Participation behaviour following a false positive test in the Copenhagen mammography screening programme

Abstract

Introduction. There is an ongoing debate concerning possible disadvantages of mammography screening, one being the consequence of receiving a false positive test-result. It is argued that receiving a false positive answer may have short- and/or long-term adverse psychological effects on women, but results from different studies are conflicting. We tested if there was a difference in continued participation behaviour between the group of women who have been subject to a false positive result and those who have not. Material and methods. The study used the registers from the first six invitation rounds of the mammography screening programme in Copenhagen (1991–2003). We estimated the relative risk of not participating in the subsequent screening round for women with a false positive test using women with a negative test as baseline. As outcome measure odds ratios (OR) with 95% confidence intervals (CI) were used. Results. There was no significant difference in participation in the subsequent round between women with a false positive test and women with a negative test. The proportion of screens resulting in false positive answers, both after assessment and after surgery, decreased from 5.54% in Round 1 to 1.79% in Round 5. Participation in the subsequent screening round was well above 80% in all five screening rounds. Discussion. Our results showed that women experiencing a false positive test at mammography screening participated in the subsequent screening round to the same extent as did women experiencing a negative screening test, regardless of whether the false positive statement was given following assessment or following surgery. The benign to malignant biopsy ratio, comparing the type B false positives to the true positives, was by the fifth round well below the desirable level of ≤1:4, recommended by the European guidelines. Other possible adverse effects should be further investigated.

Mammography has proven an efficient way of decreasing mortality of breast cancer in the population [1]. There is, however, an ongoing debate concerning possible disadvantages of mammography screening, one being the consequence of receiving a false positive test-result. It is argued that receiving a false positive answer may have short- and/or long-term adverse psychological effects on women, but results from different studies are conflicting. Several studies report increased levels of anxiety and cancer-related concerns in the short term [2–6]. Some long-term effects have also been reported, notably more active and excessive breast self-examinations [3], [6].

It would be reasonable to assume that these effects could change women's attitude towards the mammography screening programme and thereby alter their participation behaviour. However, concerning subsequent adherence to mammography programmes, earlier results are also inconclusive [7]. These variations might in part be due to different recruitment processes, varying length of follow-up and varying study design.

We tested if there was a difference in continued participation behaviour between the group of women who have been subject to a false positive result and those who have not. We used data from the population-based organised screening programme in Copenhagen, Denmark, 1991–2003.

Materials and methods

Definitions

Invitation round

The invitation round is defined administratively by an interval of dates of approximately two years.

Negative

Refers to women who have been tested negative at the initial examination.

False positive type A

Refers to women who have been tested negative after the first recall, at assessment.

False positive type B

Refers to women who have been tested negative after the second recall, at surgery.

The mammography screening process

Mammography screening in Copenhagen is conducted by one clinic, located at one of the hospitals in the municipality of Copenhagen. By appearance, after invitation, at the initial screening the women get standardized two-projections x-rays, which are evaluated by two independent radiologists. From the start of the programme in April 1991 women with mixed/dense breast tissue were screened with two projections at subsequent examinations, whereas women with fatty breast tissue were screened only with one projection in the subsequent screening rounds. From 2004 and forth it has been the policy to perform two projections of all women.

In case of suspicion of malignancy the women are called back for assessment within 14 days. At the assessment, clinical examination including palpation, mammography and ultrasonography are conducted. By presence of a palpable lesion or persisting suspicion of malignancy, ultrasound-guided fine-needle aspiration cytology (FNAC), or needle core biopsy is performed. If consensus regarding clinical examination, mammography and ultrasound and needle-biopsy (the triple diagnostic test) is present, the patient either returns to the regular screening programme or is referred to surgery. In case of an inconclusive outcome of the triple diagnostic test, repeated needle-biopsies or excision biopsy is performed. If possible, needle biopsies are ultrasound guided. Suspicious microcalcifications are investigated by stereotactic biopsy.

A false-positive answer is defined as a recall for assessment (i.e. result of first examination yields a positive test result) or surgery (i.e. result of first examination yields a positive answer as do the result of the assessment) that eventually turns out to show no signs of malignancy (Figure 1).

Figure 1.  Schematic figure showing the mammography process

Data sources

The target population constituted all women aged 50 to 69 living in Copenhagen at any time during an invitation round, and it was identified from the Central Population Register (CPR). The Danish Central Population Register includes information on all persons ever residing in Denmark since the register was set up in 1968. It also includes historical addresses, migrations and deaths. The CPR was used to define target populations, and data were linked using the CPR-number. All women in the target population are invited, unless they have specifically asked to be excluded. The study used the registers from the first six invitation rounds of the mammography screening programme in Copenhagen (1991–2003). The register contains the unique personal identification number (CPR-number), invitation date, attendance, examination date, result of screening and recalls for each screening round.

Analysis

We estimated the relative risk of not participating in the subsequent screening round for women with a false positive test using women with a negative test as baseline. Participants in one screening round (Round 1 to 5) still belonging to the target population for the next round (Round 2 to 6) constituted the study population. Women who during an invitation round turned 70, died or migrated were excluded from the target population in the subsequent round, as they were then non-eligible according to the screening programme criteria. As outcome measure odds ratios (OR) with 95% confidence intervals (CI) were used. SAS version 9.1 was used for the statistical analysis.

Ethics

This survey is entirely based on data from registers. No contact has been taken with patients, relatives or their practising doctors. The survey is registered with the Data Inspection Board.

Results

Among women experiencing a negative test at first screen, 84% came back for the next screen. This percentage increased to 89%, 89% and 88% in the next three invitation rounds, but then dropped back to 85% for those who experienced a negative test in the fifth round (Table I).

Table I.  Participation behaviour by diagnosis and invitation round, Copenhagen, Denmark, 1991–2003.

Date	Invitation round	All	Target next round	Non-participants next round	Participated next round	Participated next round %	CrudeOR	(95% CI)	Age-adjusted OR	(95% CI)
4 April 1991–23 April 1993	Round 1
	Negative	28328	23916	3914	20002	84	1		1
	False positive A	1432	1199	192	1007	84	1.03	0.87–1.21	1.02	0.87–1.20
	False Positive B	231	188	32	156	83	0.95	0.64–1.42	0.96	0.66–1.41
26 April 1993–31 May 1995	Round 2
	Negative	24914	21005	2398	18607	89	1		1
	False positive A	947	805	79	726	90	1.18	0.93–1.51	1.18	0.93–1.49
	False Positive B	59	51	6	45	88	0.97	0.40–2.52	0.99	0.42–2.32
1 June 1995–24 March 1997	Round 3
	Negative	24359	20910	2268	18642	89	1		1
	False positive A	571	489	52	437	89	1.02	0.76–1.38	1.00	0.75–1.35
	False Positive B	51	47	4	43	91	1.31	0.45–4.29	1.33	0.48–3.70
25 March 1997–19 April 1999	Round 4
	Negative	24662	21584	2653	18931	88	1		1
	False positive A	571	508	68	440	87	0.91	0.70–1.18	0.94	0.73–1.22
	False Positive B	36	35	4	31	89	1.09	0.37–3.62	1.09	0.38–3.08
20 April 1999–31 March 2001^1	Round 5
	Negative	24463	22243	3234	19009	85	1		1
	False positive A	419	385	68	317	82	0.79	0.60–1.04	0.81	0.62–1.05
	False Positive B	29	25	4	21	84	0.89	0.29–3.07	0.94	0.32–2.75

¹The subsequent round refers to Round 6, 1 April 2001–31 March 2003.

The proportion of screens resulting in false positive answers, both after assessment and after surgery, decreased from 5.54% in Round 1 to 1.79% in Round 5. False positive B decreased from 0.77% to 0.12% between Round 1 and Round 5. The proportion of type B false positives out of all false positives was highest in the first round, 13.9%, and then 5.9%, 8.2%, 5.9% and 6.5% for Rounds 2 to 5, respectively (Figure 2).

Figure 2.  False positive tests in% of all negative tests by round, Copenhagen 1991–2001

There was no significant difference in participation in the subsequent round between women with a false positive test and women with a negative test. For false positive test type A, the adjusted ORs for not participating in the subsequent round were 1.02 (95% CI, 0.87–1.20), 1.18 (95% CI, 0.93–1.49), 1.00 (0.75–1.35), 0.94 (95% CI, 0.73–1.22) and 0.81 (95% CI, 0.62–1.05) for round 1 to 5, respectively. The relative risks for women with false positive test type B were similar with ORs 0.96 (95% CI, 0.66–1.41), 0.99 (95% CI, 0.42–2.32), 1.33 (95% CI, 0.48–3.70), 1.09 (95% CI, 0.38–3.08) and 0.94 (95% CI, 0.32–2.75) for round 1 to 5, respectively. Apart from in round 1, few women received a false positive test type B and the confidence intervals are therefore broad (Table I).

Regardless of whether the patient received a negative test result or either of the two false positive test results, participation in the subsequent screening round was well above 80% in all five screening rounds.

Discussion

Our results showed that women experiencing a false positive test at mammography screening participated in the subsequent screening round to the same extent as did women experiencing a negative screening test. The results did not supply any statistical evidence supporting the hypothesis that women's participation behaviour would be affected by receiving a false positive answer. This, regardless of whether the false positive statement was given following assessment or following surgery.

In screening programmes the occurrence of false positive results is one of primary concern [3], [4], [6], [7]. In Copenhagen 12.7% of the women participating in the first five rounds of screening experienced at least one false positive screen. The predicted risk of experiencing a false positive result over an entire course of the screening programme (10 screens) is 15.8–21.5% [8]. The majority of these false positive results are found to be negative at evaluation following assessment [9], i.e. at first recall. The benign to malignant biopsy ratio, comparing the type B false positives to the true positives, improved for every screening round and where 1:1.57 in the first screening round, 1:2.76 in the second, 1:3.08 in the third, 1:4.14 in the fourth and 1:5.00 in the fifth screening round (data not shown).

Brewer et al. [7] (2007) reported, in a systematic review, results concerning the relationship between receiving a false positive mammogram and return for routine mammography from 12 studies. Most of these studies showed either a slightly higher re-attendance after a false positive mammogram than after a true negative [3], [10–12] or no difference in participation behaviour between the two groups [13–16]. Hofvin et al. [17] reported a lower re-attendance among false positives than negatives, and so did Brett et al. [18] Chiarelli et al. [19] show in a study from the Ontario Breast Screening Programme that although the re-attendance of false positives was lower than that of negatives, this did not apply for the centre where the diagnostic assessment remained the responsibility of the screening programme, with assessment clinics affiliated with screening centres. The organisational setup in Denmark follows this model. Other differences that may influence the re-participation of women with a false positive test are, according to Brewer et al. [7], regional differences in mammography screening practices such as shorter recommended screening intervals in the US than in Europe, and organisational differences between national programmes in Europe and screening activities handled by a variety of private and public providers as is the case in United States.

We found a decrease in the proportions of both types of false positive outcome by invitation round. This is in line with earlier studies that report higher proportions of false positive results when there was no earlier mammograms to compare with [10], [20]. The benign to malignant biopsy ratio improved for every invitation round and was by the fifth round well below the desirable level of ≤1:4, recommended by the European guidelines for quality assurance in breast cancer screening and diagnosis [21].

There are several studies concerning the potential short- and long-term effects of receiving a false positive test result, but the results are inconsistent [2]. Aro et al. [6] argued that, although some short-term distress after a positive result could be found, these are moderate, and that general cancer-specific distress is more of a problem. Scaf-Klomp et al. [22] argued that a false-positive test does not cause adverse effects longer than 8–10 weeks and that differences in psychological functioning between false positives and negatives rather was due to a sense of relief among the negatives. Other studies have indicated the presence of long-term effects, following a false positive result [3], [4], [6], [7]. Olsson et al. [4] found that 15% of false positive women still had increased levels of anxiety 6 months after assessment and Brewer et al. [7] also found that false positive women had higher levels of distress and anxiety. Brodersen et al. [23] stressed the fact, that it is questionable whether the measures used to examine the psychological long-term effects of a false positive result, actually adequately describes the long-term consequences, and that any conclusions must remain tentative. In some studies women with a false positive test were shown to perform more active and excessive breast self-examinations than women with normal mammography [6]. It has not been proven that breast self-examination decreases the mortality of breast cancer in the population [24] so this might be an undesired effect.

As described in Christensen et al. [25], one would over time expect a relative decrease in the proportion of false-positive tests due to improvement and refinement of the techniques used in screening, as well as to the increased expertise obtained by the radiologists.

In conclusion, our results did not indicate any difference in adherence to the organised screening programme between women who have received a false positive mammogram and those who have not. On average, women are neither deterred nor encouraged to participate in a subsequent screening round by experiencing a false positive test. Other possible adverse effects should be further investigated.

Acknowledgements

This study was financially supported by the Danish Cancer Society.