Abstract
Background
Two years away from 2015, the decline in child mortality is not fast enough to reach Millennium Development Goal 4. The Integrated Management of Childhood Illness (IMCI) is a strategy that simplifies management of child health. Beyond effective disease management, IMCI recommendations for care following illnesses are based on limited evidence from the field. The aim of this project was to find (1) the magnitude of and (2) risk factors for child mortality following discharge from a health facility in a low-income setting.
Methods
This study used an established population-based surveillance system for suspected invasive pneumococcal disease in Upper River Region, The Gambia, West Africa. Children that survived admission for suspected pneumonia, meningitis or septicaemia at the Region’s only referral centre (Basse Major Health Centre, Upper River Region) were followed for 180 days after discharge. Vitality status monitored by the DSS informed time-to-death information in a survival analysis that identified predictors of post-discharge mortality. Two multivariable Cox proportional hazards models were constructed. Model A described the clinical syndrome on admission (provisional diagnosis) and risk of post-discharge mortality. Model B used a reverse step-wise approach to find pre-discharge risk factors for mortality following discharge.
Results
The cohort that survived admission had higher mortality rates than the background rate in the community. Overall, 105 (2.8%) of 3735 patients died during the 6 months of follow-up. Half of the deaths occurred within 45 days of discharge. Approximately half as many patients died in the six months following discharge as died during hospital admission. Age stratified post-discharge mortality rates were three to six times higher than community mortality rates. In addition to demonstrating the protective effect of increasing age at discharge (HR 0.98 [95%CI: 0.96, 0.99] for every month increase in age), Model A showed that, compared to pneumonia alone, a provisional diagnosis of: pneumonia with visible signs of severe malnutrition had a HR 8.74 (95%CI: 4.93, 15.49); meningitis with visible signs of severe malnutrition had a HR of 13.90 (95%CI: 5.43, 35.58); sepsis with visible signs of severe malnutrition had a HR 18.79 (95%CI: 11.65, 30.32). Model B showed independent risk factors associated with post-discharge mortality were: the presence of neck stiffness on assessment (HR 17.60 [95%CI: 7.36, 42.10]); low mid-upper arm circumference (MUAC) (<10.5cm, HR 11.52 [4.59, 28.90]); visible signs of severe malnutrition (HR 3.94 [95%CI: 2.11, 7.36]); non- medical discharge (HR 6.22 [95%CI: 2.98, 13.01]); discharge during dry season (HR 2.33 [95%CI: 1.44, 3.77]); decreasing peripheral arterial haemoglobin oxygen saturation (HR 0.95 [95%CI: 0.93, 0.98] per percent increase); decreasing haemoglobin concentration (HR 0.82 [95%CI: 0.74, 0.90]) per unit g/dL increase); and decreasing axillary temperature (HR 0.70 [0.58, 0.84] per unit oC increase).
Conclusion
Gambian children in Upper River Region with suspected invasive pneumococcal disease are at increased risk of death following discharge from a health facility, and most of these deaths occur early. There are identifiable risk factors for death, including neck stiffness, low MUAC, visible signs of severe malnutrition, non-medical discharge, discharge during dry season, decreasing peripheral arterial haemoglobin oxygen saturation, decreasing haemoglobin concentration and decreasing axillary temperature. These data add to the evidence base needed to inform the development key guidelines and may be helpful towards development of a tool with clinical utility to identify children for intervention after discharge from hospital.