Skip to main content

Table 5 Risk ratios for failure to obtain a TST result by demographic and clinical characteristic. Results are presented for complete case analysis (n = 1878 patientsa) and analysis using combined results of 20 imputed datasets (n = 2303 patients). RRs marked with an asterisk (*) are statistically significant. CI = confidence interval

From: The latent tuberculosis infection cascade of care in Iqaluit, Nunavut, 2012–2016

Potential Risk factor

Patients with no TST results/ category total (%)

Risk ratio from univariate modelsb (95% CI)

Adjusted risk ratioc (95% CI), complete case

Adjusted risk ratioc (95% CI), multiple imputation

Age, years (per 5-year increase)

0.98 (0.95–1.02)

Sex

 Male

69/1065 (6.5%)

Reference

 Female

61/1238 (4.9%)

0.87 (0.67–1.13)

Ethnicity

 Inuit

48/1220 (3.9%)

Reference

Reference

Reference

 Non-Inuit

28/658 (4.3%)

0.76 (0.54–1.06)

0.68 (0.47–0.98)*

0.79 (0.58–1.06)

Indication for TST

 Tuberculosis exposure

20/606 (3.3%)

Reference

Reference

Reference

 Employment screening

26/506 (5.1%)

1.22 (0.82–1.83)

1.08 (0.65–1.78)

1.16 (0.77–1.76)

 School screening

25/297 (8.4%)

1.52 (0.95–2.42)

1.26 (0.68–2.29)

1.32 (0.83–2.11)

 Self or physician referral

45/746 (6.0%)

1.72 (1.22–2.42)*

1.65 (1.12–2.44)*

1.64 (1.17–2.30)*

Year of TST

 2012

28/759 (3.7%)

Reference

Reference

Reference

 2013

29/547 (5.3%)

1.27 (0.88–1.83)

1.38 (0.88–2.16)

1.25 (0.87–1.81)

 2014

31/498 (6.2%)

1.18 (0.80–1.74)

1.08 (0.65–1.79)

1.16 (0.79–1.71)

 2015

36/425 (8.5%)

1.52 (1.05–2.19)*

1.19 (0.70–2.02)

1.52 (1.04–2.22)*

 2016

6/74 (8.1%)

1.76 (0.93–3.34)

1.98 (0.94–4.15)

1.71 (0.91–3.21)

  1. aData regarding ethnicity were missing for 472 TSTs (14.9%) in 425 patients (18.5%) and were estimated using multiple imputation. Data on indication were missing for 189 TSTs (6.0%) in 148 patients (6.4%)
  2. bUnivariate models adjusted for the correlation between multiple TSTs performed on the same individual using an exchangeable correlation matrix
  3. cModel included ethnicity, indication for TST and year of TST and adjusted for the correlation between multiple TSTs performed on the same individual using an exchangeable correlation matrix