To measure the quality of post-abortion care in referral hospitals, we adapted the comprehensive WHO framework for Quality of Maternal and Newborn Health Care[24] to post-abortion care and apply it in two hospitals in humanitarian settings. We will first discuss the findings in the two hospitals and then the strengths and limitations of the quality-of-care framework.
Quality of PAC in two hospitals in humanitarian settings
Our comprehensive evaluation of the quality of PAC suggests that overall, the two hospitals provided effective and life-saving PAC. The risk of healthcare-related near-miss was low (less than 2%) as was the mortality index compared to other facilities in Africa (9.5% in the CAR hospital and 4.2% in the Nigerian hospital versus 18.3% in the WHO-MCS-A[41]). However, we noted a possible overuse of antibiotics and blood transfusion, suggesting overmedicalization and inefficient use of resources. Further, women reported mixed experiences with the quality of care provided.
Our quality-of-care measurement framework suggests that the low risk of healthcare-related near-miss can likely be explained by the adequate availability of healthcare inputs permitting the proper implementation of evidence-based practices. Almost all health professionals had received PAC training and almost all comprehensive and extended PAC signal functions were available in both hospitals, which has not been the case for many other African referral hospitals studied in more stable contexts[34, 35]. Some of the key evidence-based practices were correctly implemented in both hospitals including the provision of blood transfusion when indicated, antibiotics administered to patients with septic abortions and use of appropriate technology to evacuate the uterus. Only less than 2.5% of patients received the non-recommended and less safe D&C technology when having an instrumental uterine evacuation. This estimate is lower than in other studies in Africa (ranging from 8–100% [14, 38, 41, 54–58]) but is similar to levels found in facilities supported by international organizations[20, 59]. Research shows that once external support ceases, facilities struggle to maintain skills and supplies for MVA and so, reuse D&C[60]. Efforts should be maintained to completely abandon D&C and ensure a continuous use of the appropriate technology.
Nevertheless, our results also suggest inadequate knowledge and use of antibiotics in both settings. Although most women with septic abortion received antibiotics, they may not have received the recommended regimen. In fact, the KAPB survey identified insufficient knowledge among physicians, midwives, and nurses on this topic, particularly in the Nigerian hospital, where only 21% gave the correct answer. In contrast, while the prevention of surgical-related infection seems to be well implemented in the Nigerian hospital, only 40% of women receiving instrumental uterine evacuation received a prophylactic antibiotic in the CAR hospital. This preventative intervention is even more important in this context of restrictive abortion laws where women may have had unsafe instrumental abortions[10]. Moreover, as found in other African studies[60, 61], we identified some overprescription of antibiotics in both settings, even though some of these prescriptions might have been justified but not documented in the patient file. In referral hospitals like the two study sites, practitioners’ continuous training, and regular antibiotic stewardships should be implemented to promote adequate and rational use of antibiotics. This strategy will not only better prevent and treat infections but also avoid or delay antimicrobial resistance in the longer term[62].
In addition to an overprescription of antibiotics, the possible overuse of blood transfusion in the Nigerian hospital or the high use of instrumental uterine evacuation in the CAR hospital could suggest “overmedicalization” for some patients. It might be due to possible provider preference and practice or organizational constraints rather than evidence-based recommendations[63]. In CAR, the choice of the faster method to evacuate the uterus (MVA) might be due to the high bed occupancy rate. The fact that only 15% of patients were able to ask questions about their treatment suggest that patients’ preference might not have always been asked and taken into consideration. On the other hand, in the Nigerian hospital, some women might have had an undocumented indication of blood transfusion. Alternatively, the prescription of blood transfusions outside the guidelines’ indications may suggest the need to adapt these recommendations to the context. Indeed, women of reproductive age in this fragile State of Jigawa have some of the worst nutritional and anemia indicators in the country[10, 64]. The fact that resources in this hospital are not as restricted as in other hospitals not supported by international organizations[60] may have enabled clinicians to adapt their practice to the specificities of women in this state.
The provision of contraceptive services was found to be insufficient in the Nigerian hospital with less than 40% of women receiving counseling and less than 5% of contraception uptake at discharge, which is lower than what was found in the CAR facility (63.4%) and in other African hospitals studies (15–70%)[60, 65, 66]. Cultural factors may play a role to explain these poor results. This population traditionally puts high values on high fertility. They have one of the highest fertility rate in Nigeria[67] and women’s autonomy in accessing contraception is limited[68]. In addition, as indicated by the available signal functions, another reason for these results may be a lack of coordination of PAC services with the contraceptive services provided in another hospital unit that was not opened 7/7. Evidence shows that post-abortion contraception should be provided at the same time and location as clinical treatment for complications[69]. Such strategy increases the uptake of contraceptive methods by women and thus protects them against the risk of future unintended pregnancies[15, 34, 69]. In parallel, women empowerment programs should be implemented to enhance their health contraceptive decision-making capacity[68].
In both study hospitals, around 80% of women reported having received pain medications. However, pain management was not optimal according to the medical records. In the CAR hospital, anesthesia was recorded as provided only to six out of ten women undergoing instrumental uterine evacuation, despite paracervical block being a part of the standard protocol. In contrast, in the Nigerian hospital, while anesthesia was recorded as provided almost routinely in instrumental evacuation, only 36% of patients received analgesics according to the medical records’ review. The dissonance between the reported experience by women and the medical records’ review results may be explained by a lack of documentation of analgesic in the medical files, a desirability bias, or women misunderstandings of the treatment provided. This latter hypothesis is supported by the fact that the lack of provider-patient communication was the most important gap identified in the patient survey. Only half of the women received explanations about their care and only one in six were able to ask questions in both hospitals. Those results were worse than what was found in the WHO-MCS-A African referral facilities of more stable settings, where 82% of women reported having had explanations and 67% that they could ask questions[41]. Evidence shows that poor communication in hospitals may be a significant barrier in women’s satisfaction to care and adherence to treatment[70]. Furthermore, provider attitudes to abortion care can negatively impact provider-patient interaction, timeliness and quality of care[60]. The fact that in our study, almost 80% of health providers considered PAC as every woman’s right is encouraging. Nevertheless, there is an urgent need to initiate strategies to enhance communication with patients about their condition, care, and post-abortion contraception in a supportive, empathetic, and nonjudgmental attitude. Previous research on PAC has shown that introducing educational protocols in PAC, using job aids and leaflets to provide information about women's treatment, postabortion fertility, and contraception, as well as conducting workshops to clarify values and attitudes about abortion, improve women's satisfaction with care and increases the likelihood of timely contraception uptake in the absence of pregnancy desire.[69, 71–75].
Our evaluation of the quality of PAC in the two hospitals was comprehensive but cannot be generalized to other hospitals of the targeted areas, regions, and countries, nor to hospitals of humanitarian settings. Health providers’ and women’s answers to self-administered or face-to face surveys are prone to memory and desirability biases, limiting the validity of the surveys’ results, especially in a subject like abortion, which is prone to stigma. While the same prospective methodologies were used to collect data in the two hospitals, and the same management guidelines and standardized medical records were applied, some documentation differences in the patients’ files may have remained. This may limit the validity and comparability of the evidence-based practices and health outcomes indicators. The differences in patients’ characteristics between the two settings could affect participants responses, with the Nigerian women being more likely to be older, married, with less education and more severe complications than the CAR women. Individual experiences of care are highly subjective variables[76]. Differences between the two settings might be due to different levels of patients’ understanding and expectations of quality according to their characteristics or different social norms[47].
Overall, the good quality of care provided in these two hospitals of humanitarian settings can partly be explained by the important support of MSF to the two facilities in terms of provision of equipment, medication, staffing, continuous training, supervision, and availability of medical protocols. Other research in humanitarian contexts assessing the impact of NGO interventions found important improvement in some quality indicators[20, 59]. This suggests that even in such challenging contexts, providing and improving quality of PAC inside health facilities is feasible and that some of the potential barriers linked to fragility or insecurity can be overcome.
Strengths and limitations of the quality of PAC framework
Using this framework, we were able to identify various strengths and gaps in the quality of PAC. The analysis of the similarities and discontinuities between the inputs, process and outcome indicators allowed to strengthen our understanding of the issues identified. In these two NGO-supported hospitals, several gaps were thoroughly analyzed, enabling field-oriented recommendations to be formulated. This suggests that the framework may also be able to identify areas for improvement in all types of hospital settings, whether stable or humanitarian, and supported or not by NGO.
Regular measurement of quality of care is a core principle of quality improvement programs and the primary step in improving quality of care[77]. To our knowledge, this conceptual framework is the first to measure the three Donabedian’s domains of PAC quality provided in referral hospitals, examining the eleven dimensions included in the WHO framework for MNHC[24]. Indicators measuring the patient experience in terms of communication, respect, dignity, and emotional support ensure a person-centered approach and complement the more traditional assessment of the hospital's structural capacity to provide PAC[34, 35, 45, 78, 79] and its coverage of key clinical practices[19–21]. Furthermore, the inclusion of indicators assessing the implementation of evidence-based practices and one indicator assessing the information system is a clear added value, since most of these indicators are rarely included in PAC evaluations[14, 31, 60]. Both health and people-centered outcome indicators allow an overall assessment of the impact of the care provided. The use of multiple sources of information to measure the framework's indicators, including clinical data and surveys data from three different populations (ward managers, PAC providers, and patients), increases the robustness of the evaluation by considering and triangulating different points of view.
We applied a more robust outcome quality-of-care measure than the mortality index which has been used in other studies[44]. A limitation of this indicator is that it does not exclude inevitable deaths from the estimates and therefore doesn’t accurately reflect the outcome of the care provided in the health facility. As we have done in our study, we recommend that future quality-of-care research uses the risk of healthcare-related near-miss as their outcome indicator[50]. This indicator corrects for the flaw of the mortality index because it measures the worsening of the state of the women after 24h and eliminates most inevitable deaths or near-miss that are happening in the first 24h after presentation and for which the responsibility of the quality of care provided in the facility is difficult to determine.
Although this is not the case in many contexts, it would be optimal if all post-abortion, safe abortion, and contraceptive care were routinely implemented as part of a comprehensive abortion care package. Our framework is complementary to the recently published framework on the quality of safe abortion care[27] and we recommend that future assessments of the quality of safe abortion care should include an evaluation of PAC quality[16]. Nonetheless, PAC does not have to be solely assessed as part of safe abortion care, as it is a broader component of pregnancy-related care including the management of complications of any types of abortion (spontaneous and induced).
Our quality-of-care framework faced implementation challenges that requested some adaptations. In particular, the indicators measuring the emotional support were removed because questions were identified as ambiguous[40]. Nevertheless, we believe that this is an important drawback of our evaluation as emotional support is a key dimension of experience of care. Researchers should do more work to develop short valid scales including this dimension to allow a comprehensive evaluation of experience of care.
Whereas around 20% and 35% of clinicians reported using D&C to evacuate the uterus respectively in the Nigerian and CAR hospitals, less than 2.5% of patients having an instrumental uterine evacuation received this inappropriate technology according to the medical records’ review of both hospitals. This contradiction in the results might be explained by a possible confusion of the definition of D&C by the clinicians answering the KAPB questionnaire, especially in CAR where the expression “curettage” is often used to name different types of instrumental uterine evacuation, including MVA[80]. In fact, the MSF standardized medical records of both hospitals include a full page detailing the type of instrumental evacuation and is probably providing more reliable results for this indicator than the KAPB survey. This suggests that questionnaires need to clearly indicate the reference to sharp curettage and that measuring such indicator using clinical data or direct observation rather than survey among health professionals might be preferable[81].
Additionally, only few indicators per dimension of the complete framework were selected. While this limited number of indicators improves its usability, it limits its content validity. Furthermore, the classification of each of the eleven dimensions in the 3 Donabedian domains was a consensual but a subjective process taking the perspective of the measurement of quality of care at referral hospital level. Another group of investigators may have classified them differently. In particular, the coverage indicators are sometimes considered as outcomes and the signal functions indicators as process ones[24, 26].
Our framework tested in the context of a research study requires the use of different sources of data which can be challenging to collect routinely. Nevertheless, it can provide some guidance for routine assessment of PAC quality as all dimensions could be at least partially assessed by the following minimum package: the evaluation of the PAC signal functions in the hospital (which include human and physical resources indicators), the evaluation of the quality of the documentation, the evidence-based practices and the healthcare outcome indicators in a random sample of patients medical records and the different indicators of experience of care and person-centered outcome by short patients exit surveys including the nine proposed questions.
Finally, while we proposed a framework adapted from existing models and using existing, pretested and sometimes partially validated indicators from the literature, further research is needed to assess the reliability and the validity of the indicators in other referral hospitals of different contexts including in hospitals not supported by an NGO[82].