Skip to main content
. 2020 Oct 29;2:100045. doi: 10.1016/j.conx.2020.100045

Table 2.

Adjusted hazard ratios of multivariable risk score models on contraceptive discontinuation in the derivation cohort.

Full risk score
Simplified risk score
HR (95% CI) Points HR (95% CI) Points
Contraceptive method type
Cu-IUD/IUS 1.58 (0.67–3.72) 1 1.56 (0.66–3.66) 1
Pillsa 3.16 (1.59–6.26) 3 3.32 (1.68–6.57) 3
Other modernb 5.32 (2.46–11.42) 4 4.95 (2.32–10.59) 4
Reference: injectables, implants Ref. 0 Ref. 0



FP user type
Continuing method used in past month 2.25 (1.14–4.23) 2 2.41 (1.23–4.73) 2
Switching from one method type to another 2.10 (0.88–5.06) 2 2.07 (0.87–4.97) 2
Reference: initiating contraception Ref. 0 Ref. 0
< 9 years of completed education 1.62 (1.01–2.60) 1 -- --
Does not have child aged <6 months 1.60 (0.89–2.86) 1 1.57 (0.88–2.82) 1



Relationship status
Spouse supportive of FP 4.58 (0.63–33.24) 3 -- --
Not married (legal or presumed) 7.34 (0.97–55.21) 4 1.70 (0.98–2.93) 1
Reference: spouse unsupportive/unsure of spousal support Ref. 0 -- --
Maximum score 12 8

Notes: Variables included in the full risk score model were selected using stepwise forwards and backwards selection to identify the Cox model with the minimum AIC value. The Efron approach was used to handle ties. The simplified risk score was developed by removing features in the full risk score that are not routinely collected (either verbally or in written documentation) in Kenyan public health facilities. β is the Cox proportional hazards model coefficients (nonexpontentiated).

a

Pills include daily combined and progestin-only oral contraceptives.

b

Other modern methods include condoms, fertility-awareness-based methods (LAM, TwoDay Method, Standard Days Method) and emergency contraceptive pills.