Factors Associated With Hospitalization Outcomes for Cases of Anemia in Pregnancy at a Regional Level in Burkina Faso

Background : Complications related to anemia in pregnancy are common in referral hospitals of Burkina Faso. This study aimed to analyze the hospital discharge results of cases of anemia for the mother-pregnancy entity at a regional level in Burkina Faso. Methods : A hospital-based historical cohort study was conducted on anemia cases identified from hospital records of pregnant women admitted between 2009 and 2011 in the maternity wards of public referral hospitals of the “Cascades” region in Burkina Faso. The hospitalization result was favorable if it was satisfactory for both the mother and the pregnancy, and unfavorable in the other cases. The factors associated with unfavorable hospital discharge were identified through multivariable logistic regression. Results : A total of 1815 cases of biological anemia in pregnancy were included in the study. The result of hospitalization for the mother-pregnancy entity was unfavorable in 42.6% of cases. The factors associated with unfavorable hospital discharge were rural residence (OR = 1.32; 95% CI: 1.00-1.74), evacuation cases (OR = 1.96; 95% CI: 1.48-2.60), 1 st or 2 nd trimester of pregnancy (OR = 2.74; 95% CI: 2.03-3.68), severe anemia (OR = 2.46; 95% CI: 1.75-3.46), moderate anemia (OR = 1.39; 95% CI: 1.05-1.85) and poor quality of anemia prevention (OR = 2.81; 95% CI: 1.62-4.88). The unfavorable hospital discharge was less frequent in women in couple (OR = 0.46; 95% CI: 0.26-0.84), in nulliparous and primiparous (OR = 0.69; 95% CI: 0.54-0


Background
Anemia in pregnancy is a public health problem, particularly in low-income countries.According to the World Health Organization (WHO), more than 40% of pregnant women in Africa are anemic.The WHO defined anemia during pregnancy as a hemoglobin ˂ 11 g/dL [1] .Anemia increases the risk of maternal and perinatal morbidity and mortality [2] [3] .
Iron and vitamin deficiencies, hemorrhage, and infections including malaria are the major causes of this blood abnormality [4] .The prevention of anemia is based on health promotion and malaria control associated with iron and folate supplementation.The curative treatment of anemia is based on oral or injectable iron therapy, blood transfusion, and treatment of potential causes [5] .Burkina Faso, with more than 50% of pregnant women with anemia, is classified at severe level of prevalence [1] .A study conducted in Burkina Faso in 2019 reported among hospitalized cases of anemia in pregnancy, 14.7% of severe cases [6] .Some significant proportions of unfavorable hospital discharges have been described; they were close to 3% and 20% for maternal and perinatal deaths, respectively.These analyses were conducted separately for either pregnancy outcome or maternal health.They did not provide a comprehensive overview of the hospital care system's response to anemia for both the mother and the pregnancy.Occupational category, number of prenatal visits, duration of iron/folate supplementation, history of malaria or hemorrhage, and low brachial circumference were the common factors associated with the unfavorable hospital discharges [6] [7] .Our study aimed to quantitatively describe the outcomes and associated factors of hospitalization of anemia cases for the mother-pregnancy entity at a regional level in Burkina Faso.It should then contribute to strengthen the management of hospital care of anemia in pregnancy.

Methods
The study aimed to identify the factors associated with the unfavorable hospital discharge of anemia cases for the motherpregnancy entity at a regional level in Burkina Faso.

Variables:
The hospitalization outcome for the mother-pregnant entity was the dependent variable.The hospital discharge for the mother was either favorable in the case of her health improvement, or unfavorable in the case of evacuation or referral, discharge without medical advice, or death.The outcome of hospitalization for pregnancy was favorable in cases of live birth or ongoing pregnancy, and unfavorable in cases of abortion or stillbirth.The global outcome of hospitalization was favorable if it was favorable for both the mother and the pregnancy, and unfavorable in the other cases.The independent variables were stage of anemia, type of hospital (regional or district hospital), patient age in years (categorized into adolescent girls 14-19 years, young adults 20-34 years, and adults 35 years and older), gestational age in trimesters, residence (urban or rural), marital status (single or in union) income level (low for housewives and high for employed women), parity (nulliparous/primiparous and multiparous), mode of admission to the maternity ward (evacuation, referral, direct admission), time to referral/evacuation, reason for admission (anemia or other reason), and quality of anemia prevention before admission.Anemia in pregnancy was defined regarding WHO, as a hemoglobin (Hb) < 11 g/dl [1] .It was classified into stages: mild anemia (10 ≤ Hb < 11 g/dl), moderate anemia (7 ≤ Hb < 10 g/dl), and severe anemia (Hb < 7 g/dl).Because gestational age was very poorly reported in the hospital records, it was estimated in trimesters using the symphysis-fundal height with reference to the literature [8] .The time of referral/evacuation was categorized into acceptable (< 24 hours) and long (≥ 24 hours).In the absence of consensus on the limits of time of referral to hospital, the categorization of the time frame took into account the risks related to pregnancy, the average distance to hospitals, the quality of the roads and means of transportation in the region, and the published experiences [9][10] [11] .The quality of anemia prevention was considered as "good" if the patient had received at least one antenatal visit with both iron/folate supplementation and malaria preventive treatment.It was acceptable if the patient received only one of the two preventive treatments at the antenatal visit.Otherwise, the quality of anemia prevention was considered as "weak".The choice of the independent variables was made with reference to the literature related to anemia in pregnancy [6][12] , assuming their possible association with the results of hospital discharge.
Data collection: All data were extracted from the pregnant women's hospital records using a standardized questionnaire by trained investigators.The standardized form had three sections devoted to sociodemographic characteristics, clinical data, and anthropometric measurements.The form has been approved by specialists in public health, epidemiology, and biostatistics from the Université Libre de Bruxelles in Belgium and the Nazi Boni University in Burkina Faso.

Bias management:
To harmonize the understanding of the questionnaire, a pretest was conducted in a referral hospital in a neighboring region before the actual data collection.All questions were clarified with the investigators.At the end of the first day of data collection, a meeting was held to evaluate and correct the completion of the questionnaire; this evaluation was repeated once a week until the end of the data collection.
Sampling: All hospital records of pregnant women admitted between January 1, 2009, and December 31, 2011, were reviewed to retain those with a biological diagnosis of anemia.All records reviewed were for cases already discharged from hospital.

Statistical analysis:
The data were entered in duplicate to reduce entry errors.The purpose of the data analysis was to identify factors associated with unfavorable hospital discharge of cases of anemia for the mother-pregnancy entity.The variables were summarized with descriptive statistics (frequencies and percentages).The chi-square test was used to analyze the associations between independent and dependent variables.The odds ratios (OR) and their 95% confidence intervals (95% CI) were calculated to estimate these association strengths.
Two multivariable logistic regression models were constructed.In the first model (model 1) involving all subjects, only independent variables that were associated with the dependent variable in univariate analysis at a significance level of 0.20 were considered for inclusion.Variables in the final model were selected by a top-down stepwise procedure to retain only those with a p-value ˂ 0.05.The adjusted OR and their 95% CI were estimated from this model.The model was tested for fit and good specification using the Hosmer-Lemeshow test and the link test, respectively.Because gestational age was missing for 317 women (18%) and the proportion of unfavorable discharge was significantly higher in this case, the category of women with missing gestational age was considered in the analyses.The multivariable model included 1363 subjects (82% of the total sample); the proportions of unfavorable discharge did not differ between the subjects included in the analysis and those not included (43% vs. 40% respectively; p = 0.29).A second logistic regression model was constructed for the subgroup of evacuated or referred cases (model 2) to account for the effect of referral/evacuation time, following the same method as for model 1.The data were analyzed using SPSS Statistics version 24.

1) Participants
A total of 5586 hospital records of pregnant women admitted between 2009 and 2011 were reviewed, including 5348 at the regional hospital and 238 at the district hospital.Of these, there were 1815 cases of biological anemia (32.5% of records).All cases of biological anemia for which the hospital discharge result was known (n =1635) were included in the analyses.

2) Characteristics of the study population
The majority of cases were women in union, of low income, under 35 years of age and in their 1 st or 2 nd trimester of pregnancy.Twenty-seven percent (27%) had severe anemia, 42% had moderate anemia, and 32% had mild anemia.
Anemia accounted for more than one-tenth of the reasons for hospitalization.The common modes of hospital admission were evacuation and direct admission.The referral/evacuation time did not exceed 24 hours in the majority of cases.The prevention of anemia in the peripheral health facilities was good in the majority of cases (Table 1).Indeed, the number of antenatal visits per woman before hospitalization varied from 0 to 6 with a median of 2 visits; the majority of women had benefited from iron/folate supplementation and preventive treatment for malaria.
3) Hospitalization results for pregnant women admitted with anemia An unfavorable hospital discharge for both the mother and pregnancy was reported in 42.6% of cases (Table 2).In 71.6% of cases, the hospital discharge was favorable for the mother.In the remaining cases, the outcome was unfavorable, ending in evacuation or referral in 24.8% of cases, discharge without medical advice in 1.9% of cases, and death in 1.8% of cases.
For the cases received from the peripheral level of care system, 32.41% were again evacuated or referred to a higher level.Severe anemia was associated with referral/evacuation (p < 0.001), maternal death (p = 0.003), but not discharge without medical advice (p = 0.09).More than half (54.2%) of referred/evacuated cases had severe anemia on admission.
The proportion of referred/evacuated cases was higher (p < 0.001) in the 2nd trimester (41.1%) than in the 1st trimester (18.2%) and 3 rd trimester of pregnancy (20.2%).Sixty-one percent (61,0%) of cases admitted with severe anemia in the 2 nd trimester of pregnancy were referred or evacuated again.There were more rural (78.0%) than urban (22.0%) cases whose hospitalization ended in referral/evacuation (p < 0.001).The majority of patients who died (23/30 deaths) were from rural areas.The referral/evacuation times of less than one day corresponded to obstetric emergency and anemia, and the hospitalization result was unfavorable in 83.2% of these cases.The longest delays were observed in free referral cases without medical transport.
Concerning the prevention of anemia, unfavorable hospital discharge was observed in 72.6% of patients not supplemented with iron/folate, 64.5% of patients with no preventive treatment for malaria, and 78.6% of women who did not benefit any antenatal visit.The frequency of unfavorable hospital discharges decreased with increasing number of antenatal visits.Among patients who had at least four antenatal visits (15.7% of cases), an unfavorable hospital discharge was observed in 18.9% of cases.
Pregnancy outcome at hospital discharge was favorable in 77.7% of cases (with ongoing pregnancy in 46.1% and live birth in 31.6%) and unfavorable in 22.3% of cases (including abortion: 12.7% of cases; stillbirth: 8.7%; ectopic pregnancy: 0.9%).There were more abortions in moderate and mild anemia cases than in severe cases (p = 0.03).The stage of anemia was not associated with stillbirth (p = 0.19) and ectopic pregnancy (p = 0.29).
Of the 35 unfavorable outcomes of pregnancy in noncouple women, 32 cases (91.4%) were abortions.The marital status was not associated with maternal hospitalization outcome and stage of anemia.
The majority of maternal deaths and stillbirths (18 of 28 maternal deaths or 64.3% and 91 of 141 stillbirths or 64.5%) occurred in multiparous women; however, the frequency of severe anemia was higher in nulliparous women (30.7%).

Variable n %
Age (year)

4) Factors associated with hospitalization outcome for cases of anemia in pregnancy
Tables 3 and 4 present the results of univariate and multivariate analyses.In univariate analysis, the factors associated with unfavorable hospital discharge for the mother-pregnancy entity were rural residence, noncoupled status, referral or evacuation cases, anemia as reason for admission, the first two trimesters of pregnancy, multiparity, severe or moderate anemia, and low quality of anemia prevention.
In model 1 (Table 3), which did not include the time of referral/evacuation, unfavorable hospital discharge was more common among rural women, those admitted from an evacuation, patients in the 1 st or 2 nd trimesters of pregnancy, those with moderate or severe anemia, and those with poor quality of anemia prevention.The hospital discharge was less unfavorable in women in union, nulliparous, and primiparous women.
In the analysis of the subgroup of referred or evacuated cases (model 2, Table 4), factors associated with unfavorable hospital discharge were rural residence, evacuation cases with less than 24 hours' delay, first two trimesters of pregnancy, severe or moderate anemia, and poor quality of anemia prevention.Marital status and parity were not associated with unfavorable hospital discharge in this model.
The patient age, type of hospital, income level, and reason for admission were not associated with unfavorable hospital discharge of cases of anemia in pregnancy.

Discussion
Compared with the isolated maternal and pregnancy hospitalization outcomes, there was a high proportion of unfavorable hospitalization outcomes for the mother-pregnancy entity in the two hospitals of the "Cascades" region.This comparison was possible thanks to the definition of the aggregate indicator "unfavorable outcome of hospitalization for the motherpregnancy entity".The factors associated with these unfavorable outcomes were the quality of anemia prevention, stage of anemia, gestational age, mode of admission, and residence.
The limitations of such a study were inherent in the retrospective nature of data collection from hospital records, which were summarily filled out and sometimes poorly archived.However, the extension of the period covered by the data collection made it possible to have a representative number of subjects included in the analyses despite the extent of the missing data.
A selection bias may have been introduced by including only records with a positive diagnosis of anemia.This shortcoming was mitigated by checking the similarity of the results with the national references and those in the literature.
The aggregation of outcomes had the particularity to include referral/evacuation and discharge without medical advice.
This allowed for a more comprehensive assessment of hospital performance in providing care for anemia in pregnant women.
Thus, referrals and evacuations accounted for the largest part of unfavorable outcomes.More than one in five patients were referred or evacuated to a higher level of the care system.One-third of the cases received from referral/evacuation were again evacuated to a higher level, even though regional hospitals were supposed to take care of the majority of Qeios, CC-BY 4.0 • Article, January 16, 2024 Qeios ID: KXPD9Z • https://doi.org/10.32388/KXPD9Z3/16

Table 2 .
Distribution of cases of anemia in pregnancy according to the hospitalization outcome

Table 3 .
Factors associated with unfavorable hospital discharge for cases of anemia in pregnancy (Model 1) Qeios, CC-BY 4.0 • Article, January 16, 2024