Knowing who dies with HIV, and of what, tells health officials whether treatment programs are working — with modern antiretroviral therapy, people on treatment can expect a near-normal lifespan, so an HIV death can be the direct result of missing or ineffective treatment. Rich countries get these numbers from civil registration and vital statistics linked to HIV case reporting. Many sub-Saharan African countries cannot: their death registration is weak and its data poor.
Kenya keeps birth and death records on paper, which makes them hard to retrieve and analyze, and it does not report mortality or cause of death to the World Health Organization. Yet the Ministry of Health counts HIV as the leading cause of death among adults, and in 2016 national HIV prevalence among people aged 15–64 was 5.4%. One alternative is to go where the bodies are: the mortuary. In 2015, Kenya piloted HIV surveillance in Nairobi's mortuaries, and CDC evaluated it that November.
The pilot
| Where | the two largest mortuaries in Nairobi — Kenyatta National Hospital (KNH), attached to the national referral hospital, and Nairobi City mortuary. Nairobi County has more people living with HIV than any other county |
| When | 33 days, January 29 – March 3, 2015 |
| Who | the body of everyone aged 15 or older admitted in that time |
| What was done | details entered in a register (age, sex, county, circumstances, place and time of death); blood drawn from the heart through the chest wall of unpreserved bodies; HIV testing at a central laboratory using Kenya's national three-test algorithm, with viral load measured on positives; cause of death taken from KNH medical records or autopsy reports |
| Aims | HIV positivity among the dead, and yearly cause-specific and HIV-specific death rates |
What it found
807 bodies came in; 610 had an HIV result. The rest had no specimen — for reasons of logistics and staffing (91), burns or decomposition (26), legal release, discharge or transfer (31), or a failed attempt (19) — or a specimen too poor to test (30).
| Finding | Result |
|---|---|
| HIV-positive, unadjusted | 19.5% (119 of 610) |
| Men | 14.6% |
| Women | 29.5% |
| KNH mortuary | 23.2% |
| Nairobi City mortuary | 12.6% |
| Adjusted to the age and sex of expected deaths in Nairobi | 20.9% |
| Standardized mortality ratio, HIV-infected against uninfected adults | 4.35 |
Of the people with HIV who died in KNH and had medical files (90), 24.4% had been diagnosed before admission, 22.2% during their stay — and 53.3% were never diagnosed before they died.
Compared with the model
UNAIDS's Spectrum model is the usual source for these numbers. The mortuaries told a different story:
| Pilot, adjusted | Spectrum | |
|---|---|---|
| HIV-positive among the dead | 20.9% | 11.4% |
| — men | 14.8% | 13.8% |
| — women | 30.2% | 8.3% |
| Deaths caused by HIV | 12.6% | 8.4% |
| — men | 6.6% | 10.9% |
| — women | 25.4% | 5.3% |
Spectrum predicts higher HIV positivity among men who die than among women, because fewer men are estimated to be on treatment; the mortuaries found women's rate higher.
How the system performed
CDC judged it against its standard guidelines for evaluating surveillance, reviewing documents and the database and interviewing 20 people — mortuary and laboratory staff, the Ministry of Health, and funding and implementing partners.
| Attribute | Verdict |
|---|---|
| Simplicity | a straightforward design, written procedures and a day's classroom training plus supervised practice — but complex in practice: staff did not always follow the inclusion rules, specimens taken after hours and at weekends had to be rushed to the lab, there were several extra forms on top of routine paperwork, and KNH files were hard to find. Meetings, clearer procedures, better mortuary–lab communication and overtime pay addressed these |
| Flexibility | high: switching to a longer needle to reach the heart, drawing blood after registration but before cold storage, keeping hemolyzed samples once it was clear they tested fine, and revising the data forms — all without delays or extra cost |
| Data quality | nearly complete for admission date, age and sex; viral load for 68% of positives; cause of death the weak point, hard to get for anyone who died outside hospital without an autopsy |
| Acceptability (staff only) | low at first, because blood-drawing added work; it rose once the Ministry explained the purpose and staff were paid about $3.50 a day in overtime |
| Sensitivity | could not be measured: nothing records the true number of HIV cases or HIV deaths in Nairobi |
| Representativeness | limited. The two mortuaries registered 54.2% of the 12,796 adult deaths recorded in Nairobi County in 2014, but some deaths happened outside Nairobi (KNH takes referrals from across the country), some Nairobi deaths go to mortuaries elsewhere, people sometimes give an ancestral home as their residence, the hospital mortuary will see more HIV than the population does, and one month cannot show seasonal patterns |
| Timeliness | collection began about 3.5 months after the protocol was approved; specimens reached the lab in a median of about an hour; but the aim of drawing blood within 48 hours of death could not be checked, because time of death was rarely written down. Data entry and cleaning took about 6 months; the final report came out in July 2016 |
| Stability | the pilot cost about $50,000 (excluding mortuary and lab salaries); five cities with two mortuaries each might cost about $250,000, against about $7.5 million for the 2012 Kenya AIDS Indicator Survey. But mortuary surveillance is not in Kenya's national AIDS strategy or health policy, so it depends on outside funding |
Why it matters
- Epidemic control — the point where new HIV infections fall below deaths among people with HIV — cannot be measured without death data, and the UNAIDS 95–95–95 targets (95% of people with HIV knowing their status, 95% of them on treatment, 95% of those virally suppressed) leave mortality out. Mortuary surveillance can fill that gap.
- Its data could calibrate Spectrum, which in countries without good death registration has to estimate mortality from other data.
- Kenya could also strengthen ordinary death registration: automatic tabulation of cause of death at registration, a health practitioner at every death, and ICD-10 coding — with the staff training that requires.
Next: with PEPFAR funding, Kenya planned to repeat the system in Western Kenya, a high-prevalence region, testing oral fluid alongside blood. Oral fluid would make testing non-invasive, allow children to be included (drawing heart blood from a child's body is difficult), possibly work on preserved bodies, and cut costs.
Limits of the evaluation
- Sensitivity and predictive value could not be measured, for lack of data on the true number of cases.
- For most bodies, it could not be established independently whether the person died of HIV or simply had HIV when they died.
Sources
Based on Ali H, Kiama C, Muthoni L, et al., "Evaluation of an HIV-Related Mortuary Surveillance System — Nairobi, Kenya, Two Sites, 2015," MMWR Surveillance Summaries volume 67, number SS-14, Centers for Disease Control and Prevention; a work of the United States government in the public domain. The report gives three different figures for how often cause of death was recorded (46.5% in its abstract, 44.2% in its table, "approximately half" in its text), so this page gives none; and it prints the standardized mortality ratio with a percent sign ("4.35%"), shown here as the ratio it describes.
Licence: CC0 1.0 (public domain) · Adapted from www.cdc.gov
1
0
0
0
Comments
