healthmedicinet

This 6-month prospective cohort study was conducted on a sample of women recruited
while hospitalized for emergency treatment following either an induced or spontaneous
abortion in a SUS maternity hospital in Sao Paulo, Brazil. The institution is located
in the central area of the city and serves as a referral center for women with high-risk
pregnancies, delivering care to about 450 births and 30 post-abortion women every
month. Women who seek emergency treatment care after a pregnancy termination generally
remain hospitalized for no more than 24 h per institutional routine. The institution
follows the National Guidelines for Abortion Care 11] and usually sets a follow-up appointment for the 45th day after discharge.

Study population

The purposive sample included all women hospitalized following an abortion at this
SUS maternity, between May and December, 2011, irrespective of whether their abortion
was spontaneous or induced (?=?184). Exclusion criteria included poor understanding of Portuguese language (?=?6), not having a telephone for study contact (?=?0), and being hospitalized for legal abortion (?=?0). Eight women refused. Those who consented (?=?170) answered a structured questionnaire at baseline. After the first month, 147
women were traced (82 %) and constituted the study cohort. Losses over this period
were due to 8 refusals, 1 maternal death and 14 women who could not be located. Women
in the cohort were contacted by telephone every 30 days for 6 months. Once women reported
a confirmed pregnancy in response to the questions “Are you pregnant? If yes, have you confirmed it with a pregnancy test?”, they were excluded from the study. The last telephone interview was conducted in
August 2012. Overall, 17 % of women traced at the first post-abortion month were lost
to follow up by the sixth month; hence 105 women completed all six interviews (71 %).
The 17 % lost to follow-up (?=?25) were the result of changes in telephone numbers and two refusals (Fig. 1).

Fig. 1. Number of women participating in follow-up over the six month period of the study

Study instruments and data collection

Two questionnaires were used. The first, administered by trained midwives through
a face-to-face baseline interview during the period of hospitalization, collected
data on social and demographic characteristics and reproductive history. Due to legal
restrictions on abortion, it was expected that the reason for seeking the procedure
would be inaccurately reported and therefore underestimated. Thus, the reason was
not asked of women. To assess whether the terminated pregnancy was unplanned or planned,
we used the Brazilian Portuguese validated version 30] of the London Measure of Unplanned Pregnancy (LMUP) 31]. The second questionnaire was administered by trained social researchers by telephone
one month after discharge and every 30 days for 6 months (Months 1–6); it gathered
data on contraceptive use, utilization of primary health care services, future pregnancy
intention and pregnancy occurrence.

The outcome of interest was contraceptive use over the 6-month period following the
abortion (dichotomous variable with no/yes categories).

The main independent variable was utilization of family planning services 6 months
following the abortion. We created this variable based on responses received during
the monthly telephone interviews over the six-month period following hospital discharge,
so this information refers to the 30-day period preceding each monthly interview.
As family planning services are delivered at the primary health care level in Brazil
(exceptions are female and male sterilization), but women are free to obtain services
from private providers, we assessed the utilization of two types of services: medical
consultation and contraceptive counseling, which allowed us to collect data on a combination
of ways to obtaining a contraceptive.

We used medical consultation since doctors are the only health professionals who are
currently allowed to prescribe hormonal contraception and trained to insert IUDs in
Brazil. The exception is emergency contraception, which nurses can also prescribe.
Thus, users of public health facilities who seek non-barrier contraceptive methods
must make an appointment to see a doctor for a prescription, as in the private sector.

We also measured contraceptive counseling receipt separately because this is carried
out, both individually and in groups, by a range of other health professionals, e.g.,
nurses, pharmacists, nurse auxiliaries, community health agents, and social workers,
and can take place apart from medical consultation, such as during home visiting,
health promotion and educational meetings, and purchases at drugstores. Contraceptive
counseling was measured through two questions: whether women received information
about contraception from a healthcare worker in the 30 days preceding the interview
(no/yes), and whether they received information from a healthcare worker in the 30 days
preceding the interview regarding risks of inter-pregnancy intervals less than 6 months
after a pregnancy loss (no/yes). A positive answer in either question was counted
as receipt of contraceptive counseling.

To sum up, we asked women every month if (a) they had a medical consultation in the
preceding 30-day period (no/yes); and (b) they received any contraceptive counseling
in the preceding 30-day period (no/yes). The variable generated consisted of four
categories: (1) report of neither medical consultation nor contraceptive counseling
in the preceding 30-day period (reference); (2) only contraceptive counseling in the
preceding 30-day period; (3) only medical consultation in the preceding 30-day period;
and (4) report of both contraceptive counseling and medical consultation in the preceding
30-day period.

In this study, we assume that effective contraceptive services include a combination
of being able to simultaneously utilize contraceptive counseling and medical consultation,
irrespective of whether women utilized private services, or accessed primary health
care facilities or drugstores.

Other covariates were pregnancy planning status (baseline; categorized as unplanned,
ambivalent and planned); desire to have more children (baseline; categorized as no/yes);
age (baseline; categorized as 15–24, 25–34, and 35–44 years); and education (baseline;
measured in completed years of schooling).

For sample description purposes, we collected and report data on women’s age (mean),
age at first intercourse (mean), age at first pregnancy (mean), number of previous
pregnancies (mean), race/ethnic color (self-identified as white, brown, or black),
religion (categorized as Catholic, Protestant, Others and None), work status/employment
(no/yes), cohabitation with partner (no/yes), pregnancy planning status (classified
from LMUP as planned, ambivalent and unplanned), and sexual activity in the preceding
month (no/yes). The type of contraceptive method used is reported for the first and
sixth month (no use, pill, condom, injectable, IUD, and traditional, the latter referring
to withdrawal and rhythm). We also describe if women were provided with a contraceptive
prescription during hospitalization. This variable was based on responses to the question
“Did you receive any prescription of contraceptive while hospitalized for emergency
treatment following the pregnancy loss?”, which was asked during the telephone interview
that occurred 30 days after hospital discharge.

We obtained administrative approval to conduct the study from the University of Sao
Paulo School of Nursing and ethical approval from the Maternity Research Ethics Committee.
Informed written consent was obtained from all the participants. During recruitment,
we emphasized that women could withdraw from the study at any time, with no effect
on their health care. In order to protect women’s safety and confidentiality during
the telephone interviews, we set a schedule together, based on the most appropriate
time and day of the week to call them. We also agreed about what to say when we called
them, who we could talk to and if we could identify ourselves as researchers, or as
another person/institution. Telephone numbers were noted as being unique or shared.
In order to ensure we were talking to the participant herself, we would use check
questions, based on information obtained in the previous interview.

Statistical analysis

The analytic sample consisted of women who completed the baseline interview and were
traced at one month post-abortion (?=?147). Double data entry by two different operators was done using Epi Info version
6.04. Questionable entries were reconciled.

We describe sociodemographic characteristics of all participants at two time points:
in the first post-abortion month (Month 1) and in the sixth post-abortion month (Month
6), as well as the characteristics of those who used contraception. We also describe
women who reported a confirmed pregnancy anytime during follow-up. Additionally, we
describe the degree of medical consultation utilization and contraceptive counseling
receipt each month.

As this is a longitudinal study, we need to account for repeated and correlated observations.
Therefore, we used Generalized Estimating Equations (GEE) considering an unstructured
correlation matrix and logit link transformation function. GEE is a non-likelihood
based method utilized for marginal models of non-linear responses. It is used in longitudinal
data when the focus is the difference in the population-average response between two
groups with different risk factors 32]. The dependent variable was contraceptive use over the six-month post-abortion period;
the main independent variable was utilization of family planning services 6 months
post-abortion; and covariates were pregnancy planning status, desire to have more
children, age, and education. The multivariate analysis was adjusted for time (i.e.,
observation for six months) and did not include women who were provided with a contraceptive
prescription during hospitalization. Crude and adjusted odds ratios and 95 % confidence
intervals were estimated. All analyses were conducted using Stata 13.0.