WHO Training Alone Fails to Fix Mental Health Reporting in Nepal’s Clinics

A striking new study from Nepal has revealed that one of the World Health Organization’s flagship programs for closing the global mental health treatment gap is failing to produce one of its most basic prerequisites: reliable data. Researchers comparing primary health care facilities whose staff had received training under the WHO’s Mental Health Gap Action Programme, widely known as mhGAP, with facilities that had received no such training found that routine mental health reporting was critically poor in both groups. The finding, published in BMC Health Services Research, challenges a widespread assumption in global health that training frontline workers is sufficient to build functioning mental health services in low- and middle-income countries.

Mental health disorders represent a substantial and growing share of the global burden of disease, yet in many low- and middle-income countries the health systems responsible for detecting and treating these conditions operate largely blind. Routine health information systems, the administrative machinery through which clinics report diagnoses, cases, and service delivery to district and national authorities, are the foundation of any evidence-based response. When those systems fail to capture mental health data, planners cannot estimate need, allocate medication, deploy specialists, or evaluate interventions. The quality of routine mental health reporting has long been recognized as particularly weak in resource-constrained settings, but the question of whether internationally supported training programs improve that quality has remained poorly understood. The new study set out to answer it directly.

The research team, led by Mithun Kumar Jha and Binita Kumari Paudel of Purbanchal University School of Health Sciences, conducted an explanatory mixed-methods study in twenty-eight primary health care facilities in Parsa District in the southern plains of Nepal. Fourteen of the facilities had staff trained under mhGAP, the WHO program designed to scale up services for mental, neurological, and substance use disorders by equipping non-specialist health workers with evidence-based protocols. The other fourteen had no mhGAP training. The quantitative arm of the study involved a retrospective analysis of twelve months of routine mental health reporting data drawn from Nepal’s Health Management Information System, which runs on the open-source District Health Information Software 2 platform, or DHIS2, used by dozens of countries worldwide. The qualitative arm consisted of in-depth interviews with eleven health workers, analyzed through the Consolidated Framework for Implementation Research, a widely used implementation science framework for identifying barriers and facilitators to putting evidence-based programs into practice.

The quantitative results were unambiguous and, in places, startling. Timeliness of reporting, meaning whether facilities submitted their monthly reports on schedule, was impressively high: every single mhGAP-trained facility achieved one hundred percent timeliness, and the untrained facilities were close behind at 96.4 percent. But completeness, the proportion of expected mental health data actually reported, was zero percent in both groups. Not a single facility in either arm of the study achieved complete mental health reporting over the twelve-month period, and most facilities reported no mental health cases at all across the entire year. Accuracy fared no better. The researchers found no consistency between the paper registers kept at the facilities and the data entered into DHIS2, meaning that in none of the twenty-eight facilities could the digital record be verified against the primary source documents. Because timeliness was analyzed as a continuous indicator rather than a categorical one, the authors note it was not subjected to formal statistical comparison, but the completeness and accuracy findings alone paint a bleak picture.

Perhaps the most telling contrast in the study was between structural readiness and functional performance. The mhGAP-trained facilities did show higher structural preparedness: they had trained personnel, and presumably the nominal capacity to identify and manage priority mental health conditions under the WHO protocols. Yet this structural advantage translated into no measurable improvement in the reporting domain. Whether a facility had received mhGAP training or not, its performance in routine mental health reporting was rated as poor. The authors conclude that the facilities lack functional routine reporting systems for mental health, and that training alone, however well designed, does not create the operational machinery needed to move data from a clinic register to a national database in a complete and accurate form.

The qualitative interviews, analyzed through the CFIR lens, help explain why this gap persists. Health workers described a system in which mental health remains peripheral to everyday primary care practice, with weak integration into the regular health information workflow. Without supportive supervision to reinforce training, without system-level coordination between the mental health program and the health management information system, and without mental health indicators being genuinely embedded in routine reporting expectations, the knowledge gained in mhGAP trainings had nowhere to go. A health worker may know how to diagnose depression or psychosis using a WHO protocol, but if no one checks whether those diagnoses are recorded, if the reporting forms and DHIS2 data elements are not aligned, and if supervisors never review the mental health column of the monthly report, the diagnosis simply never enters the record.

The implications reach far beyond Parsa District. The mhGAP program has been implemented in more than a hundred countries and is the centerpiece of global efforts to address the enormous treatment gap for mental, neurological, and substance use disorders, a gap that is widest precisely in the low- and middle-income countries where specialist psychiatrists are scarce. The program’s logic is elegant: train generalist health workers to deliver evidence-based care, and specialist-level treatment can be scaled without waiting for specialists. But the Nepal study exposes a critical dependency in that logic. Care that is delivered but not recorded is invisible to the health system. It cannot be counted, funded, resupplied, or improved. If routine reporting remains at zero completeness, then even successful mhGAP implementation at the clinical level would be undetectable in national statistics, and the program’s true coverage would be unknowable.

The study also carries a broader lesson for the rapidly digitizing world of global health data. Nepal’s adoption of DHIS2 represents a significant technical achievement, and the near-perfect timeliness scores show that facilities are capable of disciplined, punctual reporting when the system demands it. The failure is therefore not one of infrastructure or motivation but of functional integration: the mental health component of the information system has not been wired into the routines, supervision structures, and accountability mechanisms that make other reporting streams work. This distinction between structural readiness and functional performance is one that implementation scientists have emphasized for years, but it is rarely quantified so starkly. A trained workforce and a functioning software platform are necessary conditions for good data, and the Nepal results demonstrate they are nowhere near sufficient.

The authors call for comprehensive system-level interventions that go well beyond training. These would include regular supportive supervision that explicitly reviews mental health reporting, coordination between the mental health program and the health management information system so that indicators, registers, and digital data elements are aligned, and genuine integration of mental health into the routine reporting expectations placed on every primary health care facility. Such measures are less glamorous than international training workshops, but the evidence suggests they are where the real work lies. Without them, the study warns, reporting performance may not improve no matter how many workers are trained.

For a world preparing to confront a mental health crisis that the burden-of-disease data only partially captures, the message from Nepal is uncomfortable but valuable. The invisible fraction of the mental health burden, the cases that never appear in any register or dashboard, may be far larger than global statistics suggest, not because the cases do not exist but because the systems meant to see them are not looking. Fixing that will require treating mental health data not as an optional add-on to primary care but as a core, supervised, and audited function of every health facility, in Nepal and in every country that hopes to close the mental health gap.

Subject of Research: Routine mental health reporting performance in primary health care facilities in Nepal and the effect of WHO mhGAP training

Article Title: Mental health reporting performance in primary health care: a comparative mixed-methods study of mhGAP-trained and untrained health facilities in Nepal

Article References: Jha, M. K., & Paudel, B. K. (2026). Mental health reporting performance in primary health care: a comparative mixed-methods study of mhGAP-trained and untrained health facilities in Nepal. BMC Health Services Research. https://doi.org/10.1186/s12913-026-15782-7

Image Credits: AI Generated

DOI: 10.1186/s12913-026-15782-7

Keywords: mental health, mhGAP, Nepal, primary health care, DHIS2, health information systems, WHO, low- and middle-income countries, implementation science, CFIR, health services research, reporting completeness