final mozambique race endline survey report 31march17 ... · race endline survey final report v...
TRANSCRIPT
ENDLINE SURVEY
FINAL REPORT
SAVE THE CHILDREN, MOZAMBIQUE
Prepared by ICF and Save the Children for WHO Rapid Access Expansion (RAcE) Program
March 2017
AUTHORS: Meghan Swor, Debra Prosnitz, Kirsten Zalisk, Tanya Gunther, Eleanor Hill, Jeanne Koepsell
RAcEEndlineSurveyFinalReport i
ACKNOWLEDGEMENTS
ICF and Save the Children would like to thank the Instituto Nacional de Saúde and the Mozambique Ministry of Health for their contributions to this work. We would also like to thank the Agente Polivalente Elementares (Mozambique’s community health workers) who work hard to provide services to caregivers and children in communities, and to the caregivers who give so much to ensure and improve the health of their children. This work was made possible by the World Health Organization through funding by the Canadian Government.
RAcEEndlineSurveyFinalReport ii
TABLE OF CONTENTS
ABBREVIATIONS ............................................................................................................................................................. iii
EXECUTIVE SUMMARY ................................................................................................................................................. iv
1 BACKGROUND ...................................................................................................................................................... 1 1.1 RAcE Program Goals and Objectives ....................................................................................................... 1 1.2 RAcE Mozambique Project Background .................................................................................................. 1 1.3 RAcE Mozambique Endline Survey Objectives ...................................................................................... 2
2 SURVEY METHODS ................................................................................................................................................ 3 2.1 Survey Implementation and Partnership .................................................................................................. 3 2.2 Survey Design ................................................................................................................................................. 3 2.3 Survey Questionnaire ................................................................................................................................... 5 2.4 Selection and Training of Survey Staff ...................................................................................................... 6 2.5 Data Collection ............................................................................................................................................. 6 2.6 Data Entry and Management ...................................................................................................................... 7 2.7 Data Analysis .................................................................................................................................................. 8 2.8 Survey Indicators ........................................................................................................................................... 8 2.9 Survey Limitations ......................................................................................................................................... 8
3 FINDINGS .................................................................................................................................................................. 9 3.1 Characteristics of Sick Children and Caregivers ................................................................................... 9 3.2 Decision-Making .......................................................................................................................................... 11 3.3 Caregiver Knowledge and Perception of APEs .................................................................................... 11 3.4 Care-Seeking ................................................................................................................................................ 12 3.5 Assessment ................................................................................................................................................... 14 3.6 Treatment Coverage .................................................................................................................................. 15 3.7 First Dose of Treatment and Counseling from APE .......................................................................... 17 3.8 Referral Adherence .................................................................................................................................... 18 3.9 Sick Child Follow-Up .................................................................................................................................. 18 3.10 Illness Management and Diagnostics by Sex ......................................................................................... 19
4 DISCUSSION .......................................................................................................................................................... 21 4.1 Implications ................................................................................................................................................... 23
Annex A. Survey Report by INS .................................................................................................................................. 24
Annex B. Endline Sample ............................................................................................................................................... 26
Annex C. Detailed Sampling Design ............................................................................................................................ 27
Annex D. Endline Survey Questionnaire ................................................................................................................... 28
Annex E. Endline Survey Training Schedule .............................................................................................................. 29
Annex F. Endline Survey Fieldwork Schedule ........................................................................................................... 34
Annex G. Details of Data Cleaning and Analysis ..................................................................................................... 37
Annex H. Key Indicators for Full Project Area ........................................................................................................ 39
Annex I. Summary of Findings from RAcE Mozambique Endline APE Survey ................................................... 42
RAcEEndlineSurveyFinalReport iii
ABBREVIATIONS
ACT artemisinin-based combination therapy
APE Agente Polivalente Elementar (community health worker)
CCM community case management
iCCM integrated community case management
INS Instituto Nacional de Saúde (National Institute of Health)
M&E Monitoring and evaluation
MISAU Ministério da Saúde (Ministry of Health)
MOH Ministry of Health
mRDT malaria rapid diagnostic test
ORS oral rehydration solution
PPS probability proportional to size
RDT rapid diagnostic test
RAcE Rapid Access Expansion
WHO World Health Organization
RAcEEndlineSurveyFinalReport iv
EXECUTIVE SUMMARY
Save the Children, in partnership with Malaria Consortium, implemented the Rapid Access Expansion (RAcE) program in four provinces in Mozambique—Inhambane, Manica, Nampula, and Zambezia—from 2013 to 2016. In October 2016 the Instituto Nacional de Saúde (INS, or National Institute of Health) conducted the RAcE endline survey, with technical assistance from ICF and logistical, administrative, and financial support from Save the Children. We compared baseline and endline data to assess changes in sick child care-seeking, assessment, and treatment coverage, as well as caregivers’ knowledge of childhood illnesses and perceptions of Agente Polivalente Elementar (APE, or community health worker) services and used the information to make inferences about project accomplishments.
The percentage of caregivers who know that an APE works in their community increased significantly from baseline (62 percent) to endline (93 percent) (p<0.001). Care-seeking from APEs increased significantly between baseline and endline, even though the overall level of care-seeking was high at baseline and remained so at endline. The percentage of children age 2-59 months who were sick in the two weeks preceding the survey and taken to a community case management (CCM)-trained APE as a first source of care overall more than doubled, from 23 percent at baseline to 57 percent at endline (p<0.001).
The proportion of sick children receiving appropriate treatment for integrated community case management (iCCM) illnesses did not change over the course of the project. The proportion of sick children receiving appropriate treatment for an iCCM illness from an APE, however, increased significantly, from 12 percent at baseline to 29 percent at endline (p<0.001). At endline, appropriate treatment by APEs accounted for more than half of all reported appropriate treatment [58.4 percent (28.8 percent of cases of illness received appropriate treatment from an APE/49.3 percent of cases of illness received appropriate treatment, from any source)] compared to about a quarter at baseline [26.0% (11.9%/45.7%)].
Appropriate treatment for diarrhea—treatment with both oral rehydration solution (ORS) and zinc—increased significantly, from 8 percent at baseline to 31 percent at endline. Among those cases that sought care from an APE, there was a nearly five-fold increase in treatment of diarrhea with both ORS and zinc by an APE, from 2 percent at baseline to 23 percent at endline, and treatment of cough with difficult or fast breathing with amoxicillin by an APE nearly doubled, from 17 percent at baseline to 33 percent at endline. There was no significant change in appropriate malaria treatment.
With the RAcE project, the supply chain management shifted from being led by Save the Children to being led by the Ministry of Health, resulting in greater challenges maintaining supply levels. The lack of a measured increase for appropriate fever treatment could be in part due to consistent and widespread stockouts of rapid diagnostic tests and artemisinin-based combination therapy (ACT), both of which need to be in stock for APEs to delivery appropriate treatment. Only 78 percent of APEs had a functional timer to use in counting a child’s respiratory rate at the time of the endline survey. Stock levels of both ORS and zinc were quite high at the time of the endline.
RAcEEndlineSurveyFinalReport v
Table 1. Changes in key indicators and end-of-project indicator estimates
Indicators highlighted in green had a statistically significant change from baseline to endline, determined by a p-value of less than 0.05 and no overlap in confidence intervals. Those determined to have statistically significant change over time based on p-values less than 0.05 but have overlapping confidence intervals are noted by red highlight.
Indicator Baseline Endline % Point
Change p-value
% (CI %) % (CI %)Caregiver Knowledge
1
Percentage of caregivers of children age 2–59 months who have been sick in the two weeks preceding the survey who are aware of the presence of the CCM-trained APE in their community
62.0 (45.4 – 76.2)
93.4 (83.3 – 97.6)
31.4% 0.0007
2
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who know the role of the CCM-trained APE in their community
49.2 (36.8 – 61.8)
69.1 (59.4 – 77.5)
19.9% 0.0142
3
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who know two or more signs of childhood illness that require immediate assessment by an appropriately trained provider
86.5 (81.8 – 90.2)
92.9 (87.3 – 96.1)
6.4% 0.0474
Caregiver perceptions of iCCM services
4
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who view CCM-trained APEs as trusted health care providers
82.9 (75.3 – 88.5)
78.2 (70.5 – 84.4)
-4.7% 0.3070
5
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who believe CCM-trained APEs provide quality services
74.8 (63.7 – 83.3)
76.7 (70.3 – 82.1)
1.9% 0.7217
6
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who found the CCM-trained APE at first visit
N/A 82.2
(74.0 – 88.3) N/A N/A
7
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who cite the CCM-trained APE as a convenient source of treatment
80.4 (70.2 – 87.7)
81.9 (72.4 – 88.6)
1.5% 0.8254
RAcEEndlineSurveyFinalReport vi
Indicator Baseline Endline % Point
Change p-value
% (CI %) % (CI %)Sick Child Care-seeking
8
Percentage of children age 2-59 months who have been sick in the two weeks preceding the survey for whom advice or treatment was sought from an appropriate provider
Overall 79.1
(73.3 – 84.0) 79.5
(73.2 – 84.7) 0.4% 0.8991
Malaria (ACT)** 84.1
(77.6 – 89.0) 77.5
(70.5 – 83.3) -6.6% 0.0982
Diarrhea (ORS and zinc) 74.2
(65.9 – 81.1) 78.1
(69.9 – 84.6) 3.9% 0.3570
Cough with difficult or fast breathing (amoxicillin)
79.1 (72.3 – 84.6)
83.1 (76.3 – 88.3)
4.0% 0.3452
9
Percentage of children age 2-59 months who were sick in two weeks preceding the survey taken to a CCM-trained APE as first source of care
Overall 23.1
(15.1 – 33.6) 57.0
(48.5 – 65.1) 33.9% 0.0000
Fever 24.4
(15.5 – 36.1) 54.1
(44.6 – 63.2) 29.7% 0.0003
Diarrhea 22.9
(14.5 – 34.2) 60.8
(51.5 – 69.3) 37.9% 0.0000
Cough with difficult or fast breathing (amoxicillin)
21.7 (14.0 – 32.2)
56.1 (47.4 – 64.4)
34.4% 0.0001
Sick Child Assessment
10
Percentage of children age 2-59 months with fever in the two weeks preceding the survey who had finger or heel stick
43.9 (35.4 – 52.8)
51.1 (43.9 – 58.3)
7.2% 0.2490
11
Percentage of children age 2-59 months for whom their caregiver received the results of the malaria diagnostic test of the children who had had finger or heel stick in the two weeks preceding the survey
84.9 (75.0 – 91.3)
93.6 (87.2 – 96.9)
8.7% 0.0603
12
Percentage of children age 2-59 months with cough and difficult or rapid breathing (suspected pneumonia) in the two weeks preceding the survey who had their respiratory rate counted to assess fast breathing
32.6 (25.7 – 40.4)
39.2 (29.6 – 49.7)
6.6% 0.2398
Sick Child Assessment by CHW
13
Percentage of children age 2-59 months with fever in the two weeks preceding the survey who had a finger or heel stick by an APE (among those who sought care from an APE)
19.1 (10.6 – 32.0)
51.2 (38.2 – 64.0)
32.1% 0.0008
RAcEEndlineSurveyFinalReport vii
Indicator Baseline Endline % Point
Change p-value
% (CI %) % (CI %)
14
Percentage of children age 2-59 months for whom their caregiver received the results of the malaria diagnostic test of the children who had a finger or heel stick by an APE in the two weeks preceding the survey (among those who sought care from an APE)
100 96.6
(89.6 – 98.9) -3.4% 0.3907
15
Percentage of children age 2-59 months with cough and difficult or rapid breathing (suspected pneumonia) in the two weeks preceding the survey who had their respiratory rate counted to assess fast breathing by an APE (among those who sought care from an APE)
23.2 (12.2 – 39.8)
39.0 (27.1 – 52.3)
15.8% 0.0915
Sick Child Treatment
16
Percentage of children age 2-59 months who have been sick in two weeks preceding the survey who received appropriate treatment
Overall 45.7
(40.7 – 50.7) 49.3
(42.1 – 56.5) 3.6% 0.4168
Fever (Confirmed malaria – ACT same or next day with positive
blood test)
77.2 (66.5 – 85.2)
70.5 (59.4 – 79.5)
-6.7% 0.3652
Diarrhea (ORS and zinc) 8.1
(4.3 – 14.7) 31.1
(21.8 – 42.1) 23.0% 0.0001
Cough with difficult or fast breathing (Amoxicillin)
69.6 (61.1 – 76.9)
58.8 (50.9 – 66.2)
-10.8% 0.0558
17
Percentage of children age 2-59 months who have been sick in the two weeks preceding the survey who received appropriate treatment from a CCM-trained APE
Overall 11.9
(7.4 – 18.5) 23.7
(18.5 – 29.7) 11.8% 0.0003
Fever (Confirmed malaria – ACT same or next day with positive
blood test)
23.5 (12.9 – 38.9)
34.2 (23.5 – 46.7)
10.7% 0.1736
Diarrhea (ORS and zinc) 2.2
(0.7 – 7.1) 22.6
(15.0 – 32.4) 20.4% 0.0000
Cough with difficult or fast breathing (Amoxicillin)
17.0 (10.2 – 27.0)
33.1 (26.0 – 41.1)
16.1% 0.0164
18
Percentage of children age 2-59 months who have been sick in the two weeks preceding the survey who received the first dose of treatment in the presence of an APE among those who received prescription medicines for a CCM condition in the two weeks preceding the survey
Overall 41.3
(29.7 – 53.9) 33.7
(24.5 – 44.5) -7.6% 0.3549
Fever 56.3
(36.8 – 73.9) 26.8
(16.5 – 40.4) -29.5% 0.0289
RAcEEndlineSurveyFinalReport viii
Indicator Baseline Endline % Point
Change p-value
% (CI %) % (CI %)
Diarrhea 0.0 17.4
(7.0 – 37.1) 17.4% 0.4418
Cough with difficult or fast breathing (Amoxicillin)
28.2 (21.0 – 36.7)
51.0 (37.4 – 64.5)
22.8% 0.0014
19
Percentage of sick children age 2-59 months for whom their caregivers received counseling on how to provide the treatment(s) among those who received prescription medicines for a CCM condition in the two weeks preceding the survey
Overall 96.7
(90.8 – 98.9) 96.0
(92.0 – 98.0) -0.7% 0.7481
Fever 95.8
(85.2 – 98.9) 98.8
(91.2 – 99.8) 3.0% 0.2781
Diarrhea 80.0
(22.8 – 98.2) 95.7
(86.4 – 98.7) 15.7% 0.1740
Cough with difficult or fast breathing
100 93.9
(85.6 – 97.5) -6.1% 0.1767
Sick Child Referral and Follow-up
20
Percentage of sick children age 2-59 who were referred in the two weeks preceding the survey whose caregiver adhered to referral advice
68.8 (54.2 – 80.3)
65.4 (55.4 – 74.2)
-3.4% 0.6689
21
Percentage of sick children age 2-59 months receiving treatment from an APE in the two weeks preceding the survey who received a follow-up visit from an APE
40.9 (25.0 – 58.9)
41.2 (32.1 – 50.9)
0.3% 0.9753
22
Percentage of sick children age 2-59 months receiving treatment from an APE in the two weeks preceding the survey whose caregiver followed-up with an APE
10.2 (4.1 – 23.1)
10.3 (6.1 – 16.9)
0.1% 0.9826
* Indicates that a figure is based on fewer than 10 cases ** Numerator: Cases of fever with a positive blood test who received ACT within 24 hours; Denominator: all cases of fever
RAcEMozambiqueEndlineSurveyFinalReport 1
1 BACKGROUND
1.1 RAcE Program Goals and Objectives
In 2013, the World Health Organization (WHO) launched the Rapid Access Expansion (RAcE) program in five sub-Saharan African countries—Democratic Republic of Congo, Malawi, Mozambique, Niger, and Nigeria. The goal of the program was to increase coverage of diagnostic, treatment, and referral services for malaria, pneumonia, and diarrhea to decrease overall mortality and the number of severe cases among children aged 2–59 months. The program would accomplish this goal through the following objectives:
Catalyze the scale-up of integrated community case management (iCCM) as an integral part of government-provided health services in sub-Saharan Africa.
Stimulate policy review and regulatory update in each country on disease case management.
Accelerate adaptation of supply management and surveillance systems to include services at the community level.
This effort came at a time when there was great momentum for iCCM at the country level and a high degree of interest among the global health community to understand how to best measure success and how to build country ownership and capacity to sustain iCCM interventions.
1.2 RAcE Mozambique Project Background
Save the Children, in partnership with Malaria Consortium, implemented the RAcE program in four provinces in Mozambique—Inhambane, Manica, Nampula, and Zambezia—from 2013 to 2016. The goal of the program was to reduce under-five mortality and morbidity by increasing access to iCCM services, improving the quality of iCCM services, strengthening links between iCCM services and communities, and advocating with the Ministry of Health (MOH) and partners to strengthen the national system to support, sustain, and scale up iCCM.
In Mozambique, iCCM is part of the Ministério da Saúde (MISAU, or Ministry of Health) Agente Polivalente Elementar (APE, or community health worker) program. The APE program started in 1978 and, after a period of decline, was revitalized starting in 2010. In 2014, the APE program became a department in the National Directorate of Public Health, which oversees planning, coordination, and monitoring of interventions. APEs, who live in the communities they service, provide iCCM services as well as conduct health promotion activities, preventive home visits, and malnutrition screening. The communities have populations between 500–2,000 residents or more and are located between 8 and 25 km from the nearest health facility.
The RAcE program began implementation in the four provinces in April 2013. Save the Children was the lead implementing partner, working closely with MISAU in Manica, Nampula, and Zambezia. In Inhambane, Malaria Consortium led the implementation, in collaboration with MISAU. As of September 2016, a total of 1,339 APEs were providing iCCM services in the four project provinces, covering a total population of approximately 4,196,074, including an estimated 719,444 children under five years of age.
RAcEMozambiqueEndlineSurveyFinalReport 2
The national program is expanding with the ongoing training of new APEs and the roll out of an additional package of interventions, introduced in October 2015, which includes family planning, distribution of vitamin A capsules for children aged 6–59 months, provision of Misoprostol for postpartum hemorrhage, and support for HIV/AIDS and tuberculosis treatment adherence.
The baseline household survey was conducted mid-November through mid-December 2013 by Save the Children, in collaboration with the Instituto Nacional de Saude (INS, or National Institute of Health) and with technical support from ICF.
1.3 RAcE Mozambique Endline Survey Objectives
The objective of the RAcE endline household survey was to assess care-seeking behavior for sick children, iCCM coverage, and caregiver knowledge, attitudes, and practices related to pneumonia, diarrhea, and malaria in the RAcE Mozambique intervention areas. We compared baseline and endline data to assess changes in sick child care-seeking, assessment, and treatment coverage as well as caregivers’ knowledge of childhood illnesses and perceptions of APE services, and used the information to make inferences about project accomplishments.
RAcEMozambiqueEndlineSurveyFinalReport 3
2 SURVEY METHODS
2.1 Survey Implementation and Partnership
INS conducted the RAcE endline survey, with technical assistance from ICF and logistical, administrative, and financial support from Save the Children. INS worked with ICF and Save the Children to finalize the questionnaire; led the training of enumerators, data entry operators, and supervisors; and provided oversight to the entire implementation process.
The survey protocol received ethical approval from ICF’s Institutional Review Board and from Mozambique’s Ministry of Health Bio Ethics Committee.
Annex A contains a complete list of the people involved in the survey and their roles.
2.2 Survey Design
Household survey: This was a cross-sectional cluster-based household survey, targeting primary caregivers of children aged 2–59 months who had recently been sick with diarrhea, fever, or fast breathing. All primary caregivers of children aged 2–59 months reported to have experienced diarrhea, fever, or cough with rapid breathing in the two weeks prior to interview were considered eligible for inclusion in the survey. ICF developed standardized sampling guidance for all RAcE projects, which was adapted for Save the Children Mozambique.
To be able to detect a 20 percent difference at 90 percent power with a two-tailed test and 95 percent confidence using cluster sampling, 263 cases were needed for each disease. ICF rounded up to 300 cases to ensure a consistent number of interviews per cluster and a slight increase in the precision of the coverage estimates.
The household survey used a 30x30 multi-stage cluster sampling methodology. At baseline, the target population comprised the RAcE project area: iCCM-eligible areas—more than 8 km and less than 25 km from a health facility, in Manica, Nampula, and Zambezia provinces.
At baseline, 30 clusters were selected using probability proportional to size (PPS). Inhambane was excluded from the baseline survey sample, because a household survey had been conducted there earlier in the year and there was concern about over-surveying those communities.
Prior to selection of enumeration areas, Save the Children and ICF worked with INS to develop the sampling frame, the list of all APEs (both active and inactive) and corresponding communities in the areas where RAcE was already being implemented. Where possible, the Instituto Nacional de Estatistica (National Institute of Statistics) provided population information about these enumeration areas.
The same clusters sampled at baseline were planned to be sampled at endline, plus an additional two clusters from Inhambane discussed further below. However, 9 of the 30 clusters were replaced using PPS, 4 due to security issues that made in the original clusters inaccessible and 5 because the clusters were no longer part of the project area because APEs had abandoned the sites sometime between 2012 and 2015.
During fieldwork, an additional five clusters were replaced by selecting the nearest accessible cluster from the sampling frame: three were misidentified in the sampling frame, and although they had APEs, they were not supported by Save the Children with RAcE funding; one was inaccessible due to conflict;
RAcEMozambiqueEndlineSurveyFinalReport 4
and one was replaced because upon arriving in the cluster the data collection team found that the APE serving that community had passed away the night before. All replacement sites were selected under the instruction of and with guidance from ICF and Save the Children.
Inhambane was included in the endline survey to get estimates for the entire RAcE project area, inclusive of all four provinces. Following discussion among project stakeholders, agreement was reached and supported by WHO, to get a project-wide estimate only. Per WHO’s request, a representative sample of Inhambane was not collected, because it did not have a baseline for comparison. To power the sample to obtain project wide estimates, we only needed to sample two clusters from Inhambane. These two clusters were selected using PPS. This sampling approach provides comparable estimates to baseline for the combined area of Manica, Nampula, and Zambezia, as well as endline estimates for the entire RAcE project area. Please see Annex B for a complete list of the 32 endline clusters (30 from the original 3 provinces and 2 from Inhambane).
Within each cluster, 10 interviews were conducted for each of the 3 illness modules—diarrhea, fever, and fast breathing—for a total of 30 interviews per cluster, or 320 interviews per each illness across the project area for the 32 clusters at endline (inclusive of Inhambane).
Within each cluster, the survey team randomly selected the first household for interview and proceeded to the household with its front door nearest to the front door of the current household until the team conducted 10 interviews for each illness.
At each household, the enumerator first determined whether an eligible child lived there. An eligible child was aged 2–59 months old and had been sick with diarrhea, fever, cough with rapid breathing, or any combination of the three illnesses in the two weeks preceding the survey.
If there was an eligible child in the household, the interviewer administered the questionnaire, including all applicable illness modules, to the caregiver of the eligible child. If more than one child was eligible, and they were sick with different illnesses, their caregiver was asked about each instance of illness. If there was more than one eligible child in the household for an illness, the interviewer randomly selected one of the eligible children and interviewed his or her caregiver. If multiple children in the same household were reported to have symptoms in the preceding two weeks but had different caregivers, interviewers could interview multiple caregivers, as long as not more than one child from each household was included for each iCCM condition.
Before survey implementation, based on experience with the baseline study, Save the Children expressed concerns about the difficulty in finding the requisite number of pneumonia cases per cluster. Therefore, if survey teams were not able to identify and survey the required number of cases per disease, the team visited an adjacent enumeration area using the same household selection methodology. In the 30 clusters in Manica, Nampula, and Zambezia, the sample included a total of 305 cases with cough with fast or difficult breathing, 308 cases of diarrhea, and 312 cases of fever. In the 32 clusters inclusive of Inhambane, the sample included a total of 326 cases with cough with fast or difficult breathing, 329 cases of diarrhea, and 341 cases of fever.
Details related to the household and respondent selection process are provided in Annex C.
APE survey: As part of the endline survey, Save the Children conducted a survey of APEs to assess the implementation strength and quality of the iCCM services they delivered. The APE survey was conducted with the endline household survey to assess care-seeking practices and treatment coverage
RAcEMozambiqueEndlineSurveyFinalReport 5
for iCCM conditions. The objective of the APE survey was to gain a better understanding of the APEs’ background characteristics, activity levels, and support and supervision received to help interpret the results of the household survey. The APEs serving the 32 clusters selected for the endline household survey formed the sample population for the APE survey, and all were interviewed.
2.3 Survey Questionnaire
ICF developed a standard household survey questionnaire for all RAcE grantees to use for their baseline surveys. Each grantee adapted the questionnaire to fit the iCCM program and country context, such as appropriate local terminology for community health workers, care-seeking locations, and treatment options. The same questionnaire used for the baseline survey was used at endline, with some additional questions added to the endline questionnaire: two questions were added to each of the illness modules to gather information about whether caregivers sought care for their sick child and whether they sought care from an APE.
The survey questionnaire contains seven modules: caregiver and household background information; caregivers’ knowledge of iCCM activities in their community; caregivers’ knowledge of childhood illness danger signs; household decision-making; and a module for each major childhood illness: fever, diarrhea, and fast breathing. In addition to collecting information about caregiver knowledge, care-seeking, and treatment coverage, the questionnaire collected standard Demographic and Health Survey data on household ownership of selected assets, materials used for housing construction, and types of water access and sanitation facilities, which ICF will analyze and use for the final evaluation.
Save the Children preferred mobile data collection over paper surveys. To avoid programming mistakes that complicated baseline survey analysis, ICF programmed the questionnaire using CommCare, and Save the Children uploaded this program onto the mobile devices (Samsung Galaxy 12 phones in Manica, Nampula, and Zambezia; iPad tablets in Inhambane).
The survey questionnaire was translated into Portuguese by a translator selected by ICF, and fine-tuned to Mozambican Portuguese by Save the Children staff. Because most households understand local languages rather than Portuguese, supervisors and enumerators, under the direction of INS staff, orally translated the questionnaire to each province’s local language during the enumerator and supervisor training. Under the direction of Save the Children and INS, the questionnaires were not translated into each local language in writing, because supervisors and enumerators do not necessarily read the local language.
Following the enumerator and supervisor training, the survey was pretested in each province in the local language. Questionnaire field testing was conducted in communities that were not in the survey sample. No adjustments were identified as being necessary to the questionnaire during field testing.
The APE survey questionnaire was developed based on tools used in the previous iCCM program funded by CIDA from 2009 to 2012 and from the more recent quality of care assessment conducted in January 2016. The questionnaire was administered to selected APEs in Portuguese using phones and tablets (same as the household survey questionnaire).
Annex D contains the survey household questionnaire.
RAcEMozambiqueEndlineSurveyFinalReport 6
2.4 Selection and Training of Survey Staff
Terms of reference for the enumerators and supervisors were modified from the baseline survey. Save the Children worked in collaboration with the Provincial Health Authorities to advertise the positions in the local papers and on the radio. INS indicated an individual in the Provincial Health Authorities to work with Save the Children on pre-selecting candidates, and the CVs for potential candidates were sent to INS for final selection.
More participants than were required to field the data collection teams were recruited from each province to participate in the training, to ensure high-quality teams with alternate enumerators available. A five-day training was held in Maputo September 27–October 1, 2016, led by INS with support from ICF and Save the Children. Some of the same INS and Save the Children staff who facilitated the baseline training also facilitated the endline training. INS supervisors from each of the four project provinces were also among the training facilitators. The endline training used the same format and structure as the baseline training, but it was conducted with all participants together in Maputo rather than as separate trainings in the provinces, as was done at baseline.1
The five-day training covered the following:
Overview of the RAcE project goals and activities
Fieldwork procedures
Overview of the roles and responsibilities of enumerators and supervisors
Review of the translated questionnaire and questions and the topics covered by the questionnaire, including consistency of the questions asked
Mobile phone data collection and practice on collecting data using mobile phones
Training participants took an exam after the second day of training, and based upon these results and their participation in the training, the project team selected the strongest participants to serve as team supervisors and the others to serve as enumerators. The training schedule is provided in Annex E.
2.5 Data Collection
Endline survey data collection took place October 9–15 in Inhambane, October 8–18 in Manica, October 10–20 in Nampula, and October 10–30 in Zambezia. A total of 12 survey teams (2 in Inhambane, 3 in Manica, 4 in Nampula, and 3 in Zambezia), each composed of 1 supervisor and 2 enumerators, conducted the data collection. Mobile devices were used to complete the survey, with smart phones (Samsung Galaxy 12) in Manica, Nampula, and Zambezia, and iPad tablets in Inhambane.
Written informed consent was obtained from each caregiver and APE prior to interview. Participation in the study was voluntary, and there was no penalty for non-participation. Survey respondents were not compensated for their time away from income-earning activities or daily duties for participating in the data collection. The questionnaire took approximately one hour to administer per caregiver interviewed.
1 Due to travel delays, the participants from Nampula arrived at the start of the second day of training. They spent additional time on second day getting an overview of material covered on the first day.
RAcEMozambiqueEndlineSurveyFinalReport 7
Quality control procedures during fieldwork included daily spot checks by supervisors, during which they observed at least one interview per enumerator per day and reviewed data entered into the mobile devices. At the end of each day, data were sent to a cloud server (CommCare) and reviewed by ICF staff, who collaborated with INS and Save the Children to address questions and follow up with the survey teams. However, many clusters did not have a mobile internet connection, so survey teams often needed to upload the data after they left the cluster area.
There were delays throughout fieldwork for a number of reasons. The distance between clusters within each province resulted in significant travel time. Often teams were not able to leave one cluster for the next in a timely manner because they first needed confirmation from ICF or INS that data from a completed cluster had been uploaded to the system. In some clusters, the survey team needed two to three days to complete the data collection. In addition, before starting data collection the survey teams had to present themselves to local authorities and then work with a community leader or guide to map the area before the household selection. These local guides were not always immediately available, thus delaying the start of fieldwork in some clusters. In Zambezia, the start date of the data collection was delayed by five days while the team waited to receive official credentials and permission to work in the province, which were provided by the Provincial Health Authorities.
In addition to various delays, and the challenge of long distances needed to travel between clusters, supervision was an issue. The survey supervisors assigned to Manica and Zambezia were unable to serve in these roles at the last moment, one due to illness and the other due to conflicts with the delayed survey schedule. The Manica survey team was consequently supervised remotely from Maputo and in Zambezia, the Inhambane INS supervisor relocated to supervise the team after completing the data collection in Inhambane.
The fieldwork schedule is provided in Annex F.
2.6 Data Entry and Management
Using mobile devices to capture data removed the additional step of data entry. Quality control checks were built into the mobile application to set allowable ranges and avoid the most common data entry errors. However, the application did allow enumerators to skip modules and questions without receiving an error report.
INS and Save the Children submitted the final dataset to ICF in Excel in separate spreadsheets for each module and for each province. Data were merged in Microsoft Excel and Stata.
The household survey dataset submitted to ICF had a number of errors that required extensive review, additional cleaning, and some reconstruction of data. This is due to two major errors: (1) some enumerators conducted interviews with caregivers who did not have a child who had been sick in the two weeks prior to the survey; these caregivers were asked the caregiver-specific modules but not the sick child modules even though they were not selected for the survey; and (2) many enumerators did not record the cluster number, household number, caregiver number, or child number correctly in each questionnaire module as they moved through the survey, meaning that ICF, in collaboration with INS, had to go back through the data and manually correct this information to ensure that questionnaire records were correctly linked.
RAcEMozambiqueEndlineSurveyFinalReport 8
2.7 Data Analysis
ICF analyzed the survey data using Stata v14 and Microsoft Excel. At the time of endline survey analysis, ICF also re-analyzed the baseline household survey and made key updates to correct a number of errors discovered in the baseline analysis to ensure comparability with endline data. Steps taken to clean and update endline survey data and key corrections made in the re-analysis of the baseline data are outlined in Annex G.
The analyst calculated point estimates and 95 percent confidence intervals for baseline and endline survey indicators accounting for cluster effects. To test for statistically significant changes between indicators at baseline and endline, a Pearson’s chi-squared test was used for binary and categorical variables and regression for continuous variables. Indicators with p-values less than 0.05 are determined to show a statistically significant change between baseline and endline.
Endline data are displayed disaggregated by a child’s sex and illness. For the comparison of indicators between baseline and endline, we only disaggregated data by sex if we found the differences between males and females to be statistically significant.
The initial analysis of the APE survey was conducted by ICF, which provided tables with point estimates and 95 percent confidence intervals for a set of indicators developed by Save the Children. Save the Children reviewed the tables and conducted additional analyses. The results of the APE survey are included in Annex I, and relevant aspects have been integrated into the main household survey results and discussion sections.
2.8 Survey Indicators
The household survey collected data on 18 key indicators related to caregiver knowledge of APEs and child illnesses; caregiver perceptions of APEs; and sick child care-seeking, assessment, treatment, referral adherence, and follow-up. The survey also collected information on household and caregiver characteristics and household decision-making.
The APE survey collected data on 16 indicators related to APE residency, functionality, medicine and diagnostics availability, supervision, service availability and activity levels, recording completeness, knowledge, and data display and use.
2.9 Survey Limitations
The survey provides estimates for the RAcE project area as a whole. The survey was not powered to provide province-level estimates. Resource limitations led RAcE stakeholders to agree that collecting an adequate sample for province-level estimates (four times the total sample size, to collect data on 300 cases of illness per disease, per province) was not feasible. However, the diversity of the RAcE provinces, and the fact that some RAcE provinces had different project implementation inputs and challenges, makes not having province-specific RAcE estimates a limitation in interpreting the changes in care-seeking and intervention coverage.
In addition, there are known potential biases and limitations with the indicators that assess caregiver recall of malaria diagnostic testing and coverage of appropriate treatment for children with fever and with cough with difficult or fast breathing. The potential biases and limitations of these indicators are further detailed in the findings section.
RAcEMozambiqueEndlineSurveyFinalReport 9
3 FINDINGS
The results presented in this report reflect only the 30 clusters from Manica, Nampula, and Zambezia where we had comparable baseline and endline data. A summary of results from the 32 cluster sample, representative of the full four-province project area, is presented in Annex H. Two additional factors should be acknowledged when interpreting the changes over time. First, iCCM was being implemented in Inhambane and Nampula provinces prior to the start of the RAcE project. The baseline survey thus reflects iCCM activity already occurring at that time. The endline survey sampled 17 clusters in Nampula. Second, RAcE household surveys used a sampling frame inclusive of only RAcE project areas. At baseline, the project area was defined as areas in which APEs were, or were planned to be, deployed during the project period. At endline, the project area was defined as areas in which APEs were deployed.
3.1 Characteristics of Sick Children and Caregivers
Table 2. Characteristics of sick children included in the survey
Characteristic Baseline EndlineSex* Male, % 50.5 47.6
Female, % 49.5 52.4 Age (months)* <12, % 22.7 24.0 6
12-23, % 24.4 23.2 24-35, % 21.8 20.5 36-47, % 18.3 19.2 48-59, % 12.8 13.1
Two week history of illness** Had fever, % 53.8 51.1
Had diarrhea, % 49.7 45.5 Had cough with fast
breathing, % 38.3 40.4
Average number of illnesses
1.4 1.4
Total number of sick children included in survey
579 696
Cases of illness Fever, n 271 307
Diarrhea, n 275 306 Cough with fast
breathing, n 230 296
Total number of sick child cases included in survey
776 909
* Age of 64 sick children and sex of 5 sick children missing at endline; age of 1 sick child missing at baseline. ** Fever information missing for 16 sick children, cough with fast or difficult breathing information missing for 1 child at baseline.
RAcEMozambiqueEndlineSurveyFinalReport 10
Table 3. Caregiver characteristics
Characteristic Baseline % Endline %Age (years)* 15-24 N/A 38.5
25-34 34.9 35-44 21.4 45-60 5.1 Mean age (years) 29.0
Education* None N/A 36.9 Primary, ≤ year 4 34.8 Primary, ≥ year 5 24.4 Secondary or higher 3.9
Marital status* Currently married or
living with partner 57.6 55.5
Not married but living with a partner
29.2 28.1 Not in union 13.2 16.4
Partner living with caregiver (among those in union) Yes 94.2 95.7 Total number of caregivers 519 652 * Age of 102 caregivers, education and relationship status of 5 caregivers at endline missing; marital status of 2 caregivers missing at baseline
Table 4. Reported distance and mode of transport to nearest health facility
Baseline Endline
% (CI%) % (CI%)
Distance to nearest facility
< 8 km 23.9
(13.7 - 38.4) 15.0
(8.4 - 25.2) 8-25 km
18.7 (12.0 - 27.9)
31.7 (20.9 - 45.1)
> 25 km 9.6
(3.3 - 25.1) 7.2
(3.0 - 16.5) Don’t know
47.8 (35.6 - 60.2)
46.1 (31.6 - 61.2)
Mean distance to nearest facility 13.7 km 14.5 km Number of caregivers 519 649 Mode of transport
Walk 81.4
(70.3 - 89.0) 79.8
(71.3 - 86.3) Motorbike/Taxi/Bus
9.7 (5.1 - 17.7)
11.1 (6.9 - 17.3)
Other 8.9
(4.3 - 17.5) 9.1
(5.3 - 15.1) Number of caregivers 516 649 Time to nearest facility (among those who go to the facility)
< 30 minutes
13.7 (6.7 - 26.1)
9.8 (4.7 - 19.3)
30–59 minutes
5.0 (2.7 - 9.0)
6.6 (2.8 - 14.6)
1–< 2 hours
38.5 (28.7 - 49.3)
16.7 (12.2 - 22.5)
2–< 3 hours
15.1 (10.6 - 20.9)
32.0 (23.9 - 41.3)
3 hours or more
15.1 (10.6 - 20.9)
32.0 (23.9 - 41.3)
Mean time to nearest facility 2 hours, 17 minutes 2 hours, 13 minutes Total number of caregivers 517 531
RAcEMozambiqueEndlineSurveyFinalReport 11
3.2 Decision-Making
Decision-making about both income and care-seeking was, at both baseline and endline, made predominantly by a caregiver’s spouse or partner. Approximately one-third of all caregivers reported making these households decisions jointly with their spouse or partner. As shown in Table 5, household decision-making about income and care-seeking did not change significantly over the course of the project period. However, at endline, among caregivers who sought care for their child aged 2–59 months who had been sick in the two weeks before the survey, slightly more than half (57 percent) made the decision to seek care jointly with their spouse or partner.
Table 5. Usual decision-maker in household about income and care-seeking
Decision-maker Income decisions
p-value Care-seeking decisions
p-value Baseline Endline Baseline Endline % (CI%) % (CI%) % (CI%) % (CI%)
Caregiver 5.1
(3.3 - 8.0) 7.4
(4.6 - 11.6) 7.4
(4.3 - 12.3) 13.3
(8.4 - 20.4) Caregiver’s husband or partner
71.7 (61.4 - 80.2)
67.5 (60.0 - 74.2)
68.8 (58.3 - 77.7)
65.4 (56.9 - 73.1)
Caregiver and partner jointly
22.5 (14.9 - 32.4)
23.8 (17.4 - 31.8)
0.8066 23.4
(15.5 - 33.7) 20.5
(15.4 - 26.8) 0.5636
Other 0.7
(0.2 - 3.0) 1.3
(0.5 - 3.1) 0.5
(0.1 - 1.9) 0.7
(0.2 - 2.4) Total number of caregivers
449 541
449 541
Table 6. Joint decision-making to seek care for sick child by illness
Decided to seek care jointly with partner/spouse
Baseline Endline Endline N % (CI %) % (CI %)
Overall N/A 56.3
(50.7 - 61.8) 751
Fever N/A 55.5
(48.8 - 61.9) 256
Diarrhea N/A 53.6
(46.5 - 60.6) 248
Cough with fast breathing N/A 59.9
(51.9 - 67.5) 247
N/A=not available
3.3 Caregiver Knowledge and Perception of APEs
The percentage of caregivers who know that an APE works in their community increased significantly from baseline (62 percent) to endline (93 percent) (p<0.001). There was a moderate increase in the percentage of caregivers who were aware of the APE in their community and were able to list at least two curative services provided by the APE. Of the many activities that APEs implement in communities, from community mobilization to assessment and treatment of iCCM illnesses, caregivers most noted malaria testing (54 percent) and malaria treatment (58 percent). Nearly all caregivers who know an APE is in their community knew where the APE was located at both baseline and endline.
RAcEMozambiqueEndlineSurveyFinalReport 12
Table 7. Caregiver knowledge of childhood illnesses
Caregiver knowledge Baseline Endline
p-value % (CI %) % (CI %)
Knows 2+ child illness signs 86.5 (81.8 - 90.2)
92.9 (87.3 - 96.1)
0.0474
Knows cause of malaria 67.4 (58.2 - 75.5)
75.9 (66.7 - 83.3)
0.1948
Knows fever is a sign of malaria 64.7 (57.7 - 71.2)
74.7 (66.8 - 81.2)
0.0348
Knows malaria treatment 72.6 (65.0 - 79.1)
86.5 (80.1 - 91.1)
0.0102
Total number of caregivers 519 644
Table 8. Caregiver knowledge of APE
Caregiver knowledge Baseline Endline
p-value % (CI %) % (CI %)
Knows APE works in community 62.0 (45.4 - 76.2)
93.4 (83.3 - 97.6)
0.0007
Total number of caregivers 518 649 Knows location of APE* 95.0
(90.6 - 97.4) 94.5
(90.1 - 97.0) 0.8075
Knows 2+ APE curative services 49.2 (36.8 - 61.8)
69.1 (59.4 - 77.5)
0.0142
Total number of caregivers 321 606 * 134 responses missing at endline
Table 9. Caregiver perceptions of iCCM APE
Caregiver perceptions Baseline Endline
p-value % (CI %) % (CI %)
View CCM-trained APEs as trusted health care providers
82.9 (75.3 - 88.5)
78.2 (70.5 - 84.4)
0.3070
Believe CCM-trained APEs provide quality services
74.8 (63.7 - 83.3)
76.7 (70.3 - 82.1)
0.7217
Found the CCM-trained APE at first visit (for all instances of care-seeking included in survey)*
N/A 82.2
(74.0 - 88.3) N/A
Cite the CCM-trained APE as a convenient source of treatment
80.4 (70.2 - 87.7)
81.9 (72.4 - 88.6)
0.8254
Total number of caregivers 321 606 N/A=not available * Denominator is 366 caregivers at endline; only those who sought care from an APE for at least one sick child are included
3.4 Care-Seeking
Care-seeking from APEs increased significantly between baseline and endline, although the overall level of care-seeking from an appropriate provider was high at baseline (79 percent) and remained high at endline (80 percent). The source of care-seeking changed substantially over the course of the project, shifting from public facilities as the location where most sought care (65 percent at baseline, 33 percent at endline) to APEs as the location where most sought care (38 percent at baseline, 70 percent at endline).
A similar shift took place in the first source of care. The percentage of cases of illness among children aged 2–59 months who were sick in the two weeks preceding the survey taken to an APE as a first source of care overall more than doubled, from 23 percent at baseline to 57 percent at endline (p<0.001). Among those who sought care, those who sought care from an APE first increased significantly over the course of the project, more than doubling overall, from 27 percent at baseline to
RAcEMozambiqueEndlineSurveyFinalReport 13
67 percent at endline. This pattern of increased care-seeking from APEs, overall and among those who sought any care, was consistent across fever, diarrhea, and fast breathing.
Among those who did not seek care at all (n=140 cases of illness), the majority did not seek care because either the condition was not serious (30 percent) or the condition passed (31 percent). Among those who sought care but did not seek care from an APE, the majority noted that they did not seek care from the APE because the APE was not available (33 percent) or the APE did not have medicines or supplies (19 percent). The APE survey results showed that although nearly all APEs (91 percent) resided in their communities, only 41 percent had also provided iCCM services at least 5 days per week, and only 37 percent had all essential iCCM medicines and supplies (see Annex I).2
Table 10. Source of care by illness
Illness
Sought care from appropriate provider* p-
value
APE was first source of care p-
value Baseline
N Endline
N Baseline Endline Baseline Endline% (CI %) % (CI %) % (CI %) % (CI %)
Overall 79.1
(73.3 - 84.0) 79.5
(73.2 - 84.7)0.8991
23.1 (15.1 - 33.6)
57.0 (48.5 - 65.1)
0.0000 776 909
Fever 84.1
(77.6 - 89.0) 77.5
(70.5 - 83.3)0.0982
24.4 (15.5 - 36.1)
54.1 (44.6 - 63.2)
0.0003 271 307
Diarrhea 74.2
(65.9 - 81.1) 78.1
(69.9 - 84.6)0.3570
22.9 (14.5 - 34.2)
60.8 (51.5 - 69.3)
0.0000 275 306
Cough with difficult or fast breathing
79.1 (72.3 - 84.6)
83.1 (76.3 - 88.3)
0.345221.7
(14.0 - 32.2)56.1
(47.4 - 64.4)0.0001 230 296
* Appropriate providers include hospital, private clinic, other health center, and APE.
Table 11. Care-seeking from APEs
Illness
APE was first source of care among those who sought any care p-
value Baseline N Endline N
Baseline Endline% (CI %) % (CI %)
Overall 27.2
(17.8 - 39.2) 67.4
(58.5 - 75.1) 0.0000 658 769
Fever 27.6
(17.3 - 41.1) 66.7
(55.8 - 76.0) 0.0001 239 249
Diarrhea 28.8
(18.4 - 42.0) 73.8
(65.0 - 81.1) 0.0000 219 252
Cough with difficult or fast breathing
25.0 (16.2 - 36.5)
61.9 (52.5 - 70.6)
0.0001 200 268
2 Artemether-lumefantrine (at least 1x6 or 2x6), amoxicillin, ORS, zinc, RDTs, and functional timer
RAcEMozambiqueEndlineSurveyFinalReport 14
Table 12. Cases of illness for which no care was sought
Illness Did not seek care
p-value
Sought care but not from APE
p-value Baseline Endline Baseline Endline % (CI %) % (CI %) % (CI %) % (CI %)
Overall 15.2
(11.2 - 20.4) 15.4
(12.5 - 18.9) 0.9266
62.0 (49.7 - 72.9)
30.8 (23.3 - 39.6)
0.0007
Fever 11.8
(7.8 - 17.6) 18.9
(14.5 - 24.2) 0.0119
45.2 (28.2 - 63.4)
31.7 (22.4 - 42.8)
0.2461
Diarrhea 20.4
(14.3 - 28.1) 17.7
(12.3 - 24.6) 0.4784
70.8 (57.7 - 81.1)
24.6 (17.7 - 33.1)
0.0000
Cough with difficult or fast breathing
13.0 (8.5 - 19.6)
9.5 (6.8 - 13.1)
0.2742 72.5
(60.5 - 81.9) 35.8
(27.2 - 45.5) 0.0001
Total number of sick child cases
776 909 658 769
3.5 Assessment
Caregiver recall of malaria diagnostic testing is poor, which could affect the malaria diagnosis and appropriate treatment indicators calculated. According to the Indicator Guide: Monitoring and Evaluating Integrated Community Case Management, “Studies have found poor sensitivity and specificity of maternal recall for malaria diagnostic tests (finger/heel stick). Consequently, the current recommendation is that household surveys track treatment coverage of fever and, where possible, supplement with data from service delivery assessment to better understand the proportion of suspected malaria cases that receive appropriate diagnosis and treatment.”3
Reported assessment of fever by APEs increased significantly over the course of the project. Among all cases of fever that were assessed by a provider, nearly all were assessed with a malaria rapid diagnostic test (mRDT)4 by either an APE or a nurse, but the percentage assessed by each shifted over the course of the project. At baseline, 73 percent of cases of fever tested were assessed by a nurse and 23 percent by an APE. At endline, 43 percent of fever cases tested were assessed by a nurse and 55 percent by an APE.
Among cases of fever managed by APEs, the percentage of cases for which the APE administered an mRDT increased significantly from baseline (19 percent) to endline (51 percent) p<0.001. Despite this 32 percentage point increase, the endline value of 51 percent is low and may be due to ongoing stockouts of malaria kits, which include rapid diagnostic tests (RDTs) and artemisinin-based combination therapy (ACT). The APE survey revealed that 75 percent of APEs had mRDTs in stock on the day of the survey, but that only 44 percent had at least one age formulation of first-line antimalarial drugs (Coartem; artemether-lumefantrine 1x6 or 2x6) in stock (see Annex I). Data on the availability of RDTs and ACTs at first-level health facilities were not collected at the time of the survey.
Assessment of cough with fast or difficult breathing by APEs did not increase over the project period.
3 Maternal and Child Health Integrated Program. (2013). Indicator Guide: Monitoring and Evaluating Integrated Community Case Management. 4 The survey questionnaire asked caregivers whether blood was drawn. For purposes of this report, we assume that any blood draw in a community and by an APE is an mRDT. However, at the health facility level it is possible that some children received a blood smear and microscopy test for malaria.
RAcEMozambiqueEndlineSurveyFinalReport 15
Table 13. Malaria assessment among children with fever
Fever assessment Cases managed by APE
p-value All cases
p-value Baseline Endline Baseline Endline % (CI %) % (CI %) % (CI %) % (CI %)
Child had blood drawn 19.1
(10.6 - 32.0) 51.2
(38.2 - 64.0) 0.0008
43.9 (35.4 - 52.8)
51.1 (43.9 - 58.3)
0.2490
Caregiver received result of blood test
100 96.6
(89.6 - 98.9) 0.3907
84.9 (75.0 - 91.3)
93.6 (87.2 - 96.9)
0.0603
Blood test positive for malaria
100 91.7
(75.6 - 97.5) 0.3211
92.9 (81.7 - 97.5)
91.0 (79.4 - 96.4)
0.7216
Received ACT after positive blood test, among those who had a positive blood test
50.0 (30.6 - 69.4)
56.6 (42.0 - 70.2)
0.5494 93.5
(85.8 - 97.1) 82.6
(68.8 - 91.1) 0.0626
Total number of fever cases
131 170 271 307
Table 14. Respiratory rate assessment
Respiratory rate assessment
Cases managed by APEp-value
All cases p-value Baseline Endline Baseline Endline
% (CI %) % (CI %) % (CI %) % (CI %)
Respiratory rate assessed 23.2
(12.2 - 39.8) 39.0
(27.1 - 52.3) 0.0915
32.6 (25.7 - 40.4)
39.2 (29.6 - 49.7)
0.2398
Total number of cough with difficult breathing cases
56 172 230 296
3.6 Treatment Coverage
Overall, appropriate treatment of iCCM illnesses did not increase over the course of the project. Appropriate treatment for diarrhea—treatment with both ORS and zinc—increased significantly by almost 23 percent, from only 8 percent at baseline to 31 percent at endline.
Of the cases of illness among children aged 2–59 months who sought care from any provider in the two weeks prior to the survey, the percentage who received appropriate treatment from the APE increased significantly, from 12 percent at baseline to 29 percent at endline (p<0.0001). Of those who sought care from an APE, the percentage who received appropriate treatment from the APE did not change significantly. Looking at appropriate treatment for each iCCM illness by an APE among only those cases that sought care from an APE, there was a slight but non-significant decrease in treatment of cough with difficult or fast breathing with amoxicillin by an APE. There was a significant (p<0.0001) and nearly five-fold increase in the treatment of diarrhea with ORS and zinc by an APE, from 8 percent at baseline to 36 percent at endline, but no difference in the percentage of cases of diarrhea among children whose caregiver continued fluids or feeding for the child during the illness. Although appropriate treatment with diarrhea improved over the course of the project, still only one-third of children with diarrhea taken to an APE received treatment with ORS and zinc as recommended. The APE survey results showed that most APEs had ORS (91 percent) and zinc (81 percent) in stock on the day of the survey, and most (75 percent for ORS and 63 percent for zinc) reported no stockouts in the past month. In comparison, 81 percent of APEs had amoxicillin in stock on the day of the survey, but only 53 percent reported having no stockouts of amoxicillin in the past month, highlighting that during the recall period of the survey quite a few APEs may have had no amoxicillin in stock. Stockouts of ACTs and RDTs were widespread, with only 28 percent of APEs reporting continuous stock of ACTs and 44 percent reporting
RAcEMozambiqueEndlineSurveyFinalReport 16
continuous stocks of RDTs during the month before the survey (refer to Annex I for more details). APEs must have both RDTs and ACTs in stock to provide appropriate treatment for fever; if RDTs are out of stock, the APEs are to refer to the health facility even if they have ACTs in stock; if they are out of ACTs but have RDTs, the APEs should test the fever and refer malaria positive cases to the nearest facility for treatment. Further analysis of the APE survey showed that only 41 percent (13 of 32) of APEs had both RDTs and ACTs in stock at the time of the survey, and only 19 percent (6 of 32) of APEs reported no stockouts of either RDTs or ACTs during the month before the survey.
Table 15. Treatment coverage
Received appropriate treatmentp-value
Baseline N
Endline N Illness (treatment)
Baseline Endline% (CI %) % (CI %)
Overall 45.7
(40.7 - 50.7) 49.3
(42.1 - 56.5) 0.4168 545 734
Confirmed malaria (ACT) 93.5
(85.8 - 97.1) 82.6
(68.8 - 91.1) 0.0626 92 132
Confirmed malaria (ACT within two days)
77.2 (66.5 - 85.2)
70.5 (59.4 - 79.5)
0.3652 92 132
Diarrhea (ORS and zinc) 8.1
(4.3 - 14.7) 31.1
(21.8 - 42.1) 0.0001 223 306
Cough with difficult or fast breathing (amoxicillin)
69.6 (61.1 - 76.9)
58.8 (50.9 - 66.2)
0.0558 230 296
Table 16. Appropriate treatment by APEs
Received appropriate treatment from
APE p-value
Baseline N
Endline N
Illness (treatment) Baseline Endline% (CI %) % (CI %)
Overall 11.9
(7.4 - 18.5) 28.8
(22.5 - 36.1) 0.0003 538 725
Confirmed malaria (ACT)* 25.9
(14.4 - 42.1) 38.2
(27.0 - 50.8) 0.1579 85 123
Confirmed malaria (ACT within two days)*
23.5 (12.9 - 38.9)
34.2 (23.5 - 46.7)
0.1736 85 123
Diarrhea (ORS and zinc) 2.2
(0.7 - 7.1) 22.6
(15.0 - 32.4) 0.0000 223 306
Cough with difficult or fast breathing (amoxicillin)
17.0 (10.2 - 27.0)
33.1 (26.0 - 41.1)
0.0164 230 296
* Denominator for received appropriate treatment from APE for confirmed malaria indicators is restricted to children aged 6–59 months. APE protocol is to refer children under five months who have fever.
Table 17. Sought care from an APE and received appropriate treatment by APE
Received appropriate treatment
from APE among those who sought care from APE p-value
Baseline N
Endline N
Illness (treatment) Baseline Endline % (CI %) % (CI %)
Overall 38.9
(31.7 - 46.5) 46.5
(37.9 - 55.4) 0.0958 162 445
Confirmed malaria (ACT within 2 days) 44.2
(25.4 - 64.8) 50.6
(35.8 - 65.3) 0.5605 43 83
Diarrhea (ORS and zinc) 7.8
(2.7 - 20.8) 35.8
(24.5 - 48.9) 0.0011 64 190
Cough with difficult or fast breathing (amoxicillin)
70.9 (55.3 - 82.8)
56.4 (46.5 - 65.8)
0.1181 55 172
* Denominators for received appropriate treatment from APE for confirmed malaria indicators is restricted to children aged 6–59 months.
RAcEMozambiqueEndlineSurveyFinalReport 17
Table 18. Continued fluids and feeding during diarrhea episode
Continued feeding and fluids
Baseline Endlinep-value
% (CI %) % (CI %)
Continued fluids 43.3
(36.1 - 50.7) 38.7
(29.9 - 48.3) 0.3962
Continued feeding N/A N/A Total number of sick children 275 297 N/A=not available
3.7 First Dose of Treatment and Counseling from APE
APE treatment protocols indicate that the first dose of treatment should be provided by the APE at the time of assessment and that the APE should counsel the caregiver on how to provide treatment to his or her child. However, according to caregivers surveyed, less than half of cases treated by an APE received the first dose of the treatment in the presence of the APE. There were no significant changes overall over the course of the project. However, there was a significant increase in the percentage of cases of cough with difficult or fast breathing that received the first dose of amoxicillin from the APE (28 percent at baseline to 51 percent at endline). The percentage of cases of confirmed malaria that received the first dose of ACT in presence of the APE dropped significantly (p<0.05), from 56 percent at baseline to only 27 percent at endline.
Counseling by APEs on how to provide treatment was high at baseline and remained high at endline.
Table 19. First dose of treatment from APE
Illness (treatment)
First dose received in presence of APE
p-value Baseline
N Endline
N Baseline Endline% (CI %) % (CI %)
Overall 41.3
(29.7 - 53.9) 33.7
(24.5 - 44.5) 0.3549 92 249
Confirmed malaria (ACT) 56.3
(36.8 - 73.9) 26.8
(16.5 - 40.4) 0.0289 48 82
Diarrhea (ORS) 26.2
(17.3 - 37.8) 28.8
(18.6 - 41.7) 0.7322 61 153
Diarrhea (zinc) 0.0 35.0
(21.3 - 51.7) 0.2039 5 80
Diarrhea (ORS and zinc) 0.0 17.4
(7.0 - 37.1) 0.4418 5 69
Cough with difficult or fast breathing (amoxicillin)
28.2 (21.0 - 36.7)
51.0 (37.4 - 64.5)
0.0014 39 98
Table 20. Counseling on treatment administration by APE
Illness (treatment)
Counseled on treatment administration
p-value Baseline
N Endline
N Baseline Endline% (CI %) % (CI %)
Overall 96.7
(90.8 - 98.9) 96.0
(92.0 - 98.0) 0.7481 92 249
Confirmed malaria (ACT) 95.8
(85.2 - 98.9) 98.8
(91.2 - 99.8) 0.2781 48 82
Diarrhea (ORS) 95.1
(85.5 - 98.4) 96.1
(91.9 - 98.2) 0.6876 61 153
Diarrhea (zinc) 80.0
(23.0 - 98.2) 98.8
(89.9 - 99.9) 0.0196 5 80
RAcEMozambiqueEndlineSurveyFinalReport 18
Illness (treatment)
Counseled on treatment administration
p-value Baseline
N Endline
N Baseline Endline% (CI %) % (CI %)
Diarrhea (ORS and zinc) 80.0
(22.8 - 98.2) 95.7
(86.4 - 98.7) 0.1740 5 69
Cough with difficult or fast breathing (amoxicillin)
100 93.9
(85.6 - 97.5) 0.1767 39 98
3.8 Referral Adherence
The levels of reported referrals from an APE increased markedly from 26 percent at baseline to 48 percent at endline. Caregiver’s adherence to a referral provided by an APE was reported at moderate levels at baseline (69 percent overall for all illnesses) and remained similar at endline (65 percent overall for all illnesses). For those who did not adhere to the referral, the most common reason given for not adhering was because the facility was too far (45 percent at baseline, 42 percent at endline), followed by the child improving (35 percent at baseline, 45 percent at endline). Other reasons reported included having no money, having no transport, having no time, and thinking that the illness was not serious.
Table 21. Adherence to APE referral
Illness Baseline Endline
p-value Baseline
N Endline
N % (CI %) % (CI %)
Overall 68.8
(54.2 - 80.3) 65.4
(55.4 - 74.2) 0.6689 64 257
Fever 80.0
(62.1 - 90.7) 70.6
(57.0 - 81.3) 0.3466 25 68
Diarrhea 75.0
(39.1 - 93.4) 60.4
(46.4 - 72.9) 0.4266 12 91
Cough with difficult or fast breathing
55.6 (33.4 - 75.7)
66.3 (52.5 - 77.9)
0.3401 27 98
3.9 Sick Child Follow-Up
According to the iCCM protocol, APEs are trained to counsel mothers to return for follow-up within three days. The percentage of cases of all illnesses for which mothers returned for follow-up with the APE with their child did not change significantly over the course of the project. The number of cases for which the mothers followed up, however, was low (7 at baseline, 55 at endline). Among the small number of caregivers who did return for follow-up with an APE, the percentage of those who followed up within three days increased from 0 percent at baseline to 22 percentage at endline. The percentage of cases of all illnesses for which APEs followed up with the child did not change over the course of the project. Among cases of fever, there was a significant increase in APE follow-up, from 15 percent at baseline to 38 percent at endline. However, it is important to note that the cases of illness captured in this survey are those that occurred at any time in the previous two weeks. Therefore, some cases of illness captured in the survey were not yet be eligible for follow-up by either the mother or the APE.
RAcEMozambiqueEndlineSurveyFinalReport 19
Table 22. Caregiver follow-up with APE
Condition Baseline Endline
p-value Baseline N Endline N % (CI %) % (CI %)
Overall 10.2
(4.1 - 23.1) 10.3
(6.1 - 16.9) 0.9826 137 532
Fever* 10.0
(2.5 - 32.8) 9.4
(4.7 - 17.9) 0.9350 20 170
Diarrhea 9.5
(3.3 - 24.6) 11.1
(6.1 - 19.1) 0.8018 63 190
Cough with difficult or fast breathing
11.1 (4.8 - 23.5)
10.5 (5.1 - 20.4)
0.9156 54 172
*112 responses missing for fever at baseline
Table 23. APE follow-up with sick child
Illness Baseline Endline
p-value Baseline N Endline N % (CI %) % (CI %)
Overall 40.9
(25.0 - 58.9) 41.2
(32.1 - 50.9) 0.9753 137 532
Fever* 15.0
(5.2 - 36.1) 37.7
(26.9 - 49.8) 0.0453 20 170
Diarrhea 47.6
(30.9 - 64.9) 38.4
(28.7 - 49.2) 0.3417 63 190
Cough with difficult or fast breathing
42.6 (23.0 - 64.9)
47.7 (37.2 - 58.4)
0.6505 54 172
*112 responses missing for fever at baseline
3.10 Illness Management and Diagnostics by Sex
There were no notable differences in appropriate assessment or treatment of iCCM illnesses between boys and girls.
Table 24. Fever management and treatment administered
Sex** Confirmed malaria treatment*
Number of children with positive RDT
Any antimalarial ACT ACT within 24 Hours % (CI %) % (CI %) % (CI %)
Overall 87.1
(76.1 - 93.5) 82.6
(68.8 - 91.1) 70.5
(59.4 - 79.5) 132
Male 87.1
(74.8 - 93.9) 82.9
(65.4 - 92.5) 71.4
(56.6 - 82.8) 70
Female 86.9
(73.0 - 94.2) 82.0
(68.0 - 90.7) 68.9
(56.9 - 78.7) 61
€ Among to those who sought care from a hospital, private clinic, other health center, or APE. ** Sex is missing for one child with fever.
Table 25. Fever diagnostics
Sex** Had blood taken from
finger or heel Among those who had blood taken Number of children
With fever Were given results Test result positive
Overall 51.1
(43.9 - 58.3) 93.6
(87.2 - 96.9) 91.0
(79.4 - 96.4) 307
Male 55.9
(46.0 - 65.3) 93.8
(87.0 - 97.2) 92.1
(77.2 - 97.6) 145
Female 46.6
(38.0 - 55.4) 93.3
(78.1 - 98.2) 87.1
(73.2 - 94.4) 161
** Sex is missing for one child with fever.
RAcEMozambiqueEndlineSurveyFinalReport 20
Table 26. Diarrhea management by provider and treatment administered
Sex Sought any advice or treatment
Sought treatment from an
appropriate provider€
Sought treatment
from an APE
Sought treatment
from an APE as first choice
Given more than usual to drink*
TreatmentTreated
with ORS AND zinc
Number of children
with diarrhea
ORS Homemade
fluid Zinc
Overall 82.4
(75.4 - 87.7) 77.5
(70.5 - 83.3) 62.1
(52.9 - 70.5) 60.8
(51.5 - 69.3) 38.7
(29.9 - 48.3) 69.9
(62.5 - 76.5) 25.5
(17.9 - 35.0) 35.0
(25.2 - 46.3)31.1
(21.8 - 42.1) 306
Male 83.0
(74.6 - 89.0) 78.9
(68.2 - 86.7) 65.3
(54.3 - 74.9) 64.0
(53.2 - 73.5) 44.7
(33.3 - 56.6) 66.7
(57.1 - 75.0) 19.7
(12.0 - 30.7) 36.7
(25.3 - 49.8)33.3
(23.2 - 45.3) 147
Female 81.8
(73.3 - 88.0) 77.4
(68.3 - 84.4) 59.1
(49.0 - 68.5) 57.9
(47.8 - 67.3) 33.3
(24.4 - 43.7) 73.0
(63.7 - 80.6) 30.8
(21.3 - 42.3) 33.3
(23.0 - 45.6)28.9
(19.0 - 41.5) 159
€ Refers to those who sought care from a hospital, private clinic, other health center, or APE. * Nine missing responses
Table 27. Cough with difficult breathing management by provider and treatment
Sex** Sought any advice or treatment
Sought treatment from an appropriate
provider€
Sought treatment from
an APE
Sought treatment from an APE as first
choice
Assessed for rapid breathing
Treatment Number of children
with cough and fast or
difficult breathing
Any antibiotic Amoxicillin
Overall 90.5
(86.9 - 93.2) 83.1
(76.3 - 88.3) 58.1
(49.2 - 66.5) 56.1
(47.4 - 64.4) 39.2
(29.6 - 49.7) 60.33
(52.28 - 67.85) 58.8
(50.9 - 66.2) 296
Male 90.0
(83.5 - 94.1) 83.9
(74.2 - 90.4) 57.7
(47.5 - 67.3) 55.4
(44.7 - 65.6) 41.5
(30.6 - 53.4) 63.9
(53.1 - 73.4) 62.3
(51.3 - 72.2) 130
Female 90.7
(85.3 - 94.3) 82.7
(74.0 - 89.0) 58.0
(47.6 - 67.8) 56.2
(46.2 - 65.7) 37.7
(26.3 - 50.6) 55.6
(45.3 - 65.4) 54.9
(45.0 - 64.5) 162
€ Refers to those who sought care from a hospital, private clinic, other health center, or APE. ** Sex is missing for four children with cough and fast or difficult breathing.
RAcEMozambiqueEndlineSurveyFinalReport 21
4 DISCUSSION
The revitalization of the APE program has increased the presence and scope of APEs in communities over the course of the RAcE project period. At the time of the baseline, iCCM services were available in most districts in Nampula and Inhambane provinces and were introduced in Manica and Zambezia provinces through RAcE. During the RAcE project, caregivers’ knowledge of APEs and care-seeking from APEs, specifically for iCCM services, increased. The percentage of caregivers who know that an APE works in their community increased significantly from baseline (62 percent) to endline (93 percent) p<0.001. Care-seeking from APEs increased significantly between baseline and endline, even though the overall level of care-seeking was high at baseline and remained high at endline. The percentage of children aged 2–59 months who were sick in the two weeks preceding the survey and were taken to a CCM-trained APE as a first source of care overall more than doubled, from 23 percent at baseline to 57 percent at endline (p<0.001). This pattern of increased care-seeking from APEs was consistent across fever, diarrhea, and fast breathing. As care-seeking from APEs increased, there was a concurrent shift in the source of care-seeking, from public facilities to APEs. At baseline, the majority of caregivers (65 percent) sought care from public facilities, and only 38 percent sought care from APEs; at endline, this flipped to most caregivers seeking care from APEs (69 percent) and only 34 percent seeking care from public facilities. Considering the overall increase in care-seeking, this suggests that deployment of APEs to provide iCCM services has indeed extended health services to those who might not have previously sought care. Among those who did seek care but did not seek care from an APE, the majority noted that they did not seek care from the APE because the APE was not available (33 percent) or the APE did not have medicines or supplies (19 percent).
These achievements suggest an overall increase in access to care in RAcE project areas. This is commendable, particularly given that episodic political violence rendered some project areas inaccessible at points during project implementation. This episodic violence also made some project areas inaccessible during both baseline and endline surveys, consequently skewing the results of the surveys to be more reflective of areas that likely had better or more consistent provision of services by APEs.
Significant improvements were found in assessment of cases of illness by APEs and in appropriately managing cases of illnesses. Reported assessment of illness by APEs increased significantly for both fever and cough with difficult or fast breathing over the course of the project.
Assessment of fever and diagnosis of malaria, in particular, significantly improved over the course of the project. Among those who sought care from an APE, the percentage of children aged 2–59 months with fever in the two weeks preceding the survey who were given an RDT (had a finger or heel stick) by an APE more than doubled from baseline to endline, from 19 percent at baseline to 51 percent at endline (p<0.05). Despite this 32 percentage point increase, the endline value of 51 percent is low, and this is most likely due to ongoing stockouts of malaria kits, which include RDTs. Data collected from RAcE teams in the provinces (source Serviço Distrital de Saúde, Mulher e Acção Social) post survey showed that stockouts of the malaria kits were registered in 6 districts in August, in 11 districts in September, and in 9 districts in October. In addition, the APE survey noted widespread stock disruptions; only 19 percent of APEs had continuous stock of both RDTs and ACTs the month before the survey (covering the 2-week recall period of the household survey).
RAcEMozambiqueEndlineSurveyFinalReport 22
The percentage of sick children receiving appropriate treatment for iCCM illnesses did not change over the course of the project. The percentage of sick children receiving appropriate treatment for an iCCM illness from an APE, however, increased significantly, from 12 percent at baseline to 29 percent at endline (p<0.001). At endline, appropriate treatment by APEs accounted for over half of all reported appropriate treatment [58.4 percent (28.8 percent of cases of illness received appropriate treatment from an APE/49.3 percent of cases of illness received appropriate treatment, from any source)] compared to about a quarter at baseline [26.0% (11.9%/45.7%)].
Appropriate treatment for diarrhea—treatment with both ORS and zinc—increased significantly by almost 23 percent, from 8 percent at baseline to 31 percent at endline. Among those cases that sought care from an APE, there was a nearly five-fold increase in treatment of diarrhea with both ORS and zinc by an APE, from 8 percent at baseline to 36 percent at endline. The increased availability of zinc among APEs between baseline and endline can help explain the improvement. There was no significant change in appropriate malaria treatment or in appropriate treatment for cough with difficult or fast breathing. The lack of a measured increase for appropriate fever treatment could be in part due to consistent and widespread stockouts of RDTs and ACTs, both of which need to be in stock for the APE to delivery appropriate treatment. It is also important to consider that caregiver recall of RDT testing is poor and that this could affect the results of malaria diagnosis and appropriate treatment. Caregivers can better recall treatment of any fever. Stockouts of amoxicillin, reported by 47 percent of APEs during the month before the survey, may have also contributed. Treatment of cough with difficult or fast breathing must be interpreted carefully. Pneumonia treatment, for which this indicator is a proxy, is globally recognized to have validity issues because this diagnosis of presumptive pneumonia is often inaccurate in comparison with clinical diagnosis of pneumonia at health facilities. Therefore, the number of cases of cough with difficult or fast breathing is likely an overestimate of actual clinical pneumonia cases, and the percentage of these treated with amoxicillin can, and should, reasonably not be 100 percent.5
The relatively low coverage of treatment with ORS and zinc among diarrhea cases managed by APEs (36 percent) is concerning. Stock levels of both ORS and zinc were quite high at the time of the endline, so lack of stock cannot explain all missed cases. A quality of care survey conducted in January 2016 found that APEs failed to identify and treat many cases of diarrhea when compared to a gold standard examiner. In cases where children presented with other iCCM conditions, such as fever or cough and fast-breathing, the APE may have overlooked the diarrhea.
Although the increases in appropriate assessment and treatment coverage between baseline and endline are positive, the values observed at endline are lower than expected in a well-supported program in which APEs were well-supplied and supervised. An endline survey conducted in 2012 for the CIDA-funded iCCM program6 in three intervention districts of Nampula (Angoche, Erati, and Monapo) found that 68 percent of children aged 6–59 months who sought care first from an APE received an RDT, compared to 51 percent of all cases managed by an APE in the RAcE endline survey, and that
5 Campbell H, el Arifeen S, Hazir T, O'Kelly J, Bryce J, Rudan I, et al. (2013). Measuring coverage in MNCH: Challenges in monitoring the proportion of young children with pneumonia who receive antibiotic treatment. PLoS Med 10(5): e1001421. doi:10.1371/journal.pmed.1001421 6 CIDA 2012 endline survey and RAcE surveys are not directly comparable because they used different methodologies and questionnaires.
RAcEMozambiqueEndlineSurveyFinalReport 23
98 percent of RDT+ cases received ACTs, compared to 57 percent in RAcE endline. An APE survey carried out alongside the 2012 CIDA endline household survey found that most APEs had RDTs (87 percent) and ACTs (80 percent) in stock, whereas the current endline APE survey found widespread shortages. With the RAcE project, the supply chain management shifted from being led by Save the Children to being led by the MOH, resulting in greater challenges maintaining supply levels. Similar decreases were seen between the 2012 CIDA endline and the 2016 RAcE endline for assessment of respiratory rate by APEs. In the 2012 CIDA endline survey, 60 percent of children presenting first to an APE had their respiratory rate counted, compared to 39 percent of the cases of cough with difficult or fast breathing managed by APE in the RAcE endline survey. Only 78 percent of APEs had a functional timer at the time of the RAcE endline survey, which may have contributed to the lower levels observed in 2016. Provision of first dose and levels of follow-up with the APE were also higher in the 2012 CIDA endline survey, with 64 percent receiving first dose of ACT, compared to 27 percent receiving first dose of ACT in the 2016 RAcE endline. Also in the 2012 CIDA endline, 70 percent of fever cases treated by APE returned to the APE for follow-up, compared to 9 percent in 2016 RAcE endline. These differences cannot be explained by differences in stock levels and suggest that greater reinforcement of the protocols through supervision is required.
Given that the population served by APEs includes individuals who are already far from a health facility, it is not surprising to see the level of referral adherence unchanged over the course of the project, from 69 percent at baseline to 65 percent at endline. The primary reason caregivers did not adhere to the APE referral was because the facility was too far, emphasizing larger issues of poor access to care for many in Mozambique living far from health facilities, including many who live father than even the 25km reach of the APEs.
4.1 Implications
The RAcE project has completed implementation in Mozambique, and the MOH will continue to implement iCCM on its own. The findings of the household survey and APE survey highlight the critical importance of ensuring that APEs and health facilities are adequately supported to provide case management. The comparison of RAcE baseline and endline survey results demonstrates that caregivers seek care from APEs for iCCM conditions and that access to appropriate treatment in these remote areas is greatly improved. However, the iCCM program cannot exist without commodities, and the widespread stockouts recorded during the APE survey, particularly of antimalarial drugs and RDTs, help explain why 2016 RAcE endline survey results for assessment and treatment practices were weaker than those observed in the 2012 CIDA endline survey, when APEs were well-supplied. The handover of supply chain management responsibilities to the MOH under RAcE resulted in pervasive disruptions in stock of key iCCM commodities. Moving forward, the MOH and other partners will need to come together to find ways to sustain supplies to APEs or risk further declines in program performance.
RAcEMozambiqueEndlineSurveyFinalReport 24
ANNEX A. PEOPLE INVOLVED IN THE SURVEY
The following table provides a list of people involved in the survey and their roles.
Persons Involved Role
Ministry of Health and National Health Institute Dr Mbofana Principal Investigador Dr Chaquisse Coordination and Oversight Dr Jose Braz Chidassicua Coordination and Oversight Dr Acacio Sabonete Trainer and Data Quality Control Sergio Chicumbe Trainer Clementina Jacares Macondzo Provincial Survey Team Supervisor Nampula
Arminda Ubisse Provincial Survey Team Supervisor Manica Hesia Chilengue Provincial Survey Team Supervisor Inhambane and Zambezia Amilcar Magaço Provincial Survey Team Supervisor Zambezia Save the Children Marla Smith Oversight and Technical Eleanor Hill Coordination Lazaro Mondlane Logistics and Administration Fatima Ibo Logistics and Administration Malumbila Basilio Logistics and Administration Paula Nhambirre Logistics and Administration Tanya Gunther Technical Jeanne Koepsall Technical Ezequiel Barreto Technical (Mobile Devices for APE survey) Malaria Consortium Sonia Casimiro Trigo Logistics and Administration Helen Counihan Technical Ana Cristina Castel - Branco Technical ICF Meghan Swor Technical Debra Prosnitz Technical Kirsten Zalisk Technical
RAcEMozambiqueEndlineSurveyFinalReport 25
Data Collection Teams
Province Team Name Role
Nampula A 1 Maria Anabela Uaraca Supervisor
Benilto Marcolino Germano Enumerator
Manuel Jacinto Paulino Enumerator
Nampula B 2 Jeirezinho O.A. Calisto Supervisor
Alvaro Calisto E. Orlando Enumerator
Mequeza Abdul Carimo Enumerator
Nampula C 3 Sérgio Vieira Supervisor
Mamudo Assane Sualehe Enumerator
Joaquina Fernando Enumerator
Nampula D 4 Heraclito S. Bras Cassimo Supervisor
Salmata Braimo Selemane Enumerator
Abubacar Abilio Paconeta Enumerator
Zambézia A 5 Zarina Jose Cussinho Supervisor
Baciao Dinis Enumerator
Carlos Marcelo Neves Enumerator
Zambézia B 6 Antonio Cipriano Calia Supervisor
Amilcar Salvador Mole Enumerator
Vania Salvador Enumerator
Zambézia C 7 Anelita Emano Assane Supervisor
Rosario da Silva Valia Enumerator
Valdmiro Albino Xavier Enumerator
Manica A 8 Albertina Lucinda J Manuel Supervisor
Mequelina Alfredo Alice Enumerator
Marcelo Brito Fernandes Enumerator
Manica B 9 Silvia Damião Supervisor
Anastância Samuel Naene Enumerator
António Simões Julio Enumerator
10 Verónica Caetano Nhancalize Supervisor
Manica C Ana Rita Faustino Boane Enumerator
Paulino Jaime Gimo Enumerator
Inhambane A 11 Merton Andre Mueeombe Supervisor
Fatima Herminio Matandalasse Enumerator
Jorge Mario Binguane Enumerator
Inhambane B 12 Natercia Martins Langa Supervisor
Samuel Saranga Enumerator
Nace do Crescencio Rungo Enumerator
RAcEMozambiqueEndlineSurveyFinalReport 26
ANNEX B. ENDLINE SAMPLE
*-denotes districts where iCCM had been rolled out at baseline
Province District Community (Cluster)
Nampula
Angoche* Siretene/ Bairro Mutolo, replaced by Ntupo- Nparame due to death of APE Zoro/ Quarteirão E
Erati* Zuleme OU Zulume Malema* Murrumbo/ Nivatha
Memba
Regulo Napita/ Chefe Nivaca replaced by community Nehequene due to death of APE Reg Puejequele, no APE in this community so was replaced by Mangane Sirissa
Mogincual* Chapueia Mogovolas* Nanhotho Moma* Cotocuane/ Reg Machangula Mossuril Namuco Muecate Napacala/ Chinatane Murrupula* Muarrapaz
Nacaroa* Cabo Juliano Mavalane
Rapale Tchaine Ribaue* Namarepo
Zambezia
Alto-Molocue Yeheiya Nauela/ Milevane (community is called Guilherme)
Gile Namerico/ Uataria (ou Uaturia) Puampuela replaced with Quirule due to absence of APE in Puampuela
Inhassunge Muterreno/ Mussangane
Lugela Comone, replaced with adjacent community Mpemula due to security situation
Milange Muriamuendo Manica Vanduzi Dongo
Gondola
Nhamagoa Amatongas/ Chipangara Mussangaze/ Nhoane Gojombe Manhere
Inhambane Panda* Polana Zavala* Nhambele
RAcEMozambiqueEndlineSurveyFinalReport 27
ANNEX C. DETAILED SAMPLING DESIGN
Within each cluster, the survey team selected the first household for interview using the selection by subdivision approach.7 The survey team proceeded from one household to the next by visiting the household with its front door nearest to the front door of the current household, until the team conducted 10 interviews for each illness.
In all surveyed clusters in Manica and Nampula the survey teams found adequate cases of each illness. In Inhambane and Zambezia, there were some clusters (the exact number was not specified by INS) in which, after visiting all selected households, the survey team did not find 10 cases of each illness, so the team proceeded to the nearest enumeration area to find the number of illness cases still needed.
At each household, the enumerator first determined whether an eligible child lived there. An eligible child was aged 2–59 months and had been sick with diarrhea, fever, cough with rapid breathing, or any combination of the three illnesses in the two weeks preceding the survey.
The enumerator administered the questionnaire, including all applicable illness modules, to the caregiver of the eligible child. If more than one child was eligible, and they were sick with different illnesses, their caregiver was asked about each instance of illness. If the children had different caregivers, each caregiver was administered a separate questionnaire and answered questions about only his or her eligible child or children. If there was more than one eligible child in the household for an illness, the interviewer randomly selected one of the eligible children and interviewed his or her caregiver.
7 Selection by subdivision instructions: Go to the population center of the sample area and identify four quadrants. Assign each quadrant a number. Write the numbers on separate slips of paper and put them in a container. Randomly select one of the slips of paper. Go to the place that equally divides the selected quadrant’s population in half. Randomly select which of the two ways to proceed, and repeat this step until you have a manageable set of households (for example, 30). Count the households in the area. Use a random number table to select the first household for interview.
RAcEMozambiqueEndlineSurveyFinalReport 28
ANNEX D. ENDLINE SURVEY QUESTIONNAIRE
See attachment.
RAcEMozambiqueEndlineSurveyFinalReport 29
ANNEX E. ENDLINE SURVEY TRAINING SCHEDULE
Agenda de Formação do Estudo de Base do Projecto de RAcE-Mozambique
Maputo Mozambique
Centro de Reciclagem, 27 de Setembro-1 de Outubro
27 de Setembro: Dia 1 Hora Sessão e Objectivos Metodologia/Material Facilitador8:00-8:30 Introduções
Apresentar todos os membros da equipe do inquérito
Introduzir projecto RAcE e objectivos da pesquisa
Detalhes administrativos Provide all members of the
survey team Enter RAcE project and
objectives of research Administrative Details
MISAU/INS/SC
8:30-8:45 Abertura de Treino Training Opening
8:45-9:15 Desenho do Estudo Study Design INS 9:15-9:45 Aspectos éticos e Política de
Salvaguarda da Criança Ethical aspects and the Protection of the Child Policy
-Ficha de salvaguarda da criança ICF/SC
9:45-10:00 Pausa Lanche 10:00-10:45 Responsabilidades dos inquiridores
e supervisores Responsibilities of enumerators and supervisors
ICF
10:45-11:30 Procedimentos do trabalho de campo field work procedures
ICF
11:30-12:15 Seleção dos agregados familiares Selection of households
ICF
12:15-13:00 Seleção dos entrevistados Selection of respondents
Fichas de Controlo de Módulos de Doença
SC
13:00-14:00 Almoço
14:00-16:00 Prática da selecção dos agregados familiares e dos entrevistados
Practice of selection of households and respondents
INS + ICF + SC
16:00-16:45 Apresentação dos telemoveis e como usá-los
Manutenção Aplicação no telemóvel
Ezequiel Barreto
RAcEMozambiqueEndlineSurveyFinalReport 30
para randomização
Presentation of mobile phones and how to use them
• Maintenance
• Application on the phone to randomisation
16:45-17:00 Avaliação das sessões
Evaluation of sessions
28 de Setembro: Dia 2
8:00-8:30 Revisão dos pontos-chave de Dia 1
Review of the Day key points 1
1 supervisor, 1 inquiridor
8:30-10:00 Seleção dos agregados familiares e dos entrevistados
Selection of households and respondents
Cenários e Situações hipotéticas INS, com dois
supervisores
10:00-10:15 Pausa Lanche
10:15-11:00 Técnicas de entrevista
interview techniques
Os participantes trabalham em conjunto para elaborar uma lista de todos os aspectos importantes e técnicas para implementar uma entrevista. Depois, dois supervisores fazem dois dramatizações (boas e maus) e os participantes a identificar e discutir os pontos fortes e áreas de melhoria.
Participants work together to draw up a list of all the important aspects and techniques to implement an interview. Then two supervisors make two dramatizations (good and bad) and participants to identify and discuss the strengths and areas for improvement.
ICF
11:00- 12:00 Como Iniciar o questionário
Apresentações no agregado
Consentimento informado Prática no telemóvel
Presentations on aggregate
• Informed consent
• Practice on your phone
Ezequiel + ICF
RAcEMozambiqueEndlineSurveyFinalReport 31
12:00-13:00 Revisão do Módulos do questionário (no telemóvel):
Identificação das crianças Antecedentes do(a)
cuidador(a)
Quiz Modules Review (on the phone):
• Identification of children
• Background to the (a) caregiver (a)
Para cada perguntas, deve-se ter:
Consenso de como dizer em língua local
Fazer revisão das respostas para cada pergunta e as respostas em língua local
For each question, you should have:
• Consensus as saying in local language
• Do review of the responses to each question and answers in local language
INS
13:00-14:00 Almoço
14:00-14:45 Prática notelemóvel (primeiros módulos)
Como revisar o questionário preenchido
Practice on your phone (first modules)
• Reviewing the completed questionnaire
Em pares Ezequiel
14:45-16:00 Revisão 2 do Módulos do questionário (no telemóvel):
Tomada de decisões Conhecimento dos APE
por parte do cuidador(a) Conhecimento dos
cuidadores
Revision 2 of the questionnaire modules (on the phone):
• Decision-making
• Knowledge of the EPAs by the caregiver (a)
• Knowledge of caregivers
SC
16:00-16:30 Prática no telemóvel
Practice on your phone
Ezequiel
16:30-17:15 Simulação (em grupos de 3: 1 supervisor, 1 inquiridor, 1entrevistado; cada grupo implementa um/dois módulos; anota observações/desafios para partilhar com o grupo)
Os voluntários fazem uma simulação de uma entrevista (a partir de chegar a casa hipotética, e o grupo identifica e discute os pontos fortes e áreas de melhoria)
ICF
RAcEMozambiqueEndlineSurveyFinalReport 32
Simulation (in groups of 3: 1 supervisor, 1 inquirer, one respondent; each group implements one / two modules; notes comments / challenges to share with the group)
17:15-17:30 Avaliação das sessões
29 de Setembro: Dia 3
8:00-8:30 Revisão dos pontos-chave de Dia2 1 supervisor, 1 inquiridor
8:30-9:30 Revisão 3 do Módulos do questionário:
Diarreia
INS
9:30-10:00 Prática no telemóvel Ezequiel
10:00-10:15 Pausa Lanche
10:15-11:15 Revisão 4 do Módulos do questionário:
Febre
ICF
11:15-11:45 Prática no telemóvel Ezequiel
11:45-13:00 Revisão 5 do Módulos do questionário:
Tosse com respiração rápida/ dificuldades de respiração
Prática no telemóvel
SC
13:00-14:00 Almoço
14:00-14:45 Revisão de documentos de apoio Cábula de seleccionar criança, Ficha de Controlo para inquiridores e supervisores, álbum de Medicamentos, Tabela de Idade das Crianças em Meses, Ficha de Listagem dos AF; anotação de observações pelos inquiridores
SC
14:45-16:15 Simulação (em grupos de 3: 1 supervisor, 1 inquiridor, 1entrevistado; cada grupo implementa um questionário inteiro
Supervisores praticam a observação com o Lista de Verificação para Controlo da Qualidade e dar retroinformação)
ICF
16:15-17:00 Como responder a cenários difíceis Trabalho em grupos INS
17:00-17:15 Avaliação das sessões
RAcEMozambiqueEndlineSurveyFinalReport 33
30 de Setembro: Dia 4
8:00-8:30 Revisão dos pontos-chave de Dia 3 1 supervisor, 1 inquiridor
8:30-9:00 Preparações para a prática no campo
SC
9:00-16:00 Prática Em equipas, realizar a selecção das famílias, a selecção dos entrevistados, o consentimento informado e entrevistas. Cada entrevistador vai ter pelo menos uma oportunidade de fazer a entrevista. Supervisores vão avaliar a implementação dos inquiridores com um formulário padronizado.
16:00-17:00 Revisão da prática no campo Cada supervisor e inquiridor compartilhará uma lição aprendida ou um desafio confrontada
INS +ICF
17:00-17:15 Avaliação do dia
1 de Outubro: Dia 5
8:00-10:00 Revisão das áreas de dificuldade durante a prática no campo
INS
10:00-10:15 Pausa Lanche
10:15-11:00 Teste Rapido de aprendizagem
11:00-11:45 Como enviar os questonários preenchidos
Ezequiel
11:45-12:30 Simulação ICF
12:30-13:30 ‘Troubleshooting’ (Resolução de problemas) no telemóvel
Ezequiel
13:30-14:30 Almoço
14:30-15:30 Selecção dos inquiridores de Quelimane e designações dos equipas de campo e planificação do trabalho de campo (e logística)
SC/INS
15:30-16:00 Tempo livre para conhecer os membros do equipe, fazer perguntas/prática no telemóvel
16:00-17:00 Jogo de conhecimentos ICF
17:00-17:15 Avaliação do dia e da formação
RAcEMozambiqueEndlineSurveyFinalReport 34
ANNEX F. ENDLINE SURVEY FIELDWORK SCHEDULE
Manica
Equipe 1
Equipe 2
Equipa 3
Equipas 1,2, 3 Pre teste
SUPERVISOR: Silvia; INQUIRIDOR: Antonio e Anastacia
SUPERVISOR:Albertina; INQUIRIDOR: Miquelima, Marcelo
SUPERVISOR: Veronica; INQUIRIDOR: Ana Rita, Paulino
WED THU FRI SAT SUN MON TUE WED THU FRI SAT SUN MON TUES
DISTRITO POSTO
ADMINISTRATIVO COMUNIDADE5‐Oct 6‐Oct 7‐Oct 8‐Oct 9‐Oct 10‐Oct 11‐Oct 12‐Oct 13‐Oct 14‐Oct 15‐Oct 16‐Oct 17‐Oct 18‐Oct
Gondola CAFUMPE CAFUMPE
Vanduzi MATSINHO DONGO
Gondola INCHOPE MANHERE
Gondola CHIPINDAUMWE NHAMAGOA
Gondola AMATONGAS MBIA MBOGUE
Gondola MACATE MUSSANGANHAZE NHOANE
Gondola MARERA GOJOMBE
RAcEMozambiqueEndlineSurveyFinalReport 35
Nampula
Distritos
Datas e locais 10/10/201
6 11/10/2016 12/10/2016 2016‐13‐10 2016‐14‐10 2016‐15‐
10 2016‐16‐
10 2016‐17‐
10 2016‐18‐
10 2016‐19‐
10 2016/20/1
0
Moma Viagem à Moma Cotocuane Cotocuane
Angoche Ntupo –Naprume
Ntupo –Naprume Zoro Zoro Zoro Zoro
Mogovolas
Viagem à Mogovola
s Nanhoth
o Nanhoth
o
Erati Viagem à Erati Zulume Zulume
Nacaroa Cabo Juliano Cabo Juliano Mavalane Mavalane
Muecate Viagem à Muecate Napacala Napacala
Malema Viagem à Malema Murrumbo Murrumbo
Ribaue Namarrepo Namarrepo
Murrupula
Muarrapaz
Muarrapaz
Rapale Viagem à Rapale Tchaiane Tchaiane
Memba Viagem à Memba
Comunidade de Nehequene
Comunidade de Nehequene
Comunidade de Mangane
Comunidade de Mangane Sirissa Sirissa
Mossuril Namuco Namuco
Mogincual Chapueia
RAcEMozambiqueEndlineSurveyFinalReport 36
Zambezia
Inhambane
Equipa Inquiridores Supervisor Distritos 9/10/2016 10/10/2016 11/10/2016 12/10/2016 13/10/2016 14/10/2016 15/10/2016 16/10/21016 17/10/2016 18/10/2016 19/10/2016 20/10/2016 21/10/2016 22/10/2016 23/10/20106 24/10/21016 25/10/2016 26/10/2016 27/10/2016 28/10/2016 29/10/2016 30/10/2016
5Baciao Dinis & Carlos Marcelo Neves
Zarina Jose CussinhoAlto Molocue e
InhassungePre test
Travel
preperations Travel
Alto Molocue/
Yeheiya
Alto Molocue/
Yeheiya
Alto Molocue/
Yeheiya
Alto Molocue/
Yeheiya Rest day
Ferry Boat
under repair,
team waiting
to cross river
for Inhassunge
Ferry Boat under
repair, team
waiting to cross
river for
Inhassunge
Inhassunge/
Mussangano
Inhassunge/
Mussangano
Inhassunge/
Mussangano
Inhassunge/
Mussangano
Inhassunge/
Mussangano
Inhassunge/
Mussangano
Inhassunge/
Mussangano and
travel to
Quelimane
6Amilcar Salvador Mole &
Vania SalvadorAntonio Cipriano Calia
Alto Molocue e
GilePre test
Travel
preperations Travel Gile/Uataria Gile/Uataria Gile/Uataria Gile/Uataria Rest day
Gile/
Puampuela
Gile/
Puampuela
(substituted by
Quirule)
Gile/Quirule Gile/Quirule Gile/QuiruleGile/Quirule &
Gile/Uaturia
Travel to Alto
Malocue/
Guilherme
Alto Malocué/
Guilherme
Alto Malocué/
Guilherme
Alto Malocué/
Guilherme
Alto
Malocué/
Guilherme
Alto Malocué/
Guilherme
Alto
Malocué/
Guilherme
Travel to
Quelimane
7Rosario da Silva Valia & Valdimiro Albino Xavier
Anelita Assane Milange &
LugelaPre test
Travel
preperations Travel
Lugela/ Comone
(substituted with
Mpemula)
Lugela/
Mpemula
Lugela/
Mpemula
Lugela/
MpemulaRest day
Travel to
Milange
Milange/
Muriamuendo
Milange/
Muriamuendo
Milange/
Muriamuendo
Milange/
Muriamuendo
Milange/
Muriamuendo
Milange/
Muriamuendo
Milange/
Muriamuendo
Milange/
Muriamuendo
Team waiting
for data check
Team
waiting for
data check
Travel to
Quelimane
Equipa Inquiridores Supervisor Distritos 7/10/2016 8/10/2016 9/10/2016 10/11/2016 11/10/2016 12/10/2016 13/10/2016 14/10/2016 15/10/2016
11Fatima Herminio
Matandalasse & Jorge
Mario Binguane
Merton Andre Mueeombe
Panda & Zavala Pre testPanda/
Polana
Panda/
Polana
Panda/
Polana
Panda/
Polana &
Travel to
Zavala
Zavala/
Nhamble
Zavala/
Nhamble &
return to Polana
to re do
interviews with
errors
Panda/
Polana
Travel to
Inhambane
12Samuel Saranga & Nace do Crescencio Rungo
Natercia Martins Langa Panda & Zavala Pre testPanda/
Polana
Panda/
Polana
Travel to
Zavala
Zavala/
Nhamble
Zavala/
Nhamble
Zavala/
Nhamble
Zavala/
Nhamble
Travel to
Inhambane
RAcEMozambiqueEndlineSurveyFinalReport 37
ANNEX G. DETAILS OF DATA CLEANING AND ANALYSIS
The dataset submitted to ICF had a number of errors that required extensive review, additional cleaning, and some reconstruction of data. This is due to two major errors in the field. First, some enumerators conducted interviews with caregivers who did not have a child who had been sick in the two weeks prior to the survey; these caregivers were asked the caregiver-specific modules but not the sick child modules despite not being selected for the survey. Second, many enumerators did not record the cluster number, household number, caregiver number, or child number correctly in each questionnaire module as they moved through the survey, meaning that ICF had to go back through and manually correct these to ensure that questionnaire records were correctly linked. The following is a summary of steps taken to clean the RAcE Mozambique endline survey data:8
Manually added sex, age, and two-week illness history information for children missing this information, when available.
Manually removed records in the Excel dataset that did not fall within valid data collection dates (e.g., practice interviews).
Manually updated incorrect cluster numbers, household numbers, child line numbers, and caregiver numbers in the Excel dataset, across all modules.
Manually deleted duplicate records. Corrected the child line numbers in Module 1, the listing of selected children, to align with their
line numbers in the household child roster. This was done in part manually in the Excel dataset and in part via a Stata RAcE Moz Endline Cleaning do file.
Corrected child line numbers in the diarrhea, fever, and fast breathing modules to align with the line number in Module 1. This was done in part manually in the Excel dataset and in part via a Stata RAcE Moz Endline Cleaning do file.
Dropped records for which there were no sick children included in the survey.
The following is a summary of errors and corrections made to baseline survey data during comparative analyses at endline:
In 2014, caregiver knowledge indicators were calculated using a denominator of 625, which included caregivers without a sick child included in the survey. In 2017, revising these data for comparison with endline data, any caregivers without sick children included in the survey were excluded. This reduced the baseline denominator to 527. For the caregiver knowledge of CCM-trained APE role indicator, the denominator was changed to 327 because only caregivers who reported having an APE in their community were asked the associated survey questions.
For the first source of care indicators, caregivers who sought care from an APE were incorrectly included in the numerator in the 2014 analysis, whether or not they sought care from an APE first. This was corrected in 2017 by removing caregivers who did not seek care from an APE first from the numerator.
8 A log of all of the manual updates is contained in “Mozambique endline survey dataset update log 12.21.2016.xlsx.” A list of modules available by cluster and household in the final cleaned dataset is contained in “Mozambique Cluster Cleaning Workbook 12-16-2016.xlsx.”
RAcEMozambiqueEndlineSurveyFinalReport 38
In 2014, the fever treatment coverage indicators did not account for the “prompt” component—the same or next day—in the calculation. In 2017, this indicator was recalculated to include only those who also sought care “promptly” or “within same or next day.”
In 2014, 53 caregivers for whom zinc treatment information was missing were included in the denominators for diarrhea treatment coverage. These caregivers were omitted in the updated analysis in 2017.
In 2014, the first dose of treatment from an APE indicator was calculated with all sick children included in the survey as the denominator. This was corrected in 2017 by including only sick children who received treatment from an APE in the denominator.
In 2017, the sick-child follow-up indicator was corrected to specify whether an APE visited the child for follow-up regardless of timeline. In 2014, the calculation included any follow-up within three days.
RAcEMozambiqueEndlineSurveyFinalReport 39
ANNEX H. KEY INDICATORS FOR FULL PROJECT AREA
The following are key indicators for the sample of 32 clusters representing the full project area across all four provinces: Inhambane, Manica, Nampula, and Zambezia.
Indicator Endline% (CI %)
End of Project Sample: Inhambane, Manica, Nampula, ZambeziaCaregiver Knowledge
1
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who are aware of the presence of the CCM-trained APE in their community
93.9 (84.4 - 97.7)
2
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who know the role of the CCM-trained APE in their community
68.1 (59.0 - 76.1)
3
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who know two or more signs of childhood illness that require immediate assessment by an appropriately trained provider
92.4 (87.3 - 95.5)
Caregiver perceptions of iCCM services
4
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who view CCM-trained APEs as trusted health care providers
78.4 (71.1 - 84.2)
5
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who believe CCM-trained APEs provide quality services
77.4 (71.4 - 82.5)
6
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who found the CCM-trained APE at first visit
83.0 (75.1 - 88.8)
7
Percentage of caregivers of children age 2-59 months who have been sick in the two weeks preceding the survey who cite the CCM-trained APE as a convenient source of treatment
81.7 (73.1 - 88.0)
Sick Child Care-seeking
8
Percentage of children age 2-59 months who have been sick in the two weeks preceding the survey for whom advice or treatment was sought from an appropriate provider
Overall78.8
(72.8 - 83.7)
Fever78.0
(71.5 - 83.3)
Diarrhea76.5
(68.2 - 83.1)
Fast breathing82.0
(75.3 - 87.2)
9
Percentage of children age 2-59 months who were sick in two weeks preceding the survey taken to a CCM-trained APE as first source of care
Overall56.5
(48.5 - 64.2)
RAcEMozambiqueEndlineSurveyFinalReport 40
Fever55.1
(46.2 - 63.6)
Diarrhea59.3
(50.1 - 67.9)
Fast breathing55.2
(47.0 - 63.1) Sick Child Assessment
10 Percentage of children age 2-59 months with fever in the two weeks preceding the survey who had finger or heel stick
55.0 (41.8 - 67.6)
11
Percentage of children age 2-59 months for whom their caregiver received the results of the malaria diagnostic test of the children who had had finger or heel stick in the two weeks preceding the survey
93.9 (88.2 - 96.9)
12
Percentage of children age 2-59 months with cough and difficult or rapid breathing (suspected pneumonia) in the two weeks preceding the survey who had their respiratory rate counted to assess fast breathing
38.2 (29.1 - 48.1)
Sick Child Assessment by APE
13
Percentage of children age 2-59 months with fever in the two weeks preceding the survey who had a finger or heel stick by an APE (among those who sought care from an APE)
55.0 (41.8 - 67.6)
14
Percentage of children age 2-59 months for whom their caregiver received the results of the malaria diagnostic test of the children who had a finger or heel stick by an APE in the two weeks preceding the survey (among those who sought care from an APE)
96.2 (90.5 - 98.5)
15
Percentage of children age 2-59 months with cough and difficult or rapid breathing (suspected pneumonia) in the two weeks preceding the survey who had their respiratory rate counted to assess fast breathing by a CHW (among those who sought care from an APE)
39.2 (27.9 - 51.8)
Sick Child Treatment
16
Percentage of children age 2-59 months who have been sick in two weeks preceding the survey who received appropriate treatment
Overall49.6
(42.9 - 56.3)
Malaria (ACT)*72.1
(61.9 - 80.5)
Diarrhea (ORS and Zinc)31.2
(22.3 - 41.7)
Fast breathing (Amoxicillin)58.0
(50.7 - 65.1)
17
Percentage of children age 2-59 months who have been sick in the two weeks preceding the survey who received appropriate treatment from a CCM-trained APE
Overall29.2
(23.0 - 36.2)
Fever36.2
(25.8 - 48.1)
Diarrhea22.9
(15.5 - 32.5)
RAcEMozambiqueEndlineSurveyFinalReport 41
Fast breathing32.5
(25.8 - 40.0)
18
Percentage of children age 2-59 months who have been sick in the two weeks preceding the survey who received the first dose of treatment in the presence of an APE among those who received prescription medicines for a CCM condition in the two weeks preceding the survey
Overall32.2
(23.5 - 42.3)
Fever26.1
(16.6 - 38.6)
Diarrhea16.0
(6.3 - 35.0)
Fast breathing49.5
(36.5 - 62.6)
19
Percentage of sick children age 2-59 months for whom their caregivers received counseling on how to provide the treatment(s) among those who received prescription medicines for a CCM condition in the two weeks preceding the survey
Overall96.3
(92.5 - 98.2)
Fever98.9
(92.0 - 99.9)
Diarrhea96.0
(87.3 - 98.8)
Fast breathing94.2
(86.2 - 97.7) Sick Child Referral and Follow-up
20 Percentage of sick children age 2-59 who were referred in the two weeks preceding the survey whose caregiver adhered to referral advice
65.6 (55.7 - 74.4)
21
Percentage of sick children age 2-59 months receiving treatment from an APE in the two weeks preceding the survey who received a follow-up visit from an APE according to country protocol
41.0 (32.5 - 50.2)
22
Percentage of sick children age 2-59 months receiving treatment from an APE in the two weeks preceding the survey whose caregiver followed-up with an APE
9.7 (5.7 - 16.0)
*Numerator: Cases of fever with a positive blood test who received ACT within 24 hours; Denominator: cases of fever with a positive blood test.
RAcEMozambiqueEndlineSurveyFinalReport 42
ANNEX I. SUMMARY OF FINDINGS FROM RAcE MOZAMBIQUE ENDLINE APE SURVEY
INTRODUCTION As part of the endline evaluation, Save the Children conducted a survey of APEs to assess the implementation strength and quality of the iCCM services they delivered. The APE survey was conducted with the endline household survey to assess care-seeking practices and treatment coverage for iCCM conditions. The objective of the APE survey was to gain a better understanding of the APEs’ background characteristics, activity levels, and support and supervision received to help interpret the results of the coverage survey. The APEs serving the 32 clusters selected for the endline household survey formed the sample population for the APE survey, and all were interviewed. Data collection was carried out during October–November 2016. RESULTS A detailed indicator table is provided in Table A. The main findings are summarized below. APE profile: Most sampled APEs were male (78 percent), and the majority had seven or more years of education. About two-thirds of APEs were aged 40 years and younger. As a result of the household survey sampling design, the APEs interviewed were primarily from Nampula province, with only small numbers sampled in Inhambane, Manica, and Zambezia. APE catchment areas: The majority of APEs (91 percent) resided in their catchment areas. The main modes of transportation were walking (59 percent) or bicycle (31 percent), and most APEs reported that it took one or more hours to reach the nearest health post. Catchment areas were geographically large, with 34 percent of APEs reporting traveling more than 2 hours and 41 percent travelling 1–2 hours to reach the farthest house within their catchment areas. Supervision: Most APEs (84 percent) reported receiving a supervisory visit in the last 3 months. District coordinators9 and health facility supervisors were the most frequently mentioned providers of supervision. Among APEs who reported supervision, the majority reported that the supervisors used a supervision checklist and that records and kit and medical supply availability were reviewed. A smaller percentage of APEs mentioned that the supervisor had observed sick child care or talked with village leaders. About two-thirds of APEs (68 percent) reported receiving clinical supervision in which sick child care was observed in the last 3 months. Availability of iCCM medicines and diagnostics: Only 44 percent of APEs had at least one age formulation of first-line antimalarial drugs (Coartem; artemether-lumefantrine 1x6 or 2x6) in stock on the day of observation. Most APEs were observed to have amoxicillin (81 percent), ORS (91 percent), zinc (81 percent), and paracetamol (81 percent) in stock. Three-quarters (75 percent) of APEs had RDTs, and 78 percent had a functional stopwatch. Overall, only 19 percent of APEs had all iCCM medicines and supplies available on the day of the survey, and 37 percent had all essential iCCM medicines and supplies.10 Stockouts were common, particularly for artemether-lumefantrine, RDTs, and amoxicillin. Only 28 percent of APEs reported that they had continuous stock of artemether-lumefantrine (1x6 or 2x6), 44 percent for RDTs, and 53 percent for amoxicillin in the month before the
9 There were both RAcE district supervisors and MOH district supervisors providing supervision at time of the study, but the questionnaire did not capture information to distinguish between the two. 10 Artemether-lumefantrine (at least 1x6 or 2x6), amoxicillin, ORS, zinc, RDTs, and functional timer
RAcEMozambiqueEndlineSurveyFinalReport 43
survey. Overall, only 16 percent reported no stockouts of essential iCCM medicines and RDTs in the past month. APE functionality: All APEs had provided iCCM services in the past month and reported providing iCCM services at least two days per week. However, only 41 percent met a stricter definition of “functionality,” in which they resided in their catchment areas, provided iCCM services in the last month, and reported providing iCCM at least 5 days per week. Activity levels and record-keeping: All APEs reported providing iCCM services at least 2 days per week, and nearly half (47 percent) reported providing iCCM 5 or more days per week. Register reviews indicated that in the last month, APEs treated an average of 16 malaria cases (range 0–70), 7.9 diarrhea cases (range 0–28), and 7.7 pneumonia cases (range 0–35). Overall, APEs treated an average of 31.6 iCCM cases per month (median 32.0; range 0–109). The majority (71 percent) of APEs had referred one or more sick children for danger signs in the last month. Completeness of recording of cases in APE registers was lacking, with 77 percent of APEs recording RDT results for the 5 most recent fever cases, and 71 percent with respiratory rates recorded for the 5 most recent cough cases. APE knowledge of danger signs and iCCM protocols: APE knowledge of danger signs for which they should refer a sick child was low, with only 13 percent able to cite all 4 of the most serious or general signs (vomiting everything, convulsions, not conscious/lethargy, and not able to drink/breastfeed). Three-quarters of the APEs (75 percent) said that they would count breaths per minute to assess for fast-breathing, and 69 percent said that they would ask whether the child had difficult or rapid breathing. About 53 percent said that they would ask about cough, and 19 percent would check for fever. All but two APEs (94 percent) correctly indicated that they would provide amoxicillin to a child determined to have pneumonia, and 91 percent gave the correct response for duration of treatment of malaria with Coartem (3 days). APE data use: Nearly 60percent of APEs (19 of 32) reported being trained in data use, and 47 percent were observed to have completed charts with data filled out for the past month. About 13 percent (4 of 32) had the chart books but did not fill them in. IMPLICATIONS These findings have several implications for the iCCM program in Mozambique:
District coordinators were the most frequently mentioned provider of supervision; however it was not possible to distinguish between RAcE-funded district coordinators and MOH district coordinators. As the RAcE program transitions activities to the MOH, the RAcE-funded district coordinator position will be discontinued, but the MOH district coordinators will continue. Further efforts will be needed to strengthen the capacity of health facilities to provide regular supervision for APEs. The MOH has already revised the policy to recommend quarterly supervision (previously supervision was monthly) to help make supervision more feasible; however, the lack of financial resources for transport and the distance to some remote APEs may continue to present challenges to reaching all APEs with routine supervision. Save the Children purchased 40 new motorcycles for supervision visits and will donate the existing motorcycles to the MOH. UNICEF also purchased 250 motorcycles for the APE program. Issues will continue to be the following: (1) fuel, maintenance, and per diem costs for the supervision visits; (2) prioritization of the motorcycles for the supervision visits when there is such a transportation need in the health facilities (3) time for the supervisions by the health facility supervisors because there is still a lack of medical staff at health facilities.
RAcEMozambiqueEndlineSurveyFinalReport 44
APEs cannot provide quality iCCM services without adequate and continuous stocks of medicines. Less than half of APEs had any artemether-lumefantrine available on the day of the survey, and nearly three-quarters reported stockouts of artemether-lumefantrine in the past month. Stockouts of RDTs and other essential iCCM medicines were also common. Medicines are still provided to APEs based on a kit system, in which supplies are not determined based on consumption. Mozambique has faced widespread shortages of antimalarial drugs and has struggled to maintain supplies at first-level health facilities. APE stocks have frequently been compromised. Systemic strengthening of the supply chain is urgently needed in Mozambique to avoid ongoing stockouts at the APE level, and there is a need to manage consumption data.
Most APEs were living in their catchment areas and providing iCCM on a regular basis, with nearly half providing iCCM at least five days per week. Activity levels were relatively high, with APEs treating more than 30 iCCM cases per month. Completeness of registers for recording of RDT and respiratory assessment results could be improved through reinforcement during supervision and data use training.
The knowledge of general danger signs and of assessment for pneumonia cases was markedly lower than in previous surveys of APEs. Only 13 percent of APEs interviewed could cite all 4 general danger signs, compared to 97 percent of APEs interviewed as part of the endline survey for the CIDA-funded iCCM program in Nampula in 2012. Reasons for these lower knowledge results are unclear. Under the previous program, supervision levels were higher and often provided through Save the Children staff with clinical training, and they may have focused more on reinforcing clinical skills. A quality of care study conducted in January 2016 found that only 25 percent of sick children assessed by APEs were assessed for all 4 general danger signs, providing further evidence of knowledge and practice gaps for general danger signs.
Table A. Summary of indicators for iCCM service delivery by APEs (n=32) Domain Indicator ResultResidency % of CCM-trained APEs residing in their catchment area 91% (73–97%) Functionality % of CCM-trained APEs who are functional (MOH definition):
i) have provided iCCM services in past month ii) report operating village clinic for at least two days/week
100%
% of CCM-trained APEs who are functional (stricter definition): i) reside in their catchment area ii) have provided CCM services in the past month iii) reports operating village clinic for at least five days per week
41% (25–59%)
Medicine and diagnostics availability
% of APEs with all key CCM medicines and diagnostics in stock on day of assessment (artemether-lumefantrine, amoxicillin, ORS, zinc, RDTs, timer)
38% (20–55%)
Artemether-lumefantrine (1x6 or 2x6) 44% RDTs 75% Amoxicillin 81% ORS 91% Zinc 81% Timer 78% Paracetamol 81% Graphic notebook (data use) 56%
% of APEs with all medicines and diagnostics in stock on the day of assessment (all above medicines and supplies)
19% (8–37%)
% of APEs reporting no stockouts of essential iCCM supplies lasting seven days or more in the month before the survey (artemether-lumefantrine 1x6 and 2x6, RDTs, amoxicillin, ORS, zinc)
16% (6–37%)
Artemether-lumefantrine (1x6 and 2x6) 28% RDTs 44% Amoxicillin 53% ORS 75% Zinc 63%
RAcEMozambiqueEndlineSurveyFinalReport 45
Domain Indicator ResultSupervision % of APEs who received at least one supervision session during the prior
three months during which registers were reviewed 84% (66–94%)
I. Reviewed treatment records 81% II. Checked medical supply quantities 75%
III. Used supervisory checklist 78% IV. Observed care/treatment for sick child 63% V. Talked to village leaders 47% VI. Talked to you about clinical activities 72%
% of APEs who received at least one mentorship session during the prior three months with observation of case management (clinical supervision)
69% (50–83%)
Service availability and activity levels
% of APEs who report typically operating their village clinic: i) <2 days/week 0% ii) 2+ days/week 100% iii) 5+ days per week 47% Number of days APEs report operating village clinic per week Mean: 4.5
Median: 4 Number of sick child cases treated in the last one month Mean: 31.6
Median: 32.0 Range: 0–109
I. Malaria cases Mean: 16.0 Range: 0-70
II. Suspected pneumonia cases Mean: 7.7 Range: 0-35
III. Diarrhea cases Mean: 7.9 Range: 0-28
% of APEs who referred one or more sick child cases for danger signs in the last 1 month
71% (52–85%)
Recording completeness
% of APEs with complete recording for use of RDTs for five most recent cases presenting with fever
77% (58–89%)
% of APEs with complete recording of RR for five most recent cases presenting with cough/difficult breathing
71% (52–85%)
APE knowledge
% of APEs who can cite all four general danger signs 13% I. Convulsions 84%
II. Lethargic or unconscious 22% III. Cannot eat or drink 63% IV. Vomits everything they eat or drink 66%
% of APEs who know how to classify a case of pneumonia (all four steps) 9% I. Ask if child has cough 53%
II. Count child's respiratory rate 75% III. Check if child has fever 19% IV. Ask if child has difficult or rapid breathing 69%
% of APEs who can cite the correct duration of treatment for artemether-lumefantrine 91% (73–97%)
Data display and use
% of APEs who were trained on the data use package 59% (41–76%) % of APEs trained in the data use package (n=19) who had template filled with data for last one month 79%