Study design and data source

We used a concurrent mixed-methods design [30] incorporating routinely collected quantitative data derived from the health management information system (HMIS) in hospitals [31]. An interrupted time series design was applied for quantitative and an exploratory design for qualitative study. Quantitative and qualitative data were collected in parallel, and the first stage of analysis separately addressed the two types. A further integration phase was conducted to compare and contrast the quantitative and qualitative findings and assess whether they were corroborative or contradictory by making mixed-methods meta-inferences. The integration was conducted by joint display technique [30, 32] (Fig 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Concurrent mixed methods research on the impact of COVID-19 on ANC provision was adapted [30].

https://doi.org/10.1371/journal.pone.0301994.g001

Data collection methods for quantitative data

Data were collected from all 15 public hospitals providing comprehensive emergency obstetric care in the Sidama region at the time of the study. The sample included all pregnant women who attended ANC in the 12 months before the COVID-19 pandemic (March 2019 to February 2020) and during the six months of the pandemic (March to August 2020), totalling 21,000 women. The first author and research assistants (all with an MSc in clinical midwifery) extracted data from the hospitals’ HMIS. The first case of COVID-19 was officially reported in Ethiopia on 13 March 2020. We used the same months before and during the pandemic to assess the impact of COVID-19 on ANC uptake. Similarly, we included data from March 2019 to February 2020 to evaluate whether ANC was affected by other incidents before COVID-19 and whether these events affected ANC outcomes from March to August 2020.

Quantitative data processing and analysis

After screening the data, any questions regarding data clarity were resolved by revisiting the hospitals and HMIS offices to address any missing data in the provided Excel spreadsheet. Data were imported from Microsoft Excel into Stata version 17 for analysis. We performed an interrupted time series analysis (ITSA) to estimate trends in the uptake of ANC across two periods: before COVID-19 (March 2019 to February 2020) and during COVID-19 (March to August 2020). ITSA can evaluate the impact of population-level interventions, including policy changes and infection prevention programmes, implemented at a clearly defined time [33]. The first official COVID-19 case in Ethiopia was reported on 13 March 2020, so we used this date to mark the start of the intervention period. The mean monthly incidence rate ratio (IRR) of ANC uptake was calculated with a 95% confidence interval (CI), using a Poisson regression model [34, 35] with pre-COVID-19 data as the reference. A Poisson regression model was suitable because the monthly reports of ANC provision comprised count data (non-negative integer values). In ITSA, a Poisson regression model [36] performs better than an autoregressive integrated moving average (ARIMA) model, which is more conventionally used for real-valued time series data. Differences are considered statistically significant at a p-value of less than 0.05 (p < 0.05).

We used a single-group ITSA for this study [35] where Y_t is the aggregated outcome; β₀ estimates the ANC uptake number before COVID-19; β₁ estimates the average monthly change in ANC uptake before COVID-19; T_t is the time since the start of the study; β₂ represents the change in ANC uptake occurring immediately during the pandemic (within three months) (designated by X_tT_t); β₃ denotes the difference between the trends in ANC uptake before and during COVID-19; and β_m represents the month and ϵ_t the random error.

In this model, time is measured as a dummy variable, taking the value 0 for the period before COVID-19 and 1 for the period during COVID-19 (the intervention period) [35, 37].

Study approach for qualitative data

We adopted an exploratory design [38] to investigate maternity care providers’ views on and experiences of the impact of COVID-19 on ANC provision in the Sidama region. In-depth interviews (IDIs) were conducted with maternity care providers (midwives, obstetric/gynaecology residents, integrated emergency surgical officers [IESOs] and obstetricians/gynaecologists) in private duty rooms and offices in the region’s public hospitals. Four public hospitals were selected for the qualitative study. These four public hospitals (including two primary hospitals, one general hospital and one specialised hospital) were chosen for the qualitative study based on the caseload maternity care services provided and the order in which COVID-19 cases were initially reported in the Sidama region. Three different types of hospitals were selected: primary, general, and one specialised hospital that served as a referral centre for the Sidama region and the surrounding population in the Oromia region. This selection allowed for a nuanced understanding of the impact of the pandemic on various tiers of hospitals and their preparedness, response efficiency and the challenges they faced.

Participant recruitment and sampling technique for qualitative data

Within each chosen hospital, we explained the study’s purpose to the hospital medical director, chief executive director, and maternity care head, seeking their permission to conduct the research. Subsequently, two research assistants (both with an MSc in clinical midwifery) explained the study’s purpose in detail to maternity care providers who volunteered to be interviewed. We used purposive sampling to recruit staff who provided maternity care both before and during the pandemic. All participants provided written informed consent prior to being interviewed. We aimed to recruit approximately 20 participants (10 midwives and 10 obstetricians/gynaecologists). Data reached saturation at 24 interviews. We conducted another four interviews to confirm that data were saturated before ending qualitative data collection.

Collection tools and procedure for qualitative data

We developed a semi-structured interview guide comprising open-ended questions concerning the following factors: availability of and access to maternal and perinatal care; availability of adopted maternal and perinatal care guidelines related to COVID-19; availability of medical supplies and skilled healthcare personnel; and how challenges were overcome. The interview guide was piloted with midwives (N = 4) not included in this study. Two research assistants were recruited to facilitate the IDIs. The first author prepared the interview guide in English and translated it into Amharic, the official language of Ethiopia. In-depth face-to-face interviews were carried out in Amharic by the first author. Each interview was conducted in the maternity care duty room or office when participants were not on duty, and we fully adhered to the Ethiopian government’s COVID-19 prevention policy. The 28 interviews were conducted between 14 February and 10 May 2022, and each was digitally audio-recorded. Interview duration was approximately 30 minutes.

Qualitative data processing and analysis

The audio recordings were transcribed immediately and listened to iteratively. Simultaneously, bilingual researchers transcribed and translated transcripts into English to check consistency. The transcriptions were imported into NVivo software (QSR International, version 12 Plus) to manage the overall data analysis. Thematic analysis [39] was employed to identify, analyse, and report themes and subthemes. We used inductive thematic analysis and followed six phases: phase 1—data familiarisation and writing familiarisation notes; phase 2—systematic coding; phase 3—generating initial themes from coded data; phase four—developing and reviewing themes; phase 5—refining, defining, and naming themes; and phase 6—writing the report [39]. All authors reviewed the themes and subthemes in the thematic analysis phases (from coding to writing a report) and approved the final themes. This study is reported according to the Standards for Reporting Qualitative Research [40] (S1 Table) to ensure that essential details are reported and the thematic analysis is of sufficient quality [41].

Ethics approval and consent to participate

An internal research review board (IRB) at Hawassa University granted ethical clearance, and the University of Technology Sydney ethics committee approved the study with reference number IRB/029/14 (approval no, ETH22-7567) respectively. The research assistants explained the goal and advantages of the research project to each study participant during the data collection process. Before beginning the data collection process, each study participant provided written informed consent. Study participants were also informed of their full right to refuse, withdraw, or reject part or all of their roles in the study. All processes were carried out in accordance with the standards and laws outlined in the Declaration of Helsinki, and data were collected anonymously and kept confidential with the investigators.

Barriers to ANC access during COVID-19

Two subthemes emerged from our analysis of interview responses relevant to ANC access during COVID-19: ‘Shortage of resources’ and ‘Community discrimination against those attending the hospital’.

Shortage of resources

Participants indicated that a lack of resources during COVID-19 affected women’s access to ANC. In particular, women’s incomes declined, and public transport became more limited and doubled in cost. There was a shift of hospital resources towards COVID-19 prevention and treatment, but a lack of medical supplies, as evidenced by the following quote:

‘The taxi cost increased during COVID-19, and the women who live in rural areas did not come to the hospital due to transport costs. For example, a woman who came from another city paid double, and she might not come to the hospital [again]’ (Midwife RMP6).

Maternity care providers described how the price of medical supplies substantially increased due to shortages, to the extent that it was impossible for most people to buy them. The combination of Ethiopia’s socio-political situation and the COVID-19 pandemic prevented foreign aid from funding public hospitals, as illustrated by the following quote:

‘Medical supplies were tough [expensive]; for example, one glove was sold for up to 100 Ethiopian birr [$2 USD]. Foreign aid has decreased due to COVID-19 and the ongoing war in the country. The NGOs donated a lot of supplies [for women’s use] before COVID-19. NGOs ceased contributing medical supplies… no medical supplies in the hospital. Due to the economic crisis, women cannot afford to pay [$2USD] for a single glove’ (IESO ALTP14).

Community discrimination against those attending the hospital

Discrimination by the community against individuals attending the hospital had a detrimental impact on women’s ability to access ANC during COVID-19. Midwives stated that the community discriminated against those who attended hospitals, including both providers and receivers of care. Many community members believed those who visited the hospital could bring the virus to the community. This discrimination also affected healthcare providers, as they could not meet with their families or maintain their daily routines. The discrimination against maternity care providers is illustrated in the following quote:

‘We [health care providers] could not find food to eat or meet basic needs because of discrimination us in the town. Women and families discriminated against healthcare providers; they did not want us to live with them as we were involved in providing care in the hospital; they believed we could bring COVID-19 to the community, and they think COVID-19 is a killer’ (Midwife MALTP17).

‘It was a tough time; in my residential area, nurses and physicians could not obtain basic necessities and could not rent a place as a result of discrimination’ (Midwife WR1).

Discrimination resulted in a decline in social capital between healthcare providers and the community, resulting in limited access to ANC As a consequence, COVID-19 further damaged cultural cohesiveness between the community and healthcare providers, as exemplified by the following quote:

‘Social cohesion between healthcare providers and the community has broken down during COVID-19; for example, we [healthcare providers] could not attend funeral ceremonies. We are confined in the home. We live in a small town; everyone knows the healthcare providers, and they discriminated against us’ (Midwife MLP5).

Midwives reported similar discrimination against women who received care in hospitals. The community discriminated against women who sought any care at the hospital because they believed that anyone visiting the hospital would contract the virus and spread it around. Consequently, this discrimination resulted in limited access to ANC. As one midwife highlighted,

‘Pregnant women’s rapport with neighbours was affected when they came to [access] care in the hospital during COVID-19. After a [pregnant] woman returned home, her neighbours discriminated against her since she was [accessing] care in the hospital, and her [neighbours] believed she had contracted COVID-19’ (Midwife MAP14).

Barriers to ANC uptake during COVID-19

Two subthemes were identified within ‘Barriers to ANC uptake during COVID-19’ theme: ‘Fear of contracting COVID-19’ and ‘Decreased attendance of ANC’.

Fear of contracting COVID-19

The midwives highlighted that a significant number of women did not attend hospitals because they feared contracting COVID-19. Due to this fear, even women who attended and received care kept their distance from maternity care providers, believing that ‘the hospital was the epicentre of the virus’ (Midwife MRP1). Other midwives concurred:

‘Women came to [uptake] care in the hospital with fear. They feared healthcare providers. They considered healthcare providers to be a source of COVID-19, ha ha ha’ (Midwife MALTP16).

‘The women were afraid; as I told you before, they believed that if they came to the hospital for ANC, they would catch COVID-19’ (Midwife MLP5).

Midwives explained that pregnant women’s fear of contracting COVID-19 considerably affected practitioner–client communication during care provision. Maternity care providers felt that this fear prevented them from developing a strong rapport with women or spending sufficient time in consultations. Physical-distance policies stipulated that individuals must remain at least 1.5 metres apart. In addition, because of the perceived risk to their privacy resulting from physical distancing and weakened communication, women were reluctant to disclose their medical information to maternity care providers when attending the hospital. One midwife has the following explanation:

‘During the pandemic, fear reduced women’s communication with midwives. Women needed to discuss their private [sexual] issues by approaching us [healthcare providers], but distance made it impossible. When women came, we comforted them by touching their shoulders, but now it is not easy. It reduces something that we have had. Furthermore, we would keep our distance from them’ (Midwife MAP12).

Obstetricians and obstetric/gynaecology residents also described how the pandemic disrupted their relationships and communication with women. For fear of spreading the infection, at the peak of COVID-19, none would even touch a woman’s medical chart, and the fetal heartbeat was checked using a Doppler rather than a Pinard fetoscope. One resident commented:

‘The rapport between women and physicians clearly declined [during COVID-19]. You could not even touch a medical chart when COVID-19 was at its peak because of the news from Italy that physicians had contracted and died from COVID-19. The disease is transmitted by touching a medical chart. There was a great distance. As a result, there was a decline in the rapport between women and physicians. If physicians were doing these things, you might assume other healthcare providers were as well, and many things were missed in giving care’ (Resident 4 RRP10).

‘Maternity care providers’ rapport with women substantially decreased. For example, before the pandemic, we used Pinard fetoscope to check the fetal heartbeat. Nevertheless, there were numerous concerns about using a Pinard fetoscope to check a fetal heartbeat during the pandemic, so we switched to a Doppler to check the heartbeat. The maternity care providers believed that women with a cough had COVID-19, and so were unwilling to care for them’ (Obstetrician SRP19).

Decreased attendance of ANC

Reporting a decrease in ANC attendance during COVID-19, participants noted that women perceived the hospital as not providing the services and failing to address their needs. In addition, maternity care providers were not actively encouraging women to visit the hospital, leading to recommendations for staying at home, as noted in the following quote:

‘After Ethiopia reported the first COVID-19 case, maternal health services significantly decreased, and the number of women who came to the hospital, especially for their ANC follow-up, decreased because they were listening to other news about COVID-19 through various media outlets. Many women were staying at home. They believed that the hospital was not providing the services and did not take their needs into consideration when they came’ (IESO ALTP15).

Due to population growth, ANC attendance is expected to grow from one year to the next. However, during COVID-19, obstetricians reported declining daily ANC attendance: ‘women did not come on their appointment day’ (Obstetrician SAP22). One obstetrician expressed an opinion that the decline in ANC attendance was in part due to the messages recommending the community stay at home, and the fact that women were not encouraged to access the services by medical and midwifery staff at the hospital. Further evidence is provided by the following quote:

‘Before COVID-19, there were around ten women [per day], but during COVID-19, it was around five to seven women. Our population is increasing, but the number of women attending decreased during COVID-19. They came, we treated them, sent them home, and made predictions. As I had predicted, fewer people attended hospitals. According to my predictions, these decreases resulted from the fact that we did not encourage women to attend the hospital …there was a lack of maternity care providers in this hospital’ (Obstetrician SRP24).

Barriers to ANC provision during COVID-19

Four subthemes were identified within the theme ‘Barriers to ANC provision during COVID-19’: ‘Absence of ANC guidelines for care provision’, ‘Inadequate PPE’, ‘COVID-19 vaccine hesitancy’, and ‘Long waiting times for ANC’.

Absence of ANC guidelines for care provision

Participants acknowledged that the availability of ANC guidelines is crucial for providing high-quality care and evidence-based care. Most study participants reported that there were no COVID-19-related ANC guidelines in the hospitals. One obstetrician said, ‘We did not have any protocol… We continued to use the previous obstetric care guidelines’ (Obstetrician SRP 23). While there was some guidance from the WHO on emergency infection prevention, there was no specific protocol for the management of pregnancy care:

‘There is no unique protocol or approach that we adapted or adopted. The hospital got some aid from the regional health bureau and gave us essential personal protection equipment like gloves, masks, and sanitiser. Using this equipment, we admitted the women who came to the service. However, we did not prepare any COVID-19 protocol or design any strategies at the curriculum level’ (IESO ALTP15).

‘The WHO developed the guideline and then nationally adopted it. By chance, we have one guideline that was adopted and updated from the previous national guideline in obstetrics in 2020. However, the guideline was not related to COVID-19. Meanwhile, there is a protocol for emergency infection prevention and COVID-19 prevention protocol. It is used for all, including pregnant women. However, specifically, there is no guideline for pregnancy-related care during COVID-19’ (Resident 4 RRP 10).

Inadequate PPE

Midwives indicated that inadequate supplies of COVID-19 PPE were a common problem in hospitals during the pandemic. Maternity care providers noted an imbalance between demand and supply of PPE in the hospitals. The scarcity of ‘facemasks, face shields, shoes and other PPE’ (Midwife MAP 13), perhaps due to increased use of infection control supplies. The following quotes are illustrative:

‘Five masks were given to each healthcare provider per week, and one sanitiser was given to many healthcare providers, but it was not given individually. [Later] the number of masks was reduced from five to four per week’ (Midwife MALTP17).

‘Women bought a mask and used it together; when someone left the hospital, they handed it over to another person’ (Midwife MLP5).

COVID-19 vaccine hesitancy

At the time of data collection, only 9 of the 28 participating maternity care providers had received two doses of the COVID-19 vaccination. If maternity care providers are hesitant about the COVID-19 vaccine, it may impact pregnant women’s confidence in the vaccine’s safety and efficacy. This could lead to lower vaccine rates and hinder ANC provision. Some mentioned that they doubted the vaccine’s efficacy or believed that the infection was a punishment for disobeying God’s commandments, as exemplified in the following quote:

‘I have chosen not to get vaccinated. I do not believe in it, and I do not want the vaccine. My belief is that the outbreak of COVID-19 may be a consequence of our sins was caused by our sins and evil actions. I consider COVID-19 to be a form of divine punishment for our transgressions. Therefore, I believe we should fast and pray in an attempt to avoid this disease’ (Midwife MAP13).

Long waiting times for ANC

Waiting times for ANC provision were longer than usual during COVID-19. Priority was given to women wearing facemasks; hence, those without facemasks had to wait longer. As one midwife commented, ‘The women who wear facemasks get [to access] the service first; those who do not wear facemasks do not get [access to] care’ (Midwife MALTP17). Moreover, maternity care providers were unable to provide the service without face masks, causing delays in providing ANC in hospitals when facemasks were unavailable in the ANC clinic. Women who did not wear facemasks were not permitted to enter ANC clinics or receive immediate care. Despite the decline in ANC attendance, those women who did attend were met with delays in receiving care due to lack of staff, as illustrated by the following quote:

‘We were nagging patients who did not wear masks, we ordered a woman to wear a mask, and we left the room if she did not wear a facemask. However, we offered ANC during COVID-19 in a manner similar to that before the pandemic; all ANC components were offered during [COVID-19]. Although we performed complete physical examinations…comparable to those performed before COVID-19, the facemask did not bring comfort to women or those who provide maternity care and had a terrible effect during physical examinations’ (Midwife MALTP16).

‘Many women were waiting [to access] the services, crowded on ANC waiting chairs, and the waiting time was long because [only] one or two maternity care providers gave services. Women were desperate to get [access to] the service, and they [sometimes] returned to their homes without getting healthcare. We [used to provide] ultrasounds for all women during each visit. During COVID-19, we [only] performed ultrasounds for selected women’ (Obstetrician SRP23).

Integrating quantitative and qualitative findings using joint display technique

In the quantitative analysis, monthly trends of ANC attendance and the availability of essential drugs in hospitals were assessed before and during the pandemic. Simultaneously, three themes and eight subthemes were also identified in the qualitative data to explore whether the qualitative findings confirmed or disconfirmed the quantitative findings.

For ANC1 and ANC4 attendance, the quantitative findings revealed decreasing trends during the pandemic, and the qualitative study corroborated these declines, providing evidence that stay-at-home recommendations, lack of (and more expensive) transport, and fear of contracting the virus contributed to reducing the number of women attending hospitals for ANC (Table 2). As one participant commented, ‘the uptake of ANC4 declined due to fear of contracting the virus during COVID-19’ (Midwife MLP5).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Joint display of quantitative and qualitative findings for each theme and subtheme on impact of COVID-19 on ANC access, uptake and provision, and mixed-methods meta-inferences.

https://doi.org/10.1371/journal.pone.0301994.t002

By contrast, the quantitative and qualitative findings on essential drug availability were contradictory (Table 2). The quantitative analysis showed no significant difference in the availability of essential drugs in hospitals before and during the pandemic. However, access was impacted for women by the cost, and according to staff, by supply shortages, as the following quote demonstrates, ‘medical supplies decreased in the hospital, and the cost of medical supplies increased. All things were going up during the COVID-19 pandemic’ (IESO ALTP14). The same discrepancy was found for iron and folic acid availability, which did not significantly differ between before and during the pandemic according to the quantitative analysis but notably declined according to the qualitative analysis: ‘pregnant women faced challenges in accessing and taking iron-folic acid supplement appropriately during the pandemic’ (Midwife MALTP18). These inconsistencies could be explained by inadequacies in the HMIS inventory. The binary nature of data in the HMIS form means that responses are limited to ‘available’ or ‘unavailable’, with no space to include details of the quantities of essential drugs. Hence, the HMIS data do not adequately represent the scenario on the ground at the hospital pharmacy level (Table 2).

There is a lack of data to demonstrate quantitatively whether ANC guidelines are in place, the amount of PPE equipment available or vaccine rates to confirm qualitative findings. Additionally, there is no data to determine wait times for ANC, presence and severity of fear of contracting COVID-19, or measures of discrimination from the community. This information reported by care providers is similarly reported in the literature for other comparable countries (Table 2).