A Validated Algorithm for Register-Based Identification of Patients with Relapse of Clinical Stage I Testicular Cancer

Figures & data

Table 1 Pre-Specified and Refined Indicators of Testicular Cancer Relapse, and Different Versions of the Algorithm Based on the Various Indicators

Indicators	Source	Codes
Pre-specified
1	Danish National Pathology Registry	SNOMED combination codes 30 or more days after the orchiectomy: any T-code except T78*** (testis) and T79*** (para-testicular structures) in combination with the Æ-code ÆF4630 (metastasis with primary tumor in the testis) and/or any of the following M-codes: M9061* (seminoma), M9065* (seminoma with syncytiothrophoblasts), M9070* (embryonal carcinoma), M907x* (embryonal carcinoma with syncytiothrophoblasts), M9071* (yolk sac tumor), M908B* (teratoma), M9084* (teratoma with somatic type malignancy), M90800 (mature teratoma), M90803 (immature teratoma), M90801 (solid teratoma), M9100* (choriocarcinoma) or M908A* (mixed germ cell tumor) with the numbers 3,4,6,7 or 9 in the fifth digit, which indicates “tumor behavior” 3: “malignant, primary”, 4: “direct spread to surrounding tissue”, 6: “malignant, metastasis”, 7: “malignant, relapse/recurrence” and 9: “malignant”, “unknown if primary or metastatic”
2	Danish National Patient Register	Procedure codes for radiotherapy (BWG*) or cytostatic therapy (BWHA***) 30 or more days after the orchiectomy
3	Danish National Patient Register	Diagnosis codes of testicular cancer relapse (ICD-10: C629X) or metastases (ICD-10: C629M) 30 or more days after the orchiectomy
4	Danish National Patient Register	Diagnosis codes of distant metastasis (ICD-10: C76*-C79*) 30 or more days after the orchiectomy
Refined
2a	Danish National Patient Register	Procedure codes for radiotherapy (BWG*) 60 or more days after the orchiectomy, or cytostatic therapy (BWHA***) 30 or more days after the orchiectomy.
2b	Danish National Patient Register	Procedure codes for radiotherapy (BWG*) 60 or more days after the orchiectomy, or cytostatic therapy (BWHA***) 30 or more days after the orchiectomy – procedure codes combined with a simultaneously registered testicular cancer diagnosis code (ICD-10: C62*)
Different versions of the algorithm		Indicators
Algorithm 1 (pre-specified)		1,2,3 and 4
Algorithm 2		1,2 and 3
Algorithm 3		1 and 2
Algorithm 4		1 and 2a
Algorithm 5		1 and 2b

Abbreviations: ICD-10, International Classification of Disease, tenth revision; SNOMED, Systematized Nomenclature of Medicine; T, topography; M, morphology; Æ, etiology; F, function.

Table 2 Characteristics of the Study Cohort, and Stratified by Data Cohort

	Total Cohort	Development Cohort	Validation Cohort
Patients, n (%)	1377 (100)	250 (18)	1127 (82)
Median age at diagnosis, years (IQR)	37 (29–46)	38 (30–46)	37 (29–47)
Tumor subtype, n (%)
Seminoma	924 (67)	173 (69)	751 (67)
Non-seminoma	453 (33)	77 (31)	376 (33)
Median follow-up, years (IQR)	5.9 (4.6–7.4)	5.9 (4.8–7.7)	5.9 (4.5–7.4)
Relapse, n (%)	284 (20.6)	51 (20.4)	233 (20.6)

Abbreviation: IQR, interquartile range.

Figure 1 Flowchart of the study population.

Note: *Reactive metaplasia of epididymis.

Abbreviations: CSI, clinical stage I; CSIS, clinical stage IS; CSII, clinical stage II; CSIII, clinical stage III; DaTeCa database, Danish Testicular Cancer database; GCNIS, germ cell neoplasia in situ; GCT, germ cell tumor; RT, radiotherapy; TC, testicular cancer.

Table 3 Concordance of Relapse Identified by the Gold Standard and Registered in the DaTeCa Database (n = 1377)

		Relapse by Gold Standard
		Yes	No	Total
Relapse data in the DaTeCa database	Yes	276	1	277
	No	8	1092	1100
	Total	284	1093	1377

Notes: Positive predictive value: 99.6% (95% CI: 98.9–100). Negative predictive value: 99.3% (95% CI: 98.8–99.8). Completeness: 97.2% (95% CI: 95.2–99.1).

Abbreviations: CI, confidence interval; DaTeCa database, Danish Testicular Cancer database.

Table 4 Performance of Different Versions of the Algorithm to Identify Patients with Relapse from Stage I Testicular Cancer in the Development Cohort (n = 250)

Algorithm	Indicators	Sensitivity % (95% CI)	Specificity % (95% CI)	Positive Predictive Value % (95% CI)	Negative Predictive Value % (95% CI)
1	1,2,3,4	98.0 (94.2–100)	90.1 (87.0–94.9)	73.5 (63.0–84.0)	99.5 (98.4–100)
2	1,2,3	98.0 (94.2–100)	92.0 (88.2–95.7)	75.8 (65.4–86.1)	99.5 (98.4–100)
3	1,2	98.0 (94.2–100)	92.0 (88.2–95.7)	75.8 (65.4–86.1)	99.5 (98.4–100)
4	1,2a	98.0 (94.2–100)	95.5 (92.6–98.4)	84.8 (75.6–93.9)	99.5 (98.5–100)
5	1,2b	96.1 (90.1–100)	98.5 (96.8–100)	94.2 (87.9–100)	99.0 (97.6–100)

Notes: Indicator 1: pathology codes: SNOMED combinations, 30 days. Indicator 2: procedure codes: radiotherapy and cytostatic therapy, 30 days. Indicator 3: diagnosis codes: testicular cancer relapse and metastasis codes, 30 days. Indicator 4: diagnosis codes: distant metastasis codes. Indicator 2a: procedure codes: radiotherapy 60 days, cytostatic therapy 30 days. Indicator 2b: procedure codes in combination with testicular cancer diagnosis code: radiotherapy 60 days, cytostatic therapy 30 days.

Abbreviation: CI, confidence interval.

Table 5 Concordance of Relapse Identified by the Gold Standard and the Final Algorithm (Algorithm 5) in the Validation Cohort (n = 1127)

		Relapse by Gold Standard
		Yes	No	Total
Relapse by algorithm	Yes	232	10	242
	No	1	884	885
	Total	233	894	1127

Table 6 Performance of the Final Algorithm (Algorithm 5) in the Validation Cohort (n = 1127) to Identify Patients with Relapse from Stage I Testicular Cancer

Algorithm Performance	All Indicators Combined	Indicators Stratified on Codes Related to Pathology and Procedure
	% (95% CI)	Pathology Codes % (95% CI)	Procedure Codes % (95% CI)
Sensitivity	99.6 (98.7–100)	57.9 (51.6–64.2)	97.0 (94.8–99.2)
Specificity	98.9 (98.2–99.6)	100 (100–100)	98.9 (98.2–99.6)
Positive predictive value	95.9 (93.4–98.4)	100 (100–100)	95.8 (93.2–98.3)
Negative predictive value	99.9 (99.7–100)	90.1 (88.3–92.0)	99.2 (98.6–99.8)

Abbreviation: CI, confidence interval.

Table 7 Accuracy of Relapse Date as Estimated from the Final Algorithm (Algorithm 5) Compared with the Gold Standard in the Full Study Cohort (n = 1377)

Algorithm Estimated Date	% (95% CI)
Same date	44.4 (38.6–50.5)
Within 7 days	70.1 (64.4–75.4)
Within 14 days	85.8 (81.1–89.6)
Within 30 days	95.7 (92.7–97.8)
Within 60 days	97.5 (94.9–99.0)
Within 90 days	97.9 (95.4–99.2)

Abbreviation: CI, confidence interval.

Figure 2 Concordance of relapse date between the final algorithm (algorithm 5) and by the gold standard.