and 12-month screening mammography adherence rates in this urban academic general medicine practice. The effect was most marked among women aged 65 years or older, women who were not employed, and those who had had fewer than 3 mammograms in the last 5 years. The efficacy of this intervention in other settings still needs to be demonstrated.
SCREENING mammography has been shown to decrease breast cancer mortality in women aged 50 years or older by up to 30%.1- 3 The benefits of breast cancer screening to reduce mortality in the population can be achieved only if screening guidelines are followed and a large proportion of women receive screening examinations regularly. While recent data show that the proportion of women reporting recent mammography has substantially increased from 1989 to 1995, 30% to 40% of women aged 40 years or older report that they have not had a mammogram within the last 2 years.4- 6 Although lack of a physician's recommendation is an important cause of underutilization,7- 12 among women seen in a physician's office who have not had a recent mammogram, adherence rates to a physician's recommendation are only 45% to 60%.13- 15
In a previous prospective observational study15 among women aged 50 years or older who received a physician's recommendation for screening mammography, we identified "inconvenience" as one of the most frequently cited reasons for not obtaining the test. Other observational studies16,17 have also suggested that factors affecting convenience of screening mammography are barriers to adherence.
These data suggest that increasing the convenience of screening mammography may increase screening rates. We hypothesized that providing women with the opportunity to get their screening mammogram immediately after the appointment at which it was recommended (same-day mammography) would improve adherence.
To test the effect of this strategy on screening mammography adherence, we conducted a controlled trial to compare same-day screening mammography availability with usual scheduling. Because many women in the intervention arm reported that they would have taken advantage of the opportunity if they had known about it in advance, we designed a second phase of the study to test the effect of advance notification of the same-day opportunity along with same-day screening availability on adherence rates.
The study site was an urban academic general internal medicine practice with a hospital mammography center located 3 blocks away. The practice site was staffed during the time of the study by an average of 30 attending physicians and 62 house staff. Approximately 77% of all patients during the study period had an attending physician and 23% had a house staff as their primary care physician. Consecutive female patients aged 50 years or older presenting for new or return visits between February 1, 1995, to September 1, 1996, were eligible for the study. Women were excluded if they were presenting for an acute care visit, had obtained a mammogram in the previous 12 months, had a history of breast cancer, had an active breast symptom at the time of the visit, or had not received a recommendation for a screening mammogram from their physician at the index appointment. Acute care visits were defined as visits scheduled within the previous 24 hours for an acute medical problem such as a cold or back pain. At the study practice, receptionists scheduling the appointments designated the visit type at the time an appointment was made. Having had a mammogram within 12 months was used as an exclusion criterion to capture all women who would be eligible to receive a physician's recommendation for a screening mammogram. The study protocol was reviewed and approved by the Institutional Review Board.
A research assistant asked eligible women to complete a study questionnaire at the time of check-in and attached screening mammography recommendation physician-prompting sheets to the charts of participating patients. Physicians documented whether they recommended a mammogram at the study visit. At the time of checkout, a research assistant documented whether patients planned to get the recommended mammogram, and where they intended to get it. Women were then assigned to the intervention or control group according to whether the fourth digit of their social security number was odd or even. Women with an even number were assigned to the intervention group, women with an odd number the control group.
Although women could have had subsequent visits to their physician after study enrollment, mammography recommendation reminders and same-day mammography intervention were provided only at the initial visit.
Women in the intervention group were offered the opportunity to obtain the screening mammogram immediately after their appointment. Responses and reasons for refusal were recorded. From February 1, 1995, until September 29, 1995 (phase 1 of trial), women who refused the same-day mammogram were asked if they would have been likely to accept if they had known about the opportunity in advance.
The research assistant notified the mammography center of those accepting the offer and directed these women to the center located 3 blocks away. A free minibus service was available for transport to the mammography center. This service was discontinued September 29, 1995, for reasons not related to the study. Women in the intervention group not accepting the same-day mammogram offer were checked out and directed to schedule the mammogram by telephone as per the usual procedure at the study site. Waiting periods for mammography ranged from 1 to 3 weeks from the time of the scheduling telephone call.
The purpose of phase 1 of the study (February 1, 1995-September 29, 1995) was to evaluate whether a same-day screening mammography opportunity increases screening mammography adherence rates among women aged 50 years or older in a general medicine practice. Because a substantial proportion of subjects in the intervention arm reported that advanced notification of the same-day opportunity would have increased their likelihood of obtaining a same-day mammogram, we designed a second study (phase 2) to test the additional intervention of advanced notification of the same-day screening mammography opportunity. Because phase 1 provides a reference against which phase 2 screening mammography adherence rates can be compared, we elected to report the results of both phases 1 and 2 in a single article.
Phase 2 of the study began October 1, 1995. Two weeks before their scheduled appointments, potential study participants were assigned to the control or intervention group. Potential control group women were sent an informational postcard on screening mammography. The intervention group women were sent the same information as well as notification of the availability of same-day screening mammography if their physician recommended it. When a woman arrived for her appointment, a research assistant asked whether she remembered receiving the postcard and documented this on the questionnaire.
The primary outcome measure was the 3-month rate of adherence to physicians' screening mammography recommendations. Three months was chosen to allow women not undergoing same-day mammography adequate time to complete the screening mammogram. To allow for the effects of delayed adherence among both groups, we looked at 6- and 12-month adherence rates as secondary outcome measures. Adherence rate was defined as the percentage of women who had documentation of having had a screening mammogram within the defined period (3, 6, and 12 months) from their physician's recommendation. Adherence was determined for both groups using computerized radiology records at the study institution. If a woman indicated she was going to obtain the mammogram at another mammography center, the specified site was contacted to determine whether the mammogram had been performed.
To determine whether specific patient characteristics were associated with a greater intervention effect, we also analyzed 3-month adherence rates stratified by the following variables: calendar period (phase 1 vs phase 2), age (<65 years vs ≥65 years), education (high school and below vs more than high school), race (white vs African American), employment status (employed vs not employed), and number of mammograms within the last 5 years (<3 vs ≥3). Other outcome measures analyzed were the percentage of women in the intervention group undergoing same-day mammography, and for phase 1 of the trial, the percentage of women in the intervention group who reported they would have accepted the intervention with advanced notification of the opportunity. To measure patient satisfaction with the intervention, a research assistant called women who underwent same-day screening mammography 1 day after the test. Women were asked to rate their satisfaction with the experience on a Likert scale (1, very satisfied; 5, dissatisfied).
χ2 Tests were used to compare categorical variables and adherence rates between the intervention and control groups. Two-sample t tests were used to compare continuous variables between groups. We performed these analyses separately for phases 1 and 2 and combined. Because the characteristics of control group patients in phases 1 and 2, and intervention groups in phases 1 and 2 were similar, only the combined data are shown. Three-month adherence rate ratios and 95% confidence intervals (CIs) were calculated to compare adherence rates among the subgroups of intervention and control individuals. Because the 3-month screening mammography rates were similar between the control groups in phases 1 and 2 and between the intervention groups in phases 1 and 2, we chose to report the combined results for our subset analyses. When the subset analyses were analyzed separately for phase 1 and 2 participants, our findings were similar to those for the combined analyses. All women who were entered during phase 2 of the study were analyzed in the same subgroup, regardless of whether women actually reported receiving the postcard (intention-to-treat).
Using combined data from phases 1 and 2, unadjusted and adjusted logistic regression analyses were performed to evaluate the effect of the intervention alone and the effect of the intervention after controlling for potential confounding variables. Variables with significant baseline differences (P≤.05) between the control and intervention groups were included in the adjusted logistic regression analysis. The independent variables entered into the model were group status (intervention vs control), age, education level, employment status (employed vs not employed), and primary insurance type (Medicare vs other).
Of 2039 women aged 50 years or older presenting to the office for new or return visits, 722 had had a mammogram within the previous 12 months, 119 had a history of breast cancer, 45 had active breast symptoms at the time of the visit, 57 had not received a physician's recommendation, and 176 declined to participate. Nine hundred twenty women were enrolled in the study, 533 in phase 1 (249 intervention and 284 control) and 387 in phase 2 (159 intervention and 228 control). Intervention and control groups demographic characteristics combined for phases 1 and 2 are summarized in Table 1. Women in the intervention group were older, less well educated, more likely to have Medicare, and less likely to be employed compared with control group women. The groups were well balanced with respect to family history of breast cancer, history of breast biopsy, and prior use of screening mammography.
During phase 1 of the trial, 67 (27%) of 249 women in the intervention group underwent a same-day screening mammogram. One hundred two (56%) of the 182 intervention women who did not undergo a same-day mammogram during phase 1 of the trial stated that they would have taken advantage of the opportunity if they had known about it earlier. Of the 159 intervention women enrolled during phase 2 of the trial when advance notification postcards were sent, 95 (60%) reported receiving the postcards. Among these 95 women, 20 (21%) had a same-day mammogram. Among the 64 women who did not receive the postcard, 17 (27%) accepted the same-day screening mammography opportunity.
Table 2 summarizes the characteristics of phase 1 and 2 women in the intervention group undergoing same-day screening mammography compared with those who did not. Women who took advantage of the same-day screening opportunity had slightly more education than those who did not and tended to be more likely to take public transportation to their appointments, but did not differ significantly with respect to age, race, employment status, and past use of mammography. Among women who underwent same-day mammography, overall satisfaction with the experience was 1.4 ± 1.0 (mean ± SD) on a 5-point scale with 1 being most satisfied.
Three months after the recommendation was made, 144 (58%) of 279 women in the intervention group had obtained the recommended mammogram compared with 120 (42%) of 284 in the control group (P<.001), increasing to 152 (61%) of 249 vs 140 (49%) of 284, respectively, at 6 months (P = .006) and 156 (64%) of 242 vs 158 (58%) of 271 at 12 months (P = .15).
Three- and 6-month adherence rates for phase 2 participants were identical to those of phase 1. Three months after the recommendation was made, 92 (58%) of 159 of those in the intervention group had obtained the recommended mammogram compared with 98 (43%) of 228 in the control group (P = .003), increasing to 97 (61%) of 159 vs 111 (49%) of 228, respectively, at 6 months (P = .006) and 106 (67%) of 159 vs 123 (54%) of 227 at 12 months (P = .01). Figure 1 illustrates the overall adherence rates of the intervention and control group women combined for phases 1 and 2 of the trial.
The results of the subgroup analyses for combined data from phases 1 and 2 are summarized in Table 3. All subsets of women except those who had 3 or more mammograms in the last 5 years benefited from the same-day screening intervention. The difference between the intervention group's and control group's 3-month adherence rates was most marked among women aged 65 years or older (58% vs 34%; P<.001), women who were not employed (54% vs 36%; P<.001), and women with a history of either no mammograms (39% vs 20%; P = .02), or only 1 to 2 mammograms (57% vs 38%; P<.001) within the last 5 years.
In a logistic regression analysis controlling for age, education, race, employment status, and primary insurance type, the odds ratio for the intervention group undergoing mammography was 1.9 (95% CI, 1.7-2.2) at 3 months, 1.7 (95% CI, 1.4-1.9) at 6 months, and 1.5 (95% CI, 1.1-2.1) at 12 months.
The results of this study suggest that the availability of same-day screening mammography increases rates of adherence to physicians' screening mammography recommendations among women aged 50 years or older and is associated with high levels of satisfaction. Our data also suggest that advance notification of this opportunity may not increase its use.
Previously studied patient-directed interventions designed to increase screening mammography rates have included mailed invitations to participate in screening, mailed reminders, mailed booklets based on the Health Belief Model, educational videos, tailored letters, and telephone counseling.18 With the exception of programs using mobile mammography vans, however, few studies have used access-enhancing interventions.18
Although this is the only controlled trial we are aware of that evaluates the effectiveness of same-day screening mammography availability on adherence, previous data suggest that a same-day mammography opportunity may be associated with greater use of mammography.17 McBride et al,17 in a study of women in a health care maintenance organization, found that nonparticipants in screening mammography had more trouble getting to the facility, had to travel farther, and were more likely to rate the facility as being inconvenient. Margolis et al16 studied 907 women with scheduled mammography appointments at a public teaching hospital and determined that long waiting intervals for appointments were associated with decreased adherence.
There are several potential explanations for the improved adherence rates observed with same-day screening mammography availability. First, the impact of a physician's recommendation is likely to be strongest at the time it is made. Longer intervals between the time a recommendation is made and the point at which mammography is available may weaken the initial motivation inspired by the physician's recommendation. Second, because same-day screening mammography availability eliminates the need for a separate visit, it saves time, is more efficient, and reduces or eliminates transportation-related problems and costs.
During phase 1 of the trial, 27% of women in the intervention group actually took advantage of the same-day screening mammography opportunity. We found no patient characteristic among women in the intervention group, other than education, associated with acceptance of the same-day screening mammography opportunity. Logistical factors may have contributed to the relatively low rate of acceptance. The mammography center was located 3 blocks from the office site. Inclement weather or difficulty with ambulating may have deterred some women from taking advantage of the opportunity. Another potential explanation is that women were unable to take the time for the mammogram because of other previously scheduled commitments. In fact, a large proportion of intervention group women in phase 1 of the trial indicated that they would have gotten a same-day screening mammogram if they had known about it in advance. To address this second potential barrier, we conducted phase 2 of the trial to test whether use of the same-day opportunity would increase if women knew about it in advance. Our results showed that the percentage of women accepting the opportunity did not increase when women were notified in advance. Many women in phase 2 of the study, when actually faced with that option, were perhaps still not in a state of readiness to comply with the recommendation. Another explanation for the lack of effect of the advanced notification postcards might be that the potential positive effect of the intervention was offset by the discontinuation of the minivan at the same time. Women entered during phase 2 of the study when there was no minivan but who reported not receiving a postcard, however, had the same rate of same-day mammography acceptance as that of intervention women in phase 1. This finding suggests that the discontinuation of the minivan did not have a significant effect on the use of the same-day mammography opportunity, and therefore is not likely to be the explanation for the lack of an effect of the postcards on same-day mammography use.
Our data suggest that women at highest risk for not obtaining a screening mammogram benefited the most from this intervention. Specifically, older age, unemployment, and fewer previous mammograms, factors previously associated with decreased use of mammography,7- 12,19 were associated with the strongest intervention effects. In contrast, women with frequent previous use of mammography had high rates of adherence regardless of whether they were in the control or intervention group. Therefore, targeting this intervention to those at greatest risk of nonadherence might be an effective strategy for improving adherence while minimizing the potential burden of same-day screening on mammography units.
Several considerations should be taken into account when interpreting these study results. First, the study took place in an urban academic practice with a nearby mammography center. The results may not be generalizable to practices that do not have mammography units in such close proximity. Alternatively, same-day mammography use and subsequent adherence may be even higher among practices that have on-site mammography units. Second, all physicians received a prompt to recommend a screening mammogram to eligible women and only women who received a recommendation were included in the study. Physician prompts are a separate intervention that may have led to overall higher use of mammography in both intervention and control groups. Third, since women who had had a mammogram within the previous 12 months were excluded, the population studied was a relatively select group whose adherence would be expected to be less than that of the entire practice. This is the group, however, in which interventions to improve adherence are most necessary.
A limitation of this study is the unbalanced allocation. Study participants were allocated to the intervention or control group based on whether the fourth digit of their social security number was even or odd, respectively. Digits 4 and 5 of the social security number denote the group number. In general, "even" group numbers (ie, 10, 12, 14, or 16) are assigned consecutively followed by "odd" group numbers (ie, 11, 13, 15, or 17) within a given state or area. Because social security numbers are not randomly assigned, our method of allocation to the intervention vs control group was not random, but instead reflected the social security numbers of patients seen in our general medicine practice. As a result, more women were randomized to the control group, and women allocated to the intervention arm were older, less educated, and more likely to have Medicare insurance compared with the controls. Previous observational data from our institution and others show that increasing age, lower educational status, and Medicare insurance are all associated with lower rates of adherence to screening mammography.7- 12 Therefore, our finding that the same-day mammography intervention resulted in higher rates of screening mammography is especially noteworthy, since the distinguishing characteristics of the intervention group at baseline are known to be associated with lower screening mammography adherence rates. This phenomenon is underscored by our finding that the odds ratio for screening mammography adherence increased after adjusting the bivariate analyses for the observed differences between the intervention and control groups.
In conclusion, same-day mammography availability appears to effectively increase adherence to screening recommendations. This effect is most marked among older women and those who were previously low users of mammography. Targeting women with a history of fewer mammograms for this intervention would be an effective strategy for mammography centers that are unable to accommodate a large mammography-on-demand population. Whether this strategy is effective in combination with other intervention strategies and in other practice settings are areas for future investigation.
Accepted for publication June 2, 1998.
Dr Dolan is supported by an American Cancer Society Cancer Control Career Development Award for Primary Care Physicians. Dr McDermott is a Robert Wood Johnson Generalist Physician Faculty Scholar. This study was supported in part by the Illinois Department of Public Health Breast, Springfield, and Cervical Cancer Research Fund, Frederick, Md, and by the Department of Defense Breast Center Research Grant (DAMD 17-96-2-6013).
Presented at the Annual Meeting of the Society of General Internal Medicine, Washington, DC, May 4, 1996.
We thank Joan Chmiel, PhD, for her statistical guidance and critical review of the manuscript, and Catherine Eusebio for her assistance with the logistics of the study.
Corresponding author: Nancy C. Dolan, MD, Division of General Medicine, Department of Medicine, Northwestern University Medical School, 303 E Ohio, Suite 300, Chicago, IL 60611 (e-mail: email@example.com).