Validation of an Algorithm to Ascertain Late Breast Cancer Recurrence Using Danish Medical Registries

Figures & data

Figure 1 Flowchart of study populations.

Abbreviations: CBC, contralateral breast cancer; ER, estrogen receptor.

Figure 2 Overview of the algorithm.

Abbreviation: CBC, contralateral breast cancer.

Table 1 Descriptive Characteristics on Patients Diagnosed with Stage I–III Operable Breast Cancer Registered in the DBCG Between January 1, 1987 and December 31, 2004 Who Were Alive, Living in Denmark and Without a Recurrence or Second Cancer 10 Years After Diagnosis, According to Study Population

Characteristics	Study Population I^a	Study Population II^b
Total Numbers of Patients	105 (100%)	114 (100%)
Age at primary breast cancer diagnosis (years)
<40	7 (6.7)	<5
40–49	38 (36.2)	22 (19.3)
50–59	40 (38.1)	39 (34.2)
60–69	15 (14.3)	38 (33.3)
≥70	5 (4.7)	<15
Age, median (years)	52	57
Calendar period of primary breast cancer
1987–1991	23 (21.9)	15 (13.2)
1992–1996	33 (31.4)	24 (21.1)
1997–2000	30 (28.6)	26 (22.8)
2001–2004	19 (18.1)	49 (43.0)
Menopausal status at primary breast cancer diagnosis
Premenopausal	48 (45.7)	27 (23.7)
Postmenopausal	46 (43.8)	77 (67.5)
Unknown	11 (10.5)	10 (8.8)
Charlson Comorbidity Index Score
0	61 (58.1)	45 (39.5)
1–2	38 (36.2)	63 (55.3)
≥3	6 (5.7)	6 (5.2)
Stage^c
I	45 (42.8)	49 (43.0)
II	54 (51.4)	60 (52.6)
III	<10	5 (4.4)
Unknown	<5	–
Grade
Low	29 (27.6)	32 (28.1)
Moderate	45 (42.9)	42 (36.8)
High	11 (10.5)	14 (12.3)
Unknown	20 (19.0)	26 (22.8)
Number of positive lymph nodes
0	58 (55.2)	74 (64.9)
1–3	42 (40.0)	35 (30.7)
≥4	5 (4.8)	5 (4.4)
Tumor size
<20mm	72 (68.6)	75 (65.8)
>20mm	33 (31.4)	39 (34.2)
ER status
ER+	60 (57.1)	70 (61.4)
ER-	9 (8.6)	12 (10.5)
Unknown	36 (34.3)	32 (28.1)
Type of primary surgery
Mastectomy	39 (37.1)	62 (54.4)
Mastectomy + RT	18 (17.2)	20 (17.5)
BCS + RT	48 (45.7)	32 (28.1)
Adjuvant chemotherapy received
No	73 (69.5)	93 (81.6)
Yes	32 (30.5)	21 (18.4)
Endocrine therapy received
No	68 (64.8)	62 (54.4)
Yes	37 (35.2)	52 (45.6)

Notes: Cell sizes less than 5 are reported in aggregate to reduce identifiability of individuals in the data. ^aWomen diagnosed with a late recurrence according to our algorithm at Aarhus University Hospital. ^bWomen with primary breast cancer diagnosed at Aalborg University Hospital. ^cStage was calculated using the TNM staging system.

Abbreviations: DBCG, Danish Breast Cancer Group; ER, estrogen receptor status; BCS, breast conserving surgery; RT, radiation therapy.

Table 2 Positive Predictive Values (PPVs) and Corresponding 95% Confidence Intervals (CIs) of Late Breast Cancer Recurrence Identified by Our Algorithm During Follow-Up

	Confirmed Cases/Potential Cases	PPV (%)	95% CI
Algorithm criteria, overall	90/105	85.7	77.5–91.3
DNPR-registered metastases^a	63/69	91.3	81.6–96.1
Pathology codes for recurrence^b	63/66	95.5	86.5–98.6
DNPR-registered treatment codes^c	81/92	88.0	79.5–93.3
Neo-adjuvant/adjuvant treatment	75/80	93.8	85.6–97.4
Surgery (mastectomy, BCS, etc.)	16/23	69.6	46.6–85.7
Specific codes for recurrence in DNPR^d	14/14	100.0	–
Local breast cancer recurrence	13/13	100.0	–
Recurrence surgery	<5/<5	100.0	–

Notes: Cell sizes less than 5 are reported in aggregate to reduce identifiability of individuals in the data. ^aICD10 DC76-DC80. ^bT code in the breast with morphology codes M8 or M9 with 4, 6, 7 or 9 in the fifth position (eg, M8XXX4), or any T code (excluding the breast T codes) with morphology codes M8 or M9 with the numbers 6 or 9 in the fifth position. ^cSKS treatment codes: BWG, BWHA, BWHC, BHHH, BOHJ13; Danish version of the Nordic Medico-Statistical Committee Classification of Surgical Procedures: KHAC, KHAB, KGAE16. ^dICD10 DC509X; Danish version of the Nordic Medico-Statistical Committee Classification of Surgical Procedures: KHAF.

Abbreviations: DNPR, Danish National Patient Registry; BCS, breast conserving surgery; PPV, positive predictive value.

Table 3 Estimation of Sensitivity, Specificity, Negative Predictive Value and Likelihood Ratios Among Women Diagnosed with Primary Breast Cancer at Aalborg University Hospital, 1987–2004

Algorithm	Medical Record Review
	Late Recurrence	No Late Recurrence	In Total
Late recurrence	<5	<5	7
No late recurrence	0	107	107
In total	<5	<114	114
Results (95% confidence intervals)
Sensitivity (TP/(TP+FN))	100.0% (39.8–100.0%)
Specificity (TN/(TN+FP))	97.3% (92.2–99.4%)
NPV (TN/(TN+FN))	100.0% (96.6–100.0%)
PLR (sensitivity/1- specificity)	36.7 (12.0–111.9)
NLR (1-sensitivity/specificity)	0

Notes: Cell sizes less than 5 are reported in aggregate to reduce identifiability of individuals in the data.

Abbreviations: TP, true positive; FN, false negative, TN; true negative; NPV, negative predictive value, PLR, positive likelihood ratio; NLR, negative likelihood ratio.