Original quantitative research – Breast cancer detection method, diagnostic interval and use of specialized diagnostic assessment units across Ontario, Canada

Abstract

Introduction: Breast cancer is detected through screening or through signs and symptoms. In Canada, mammograms for breast cancer screening are offered in organized programs or independently (opportunistic screening). Province of Ontario breast Diagnostic Assessment Units (DAUs) are facility-based programs that provide coordinated breast cancer diagnostic services, as opposed to usual care, in which the primary care provider arranges the tests and consultations. This study describes breast cancer detection method, diagnostic interval and DAU use across Ontario.

Methods: The study cohort consisted of 6898 women with invasive breast cancer diagnosed in 2011. We used the Ontario Cancer Registry linked to administrative health care databases. We determined the detection method using the Ontario Breast Screening Program (OBSP) data and physician claims. The diagnostic interval was the time between the initial screen, specialist referral or first diagnostic test and the cancer diagnosis. The diagnostic route (whether through DAU or usual care) was determined based on the OBSP records and biopsy or surgery location. We mapped the diagnostic interval and DAU coverage geographically by women's residence. 

Results: In 2011, 36% of Ontario breast cancer patients were screen-detected, with a 48% rate among those aged 50 to 69. The provincial median diagnostic interval was 32 days, with county medians ranging from 15 to 65 days. Provincially, 48.4% were diagnosed at a DAU, and this ranged from zero to 100% across counties.

Conclusion: The screening detection rate in age-eligible breast cancer patients was lower than published population-wide screening rates. Geographic mapping of the diagnostic interval and DAU use reveals regional variations in cancer diagnostic care that need to be addressed.

Keywords: breast neoplasms/diagnosis, female, diagnostic services, early detection of cancer, early diagnosis, cross-sectional studies, health services research

Introduction

Screening is important for achieving an early diagnosis of cancer and thereby improving the chance of survival.^{Footnote 1} In Canada, mammograms for breast cancer screening are offered by organized programs or independently (opportunistic screening).^{Footnote 2} The breast screening participation rate in the screen-eligible population is viewed as an important indicator of a screening program's effectiveness and is routinely monitored by organized programs.^{Footnote 3} But a high screening participation rate does not necessarily mean that all breast cancers are detected by the screening program. In fact, many are detected when the patient presents to her doctor with breast cancer signs or symptoms.^{Footnote 4} Documentation of the proportion of breast cancer patients detected by screening (vs. symptomatic presentation) is lacking.

The breast cancer diagnostic period is characterized by multiple appointments for diagnostic tests and consultations and it often provokes considerable distress and anxiety for women and their families.^{Footnote 5Footnote 6Footnote 7} A diagnostic delay of three months or more can lead to advanced cancer stage and lower chances of survival.^{Footnote 8} The 2007 guideline for monitoring breast screening program performance recommended 90% of abnormal screening results should be resolved within 5 weeks, or within 7 weeks if a tissue (core or open) biopsy is required.^{Footnote 9} Consequently, the time interval from an abnormal screen to a final diagnosis is routinely assessed against the national targets by organized programs. In 2011, the Ontario Breast Screening Program (OBSP) reported significant regional variation in meeting the national timeliness target for women with abnormal screens.^{Footnote 3} However, little is known about the length of the diagnostic interval at the population level and whether it varies in different regions of the province.

Ontario Diagnostic Assessment Units (DAUs) are innovative, facility-based programs designed to improve the timeliness and quality of diagnostic care services.^{Footnote 7} Similar initiatives exist in many jurisdictions around the world.^{Footnote 10Footnote 11} Ontario breast DAUs take two forms: organized Breast Assessment Affiliates (BAAs),^{Footnote 12} and regionally developed breast assessment centres. Diagnostic care at BAAs is provided by a multidisciplinary health care team that includes a nurse navigator who coordinates the process. This differs from the usual care diagnostic route (UC) where the patient's primary care provider plays a central role in arranging referrals and recommending appropriate diagnostic assessments and specialist consultations.^{Footnote 12Footnote 13Footnote 14} The nurse navigator also provides patient psychological and informational support. Each designated facility has to meet minimum organizational criteria regarding breast cancer diagnostic service resources and availability to maintain its BAA status.^{Footnote 13} The diagnostic process at regional breast assessment centres is less well documented, but those centres have similar goals and structures to those of the BAAs.

This study aimed to describe, among all Ontario breast cancer patients: 1) the cancer detection method; 2) the length of the diagnostic interval and its geographic pattern; and 3) the coverage of DAU services and its geographic pattern. This is one of three reports arising from our population-based study on DAU use and the diagnostic interval.^{Footnote 14Footnote 15Footnote 16}

Methods

Study population

The study population consisted of women with a first primary invasive breast cancer diagnosed in 2011 in Ontario, Canada. We excluded: 1) those whose cancer registry record could not be linked to administrative data; 2) those who were living outside of Ontario at the time of diagnosis; and 3) those who did not have Ontario Health Insurance Plan (OHIP) coverage for at least three years prior to diagnosis.^{Footnote 14Footnote 15}

Data sources

Study data included the Ontario Cancer Registry linked to administrative databases at the Institute for Clinical Evaluative Sciences (ICES) and Cancer Care Ontario (CCO). We used the Ontario Cancer Registry to identify breast cancer cases and determine the date of diagnosis. The Collaborative Stage Data provided information on cancer stage and histology. We defined stage at diagnosis using the TNM classification, 7th edition.^{Footnote 17Footnote 18} The OBSP database provides information on screening date, screening test results and use of BAAs. The OHIP claims database, the National Ambulatory Care Reporting System (NACRS), the Canadian Institute for Health Information Discharge Abstract Database (CIHI-DAD) and the Same-Day Surgery Database (CIHI-SDS) contain information on breast cancer diagnostic procedures and associated dates and physicians. A list of BAA hospitals was provided by the OBSP and we developed a separate list of regional breast assessment centres by conducting an email survey among CCO Regional Primary Care Leads and OBSP Regional Program Managers.

Detection method, diagnostic interval and DAU use definitions

We determined screen-detected cancers by looking back 12 months from the date of the cancer diagnosis for abnormal OBSP screens and 6 months for opportunistic screens. We assumed a patient presented with signs or symptoms if her cancer was not screen-detected. We shortened the look-back time window to 6 months from 12 months when identifying opportunistic screens because those screens were identified using billing claims, which do not contain test results (which were available in the OBSP database). This 6-month window was based on Canadian guideline evidence^{Footnote 9Footnote 19} and our observation that fewer than 5% of abnormal OBSP screens occurred in the 6 to 12 months before diagnosis.^{Footnote 15}

We defined the diagnostic interval as the number of days from the date of the cancer diagnosis back to the initial screen, or referral to a specialist or the first diagnostic test in the absence of a screening test (Figure 1). This strategy is an adaptation of one used in colorectal cancer.^{Footnote 20} The cancer diagnosis date was contained in the Ontario Cancer Registry and is normally the first occurrence of a histology- or cytology-confirmed malignancy or a hospital admission or outpatient consultation. The diagnostic interval was analyzed as both a continuous and a dichotomous outcome at 7 weeks, with the latter based on the longer of the two timeliness targets mentioned above.^{Footnote 9Footnote 19}

Source: Based on methodology outlined in Singh et al.²⁰

Text Description

We defined the diagnostic interval as the number of days from the date of the cancer diagnosis back to the initial screen, or referral to a specialist or the first diagnostic test in the absence of a screening test. This figure is a flow chart illustrating this process and is replicated in the table below:

Index contact date	Visit date	Claim	Treatment date	Date of diagnosis
Index contact date = the earliest of A (if A did not exist, take the earliest of B, C, D, E, F or G)	A. Date of the first OHIP screening mammogram in 6 months before diagnosis or the rst abnormal OBSP screening test in 12 months before diagnosis	N/A	Date of the first OHIP screening mammogram in 6 months before diagnosis or the ­first abnormal OBSP screening test in 12 months before diagnosis	Date of the breast cancer diagnosis
Index contact date = the earliest of A (if A did not exist, take the earliest of B, C, D, E, F or G)	B. Date of the last visit with the ordering physician prior to the date of diagnostic mammogram	Claim # of the ordering physician	Date of the first diagnostic mammogram in 6 months prior to diagnosis	Date of the breast cancer diagnosis
Index contact date = the earliest of A (if A did not exist, take the earliest of B, C, D, E, F or G)	C. Date of the last visit with the ordering physician prior to the date of breast ultrasound	Claim # of the ordering physician	Date of the first breast ultrasound in 6 months prior to diagnosis	Date of the breast cancer diagnosis
Index contact date = the earliest of A (if A did not exist, take the earliest of B, C, D, E, F or G)	D. Date of the last visit with the ordering physician prior to the date of breast MRI	Claim # of the ordering physician	Date of the first breast MRI in 6 months prior to diagnosis	Date of the breast cancer diagnosis
Index contact date = the earliest of A (if A did not exist, take the earliest of B, C, D, E, F or G)	E. Date of the last visit with the ordering physician prior to the date of breast biopsy	Claim # of the ordering physician	Date of the fi­rst breast biopsy in 6 months prior to diagnosis	Date of the breast cancer diagnosis
Index contact date = the earliest of A (if A did not exist, take the earliest of B, C, D, E, F or G)	F. Date of the last visit with the ordering physician prior to the date of breast surgeon consultation	Claim # of the ordering physician	Date of the first surgeon consultation in 6 months prior to diagnosis	Date of the breast cancer diagnosis
Index contact date = the earliest of A (if A did not exist, take the earliest of B, C, D, E, F or G)	G. Date of the first breast cancer-related ED visit in 6 months prior to diagnosis	N/A	Date of the first breast cancer-related ED visit in 6 months prior to diagnosis	Date of the breast cancer diagnosis

We determined DAU use for patients who had an initial screen within the OBSP (OBSP patients) separately from the rest, because the OBSP has a database that tracks the use of the Breast Assessment Affiliates. The OBSP patients were considered to be diagnosed through a DAU if they had either an OBSP BAA payment record or a biopsy/surgery performed at a DAU hospital. Otherwise, they were considered to be diagnosed through the UC. The remaining patients (opportunistically screened and symptomatic patients) were considered to be diagnosed through a DAU if they had a biopsy/surgery performed at a DAU hospital. We validated this strategy using the OBSP record as the criterion standard and found that using the biopsy/surgery hospital had a sensitivity of 90.1% and a specificity of 84.6%.^{Footnote 15}

Statistical analysis

We used proportions to describe the breast cancer screening detection rates and DAU use. Screening detection rates were categorized into three age groups (< 50 years, 50–69 years, > 69 years) based on the most recent Canadian guideline at the time of this study, although the updated guideline increased the upper age limit for screening to 74 in November of 2011.^{Footnote 21} We report the median diagnostic interval because its distribution is right-skewed. We report the diagnostic interval and DAU coverage by Local Health Integration Network (LHIN) and by county, both based on the patient's residence location at diagnosis. The county-level diagnostic interval was categorized for mapping purposes with category cut-points chosen to ensure a reasonable number of counties within each category. DAU service coverage was also categorized for mapping. To assess whether adjacent counties could be combined, we calculated a spatial autocorrelation on the diagnostic interval using the Global Moran's I statistic, which takes the values between −1 (perfect spatial dispersion) to 1 (perfect spatial clustering), with 0 representing a random geographic pattern.^{Footnote 22} For the county-level analysis, we conducted univariable and multivariable median regressions to assess the unadjusted and stage-adjusted geographic variation in the diagnostic interval because earlier stage at diagnosis has repeatedly been shown to be associated with a longer diagnostic interval.^{Footnote 23} We received ethics approval from the Health Sciences and Affiliated Teaching Hospitals Research Ethics Board at Queen's University at Kingston, Ontario, Canada.

Results

The final cohort size was 6898. The flow chart depicting the cohort selection process is shown in Figure 2. The mean age was 61.2 (SD: 13.5).

Text Description

The final cohort size was 6898. The flow chart depicting the cohort selection process is shown in this figure, as follows:

Patients who met the inclusion criteria:

• female
• invasive breast cancer
• a single primary cancer
• diagnosed between 1 January, 2011, and 31 December, 2011 (n = 8719)

Patients excluded

• whose cancer registry record could not be linked to other administrative data (n = 46)
• who were living outside of Ontario at the time of diagnosis (n = 674)
• who did not have an OHIP coverage for at least three years prior to diagnosis (n = 169)

Eligible study cohort (n = 7830)

Further exclusions

• missing index contact dates (n = 271)
• unable to assign a DAU use (n = 661)

Final study cohort (n = 6898)

Of the 6898 patients, 2499 (36.2%) were screen-detected and 4399 (63.8%) were symptomatic. Table 1 reports screening detection rates by age group. Of screen-detected patients, 1986 (79.5%) were screened via the OBSP and 513 (20.5%) were screened opportunistically.

Table 1. Breast cancer detection method rates by age group, Ontario, Canada, 2011 (N = 6898)

Age group	Screen-detected	Symptomatic
	Total (row %)	Total (row %)
< 50 years	158 (11.1)	1270 (88.9)
50–69 years	1736 (48.4)	1851 (51.6)
> 69 years	605 (32.1)	1278 (67.9)
All ages (total)	2499 (36.2)	4399 (63.8)

The provincial median time to diagnosis was 32 days (interquartile range: 17–60). Sixty-eight percent of breast cancer patients were diagnosed within the 7-week target, while 10% waited 107 days or more. Nearly half (48.4%) of patients were diagnosed at a DAU.

Table 2 describes the diagnostic interval and DAU use by LHIN. The median diagnostic interval ranged from 24 days (95% CI: 20.6–27.4) in the South East LHIN to 41 days (95% CI: 34.7–47.3) in the North West LHIN (p < .001). The percent of women diagnosed through a DAU varied from 2.6% (95% CI: 0.7%–4.6%) in the North Simcoe Muskoka LHIN to 93% (95% CI: 90.2%– 95.8%) in the South East LHIN (p < .001), reflecting regional differences in DAU availability.

Table 2. Descriptive statistics of the breast cancer diagnostic interval and diagnostic assessment unit coverage by Local Health Integration Network, Ontario, Canada, 2011 (N = 6898)

LHIN	n	Diagnostic interval median (95% CI)	DAU coverage rateTable 2 Footnote a (95% CI)
Erie St. Clair	412	27.5 (25.1–30.0)	81.6 (77.8–85.3)
South West	594	35.0 (30.6–39.4)	81.6 (78.5–84.8)
Waterloo Wellington	380	30.0 (27.9–32.1)	51.3 (46.3–56.3)
Hamilton Niagara Haldimand Brant	697	29.0 (27.7–30.3)	53.4 (49.7–57.1)
Central West	311	37.0 (33.7–40.3)	12.9 (9.1–16.6)
Mississauga Halton	571	36.0 (32.4–39.6)	29.1 (25.3–32.8)
Toronto Central	553	34.0 (30.8–37.2)	37.4 (33.4–41.5)
Central	863	36.0 (33.3–38.7)	21.2 (18.5–23.9)
Central East	740	35.5 (34.3–36.6)	38.0 (34.5–41.5)
South East	315	24.0 (20.6–27.4)	93.0 (90.2–95.8)
Champlain	758	29.0 (27.8–30.2)	69.5 (66.2–72.8)
North Simcoe Muskoka	266	29.8 (23.7–35.8)	2.6 (0.7–4.6)
North East	316	28.0 (25.3–30.7)	45.6 (40.1– 51.1)
North West	120	41.0 (34.7–47.3)	80.8 (73.8–87.9)
Unknown/shared LHINs	—	—	—
Abbreviations: CI, confidence interval; DAU, diagnostic assessment unit; ICES, Institute for Clinical Evaluative Services; LHIN, Local Health Integration Network. Note: — indicates cells with counts less than 10, which were not reported due to ICES privacy regulations. Table 2 Footnote a Calculated as the percent of patients diagnosed through DAU for each LHIN. Table 2 Return to footnote a referrer

Figure 3 further maps the median diagnostic interval across Ontario counties. There was no evidence of a spatial correlation in the diagnostic interval (Global Moran's I = −0.02, p = .95), indicating that grouping neighbouring counties, as is done when reporting by LHIN, masks variation. Counties with the longest diagnostic intervals were scattered across the province, and patients in the most populous area (Greater Toronto Area) experienced waits in the second longest category. As Table 3 shows, 13 counties out of 47 (two counties with small numbers are not reported) had a median diagnostic interval greater than 7 weeks, and the median ranged from 15 days (95% CI: 7.6–22.4) to 65 days (95% CI: 39.6–90.4). This variation did not materially change with stage adjustment (results not shown).

Text Description

Figure 3 data

County	Diagnostic interval, in days (95% confidence interval)
Brant County	31.0 (23.8–38.2)
Bruce County	21.9 (14.6–29.2)
Dufferin County	46.9 (32.5–61.3)
Elgin County	49.0 (36.2–61.8)
Essex County	25.0 (20.8–29.2)
Frontenac County	21.0 (17.7–24.3)
Grey County	30.0 (23.3–36.7)
Haldimand-Norfolk Regional Municipality	37.8 (32.6–43.0)
Haliburton County	—
Halton Regional Municipality	36.0 (30.1–41.9)
Hastings County	22.0 (16.4–27.6)
Huron County	30.1 (18.0–42.3)
Chatham-Kent Division	40.5 (23.9–57.1)
Lambton County	25.4 (15.5–35.3)
Lanark County	38.0 (30.5–45.5)
Leeds and Grenville United Counties	31.7 (26.1–37.3)
Lennox and Addington County	27.0 (9.5–44.5)
Toronto Division	38.0 (36.0–40.0)
Middlesex County	42.0 (36.6–47.4)
Muskoka District Municipality	65.0 (39.6–90.4)
Niagara Regional Municipality	35.0 (31.3–38.7)
Northumberland County	20.0 (13.7–26.3)
Durham Regional Municipality	33.0 (28.6–37.4)
Ottawa Division	28.0 (24.2–31.8)
Oxford County	16.4 (10.7–22.1)
Peel Regional Municipality	37.0 (34.0–40.0)
Perth County	23.0 (15.5–30.5)
Peterborough County	20.0 (14.3–25.7)
Prescott and Russell United Counties	48.7 (36.9–60.5)
Prince Edward Division	15.0 (7.6–22.4)
Renfrew County	15.0 (6.8–23.2)
Simcoe County	24.0 (19.4–28.6)
Stormont, Dundas, Glengary United Counties	40.8 (32.3–49.2)
Kawartha Lakes Division	23.0 (10.6–35.4)
Waterloo Regional Municipality	30.0 (26.8–33.2)
Wellington County	32.0 (27.5–36.5)
Hamilton Division	19.7 (16.5–22.9)
York Regional Municipality	32.0 (28.8–35.2)
Algoma District	24.0 (13.5–34.5)
Cochrane District	35.0 (28.0–41.9)
Kenora District	50.0 (30.8–69.2)
Manitoulin District	19.7 (6.3–33.2)
Nipissing District	22.0 (13.2–30.8)
Parry Sound District	32.5 (19.6–45.3)
Rainy River District	—
Greater Sudbury Division	24.0 (18.1–29.9)
Sudbury District	22.0 (2.9–41.1)
Thunder Bay District	40.0 (34.8–45.2)
Timiskaming District	(14.2–91.8)
– indicates cell counts less than 10, which are not reported due to privacy regulations.

Table 3. Descriptive statistics of the breast cancer diagnostic interval and diagnostic assessment unit coverage by county, Ontario, Canada, 2011 (N = 6898)

County	n	Diagnostic interval median (95% CI)	DAU coverage rateTable 3 Footnote a (95% CI)
Brant CountyTable 3 Footnote b	60	31.0 (23.8–38.2)	6.6 (0.4–13.0)
Bruce CountyTable 3 Footnote b	46	21.9 (14.6–29.2)	73.9 (61.2–86.6)
Dufferin CountyTable 3 Footnote b	30	46.9 (32.5–61.3)	6.7 (−2.3–15.59)
Elgin CountyTable 3 Footnote b	59	49.0 (36.2–61.8)	18.6 (8.7–28.6)
Essex County	251	25.0 (20.8–29.2)	85.7 (81.3–90.0)
Frontenac County	89	21.0 (17.7–24.3)	98.9(96.7–101.1)
Grey County	63	30.0 (23.3–36.7)	76.2 (65.7–86.7)
Haldimand-Norfolk Regional MunicipalityTable 3 Footnote b	68	37.8 (32.6–43.0)	17.6 (8.6–26.7)
Haliburton CountyTable 3 Footnote b	—	—	—
Halton Regional MunicipalityTable 3 Footnote b	215	36.0 (30.1–41.9)	26.0 (20.2–31.9)
Hastings County	89	22.0 (16.4–27.6)	96.6 (92.9–100.4)
Huron CountyTable 3 Footnote b	36	30.1 (18.0–42.3)	88.9 (78.6–99.2)
Chatham-Kent Division	62	40.5 (23.9–57.1)	91.9 (85.2–98.7)
Lambton County	98	25.4 (15.5–35.3)	64.3 (54.8–73.8)
Lanark CountyTable 3 Footnote b	41	38.0 (30.5–45.5)	53.7 (38.4–68.9)
Leeds and Grenville United Counties	68	31.7 (26.1–37.3)	86.8 (78.7–94.8)
Lennox and Addington CountyTable 3 Footnote b	19	27.0 (9.5–44.5)	100.0 (100.0–100.0)
Toronto Division	1289	38.0 (36.0–40.0)	25.5 (23.1–27.9)
Middlesex County	285	42.0 (36.6–47.4)	94.0 (91.3–96.8)
Muskoka District MunicipalityTable 3 Footnote b	49	65.0 (39.6–90.4)	0.0 (0.0–0.0)
Niagara Regional Municipality	241	35.0 (31.3–38.7)	90.9 (87.2–94.5)
Northumberland County	57	20.0 (13.7–26.3)	94.7 (88.9–100.5)
Durham Regional Municipality	283	33.0 (28.6–37.4)	54.1 (48.3–59.9)
Ottawa Division	549	28.0 (24.2–31.8)	74.7 (71.0–78.3)
Oxford County	62	16.4 (10.7–22.1)	85.5 (76.7–94.3)
Peel Regional Municipality	567	37.0 (34.0–40.0)	27.5 (23.8–31.2)
Perth County	43	23.0 (15.5–30.5)	100.0 (100.0–100.0)
Peterborough County	73	20.0 (14.3–25.7)	41.1 (29.8–52.4)
Prescott and Russell United Counties	52	48.7 (36.9–60.5)	48.1 (34.5–61.7)
Prince Edward Division	19	15.0 (7.6–22.4)	94.7 (84.7–104.8)
Renfrew County	51	15.0 (6.8–23.2)	90.2 (82.0–98.4)
Simcoe County	245	24.0 (19.4–28.6)	4.1 (1.6–6.6)
Stormont, Dundas, Glengary United Counties	74	40.8 (32.3–49.2)	36.5 (25.5–47.5)
Kawartha Lakes Division	45	23.0 (10.6–35.4)	28.9 (15.6–42.1)
Waterloo Regional Municipality	281	30.0 (26.8–33.2)	61.9 (56.2–67.6)
Wellington CountyTable 3 Footnote b	94	32.0 (27.5–36.5)	19.1 (11.2–27.1)
Hamilton Division	266	19.7 (16.5–22.9)	41.0 (35.1–46.9)
York Regional Municipality	528	32.0 (28.8–35.2)	23.9 (20.2–27.5)
Algoma District	68	24.0 (13.5–34.5)	88.2 (80.6–95.9)
Cochrane District	42	35.0 (28.0–41.9)	38.1 (23.4–52.8)
Kenora DistrictTable 3 Footnote b	31	50.0 (30.8–69.2)	41.9 (24.6–59.3)
Manitoulin DistrictTable 3 Footnote b	14	19.7 (6.3–33.2)	50.0 (23.8–76.2)
Nipissing DistrictTable 3 Footnote b	53	22.0 (13.2–30.8)	13.2 (4.1–22.3)
Parry Sound DistrictTable 3 Footnote b	28	32.5 (19.6–45.3)	17.9 (3.7–32.0)
Rainy River DistrictTable 3 Footnote b	—	—	—
Greater Sudbury Division	87	24.0 (18.1–29.9)	46.0 (35.5–56.4)
Sudbury DistrictTable 3 Footnote b	15	22.0 (2.9–41.1)	40.0 (15.2–64.8)
Thunder Bay District	84	40.0 (34.8–45.2)	97.6 (94.4–100.9)
Timiskaming DistrictTable 3 Footnote b	11	53.0 (14.2–91.8)	36.4 (7.9–64.8)
Note: — indicates cells with counts less than 10, which were not reported due to Institute for Clinical Evaluative Services (ICES) privacy regulations. Table 3 Footnote a Calculated as the percent of patients diagnosed through DAUs for each county. Table 3 Return to footnote a referrer Table 3 Footnote b Counties without a DAU. Table 3 Return to footnote b referrer

Figure 4 maps county-level DAU service use across Ontario. Rates range from 0% to 100%, partially reflecting that, at the time of this study, DAUs did not exist in 19 (38.8%) counties (see Table 3). Of these 19, 11 had a median diagnostic interval greater than or equal to the provincial median.

Text Description

Figure 4 data

County	Diagnostic assessment unit coverage rate (95% confidence interval)
Brant County	6.6 (0.4–13.0)
Bruce County	73.9 (61.2–86.6)
Dufferin County	6.7 (−2.3–15.59)
Elgin County	18.6 (8.7–28.6)
Essex County	85.7 (81.3–90.0)
Frontenac County	98.9 (96.7–101.1)
Grey County	76.2 (65.7–86.7)
Haldimand-Norfolk Regional Municipality	17.6 (8.6–26.7)
Haliburton County	—
Halton Regional Municipality	26.0 (20.2–31.9)
Hastings County	96.6 (92.9–100.4)
Huron County	88.9 (78.6–99.2)
Chatham-Kent Division	91.9 (85.2–98.7)
Lambton County	64.3 (54.8–73.8)
Lanark County	53.7 (38.4–68.9)
Leeds and Grenville United Counties	86.8 (78.7–94.8)
Lennox and Addington County	100.0 (100.0–100.0)
Toronto Division	25.5 (23.1–27.9)
Middlesex County	94.0 (91.3–96.8)
Muskoka District Municipality	0.0 (0.0–0.0)
Niagara Regional Municipality	90.9 (87.2–94.5)
Northumberland County	94.7 (88.9–100.5)
Durham Regional Municipality	54.1 (48.3–59.9)
Ottawa Division	74.7 (71.0–78.3)
Oxford County	85.5 (76.7–94.3)
Peel Regional Municipality	27.5 (23.8–31.2)
Perth County	100.0 (100.0–100.0)
Peterborough County	41.1 (29.8–52.4)
Prescott and Russell United Counties	48.1 (34.5–61.7)
Prince Edward Division	94.7 (84.7–104.8)
Renfrew County	90.2 (82.0–98.4)
Simcoe County	4.1 (1.6–6.6)
Stormont, Dundas, Glengary United Counties	36.5 (25.5–47.5)
Kawartha Lakes Division	28.9 (15.6–42.1)
Waterloo Regional Municipality	61.9 (56.2–67.6)
Wellington County	19.1 (11.2–27.1)
Hamilton Division	41.0 (35.1–46.9)
York Regional Municipality	23.9 (20.2–27.5)
Algoma District	88.2 (80.6–95.9)
Cochrane District	38.1 (23.4–52.8)
Kenora District	41.9 (24.6–59.3)
Manitoulin District	50.0 (23.8–76.2)
Nipissing District	13.2 (4.1–22.3)
Parry Sound District	17.9 (3.7–32.0)
Rainy River District	—
Greater Sudbury Division	46.0 (35.5–56.4)
Sudbury District	40.0 (15.2–64.8)
Thunder Bay District	97.6 (94.4–100.9)
Timiskaming District	36.4 (7.9–64.8)

Discussion

To the best of our knowledge, this is the first Ontario study and among the first Canadian studies describing breast cancer screening detection rates and waiting time to diagnosis in a population-based breast cancer cohort.^{Footnote 24Footnote 25} Previous Ontario reports have been restricted to those seen within organized breast cancer screening programs^{Footnote 12Footnote 26} while our study also includes the 63.8% of breast cancer patients who were diagnosed symptomatically and the 7.4% detected through opportunistic screening.

Our observed screening detection rate (48.4% in those breast cancer patients aged 50–69 years) was lower than the general population screening participation rates reported by the OBSP. In calendar years 2007 and 2008, the OBSP calculated that 66.3% of the Ontario women aged 50 to 69 years had been screened (through OBSP or opportunistic screening),^{Footnote 27} and the coverage was estimated to be similar in 2011.^{Footnote 3} The discrepancy between screening program participation rates and screening detection rates aligns with evidence that those at higher risk with breast cancer risk factors such as higher age and lack of exercise are less likely to get screened^{Footnote 28} and that some women are being diagnosed between screenings. To this latter point, Kirsh et al. estimated that 13.8% of breast cancer patients nested within a cohort of OBSP participants from 1994 to 2002 were diagnosed within 24 months after a negative screening mammogram (known as interval cancers).^{Footnote 29}

The only other Canadian evidence on the breast cancer screening detection rate comes from two recent studies in Alberta, with one reporting that 44% of all breast cancers were detected through screening from 2007 to 2010^{Footnote 24} and the other reporting a screening detection rate of 38% from 2004 to 2010^{Footnote 25} (we observed 36% in Ontario in 2011). Screening eligibility criteria differ between the two provinces, as the Alberta Breast Cancer Screening Program additionally accepts patients aged 40 to 49 years.^{Footnote 30} However, the biennial breast cancer screening participation rate (programmatic and nonprogrammatic) for women aged 50 to 69 years is lower in Alberta (55.1%–57.3% in 2007–2010)^{Footnote 31} than Ontario (66.3% in years 2007–2008).^{Footnote 27}

Two population-based studies conducted in United States reported the proportions of screen-detected cancer at 22% and 30.2%, respectively,^{Footnote 32Footnote 33} with the latter restricted to women aged 40 to 49 years. These lower rates may have changed since these studies were conducted in the 1990s. A more recent study in Mexico^{Footnote 34} reported a screening detection rate of 58.3% among breast cancer patients of all ages. This higher rate may be partly attributed to a more liberal definition of “screen-detected cancer”: cancer diagnosed within 270 days of a bilateral (both breasts) screening mammogram.

Waiting for a diagnosis can have considerable psychological consequences for patients and their families and wait time reductions can significantly reduce anxiety levels.^{Footnote 6} In Canada, there is little guidance about breast cancer diagnostic wait times; breast cancer screening timeliness targets at the time of this study recommended that 90% of abnormal screenings should be resolved within 5 weeks, or within 7 weeks if a tissue biopsy is performed.^{Footnote 9} Applying the more liberal 7-week cut, we observed that 68.3% of breast cancer patients in our study were diagnosed within 7 weeks. Our observed rate is similar to a 64% rate reported in 2010 for resolution of all Ontario abnormal breast cancer screens requiring a tissue biopsy.^{Footnote 30}

In our study, the median breast cancer diagnostic interval for Ontario in 2011 was 4.6 weeks (32 days), and 690 patients (10%) waited 3.6 months or longer, which is a level of delay that has been associated with worse survival.^{Footnote 8} Our observed median was shorter than the reported median time from an abnormal screen to diagnosis requiring tissue biopsy for other Canadian provinces in the year 2010 (range: 5–7 weeks).^{Footnote 30} This is likely explained by our restriction to the cancer population. We reported the diagnostic interval separately for the screen-detected and the symptomatic breast cancer patients in two companion papers,^{Footnote 14Footnote 16} with medians of 29 days and 34 days, respectively. Our observations differed from findings in Alberta, which reported medians of 19 days for the screen-detected patients and 21 days for the symptomatic patients in the years 2004 to 2010.^{Footnote 25}

Within Ontario, we observed significant regional variation in the diagnostic interval. The maximum difference was 2.4 weeks (17 days) between LHINs. Women who live in the North West, Central West and Mississauga Halton LHINs experienced the longest waits, exceeding 5 weeks. In 2011, the OBSP reported that these regions were among those that had the lowest 7-week tissue biopsy target rates.^{Footnote 3} Reasons for these low rates likely vary across these three LHINs as they are very different geographically and demographically. Further study is required to better understand these differences.

Compared to the LHIN-level analysis, we observed greater variation between counties; there was a maximum difference of 7.1 weeks (50 days) between counties, indicating intra-LHIN variation. Consistent with the LHIN-level findings, the longest diagnostic intervals were observed in both the most populous (Greater Toronto Area) and the least populous (northern Ontario) areas. We also observed significant variation in DAU service use across counties. The county with 0% of DAU use had the longest diagnostic interval at both the median (65 days) and 75th percentile (108 days), and the county where 100% of patients used DAU service had a median diagnostic interval (27 days) that was among the shortest.

We previously hypothesized that the variation in the diagnostic interval may be partially attributable to different DAU service availabilities across regions.^{Footnote 14Footnote 15} We found that DAU use was associated with reduced time-to-diagnosis for both screen-detected and symptomatic patients after control for potential confounders in multivariable analyses.^{Footnote 14Footnote 16} It should be noted that other regional factors such as patient demographics, care referral processes and local programs may have contributed to the variation that we have observed, since regions with high DAU coverage are not consistently associated with shorter diagnostic intervals. The World Health Organization has identified geographic variation as a health care quality concern pertinent to two dimensions: equity and accessibility.^{Footnote 35} Therefore, the reasons for this observed variation warrant further investigation in order to improve access and equity of breast cancer diagnostic care.  

Strengths and limitations

This study has several strengths. We have demonstrated the feasibility of using routinely collected health administrative data to understand the breast cancer diagnostic process. Our approach provides methodological guidance for determining breast cancer detection method, diagnostic interval and use of specialized breast diagnostic assessment units that may be adapted for use over time in Ontario and in other Canadian provinces. This is the first Ontario study to provide population-based evidence on the breast cancer screening detection rate, diagnostic interval and use of diagnostic assessment units. This population-wide information highlighted the low screening detection rate (36%) when breast cancer patients of all ages are considered, and it provides a full picture of the breast cancer experience by including those who are opportunistically screened and those who presented symptomatically: groups who are routinely excluded from breast cancer diagnostic studies. Therefore, our findings provide a unique population perspective on the role of breast cancer screening in cancer detection, access to DAUs and the amount of time it takes for a diagnosis, which can inform cancer program policies and interventions and their evaluation. The geographic variations we observed provide important information for system administrators to compare across regions and identify successful regional initiatives for wider implementation. This information could prompt knowledge sharing and facilitate collaboration across regions aimed at improving the timeliness of care and, ultimately, patients’ experience. The existence of such wide variation at a county level is a caution to other health care systems.

This study also has a number of limitations. First, we did not estimate the time-to-diagnosis or time-to-resolution interval for all patients being investigated for breast cancer, but instead focussed on the group who were ultimately diagnosed with the disease. Although patients under investigation are identifiable in the screening program, they are much harder to find in administrative data, which we needed in order to include patients who presented symptomatically. Evidence suggests that women with invasive breast cancer get a quicker diagnosis compared to those with benign diseases,^{Footnote 36Footnote 37} so the intervals we report are likely shorter than the benign group would experience and therefore apply only to women with breast cancer.

Second, we did not have test results in the administrative databases, so a small portion of the 7.4% whom we considered as being opportunistically screened may have actually had a negative screening mammogram, making our estimate of their diagnostic interval longer than it actually was and erroneously labelling that small group as having been screen-detected. Conversely, a small proportion of apparently symptomatic patients may have had an abnormal opportunistic screening test more than 6 months before diagnosis, thereby underestimating their diagnostic interval (and erroneously labelling them as having been symptomatic at presentation).

Third, defining the symptomatic index contact date as the most recent referring physician visit preceding the earliest test procedure was a conservative choice regarding the calculation of the diagnostic interval because the actual referral may have occurred earlier.

Fourth, new physician billing codes that specify the reason for the mammogram (symptomatic vs. asymptomatic) were introduced on 1 October, 2010, and may not have been fully adopted by the time of our study. Therefore, some opportunistically screened patients may be misclassified as symptomatic and thus we may have underestimated the proportion of screen-detected patients. We expect this influence to be small, because the frequency of use of this new code increased dramatically during the first three months of its introduction and had levelled off by the time of this study.^{Footnote 38}

Finally, the determination of DAU use was subject to misclassification, with the sensitivity estimated at 90.1% and specificity at 84.6%.^{Footnote 15}

Conclusion

This report provides diagnostic access information about those who are ultimately diagnosed with a breast cancer. Among age-eligible breast cancer patients, we observed lower screening detection rates than screening participation rates in the general population. This result suggests that future cancer screening efforts should target women at increased risk of breast cancer and/or that the rate of interval cancers is in need of improvement. We also observed considerable variation in the length of diagnostic interval and DAU across geographic regions. There is a need for further research to understand the factors attributable to this variation and identify opportunities for system improvement. Finally, future efforts to understand diagnostic wait times and the influence of programs designed to mitigate them must occur at the population level rather than be restricted to organized screening programs.

Acknowledgements

The authors thank Ms. Marlo Whitehead for data linkage and data preparation, Dr. Marcy Winget for methodological advice, Mr. Peter Gozdyra for assistance with the geographic mapping and Ms. Melissa Enmore for assistance in circulating the survey. This research is funded by a grant from the Canadian Institutes of Health Research. Li Jiang was supported by an Ontario Graduate Scholarship and a Queen’s University Graduate Award. Parts of this material are based on data and information provided by CCO. The opinions, results, view and conclusions reported in this paper are those of the authors and do not necessarily reflect those of CCO. No endorsement by CCO is intended or should be inferred. Parts of this material are based on data and information compiled and provided by CIHI. However, the analyses, conclusions, opinions and statements expressed herein are those of the authors, and not necessarily those of CIHI. This study was supported by the Institute for Clinical Evaluative Sciences (ICES), which is funded by an annual grant from the Ontario Ministry of Health and Long-Term Care (MOHLTC). The opinions, results and conclusions reported in this paper are those of the authors and are independent of the funding sources. No endorsement by ICES or the Ontario MOHLTC is intended or should be inferred.

Conflicts of interest

The authors declare no conflicts of interest.

Authors’ contributions and statement

All authors contributed to study design, results interpretation and manuscript revision. PG oversaw the research methodology. LJ performed analysis and had a major role in drafting and revising the manuscript.

The content and views expressed in this article are those of the authors and do not necessarily reflect those of the Government of Canada.

References