The numbers landed on my desk three years ago and I still remember the quiet that settled in the room. Not because they were shocking — though some were. But because they told a story that no press release, no policy brief, no carefully worded ministry statement had managed to say out loud.
Nepal legalized abortion in 2002. Think about it: that's not new. What's newer is what happens after the law changes, after the clinics open, after the training manuals get printed. The 2021 health management information system (HMIS) data on post-abortion complications by province doesn't just sit in a spreadsheet. It breathes. It moves. And if you know how to read it, it tells you exactly where the system is holding — and where it's cracking Small thing, real impact..
What the 2021 HMIS Data Actually Shows
Let's start with what we're looking at. So the Health Management Information System, run by the Ministry of Health and Population, aggregates monthly reports from public facilities across all seven provinces. Here's the thing — the 2021 fiscal year (mid-July 2020 to mid-July 2021) captured over 98,000 abortion-related service visits. Of those, roughly 11,300 were recorded as post-abortion complication cases Easy to understand, harder to ignore..
That's about 11.Even so, on paper, it's a rate. 5 percent. In practice, it's a map.
Province 1: High Volume, Mixed Signals
Province 1 reported the highest absolute number of complication cases — just over 2,400. 8 percent. But it also recorded the highest total abortion service volume. On top of that, when you adjust for volume, the complication rate sits around 10. Slightly below the national average.
Here's what's interesting: Morang and Sunsari districts, both with major referral hospitals, drove a disproportionate share of the complications. So naturally, not because care is worse there. Because they're referral hubs. Plus, women arrive from the hills — Ilam, Panchthar, Taplejung — after something goes wrong at home or at a lower-level facility. Because of that, the data doesn't distinguish between complications that started at the referral hospital and those that arrived there. Now, that distinction matters. A lot.
Madhesh Province: The Silent Spike
Madhesh Province tells a different story. In real terms, lower total volume — about 14,000 abortion visits — but a complication rate pushing 14. 2 percent. The highest in the country The details matter here..
Why? Even so, the data doesn't say. The lowest female literacy. The highest early marriage rate. But the context fills in blanks. Still, madhesh has the lowest facility delivery rate in Nepal. And a health system that's chronically understaffed — especially for reproductive health cadres. Many women still access medication abortion through informal pharmacies without clinical follow-up. When bleeding doesn't stop, they show up at district hospitals in Siraha, Saptari, Rautahat — often late, often unstable Worth knowing..
The 2021 numbers don't capture the near-misses. The women who made it. The ones who didn't reach a facility at all.
Bagmati: Low Rate, High Complexity
Bagmati Province — home to Kathmandu — reported the lowest complication rate at 7.Think about it: looks good on a dashboard. 9 percent. But dig into the district-level data and the picture fractures.
Kathmandu, Lalitpur, and Bhaktapur pull the average down. In practice, they have 24/7 emergency obstetric care, blood banks, trained providers, and private clinics that report into HMIS. But move to Sindhupalchok, Rasuwa, or Dolakha — same province, same year — and the rate climbs past 13 percent. In practice, same policy. Same legal framework. Completely different reality But it adds up..
The province-level aggregate hides more than it reveals. And that's not a flaw in the data. It's a feature of how we choose to read it.
Gandaki and Lumbini: The Middle That Isn't
Gandaki (9.Even so, 1 percent) and Lumbini (10. That's why 3 percent) sit comfortably near the national average. Plus, easy to overlook. But Lumbini's Banke and Bardiya districts — bordering India, with high cross-border mobility — showed a 30 percent jump in complication reports between 2020 and 2021. Was it better reporting? More actual complications? A shift in where women seek care? The dataset doesn't say. But the spike is real, and it didn't make the annual report's executive summary.
Quick note before moving on.
Gandaki's mountain districts — Manang, Mustang — reported zero complications in 2021. Zero. Also, because the reporting system doesn't reach there reliably. Not because abortion doesn't happen there. Or because women travel to Pokhara or beyond, and the complication gets recorded at the destination, not the origin Surprisingly effective..
Karnali and Sudurpashchim: The Data Desert
Karnali Province reported a 12.Sudurpashchim: 13.8 percent. Also, 6 percent complication rate. Both have the fewest facilities per capita, the longest travel times to comprehensive abortion care (CAC) sites, and the highest proportion of second-trimester procedures — which carry inherently higher complication risks Turns out it matters..
Short version: it depends. Long version — keep reading That's the part that actually makes a difference..
But here's the kicker: both provinces also had the lowest absolute numbers of reported abortion services. Sudurpashchim: just over 5,000. That said, karnali: under 4,000 total visits. But in a combined population of nearly 3. 5 million women of reproductive age.
That's not low demand. Worth adding: that's low access. And low reporting Most people skip this — try not to..
Why This Data Matters — And Why It's Easy to Misread
The 2021 HMIS dataset is the most comprehensive national picture we have. But it's not a prevalence study. It's a facility-based reporting system. In real terms, every number represents a woman who made it to a government facility and got recorded. That's a filter. A powerful one And that's really what it comes down to..
The Denominator Problem
Complication rates need denominators. The HMIS gives us "total abortion service visits" as the denominator. But that includes:
- Medication abortion follow-ups
- Post-abortion family planning counseling
- Manual vacuum aspiration (MVA) procedures
- Dilation and evacuation (D&E) procedures
- Treatment for complications (which gets counted again)
A woman who takes misoprostol at home, bleeds heavily, goes to a health post, gets referred to a district hospital, and receives MVA — she might appear in the data two or three times. Even so, as a medication abortion client. Still, as a complication. On the flip side, as a surgical evacuation. The system wasn't built to deduplicate her journey And that's really what it comes down to..
The Classification Mess
"Post-abortion complication" in HMIS bundles everything: incomplete abortion, hemorrhage, infection, uterine perforation, cervical laceration, shock. No severity grading. No etiology. A woman with retained products managed with a single dose of misoprostol at a health post sits in the same column as a woman who arrives at a zonal hospital with septic shock after an unsafe procedure The details matter here..
They're not the same. But the data treats them that way Easy to understand, harder to ignore..
The Private Sector Blind Spot
Over 60 percent of abortions in Nepal happen outside public facilities — private clinics, NGOs, pharmacies. Most don't report to HMIS. Practically speaking, when complications from those services land in public hospitals, they do get counted. But the numerator (complications) gets inflated while the denominator (total procedures) stays artificially low. Also, the rate looks worse than it is. Or better. Depending on which way the referral flows.
People argue about this. Here's where I land on it.
How the System Responds — And
How the System Responds — And Where It Doesn't
The Ministry of Health and Population knows these gaps. On the flip side, the 2020 Safe Abortion Service Program (SASP) guidelines explicitly mandate complication tracking, referral protocols, and quarterly facility audits. But implementation lives in the gap between Kathmandu directives and a health post in Humla where the auxiliary nurse midwife hasn't received refresher training in three years Easy to understand, harder to ignore..
The Referral Chain That Isn't
On paper, the referral pathway is clean: Health Post → Primary Health Care Center → District Hospital → Zonal/Provincial Hospital → Central Hospital. Each level has defined capacity: MVA at health posts, MVA plus limited D&E at PHCCs, full CAC including second-trimester at district hospitals and above.
At its core, where a lot of people lose the thread.
In practice, the chain fractures at the first link. A 2022 UNFPA facility assessment found that 41 percent of health posts in Karnali lacked functional MVA kits. Even so, thirty-seven percent had no provider trained in medication abortion follow-up. That's why when a complication arises — say, incomplete abortion with moderate bleeding — the health post refers. But the ambulance is broken. Think about it: the district hospital is six hours away on a dirt road that monsoon landslides have cut. The woman's family decides to wait. Or they go to a traditional healer. Or she dies at home, uncounted That's the part that actually makes a difference..
The HMIS captures none of this. It only knows what walks through the door.
The Training Paradox
Nepal has trained over 12,000 providers in safe abortion techniques since 2004. Which means impressive. But training ≠ competency retention. A 2021 study in BMC Health Services Research found that providers who performed fewer than five MVAs per quarter lost procedural confidence within six months. In remote facilities, annual caseloads often fall below that threshold No workaround needed..
The response? But a health post in-charge in Bajura can't leave for three days — there's no replacement. Effective for those who attend. Periodic "skills labs" at provincial hubs. The system trains people, then strands them Worth keeping that in mind..
The Commodity Trap
Misoprostol and mifepristone are on the essential medicines list. LMIS shows 500 combi-packs delivered. Here's the thing — hMIS shows it. Federal procurement exists. But the logistics management information system (LMIS) doesn't talk to HMIS. Still, a district hospital in Doti reports zero medication abortion visits for four months. Where did they go?
No fluff here — just what actually works.
Stockouts at the point of care. Now, expired stock not rotated. Diversion to private pharmacies. The data systems don't connect, so the accountability loop stays open The details matter here..
What the Numbers Can Tell Us — If We Ask Differently
Despite the noise, signals emerge when we stop treating HMIS as a census and start treating it as a sentinel.
Geographic Clustering Is Real
The Karnali-Sudurpashchim pattern holds across three reporting years. It's not random. These provinces share: lowest road density, highest poverty quintiles, lowest female literacy, highest child marriage rates. The complication rate isn't a clinical failure — it's a structural one. Women present later, sicker, because every barrier delayed them: distance, cost, stigma, spousal permission, lack of knowledge that abortion is legal up to 12 weeks (18 for rape/incest, 28 for fetal anomaly).
The Second-Trimester Signal
Nationally, second-trimester procedures constitute 8 percent of reported services but 34 percent of reported complications. Think about it: in Karnali, that share jumps to 22 percent of services. The math is brutal: later gestation → more complex procedures → higher complication risk → fewer facilities equipped to manage them → longer referrals → worse outcomes Easy to understand, harder to ignore..
Quick note before moving on Simple, but easy to overlook..
This isn't a training gap. It's an access gap masquerading as a clinical one.
The Medication Abortion Paradox
Medication abortion (MA) now accounts for 68 percent of first-trimester services nationally. Consider this: complication rates for MA are lower than surgical — but absolute numbers of MA complications are rising because volume is rising. And the complications that do occur — incomplete abortion, prolonged bleeding — require follow-up that the system struggles to provide Surprisingly effective..
The 2021 data shows MA follow-up completion rates below 40 percent in mountain districts. The system assumes success. Women take the pills. They don't return. The reality is unknown.
From Data to Action: Three Non-Negotiables
1. Fix the Denominator — Now
Nepal needs a unique identifier for abortion care episodes. Not patient names — a hashed, facility-generated episode ID that links: initial visit → method → follow-up → complication (if any) → outcome. Pilot it in five districts. Day to day, use the existing DHIS2 platform. The technology exists Less friction, more output..
ortion reporting system is what’s lacking. Without this, we’re measuring symptoms, not the disease.
2. Decentralize the Follow-Up — The current model assumes women will return for post-abortion care, but in reality, they don’t. Why? Because the system hasn’t adapted to their lives. Mobile clinics, community health workers, and telehealth consultations must be integrated into the abortion continuum. In Karnali, where roads are unpaved and weather disrupts travel, a woman shouldn’t have to choose between her health and her livelihood. The follow-up must come to her — not the other way around.
3. Map the Unmapped — The data we have is siloed, incomplete, and often outdated. We need a real-time dashboard that aggregates HMIS, LMIS, and even pharmacy supply chain data into one platform. This would allow district health offices to see not only how many abortions are being performed, but where medications are being diverted, where stockouts are chronic, and where complications are clustering. With this intelligence, resources can be pre-positioned, not just reacted to.
Conclusion
Nepal’s abortion data is not broken — it’s being used the wrong way. In practice, we’ve built systems that tally numbers but not narratives, that count procedures but not people, that track deliveries but not destinies. The silence in the Karnali districts isn’t just about missing data — it’s about missing voices.
To truly understand and improve abortion care, we must stop asking, “How many?In practice, ” Only then can we turn data into direction — and direction into dignity. ” and start asking, “Who, where, why, and what next?Until then, the numbers will lie, and women will pay the price And it works..
