Monitoring Pharmacy and Test Kit Stocks in Rural Mozambique: U.S. President's Emergency Plan for AIDS Relief Surveillance to Help Prevent Ministry of Health Shortages

Abstract

Support of human immunodeficiency virus (HIV) and tuberculosis (TB) testing and treatment supported by President's Emergency Plan for AIDS Relief (PEPFAR) in Africa requires immense quantities of tests and medications. We sought to use central pharmacy supply data of Mozambique's rural Zambézia Province (2017 population ≈5.11 million persons; ≈12.6% adult HIV prevalence in 2016) to examine shortages, stockouts, and trends in availability. Using stock surveillance for 60 weeks in 2014–2015, we assessed availability of 36 medications [4 classes: adult antiretroviral (ARV) medications, pediatric ARVs, anti-TB medications, and antibiotics] and diagnostic test kits (2 rapid tests for HIV; 1 each for malaria and syphilis). We contrasted these to 2018–2019 data. We modeled pharmacy data using ordinal logistic regression, characterizing weekly product availability in four categories: good, adequate, shortage, or complete stockout. We found 166 (7_.7%) stockouts and 150 (6.9%) shortages among 2,160 weekly records. Earlier calendar time was associated with reduced medication supplies (p < .001). Certain medication/test kit classes were associated with reduced supply (p < .001). We found an interaction between time and medication class on the odds of reduced supply (p < .001). Pediatric ARVs had a 17.4 (95% confidence interval: 8.8–34.4) times higher odds of reduced medication supply compared with adult ARVs at study midpoint. Trends comparing the first and last weeks showed adult ARVs having 67% and pediatric having 71% lower odds of reduced supplies. Only adult ARV shortages improved amid growing demand. Data from 2018 to 2019 suggest continuing inventory management challenges. Monitoring of drug (especially pediatric) and test kit shortages is vital to ensure quality improvement to guarantee adequate supplies to enable patients and care providers to achieve sustained viral suppression. A central Mozambican drug repository in the nation's second largest Province continues to experience drug and rapid test kit stockouts.

Introduction

The pharmacy supply chains of sub-Saharan African primary care systems were not prepared to meet the surging demands of the emergency response to the human immunodeficiency virus (HIV) and tuberculosis (TB) epidemics, funded intensively since 2004.^1–3 Hence, national governments, the U.S. President's Emergency Plan for AIDS Relief (PEPFAR), and the Global Fund to Fight AIDS, Tuberculosis and Malaria (Global Fund) have assisted Ministries of Health to better capacitate national systems, often setting up parallel supply systems with private contractors to meet the urgent demands of combination antiretroviral therapy (ART) and anti-TB medications in resource-constrained settings.

Capacitating pharmaceutical supply systems quickly and sustainably to meet the demands of burgeoning HIV-TB programs is more than a funding challenge; human workforce and supply chain logistics can be daunting. This is especially difficult within the vast rural regions of sub-Saharan Africa where many recent scale-up initiatives have been implemented.^4–9

Availability of HIV rapid test kits is an essential element of achieving the first of the three international goals of the World Health Organization and UNAIDS in achieving the 90-90-90 objectives by 2020, namely the “first 90” or having at least 90% of persons living with HIV (PLHIV) knowing their status. When a motivated individual wants an HIV test, but no test kits are available, the health system has lost a moment of opportunity that may not present itself again when kits become available again. Similarly, lack of antiretroviral (ARV) medications undermines the second and third “90” goals, namely having at least 90% of persons testing HIV-positive taking ART and having at least 90% of the persons on ART being virally suppressed. TB and HIV resistance can ensue from drug shortages.

Pharmacy-related health system strengthening has been essential, especially in the many venues that lack experienced supply chain professionals.^8–12 Delivery of medications or diagnostic test kits to health facilities depends on warehouse managers, shipping and logistics managers, quality assurance experts, data managers, pharmacists/pharmacy technicians, loaders and truckers experienced with handling medications and laboratory supplies, and financial managers and accountants. Even one of the simplest of these jobs, that of a truck or train loader, has many elements of required training and skills (Supplementary Table S1).

Pressure to attain higher and higher coverage targets for persons newly initiated on lifelong ART and anti-TB medication regimens can compromise the quality of pharmacy systems. In the extreme case of a pharmacy medication stockout (i.e., a complete lack of product in the pharmacy), patients who need and want medications cannot take them, undermining adherence and likely contributing to loss to follow-up when disgruntled patients do not return in the future.¹³ Similarly, a stockout in rapid diagnostic test kits in an under-resourced rural area means that someone desiring an HIV test cannot get it.

Data management systems in sub-Saharan Africa do not typically provide real-time assessment of supply chain system stocks.¹⁴ Supply chain managers and health officials with control over medication logistics may not know, in real time, what sites are stocked out, on what days, and for how long.^15,16 In addition, health care workers are often uncertain when nonadherence is due to patient-related factors or due to interruptions in medication availability for people receiving ART or anti-TB medications. The literature on this topic is not as detailed as in other areas of HIV/AIDS program management; in our experience, many health and donor agencies are reluctant to permit publication of drug and test kit stockouts and other shortages, perhaps because they do not wish to advertise suboptimal programmatic functioning.

Pediatric ARV medications and anti-TB medications may be especially vulnerable to stockouts as they may not be used in as high a volume as the adult ARVs for which supplies may have been planned and alternative supply paths may have been created. A 2015–2016 initiative of PEPFAR and the Children's Investment Fund Foundation (CIFF) has supported the Accelerating Children's HIV/AIDS Treatment (ACT) initiative, investing in 9 African nations (including Mozambique) to double the number of children receiving potentially life-saving ART to 300,000 over a period of 2 years. Because of increased need for HIV testing and care under the new PEPFAR/CIFF ACT program, there have been substantial challenges to supply enough rapid HIV test kits and pediatric ARVs and anti-TB medications to meet heightened demands.

Similarly, a reorientation of funding priorities to redirect resources both to care and treatment and also to areas of highest HIV prevalence (the PEPFAR 3.0 strategy) has resulted in an increase in the number of persons being placed on ART and anti-TB medications in many regions, as well as an increased demand for HIV rapid test kits that are supplied along with medications.

Pharmacy continuous quality improvement (CQI) programs are designed to do what any business seeks to do, ensuring availability by keeping products that are needed by customers in optimal stock, always providing needed drugs, but not wasting any drugs through overstocking and consequent expiration. Models in high income countries highlight the value of process reporting and routine audits to identify, correct, and institute systemic changes to avoid future mistakes in stocking and dispensing drugs; the New Hampshire Pharmacists Association in the United States highlights their policies and state regulations to ensure CQI (www.nhpharmacists.net/resources/CE%20Presentations/2019/2019%20EV1/CQI.CE.4.19.pdf). Such programs are in their comparative infancy in low income countries like Mozambique, but new high-volume programs such as PEPFAR have made such programs essential.

We sought to use pharmacy supply data in the central stores of Zambézia Province, 2016 population of ≈4.9 million people, to examine shortages, stockouts, and trends in availability over time. The Mozambican government cooperated and the PEPFAR program of the U.S. Centers for Disease Control and Prevention (CDC) supported this implementation science project to aid in quality improvement.

Materials and Methods

To identify achievements and challenges in pharmacy demands in rural Mozambique, we used pharmacy records of the central stores of the Zambézia Province central pharmacy run by the provincial health directorate (Direcção Provincial de Saúde or DPS), with technical assistance from the nongovernmental organization Friends in Global Health (FGH), affiliated with Vanderbilt University Medical Center. When medication stocks at these central pharmacy stores are low or fully depleted, adverse downstream effects are seen in the availability of medications at the distal individual health facility-level pharmacies.

Pharmacy technicians request replenishment of supplies based on recent usage, and the central pharmacy relies on these recent usage data to determine near term allocations/resupply needs. Motivating this analysis, the provision of medications and diagnostic test kits depends upon historical usage statistics, an approach that may not be suited to a rapidly growing program, since historic usage may not reflect imminent needs within a growing program.

All medications and diagnostic test kits are supplied by the Central de Medicamentos e Artigos Médicos (CMAM: the Center for Medicines and Medical Supplies) in Maputo with help from a private PEPFAR contractor (Fig. 1). For the central part of the nation, including Zambézia Province, the CMAM regional supply office in Beira supplies general medications (predefined by size of the heath unit), ARV medications, and anti-TB medications (monthly) and diagnostic test kit supplies (quarterly) to Zambézia's Depósito Provincial de Medicamentos (DPMZ: Provincial Medication Repository). The DPMZ, in turn, distributes medications/test kits to each district's clinic pharmacies (Fig. 1). Since our data only reported the data from the Provincial Medication Repository itself, we did not assess or compare stocks in distal pharmacies of provincial clinics and hospitals.

FIG. 1.

Flow of medications through the Mozambican health system to rural clinic pharmacies in Zambézia Province.

For our CQI work, the DPS and FGH have been collaborating in four main pharmacy areas: (1) training of supply chain professionals; (2) improving the documentation of low supplies or complete stockouts of adult and pediatric ARVs, anti-TB medications, antibiotics, and diagnostic test kit supplies for HIV, malaria, and syphilis; (3) early action plans when supplies are low; and (4) improving projections of supply chain needs by pharmacies and individual health facilities. In addition, plans have been developed for emergencies, as with natural disasters like floods that occur in the province.

To document HIV and anti-TB medication availability, color coded Microsoft Excel^® (Microsoft Corporation, Redmond, WA) tracking spreadsheets are used for each facility utilizing a four color “heat-map” scheme. These spreadsheets are updated monthly by DPS and FGH pharmacy leaders for both the Provincial Medication Repository stores. Green means that the given medication is in good supply for the corresponding week (good stock; “Stock Bom” in Portuguese), while yellow (adequate; Reasonable stock or “Stock Razoável”) and red (shortage; Stockout is imminent or “Eminência de Rutura”) indicate a potential or urgent shortage, respectively. Black means that the medication is completely unavailable (stockout; “Rutura”).

Good, adequate, shortage, and stockout definitions as translated from Ministry of Health guidelines are, respectively: Good Stock—Sufficient stock to distribute and fulfill all requirements, with additional safety stock on hand; Reasonable Stock—Sufficient stock to satisfy requests, but no additional safety stock on hand; Imminent shortage—Insufficient stock to meet all requests; Inadequate stock for Average Monthly Consumption; and Stockout—No product available. We converted the four colors into an ordinal numeric scheme and assessed the frequency of stockout for each medication. We then assessed the progress being made in preventing medication and diagnostic test kit stockouts.

Weekly tracking data of the Provincial Medication Repository from April 2014 through June 2015, spanning 60 weeks, were used to investigate time trends by six medication/test kit classes. We did not analyze stocks at the distal clinics and hospitals, so we could not assess risk factors for shortages for individuals at their points-of-care, although a shortage or stockout at the Provincial Medication Repository does not bode well for availability of drug for distal venues. The six classifications are as follows: (1) adult ARVs, (2) pediatric ARVs, (3) adult anti-TB medications, (4) pediatric anti-TB medications, (5) antibiotics, and (6) rapid diagnostic test kits. Due to the limited number/volume of pediatric TB medications, pediatric and adult anti-TB medications were combined as one medication class for some analyses.

We modeled the ordinal pharmacy tracking data (good, adequate, shortage, or stockout) using ordinal logistic regression. We included medication/test kit class (categorical), calendar week (continuous), and an interaction term as covariates. Week was modeled using a restricted cubic spline with four knots.¹⁷ To illustrate the interpretation of our data, a proportional odds ratio (POR) equal to 3.2 comparing medication class A with medication class B as reference on ordinal pharmacy tracking data would have the following interpretation: for medication class A, the odds of stockout versus the combined categories of good/adequate risk are 3.2 times higher than for medication class B, given that other covariates are held constant. Likewise, odds might apply to the combined categories of stockout/shortage versus good/adequate. We refer to these PORs as the odds of “reduced medication supply.” R version 3.3.1 (www.r-project.org) was used for analysis.

To assess progress in the past 4 years, we obtained data from a 12-month period (October 2018 to September 2019) that are from the same Mozambican government source and are recorded in the same CQI system as in 2014–2015, the Provincial Medication Repository. We compared the time periods for ARV and anti-TB drugs, antibiotics/antifungals, and test kits. We also share the 2018–2019 data for newly added products that are tracked, but were not tracked in the 2014–2015 period: anticancer drugs; hormonal and barrier contraceptives; and Plumpy Nut nutritional supplement. We assessed the extent to which weekly stockouts and near stockouts were comparable or different from the situation in 2014–2015. Note that each month is assigned four near-equal time blocks with each month having four blocks. Hence, the “week” averages 7.6 days or 48 “weeks” per year. Weeks in which the given adult or pediatric ARV or anti-TB drug/test kit/antibiotic/contraceptive/nutritional supplement was not in the formulary were not counted in the time denominator.

The study was approved by four entities in Mozambique and in the United States: Vanderbilt University Medical Center Institutional Review Board (170970); Institutional Research Ethics Committee for Health of Zambézia Province (Comité Institucional de Bioética para Saúde-Zambézia; 02-CIBS-Z-17); Provincial Directorate of Health of Zambézia Province (DPS Zambézia; 1859-DPS-Z-240-2017); and U.S. CDC office in Mozambique (2016-163a).

Results

To study the supply chain and inventory status for drugs and test kits in central provincial pharmacy stores in Zambézia Province, we tracked 36 drugs within 6 medication classes for 60 calendar weeks from April 2014 through June 2015 (Fig. 2 and Table 1). There were no missing data from the pharmacy reports in this time period, a consequence of intensive prior data quality improvement efforts.

FIG. 2.

Pharmacy stocks in the Zambézia, Mozambique, Provincial Medication Repository, April 2014 to June 2015. This is a heatmap reproduction of the provincial tracking data. The color key is: good stock (green), adequate stock (yellow), shortage (red), and stockout (black). 3TC, lamivudine; ABC, abacavir; ART, antiretroviral therapy; AZT, zidovudine; Comp, comprimidos (tablets); d4t, stavudine; EFV, efavirenz; LPV/r, ritonavir-boosted lopinavir; NVP, nevirapine; Susp, suspension; TB, tuberculosis; TDF, tenofovir disoproxil fumarate.

Table 1.

Frequency of Stockout and Shortage by Medication and Test Kit from April 2014 Through June 2015

	Good	Adequate	Shortage	Stockout
Adult—antiretroviral medications, n (%)
3TC+AZT	60 (100)	0 (0)	0 (0)	0 (0)
3TC+AZT+NVP	59 (98)	0 (0)	1 (2)	0 (0)
3TC+d4t	55 (92)	5 (8)	0 (0)	0 (0)
ABC	42 (70)	14 (23)	4 (7)	0 (0)
AZT	57 (95)	3 (5)	0 (0)	0 (0)
d4t+3TC+NVP	56 (93)	2 (3)	2 (3)	0 (0)
EFV	57 (95)	1 (2)	2 (3)	0 (0)
LPV/r	52 (87)	3 (5)	1 (2)	4 (7)
TDF+3TC	48 (80)	9 (15)	3 (5)	0 (0)
TDF+3TC+EVF	60 (100)	0 (0)	0 (0)	0 (0)
Pediatric—antiretroviral medications, n (%)
3TC+AZT	39 (65)	6 (10)	15 (25)	0 (0)
3TC+AZT+NVP	47 (78)	10 (17)	3 (5)	0 (0)
3TC+d4t	28 (47)	10 (17)	3 (5)	19 (32)
LPV/r (Aluvia®)	35 (58)	11 (18)	5 (8)	9 (15)
LPV/r (Kaletra®)	38 (63)	12 (20)	2 (3)	8 (13)
NVP	37 (62)	11 (18)	7 (12)	5 (8)
Rapid diagnostic test, n (%)
Determine® HIV	30 (50)	16 (27)	10 (17)	4 (7)
Malaria	51 (85)	1 (2)	0 (0)	8 (13)
Syphilis (RPR)	30 (50)	8 (13)	12 (20)	10 (17)
Unigold® HIV	29 (48)	18 (30)	10 (17)	3 (5)
Adult—anti-TB medications, n (%)
Isoniazid 100 mg	52 (87)	6 (10)	2 (3)	0 (0)
Isoniazid 300 mg	39 (65)	13 (22)	1 (2)	7 (12)
Isoniazid+rifampin	54 (90)	6 (10)	0 (0)	0 (0)
Isoniazid+rifampin+ethambutol	49 (82)	11 (18)	0 (0)	0 (0)
Isoniazid+rifampin+ethambutol+pyrazinamide	33 (55)	13 (22)	11 (18)	3 (5)
Pediatric—anti-TB medications, n (%)
Isoniazid+rifampin	46 (77)	11 (18)	3 (5)	0 (0)
Isoniazid+rifampin+ethambutol	52 (87)	8 (13)	0 (0)	0 (0)
Antibiotics, n (%)
Azithromycin	35 (58)	2 (3)	11 (18)	12 (20)
Benzathine penicillin	21 (35)	18 (30)	5 (8)	16 (27)
Ceftriaxone	36 (60)	8 (13)	6 (10)	10 (17)
Ciprofloxacin	29 (48)	17 (28)	10 (17)	4 (7)
Cotrimoxazole (tablets)	36 (60)	9 (15)	4 (7)	11 (18)
Cotrimoxazole (suspension)	36 (60)	9 (15)	8 (13)	7 (12)
Metronidazole	29 (48)	10 (17)	3 (5)	18 (30)
Penicillin	38 (63)	14 (23)	1 (2)	7 (12)
Procaine penicillin	36 (60)	18 (30)	5 (8)	1 (2)

Data from the Central de Medicamentos e Artigos Médicos (CMAM: the Center for Medicines and Medical Supplies of Zambézia Province, Mozambique), as analyzed and summarized by the authors.

3TC, lamivudine; ABC, abacavir; AZT, zidovudine; d4t, stavudine; EFV, efavirenz; HIV, human immunodeficiency virus; NVP, nevirapine; LPV/r, ritonavir-boosted lopinavir; RPR, rapid plasma reagin; TB, tuberculosis; TDF, tenofovir disoproxil fumarate.

Among 2,160 weekly medication/test kit records, there were 166 (7.7%) recorded stockouts in Zambézia's Provincial Medication Repository (Table 1). Another 150 (6.9%) weekly medication records had supply shortages, that is, deemed at high risk of stockout by the pharmacist. Of the remaining weekly records, 313 (14.5%) were recorded as being in adequate supply and 1,531 (70.9%) as being in good supply. Using ordinal logistic regression, there was strong evidence that earlier calendar time was associated with a reduced medication supply (p < .001); the time association was nonlinear (p < .001), select medication classes were associated with reduced supply (p < .001), and there was an interaction effect between time and medication class on the odds of reduced supply (p < .001; Fig. 3). We illustrate both time trends (Fig. 3) for the five medication/diagnostic test kit classes (adult and pediatric anti-TB medications merged) and present the PORs (Table 2).

FIG. 3.

Marginal ordered log-odds plots of medication or diagnostic kit class and time interaction for Zambézia Province reduced supply of pharmacy stocks in the Provincial Medication Repository, April 2014 to June 2015. Due to small numbers, we merged the pediatric and adult TB medication classes.

Table 2.

Ordinal Logistic Regression Results for Reduced Medication Supply in Zambézia Province Central Pharmacy Store Supplies, April 2014 to June 2015

	Proportional odds ratio (95% CI)
Medication class (adjusted to week 1 of November 2014)
Adult antiretroviral medications (Ref.)	1
Pediatric antiretroviral medications	17.4 (8.82–34.4)
HIV rapid test	10.6 (5.16–21.8)
Anti-TB medications	3.21 (1.40–7.33)
Antibiotics	4.78 (2.18–10.5)
Week 60 versus week 1 by medication class
Adult antiretroviral medications	0.33 (0.11–1.04)
Pediatric antiretroviral medications	0.29 (0.11–0.79)
HIV rapid tests	0.54 (0.17–1.68)
Anti-TB medications	2.14 (0.73–6.33)
Antibiotics	2.17 (0.99–4.76)

Due to small numbers, we merged the pediatric and adult TB medication classes. This table summarizes the model using proportional odds ratios for various comparisons of the interaction effect. Week 1 is week 1 of April 2014, and week 60 is week 4 of June 2015; these are the first and last weeks of available data.

CI, confidence interval.

We conducted a group test of the interaction terms, testing the following null hypothesis: “changes in stockout over time are not different by medication class.” While there was significant medication/diagnostic kit and time interaction, the ordered log-odds plots indicate that the interaction was only relevant for rapid test kits. Adult ART showed improvement over time, but pediatric ART, TB drugs, and antibiotic did not show much difference. That one group has a different trajectory from the other groups is consistent with a finding rejecting the null hypothesis.

At the midpoint of our study period (first week of November 2014), pediatric ARV medications had a 17.4 [95% confidence interval (CI): 8.8–34.4] times higher odds of a reduced supply compared with adult ARV medications. Similarly, rapid diagnostic test kits had 10.6 times higher odds (95% CI: 5.2–21.8) of reduced supply compared with adult ARV medications in early November 2014. When comparing the last week of the study period to the first week, both adult and pediatric ARV medications had lower odds (67% and 71%, POR 0.33, 95% CI: 0.11–1.04 and POR 0.29, 95% CI: 0.11–0.79, respectively) of reduced supply, indicating that supply chain management had improved over time (Table 2).

The 2018–2019 data indicated the following weekly stockout rates over the 12-month span (October 2018 to September 2019): adult ARVs 5.1%; pediatric ARVs 23.8%; adult anti-TB drugs 11.6%; pediatric anti-TB drugs 1.4%; test kits 9.4%; and antibiotics 24.5%. Among 2,811 weekly medication/test kit records, there were 368 (13.1%) recorded weeks with stockouts in the Zambézia Province central pharmacy (Fig. 4) versus 7.7% in the 2014–2015 time period. There were more drug types being stocked and tracked in the latter time period. Among the new categories not tracked in 2014–2015, anticancer drugs were likely to be stocked out in 41.3% of drug class weeks, hormonal/barrier contraceptives stocked out in 16.9% of product weeks, and Plumpy Nut nutritional supplement was stocked out in 4.2% of the weeks.

FIG. 4.

Pharmacy stocks in the Zambézia, Mozambique, Provincial Medication Repository, October 2018 to September 2019. This is a heatmap reproduction of the provincial tracking data. The color key is: good stock (green), adequate stock (yellow), shortage (red), and stockout (black).

Discussion

We have documented improvements in medication availability over time, even in the face of burgeoning demands from new policies (e.g., PEPFAR 3.0, rapid expansion of prevention of mother-to-child HIV transmission [PMTCT] Option B+) and programs. Programs included the ACT initiative and the DPS-led pediatric testing campaign, termed “testagem massiva” (widespread pediatric HIV testing) that increased demand for HIV rapid diagnostic test kits and both ARV and anti-TB medications in this time period. However, we also documented periodic medication stockouts, a topic that has generated surprisingly little attention in the global HIV care and treatment literature given the harm that it may have on individual patient outcomes, specifically the risk of development of genotypic drug resistance to one or more ARV medications.

The most recent data from the same central Provincial Medication Repository document continued difficulties in avoiding stockouts. In fact, comparing the same drug and test kit weekly stockout rates, stockouts are more common in 2018–2019 than in 2014–2015. This is in the context of both more drugs being stocked and also much higher drug use volumes with more persons being treated for HIV or TB. We speculate that while there has been system improvement, the CQI system is not keeping up with rapid program expansion, such that stockouts and shortages still occur at high levels.

For quality improvement, the organizational structure of the FGH pharmacy team has been restructured to better support coordination at the district and provincial levels. In addition to the placement of District Clinical Officers, the team was strengthened with Regional Coordinators and the “last mile” Provincial Deposit that coordinates directly with the staff of the Provincial Medications Deposit. The stock monitoring form was then deployed at the district level. In coordination with the Provincial Health Directorate, an enhanced version was piloted to actively report stocks on a weekly basis. Throughout the province, efforts continue toward improvement of communication on inventory management, drug requisition, coordination in the management of therapeutic lines, and collaboration for the timely submission of reports. However, this effort was not without its substantial challenges, and recent data suggest that the CQI efforts must be revitalized to discover why stockouts continue to be common.

The context of our circumstances is reported by other HIV/AIDS/TB programs in rural sub-Saharan Africa.^13,14 With suboptimal funding, the DPS in Zambézia Province provides health care to a very impoverished rural population of subsistence farmers and fisherfolk. As the nation's second most populous province with a comparatively high adult HIV seroprevalence (in 2009, 12.6% in persons 15–49 years of age), Zambézia is challenging logistically. Our findings may be a “best case” scenario, since sufficient stock at the provincial level does not always correlate with sufficient stock at the district level (data not shown). For example, two of our PEPFAR-supported districts, Inhassunge and Chinde, are accessible only by boat. Other districts have experienced extreme flooding during two out of the three rainy seasons proximate to this study. In January 2015, for example, there was severe flooding of the Chire River within the district of Morrumbala, resulting in the displacement of thousands of residents and disruption of clinical services. In the same season, the Nipiode river flooded in the district of Pebane district and the Licungo river flooded, affecting the districts of Maganja de Costa, Alto Molócuè, and Gilé.

With frequent impassable dirt roads, the rainy season is a particularly difficult time to guarantee that pharmacy and laboratory stocks can be successfully transported from the central provincial deposit to peripheral sites. While it is unknown which year will present these seasonal difficulties, one can anticipate that they are possible in any given year related to the rainy season. Hence, DPS/FGH can send additional medications when access is anticipated to be compromised, even though predictions of service interruptions are imperfect, compromising efficient mitigation efforts.

Plans for climate-related logistical complexities may be increasingly important, too, in coastal districts as global climate change and rising ocean levels increase risk of storm surges into coastal communities. In past floods of the Zambezi River, we have gone to refugee sites and provided extra drug supplies to cope with needs of PLHIV whose clinics were closed or who could not access clinics in flooded zones. Seasonal migration might be a concern in parts of Mozambique, but this is not common in Zambézia Province since subsistence farming and fishing require continuous agricultural work or boat-based fishing.

Fiscal resources alone did not prevent continuing pharmacy stockout vulnerability, since PEPFAR works with the government of Mozambique to provide ARV and anti-TB drugs and HIV/syphilis/malaria test kits. Additional care and treatment funds for HIV and TB due to the increased coverage of persons in high prevalence regions facilitated by the “PEPFAR 3.0” strategy and special programs like the ACT have resulted in sharply upward trends in ART enrollment in the 15 districts in which FGH implementation partner staff presently work. This underscores the urgency of effectively managing the supply chain to cope with surging patient numbers. Both pediatric ARV medication and HIV diagnostic test kits were especially vulnerable for stockout and shortage, unfortunate in the face of the ACT initiative. Given comparatively poor program results for pediatric HIV/AIDS management,¹⁸ medication shortages for children require special attention. Similarly, the importance of HIV testing as the key entry point into care and HIV education suggests the urgency of resolving diagnostic test kit shortages at the Provincial Medication Repository level.

Supply chain systems in resource-limited settings often fail to have real-time monitoring to help predict what sites will run out of essential medications and/or diagnostic test kits, on what days and for how long.^4–12 Because of increased need for HIV testing, there have been stockouts of HIV rapid test kits; these were predictable, but not compensated for in supply chain management. Viral load implementation began in July 2015 within Zambézia province and adds new complexities within the context of challenges in supply chain management for even some of the more basic elements of drug distribution.

Supply chain human resources are limited throughout PEPFAR target nations in sub-Saharan Africa, but the scarcity of qualified supply chain personnel is especially pronounced in Mozambique. Many PEPFAR services in Mozambique are focused on operational assistance rather than technical assistance, suggesting that the transition to country “ownership” is not imminent as it is in other better capacitated nations.^19,20 We were pleased to see that central pharmacy shortages in Zambézia Province declined over time, despite the expansion of persons being covered. We investigated the correlation of central supply stocks and district level supplies that had been well documented to be suboptimal in an adjacent Mozambican province.^21,22

The PEPFAR 3.0 plan has moved resources from lower to higher prevalence (or burden) regions.²³ Challenges include the training, retention, and accountability of supply chain staff, including warehouse managers, logistics managers, quality assurance people, accountants, stockers, shippers, and transportation managers. In Mozambique, the sometimes-bleak human resource situation slows sustainable improvements in the supply chain. Staffing up supply chain projects with expatriates to deal with the expanding demands is a short-term costly solution. Governments must address the human resource crisis in supply chain management, building local capacity, and make improvements sustainable.

To avoid the unintended consequences of increasing dependency of nations on PEPFAR and Global Fund foreign operational assistance, an action plan for sustainability is needed. We believe that the problem illustrated by our study findings, along with evidence from the literature, points toward these needed elements:

• (1)Intense training and mentorship toward human resource development of indigenous supply chain experts, with career incentives for their retention in the public sector.^15,24–27
• (2)Temporary enhancement of expatriate staff toward health systems strengthening in the supply chain arena. All such staff must have a training and human capacity-building role, to “work themselves out of a job” through knowledge transfer and training of Mozambicans.
• (3)Clarity in the relationships between global funders (e.g., PEPFAR and Global Fund) and national and provincial health ministries as to which entity is fiscally responsible for which operational or technical task.
• (4)New standard operating procedures with Ministries of Health that transfer clearly the responsibilities from externally supported logistics partners to the Ministry on a realistic timetable, to ensure long-term sustainability and enhancement of overall drug/supply logistics.
• (5)Effective programmatic adjustments to meet expanding needs in areas where new PEPFAR 3.0 and ACT resources have increased demand for test kits, adult and pediatric ARV and anti-TB medications, and laboratory demands [e.g., GeneXpert^® MTB/RIF (Cepheid, Sunnyvale, CA) or TB, CD4⁺ cell counts, viral loads, syphilis testing, and others].

In Mozambique, supply chain projects are performed by operational contractors, not just through technical assistance to the Mozambican Ministry of Health (MISAU), a reasonable emergency response, but not what anyone wants for long-term management. Arrangements between PEPFAR and Global Fund authorities remain awkward with contractors responsible for distribution logistics depending on other contractors responsible for drug procurement and importation. Pressure to reach ART targets can come at the expense of quality, namely poor performance in patient retention and medication adherence in the face of stockouts and staff shortages under rising numbers of patients over time.

If supply chain systems could predict what sites might be stocked out, on what days, and for how long, the frequency of interruption to treatment for people already initiated on ART will be minimized. Modern surveillance and inventory management business practices are needed, integrated into MISAU systems even as these systems are upgraded, attempted with some success in other Mozambican health contexts.^14,28–31 The 2018–2019 data suggest that our CQI efforts need to be revitalized to renew the salutary trends seen in the 2014–2015 time period.

The goals of PEPFAR 3.0 and ACT are undermined by inadequate pharmacy inventories.^32–34 This problem is familiar to anyone engaged in primary care and provision of essential medicines such as antimalarial drugs in low and middle income counties.^35–46 Starting HIV-infected or HIV/TB coinfected persons on therapy but then failing to sustain their continuous therapy when the medications are unavailable nurtures the development of medication resistance and certainly contributes, we hypothesize, to loss to follow-up. In our program, this was especially urgent for pediatric ARV medications and anti-TB medications.

Our real-time monitoring and evaluation of pharmacy system performance highlights the urgency of pharmacy and pharmacy supply chain capacity building, the topic of much innovation in programs elsewhere in sub-Saharan Africa.^47–50 We believe that our findings, demonstrating drug and test kit shortages particularly for selected anti-TB drugs and pediatric ARVs, reinforce how urgent a challenge this is, even within the world's largest HIV donor investment, PEPFAR. Continued surveillance of pharmacy stocks is a vital tool for system improvement, identifying supply chain system vulnerability amenable to quality improvement evaluation and corrective action.^51–53 Our struggle with this dilemma in 2018–2019 is shared in other low-resource settings.^54–57

Authors' Contributions

Co-authors M.P.B., A.S.M., and M.F.S.A. provided data collection for the study and M.B.P. provided study design and statistical support, all in addition to their contributions to article writing and review. E.M., A.F.G., C.W.W., and S.H.V. each provided contributions to article writing and review.

Data Sharing

The Government of Mozambique controls their own data. If they approve a particular request, data are available from the authors.

Funding Information

This work was supported, in part, by the President's Emergency Plan for AIDS Relief (PEPFAR) through the Centers for Disease Control and Prevention (CDC) under the terms of number U2GGH002071, formerly U2GGH000812, and by the National Institutes of Health/National Institute of Allergy and Infectious Diseases (NIH/NIAID) supported Tennessee Center for AIDS Research grant number P30AI110527. Dr. Vermund was funded, in part, by National Institutes of Health/National Institute of Mental Health (NIH/NIMH) grant number P30MH062294.

Supplementary Material

Supplementary Table S1