Improving maternal, newborn and women’s reproductive health in crisis settings

Cochrane Database of Systematic Reviews, Volume 2015, No. 8, Year 2015

Interpretation

This is the protocol for a review and there is no abstract. The objectives are as follows: To identify, synthesise and evaluate the effects of health system and other interventions aimed at improving maternal, newborn and women’s reproductive health in crisis settings. We will include: We will only include cluster RCTs, non-randomised cluster trials, and CBA studies with at least two intervention sites and two control sites. The review will include studies conducted in any country as disasters and conflicts are not only limited to low- and middle-income countries (LMICs). Although high-income countries’ (HICs) response to RMNH following a crisis may vary from those in LMICs, documenting the nature of such a response may be valuable for LMICs in their efforts to mitigate the impact of these crises. Women of reproductive age and newborns in crisis and post-crisis settings. Due to the already broad scope of this review, we will exclude interventions directed at men and adolescents to improve their sexual and reproductive health. These interventions would be better addressed in separate reviews. The review will identify and synthesise evidence of the effects of health system and population health interventions aimed at improving RMNH in crisis/emergency situations. Potential interventions will meet the following criteria: they should For the purpose of the review, a crisis setting is defined as ‘situations of armed conflict or natural disaster, often involving the displacement of populations, sometimes as refugees, other times as IDPs’ (WHO 2007). We will focus on: We will not consider studies in crisis settings emanating from man-made disasters with limited impacts in terms of population displacement, and morbidity and mortality (such as chemical spills, plane crashes, terrorist attacks, industrial explosions, armoury blasts etc.), unless there is a strong and compelling reason for inclusion. In such situations, we will clearly state the reason for such a decision. While acknowledging the difficulties in classifying a situation as ‘in crisis’ or ‘post-crisis’, we will include all eligible studies undertaken in countries where the state is unable or unwilling to provide basic health care for all the population because of crises arising from armed conflict or disasters. For each crisis, the underlying cause must be linked to a conflict or disaster. RMNH will be defined as above. The review will focus on health system interventions aimed at improving the effectiveness, efficiency and equity in the delivery of clinical and public health services (Lewin 2010) (See Table 1 and Appendix 1 for detailed description). The interventions must be specifically developed to improve women’s newborn, or reproductive health. or a combination of all three, in such settings, and may or may not be part of a bigger package of interventions. The clinical and public health services addressed by the health system intervention(s) could or should be for the purpose of prevention, treatment or rehabilitation, and aimed at improving maternal and newborn health as well as reproductive health of women in its broadest sense as defined by the 1994 International Conference on Population and Development (UNFPA 1995). These could include the delivery of service packages to reduce maternal and neonatal morbidity and mortality, and service packages that enhance/promote the sexual and reproductive wellness/well being of women in crisis/emergency settings (e.g. access to basic reproductive health/family planning services such as contraceptives, ANC) etc. prevention of all forms of sexual violence; prevention and treatment of reproductive tract infections and diseases etc.). We will not include clinical interventions specifically for post-traumatic stress disorders (PTSD), as they constitute a major concern for populations in emergency settings and would be best examined in a separate review. Also, a Cochrane systematic review, ‘Non-specialist health worker interventions for mental health care in low- and middle- income countries’ (van Ginneken 2013) has recently been completed and covers task-shifting for the delivery of mental health care interventions. We will, however, include health system interventions for PTSD linked to maternal and women’s reproductive health, such as interventions for preventing or managing the consequences of sexual violence against women. Furthermore, while acknowledging the broad scope of these issues, we will focus on women’s reproductive health, maternal health (health of the woman during pregnancy, childbirth and the postpartum period), and the health of the newborn during the first 28 days of life. These are some of the most serious concerns in crisis settings that if not promptly addressed may lead to mortality, morbidities and disabilities. We will consider both health system support and health system strengthening interventions. While health system support ‘includes any activity that improves services, from upgrading facilities and equipment to distributing mosquito nets to promote healthy behaviour, improving the health system’s functionality primarily through increasing inputs, for a short term and with a narrow focus, health system strengthening is achieved by more comprehensive changes to policies and regulations, organizational structures, and relationships across the health system building blocks that motivate changes in behavior, and/or allow more effective use of resources to improve multiple health services’ (Chee 2012). Arguably, differentiating between those types of interventions in such contexts might be very difficult as both supporting and strengthening interventions might be packaged and delivered simultaneously. However, in a crisis setting, it is widely expected that initial efforts will be focused on immediate inputs to provide health services (perceived as supporting), while identifying priority areas for strengthening is a long term priority (Chee 2012). We will include and describe studies that meet the inclusion criteria in the characteristics of included studies table, even if they do not report usable results (EPOC 2015). We will consider differential effects across advantaged and disadvantaged populations for all of the outcomes listed above. We will search the following electronic databases : Please see Appendix 2 for the MEDLINE search strategy We will also All searches of all resources will be applied without language restrictions. The definition of some major concepts used throughout the review can be found at Appendix 3. Three authors (CPC, DD and OUJ) will independently read the titles and abstracts of identified studies to eliminate obviously irrelevant studies. For studies identified by any of the review authors that potentially meet our inclusion criteria, the full texts will be retrieved and two review authors (CPC and DD) will assess their final eligibility against the review inclusion criteria. Disagreements will be resolved by discussion and recourse to a third author on the team or the full team if appropriate. We will develop a data extraction form based on the data extraction template provided by the Cochrane EPOC Group. Two reviewers (CPC and OUJ) will independently extract the following details of the eligible studies. We will assess the risk of bias for each included study. Three of the study authors (CPC, OUJ and DD) will independently undertake the ‘Risk of bias’ assessment by using a form with the standard criteria described by the Cochrane EPOC Group (EPOC 2015). We will use the EPOC nine point criteria for RCTs, non-RCTs, and CBA studies and seven point criteria for ITS studies to determine the certainty of all included studies. We will grade each eligible study against each criterion into ‘Low risk’, ‘High risk’ or ‘Unclear risk’. In situations where the information is not reported in the paper, we will contact the study authors for further information. These assessments will be presented in a ‘Risk of bias’ table. For dichotomous outcomes we will use risk ratios (RR),while for continuous outcomes the mean difference (MD) and the standardised mean difference (SMD) will be used. The MD will be used when the outcomes of interest from the included studies are measured in the same manner while the SMD will be used when similar outcomes are measured or assessed using different methods. We will base our analysis on the change in scores before and after the intervention for RCTs and non-RCTs. Where these treatment effects are not directly presented in the papers, we will contact the study author for these data and if we are unable to secure these we will not impute the missing data but will include the study in the review and address the potential impact of the missing data in the Assessment of risk of bias and Discussion sections of the review. For ITS studies we will record the change in level and the change in trend before and after the intervention. A change in level is defined as the difference between the observed level at the first intervention time point and that predicted by the pre-intervention time trend, while a change in trend is defined as the difference between post- and pre-intervention slopes (Ramsay 2003).This is measured as the difference between the fitted value for the first post-intervention data point (one, two, three etc months after the intervention) minus the predicted outcome (one, two, three etc months after the intervention) based on the pre-intervention slope only (EPOC 2015). Our preferred method of obtaining these will be a statistical comparison of time trends before and after the intervention using autoregressive integrated moving average (ARIMA) models, depending on the characteristics of the data, the number of data points available and whether autocorrelation is present (EPOC 2013). An ARIMA model is a form of regression analysis that uses time series data to predict future trends. Additionally, ARIMA models take into account seasonality, cycles, errors and non-stationary aspects of a data set when making forecasts. However, where the use of ARIMA models is inappropriate but the data in the original paper are presented in tables or graphs, with at least three data points before and three data points after the intervention, along with a clearly defined intervention point, we will re-analyse using the segmented time series regression techniques as described in the EPOC guidelines (EPOC 2013). To address the possibility of overestimation or underestimation of the intervention effect in ITS studies, we will avoid solely comparing the means before and after an intervention, without taking into account any secular trends. A secular trend is the smooth long-term direction of a time series; the act of a variable that continues to move in a somewhat consistent way over a long period of time. For CBA studies, the difference-in-difference approach will be used to measure the treatment effects. Here, the treatment effect will be obtained by subtracting the average gain score at baseline and endpoint in the control group from the average gain score at baseline and endpoint in the treatment group. In the case of RCTs, non-RCTs and CBA studies where the pre-intervention or baseline data are not available, we will compare the post-intervention data for the treatment and control groups. For cluster RCTs, non- RCTs and CBA studies we will assess whether an appropriate analysis has been done that adjusts for clustering in calculating measures of precision. If unit-of-analysis errors are identified in the studies we will undertake a re-analysis using an estimate of the intra-cluster correlation co-efficient (ICC) derived from the trial (if possible), from a similar trial or from a study of a similar population. Where this approach is not possible, we will use the inflated standard errors method to address any existing unit-of-analysis issues as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011 ). In the case of ITS studies, analysis will be done using either a regression analysis with time trends and change in levels before and after the intervention or ARIMA analysis, depending on the characteristics of the data, the number of data points available and whether autocorrelation is present (EPOC 2015). For included studies, we will note levels of attrition. We will explore the impact of including studies with high levels of missing data (= 20% for primary outcome) in the overall assessment of treatment effect by using sensitivity analysis. In cases of missing data, we will make efforts to contact the authors concerned. When this is unsuccessful, we will report the data as missing and will not attempt to impute missing values. We will include such studies and undertake an available case analysis. The potential impact of the missing data will be explored in the Assessment of risk of bias and Discussion sections of the review. For all outcomes, we will carry out available case analyses. The denominator for each outcome in each trial will be the number of women or newborns or both randomised minus any women or newborns or both whose outcomes are known to be missing. However, if we are able to secure all the missing data we will undertake the analyses on a full intention-to-treat basis, i.e. we will attempt to include all women or newborns or both randomised to each group in the analyses, and all women or newborns or both will be analysed in the group to which they were allocated, regardless of whether or not they received the allocated intervention. We will asses methodological and clinical heterogeneity. Methodologically, we will explore the effects of different study designs employed (e.g. randomised and non-randomised). We will assess clinical heterogeneity at the following levels as described in Gagnier 2012: We do not anticipate obtaining sufficient studies to allow meta-analysis. As such, we might not be able to undertake a statistical assessment of heterogeneity across the included studies. However, should we find many related studies and to the extent possible, we will use forest plots, the I2 statistic and the Chi2 test to assess heterogeneity (Higgins 2003). We will regard statistical heterogeneity as potentially substantial if I² is greater than 30% and either T² is greater than zero, or there is a low P value (less than 0.10) in the Chi² test for heterogeneity. We will interpret the I2 taking into consideration the magnitude and direction of the treatment effects and the strength of the evidence for heterogeneity. We will undertake assessment of reporting bias according to the criteria described in the Cochrane Handbook for Systematic Reviews of Interventions (Sterne 2011). We will assess the risk of publication bias based on the direction of the findings/results of the included studies, coupled with information obtained from experts in the field of RMNH in crisis settings. For example, if the review identifies only a small number of studies with overwhelming positive results, this would highlight the possibility of publication bias. We will contact experts in the field for their impressions about the findings and also the possibility of unpublished studies with negative or non-significant findings. We will further assess the reporting bias by testing for asymmetry of funnel plots. Tests for funnel plot asymmetry will be done only when there are at least 10 studies included in the meta-analysis. We will interpret the results of tests for funnel plot asymmetry by visual inspection of the funnel plots.To test for funnel plot asymmetry we will use Egger 1997 for continuous outcomes with intervention effects measured as mean difference, and Harbord 2006 for dichotomous outcomes with intervention effects measured as odd ratios (ORs), as suggested in the Cochrane Handbook for Systematic Reviews of Interventions (Sterne 2011). If possible, we will undertake a main analysis involving eligible studies reporting similar outcomes and undertaken in related settings. We will stratify this analysis by the type of crisis, and the category of the intervention. We do not anticipate finding many studies recording similar outcomes in similar settings that will justify merging and calculating an overall effect size and undertaking a meta-analysis. However, if there are two or more studies that evaluate similar interventions and report similar outcomes, we will calculate pooled RRs, MDs or SMDs using a random-effects model. Otherwise, we will report the median and range of effects, if relevant, or measures of effect from individual studies when there are no other studies evaluating a similar intervention and reporting a similar outcome. In the event that we cannot synthesise the data across the studies, we will undertake a structured synthesis of the results. For ITS studies, we do not anticipate finding many identical studies that will necessitate data pooling. However should we find up to 10 identical studies (methodologically homogenous) we will pool the adjusted effect estimates for possible meta-analysis. We will use GRADE to assess the certainty of the evidence for each study outcome (Schunemann 2011a).We will grade the certainty of the evidence for each major study outcome as ‘High’, ‘Moderate’, ‘Low’, or ‘Very Low’. We will present assessments of the certainty of evidence in a ‘Summary of findings’ table (Schunemann 2011b). We expect that there may be variations in the findings of the different studies included in the review due to various explanatory factors. If we identify sufficient studies for the review, we will further investigate this heterogeneity within the following subgroups. We anticipate more favourable outcomes for facility-based care as the probability of having access to skilled care is much higher compared to other structures outside the health facility. Refugees tend to have better health outcomes compared to IDPs possibly due to better international support, as international support for IDPs has been a long-neglected and complex issue within the humanitarian community. Furthermore, we expect that populations receiving healthcare from skilled personnel should have better outcomes compared to when the care is provided by less skilled or unskilled individuals. Health care from a stationary health facility, including some specialised services that provided on a 24-hour basis should be more comprehensive than health care provided by mobile facilities, that may provide only a limited range of services within a specific period. Concerning the nature of the crisis, we expect that those with a slow onset should experience a better response compared to crisis with a rapid onset that will provide little room for an effective response. However, it is possible that crises with a rapid onset may receive better international publicity and support and hence inhabitants may receive better services, compared to inhabitants in areas where the crisis is more gradual in nature. Provided there are sufficient included studies, we will conduct sensitivity analyses to assess how robust the synthesis is in relation to any assumptions that are made with respect to the risk of bias for included studies and how these studies should be grouped (EPOC 2015). Sensitivity analyses will done by excluding: If appropriate, we will also undertake a sensitivity analysis of varying the ICC used for re-analysis of results from clustered designs.

AI Digest

Study Justification:

The study aims to evaluate the effects of health system and other interventions on improving maternal, newborn, and women’s reproductive health in crisis settings. This is important because crises such as armed conflicts and natural disasters can have a significant impact on the health of women and newborns. By identifying and synthesizing evidence on effective interventions, this study can inform policies and programs aimed at mitigating the impact of crises on reproductive health.

Highlights:

– The study will include cluster randomized controlled trials, non-randomized cluster trials, and controlled before-after studies with at least two intervention sites and two control sites.
– It will focus on women of reproductive age and newborns in crisis and post-crisis settings.
– The review will consider health system and population health interventions aimed at improving reproductive, maternal, and newborn health.
– The study will assess the risk of bias for each included study and grade the certainty of evidence for each outcome.
– Sensitivity analyses will be conducted to assess the robustness of the findings.

Recommendations:

Based on the findings of the study, the following recommendations can be made:

1. Increase investment in health system interventions to improve reproductive, maternal, and newborn health in crisis settings.
2. Strengthen coordination and collaboration between humanitarian and health sectors to ensure effective delivery of interventions.
3. Prioritize the provision of skilled care and access to reproductive health services in crisis settings.
4. Enhance support for displaced populations, including refugees and internally displaced persons, to ensure their reproductive health needs are met.
5. Improve data collection and monitoring systems to track the impact of interventions and inform evidence-based decision making.

Key Role Players:

1. Government health departments and ministries
2. International organizations (e.g., WHO, UNFPA)
3. Non-governmental organizations (NGOs) working in humanitarian and health sectors
4. Health professionals and service providers
5. Researchers and academics in the field of reproductive health in crisis settings

Cost Items:

1. Training and capacity building for health professionals and service providers
2. Procurement and distribution of essential reproductive health supplies and equipment
3. Infrastructure development and maintenance of health facilities
4. Data collection and monitoring systems
5. Coordination and collaboration activities between different stakeholders
6. Research and evaluation activities to assess the effectiveness of interventions
Please note that the above recommendations and cost items are based on the information provided in the study protocol and may need to be further refined based on the actual findings of the study.

Strength of Evidence

The strength of evidence for this abstract is 5 out of 10.
The evidence in the abstract is limited as it only provides the protocol for a review and does not include any specific findings or results. To improve the strength of the evidence, the authors should conduct the review and include the results of the included studies in the abstract.

Abstract

We will include: We will only include cluster RCTs, non-randomised cluster trials, and CBA studies with at least two intervention sites and two control sites. The review will include studies conducted in any country as disasters and conflicts are not only limited to low- and middle-income countries (LMICs). Although high-income countries’ (HICs) response to RMNH following a crisis may vary from those in LMICs, documenting the nature of such a response may be valuable for LMICs in their efforts to mitigate the impact of these crises. Women of reproductive age and newborns in crisis and post-crisis settings. Due to the already broad scope of this review, we will exclude interventions directed at men and adolescents to improve their sexual and reproductive health. These interventions would be better addressed in separate reviews. The review will identify and synthesise evidence of the effects of health system and population health interventions aimed at improving RMNH in crisis/emergency situations. Potential interventions will meet the following criteria: they should For the purpose of the review, a crisis setting is defined as ‘situations of armed conflict or natural disaster, often involving the displacement of populations, sometimes as refugees, other times as IDPs’ (WHO 2007). We will focus on: We will not consider studies in crisis settings emanating from man-made disasters with limited impacts in terms of population displacement, and morbidity and mortality (such as chemical spills, plane crashes, terrorist attacks, industrial explosions, armoury blasts etc.), unless there is a strong and compelling reason for inclusion. In such situations, we will clearly state the reason for such a decision. While acknowledging the difficulties in classifying a situation as ‘in crisis’ or ‘post-crisis’, we will include all eligible studies undertaken in countries where the state is unable or unwilling to provide basic health care for all the population because of crises arising from armed conflict or disasters. For each crisis, the underlying cause must be linked to a conflict or disaster. RMNH will be defined as above. The review will focus on health system interventions aimed at improving the effectiveness, efficiency and equity in the delivery of clinical and public health services (Lewin 2010) (See Table 1 and Appendix 1 for detailed description). The interventions must be specifically developed to improve women’s newborn, or reproductive health. or a combination of all three, in such settings, and may or may not be part of a bigger package of interventions. The clinical and public health services addressed by the health system intervention(s) could or should be for the purpose of prevention, treatment or rehabilitation, and aimed at improving maternal and newborn health as well as reproductive health of women in its broadest sense as defined by the 1994 International Conference on Population and Development (UNFPA 1995). These could include the delivery of service packages to reduce maternal and neonatal morbidity and mortality, and service packages that enhance/promote the sexual and reproductive wellness/well being of women in crisis/emergency settings (e.g. access to basic reproductive health/family planning services such as contraceptives, ANC) etc. prevention of all forms of sexual violence; prevention and treatment of reproductive tract infections and diseases etc.). We will not include clinical interventions specifically for post-traumatic stress disorders (PTSD), as they constitute a major concern for populations in emergency settings and would be best examined in a separate review. Also, a Cochrane systematic review, ‘Non-specialist health worker interventions for mental health care in low- and middle- income countries’ (van Ginneken 2013) has recently been completed and covers task-shifting for the delivery of mental health care interventions. We will, however, include health system interventions for PTSD linked to maternal and women’s reproductive health, such as interventions for preventing or managing the consequences of sexual violence against women. Furthermore, while acknowledging the broad scope of these issues, we will focus on women’s reproductive health, maternal health (health of the woman during pregnancy, childbirth and the postpartum period), and the health of the newborn during the first 28 days of life. These are some of the most serious concerns in crisis settings that if not promptly addressed may lead to mortality, morbidities and disabilities. We will consider both health system support and health system strengthening interventions. While health system support ‘includes any activity that improves services, from upgrading facilities and equipment to distributing mosquito nets to promote healthy behaviour, improving the health system’s functionality primarily through increasing inputs, for a short term and with a narrow focus, health system strengthening is achieved by more comprehensive changes to policies and regulations, organizational structures, and relationships across the health system building blocks that motivate changes in behavior, and/or allow more effective use of resources to improve multiple health services’ (Chee 2012). Arguably, differentiating between those types of interventions in such contexts might be very difficult as both supporting and strengthening interventions might be packaged and delivered simultaneously. However, in a crisis setting, it is widely expected that initial efforts will be focused on immediate inputs to provide health services (perceived as supporting), while identifying priority areas for strengthening is a long term priority (Chee 2012). We will include and describe studies that meet the inclusion criteria in the characteristics of included studies table, even if they do not report usable results (EPOC 2015). We will consider differential effects across advantaged and disadvantaged populations for all of the outcomes listed above. We will search the following electronic databases : Please see Appendix 2 for the MEDLINE search strategy We will also All searches of all resources will be applied without language restrictions. The definition of some major concepts used throughout the review can be found at Appendix 3. Three authors (CPC, DD and OUJ) will independently read the titles and abstracts of identified studies to eliminate obviously irrelevant studies. For studies identified by any of the review authors that potentially meet our inclusion criteria, the full texts will be retrieved and two review authors (CPC and DD) will assess their final eligibility against the review inclusion criteria. Disagreements will be resolved by discussion and recourse to a third author on the team or the full team if appropriate. We will develop a data extraction form based on the data extraction template provided by the Cochrane EPOC Group. Two reviewers (CPC and OUJ) will independently extract the following details of the eligible studies. We will assess the risk of bias for each included study. Three of the study authors (CPC, OUJ and DD) will independently undertake the ‘Risk of bias’ assessment by using a form with the standard criteria described by the Cochrane EPOC Group (EPOC 2015). We will use the EPOC nine point criteria for RCTs, non-RCTs, and CBA studies and seven point criteria for ITS studies to determine the certainty of all included studies. We will grade each eligible study against each criterion into ‘Low risk’, ‘High risk’ or ‘Unclear risk’. In situations where the information is not reported in the paper, we will contact the study authors for further information. These assessments will be presented in a ‘Risk of bias’ table. For dichotomous outcomes we will use risk ratios (RR),while for continuous outcomes the mean difference (MD) and the standardised mean difference (SMD) will be used. The MD will be used when the outcomes of interest from the included studies are measured in the same manner while the SMD will be used when similar outcomes are measured or assessed using different methods. We will base our analysis on the change in scores before and after the intervention for RCTs and non-RCTs. Where these treatment effects are not directly presented in the papers, we will contact the study author for these data and if we are unable to secure these we will not impute the missing data but will include the study in the review and address the potential impact of the missing data in the Assessment of risk of bias and Discussion sections of the review. For ITS studies we will record the change in level and the change in trend before and after the intervention. A change in level is defined as the difference between the observed level at the first intervention time point and that predicted by the pre-intervention time trend, while a change in trend is defined as the difference between post- and pre-intervention slopes (Ramsay 2003).This is measured as the difference between the fitted value for the first post-intervention data point (one, two, three etc months after the intervention) minus the predicted outcome (one, two, three etc months after the intervention) based on the pre-intervention slope only (EPOC 2015). Our preferred method of obtaining these will be a statistical comparison of time trends before and after the intervention using autoregressive integrated moving average (ARIMA) models, depending on the characteristics of the data, the number of data points available and whether autocorrelation is present (EPOC 2013). An ARIMA model is a form of regression analysis that uses time series data to predict future trends. Additionally, ARIMA models take into account seasonality, cycles, errors and non-stationary aspects of a data set when making forecasts. However, where the use of ARIMA models is inappropriate but the data in the original paper are presented in tables or graphs, with at least three data points before and three data points after the intervention, along with a clearly defined intervention point, we will re-analyse using the segmented time series regression techniques as described in the EPOC guidelines (EPOC 2013). To address the possibility of overestimation or underestimation of the intervention effect in ITS studies, we will avoid solely comparing the means before and after an intervention, without taking into account any secular trends. A secular trend is the smooth long-term direction of a time series; the act of a variable that continues to move in a somewhat consistent way over a long period of time. For CBA studies, the difference-in-difference approach will be used to measure the treatment effects. Here, the treatment effect will be obtained by subtracting the average gain score at baseline and endpoint in the control group from the average gain score at baseline and endpoint in the treatment group. In the case of RCTs, non-RCTs and CBA studies where the pre-intervention or baseline data are not available, we will compare the post-intervention data for the treatment and control groups. For cluster RCTs, non- RCTs and CBA studies we will assess whether an appropriate analysis has been done that adjusts for clustering in calculating measures of precision. If unit-of-analysis errors are identified in the studies we will undertake a re-analysis using an estimate of the intra-cluster correlation co-efficient (ICC) derived from the trial (if possible), from a similar trial or from a study of a similar population. Where this approach is not possible, we will use the inflated standard errors method to address any existing unit-of-analysis issues as described in the Cochrane Handbook for Systematic Reviews of Interventions (Higgins 2011 ). In the case of ITS studies, analysis will be done using either a regression analysis with time trends and change in levels before and after the intervention or ARIMA analysis, depending on the characteristics of the data, the number of data points available and whether autocorrelation is present (EPOC 2015). For included studies, we will note levels of attrition. We will explore the impact of including studies with high levels of missing data (= 20% for primary outcome) in the overall assessment of treatment effect by using sensitivity analysis. In cases of missing data, we will make efforts to contact the authors concerned. When this is unsuccessful, we will report the data as missing and will not attempt to impute missing values. We will include such studies and undertake an available case analysis. The potential impact of the missing data will be explored in the Assessment of risk of bias and Discussion sections of the review. For all outcomes, we will carry out available case analyses. The denominator for each outcome in each trial will be the number of women or newborns or both randomised minus any women or newborns or both whose outcomes are known to be missing. However, if we are able to secure all the missing data we will undertake the analyses on a full intention-to-treat basis, i.e. we will attempt to include all women or newborns or both randomised to each group in the analyses, and all women or newborns or both will be analysed in the group to which they were allocated, regardless of whether or not they received the allocated intervention. We will asses methodological and clinical heterogeneity. Methodologically, we will explore the effects of different study designs employed (e.g. randomised and non-randomised). We will assess clinical heterogeneity at the following levels as described in Gagnier 2012: We do not anticipate obtaining sufficient studies to allow meta-analysis. As such, we might not be able to undertake a statistical assessment of heterogeneity across the included studies. However, should we find many related studies and to the extent possible, we will use forest plots, the I2 statistic and the Chi2 test to assess heterogeneity (Higgins 2003). We will regard statistical heterogeneity as potentially substantial if I² is greater than 30% and either T² is greater than zero, or there is a low P value (less than 0.10) in the Chi² test for heterogeneity. We will interpret the I2 taking into consideration the magnitude and direction of the treatment effects and the strength of the evidence for heterogeneity. We will undertake assessment of reporting bias according to the criteria described in the Cochrane Handbook for Systematic Reviews of Interventions (Sterne 2011). We will assess the risk of publication bias based on the direction of the findings/results of the included studies, coupled with information obtained from experts in the field of RMNH in crisis settings. For example, if the review identifies only a small number of studies with overwhelming positive results, this would highlight the possibility of publication bias. We will contact experts in the field for their impressions about the findings and also the possibility of unpublished studies with negative or non-significant findings. We will further assess the reporting bias by testing for asymmetry of funnel plots. Tests for funnel plot asymmetry will be done only when there are at least 10 studies included in the meta-analysis. We will interpret the results of tests for funnel plot asymmetry by visual inspection of the funnel plots.To test for funnel plot asymmetry we will use Egger 1997 for continuous outcomes with intervention effects measured as mean difference, and Harbord 2006 for dichotomous outcomes with intervention effects measured as odd ratios (ORs), as suggested in the Cochrane Handbook for Systematic Reviews of Interventions (Sterne 2011). If possible, we will undertake a main analysis involving eligible studies reporting similar outcomes and undertaken in related settings. We will stratify this analysis by the type of crisis, and the category of the intervention. We do not anticipate finding many studies recording similar outcomes in similar settings that will justify merging and calculating an overall effect size and undertaking a meta-analysis. However, if there are two or more studies that evaluate similar interventions and report similar outcomes, we will calculate pooled RRs, MDs or SMDs using a random-effects model. Otherwise, we will report the median and range of effects, if relevant, or measures of effect from individual studies when there are no other studies evaluating a similar intervention and reporting a similar outcome. In the event that we cannot synthesise the data across the studies, we will undertake a structured synthesis of the results. For ITS studies, we do not anticipate finding many identical studies that will necessitate data pooling. However should we find up to 10 identical studies (methodologically homogenous) we will pool the adjusted effect estimates for possible meta-analysis. We will use GRADE to assess the certainty of the evidence for each study outcome (Schunemann 2011a).We will grade the certainty of the evidence for each major study outcome as ‘High’, ‘Moderate’, ‘Low’, or ‘Very Low’. We will present assessments of the certainty of evidence in a ‘Summary of findings’ table (Schunemann 2011b). We expect that there may be variations in the findings of the different studies included in the review due to various explanatory factors. If we identify sufficient studies for the review, we will further investigate this heterogeneity within the following subgroups. We anticipate more favourable outcomes for facility-based care as the probability of having access to skilled care is much higher compared to other structures outside the health facility. Refugees tend to have better health outcomes compared to IDPs possibly due to better international support, as international support for IDPs has been a long-neglected and complex issue within the humanitarian community. Furthermore, we expect that populations receiving healthcare from skilled personnel should have better outcomes compared to when the care is provided by less skilled or unskilled individuals. Health care from a stationary health facility, including some specialised services that provided on a 24-hour basis should be more comprehensive than health care provided by mobile facilities, that may provide only a limited range of services within a specific period. Concerning the nature of the crisis, we expect that those with a slow onset should experience a better response compared to crisis with a rapid onset that will provide little room for an effective response. However, it is possible that crises with a rapid onset may receive better international publicity and support and hence inhabitants may receive better services, compared to inhabitants in areas where the crisis is more gradual in nature. Provided there are sufficient included studies, we will conduct sensitivity analyses to assess how robust the synthesis is in relation to any assumptions that are made with respect to the risk of bias for included studies and how these studies should be grouped (EPOC 2015). Sensitivity analyses will done by excluding: If appropriate, we will also undertake a sensitivity analysis of varying the ICC used for re-analysis of results from clustered designs.

Materials

N/A

Innovations

Based on the information provided, it seems that the document you shared is a protocol for a review on improving maternal, newborn, and women’s reproductive health in crisis settings. The document outlines the objectives, inclusion criteria, search strategy, data extraction process, and analysis methods for the review.

However, it does not provide specific innovations or recommendations for improving access to maternal health in crisis settings. To identify innovations, it would be necessary to review the findings and conclusions of the actual study or review once it is completed.

AI Innovations Description

Based on the provided information, the recommendation to improve access to maternal health in crisis settings is to conduct a systematic review of health system and other interventions aimed at improving maternal, newborn, and women’s reproductive health. The review should include cluster randomized controlled trials (RCTs), non-randomized cluster trials, and controlled before-after studies with at least two intervention sites and two control sites. The review should encompass studies conducted in any country, as crises are not limited to low- and middle-income countries. The focus should be on women of reproductive age and newborns in crisis and post-crisis settings.

The interventions to be considered should be specifically developed to improve women’s reproductive health, maternal health, or a combination of both. They may include the delivery of service packages to reduce maternal and neonatal morbidity and mortality, as well as service packages that enhance sexual and reproductive wellness in crisis/emergency settings. The review should also consider health system support and health system strengthening interventions.

The review should assess the effects of the interventions on various outcomes, such as maternal and newborn health, reproductive health, and the effectiveness, efficiency, and equity in the delivery of clinical and public health services. The risk of bias in the included studies should be assessed, and data extraction and analysis should be conducted using appropriate methods. The certainty of the evidence should be evaluated using the GRADE approach.

In addition, the review should explore potential heterogeneity across studies and assess reporting bias. Sensitivity analyses should be conducted to test the robustness of the findings. If feasible, subgroup analyses can be performed to investigate variations in outcomes based on factors such as the type of crisis, the category of intervention, and the nature of the healthcare facility.

Overall, conducting a comprehensive systematic review of interventions to improve maternal health in crisis settings will provide valuable evidence to guide policy and practice in this important area.

AI Innovations Methodology

The protocol you provided is for a review on improving maternal, newborn, and women’s reproductive health in crisis settings. The objective of the review is to identify, synthesize, and evaluate the effects of health system and other interventions aimed at improving maternal, newborn, and women’s reproductive health in crisis settings.

To simulate the impact of the recommendations on improving access to maternal health, a methodology is described in the protocol. The methodology includes the following steps:

1. Inclusion criteria: The review will include cluster randomized controlled trials (RCTs), non-randomized cluster trials, and controlled before-after studies with at least two intervention sites and two control sites. The studies should focus on women of reproductive age and newborns in crisis and post-crisis settings.

2. Data extraction: A data extraction form will be developed based on the Cochrane Effective Practice and Organization of Care (EPOC) Group’s template. Two reviewers will independently extract relevant details from the eligible studies.

3. Risk of bias assessment: The risk of bias for each included study will be assessed using standard criteria described by the Cochrane EPOC Group. The certainty of the included studies will be determined using the EPOC nine-point criteria for RCTs, non-RCTs, and controlled before-after studies.

4. Analysis: For dichotomous outcomes, risk ratios (RR) will be used, while for continuous outcomes, mean difference (MD) and standardized mean difference (SMD) will be used. The analysis will be based on the change in scores before and after the intervention for RCTs and non-RCTs. For controlled before-after studies, the change in level and trend before and after the intervention will be recorded.

5. Heterogeneity assessment: Methodological and clinical heterogeneity will be assessed. Statistical heterogeneity will be assessed using forest plots, the I2 statistic, and the Chi2 test. Reporting bias will also be assessed.

6. Sensitivity analysis: Sensitivity analyses will be conducted to assess the robustness of the synthesis in relation to assumptions made about the risk of bias and grouping of studies.

7. Summary of findings: The certainty of the evidence for each major study outcome will be assessed using the GRADE approach. A summary of findings table will be presented.

8. Subgroup analysis: Subgroup analyses will be conducted to investigate heterogeneity within different subgroups, such as facility-based care, refugees vs. internally displaced persons (IDPs), and type of crisis.

9. Sensitivity analysis: Sensitivity analyses will be conducted to assess the impact of excluding certain studies or varying the intra-cluster correlation coefficient (ICC) used for re-analysis of clustered designs.

Overall, this methodology aims to systematically evaluate the effects of health system and other interventions on improving access to maternal health in crisis settings. By following this methodology, the impact of the recommendations can be simulated and assessed.

Authors & Co-Authors

Statistics:

Citations: 8

Authors: 6

Identifiers:

DOI: 10.1002/14651858.CD011829

Research Areas:

Community Interventions, Disability, Disparities, Genetics and Genomics, Health System and Policy, Maternal and Child Health, Mental Health, Quality of Care, Sexual and Reproductive Health, Social Determinants, Violence and Injury

Study Design:

Cross Sectional Study, Quasi Experimental Study, randomised-control-trial

Study Approach:

Mixed-methods, Quantitative, Systematic Review

Participants Gender:

Male & Female