The management of childhood pneumonia in settings with a high burden of HIV (part 1)

Why does childhood pneumonia matter?

This review owes much to conversations and correspondence over the last month with a number of people. We’d like to thank Dr Shamim Qazi and Dr Lulu Muhe both of the Department of Child and Adolescent Health and Development at WHO, Dr Siobhan Crowley of the Department of HIV/AIDS at WHO, Dr Annelies Van Rie of the University of North Carolina, Dr Henry Barigye, who has just taken up a post with ICAP in Tanzania, Dr Tunga Namjilsurent, Communications Officer for the Partnership for Maternal Newborn and Child Health, Dr Joy Lawn of Saving Newborn Lives/Save the Children US and Dr Stephen Graham of the Centre for International Child Health at the University of Melbourne Australia.

We’d also like to thank Dr Shamim Qazi, Dr Stephen Graham, Dr Heather Zar, Dr Prakash Jeena and Dr Doug Wilson for reviewing the article.

This clinical review was kindly supported by the Diana Princess of Wales Memorial Fund. 

Pneumonia is an acute infection affecting the lungs that interferes with breathing and oxygen absorption. It is characterised by coughing and fast or difficult breathing, but there are additional symptoms or danger signs indicating more severe cases described below.

Pneumonia is the single biggest killer of children worldwide , causing an estimated 2.4 million deaths in neonates and children under the age of five every year.¹ Thus, it accounts for nearly one out of every 5 deaths in young children. Most of the deaths occur in resource-constrained countries (50% in sub-Saharan Africa and 20% in South-East Asia) where 151 million cases of childhood pneumonia are estimated to occur each year — 11 to 20 million of which may require hospitalisation.²

Such staggering figures ought to mobilise a global effort to address pneumonia, but instead “pneumonia is a forgotten pandemic” according to a 2006 report from WHO and UNICEF .³  This childhood illness receives little press, modest funding, inadequate attention from governments, local health programmes, bi- and multilateral funding partners, and there is a shortage of modern era research into how to better manage the disease.

One of the reasons could be the lack of advocacy for issues affecting children at most risk from poor and remote communities.

But the low profile of pneumonia as a global health issue could also be due in part to a perception that it is “manageable” with antibiotics (so is tuberculosis — and look at the scale of that problem). Plus, now there are vaccines for the leading causes of bacterial pneumonia— though most children in low resource settings are not getting them.

It may also be partly because a simple plan to manage most cases of pneumonia already exists, and has long been incorporated into the Integrated Management of Childhood Illnesses (IMCI) strategy (a holistic approach to childhood illness aimed at improving health worker skills, strengthening health systems and engaging and improving community and family practices to deal with the main causes of mortality and morbidity in children under 5 in resource limited settings).⁴ Unintentionally, pneumonia may have been so well integrated into Maternal, Newborn and Child Healthcare (MNCH) policy that it has become virtually invisible as a global health issue.

As a result, there’s a sense that the challenges to pneumonia management are primarily operational. Partly, this is true —a recent report from Tanzania, one of the countries where IMCI was first piloted in the mid 90’s, noted shockingly low implementation of the local IMCI-based guidelines for facilities; and a health systems audit in the northeast of the country found that documentation at the health facilities surveyed was so poor that it was impossible for the researchers to conclude whether children were getting the right diagnosis and treatment or not.⁵ What records they could find suggested that even the most “basic clinical signs were often not checked.”

“The main cause of the high mortality from pneumonia (mainly in developing countries) is a lack of access to effective health services,” according to a report on an informal consultation held last year by WHO and UNICEF to develop a Global Action Plan for the Prevention and Control of Pneumonia (GAPP).

Weak MNCH services are a perennial problem — one that should grab more people’s attention.

But there is another barrier to implementing improved pneumonia management: HIV.

Home treatment for children with severe pneumonia now possible — in non-HIV settings

A landmark study published earlier this year in the Lancet reported that children with a presumptive diagnosis of severe pneumonia (made by a physician or trained health worker) could be just as effectively treated with oral antibiotic pills (5 days of amoxicillin) in their own homes as they could be in a hospital with more intensive therapy (injected ampicillin, followed by oral amoxicillin) although ‘very’ severe cases (with danger signs as defined below) should still be taken immediately to a well-equipped facility.⁶

Similar community case management strategies for pneumonia — in which community health workers are trained to recognise pneumonia and give the mother or caregiver appropriate antibiotics to treat “non-severe” cases at home, and refer the more severe cases — are already integrated into IMCI and have a proven track record in some resource-constrained countries, particularly in Asia. ^{7, 8}  Now, the findings of this large randomised study in over 2000 children in Pakistan suggest that the approach could potentially be extended to manage severe and life-threatening pneumonia.

Since large numbers of children actually die of severe pneumonia before ever making it to a hospital, local assessment and home-based care for pneumonia could present families with a treatment option that is effective, convenient, more equitable and more compassionate — while at the same time reducing the burden on public health systems. One might expect it would offer similar potential in sub-Saharan Africa, particularly in rural areas where it can be hard or impossible to get to a hospital in time, and where task shifting is increasingly being employed to improve health service delivery.

There’s just one caveat:

“This treatment strategy for severe pneumonia will not be useful in high HIV prevalence settings,” WHO said in a press release earlier this year.^{9, 10}  Yet the majority of pneumonia deaths occur in  sub-Saharan Africa.

“The standard case management guidelines for pneumonia recommended by WHO for use in areas with low HIV burdens are less effective in areas where HIV burdens are high,” Dr Prakash Jeena of the Department of Paediatrics and Child Health, of the University of KwaZulu-Natal wrote in an editorial in a recent issue of the Bulletin of the World Health Organization.¹¹  This special issue, one of the first outcomes of GAPP, contains 16 reviews and articles dedicated to different aspects of the prevention and control of childhood pneumonia. .

Some steps in the approach — such as getting a mother to take a child with signs of pneumonia to see a community health worker for immediate assessment — remain important to saving children’s lives everywhere. But unfortunately, home care may not always be possible at present because the oral antibiotics currently prescribed for children with pneumonia don’t work (or work as well) for some of the infections causing severe pneumonia in infants and young children with, or exposed to, HIV. In addition, pneumonia in children with HIV may rapidly become so severe that they may need to be given oxygen and other services only available at larger health facilities — and even in hospitals outcomes are often poor.

Dr Jeena cited a recent South African study in a hospital setting (which  we’ll refer to as the Durban Paediatric Pneumonia Study or DPPS) that found that even when using a treatment regimen optimised for severe disease in HIV-infected children, the cumulative rate of failure was still quite high — especially in infants under one year of age with or exposed to HIV.¹²

So what exactly should be done to benefit the most children in settings with a high burden of HIV? Can the strategy be improved upon, somehow, in a way that would benefit all children? There are few easy answers.

“There are a large number of problems in designing sensible guidelines for pneumonia in practice, and there is very little work being done to try and address what should actually happen in terms of modifying guidelines,” said Dr Mike English of the Child and Newborn Health Group, Kemri in Nairobi, Kenya at a symposium on childhood pneumonia and HIV in Africa at the World Union Conference on Lung Health in Cape Town last November.¹³

“HIV has made life doubly difficult… and we are paying the price for having neglected key areas of clinical pneumonia research. For the clinician and the public policy makers, there are many more questions than answers in the search for optimum treatment approaches.”

It’s possible that the perfect could be the enemy of the good when it comes to pneumonia management. As we hope to show in this article, there are a number of policies and actions that could help mitigate the impact of HIV on pneumonia and reduce the suffering and death it causes in sub-Saharan Africa. But to achieve the best possible results, ART/PMTCT programmes and MNCH will need to work together and pool their financial, logistical resources and technical know-how, and engage all the potential partners involved in providing supportive and palliative care services within the community.

HIV’s impact upon pneumonia in children

As high as the estimates of the global pneumonia burden are, they may not accurately reflect the full impact of HIV on pneumonia over the past decade.¹⁴ In fact, the most recent figures are essentially the result of incidence rate estimates drawn from studies conducted between 1969 and 1999 and then applied to the current populations of at-risk children.¹⁵  However, studies have shown that in high burden countries, HIV has changed the formula by dramatically increasing the incidence, severity and mortality associated with pneumonia.

For instance, according to the GAPP report, the incidence of childhood pneumonia in South Africa has increased by 45% since 1995.¹⁶ So in high burden settings, HIV could be fuelling the pneumonia pandemic in much the same way as it does tuberculosis (TB). At the same time, the impact of pneumonia on children with, or even exposed to HIV, is devastating.

“Pneumonia is the most common cause of illness, hospitalisation and death in HIV-infected children,” said Dr Heather Zar, of the Red Cross Children’s Hospital and the University of Cape Town, who also spoke at the symposium on childhood pneumonia last November.¹⁷  In fact, 90% of HIV-infected children will develop a respiratory illness during the course of their HIV disease — and most of these illnesses will be due to pneumonia.¹⁸

Studies also show that simply having a mother who is HIV-infected puts infants at a greater risk of pneumonia, treatment failure and death — perhaps because they are exposed to more infections from their parents, are more likely to be malnourished or because they do not receive protective immunity transplacentally or from their mother’s breast milk.^{19, 20} But it is the HIV-infected infants who fare the worst, with a case fatality rate that is 3–8 times higher than in HIV-uninfected children.²¹

“Although children with HIV comprise less than 5% of the childhood population, [in] sub-Saharan African countries, these children suffer disproportionately from pneumonia (nine times greater risk) and are susceptible to pneumonia caused by a greater variety of pathogens. In South Africa, HIV-infected children account for 45% of all childhood pneumonia morbidity and 90% of pneumonia mortality,” Madhi et al wrote in the recent WHO Bulletin.²²

The same infections causing pneumonia can also seriously involve other parts of the respiratory tract or body. For instance, HIV-infected infants and children with bacterial pneumonia are also far more likely to become bacteraemic. In one pneumococcal vaccine trial, 22% of the HIV-positive children were bacteraemic compared to only 7% among the HIV-negative children.²³

The emergence of previously less common causes of pneumonia, and changes in the pattern and frequency of others are described further below. This wider range of infections poses challenges for diagnosis, which in turn results in inappropriate or sub-optimal treatment, and increases the risk of drug resistance. There is a significant chance this resistance could spread within the community— altering the effectiveness of the antibiotics used for empiric treatment.²⁴ And since treatment failure is more common, there is also a greater risk of recurrent pneumonia.

Additionally, while vaccination for the major causes of bacterial pneumonia is still recommended, data suggest it is less effective in children with HIV. Since at least part of some vaccines’ effectiveness comes from “herd immunity” (protecting people, whether vaccinated or not, by reducing the pool of infectious cases who can transmit the infection), a large population with untreated advanced HIV disease suffering frequent infections could theoretically diminish a vaccine’s effectiveness at the population level.

Finally, HIV drains health care resources from MNCH services since more children require treatment, hospital admission and intensive care for pneumonia.²⁵ HIV/PMTCT programmes and funding partners could offset the system-wide costs of HIV by taking a more holistic approach but are often rather narrow in focus, which may be in the long run be self-defeating.

For instance, children with HIV are more likely to have severe pneumonia with hypoxia (oxygen deficiency in body tissues) and need to be given oxygen. But in many small hospitals, not to mention primary health care clinics, oxygen is already in short supply if it is available at all.²⁶  According to the GAPP report, “WHO has been promoting the availability of oxygen in small hospitals in developing countries [but] several surveys of these hospitals in recent years, however, have shown that oxygen is either not available at all in many hospitals or is not used appropriately.”²⁷ Hypoxia increases the risk of dying from severe to very severe pneumonia fivefold²⁸; so stock outs of oxygen can lead to the death of any child with severe pneumonia, with or without HIV.

Causes of pneumonia

Over the last hundred years of medical science, the major causes of pneumonia in HIV-negative children have been identified, including such pathogens as Streptococcus pneumoniae (the leading overall cause of pneumonia in children), Haemophilus influenzae type b, respiratory syncitial virus (RSV) and influenza, Staphylococcus aureus and Klebsiella pneumoniae, along with a number of less common infections (see Table). But data on the role of many pathogens in pneumonia — or the interpretations of the data — are conflicting, and the methods used to diagnose or report on infections vary from study to study. Thus at present, reliable data on the current distribution and relative frequencies of the different pathogens causing pneumonia in different populations are limited, and further research and surveillance is urgently needed in order to determine optimal treatment and prevention strategies for childhood pneumonia.²⁹

Table: infectious organisms causing or associated with pneumonia*

_{*Drawn mostly from Dr Heather Zar’s presentation at the World Union meeting, Rudan et al’s review in the Bulletin of the WHO, and Pneumonia: The Forgotten Killer of Children.}

But HIV has broadened the spectrum of infections that can cause pneumonia — perhaps most notably Pneumocystis jirovecii pneumonia (PCP), which has emerged as a major cause of pneumonia and death in infants. Since, again, methodology varies from study to study, it is difficult to say with any certainty how much each of the infections contributes to pneumonia seen in children exposed to or with HIV. (Again, much of the following is based upon the review presented by Dr Heather Zar at the World Union on Lung Health).

In general, it is safe to say that many of the less common infections in the Table are more likely to be found in infants and children with HIV or exposed to HIV.

Bacterial infections

The most common cause of pneumonia generally remains the most important cause in children with HIV, and there has been an increasing incidence of S. pneumoniae in settings with a high burden of HIV, especially among infants under one year old. But commonly, S. aureus — with an increasing prevalence of methicillin resistance (MRSA) — and gram-negative bacteria such as K. pneumoniae or P. aeruginosa occur — which makes the selection of a good empiric treatment more difficult.   

M. tuberculosis has also been reported in about 8% of children hospitalised with acute pneumonia, but since TB in children is particularly difficult to diagnose, the burden may actually be greater.^30,31 In the DPPS study 53/358 (15%) children with severe or very severe pneumonia had TB.³²  

Viral infections

Although a study by Madhi et al found that viral causes accounted for a lower percentage of the cases of severe lower respiratory track infections in children with HIV (probably because other infections have become more common), the relative incidence rate of severe viral-associated pneumonia was much higher, ranging from 1.9 for RSV (range 1.2-2.8) to 8.03 for influenza A or B (range 5.0-12.7), and 15.07 for adenovirus (6.62-34.33).³³

Some opportunistic viral infections may also be associated with pneumonia. For instance, CMV has been found upon diagnostic evaluations in children with primary pneumonitis and HIV, especially those with very low CD4 cells — but its contribution to the illness and the best form of management are controversial (see below).

PCP

Most fungal opportunistic infections are a rare cause of pneumonia, only occurring with disseminated disease in very immunosuppressed children.

However, P. jirovecii, formerly called Pneumocystis carinii, or PCP, and classified as a protozoan, is now considered to be an atypical fungal organism — and far more importantly, a major cause of severe pneumonia in HIV-infected and perhaps even some HIV-exposed infants.³⁴

At one time, PCP was thought to be relatively rare in sub-Saharan Africa because it can be difficult to diagnose in resource-constrained settings, and there was much disagreement over the interpretation of some of the studies that did report it. Furthermore, in the pivotal cotrimoxazole study in Zambian children (CHAP) with HIV, PCP was not observed in the placebo arm — but that study only included children over 1 year of age.³⁵ In the last decade, and especially since the HIV pandemic swept into South Africa, where hospitals are better equipped, the preponderance of evidence has shown that PCP preferentially targets infants, especially those between 3 to 6 months of age.^{36,37,38,39,40,41,42,43,44,45}

“Initially we all believed that PCP didn’t exist in Africa,” said Dr Jeena, speaking at the symposium on childhood pneumonia at the World Union Lung Conference in Cape Town.⁴⁶ “But papers first published in 1995 from Côte d’Ivoire showed it accounted for a high proportion of mortality and that’s been replicated across most parts of Africa.⁴⁷ We were also silly at that point because if we looked at the data from Wakefield et al, we found that over 80% of patients had got antibodies to PCP at 8 years [of age].⁴⁸ So obviously the fungus was around long ago, we were just missing it somewhere along the line.”

“Consistently data from studies in central hospitals - so there is a likely bias to more severe pneumonia and therefore PCP - have shown that PCP is a very common cause of death in HIV-infected infants,”  Dr Stephen Graham told HATIP.

Dr Graham is currently at the Centre for International Child Health, Royal Children’s Hospital, Melbourne, Australia but previously worked in Malawi for ten years. “Of course, causes of death do not accurately reflect burden of disease — so bacteria such as pneumococcus are likely to be a much more common cause of pneumonia but because the case fatality rate is lower for pneumococcal pneumonia than for PCP, the difference in burden of disease is not well reflected in autopsy data.”

“PCP is the commonest cause of death from pneumonia in those below 6 months of age, and accounts for a quarter of the cases in infants over 6 months of age,” Dr Zar said in Cape Town.

In fact, according to a report of a WHO consultation on cotrimoxazole, PCP causes at least one out of every four deaths in HIV-infected infants under the age of one.⁴⁹ And Dr Graham thinks it may be even higher:

“Overall, PCP seems to be the cause of death in about 35% of HIV-infected infants so it’s worth preventing,” he said.

Even more disturbing figures have recently come out due to better auditing of the causes of death in children in some hospitals in South Africa. For instance, according to the Saving Children 2005 report, “during analysis of the Witbank Hospital’s Child Healthcare Problem Identification Programme data at the end of 2004 and beginning of 2005, it became clear that deaths due to suspected pneumocystis pneumonia (PCP) constituted the biggest single cause of mortality (44% of all child deaths were due to PCP). Further analysis revealed that the failure of the PMTCT programme in the district could be linked to this high mortality rate. In every single PCP death either the mother had never been on the PMTCT programme or some mistake had been made during the execution of the programme (e.g. no cotrimoxazole prophylaxis from six weeks).”

Dr Graham noted that, even though there are fewer data, PCP does occur in young HIV-negative infants as well — perhaps because they receive little passive immunity to it from their mothers.^50,51,52 There were 3 cases in HIV-negative infants in the DPPS study, and Dr Jeena suggested that a mother with PCP might even transmit it directly to her breastfeeding child.

Polymicrobial disease (mixed infections)

One of the reasons why outcomes are often poor in children with HIV is that they may have more than one respiratory infection at once. For instance, Madhi et al reported that coinfections, such as influenza and bacterial infections were more common in HIV-infected children (78% vs 31%).⁵³ And the DPPS study reported that more than one organism were found in 70% of children with severe or very severe pneumonia.⁵⁴ These infections were more likely to be unresponsive to usual antibiotics and were associated with a higher risk of mortality.

Here’s where infections such as CMV may become more important.

“We all said [CMV] was probably an innocent bystander,” said Dr Jeena. But in [the DPPS study which he co-authored] more than PCP, we found CMV to be the commonest organism that we identified [in infants]… CMV occurred mainly between 1 to 12 months of age.”

One possibility, however, is that polymicrobial disease and the poor outcomes are simply a consequence (and a sign of) extremely advanced HIV disease. Mixed infections are more likely to occur in children without a viable immune system — especially given with poor infection control practices at home and in health facilities.

But CMV is probably not innocent in these infants. It is also likely that CMV or other infections could be driving HIV replication, and indeed, Dr Jeena said it was found in the children with extremely high HIV RNA loads. In fact, in a small study that Dr Jeena conducted treatment with ganciclovir did seem to improve survival versus a historical control. The treatment effect suggests that the CMV is indeed contributing to mortality (more below).