You are here

  • Home
  • Archive
  • Volume 12, Issue 6
  • Indirect impacts of the COVID-19 pandemic at two tertiary neonatal units in Zimbabwe and Malawi: an interrupted time series analysis

Article Text

Article menu
Global health
Original research
Indirect impacts of the COVID-19 pandemic at two tertiary neonatal units in Zimbabwe and Malawi: an interrupted time series analysis
  1. Simbarashe Chimhuya1,
  2. Samuel R Neal2,
  3. Gwen Chimhini1,
  4. Hannah Gannon2,
  5. Mario Cortina Borja2,
  6. Caroline Crehan2,
  7. Deliwe Nkhoma3,
  8. Tarisai Chiyaka4,
  9. Emma Wilson2,
  10. Tim Hull-Bailey2,
  11. Felicity Fitzgerald2,
  12. Msandeni Chiume5,
  13. Michelle Heys2
  1. 1Child and Adolescent Health Unit, University of Zimbabwe, Harare, Zimbabwe
  2. 2UCL Great Ormond Street Institute of Child Health, University College London, London, UK
  3. 3Parent and Child Health Initiative, Lilongwe, Malawi
  4. 4Biomedical Research and Training Institute, Harare, Zimbabwe
  5. 5Department of Paediatrics, Kamuzu Central Hospital, Lilongwe, Malawi
  1. Correspondence to Dr Michelle Heys; m.heys{at}ucl.ac.uk

Abstract

Objectives To examine indirect impacts of the COVID-19 pandemic on neonatal care in low-income and middle-income countries.

Design Interrupted time series analysis.

Setting Two tertiary neonatal units in Harare, Zimbabwe and Lilongwe, Malawi.

Participants We included a total of 6800 neonates who were admitted to either neonatal unit from 1 June 2019 to 25 September 2020 (Zimbabwe: 3450; Malawi: 3350). We applied no specific exclusion criteria.

Interventions The first cases of COVID-19 in each country (Zimbabwe: 20 March 2020; Malawi: 3 April 2020).

Primary outcome measures Changes in the number of admissions, gestational age and birth weight, source of admission referrals, prevalence of neonatal encephalopathy, and overall mortality before and after the first cases of COVID-19.

Results Admission numbers in Zimbabwe did not initially change after the first case of COVID-19 but fell by 48% during a nurses’ strike (relative risk (RR) 0.52, 95% CI 0.41 to 0.66, p<0.001). In Malawi, admissions dropped by 42% soon after the first case of COVID-19 (RR 0.58, 95% CI 0.48 to 0.70, p<0.001). In Malawi, gestational age and birth weight decreased slightly by around 1 week (beta −1.4, 95% CI −1.62 to −0.65, p<0.001) and 300 g (beta −299.9, 95% CI −412.3 to −187.5, p<0.001) and outside referrals dropped by 28% (RR 0.72, 95% CI 0.61 to 0.85, p<0.001). No changes in these outcomes were found in Zimbabwe and no significant changes in the prevalence of neonatal encephalopathy or mortality were found at either site (p>0.05).

Conclusions The indirect impacts of COVID-19 are context-specific. While our study provides vital evidence to inform health providers and policy-makers, national data are required to ascertain the true impacts of the pandemic on newborn health.

  • COVID-19
  • paediatrics
  • neonatal intensive & critical care
  • quality in health care
  • international health services

Data availability statement

Data are available on reasonable request. Data collected for the study cannot yet be made publicly available because primary analysis for the pilot implementation evaluation of the Neotree, as well as secondary analysis are ongoing. A goal of our pilot implementation is to establish an open-source anonymised research database of Neotree data to maximise the reach and utility for researchers aiming to improve outcomes for neonates in low-income settings. This database is under development and subject to negotiation with relevant Ministries of Health.

https://creativecommons.org/licenses/by/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See: https://creativecommons.org/licenses/by/4.0/.

http://dx.doi.org/10.1136/bmjopen-2021-048955

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    Strengths and limitations of this study

    • We address the need for increased research into the indirect impacts of the COVID-19 pandemic on neonatal care in low-income and middle-income countries.

    • We collected data digitally and in real time using the Neotree application, which enabled a large sample size of 6800 neonates with minimal missing data.

    • It is possible that unobserved events occurred close to the first case of COVID-19 in either country, which could have influenced our results.

    • We only collected data on neonates admitted to the neonatal unit and did not capture stillbirths or neonatal deaths that occurred in the community.

    Introduction

    The WHO declared COVID-19 a Public Health Emergency of International Concern on 30 January 2020.1 Almost 2 years later, confirmed cases have exceeded 281 million globally with over 5.4 million deaths to the end of 2021.2 Zimbabwe recorded its first case on 20 March 2020 and, to date, has reported over 200 000 cases with nearly 5000 deaths.2 Malawi confirmed its first three cases on 3 April 2020 and has reported more than 72 000 cases and over 2000 deaths in this same period.2

    Before the COVID-19 pandemic, considerable improvements were made in global child health: the global neonatal mortality rate fell from 31 to 18 deaths per 1000 live births between 2000 and 2018.3 Yet there were disparities in the rates of decline with the sub-Saharan Africa region facing highest neonatal mortality rates.3 Now, there is a danger that health outcomes in low-income and middle-income countries (LMICs) will fall further behind high-income countries. While countries worldwide face challenges related to the COVID-19 pandemic, LMICs are particularly struggling with financial constraints, limited testing capacity, lack of personal protective equipment, staff shortages,4 5 and limited access to vaccines.6 As children are at low risk of infection or severe disease from COVID-19,7–11 any impacts on their health outcomes will likely be attributable to the indirect effects of the pandemic on health systems, as in previous disease outbreaks.12 13 These include increased rates of parental unemployment, food and housing insecurity, and reduced access to routine care, including antenatal and perinatal care, with potentially damaging downstream impacts on neonatal outcomes.14 15

    We hypothesised that the COVID-19 pandemic would negatively impact care seeking behaviours, neonatal care provision and, ultimately, neonatal outcomes in LMICs. To test this hypothesis, we aimed to examine trends in markers of neonatal care before and during the initial months of the COVID-19 pandemic at Sally Mugabe Central Hospital (SMCH), Zimbabwe and Kamuzu Central Hospital (KCH), Malawi. Specifically, we compared the:

    1. Number of admissions to the neonatal unit (NNU).

    2. Gestational age and birth weight of admitted neonates.

    3. Source of admission referrals.

    4. Prevalence of neonatal encephalopathy (NE).

    5. Overall mortality rate before and after the first reported cases of COVID-19.

    Methods

    This study is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology statement (online supplemental appendix 1).

    Setting

    Health facilities

    SMCH is a public referral hospital in Harare, Zimbabwe. It has the largest of three tertiary NNUs nationwide with 100 cots. KCH, Lilongwe, is one of four regional referral hospitals in Malawi and the NNU has 75 cots. Neonatal care at SMCH is predominantly doctor led while neonatal care at KCH is mostly nurse led. Both units accept local and national referrals for specialist surgical care.

    Government response to the pandemic

    In response to the COVID-19 pandemic, Zimbabwe and Malawi both implemented response measures in an attempt to control the outbreak. In Zimbabwe, the Government closed borders to non-essential travel within days of the first in-country confirmed case of COVID-19 and imposed a full national lockdown that lasted from 30 March to 11 June 2020, which was followed by phased relaxations of the restrictions.16 In Malawi, public events were banned and public gatherings restricted to fewer than 100 people on 20 March 2020, with all educational institutions closed several days later.17 Borders were closed to non-essential travel on 1 April 2020 and a full national lockdown was announced to last for 21 days from 18 April 2020; however, a High Court injunction prevented this. Further restrictions were announced on 9 August 2020, mandating the wearing of face masks in public, closing places of worship, restaurants, and bars, and restricting public gatherings to less than 10 people initially, although these were revised within days to reallow gatherings up to 100 people.18

    Industrial action by health workers in Zimbabwe

    Two periods of national industrial action occurred in Zimbabwe during our study. Doctors went on strike from 3 September 2019 to 22 January 2020 (pre-COVID-19 period) citing insufficient pay and poor working conditions, which put significant pressure on the public health system.19 Additionally, there was a period of strikes by nurses from 17 June to 9 September 2020 (post-COVID-19 period) over pay and availability of personal protective equipment during the pandemic.20

    Participants

    All neonates admitted to each NNU over a 16-month period from 1 June 2019 to 25 September 2020 (69 complete weeks) were eligible for inclusion. We applied no specific exclusion criteria.

    Data collection

    Data were collected prospectively using the Neotree application (app), an Android tablet-based quality improvement platform that aims to reduce neonatal mortality in low-resource settings.21 Developed in collaboration with local stakeholders, it is embedded in routine practice at two NNUs in Zimbabwe and Malawi, providing real-time clinical decision support, neonatal care education and digital data capture.22 23

    Health workers complete a digital form when a neonate is admitted to the unit (admission form) and when they are discharged or die (outcome form). The app guides assessment of the neonate and collects data on patient demographics, examination findings, diagnoses and interventions. Pseudonymised forms are uploaded monthly to University College London servers (Zimbabwe data) and Amazon Web Services (Malawi data). Admission and outcome forms are linked by a unique identifier generated by the app at admission.

    Outcomes

    We evaluated five outcomes:

    1. Number of admissions: determined from the admission date of each completed admission form.

    2. Gestational age at birth (weeks) and birth weight (grams): as entered into the admission form from obstetric records.

    3. Source of admission: defined as ‘within’ (labour ward, postnatal ward, antenatal ward, obstetric theatre or fee-paying ward (KCH only)) or ‘outside’ (referral from another health facility or postnatal self-referral from home).

    4. Diagnosis of NE: defined as ‘hypoxic ischaemic encephalopathy’ or ‘birth asphyxia’ recorded as a diagnosis, cause of death or contributory cause of death on the outcome form.

    5. Mortality: defined as an outcome of “neonatal death” on the outcome form. All other neonates, including those discharged, transferred to another facility or who left on parental request, were considered alive.

    Statistical analysis

    Analyses were performed in R V.3.6.3,24 running on RStudio V.1.2.5033.25 First, admission forms were matched with their corresponding outcome form based on the unique identifier generated at admission. Lack of completed outcome forms (SMCH: n=325 (9.4% of admission forms completed); KCH: n=245 (7.3%)) or errors in entry of the unique identifier at discharge (SMCH: n=310 (9.9% of outcome forms completed); KCH: n=182 (5.9%)) meant we were unable to match some admission forms with outcome forms (SMCH: n=635 (18.4% of admission forms completed); KCH: n=427 (12.7%)). For outcomes 1–3, we based analyses on data from all admission forms, regardless of match status. For outcomes 4 and 5, we based analyses on matched records only. Matched records implying a negative admission duration (ie, outcome date prior to admission date) were excluded (SMCH: n=57 (2.0% of matched records); KCH: n=24 (0.8%)). See online supplemental appendix 3 for a flow diagram of record inclusion. Missing data were excluded using pairwise deletion for each analysis as frequencies of missing values were minimal (online supplemental appendix 4).

    This study used an interrupted time series design with weekly data windows. We considered the first confirmed case of COVID-19 in each country as the intervention (Zimbabwe: 20 March 2020; Malawi: 3 April 2020).2 For all outcomes, we hypothesised a level change impact model without a lag, and this was tested using interrupted time series regression models.26 Gestational age and birth weight were modelled with linear regression. Count data were modelled using generalised linear models with Poisson or negative binomial responses and logarithmic link functions. We assessed for dispersion by dividing the residual deviance by the df for the Poisson model. Where this quotient was much greater than one (greater than approximately 1.10) we instead used a negative binomial model to account for overdispersion. Accordingly, source of admission referral, prevalence of NE and overall mortality at SMCH were modelled using Poisson models, while number of admissions and overall mortality at KCH were modelled using negative binomial models.

    All models for SMCH were adjusted for the periods of doctors’ strikes (3 September 2019 to 22 January 2020) and nurses’ strikes (17 June to 9 September 2020). For count data, we adjusted for variation in the number of admissions over time by including the logarithm of the number of admissions in each weekly window as an offset term. Presence of autocorrelation was assessed using autocorrelation function plots and by examining models’ residuals. Seasonality was included in the interrupted time series models with cosine functions with variable amplitude and shift. We tested models fitting cosine functions on week of admission with 6-month and 12-month periods, and a model including these two harmonic terms. To achieve this, we transformed each cosine function into a sine term and cosine term, and included these terms in the regression models for each outcome (as described by Stolwijk et al).27 The final models presented were selected by minimising the Bayesian information criterion and by comparing goodness-of-fit with the χ2 test for nested models. Adjusting for seasonality did not improve the fit of any of the models tested and, thus, all presented models are unadjusted for seasonality. See online supplemental appendix 5 for model selection and estimates.

    Patient and public involvement

    Although patients and the public were not directly involved in this study, within the broader Neotree co-development project we are carrying out a series of workshops and focus group discussions with healthcare workers and parents of admitted babies to ensure local ownership and relevance of this digital quality involvement tool aimed at improving healthcare outcomes for vulnerable neonates.

    Results

    Outcome 1: admissions to the NNU

    We included 3450 neonates at SMCH and 3350 neonates at KCH. Figure 1 shows the 7-day moving average of admissions to the NNU.

    Figure 1
    Figure 1

    Trend in daily admissions to the neonatal unit. The 7-day moving average of daily admission numbers has been plotted. Smoothed line: local regression (LOESS) model fitted on the 7-day moving average of daily admission numbers; shaded region: 95% CI. Solid vertical line: first confirmed case of COVID-19 in each country. Shaded periods on SMCH, Zimbabwe panel: industrial action by doctors (3 September 2019 to 22 January 2020) and nurses (17 July 2020 to 9 September 2020). Counts based on all admission forms completed, irrespective of match status. KCH, Kamuzu Central Hospital; SMCH, Sally Mugabe Central Hospital.

    At SMCH, the mean (SD) number of weekly admissions was 54.6 (23.5) before the first case of COVID-19 (pre-COVID-19) and 42.8 (19.9) afterwards (post-COVID-19). The negative binomial regression model showed no evidence of a change in admissions after the first case of COVID-19 (relative risk (RR) 0.87; 95% CI 0.65 to 1.17; p=0.37) (figure 2A). However, this model estimated that admissions fell by 48% during the nurses’ strike period (RR 0.52, 95% CI 0.41 to 0.66; p<0.001) and by 51% during the pre-COVID-19 doctors’ strikes (RR 0.49, 95% CI 0.41 to 0.60; p<0.001).

    Figure 2
    Figure 2

    Interrupted time series for weekly admissions to the neonatal unit. White background: pre-COVID-19 period; grey background: post-COVID-19 period. Solid line: predicted trend from negative binomial regression model; dashed line: counterfactual scenario. SMCH model (A) adjusted for doctors’ and nurses’ strike periods; KCH model (B) unadjusted. Counts based on all admission forms completed, irrespective of match status. KCH, Kamuzu Central Hospital; SMCH, Sally Mugabe Central Hospital.

    At KCH, the mean (SD) number of weekly admissions was 54.5 (10.8) in the pre-COVID-19 period and 38.0 (10.9) in the post-COVID-19 period. The negative binomial regression model yielded a 42% reduction in admissions after the first case of COVID-19 (RR 0.58; 95% CI 0.48 to 0.70; p<0.001) (figure 2B).

    Outcome 2: gestational age and birth weight

    At SMCH, the mean (SD) gestational age at birth was 36.1 (4.4) weeks in the pre-COVID-19 period and 36.0 (4.2) weeks in the post-COVID-19 period. The mean (SD) birth weight was 2500 (908) g in the pre-COVID-19 period and 2487 (896) g in the post-COVID-19 period. Linear regression analysis indicated no significant change in gestational age at birth nor birth weight after the first case of COVID-19 (gestational age: beta 0.07; 95% CI −0.50 to 0.64; p=0.81, birth weight: beta 3.4; 95% CI −117.0 to 123.8; p=0.96) (online supplemental figure 1A, C).

    At KCH, the mean (SD) gestational age was 35.0 (3.9) weeks in the pre-COVID-19 period and 34.8 (3.9) weeks in the post-COVID-19 period. The mean (SD) birth weight was 2402 (883) g in the pre-COVID-19 period and 2299 (870) g in the post-COVID-19 period. Gestational age significantly decreased by 1 week in the post-COVID-19 period (beta −1.14; 95% CI −1.62 to −0.65; p<0.001) (online supplemental figure 1B) and birth weight significantly decreased by 300 g (beta −299.9; 95% CI −412.3 to −187.5; p<0.001) (online supplemental figure 1D).

    Outcome 3: source of admission referral

    At SMCH, the mean (SD) percentage of outside referrals to the NNU was 39 (11)% in the pre-COVID-19 period and 35 (9)% in the post-COVID-19 period. The Poisson regression model showed no evidence of a change in the percentage of outside referrals after the first case of COVID-19 (RR 0.97; 95% CI 0.77 to 1.22; p=0.81) (figure 3A). However, this model did imply a 39% relative increase in the percentage of outside referrals during the doctors’ strikes in the pre-COVID-19 period (RR 1.39; 95% CI 1.20 to 1.61; p<0.001).

    Figure 3
    Figure 3

    Interrupted time series for outside referrals to the neonatal unit. White background: pre-COVID-19 period; grey background: post-COVID-19 period. Solid line: predicted trend from Poisson regression model; dashed line: counterfactual scenario. SMCH model (A) adjusted for doctors’ and nurses’ strike periods, KCH model (B) unadjusted. Data from all admission forms completed, irrespective of match status. KCH, Kamuzu Central Hospital; SMCH, Sally Mugabe Central Hospital.

    At KCH, the mean (SD) percentage of outside referrals was 61 (8)% in the pre-COVID-19 period and 51 (10)% in the post-COVID-19 period. Poisson regression analysis resulted in a 28% relative reduction in outside referrals after the first case of COVID-19 (RR 0.72; 95% CI 0.61 to 0.85; p<0.001) (figure 3B).

    Outcome 4: prevalence of NE

    At SMCH, the mean (SD) percentage of admitted neonates diagnosed with NE was 16 (6)% in the pre-COVID-19 period and 21 (12)% in the post-COVID-19 period suggesting a possible increase. Poisson regression analysis showed no statistically significant change in the percentage of neonates diagnosed with NE post-COVID-19 (RR 1.06; 95% CI 0.74 to 1.52; p=0.74) (online supplemental figure 2A).

    At KCH, the mean (SD) percentage of admitted neonates diagnosed with NE was 15 (6)% in the pre-COVID-19 period and 13 (5)% in the post-COVID-19 period. The Poisson regression model implied a possible increase in diagnoses of NE after the first case of COVID-19, but this was not statistically significant (RR 1.31; 95% CI 0.91 to 1.88; p=0.15) (online supplemental figure 2B).

    Outcome 5: overall mortality

    For SMCH, the mean (SD) percentage of deaths per week of admission was 25 (10)% in the pre-COVID-19 period and 26 (16)% in the post-COVID-19 period. The negative binomial regression model pointed towards a possible decrease in mortality after the first case of COVID-19, but this was not statistically significant (RR 0.72; 95% CI 0.52 to 1.00; p=0.05) (figure 4A). However, this model did show an 81% relative increase in mortality during the nurses’ strike period (RR 1.81; 95% CI 1.31 to 2.49; p<0.001).

    Figure 4
    Figure 4

    Interrupted time series for overall mortality. White background: pre-COVID-19 period; grey background: post-COVID-19 period. Solid line: predicted trend from negative binomial regression model (SMCH, A) or Poisson regression model (KCH, B); dashed line: counterfactual scenario. SMCH model (A) adjusted for doctors’ and nurses’ strike periods; KCH model (B) unadjusted. Data from matched admission and outcome forms only. KCH, Kamuzu Central Hospital; SMCH, Sally Mugabe Central Hospital.

    For KCH, the mean (SD) percentage of deaths per week of admission was 19 (6)% in the pre-COVID-19 period and 23 (10)% in the post-COVID-19 period. The Poisson regression model implied a possible increase in mortality after the first case of COVID-19, but this was not statistically significant (RR 1.31; 95% CI 0.97 to 1.76; p=0.08) (figure 4B).

    Discussion

    Summary

    We performed an interrupted time series analysis to examine changes in neonatal care provision at two tertiary NNUs in Zimbabwe and Malawi after the first cases of COVID-19 in each country. We found that admissions at SMCH did not change significantly after the first case of COVID-19 when considering this period as a whole, but there was a considerable decrease (around 50%) in the number admissions in June to August 2020, coinciding with a nurses’ strike. We did not find significant changes in gestational age or birth weight, source of admission referrals, prevalence of NE or mortality at SMCH. Conversely, we found several changes in markers of neonatal care at KCH after the first case of COVID-19 in Malawi. The number of admissions fell by 42% and we noted a decrease in the gestational age and birth weight of admitted neonates (by around 1 week and 300 g, respectively), and a 28% relative decrease in outside referrals after the first case of COVID-19. Although this study is descriptive, we can speculate about explanations for our results based on existing literature and discussions with local health workers.

    Interpretation

    The number of admissions at SMCH fell by around 50% between June to August 2020, but we noted no change outside this strike period, suggesting some resilience to the impact of the pandemic. However, nurses went on strike over pay and availability of personal protective equipment,20 so the strike is itself an indirect consequence of COVID-19. A recently published audit of maternal health service provision at two tertiary hospitals in Harare, Zimbabwe (including SMCH) found a 25% reduction in hospital deliveries and an increased odds of stillbirth (OR 1.8; 95% CI 1.5 to 2.2) in March to August 2020 compared with the same period in 2019,28 which might partially explain the reduction in admissions to the NNU. A similar reduction in admissions was seen at KCH, but, unlike at SMCH, this 42% decrease was noted within a week of the first case of COVID-19. In figure 5, we propose several interlinked factors that might explain reduced admissions to the NNU. Several of these factors, such as fear of using health services, disrupted transport networks and staff shortages have been directly reported by local sources in low-resource settings and were highlighted in a recent report by Graham et al.29

    Figure 5
    Figure 5

    Possible factors influencing the decrease in admissions to the neonatal unit. Delays (red boxes) derived from the ‘three delays’ model of pregnancy-related mortality.36 PPE, personal protective equipment.

    We found a slight decrease in gestational age and birth weight of neonates at KCH, but not SMCH. Studies have reported increased rates of preterm birth in pregnant women with COVID-19 compared with those without the disease, mostly from medically induced preterm birth; although none of these studies were conducted in LMICs.30 Preliminary analysis suggests rates of emergency caesarean section increased at SMCH and KCH, with a more marked increase at KCH (online supplemental appendix 6). This is one potential explanation for our findings. However, we noted that the number of outside referrals decreased by 28% at KCH, and neonates referred from outside KCH are more likely to be from lower-risk pregnancies that delivered in a health centre with higher gestational ages and birth weights. Further analysis should stratify by source of admission referral to clarify this finding, but the relative reduction in outside referrals is supported by the fact that referrals were rigorously triaged by the on-call paediatrician during the pandemic, and that referrals from some areas were diverted away from KCH to more appropriate centres for the level of care required.

    We hypothesised that rates of NE would increase during the pandemic. NE is the clinical manifestation of disordered brain function and can have multiple aetiologies.31 The term ‘hypoxic-ischaemic encephalopathy’ is reserved for cases where there is evidence of intrapartum asphyxia.31 In LMICs, obstructed labour is a major cause of maternal mortality and can lead to intrapartum asphyxia with subsequent neonatal morbidity and mortality, including NE.32 Therefore, the prevalence of NE might be expected to increase as a marker of delayed presentation to a health facility. It is reassuring that we did not find increased rates of NE at SMCH or KCH. However, these findings should be interpreted cautiously as some neonates with NE may not have presented to a health facility at all, for example, due to an increased number of home deliveries, as documented in other sub-Saharan countries.33

    Finally, we did not observe a significant change in overall mortality at KCH nor SMCH, except during the nurses’ strikes at SMCH. In fact, there was a suggestion that mortality decreased after the first case of COVID-19 in Zimbabwe when adjusted for the nurses’ strike period, but this was not statistically significant. The reasons for this are unclear but could include factors such as increased stillbirth rates or improved care for the smaller number of neonates on the NNU. More complete analysis of facility-based and community-based neonatal mortality is greatly needed.

    Limitations and future work

    A limitation intrinsic to interrupted time series analysis is the possibility that another event occurred close to the first case of COVID-19 in either country causing spurious observations. Another potential threat to validity is changing data collection practices. For example, overstretched clinicians might not input data into the Neotree app for all admitted neonates. However, this is unlikely as the Neotree app is embedded into routine practice at SMCH and KCH and discussions with local collaborators suggest use of the app has continued without issue. At present, there is limited guidance on power and sample size calculations for interrupted time series analyses.34 Therefore, we did not perform specific power calculations and relied on the data available at the time of analysis. Also, our results suggest that our study has relatively low power to detect true changes in some outcomes, particularly NE, so these results should be interpreted cautiously in the absence of further data.

    The Neotree app only collects data on neonates admitted to the NNU. Therefore, our analysis does not capture stillbirths or neonatal deaths that occur in the community. It is troubling to see a dramatic fall in admissions at both sites, raising the possibility that many unwell neonates did not attend a health facility and died at home. A recent study found that facility births decreased by over 50% during the lockdown in Nepal, and facility stillbirth and neonatal mortality rates increased significantly.35 The Neotree research team is currently collecting data on stillbirths at SMCH and KCH, but these data will still only represent stillbirths that occurred in a health facility. Given the COVID-19 pandemic is not over, it will be important to repeat our analysis to further examine longer-term trends in neonatal care provision.

    Conclusion

    The indirect impacts of COVID-19 are context-specific, with more significant and evident effects on neonatal care provision seen at KCH (Malawi) than SMCH (Zimbabwe). While this study provides vital evidence to inform health providers and policy makers, national data are required to ascertain the true impacts of the pandemic on newborn health.

    Data availability statement

    Data are available on reasonable request. Data collected for the study cannot yet be made publicly available because primary analysis for the pilot implementation evaluation of the Neotree, as well as secondary analysis are ongoing. A goal of our pilot implementation is to establish an open-source anonymised research database of Neotree data to maximise the reach and utility for researchers aiming to improve outcomes for neonates in low-income settings. This database is under development and subject to negotiation with relevant Ministries of Health.

    Ethics statements

    Patient consent for publication

    Ethics approval

    Research ethics approval was granted by the UCL Research Ethics Committee (17123/001) and ethics committees in Malawi (P.01/20/2909) and Zimbabwe (MRCZ/A/2570) (online supplemental appendix 2). The need to obtain informed consent was waived as we collected only pseudonymised data routinely documented for clinical care.

    Acknowledgments

    We are very grateful to the families at SMCH and KCH, and the staff members at both hospitals for their enthusiasm and commitment to the Neotree project, without which this work would not be possible.

    References

    1. ↵
      1. World Health Organization
      . Statement on the second meeting of the International health regulations (2005) emergency Committee regarding the outbreak of novel coronavirus (2019-nCoV), 2020. Available: https://bit.ly/3jihlAU [Accessed 04 Sep 2020].
    2. ↵
      1. World Health Organization
      . WHO coronavirus disease (COVID-19) Dashboard, 2021. Available: https://covid19.who.int [Accessed 31 Dec 2021].
    3. ↵
      1. United Nations Inter-agency Group for Child Mortality Estimation
      . Levels & Trends in Child Mortality: Report 2019. New York United Nations Children’s Fund; 2019.
    4. ↵
      1. Makoni M
      . COVID-19 worsens Zimbabwe's health crisis. Lancet 2020;396:457.doi:10.1016/S0140-6736(20)31751-7pmid:http://www.ncbi.nlm.nih.gov/pubmed/32798478
      OpenUrlPubMed
    5. ↵
      1. Truscott R
      . Covid-19: health worker strikes, limited testing, and clinic closures hamper Zimbabwe's response. BMJ 2020;370:m3267.doi:10.1136/bmj.m3267pmid:http://www.ncbi.nlm.nih.gov/pubmed/32816753
      OpenUrlFREE Full Text
    6. ↵
      1. World Health Organization
      . Less than 10% of African countries to hit key COVID-19 vaccination goal, 2021. Available: https://bit.ly/3EDj6mN [Accessed 31 Dec 2021].
    7. ↵
      1. Castagnoli R,
      2. Votto M,
      3. Licari A, et al
      . Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection in children and adolescents. JAMA Pediatr 2020;174:882.doi:10.1001/jamapediatrics.2020.1467
    8. ↵
      1. Götzinger F,
      2. Santiago-García B,
      3. Noguera-Julián A, et al
      . COVID-19 in children and adolescents in Europe: a multinational, multicentre cohort study. Lancet Child Adolesc Health 2020;4:653–61.doi:10.1016/S2352-4642(20)30177-2pmid:http://www.ncbi.nlm.nih.gov/pubmed/32593339
      OpenUrlPubMed
    9. ↵
      1. Ludvigsson JF
      . Systematic review of COVID-19 in children shows milder cases and a better prognosis than adults. Acta Paediatr 2020;109:1088–95.doi:10.1111/apa.15270pmid:http://www.ncbi.nlm.nih.gov/pubmed/32202343
      OpenUrlCrossRefPubMed
    10. ↵
      1. Mantovani A,
      2. Rinaldi E,
      3. Zusi C, et al
      . Coronavirus disease 2019 (COVID-19) in children and/or adolescents: a meta-analysis. Pediatr Res 2021;89:733–7.doi:10.1038/s41390-020-1015-2pmid:http://www.ncbi.nlm.nih.gov/pubmed/32555539
      OpenUrlPubMed
    11. ↵
      1. Viner RM,
      2. Mytton OT,
      3. Bonell C
      . Susceptibility to and transmission of COVID-19 amongst children and adolescents compared with adults: a systematic review and meta-analysis. medRxiv 2020.
    12. ↵
      1. Sochas L,
      2. Channon AA,
      3. Nam S
      . Counting indirect crisis-related deaths in the context of a low-resilience health system: the case of maternal and neonatal health during the Ebola epidemic in Sierra Leone. Health Policy Plan 2017;32:iii32–9.doi:10.1093/heapol/czx108pmid:http://www.ncbi.nlm.nih.gov/pubmed/29149310
      OpenUrlCrossRefPubMed
    13. ↵
      1. Yerger P,
      2. Jalloh M,
      3. Coltart CEM, et al
      . Barriers to maternal health services during the Ebola outbreak in three West African countries: a literature review. BMJ Glob Health 2020;5:e002974.doi:10.1136/bmjgh-2020-002974pmid:http://www.ncbi.nlm.nih.gov/pubmed/32895217
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Ahmed S,
      2. Mvalo T,
      3. Akech S, et al
      . Protecting children in low-income and middle-income countries from COVID-19. BMJ Glob Health 2020;5:e002844.doi:10.1136/bmjgh-2020-002844pmid:http://www.ncbi.nlm.nih.gov/pubmed/32461228
      OpenUrlFREE Full Text
    15. ↵
      1. Roberton T,
      2. Carter ED,
      3. Chou VB, et al
      . Early estimates of the indirect effects of the COVID-19 pandemic on maternal and child mortality in low-income and middle-income countries: a modelling study. Lancet Glob Health 2020;8:e901–8.doi:10.1016/S2214-109X(20)30229-1pmid:http://www.ncbi.nlm.nih.gov/pubmed/32405459
      OpenUrlPubMed
    16. ↵
      1. Government of Zimbabwe
      . Public health (COVID-19 prevention, containment and treatment) (national Lockdown) order, 2020. Zimbabwe Veritas Zimbabwe; 2020. https://www.veritaszim.net/node/4046
    17. ↵
      1. Mzumara GW,
      2. Chawani M,
      3. Sakala M, et al
      . The health policy response to COVID-19 in Malawi. BMJ Glob Health 2021;6:e006035.doi:10.1136/bmjgh-2021-006035pmid:http://www.ncbi.nlm.nih.gov/pubmed/34006521
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Kaponda C
      . No COVID-19 lockdown still threatens livelihoods and trade in Malawi. London School of Economics 2020 https://blogs.lse.ac.uk/africaatlse/2020/09/25/no-covid19-lockdown-threatens-livelihoods-trade-trust-malawi/
    19. ↵
      1. BBC News
      . Zimbabwe doctors end strike after billionaire’s offer, 2020. Available: https://www.bbc.co.uk/news/world-africa-51205619 [Accessed 04 Aug 2020].
    20. ↵
      1. Reuters
      . Zimbabwe nurses end three-month strike over pay, 2020. Available: https://www.reuters.com/article/us-health-corononavirus-zimbabwe-strike-idUSKBN26011R [Accessed 20 Oct 2020].
    21. ↵
      1. NeoTree
      . NeoTree [GitHub repository], 2020. Available: https://github.com/neotree/neotree [Accessed 03 Oct 2020].
    22. ↵
      1. Crehan C,
      2. Kesler E,
      3. Nambiar B, et al
      . The NeoTree application: developing an integrated mHealth solution to improve quality of newborn care and survival in a district hospital in Malawi. BMJ Glob Health 2019;4:e000860.doi:10.1136/bmjgh-2018-000860pmid:http://www.ncbi.nlm.nih.gov/pubmed/30713745
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Gannon H,
      2. Chimhuya S,
      3. Chimhini G, et al
      . Electronic application to improve management of infections in low-income neonatal units: pilot implementation of the NeoTree beta APP in a public sector hospital in Zimbabwe. BMJ Open Qual 2021;10:e001043.doi:10.1136/bmjoq-2020-001043pmid:http://www.ncbi.nlm.nih.gov/pubmed/33472853
      OpenUrlAbstract/FREE Full Text
    24. ↵
      1. R Foundation for Statistical Computing
      . R: A language and environment for statistical computing [program]. 3.6.3 version. Vienna, Austria R Foundation for Statistical Computing; 2020.
    25. ↵
      1. RStudio
      . RStudio: Integrated Development Environment for R [program]. 1.2.5033 version. Boston, MA RStudio, Inc; 2019.
    26. ↵
      1. Bernal JL,
      2. Cummins S,
      3. Gasparrini A
      . Interrupted time series regression for the evaluation of public health interventions: a tutorial. Int J Epidemiol 2017;46:348–55.doi:10.1093/ije/dyw098pmid:http://www.ncbi.nlm.nih.gov/pubmed/27283160
      OpenUrlCrossRefPubMed
    27. ↵
      1. Stolwijk AM,
      2. Straatman H,
      3. Zielhuis GA
      . Studying seasonality by using sine and cosine functions in regression analysis. J Epidemiol Community Health 1999;53:235–8.doi:10.1136/jech.53.4.235pmid:http://www.ncbi.nlm.nih.gov/pubmed/10396550
      OpenUrlAbstract
    28. ↵
      1. Bikwa Y,
      2. Murewanhema G,
      3. Kanyangarara M, et al
      . Impact of COVID-19 on maternal and perinatal outcomes in Harare, Zimbabwe: a comparative maternal audit. J Glob Health Rep 2021;5.doi:10.29392/001c.28995
    29. ↵
      1. Graham WJ,
      2. Afolabi B,
      3. Benova L, et al
      . Protecting hard-won gains for mothers and newborns in low-income and middle-income countries in the face of COVID-19: call for a service safety net. BMJ Glob Health 2020;5:e002754.doi:10.1136/bmjgh-2020-002754pmid:http://www.ncbi.nlm.nih.gov/pubmed/32499220
      OpenUrlFREE Full Text
    30. ↵
      1. Khalil A,
      2. Kalafat E,
      3. Benlioglu C, et al
      . SARS-CoV-2 infection in pregnancy: a systematic review and meta-analysis of clinical features and pregnancy outcomes. EClinicalMedicine 2020;25:100446.doi:10.1016/j.eclinm.2020.100446pmid:http://www.ncbi.nlm.nih.gov/pubmed/32838230
      OpenUrlPubMed
    31. ↵
      1. Robertson NJ,
      2. Groenendaal F
      . Hypoxic-ischaemic brain injury. In: Rennie JM, ed. Rennie and Roberton’s Textbook of Neonatology. 5th ed. London: Churchill Livingstone, 2012: 1114–55.
    32. ↵
      1. Say L,
      2. Chou D,
      3. Gemmill A, et al
      . Global causes of maternal death: a who systematic analysis. Lancet Glob Health 2014;2:e323–33.doi:10.1016/S2214-109X(14)70227-Xpmid:http://www.ncbi.nlm.nih.gov/pubmed/25103301
      OpenUrlPubMed
    33. ↵
      1. das Neves Martins Pires PH,
      2. Macaringue C,
      3. Abdirazak A, et al
      . Covid-19 pandemic impact on maternal and child health services access in Nampula, Mozambique: a mixed methods research. BMC Health Serv Res 2021;21:860.doi:10.1186/s12913-021-06878-3pmid:http://www.ncbi.nlm.nih.gov/pubmed/34425807
      OpenUrlPubMed
    34. ↵
      1. Hawley S,
      2. Ali MS,
      3. Berencsi K, et al
      . Sample size and power considerations for ordinary least squares interrupted time series analysis: a simulation study. Clin Epidemiol 2019;11:197–205.doi:10.2147/CLEP.S176723pmid:http://www.ncbi.nlm.nih.gov/pubmed/30881136
      OpenUrlPubMed
    35. ↵
      1. Kc A,
      2. Gurung R,
      3. Kinney MV, et al
      . Effect of the COVID-19 pandemic response on intrapartum care, stillbirth, and neonatal mortality outcomes in Nepal: a prospective observational study. Lancet Glob Health 2020;8:e1273–81.doi:10.1016/S2214-109X(20)30345-4pmid:http://www.ncbi.nlm.nih.gov/pubmed/32791117
      OpenUrlPubMed
    36. ↵
      1. Thaddeus S,
      2. Maine D
      . Too far to walk: maternal mortality in context. Soc Sci Med 1994;38:1091–110.doi:10.1016/0277-9536(94)90226-7pmid:http://www.ncbi.nlm.nih.gov/pubmed/8042057
      OpenUrlCrossRefPubMedWeb of Science

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Footnotes

    • SC and SRN are joint first authors.

    • MC and MH are joint last authors

    • Contributors Concept and study design by SC, SRN, GC, FF, MCB, CC, MC and MH with input from other authors. Data collected by HG, DN, TC, CC and TH-B. Analysis performed by SRN and MCB with contributions from FF, SC, EW and MH. Manuscript drafted by SC and SRN with input from GC, FF, MCB, MC and MH. All authors proof-read and approved final draft. Underlying data accessed and verified by SRN, MCB, HG, FF and MH. MH is the guarantor of the study.

    • Funding We would like to thank the funders of this study. SRN was awarded the International Child Health Group David Morley Elective Bursary for this elective project. Funders of the wider Neotree project, past and present, include the Wellcome Trust Digital Innovation Award (215742/Z/19/Z: PI: Heys), RCPCH, Naughton-Cliffe Mathews, UCL Grand Challenges and Global Engagement Fund, and the Healthcare Infection Society (SRG 201802004). FF is supported by the Academy of Medical Sciences and the funders of the Starter Grants for Clinical Lecturers scheme. This study and MH and FF are further supported by the National Institute for Health Research Great Ormond Street Hospital Biomedical Research Centre. The funders had no role in study design, data collection and analysis, or preparation of this report.

    • Disclaimer The views expressed are those of the authors and not necessarily those of the National Health Service (NHS), the NIHR or the UK Department of Health. The funders had no role in study design, data collection, data analysis, data interpretation, or preparation of this manuscript.

    • Competing interests None declared.

    • Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.