Perspectives on Sexual and Reproductive Health
Volume 34, Number 3, May/June 2002

 

Predictors of Contraceptive Discontinuation in a
Sexually Transmitted Disease Clinic Population
TABLES

TABLE 1. Percentage distribution of female sexually transmitted disease clinic clients and cumulative percentage who discontinued effective method use after one year, by selected demographic characteristics at baseline; and percentage who ever used contraceptives and who experienced side effects, and cumulative discontinuation rates, by method, Denver, 1996-1999
Characteristic All women
(N=406)		 Cumulative % who discontued after one year
% DISTRIBUTION
Age
13-17 18.9 30
18-20 29.0 32
21-23 23.1 22
>=24 28.8 32
Race/ethnicity
White 37.9 32
Black 27.4 25
Hispanic 29.8 31
Other 4.9 30
Marital status
Never-married 87.4 29
Married 5.5 31
Divorced 7.0 22
Education
<high school 39.9 32
>=high school 60.0 24
Working
Yes 62.3 31
No 37.6 26
Has public or private medical insurance
Yes 30.5 22
No 69.4 33
No. of births
0 64.0 39
1-3 32.3 26
>=4 3.6 20
Perceived likelihood of pregnancy in next year
Not at all likely 68.2 27
Slightly to extremely likely 31.7 34
Risky sexual behavior in past year*
Yes 9.1 15
No 90.8 30
Study arm
Intervention 50.7 28
Control 49.2 32
Total 100.0 29
PERCENTAGE
Ever used
Pill 56.8 30
Injectable 28.5 27
Implant 5.4 22
Male condom 95.8 29
Diaphragm 1.4 34
Spermicides 14.2 33
Ever had side effect†
Pill 22.5 30
Injectable 73.2 23
Implant 90.9 25
Male condom 20.3 31
Diaphragm 100.0 34
Spermicides 22.4 54
*Risky sexual behavior was defined as having exchanged sex for money or drugs, or having had a sexual partner who was gay or bisexual, was a commercial sex worker, injected drugs or was HIV-positive. †Percentages are based on women who have ever used each method. Note: For the distribution by marital status, data are missing for seven women.

back to text


TABLE 2. Hazard ratios (and 95% confidence intervals) from univariate Cox proportional hazards models showing the association between the risk of discontinuing effective method use and selected time-varying characteristics
Characteristic Hazard ratio p-value
No. of side effects with method used in past*
0 (ref) 1.00  
1 0.60 (0.35-1.02) .061
2 0.71 (0.27-1.85) .481
>=3 1.15 (0.38-3.51) .800
No. of side effects with current hormonal method
0 (ref) 1.00  
>=1 0.84 (0.63-1.11) .211
No. of side effects with current coitus-dependent method
0 (ref) 1.00  
>=1 1.12 (0.77-1.62) .563
No. of problems using current method†
0 (ref) 1.00  
1 3.00 (1.66-5.39) <.001
>=2 4.75 (3.00-7.50) <.001
Predominant type of current method
Hormonal 2.26 (0.76-6.74) .145
Coitus-dependent 2.94(1.04-8.34) .043
Dual‡ (ref) 1.00  
No. of times switched methods§ .794
0 0.93 (0.55-1.58)  
1 0.87 (0.51-1.47)  
2 0.81 (0.48-1.37)  
Frequency of sex§ .047
1-3 times per mo. 1.16 (1.00-1.34)  
1 time per week 1.34 (1.16-1.55)  
>=2 times per week 1.56 (1.35-1.80)  
No. of partners§ .089  
1 1.43 (0.95-2.15)  
>=2 2.04 (1.35-3.08)  
No. of new partners§ .743
0 1.05 (0.79-1.38)  
1 1.10 (0.83-1.45)  
>=2 1.15 (0.87-1.52)  
Perceived likelihood of pregnancy in next year
Not at all likely 0.61 (0.41-0.91) .016
Slightly to extremely likely (ref) 1.00  
*The sum of the number of side effects associated with past use of hormonal, coitus-dependent or dual methods divided by the number of methods of that specific type ever used; the hazard ratio is based on the current predominant method type (hormonal, coitus-dependent or dual) in that interval. †Based on the number of problems (out of a possible 11) that resulted in intercourse without the use of contraception. ‡Dual-method use defined as the use of both hormonal and coitus-dependent methods at more than 50% of recent sexual encounters. §Variable was analyzed as an ordered categorical variable; the hazard ratio thus represents the risk of discontinuation relative to that of the other respective categories for the variable. Note: ref=reference category.

back to text


TABLE 3. Hazard ratios (and 95% confidence intervals) from multivariate Cox proportional hazards models showing the association between the risk of discontinuing effective method use and selected characteristics
Characteristic Hazard ratio p-value
No. of problems using current method*
0 (ref) 1.00  
1 2.99 (1.63-5.48) .000
>=2 5.03 (3.14-8.06) .000
Had medical coverage at baseline
Yes 0.53 (0.32-0.88) .014
No (ref) 1.00  
Gained medical coverage by follow-up
Yes 0.50 (0.26-0.95) .036
No (ref) 1.00  
Lost medical coverage by follow-up
Yes 1.41 (0.66-3.02) .379
No (ref) 1.00  
Risky sexual behavior in year before enrollment
Yes 0.36 (0.15-0.88) .025
No (ref) 1.00  
Study arm
Intervention 0.90 (0.59-1.36) .604
Control (ref) 1.00  
*Based on the number of problems (out of a possible 11) that resulted in intercourse without the use of contraception. Notes: Model controlled for all of the variables listed in Table 1 (except for perception of pregnancy in next year), plus the following time-varying variables: side effects with current method, new and total number of sexual partners in the past four months, frequency of sexual intercourse, predominant type of current method, number of coitus-dependent methods used before enrollment, number of hormonal methods used before enrollment, cumulative number of method switches, insurance coverage and problems using current contraceptive method. Note: ref=reference category.

back to text



© copyright 2002, the Alan Guttmacher Institute.