Assessment of Community Health Workers Activities in Zambézia Province, Mozambique

Introduction: Mozambique’s Community Health Worker (CHW) or Agentes Polivalentes Elementares in Portuguese (APE) in Mozambique was reinitiated in 2010 after an unsuccessful rst attempt decades prior. Years after this reinitialization, the Ministry of Health of Mozambique (MoH) and other interested parties sought to understand how the implementation of APEs interventions is progressing. Given its low coverage of many health intervention, the province of Zambézia was of specic interest. Methodology: We carried out a cross-sectional study in six districts of the Province of Zambézia, namely: Ile, Inhassunge, Milange, Mocubela, Mulevala and Pebane. We designed a questionnaire based on government plans, APE training material, and best practices for the implementation of CHW programs found in the literature. We conducted descriptive analysis to summarize the different components of the APEs program according to responses. Results: A total of 144 APEs were interviewed, of which 67% were male, 63% aged between 18 to 39 years, 60% had a rst cycle of education and 46% had worked as APEs for 5 or more years. While 91% reported having received training after their initial training, only 19% had received training in the past 3 months. Only 48% of APEs replenished their kits every month, and stock-outs of medications ranged from 19% to 92%, with expired medication being found for 5-13% of the medications. Although the quality of supervision appears high with APEs reporting service delivery observation (88%), discussion of work performance (81%) and work-related problems (91%), checking of supplies (90%), review of records (93%), and praise (92%) during their last supervision, only 59% of our sample reported having had their last meeting with their supervisor less than 1 month ago. Conclusions: While there have been some successes in this renewed implementation, challenges persist around ensuring the proper implementation of the program as it is designed. Additionally, a review of the program to match the current realities of eld (e.g., increased population per APE, increase of the stipends) is becoming increasingly necessary.

APEs are members of medically underserved communities who are designated by their communities and then trained to (1) perform health promotion activities, and to (2) provide basic health services in their communities (3). APEs provide integrated community case management (iCCM) of childhood illness (1).
The APE program has made important contributions to expanding access to health services in remote communities within Mozambique (4), and if carefully implemented, can contribute to achieving increased health equity in the country (5)(6)(7)(8).
In 2015, a technical revision to the APEs training curriculum was issued and implemented, which added the administration of family planning interventions (4). In theory, each APE is meant to be responsible for 500 to 2,000 inhabitants depending on the population density and access to health facilities (3). As part of the basic healthcare services they provide, each APE is given a medical kit which contains basic medication and commodities to diagnose and treat common minor illnesses (3). The kits are assembled by Ministry of Health and are distributed to local health centers on a monthly basis for APEs to collect and use in their communities. Each APEs' entire kit is intended to be replaced every month, regardless of the quantity of commodities used (3). The type and quantity of components in the APE kits were based on rough estimates of APEs' commodity monthly use originally calculated by the MoH. These quantities have not been signi cantly altered since the initiation of iCCM in Mozambique and are not based on current APE consumption data or on regional and seasonal variations in disease burden (9).
The coverage of health interventions in Zambézia is generally lower than the national average. For example, the contraceptive prevalence rate in Zambézia is 18%, compared to 27.1% nationally. 20% of women in Zambézia have not attended any pre-natal consultations, compared to the national average of 6.7%, and 75% of Zambézia's babies do not receive any post-natal consultation, compared to 55.6% nationally (4). Additionally, most of Zambézia's population is rural and 39.1% of its population belongs to the poorest wealth quintile (4). Zambézia is therefore particularly well positioned to bene t from an APE program.
We undertook this study to assess how the APE program is currently being implemented. We hope that the ndings from this study will help identify where roadblocks exist and help guide decision-makers in how to address them.

Sampling
We conducted a cross-sectional survey of APEs in Zambézia province, using single-stage cluster sampling. We chose this sampling strategy as, for indicators that take individual APEs as the unit of analysis, the single-stage cluster sampling would in result in precision of +/-10 percentage points, assuming an observed estimate of 50%, a design effect of 1.5, and an alpha of 5%. Given that there were 22 districts in Zambézia overall, this meant selecting 6 districts as our sampling units. To select districts, we obtained a list of all 22 districts in Zambezia province. We rst strati ed this list by urban (n=5) and rural (n=17). We then performed systematic random sampling on each stratum to select our 6 districts, 5 of which were rural (Inhanssunge, Ile, Mocubela, Mulevala, Pebane), and 1 urban (Milange). We then obtained an updated list of all APEs currently working in the selected districts from the DDSMAS. All APEs included on this list for the selected districts were included in our sample.

Questionnaire design
We built a questionnaire to capture how well the APE program (as described in government plans) is being implemented. To do so, we examined national strategies, government documents, and the APE training curriculum to identify what knowledge, tools, medication, and activities an APE should theoretically have or do. Based on this information, we developed a questionnaire that asks about each expected activity (e.g. frequency of occurrence, knowledge about the activity), as well as each individual medication (e.g. whether the APE has it, expired status, stock-outs). Additionally, we used the Community Health Worker Assessment and Improvement Matrix (10) which summarizes the key factors necessary for a successful APE program to help our analysis. Building on both elements, we developed a questionnaire which is available as supplementary material.

Data collection
We used Open Data Kit (ODK) as our data collection platform. We carried out 14 days of data collection in the selected districts. Given that data collection occurred in January of 2020 (i.e. in the rainy season), road conditions for travel in the districts were poor. Teams were given a current listing of APEs for their allocated district. Additional APEs found during data collection were not included in the sample.
Interviews took place either in APEs' homes when road conditions permitted it, and at the health center when reaching APEs' homes was not possible. Data collectors performed an inspection of kits only when they were able to reach APEs' homes, as the kits and complete inventories are too large for APEs to travel with them.

Data analysis
We examined APEs' knowledge, activities, and presence of t-for-use medication and equipment. We also assessed APE work satisfaction. Data were analyzed using descriptive statistics, such as frequencies and means, to summarize APE responses. Estimates were calculated accounting for our study design and sampling weights. We constructed weights for each APE based on the joint probability of being in the nal sample of responding APEs, i.e., the probability that: 1) the district was sampled 2) the APE was sampled within the district (which given our study design was 100% for all APES), and 3) the APE was contactable (different for each district). To account for clustering (in this case, districts), Taylor-series linearization was used to estimate variance and calculate con dence intervals. A Finite Population Correction factor (FPC) was also applied to our estimated standard errors to account for how close our sample size is to the whole province-level population of APEs. To apply these adjustments for our sampling design and response rates to estimate ndings for Zambezia province as a whole, we employed the "svy" set command in Stata. All data presented in this paper is weighted estimates along with their 95% con dence interval for all APEs in the province of Zambezia province. Statistical analysis was performed using Stata (11).

Sociodemographic characteristics
Of the 171 APEs planned for this survey, 144 (84%) were interviewed and 23 (16%) were not, due to various reasons. Of note, of those 23, 11 individuals were no longer working as APEs, it was impossible to contact 10 individuals during data collection, and 2 individuals were deceased for a total of 23 (13%). Most APEs were between 20 and 39 years old (62%), male (67%), and married (79%). More than half of the respondents (57%) have a junior secondary level and were able to read (93%) and solve a simple arithmetic equation (81%) ( Table 1). APEs workload APEs are on average responsible for 3,926 individuals spread across an average of 3.4 communities (table 2). They live on average 38 km away from the health facility and 52 km from the center of the district (table 2). The performance of APEs in terms of training is described in table 3. While 99% of the interviewed APEs reported having received training to be an APE, only 20% reported having received some training in the last 3 months before the survey.

Challenges reported by APEs
We observed the kits of 64% of the APEs that we were able to reach. We found that 81% of the kits were in acceptable conditions (Annex 1). As for the condition of the medications in the kit, the percentage of expired medications varied from 5 to 13% (Figure 1).
Regarding supervision, 99% of APEs reported that they had ever received supervision but only 18% had received supervision in the last two months. Most APEs (94%) reported that they were given suggestions and advice during the last supervision they received. During supervision only 80% of APEs mentioned that the supervisor used the checklist during supervision and 88% mentioned that the supervisor observed them while working during their last supervision. When asked about frequency of activity reports sent to supervisors, 92% had sent the activity report in the last month, 99% submitted the reports directly to the health unit and 79% received feedback from the reports (Annex 2).
With regard to the barriers mentioned for not going to the health facility when patients are referred, 45% of the APEs reported that patients do not have transport to the health facility, 20% stated that patients do not have money to go to the health facility and 15% of patients did not want to go to the facility (Annex 3). Regarding barriers to replenishing their equipment and medications, 58% of the APEs reported not facing any challenge and 31% said that the health facility had a stockout (Figure 2).

APE and the health system
Although the program prioritizes the recruitment of women to be APEs, our ndings show that men continue to outnumber women -which has been a concern from the very beginning of the program (1).
Encouragingly however, our ndings suggest a shift towards hiring more women as of 2015 (year of the introduction of family planning to the APE program).
In terms of recruitment, only APEs who are literate and able to solve simple arithmetic equations are supposed to be allowed into the program. Nearly a fth of the APEs are unable to solve a simple equation, which is concerning since APE need to be able to dispense a speci c amount of medication and this may impact the delivery of their services. Although some evidence suggests that interpersonal skills outweigh educational background in terms of what makes a good CHW (12), literacy and basic numeracy is also very important for anyone managing medication.
Community embeddedness is a key determinant for the success of CHW programs -in part because CHWs are typically most familiar with their community, and in part because it protects CHWs from uctuations in political interests (12,13) . Per our ndings, although most APEs were recruited by their own community, more than a tenth were recruited by health facility staff. APEs also reported that their patients did not heed their referral to the health facility because they did not want to go to the health facility (15%) and because of the APE's own poor relationship with the health facility (5%). Taken together, these elements suggest that there is room for the health system to get more engaged with the community and with the APEs themselves.
The discrepancies between the lists of APEs provided at central and local level by the APES Program may indicate the existence of a disconnect at the level of updating records. This disconnection leads to a lack of precision and integrity of information, for example, by not including the systematic updating of APEs that have already died, others that are no longer working as APEs, and new APEs found in the eld and that were not included in the aforementioned lists. Proper record-keeping and quality assurance have been linked to improved performance of CHW programs (14), suggesting that there is also room for improving the monitoring system.
Infrequent supervision, supervision that does not include eld visits, and supervision that focuses excessively on reports can be demotivating for CHWs (14, 15). Upwards of 90% of APEs in our sample reported that their supervisor provided praise, provided advice, and asked about work-related problems during their last supervision. This suggests that the supervisor-APE relationship typically does go beyond simple box-checking exercises. However, not all supervisors complete expected tasks, as a fth of APEs reported that their supervisor did not complete the supervision checklist during the last supervision, and another fth reported not receiving feedback on their reports -both of which are required by the APE program.

Medication and commodities
We found that a concerningly high percentage of APEs had expired medications in their kits at the time of inspection. This nding was surprising, given that 90% of APEs reported that their supervisor had reviewed their medication during the most recent supervision. However, APEs report receiving on average 8 supervisory visits per year -4 fewer than planned.
We found stock-outs across all medications at the time of the study, with APEs reporting having had stock-outs within the past three months for all medications as well. Given that APEs serve on average approximately two to four times the population that their kits were designed for (3,926 individuals, compared to the 500-2,000 planned), stock-outs are unsurprising. APEs who run out of medication early in the month have no recourse to turn to and must refer patients to the health facility -who also experience frequent stock outs (16,17). Although some diseases exhibit seasonal patterns in Mozambique (18), the medication provided in the kits does not take these patterns into account -as supported by the low availability of malaria medication in our results. Frequent occurrences of the community seeking assistance from CHWs who are unable to provide care due to stock-outs can lead to the undermining of community trust (15,19). Similar factors are likely at play with the APE program in Zambézia as well.
The availability of all family planning items -from injectable capsules to male condoms -is particularly low, which contrasts with the APE program's focus on family planning as a priority area. This is of particular concern since interruptions in contraceptive use puts patients at increased risk for unintended pregnancies. Relatedly, misoprostol was also very infrequently available, and many APEs reported never having had it at all. Although women's health is prioritized at the policy level in Mozambique (20), breakdowns in the supply chain such as those found here threaten potential gains in women's health.
Training Almost all APEs have been trained to carry out health activities in their communities, we found that refresher trainings over time are infrequent. This suggests that APEs have not had su cient or recent training in practice, which can affect performance and the impact of their activities on communities (21), in-service training aims to update CHW with new policies and programs, reinforce initial training, and ensure they are practicing skills learned (21). In-service training is provided at regular intervals, varying from monthly to twice a year, and offered in the form of workshops, monthly meetings, and refresher courses (22). According to James et al., continuous training programs are more likely to contribute to improving the level of systems in environments with limited resources (23). Continuing, or refresher, training provide an opportunity to update APEs on new skills, to reinforce their initial training, and to ensure they are practicing skills learned. Beyond pre-service training, many countries also provide inservice trainings to extend or refresh the skills of the CHWs (24).

Barriers and bottlenecks
Transport was the most frequently reported di culty for APEs to both to collect their kits and to refer patients. This is a common issue for CHW programs generally (25) since these programs tend to be implemented in remote regions where access to a health facility is di cult. Although the APE program does provide bicycles, no provisions are made for repairs and maintenance.
Only a quarter of APEs reported having received their stipend within the past month, and approximately a third reported not having received it within the past three months. The amount of the monthly subsidy (1200 Meticais, roughly equivalent to 16 United States dollars at time of writing) has also not increased since the inception of the program a decade ago, despite the fact that APEs were informed that it would (15) and that the guidelines do plan for adjusting for in ation over time (3). Small payments that are delivered infrequently have been identi ed as demotivating factors for CHWs in other similar settings (19,25).
Although we were unable to capture it quantitatively, some APEs accidentally revealed wearing more than one hat by showing the research team cellular phones, contraceptives of a different type, and medications of different brands when asked to show their APE kits. All of these are items that are not included in the APE kit but are part of kits of other NGO-based health outreach programs in the region. A complex funding landscape and vertical programming is a well-documented issue in Mozambique (26), and that APEs would diversify both their income and medical source is unsurprising. APEs in Mozambique already report feeling overburdened (15,27), and having to resort to adding on additional positions may worsen the situation.

Limitations
Data collection took place during the rainy season, which meant that road conditions made it impossible to reach all of the APEs' communities to observe their kits. The kits that we were able to observe belonged to APEs who resided close enough to a functional road to be reached by car, short walk, or motorbikeand who therefore have easier access to their health facility. Additionally, given the seasonality of certain diseases, data collected during the dry season may have presented a different picture in terms of consumption of medication.

Conclusion
A decade after it was reinitiated, Mozambique's APE program has had some successes, but challenges persist. Most urgently, more focus on ensuring a steady supply of APE kits and subsidies being delivered regularly and on time is necessary. While supervision seems to be of high quality, the health system must ensure that it occurs more frequently. Since the workload has increased for the APEs, a review of the program to match the current reality is necessary -either more APEs need to be hired, or the supplies and support -both nancial and through supervision -planned for each APE must be increased.  Figure 1