Farah had taped three large sheets of paper to the wall in the records room. At the top of the first she wrote `WHAT WE KNOW`. The second said `WHAT WE ASSUME`. The third: `WHAT WE NEED TO KNOW`.
When Malik arrived at eight that evening, Achieng was already at the table with a stack of referral files from the previous three months. The room smelled of paper, dust, and toner. Outside, the hospital had quieted a little. Inside, the argument around Kevin was growing larger.
Farah pointed at the first column. “Fact: the first call to Bahari was at 08:15. Fact: the 08:31 callback did not reach the emergency room. Fact: the transfer decision was documented at 08:47. Fact: the ambulance left at 09:16.”
Achieng added, “Fact: the KPI memo says manage locally whenever possible.”
Farah wrote it down.
“And fact,” Malik said, “there is no memo saying referrals should be delayed because of payment.”
That went into the first column too.
“Assumption?” Farah asked.
“That staff delayed because of the KPI,” Achieng said.
Malik added, “That the first ECG was enough to mandate immediate transfer.”
“Need to know?”
“The pattern of referral ownership,” Malik said. “And whether similar gaps happened even without the payment controversy.”
Farah tapped the stack of files. “That is why we’re here.”
They had selected twenty referrals: emergency, maternity, pediatrics, and trauma. Not all had been delayed. The purpose was not to prove a theory. It was to see whether the system behaved in a predictable way.
The first file was clean. The trigger was documented, the consultant answered, and the ambulance left within twenty minutes. The second contained a gap of almost forty minutes between `consider referral` and the formal order. No harm had been recorded.
“Is this a delay?” Achieng asked.
“We don’t have enough context,” Malik said. “Put it under need to know.”
Farah marked it with a small code: `ownership unclear`.
The fifth file involved a pregnant woman with bleeding. A nurse had documented deterioration early. A junior doctor wrote `senior informed`, but there was no time recorded for the senior response. Transfer happened later.
“There’s no payment issue here,” Farah said.
“And no Bahari bed problem,” Achieng added.
Malik read the notes twice. “But ownership still disappears. Who had the next action after `senior informed`?”
Nobody had an answer.
They kept going. As the night deepened, the pattern was not dramatic enough for radio. There was no secret order. There was no single person blocking referrals. There was something more dangerous because it was ordinary: sentences without an owner. `Observe.` `Discuss with senior.` `Await callback.` `Consider transfer.` Each sounded reasonable on its own. Together they could create minutes that belonged to nobody.
Farah wrote on the board: `AMBIGUOUS VERB = AMBIGUOUS OWNER`.
“Let’s not rush to a conclusion,” Malik said.
“It’s a hypothesis,” Farah replied.
Achieng laughed. “Now we’re researchers.”
“Better than prosecutors.”
Another file involved a young man with a head injury. Bahari City had no immediate CT slot. The Mwangaza team kept monitoring and called every thirty minutes. Documentation was excellent because one nurse had recorded every call.
“This shows scarcity itself doesn’t have to produce silence,” Malik said.
“The owner was Achieng,” Farah said.
Achieng frowned. “I was on that shift.”
“Exactly. You wrote the next action every time.”
“Because I was tired of receiving callbacks when nobody knew what had happened on the previous call.”
Malik looked at her. “Then your workaround may be a design principle.”
“Don’t build policy out of my frustration.”
They all laughed, but Farah became serious again.
“If we fill the gap with anger, we create a new mistake,” she said. “The public wants a culprit. The board may want one sentence rewritten so they can say the problem is solved. The evidence says the problem has layers.”
Malik listed them: communication channel, ownership, escalation criteria, bed scarcity, interpretation of metrics, family understanding.
“And pride,” Achieng said.
“Pride is hard to measure.”
“But it exists.”
Farah added `culture signals` under `need to know`.
By midnight they had reached file sixteen. Malik’s eyes felt heavy. Achieng wanted to stop, but Farah pulled out one more file.
“This is the near-delay I told you about,” she said.
It was from two months earlier. The patient had a severe infection and falling blood pressure. The first note read `review senior, possible HDU transfer`. Fifty-two minutes later another note said `senior agrees transfer`. Nothing was documented between the two.
“Outcome?” Malik asked.
“Reached Bahari and recovered.”
“Payment issue?”
“None.”
“Was the KPI memo already in effect?”
“It was still a draft. It hadn’t been circulated.”
Malik sat quietly. That mattered. If the same kind of gap existed before the current wording, the memo could not be the only root cause.
“So Wekesa will use this to say the KPI is not the problem,” Achieng said.
“And others will use Kevin to say the KPI is the problem,” Farah replied. “We will say what the evidence shows.”
Under `known`, Malik wrote: `Escalation ownership vulnerability predates current KPI.` Then under `assumed`: `KPI may amplify existing hesitation.`
“That is less satisfying than a headline,” Achieng said.
“But more useful.”
Farah proposed formally expanding the review from one case to a three-month sample. Malik understood the cost: more files, more interviews, more chances for management to feel the hospital was being opened up in public.
“If we expand it, we won’t finish quickly.”
“Root causes do not have to respect the press cycle,” Farah said.
They sorted the files by codes instead of patient names. Malik lifted a stack to return it to the shelf, but Farah stopped him and pulled out one chart they had not discussed in detail.
“Look at this first.”
It was a referral from three weeks earlier. The same phrase: `await callback`. The same lack of an owner. The same gap of almost thirty minutes. The patient recovered, so nobody had asked questions.
Farah pushed the chart across the table toward Malik.
“Kevin was not the first.”
---