A Validated Register-Based Algorithm to Identify Patients Diagnosed with Recurrence of Surgically Treated Stage I Lung Cancer in Denmark

Figures & data

Figure 1 Schematic overview of the algorithm.

Table 1 Characteristics of the 217 Included Patients with Stage I Lung Cancer, Stratified on Cancer Recurrence Status in the Gold Standard^a

Population Characteristics	Cancer Recurrence	No Cancer Recurrence
	n (%)	n (%)
Number	72 (100)	145 (100)
Sex
Female	27 (38)	58 (40)
Male	45 (63)	87 (60)
Age, median (IQI^b)	74 (68;78)	73 (67;76)
Comorbidity^c
Low	29 (40)	60 (41)
Moderate	23 (32)	60 (41)
Severe	20 (28)	25 (17)
Tumor histology
Adenocarcinoma	79 (54)	33 (46)
Squamous cell carcinoma	37 (26)	25 (35)
Non-small cell carcinoma	24 (17)	8 (11)
Other	5(4)	6 (8)
Follow-up time^d, months (IQI)	18 (11;27)	36 (23;59)

Notes: ^aNumbers are stated as n (%) if nothing else is stated; ^bInterquartile interval; ^cCharlson Comorbidity Index score calculated at primary lung cancer diagnosis date, divided into score 0 (low), scores 1–2 (moderate) and scores 3+ (severe); ^dTime from primary lung cancer diagnosis to the first of the following events: recurrence, emigration, death, or end of study (1 May 2016).

Figure 2 Flowchart of study population.

Table 2 Concordance of Recurrence of Lung Cancer Identified by the Gold Standard and the Algorithm

Status by Algorithm	Status by Gold Standard
	Cancer Recurrence	No Cancer Recurrence		Total
		No New Cancer Event	New Primary Lung Cancer
Cancer recurrence	60	4	5	69
No cancer recurrence	12	131	5	148
Total	72	135	10	217

Table 3 Performance of the Algorithm to Identify Recurrence from Lung Cancer in Patients Treated for Stage I Lung Cancer with Curative Intent

Algorithm Performance	Main Analysis n=217	Excluding 31 Patients with a History of Cancer Prior to Lung Cancer, n=186	Second Primary Lung Cancer and Cancer Recurrence Pooled in the Gold Standard, n=217
	% (95% CI)	% (95% CI)	% (95% CI)
Sensitivity	83.3 (72.7–91.1)	85.0 (73.4–92.9)	79.3 (68.9–87.4)
Specificity	93.8 (88.5–97.1)	93.7 (87.9–97.2)	97.0 (92.6–99.2)
Positive predictive value	87.0 (76.7–93.9)	86.4 (75.0–94.0)	94.2 (85.8–98.4)
Negative predictive value	91.9 (86.3–95.7)	92.9 (87.0–96.7)	88.5 (82.2–93.2)

Abbreviation: CI, confidence interval.

Table 4 Concordance of Cancer Recurrence Date in the Gold Standard and Estimated by the Algorithm

Estimated Date by the Algorithm	% (95% CI)
On same date	0
Within 7 days	8.3 (3.6–18.9)
Within 14 days	20.0 (11.9–32.5)
Within 30 days	40.0 (29.9–53.4)
Within 60 days	70.0 (58.3–81.0)

Abbreviation: CI, confidence interval.

Table 5 Simulated Positive Predictive and Negative Predictive Value of the Sensitivity and Specificity of the Algorithm to Identify Recurrence from Lung Cancer in Constructed Populations with Different Recurrence Rates

Algorithm Performance	Population n=10,000 Recurrence Rate 15%	Population n=10,000 Recurrence Rate 25%	Population n=10,000 Recurrence Rate 50%
	% (95% CI)	% (95% CI)	% (95% CI)
Positive predictive value	70.3 (68.2–72.5)	81.8 (80.2–83.2)	93.1 (92.3–93.8)
Negative predictive value	96.9 (96.6–97.3)	94.4 (93.8–94.9)	84.9 (83.9–85.8)

Abbreviation: CI, confidence interval.