The status of Ghanaian community health workers’ supervision and service delivery: descriptive analyses from the 2017 Performance Monitoring and Accountability 2020 survey

Introduction: Community-based services are a critical component of high-quality primary healthcare. Ghana formally launched the National Community Health Worker (CHW) program in 2014, to augment the pre-existing Community-based Health Planning and Services (CHPS). To date, however, there is scant data about the program’s implementation. We describe the current supervision and service delivery status of CHWs throughout the country. Methods: Data were collected regarding CHW supervision and service delivery during the 2017 round of the Performance Monitoring and Accountability 2020 survey. Descriptive analyses were performed by facility type, supervisor type, service delivery type, and regional distribution. Results: Over 80% of CHWs had at least monthly supervision interactions, but there was variability in the frequency of interactions. Frequency of supervision interactions did not vary by facility or supervisor type. The types of services delivered by CHWs varied greatly by facility type and region. Community mobilization, health education, and outreach for loss-to-follow-up were delivered by over three quarters of CHWs, while mental health counseling and postnatal care are provided by fewer than one third of CHWs. The Western region and Greater Accra had especially low rates of CHW service provision. Non-communicable disease treatment, which is not included in the national guidelines, was reportedly provided by some CHWs in nine out of ten regions. Conclusions: Overall, this study demonstrates variability in supervision frequency and CHW activities. A high proportion of CHWs already meet the expected frequency of supervision. Meanwhile, there are substantial differences by region of CHW service provision, which requires further research, particularly on novel CHW services such as non-communicable disease treatment. While there are important limitations to these data, these findings can be instructive for Ghanaian policymakers and implementers to target improvement initiatives for community-based services.

country 10 . This program was designed to support the pre-existining community health programs that had been built to date 10 .
In order to address these challenges, a new cadre of health worker, was introduced in the National 1 Million CHW Program 10 . These CHWs are fully-employed workers, with a salary of approximately $142 USD per month, under the auspices of the Youth Employment Agency 10 . According to the program design, these CHWs report directly to the CHOs, supporting them to provide first-level health care throughout the communities. Detailed descriptions of the CHW roles and responsibilities are included in Table 3. The program set a goal of deploying over 31,000 CHWs throughout the country between 2014 and 2023 10 . By the end of 2019, the goal is to have achieved full rural coverage of the CHW program, involving approximately 28,000 CHWs. As of July 2019, approximately 26,000 have been trained and deployed, the distribution of which can be viewed on the program's online data dashboard 11 and coverage map 12 .
CHWs are expected to spend 80% of their time in the community, providing these services via household visits. Per the program guidelines, the CHWs are intended to support the CHPS work, and are not supposed to be specifically attached to any hospitals 8, 10 . In practice however, after the program's initiation in 2014, anecdotal evidence suggests that many CHWs have been functionally reporting to, or interacting with, facility managers at hospitals.
To ensure the quality of their work, CHWs are expected to meet with their CHO supervisors at least quarterly and also interface with the CHVs during the course of their work, especially in the context of organizing community health-related gatherings and educational campaigns 8,10 .
While the policies for training, supervision, and the responsibilities of CHWs are clearly delineated 10 --including twentyeight weeks of pre-service training and one week update training twice yearly --there is a paucity of data describing the current state of CHW service scale-up across the country, including how the CHWs' work relates to the work of the CHOs and CHVs. Given the extensive efforts that have gone into strengthening community-based health services in Ghana, understanding the present status of CHW services is important for policy makers and program implementers to target improvement initiatives for the future.
Here, we present data describing the supervision and activities provided by CHWs throughout the country. These data were collected from the facility surveys done as part of the 2017 round of the Performance Monitoring and Accountability 2020 (PMA2020) national survey 13 . Given the anecdotal evidence that some CHWs were directly interacting with hospitallevel facilities, the survey asked these questions at all facility types, to best characterize the landscape of CHW work nationally.

Survey
The PMA2020 survey is a nationally representative, rapidturnaround cross-sectional survey of family planning indicators among women of reproductive age (ages 15-49), and

Amendments from Version 2
We are grateful to the Dr. Ballard and Dr. Perry for their insightful peer-review comments. In this updated version, in response to their feedback, we have provided additional details regarding the community health worker training and salary structures; contextualized the community health officers as a type of community health worker; provided more detailed discussion about the supervision of community health workers at hospitals; discussed the implications of the 2016 guidelines in greater detail; included additional citations for some of the contextual data regarding the community health worker program in Ghana; and added to the limitations section describing survey biases in more detail. We have also included some clarifying text regarding the 2014 launch of the National 1 Million CHW program, which is distinct from other prior community health-related programs. Finally, we have corrected several prior typographical and grammatical errors.
No data, analyses, or figures have been added or changed. No changes have been made to authorship. We have no competing interests in regards to these updated changes, nor to the original content of the manuscript.

Introduction
As the world strives to achieve Universal Health Coverage and the Sustainable Development Goals, primary healthcare is foundational to meeting these goals 1,2 . Community healthcare systems serve critical roles within strong primary healthcare delivery 2-4 . The World Health Organization's recent guidelines 5 for best practices of community health workers (CHWs) offer important guidance to policy makers and program implementers about how to develop strong community health service delivery and support low-and middle-income countries along the path towards universal health coverage. Among other key recommendations, these guidelines highlight the importance of professionally-trained CHWs with clear roles and responsibilities, supported by strong supervision systems to ensure quality service delivery 5 .
Ghana has a strong history of high-quality community-based primary healthcare delivery, including the development of the Community-based Health Planning and Services (CHPS) in 1994 6 , with significant expansion and strengthening of those services over the past 25 years. In recent years, the Ghana Health Service has developed a set of 15 steps and six milestones to guide CHPS implementation across the country 7,8 . CHPS service delivery is based on the deployment of Community Health Officers (CHOs) throughout the country in CHPS zones. These CHOs -a type of community health worker in and of themselves 9 --work closely with the Community Health Volunteers (CHVs), who are responsible for home visits, community mobilizations, participation in health outreach services with the CHOs, and household health education 8 . More detailed descriptions of the roles and responsibilities of CHOs and CHVs are provided in Table 1 and Table 2. In 2014, in conjunction with the global One Million Community Health Workers Campaign, the government of Ghana formally launched the National 1 Million CHW Program, with the goal of expanding high-quality community health services throughout the

ANC
History taking, identification and management of anemia, malaria in pregnancy, syphilis in pregnancy, implementation of PMTCT activities, counselling pregnant women based on findings, and teaching danger signs in pregnancy 14 Safe emergency delivery and newborn resuscitation Immediately assess mother, prepare for delivery, monitor labor, deliver baby, resuscitate if baby is not breathing well, and conduct active management of the third stage of labor.

15
Postnatal care (PNC) and essential newborn care Conduct immediate PNC to mother and baby, educate family on PNC, assess baby and mother at 6 weeks. water, sanitation, and hygiene indicators among households, in 10 countries 13 . Using a two-stage cluster design, households were selected to estimate the national modern contraceptive prevalence rate within 3%. In order to better understand access to family planning and primary health care in these countries, data were also collected on health care facilities where women received care. The methods used to collect data from health facilities in the PMA2020 survey have been described in detail elsewhere 13 . Briefly, health care facilities in each enumeration area were surveyed by trained enumerators, who used mobile data collection technology to interview the heads of facilities and upload the data into a secure cloud server. Data is uploaded as direct responses to the survey tool, as described elsewhere 13 . We analyzed the PMA2020 survey data collected in Ghana from September 2017 to November 2017 in the 100 enumeration areas surveyed throughout the country 14 .

A3. Other clinical services
In each enumeration area, a census of the public health facilities that serve the enumeration area was conducted to populate the list of survey facilities. Since the survey focused on the primary level of care, the district hospital that serves as the referral facility for all the surveyed facilities was also studied. Facilities of different sizes and levels, from CHPS facilities to health centers and hospitals, were selected to be included in the overall PMA2020 survey sample with the intent to represent the variety of available health facilities in each enumeration area, which are utilized by the nationally representative sample of women of reproductive age.
We explored several aspects of CHW service delivery in Ghana. The PMA2020 survey collected data on whether facilities supported CHWs with supervision and/or supplies (yes/no), what type of facility was reporting CHW data (CHPS/health center/ hospital), who at the facilities supervised the CHW (community health officer/public health nurse/midwife/health assistant/ physician assistant), and how frequently the CHW was supervised. Frequency of supervision was categorized as days between supervision interactions. If "monthly" was reported, that was categorized numerically as every 30 days.
We also investigated the different types of activities CHWs were involved in, and how these varied by facility type and region.
Supervisors were asked about activities and services offered by CHWs from their facility, in reference to CHW activities as defined in the National 1 Million CHW Program documentation 10 .
While not included in the expected scopes of work for CHWs, we also investigated non-communicable disease treatment as a key priority area for potential future service expansion 8,10 . All data analyzed had been collected as part of the PMA2020 survey, using the methods previously described.

Data analyses
Analyses were conducted using descriptive statistics and figures to report on facility-reported supervision and activities of CHWs within the survey. To assess central tendencies and distributions of CHWs and how frequently they were supervised across different facility types we calculated medians, standard deviations (SD), and interquartile ranges (IQRs) by each facility type. We also calculated counts and percentages to determine who supervised CHWs at each facility type, as well as how frequently they were supervised by each facility and supervisor type. Finally, we examined the types of activities CHWs were performing by examining counts and percentages of each activity by facility type and region and created a heat map based on frequency of each activity. As the purpose of this study was descriptive rather than inferential, no null hypothesis testing was conducted. Any missing data are noted in the data tables. All study participants provided informed, written consent.
Nationally, there were more CHWs supervised on a perfacility basis at the hospital and health center levels than the Table 4. Regional distribution of facilities supporting community health workers (CHWs) included in the PMA2020 survey.    CHPS facilities (median number of CHWs per facility: 20, 10, and 4, respectively) ( Table 2). Most CHWs were supervised by CHOs at health centers and CHPS facilities (74% and 78%, respectively), while hospital-based CHW supervision was managed by both CHOs (38%) and Public Health Nurses (62%) ( Table 5).

CHPS Total
Nationally, there was considerable variability in the frequency of supervision interactions between CHWs and their supervisors, and these data show that the majority (55.8%) of CHWs interacted with their supervisors approximately once per month ( Table 6). An additional 25.6% of CHWs interacted with their supervisors more than once per month, meaning than over 80% of CHWs described in these data had at least monthly supervision interactions ( Table 6). The frequency of interactions Table 7. Frequency of community health worker (CHW) supervision interactions by facility and supervisor types. did not seem to vary substantially by facility or supervisor type. CHWs based at hospitals, health centers, and CHPS all interacted with their supervisors at approximately the same frequency (median number of days between interactions: 30, 30, and 30, respectively) ( Table 7). The frequency of supervision interactions did not differ between types of supervisors (public health nurses, CHOs, midwives), with a median of 30 days between interactions for all supervisor types, except for the single Health Assistant supervisor included in the sample (7 days) ( Table 7).

Number of days between supervision of CHWs by facility
There was wide variability in the types of services delivered by CHWs, by both facility type and region, as described in Table 8 and Table 9. Of the activities that are expected to be  delivered by CHWs according to the National 1 Million CHW Program policies 10 , some services, such as community mobilization, health education, and outreach for loss-to-follow-up, were delivered by over three-quarters of all CHWs (Table 8). In contrast, other services, such as mental health counseling and postnatal care were much less common, being delivered by less than one third of CHWs nationally. Notably, while not included in the expected scope of work by national guidelines, 22.4% of CHWs were reported to be providing non-communicable disease treatment services. Regionally, there was great variation in service delivery, with some services, such as active case finding or immunizations, being delivered by all CHWs in one region but not delivered by any CHWs in other regions (Table 9).

Discussion
In Ghana, where there is a long-standing commitment to quality community-based primary healthcare, the 2014 National 1 Million CHW program was designed to strengthen the pre-existing community-based service provision. To date, however, there is scant data to understand the success of the program implementation. We have presented data that show variability in both supervision and the CHW activities provided across the country. Additionally, these data show very clearly that, while not designed to be posted to hospitals, hospital-supervised CHWs are common across the country. The details of these data offer several important insights to program implementers and policy makers for the future of strong community-based primary healthcare services in Ghana.
The variability in the frequency of supervision interactions between CHWs and their supervisors is notable, in light of national 10 and global 5,15 guidelines that aspire to consistent, frequent supervision systems for CHWs to ensure quality service delivery. The variability seems to be agnostic of facility type or supervisor type, and over 80% of the CHWs described here were reported to be interacting with their supervisors at least monthly, which is much more frequently than the quarterly goals set forth in the National CHW Program guidelines 10 . While more frequent supervision is likely beneficial, this reported variability in frequency of interactions offers a clear area for standardization throughout the program. Additionally, even amongst the CHW-supervisor pairs that are meeting national goals, it would be informative to investigate the ideal frequency of supervision in order to optimize limited resources.
Our data show considerable variability in the type of activities performed by the CHWs, and the degree of availability of each activity, across the different regions of the country. While this survey inquired about only a sample of the expected services included in the national guidelines 10 , it is clear that many expected activities are not yet being provided by CHWs, or only minimally provided in certain regions. Only three activities -community mobilization, health education, and outreach for loss to follow-up patients -were reported to be provided by the CHWs affiliated with more than three-quarters of surveyed facilities nationally, and even these were not universally available throughout all regions. Multiple other services that are included in the national guidelines, including antenatal care (ANC) counseling, community-based integrated management of childhood illness, immunization services, mental health counselling, and post-natal care, were reported to be provided by less than half of CHWs nationally, and far fewer in some regions. At the regional level, we also found variability in service provision, with some regions' facilities reporting much higher provision of CHW activities than others. In particular, the Western region reported especially low rates of CHW services provided, with all activities except family planning provision (88.9%) being provided by CHWs affiliated with less than half the facilities, and six expected activities being provided by no facility at all. The Greater Accra region also had lower provision rates of many activities, which may be related to differential implementation of the CHW program within the larger urban area, where services might be provided by other actors and facility types, unlike the more remote areas.
Our data show evidence of an expanded role for CHWs, beyond that specified in the national guidelines. All regions except the Greater Accra region reported CHW provision of noncommunicable disease treatment. While these data only describe what the facility managers reported, and thus cannot provide insights into the details of these non-communicable disease services, nor the technical quality of their provision, this is an important finding. Given that these are not included in the national CHW guidelines, this demonstrates that there is at least some implementation of novel service delivery throughout the country. Some of these activities may be provided in the context of local pilot programs or community-based programs, although our survey data are not specific enough to elucidate those details. Regardless, given that non-communicable diseases are priorities for the national health sector 8 , this finding warrants further investigation to better understand the feasibility of CHWs providing these services at a high level of quality, and planning for potential inclusion in the national program in a more standardized manner.
Finally, our data show that, in eight of the ten regions, at least some CHWs are supervised by CHOs who operate from hospitals. These CHOs have been assigned CHPS zones in which they work with the CHWs, as mentioned on the data summary page 12 . Given that the program is intended to support the CHPS work, and that the CHWs are supposed to spend more than 80% of their time in the community, this finding has important implications for the future of the program. Notably, it is plausible that the multiple types of community health cadres, with often-times overlapping or conflicting sets of job descriptions and service delivery guidelines, may have contributed to this phenomenon of CHWs being supervised by CHOs at hospitals. The new guidelines for CHPS were released in 2016 8 , which may help to clarify scopes of work among the different cadres supporting community health activities throughout the country.

Limitations
Our data have several important limitations. First, they are descriptive data only, which were collected in the process of the PMA2020 survey, which is not explicitly designed to study CHW activities. Thus, their level of detail is limited, and further investigation is required to better characterize and understand the aforementioned findings.
Second, these data are from facility manager reports, who may have limitations in their knowledge, which may impact the quality and accuracy of these data. Relatedly, it is not possible for us to determine what percentage of the entire CHW population is accurately reflected in these data; there may be many CHWs who are not in frequent contact with these managers and thus not well-represented by these data. Additionally, since these data are all from facility managers, who may have their own inherent biases, it is quite possible that some of these data represent over-estimates of CHW supervision and activities.
Third, given that the methodology of the PMA2020 sampling strategy is not designed around CHW staffing, the collected data may not be optimal in all regions of Ghana, and importantly do not reflect the new 16-region geographical distribution, which was expanded from the prior 10-region distribution in early 2019. The new 16-region geographical distribution can be seen on the Ghanaian Embassy site.
Finally, our survey inquired very specifically about "community health workers" during each facility survey, but given the multiple cadres involved in community health-related services throughout the country (including, for example, CHOs and CHVs 9 ), it is plausible that some survey respondents may have provided answers that were not exclusively about the CHWs affiliated with their facility. Thus, our data may represent information about other community health-related cadres in Ghana. Further research and program planning should include survey methods to more explicitly differentiate CHWs from the other cadres, to ensure that the correct conclusions are attributed to the appropriate cohort of health workers.

Conclusions
We have presented descriptive data summarizing the current status of CHW supervision and activities in Ghana. These data provide policy makers and program implementers helpful insights to inform targeted improvement initiatives throughout the country. Furthermore, these data can help to better inform ongoing monitoring and evaluation strategies of community health programming in Ghana. Other countries that utilize the PMA2020 survey methodology, or comparable survey methods, may consider using similar survey techniques, as described here, to better understand their national community health programming.

Data availability
Underlying data All data used in this study are available via the PMA2020 website. Per the data use guidelines of the PMA2020 databases, all PMA2020 datasets are free to download and use, although users are required to register for a PMA2020 dataset account. This is to ensure that data use can be appropriately tracked by the PMA2020 database managers. The request form must include a brief description of the research or analysis that the user would like to conduct using the requested data. If the research question is not clear, the database managers of PMA2020 may follow-up for further clarification. Once users are granted access, a zipped folder with the compressed dataset, brief user notes, and survey questionnaires will be made available to the user. All data sets will be de-identified. Users can download the codebooks as well.

Grant information
This work was supported by the Bill and Melinda Gates Foundation [OPP1149078].
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. and Zimbabwe. United States Agency for International Development. 2017. 1.

4.
A few small edits worth considering: The 2016 CHPS Policy notes that there had been confusion about the basic minimum service package due to conflicting messages from different levels of the health system. While the 2016 policy clearly defines a minimum package of services, it may be worth noting in text that this earlier confusion may be one reason for the variability noted and that the newly issued guidance may *potentially* have rectified some of this variability in the years between the survey and now.
In many countries, CHWs are only attached to the lowest level of care (e.g. community clinic), not district hospitals. It would be worth noting for the reader the policy in Ghana -are CHWs supposed to be attached to district hospitals or is that a quirk of implementation?
In the limitations section, it is noted that the quality and accuracy of the data may have suffered due to incomplete knowledge on the part of facility managers who provided it. Given the interests and responsibilities of the managers, would it not also be fair to consider the strong possibility of bias and potential that -if anything -the supervision frequency was overestimated rather than underestimated?
There are a few small typos: - Table 1: Incorrect bolding of #1.
- Table 1: Inconsistent capitalization in key tasks of item 3.
- Table 3: Inconsistent use of periods/full stops throughout the table.
-p. 12: Missing period/full stop "attributed to the appropriate cohort of health workers."