In a district pharmacy in eastern Uganda, the shelves hold fluoxetine and amitriptyline, stocked through the national medical stores. The drugs arrived, as they have across many districts, after years of advocacy and a push to integrate mental health into primary care. But when the district supervision team makes its quarterly visit, the checklist rarely includes the dispensing bench. The team counts boxes, checks expiry dates, and inspects the cold chain. No one sits with the nurse to watch how a patient with depression is counselled, how the dose is explained, or whether a follow-up appointment is scheduled. The pills are present. The practice around them is not.

A Shelf Full of Hope, a Bench Left Unwatched

Uganda's central medical stores have steadily expanded the list of essential medicines, and antidepressants are no longer a rare sight in district pharmacies. Fluoxetine, amitriptyline, and occasionally sertraline appear on shelves that a decade ago held little beyond painkillers and antimalarials. For clinicians in rural health centres, this is a tangible victory: a patient diagnosed with depression can now leave with a prescription rather than a referral to a distant hospital.

Yet the same supervision system that ensures vaccines are kept cold and antibiotics are not expired has no routine mechanism for watching how these psychiatric drugs are dispensed. A typical supervision visit, as described by district health officers, involves checking stock registers, verifying that reporting forms are complete, and inspecting infrastructure. The dispensing bench, where a nurse hands over the medicine and perhaps a few words of advice, is often the last place a supervisor looks.

The result is a quiet divergence between policy and practice. The policy says antidepressants should be prescribed with counselling and follow-up, as outlined in the World Health Organization's Mental Health Gap Action Programme (mhGAP). The practice, observed in many facilities, is that the prescription is written, the drug is handed over, and the patient is sent home with little more than a date to return if symptoms do not improve. Supervision, which could catch this gap, instead reinforces the focus on supply.

Clinicians on the ground feel this tension. A nurse in a health centre in the east told me that she screens patients with the PHQ-9, a standard depression questionnaire, and prescribes fluoxetine when the score is high. But she has no way to track whether the patient returns, whether the dose is adequate, or whether side effects are managed. The supervision team, she said, asks about stock levels, not about the patients who received the stock.

The Evidence Demands More Than a Pill

Antidepressants are effective, but the evidence is clear that they work best when combined with regular review and psychosocial support. Meta-analyses of trials in low- and middle-income countries, where most of the world's depression burden lies, show that medication alone produces modest effect sizes. The gains are larger when the drug is paired with structured follow-up, whether that is a brief counselling session, a phone call, or a scheduled revisit.

The WHO mhGAP guidelines, which Uganda has adopted, are explicit about this. They recommend that antidepressant initiation should be accompanied by psychoeducation, that side effects should be monitored, and that a review appointment should be offered within a few weeks. The guidelines also stress the importance of adherence, noting that many patients stop medication prematurely, often because of side effects or a belief that the drug is no longer needed.

Monitoring for side effects is not a luxury; it is a safety requirement. Antidepressants can cause nausea, insomnia, agitation, and, rarely, more serious effects such as hyponatremia or serotonin syndrome. In settings where laboratory monitoring is scarce, the clinical review is the only safety net. Without it, a patient who develops troubling side effects may simply stop the drug, no one will know, and the depression may worsen or return.

Brief interventions, such as a five-minute adherence check or a simple visual analogue scale for mood, have been shown to improve outcomes. These are cheap and feasible in district settings. Yet they are rarely part of the routine, because supervision does not require them and clinicians are not trained to deliver them consistently.

Why Supervision Tours Miss the Counter

Supervision visits in Ugandan districts are under-resourced and overstretched. A district health officer may be responsible for dozens of facilities, with a small team and a limited budget for transport. The visits are often rushed, and the focus is on the things that can be counted: stock, reports, and infrastructure. The dispensing bench, where patient interaction happens, is not easily counted.

Checklists used during supervision visits typically include items on drug availability, storage conditions, and reporting completeness. They rarely include questions about prescribing quality, such as whether the prescriber checked for contraindications, whether the patient was told about side effects, or whether a follow-up appointment was made. A supervisor could observe a dispensing interaction, but the checklist does not prompt it, and the time pressure makes it an afterthought.

Most supervisors have limited mental health training. They are often generalist health workers or administrators whose expertise lies in logistics and public health. They may feel uncomfortable evaluating a clinical interaction involving a psychiatric drug, unsure of what good practice looks like. Without training, they default to what they know: counting pills.

The reporting forms themselves reinforce this bias. They ask for quantities of drugs dispensed, stock-outs, and expiry dates. They do not ask about patient outcomes, adherence rates, or the number of patients who returned for a follow-up. The data system rewards procurement and supply, not care. As one district pharmacist put it, the forms would tell you if a facility ran out of fluoxetine, but not if a single patient got better.

The Clinician's Dilemma in Real Practice

Nurses and clinical officers in district facilities are the de facto mental health providers. They screen with PHQ-9, a tool that is quick and validated, but their training in psychiatric prescribing is minimal. They are taught to start low and go slow, but they are not always sure what to do when a patient does not respond or reports side effects.

Patients often travel long distances to reach a health centre. A follow-up visit may mean a day of lost wages and transport costs. For many, the default is to stop the medication when the symptoms ease or when the side effects become intolerable. Without a structured follow-up, the clinician never learns of these outcomes, and the cycle of prescription and abandonment continues.

Clinicians also fear dependence. In my own conversations with health workers, several expressed reluctance to prescribe antidepressants for more than a few weeks, worried that patients would become 'addicted'. This caution, while understandable, leads to underdosing and premature discontinuation, which undermines the evidence base. The mhGAP guidance is clear that antidepressants are not addictive in the way that benzodiazepines are, but the message has not reached every clinician.

Referral pathways to psychiatric specialists are thin. In many districts, the nearest psychiatrist is hours away, and the referral may be a formality rather than a genuine option. The primary care clinician is the only provider, and the supervision system does little to support them in this role.

Counting Pills, Not People: The Data Trap

The data that flows from district pharmacies to the central level is dominated by stock reports. Stock-out rates are tracked, and facilities are rewarded for maintaining supply. This is important, but it creates a perverse incentive: a facility that dispenses many pills, regardless of whether patients improve, looks good on paper.

Patient outcomes are rarely tracked. There is no systematic collection of depression remission rates, adherence, or functional improvement. The PHQ-9 is used for screening, but the score is not routinely recorded in a way that allows follow-up comparison. A patient who scores 18 at baseline and 12 at six weeks may be improving, but no one knows, because the second score is not captured.

Dispensing logs show quantities, not quality. They tell you how many tablets left the shelf, but not whether the right drug was given to the right patient at the right dose, or whether the patient understood the instructions. The log cannot capture a conversation.

There is no feedback loop to prescribers. A clinician who prescribes fluoxetine has no way to learn, a month later, whether the patient refilled the prescription. The pharmacy data, if it were shared, could show that, but it is not. The prescriber operates in the dark, and the supervision system does not turn on the light.

Cheap Fixes That Fit the District Reality

Improving supervision does not require a new building or a new drug. It requires a shift in what is observed and what is rewarded. A simple addition to the supervision checklist, a five-minute observation of a dispensing interaction, could yield more insight than an hour of stock counting.

A visual analogue scale for mood, a line from 'worst' to 'best', takes seconds to administer and can be tracked over time. It is not a diagnostic tool, but it provides a crude measure of change that is better than nothing. Pairing it with a brief counselling script, a few standard sentences about side effects and the importance of continuing, could improve adherence at negligible cost.

Training supervisors in basic mental health triage is feasible. A one-day workshop on what good prescribing looks like, what side effects to ask about, and how to spot a patient at risk could transform the supervision visit. The supervisors would still count pills, but they would also ask about people.

Peer supporters, people with lived experience of depression who are trained to provide support, have been used successfully in other low-income settings. They can make follow-up phone calls, remind patients about appointments, and offer encouragement. In Uganda, where mobile phone coverage is widespread, a simple call log could be a low-tech solution to the follow-up gap.

Beyond the Checklist: Reimagining Supervision

The supervision visit itself could be reimagined as a learning opportunity rather than a policing exercise. In some districts, a move toward 'supportive supervision' has been piloted, where the supervisor's role shifts from inspector to mentor. During these visits, the supervisor might sit with the nurse during a patient consultation, observe the interaction, and then discuss what went well and what could be improved. This approach, borrowed from successful programs in other countries, requires a smaller supervisor-to-facility ratio and a different set of skills, but it has the potential to build capacity rather than merely enforce compliance.

One challenge is the fear of blame. Nurses may be reluctant to have a supervisor observe their work if they worry about being penalized for mistakes. Building trust is essential. In a pilot in one district, supervisors began by observing only the most experienced nurses, then gradually included others as the team became comfortable. The feedback was framed as 'we are learning together', not 'you are doing it wrong'. This subtle shift made the observation acceptable, and over time, the nurses began to request feedback on specific cases.

Another practical step is to integrate mental health indicators into the existing District Health Information System (DHIS2). Currently, the system captures aggregate data on outpatient visits and drug consumption, but not on mental health outcomes. Adding a few simple fields—such as the number of patients initiated on antidepressants, the number who returned for a follow-up within four weeks, and the number who reported side effects—would provide a low-cost way to track progress. These indicators are not perfect, but they would create a baseline and allow districts to see trends over time.

There is also a role for community health workers. In many areas, these volunteers are the first point of contact for patients with mental health complaints. Training them to recognize the signs of depression, to encourage adherence, and to report back to the health centre could extend the reach of the system. One program in northern Uganda trained community health workers to use a simplified version of the PHQ-2, a two-question screen, and to refer patients with positive screens to the health centre. The program increased the number of patients seeking help, but the follow-up still depended on the health centre's capacity. Supervision could include checking whether these referrals are actually happening and whether the health centre is responding.

The cost of these changes is modest compared to the cost of the drugs themselves. A day of training for supervisors might cost a few hundred dollars per district, and the redesign of forms is a one-time expense. The potential benefit—better patient outcomes, reduced disability, and fewer wasted medications—is likely to far outweigh the cost. Yet the health system, already stretched thin, may resist any additional burden. The key is to frame these changes not as extra work but as a way to make existing work more effective.

When the Bench Becomes the Standard

The dispensing bench is a metaphor, but it is also a literal place. It is where the patient meets the system, where the prescription becomes a conversation, and where the drug becomes a treatment. Making that bench a regular stop on the supervision tour would signal that mental health care is not just about supply.

Integrating mental health into routine audits is another step. The audit could ask about the proportion of patients who received a follow-up, the median PHQ-9 score change, or the number of side-effect reports. These indicators are not difficult to collect if the forms are redesigned.

A shift from counting to coaching would change the tone of supervision. Instead of a checklist that finds faults, the visit could be an opportunity for on-site mentoring, where the supervisor watches a dispensing and then offers feedback. This requires a different mindset and a different skill set, but it is within reach.

A pilot outcome registry in two districts, tracking a few hundred patients with depression, could demonstrate whether such a system is feasible and what it would cost. The registry would be simple: patient ID, baseline PHQ-9, drug prescribed, follow-up date, and a repeat score. It would not solve every problem, but it would provide a model for what is possible.

The measure of success would then be patients improved, not pills dispensed. That is a harder metric, but it is the one that matters. The antidepressants are on the shelf, a testament to years of effort. The next step is to make sure they are not just handed over, but used well.

This article is for informational purposes only and does not constitute personalised medical advice. Always consult a qualified healthcare professional for guidance on your specific situation.