Autopsies and better data on causes of death in Africa

The global burden of tuberculosis remains immense. In 2014, WHO reported that as many as 3 million people with active tuberculosis remained undiagnosed and thus untreated, 1·5 million people died from tuberculosis in 2014, and the situation with multidrug resistant tuberculosis is worsening.⁵ These estimates remain imprecise because of ongoing challenges with diagnosis, inadequate case reporting, and the poor state of vital statistics and death registration in most low-income and middle-income countries. More accurate estimates of the burden of tuberculosis and the number of deaths from tuberculosis are crucial for appropriate resource allocation to improve case detection, reduce tuberculosis transmission, and achieve global tuberculosis control.

The results of the meticulous autopsy study in Zambia by Matthew Bates and colleagues⁶ published in The Lancet Infectious Diseases are thus timely and important. Bates and colleagues highlight the substantial burden of active tuberculosis in adult patients (≥16 years) who died on adult general medical wards at a tertiary care centre in Lusaka. Overall, over 80% of patients had underlying HIV infection. Active tuberculosis was present in nearly two-thirds of patients, with most having substantial comorbidity that probably contributed to death. Nearly 20% of patients had multidrug-resistant tuberculosis. Health-seeking behaviour seemed compromised because the median time from admission to death was 7 days, suggesting that most patients who died were very sick when they arrived at the wards and there was insufficient time for diagnostic investigations and therapeutic interventions. A quarter of the patients with active tuberculosis were only diagnosed at autopsy and were not on anti-tuberculosis treatment. The attending physicians judged tuberculosis to be the cause of death in just 60% of those with active tuberculosis found at autopsy.

From a service perspective, Bates and colleagues concluded that increased clinical awareness of active tuberculosis is needed in the general medical inpatient services, backed up by more proactive screening for tuberculosis and multidrug resistant tuberculosis. Since most patients who died with undiagnosed tuberculosis had presented very late to the hospital, the effect this would have on reducing tuberculosis mortality would probably be small unless substantial efforts are made to support better health-seeking behaviour, coupled with strategies to improve the health-care cascade with earlier identification of active tuberculosis at the primary care level. Patients who end up in a tertiary centre are not representative of the patterns of morbidity and mortality in the general population, and without data on the rates of multidrug-resistant tuberculosis across Zambia, whether tuberculosis screening in all general medical services should routinely include drug resistance testing or not is unclear. Adding this test would have substantial cost and resource implications and highlights the need for population-level surveillance data in addition to autopsy data to put into context many of the study's key findings.

From a system perspective, the findings of the study by Bates and colleagues⁶ further challenges the assumption that causes of death are being accurately recorded and reported in hospitals.⁷ Health systems worldwide depend on reliable information about causes of mortality to be able to effectively respond to changing epidemiological circumstances and appropriately allocate resources. Unfortunately, the poor state of health information systems, particularly mortality statistics, is widely documented.⁸ The gold standard for cause-of-death reporting is to have the cause certified by a medical practitioner using the rules and procedures of the International Classification of Diseases (ICD), which is available in its tenth revision.⁹ Unfortunately, clinicians might not have the time, incentives, diagnostic facilities, or training to correctly certify causes of death, and they seldom appreciate that these diagnoses guide national health priorities.¹⁰ In this regard, information on how thoroughly this university teaching hospital routinely reports and codes inpatient deaths, and whether the additional autopsy data were fed back and included to improve the accuracy of hospital death reports, would have been useful.

Ultimately, one study alone has limited ability to improve patient outcome and contribute to better appreciation of the burden of tuberculosis disease and deaths in Zambia. Rather than recommending further autopsy studies to ascertain the validity of these findings, we argue that the best investment in the short-to-medium term would be to develop strategies to improve the accuracy, recording, and reporting of all causes of death. Autopsies provide the most accurate data about cause of death, but are prohibitively expensive and impossible to scale up to cover a whole country.¹¹ Improved training of selected physicians using ICD-compliant death certification guidelines to review medical records in a sample of hospitals is more practical and affordable, and can cover more facilities.⁷ Funding is likely to be available, because the World Bank with WHO have recently produced a plan to scale up investment in global civil registration and vital statistics to help address the deficit in this crucial but neglected area.¹² This comprehensive autopsy study could be the catalyst to unlock funds to improve Zambia's health information system, with particular reference to cause-of-death reporting. That would be an unexpected but extremely important outcome, with benefits above and beyond tuberculosis control for general health planning and policy development.

By Rasika Rampatige, Charles F Gilksemail

We declare no competing interests.

References