Mobile clinics have been proposed as a viable, alternative delivery model to increase ANC coverage, however, there is little prior evidence about quality of care in such clinics and how these compare to fixed clinics in the same regions. Our study compared fixed and mobile clinic quality in Central Haiti and identified differences in half of the care components studied (intake, receipt of lab exams, infection control and distribution of supplies, IFA and Tetanus Toxoid). Based on the relatively low percent of all possible services delivered in six of eight care components (excluding distribution of supplies, IFA and Tetanus Toxoid and documentation), we conclude that ANC quality was weak in both delivery models. Poorly delivered ANC is less likely to improve knowledge, behavioral or health outcomes, including detection of underlying health problems or pregnancy-related complications and proper management of danger signs, regardless of improvements in coverage.
Care components (process measures)
A potential reason for the low adherence to the Haitian guidelines in both clinic models is the short duration of consultations, likely caused by the high volume of women in some clinics and large range of numbers of women who seek ANC: 2–37 per day in fixed clinics and 3–55 per day in mobile clinics. Additionally, some providers may have other responsibilities, such as performing post-natal or well-baby visits, but these were not counted or observed in our study. Fixed clinic consultations lasted on average five and three minutes longer for first and follow-up visits, respectively, than those in mobile clinics. However, it is not evident how this additional time was spent; women were asked a slightly higher proportion of the intake questions and lab exams (usually performed in a lab, not the provider) yet did not receive more time-consuming services, such as the physical exam or education and counseling. Possibly providers spent more time on counseling but if so, the effect of this was not seen in women’s knowledge assessed in the survey. Potential reasons for variation in consultation duration could include level of health care worker training (although this was not significant between clinic models) and experience or maternal factors, such as educational levels or socio-cultural issues. Short duration of consultations has been shown to limit adherence to ANC guidelines in other countries and likely prevents providers from making correct diagnoses and management plans [
29].
Referrals were low for the three conditions studied in mobile clinics, most notably, for high blood pressure. High blood pressure and proteinuria are clinical criteria for identifying pre-eclampsia [
49]. Although we were only able to assess referral for one of these criteria, 95% of women with high blood pressure were not referred to a higher-level center so that proteinuria could be measured. This is a grave missed opportunity for identifying women with hypertensive disorders, the cause of 22% of maternal deaths in Latin America and the Caribbean [
50].
Two general trends emerge when we compare our study to other similar studies of ANC quality: 1) selective adherence to clinical guidelines and 2) better performance in the physical exam and health provider communication and interpersonal delivery components than education and counseling and intake. In our study and others, providers “pick and choose” services, providing some basic services to a high proportion of women and performing other services infrequently [
51‐
53]. For example, across studies, measurement of blood pressure and uterine height, and listening to the fetal heart rate, are much more commonly performed in the physical exam than checking for edema or weighing patients [
51‐
53]. This indicates that providers offer their version of “routine” care, which is incomplete according to clinical guidelines and not individualized to patient needs.
Women’s knowledge and perceived quality of care
Group education sessions were the primary means of education, commonly performed by community health workers or nurses aides while women waited for their consultation. Yet, had this group education been effective, a higher proportion of women would have been able to provide at least one correct message per educational topic. Other studies have also documented low levels of education, counseling and knowledge in essential areas of birth preparedness and emergency readiness, indicating that this is a wide-spread problem of ANC programs [
11,
53,
54].
Even with a low percent of clinical services being provided and poor education effectiveness, recipients of care perceived that they received high-quality care in both clinic types, which is slightly higher in mobile clinics. This paradox has been seen in other health contexts, where provider respect and politeness are deemed more important to patients than technical measures of care [
55]. In this study, providers were consistently seen as treating patients politely and with respect, which likely mediated women’s perceived quality of care [
56,
57]. Additionally, low consultation cost has been shown to increase patients’ perceptions of care quality in other developing countries [
58].
Although we found that the quality of antenatal care differed between fixed and mobile clinics for half of the care components studied, differences in intake and supplies, IFA and Tetanus Toxoid were small and likely not clinically significant, while differences in receipt of lab exams and infection control were large and likely clinically meaningful. We therefore conclude that mobile clinics can provide similar quality of ANC as fixed clinics in the majority of care components studied. Differences in infection prevention and control measures could be improved, even within the current structure of the mobile clinics. The lack of lab exams offered in the mobile clinics is a potential structural weakness of the model, as carried out in this context. However this limitation does not negate its other potential benefits, and instead suggests the need for an integrated system of ANC with strong coordination of care between mobile and fixed clinics to optimize the continuum of care. For example, with strong communication and referral structures, women could attend mobile clinics for routine care much closer to home, travelling to distant, fixed sites for initial lab exams and when referred. This will only be effective, however, with system-level improvements in both models of care to reduce the number of consultations performed by providers per day and lengthening the duration of visits. This could be achieved by hiring more health staff to perform ANC or possibly through scheduling more mobile clinics to reduce the number of ANC consultations per day. Both options would require additional financial resources. In conjunction, institution- and provider-level interventions, including education, active supervision, audit and feedback and job aids, can improve provider adherence to guidelines and making correct diagnoses [
29,
54,
59,
60].
This study took place in one of the poorest regions of Haiti, albeit one that receives substantial external support from NGOs. One example of this support is the strong supply chain for consumables, resulting in a high proportion of women who received IFA at their ANC visit. This finding is in contrast to many other studies in regions with less support, where supply chains function poorly [
11,
51,
53]. Small scale and short-term mobile clinics are commonly employed in Haiti by NGOs to deliver general or highly-specialized health services (when foreign medical specialists visit, for example), for the purpose of increasing coverage in a country with weak infrastructure, a low provider to population ratio and poor access to health services [
61]. However, the mobile clinic program described in this paper was unique in its large scale and duration (approximately 10 years, through two, five year grants). These results are therefore generalizable only to the Central Plateau of Haiti or similar regions where the health infrastructure receives external financial and technical support, yet poor health outcomes persist. Additionally, mobile clinics can be adapted to different contexts, and these adaptations will have varying impacts on quality.
There are some limitations of our study. This data was collected in 2012. Since that time the WHO released new, women-centered ANC guidelines [
62]. We are unaware of any efforts to align national guidelines with these new guidelines, nor any large-scale efforts to improve ANC nationally. Additionally, we removed two important variables from the dataset, checking the woman’s conjunctiva and palpation of abdomen for fetal presentation due to poor reliability.
Observations are considered the “gold standard” for assessing quality of health care implementation, because they reduce potential risk of recall bias and poor or incomplete documentation [
63,
64]. However, observations are limited by what can be visually or audibly assessed, and have the potential to alter client and patient behavior [
63]. It is commonly assumed that observed health care providers are on their “best behavior” while observed. However, evidence has shown that providers quickly grow accustomed to being observed, minimizing this effect, especially if they do not know the study objective [
24,
65].
We attempted to minimize any observer effect in the study design and analysis. First, neither the purpose of the study nor the study instruments were shared with care providers. Second, observers spent at least a full day with care providers, enough time for providers to resume their normal behaviors [
65]. We tested if the order patients were seen each day altered the analyses, but it did not. Third, we explored the possibility that providers observed more than once might become used to being observed, resulting in a qualitative difference between being observed the first or future days, but did not find differences for any outcome. Given these results, and no reason to suspect a differential observation effect between fixed and mobile clinics, we believe that an observation effect did not substantially affect our conclusions.
It is possible that respondents in the exit survey were influenced by courtesy bias, compelling them to inflate their responses of perception quality of care. We attempted to minimize this effect by asking a mix of objective and subjective questions, as objective questions have been shown to be more comparable to results from community-based surveys of health care quality [
58]. We found no difference in responses between these two types of questions.
Finally, observation of a single ANC visit, while informative about routine quality of care delivered in this region of Haiti, does not describe the sum of care received by women over the course of pregnancy. Our previous work on mobile clinics revealed that 54% of women who sought ANC at mobile clinics also sought care with another ANC provider [
45]. It is possible, therefore, that by visiting multiple providers, women are, in fact, receiving more care than we were able to observe, although how the sum of this care compares to Haitian guidelines is not known.