Delays in access to care for abortion-related complications: the experience of women in Northeast Brazil

Around 18 million unsafe abortions occur in low and middle-income countries and are associated with numerous adverse consequences to women’s health. The time taken by women with complications to reach facilities where they can receive appropriate post-abortion care can influence the risk of death and the extent of further complications. All women aged 18+ admitted for abortion complications to public-sector hospitals in three capital cities in the Northeastern Brazil between August-December 2010 were interviewed; medical records were extracted (N = 2,804). Nearly all women (94%) went straight to a health facility, mainly to a hospital (76.6%); the rest had various care-seeking paths, with a quarter visiting 3+ hospitals. Women waited 10 hours on average before deciding to seek care. 29% reported difficulties in starting to seek care, including facing challenges in organizing childcare, a companion or transport (17%) and fear/stigma (11%); a few did not initially recognize they needed care (0.4%). The median time taken to arrive at the ultimate facility was 36 hours. Over a quarter of women reported experiencing difficulties being admitted to a


Introduction
As the illegality of abortion results in unsafe procedures that pose a risk to women's health, abortion complications are common and important contributors to the global disease burden 1,2,3 .Abortion is a common event in Brazil, even though it is allowed only in cases of rape, when a woman's life is threatened or, more recently, for fetal anencephaly.The time taken by women with complications to reach facilities where they can receive appropriate post-abortion care can influence the risk of death and the extent of further complications or morbidity sequalae 4 .Delays in care-seeking for complications and accessing health care in pregnancy, delivery and postpartum have long been conceptualized as important contributors to maternal mortality and morbidity 5 .Specifically, the three-delays framework of Thaddeus & Maine 6 identified: (1) delays in recognizing symptoms and seeking care; (2) delays in arriving at an appropriate facility; and finally (3) delays in receiving appropriate treatment within a facility.While abortion complications fall within this framework, there are reasons to consider them separately, because the extent of delays may be exacerbated where abortion is illegal, severely restricted or highly stigmatized.Women may fear social or even legal consequences if other come to know, and healthcare providers may be afraid to treat women with complications 7 .Women may also hesitate for fear of experiencing disrespect or abuse from providers who suspect that an abortion is induced 8,9 .
In this paper, we use the three-delays framework to look at care-seeking behavior for abortion complications in the Northeast of Brazil.In 2013, induced abortions in the country were estimated between 687,347 and 865,160, with the Northeast having the highest ratio of induced abortions among women between 15 and 49 years, calculated by 100 live births, among the regions of the country 10 .The most-recent national Brazilian abortion survey revealing that 13% of women had induced at least one abortion, and that about a half of these women had attended a health facility post-abortion and had been hospitalized 11 .Hospitalizations for abortion complications, especially severe cases, have been declining in the last decades in Brazil 12 , a trend thought to stem from the widespread of misoprostol 13 , which, with correct dosage taken, results in complete abortion in 75-90% 14 of times, with low complication levels 9 .In 2013, 205,075 women were admitted to public hospitals for abortion-related complications 10 , and abortion has been estimated as the fourth cause of maternal mortality 15 .Brazil has compulsory maternal-death notification, including for abortions, and deaths are investigated, but this is not done in all municipalities 16 .
Hospitals are the appropriate level of facility in which to seek post-abortion care in Brazil, and Brazilian Ministry of Health guidelines 17 explicitly require that post-abortion complications are managed in a humane and technically effective way, so providers managing complications are unlikely to fear legal sanctions.
Nevertheless, the restricted nature of induced abortion in Brazil and the associated stigma may well influence where and how quickly women seek and obtain care, adding a complexity layer to delays that may not apply to different pregnancy complications, such as eclampsia.
Cad. Saúde Pública 2018; 34(6):e00168116 that provide obstetric and post-abortion care.Hospitals with over 120 hospitalizations in the previous year were included (19 hospitals).Brazilian law and ethics approval procedures require parents/ guardians to sign consent forms for research participants aged below eighteen.We deemed it inappropriate to seek such consent for young women, and so excluded minors, and all resident women, aged 18 or over, admitted for abortion complications in the 19 eligible hospitals, were invited to participate in the study, irrespective of whether their abortion (defined as pregnancy ending up to 22 weeks of gestation) was induced or spontaneous.Women admitted for a legal induced-abortion procedure, or whose abortion stemmed from an ectopic pregnancy, hydatidiform mole or another abnormal product of conception (International Classification of Diseases, 10 th revision -ICD-10), were ineligible.Eligible women who consented were interviewed by trained female health-professionals using standardized questionnaires, and additional information was obtained from medical records.Data were anonymized and confidentiality assured.
To classify the women's clinical condition on admission to the ultimate hospital, we convened an expert committee to reach a consensus for defining the degree of complications in the context of Brazil, with the consensus used shown in Box 1.
Data were analyzed using parametric and non-parametric methods, as appropriate, with both IBM SPSS (IBM Corp., Armonk, USA) and Stata software (StataCorp LP, College Station, USA).We analyzed the delay data using non-parametric methods, because the times reported were not normally distributed, and because some values were extreme outliers, possibly because of errors in data collection or in women's interpretation of the key questions.

Results
All 19 eligible hospitals (8 in Recife, 7 in Salvador and 4 in São Luís) participated.A total of 3064 women were eligible, of whom 2.7% refused to participate and 5.8% were discharged or died (2 women) before they could be interviewed, leaving 2,804 women interviewed.
Fifty-eight percent of women were in Salvador, 27.1% in São Luís, and 13.9% in Recife.Most women (91.7%) were under 30 years-of-age, with a mean age of 27.6 years (SD = 6.3) and were mostly (88.6%) married, cohabiting or in a stable relationship.The majority (89.1%) declared themselves as non-white, but 0.4% did not classify themselves.Nearly two-thirds (64.3%) had eight years of education or more.A third (35.5%) was not working, but only 7% had never worked.

Box 1
Criteria used to assess the clinical condition of abortion admissions.

Very serious condition
Any of the following: low level of consciousness (disoriented, lethargic or unconscious), severe hypotension (systolic < 90mmHg/diastolic < 70mmHg) or a heart rate ≥ 140 beats/minutes, septicaemia, septic shock, multiple organ failure, generalised peritonitis or intrapelvic collection, or any sign of organ failure (disseminated intravascular coagulation, hypovolemic shock, renal failure, respiratory distress syndrome or liver failure), or by management-based criteria, such as, requirement for respiratory assistance, admission to intensive care, or having a hysterectomy or blood transfusion.

Serious condition
Oriented, orthostatic hypotension, requiring replacement of blood or fluids, or heart rate between 100-139 beats/ min, signs of infection (temperature > 38ºC or < 36ºC or intrapelvic collection), or pain at uterine mobilization with localised peritonitis.

Mild condition
Having normal conversation, with signs of infection or offensive discharge, normal blood pressure or mild hypotension (systolic < 140mmHg and ≥ 90mmHg/diastolic ≥ 70mmHg), and not requiring blood or fluids transfusion.

Good condition
Temperature between 37.3-37.9ºCand normal blood pressure and no signs of infection, and not requiring fluid reposition.
Cad. Saúde Pública 2018; 34(6):e00168116 Table 1 and Figure 1 show where women first sought care, indicating the numbers who went directly to an appropriate facility, namely a hospital.They also show the types of places and the numbers of hospitals visited.
Women's means of transport for this first care-seeking attempt were private car (family, boyfriend, friend or neighbor's car) (32.5%), public transport (32.2%), taxi (18%) ambulance (7.8%), on foot (5%), or by bicycle or motorbike (0.9%).Over three-quarters (76.6%) of women went to a hospital first (of which 97% were public hospitals), however, only about half (47%) remained there for the entirety of their care.Nearly a quarter of women (23%) did not go to an appropriate facility level , although 16% sought care within the health system, going first to a health facility (outpatient clinic or primary health unit) for the care they needed.The percentage whose first care-seeking venue was outside the health system (i.e., a drugstore, use of traditional medicine at home, or a visit to a traditional birth attendant or other) was less than 7%.
The median number of hospitals visited was 2 (interquartile interval: 1-3).Just under half of all women went to one hospital only, while a quarter went to 2 hospitals, and the remainder went to 3 or more, up to a maximum of 8 hospitals.The complex care-seeking combinations involving other health levels and other non-health sector provision are also shown in Table 1 and Figure 1.
Almost all women arrived at their ultimate hospital in good condition (92%), while 5% arrived in serious or very serious condition.Women who went to only one hospital were slightly more likely to have arrived in serious or very serious condition (Table 2).84.3% of women reported that abortion occurred up to 8 weeks and 97.2% up to 12 weeks of gestation.
Table 3 displays the time from symptoms to the care-seeking decision (first delay), and the time from symptoms to ultimate care-seeking destination (first and second delay), overall and by the number of hospitals visited.The median time from when women recognized symptoms to the decision to seek care, was 10 hours, but was slightly longer for women who subsequently ended up going to more than one hospital (12 hours, difference not statistically significant).By contrast, the median hours from symptoms to arriving at the ultimate destination, markedly grew with increasing num-

Figure 1
Place women first sought care and places visited before arrive at the ultimate hospital.
bers of visited hospitals, from 15 hours, to 48 hours and 72 hours, for those going to two, or three, or more hospitals, respectively.Table 3 also shows times stratified by clinical condition on arrival at the ultimate hospital.About 31% of women reported having to overcome difficulties in searching for care.Reasons included shame and/or fear of being mistreated or humiliated by the health services, not having anyone to go with or to care for her children, lack of funds to pay for transport (or lack of means of transport) and fear of missing work.Only a few women mentioned underestimating the need of care or difficulty in knowing where to seek care.The percentage who declared they had experienced some difficulty in reaching the ultimate hospital was even lower (10%), although reasons given for difficulties were quite similar.Around 27% of women experienced delays in being admitted to a hospital even after they arrived there.Reasons included waiting a long time for medical attention, having pregnant women prioritized over abortion cases, and having to wait for a hospital bed to become available (Table 4).

Discussion
We investigated delays in care-seeking for abortion complications in Northeastern Brazil, contributing to a sparse literature in Brazil, and globally.The picture gained is mixed, indicating scope for improvement.
We sampled all women hospitalized for abortion-complications in the public sector and had an excellent response rate (91.5%).Our limitations relate to the exclusion of minors 18 who face great barriers to access 19 , and of abortions managed entirely within the private sector.In addition, despite the apparent homogeneity of the study population, another limitation of this study is that it did not analyze the influence of social position and especially race/ethnicity in determining institutional barriers to access to services.
Our measurement of the time at various stages was also problematic, with heaped response-times, and questions misinterpreted by some women.For example, we could not differentiate the length of time spent between arrival at the first hospital, between subsequent hospitals and to the hospital in which women finally received uterine evacuation.In hindsight, care-seeking patterns should have been measured with a grid, and not structured questionnaires with skip patterns.
Women hospitalized for abortion complications took nearly half-a-day (10 hours) from first symptoms to decide to seek care.Comparable research in other settings is scarce, but this is shorter than in Kenya 20 where 54% waited over a week, or Mexico where a qualitative study found most women waited two+ days 21 .Women recognize the need for care (99.6%), but nearly a third experienced difficulties initiating care-seeking, including in finding money, transport, childcare or a companion, or getting time-off from work (total 17%), and delays for fear, shame/embarrassment or stigma (11%).Weak social support networks and precarious living conditions were clear issues for some women, though comparative data are few.Our setting, three state-capital cities, meant transport was a minor barrier, unlike settings with poor public transport 22 .Women's fear/stigma were important reasons Table 3 Time from symptoms to first care seeking decision (1st delay) and from symptoms to ultimate care seeking destination (1st and 2nd delay).Median   for delay in Mexico 21 , and dehumanized care, disrespect, abuse and judgmental behavior by health professionals were documented in Brazil 9 , and elsewhere 8,23 ; even women with spontaneous abortions face discriminatory attitudes when health professionals assumed they were induced.Women's life situations only partly led to delays; service-related factors and their interactions with the legal context are also important.We found that women are unclear about the referral system and are sometimes uncertain where to go.Nearly all women went straight to a health facility, but only a third went directly to the public sector hospital, where they were ultimately treated.The rest had various care-seeking combinations involving lower health system levels that were incapable of treating complications or treatment outside the health system.Some went to facilities capable of managing complications at some moments, but not others.Ultrasound is central to managing abortion in Brazil 9 , shortages of ultrasound professionals in the public health network at times limit care-capacity 24 .In such facilities, women may be asked to return later or be directed elsewhere.Occasionally, women went to high-risk reference hospitals, but were referred downward to less busy facilities.Such prolonged and complex seeking-care processes have been reported for stigmatized diseases such as HIV or tuberculosis 25,26 and are potentially aggravated by the illegality and stigma surrounding abortion.
Half the women spent 36+ hours before reaching the ultimate hospital, and delay increased sharply with the number of hospital visited, from a median of 15 hours if only one hospital was visited, to 48 hours if two were visited, to 72 hours if 3+ were visited.Each transfer, thus, added around 1.5 days per additional hospital, and may have led to more severe post-abortion complications 27,28 .In Kenya 28 , women who delayed 6+ hours arriving at a health facility had at least twice the odds of a moderate/ severe complication compared to those who sought care earlier (delay of 7-48 hours: OR = 2.12; of 3-7 days: OR = 2.01; and 7+ days: OR = 2.35).
We found 92% of women arrived at the ultimate hospital in good condition, differing markedly from results in Kenya 20,29 , Cambodia 30 and South Africa 31 , which reported high-severity complications in 10%-42% of women, moderate-severity in 16%-40%, and low-severity in 26%-72%.This Cad.Saúde Pública 2018; 34(6):e00168116 heterogeneity reflects differences in local context, and methods such as case severity definitions, age groups studied and whether spontaneous abortions were included.However, perhaps the main difference is that the public health system is free for all Brazilians, and most abortions in Brazil are induced with misoprostol, which can be effective and safe even when used illegally 9,12,14 , resulting in shorter delays 32 .
We found women who arrived at the ultimate hospital in more compromised clinical conditions were slightly more likely to have delayed care-seeking from first symptoms.We cannot be certain when the signs of more severe complications started to emerge, but we had anticipated that women with more severe complications would have sought care earlier.The 3-hour difference seen may have been due to measurement error, or these women may have been either less likely to act or were less able to act swiftly.However, women with complications were little more likely to have been admitted to the first hospital they visited (59% vs. 40%), but those with complex pathways and long delays were more likely to have been admitted with complications.We also recognize that under Brazil's restrictive law, some women who unsuccessfully self-administered misoprostol and started bleeding may have had to seek care several times before providers were prepared to hospitalize them for "inevitable abortion" 9 .In Kenya 20 , the longer women waited to seek care, the more severe their complications on admission.We also found the longer a woman delayed before seeking care, the more severe was her clinical condition when presenting to the ultimate hospital.In contrast, those women with severe complications on arrival spent less time than the others in being admitted to a hospital, and a slightly lower percentage of them had visited a multiple hospital before being admitted.
Over a quarter of women reported difficulties in being admitted, suggesting a lack of beds and overburdened health facilities may be an important barrier, despite almost universal access to childbirth care 33 .A multi-centered national study 27 found that the risk of maternal death/near-miss was higher among women presenting with post-abortion complications, compared to other pregnancy complications.Pregnant women were attended more quickly than women with abortion complications, a finding also reported in Ghana 8 , and Gabon 4 .Obstetrician-gynecologists in Salvador 34 , expressed their views that maternity-hospital beds should be reserved for pregnant women, although they also affirmed they would never refuse to attend women with post-abortion complications.Many thought women should not have the right to choose induced-abortion based on their needs.

Conclusion
Women in our study faced difficulties obtaining post-abortion care, and the time taken to access an appropriate facility had consequences for the severity of complications.The stigma surrounding abortion delayed women's care-seeking and gave them less priority in health facilities.Women also faced difficulties stemming from their less favorable life conditions and had unmet need for emergency obstetric care.To reach the Sustainable Development Goals, Brazil's challenge is to consolidate maternal/child health gains, to improve women's access and quality of care, and further reduce inequities by tackling marginalized health issues and marginalized women.
Further research on delays is recommended in settings with restrictive laws or access to abortion to clarify the impact of abortion stigma on women's decision-making process and providers' treatment behavior.New analyzes will be carried out to distinguish the inequities between women, which is already being done considering women's race/skin color.Recommendations include continuous surveillance on death and severe abortion complications, and changes in the process of training health professionals to recognize abortion not only as a public health problem, but as issue to the field of women's sexual and reproductive health.Furthermore, review of the Brazilian restrictive Law would contribute for reducing delays and allowing timely access to abortion-related complications.Legalization of abortion in other settings has reduced abortion-related complications and probably the time taken by women to seek care.

Table 1
Place women first sought care, type of places and hospitals visited.

Table 2
Clinical condition on arrival at the ultimate hospital, overall and by numbers of hospitals visited.
* 18 women were classified as very serious.

Table 4
Difficulties experienced by women at various stages of care seeking *.
* Percentages and totals are based on respondents; more than one answer was accepted.
Cad. Saúde Pública 2018; 34(6):e00168116 Contributors T. V. B. Araújo participated in designed and conducted the fieldwork; designed and conducted the analysis; prepared the first draft, participated in interpreting the results, drafting, and commenting on the paper.E. M. L. Aquino, G. M. S. Menezes, S. V. Alves, and L. Coimbra contributed in designed and conducted the fieldwork, and participated in interpreting the results, drafting, and commenting on the paper.M. T. S. S. B. Alves participated in designed and conducted the fieldwork and participated in interpreting the results, drafting, and commenting on the paper.M. C. C. Almeida participated in interpreting the results, drafting, and commenting on the paper.O. M. R. Campbell participated in designed the analysis, and in interpreting the results, drafting, and commenting on the paper.