Interventional based study to reduce child mortality in rural Uganda

Introduction

One of the World Health Organization (WHO) Millennium Development Goals is to reduce child mortality by two-thirds between 1990 and 2015. ¹ Many countries, including Uganda, are not on target. In 2009, the child mortality rate in Uganda was 128/1000 live births ² and it is one of the countries that has made ‘insufficient progress’. ³

This study looks at the use of observation charts as a simple and cost effective way of reducing child mortality on a paediatric ward in rural Uganda. Prior to the introduction of observation charts, observations including temperature, heart rate, respiratory rate and saturations on the paediatric ward were only done sporadically and were often not complete. Blood sugar measurements were also only taken sporadically even in children with previously documented hypoglycaemia. Together with the nursing staff an observation chart was developed that could be used 1, 2, 4 or 6 hourly and introduced for all sick children. Training sessions on the use of the charts, measuring observations in children, interpreting the results and managing deteriorating children were undertaken. The mortality prior to and after introducing the charts was compared in order to see if this simple intervention reduced inpatient mortality on the paediatric ward.

Methods

This study was a retrospective study using the admissions book to calculate the number of admissions, discharges and deaths during the 6 months prior to the introduction of the observation charts (1 June to 30 November 2010) and the corresponding 6 months in the following year (1 June to 30 November 2011). The files of the deceased children were obtained from the records office and the patient’s demographics, cause of death and length of stay for both groups were analysed. If information on outcome was not indicated in the ward book the case notes were examined. These results were statistically analysed using a two-sided Fisher’s exact test.

Results

The total number of admissions for period 1 (1 June to 30 November 2010) was 496 and for period 2 (1 June to 30 November 2011) was 706. There were two cases in each period where the ward book did not include outcome and the case notes were missing, and they were excluded from the study. There were 34 deaths in period 1 and 27 in period 2.

The paediatric ward is used for children aged 1 month to 5 years. Older children are nursed on the adult wards. The mean age of the children who died in period 1 was 24 months and in period 2 was 22 months. In both periods 56% (19 in period 1, 15 in period 2) of the deaths were girls and 44% (15 in period 1, 12 in period 2) were boys.

The causes of death were similar in both groups, malaria (26% (9) in period 1; 30% (8) in period 2), HIV related illnesses (24% (8) in period 1; 26% (7) in period 2), and malnutrition (21% (7) in period 1, 19% (5) in period 2) being the three most common causes of death. In the second period, gastroenteritis was the cause of death in 22% (6) compared to just 11% (4) in the first period.

In both groups, the highest numbers of deaths were on the first day following admission (53% (18) in period 1; 41% (11) in period 2) with a decreasing number dying each day after that with 3% (1) of deaths at more than 2 weeks in period 1 and 7% (2) in period 2. Mortality for period 1 was 6.85% (34/496) and period 2 was 3.82% (27/706). The decrease in mortality is statistically significant (P = 0.023).

Discussion

Studies have shown that paediatric patients who die in hospital have had deteriorating observations prior to cardiac arrest^4,5 and that early warning systems looking at simple observation charts are of help in the recognition and timely intervention for these children and lead to a reduction in mortality. ⁶

These studies were all undertaken in developed countries with expensive resources at their disposal. This study demonstrates that, even in low resource settings, such systems can reduce mortality despite a lack of expensive equipment and/or intensive care units.

The increase in admissions in period 2 compared to period 1 is something that was seen in all departments of the hospital and reflects the year-on-year increase in workload. There were no identified outbreaks or natural disasters that contributed to this. Other factors that may have had an effect on mortality are: changes in staff on the paediatric ward; differences in the number of doctors at the hospital; improvement in community care; and improved PMTCT (prevention of mother to child transmission) of HIV. These factors are constantly changing. Age, sex, causes of death and length of stay were similar in both groups and, therefore, were not confounding factors. The same months were used in both periods in order to eliminate seasonal variation.

During a return visit to Kiwoko in November 2012 these charts were still being used which confirmed that the clinical staff agrees that they are a valuable tool and that they are willing to invest the necessary time completing them. Students at the local nursing school are introduced to the charts during their ward placement training and are encouraged to be involved in taking the observations and discussing them with senior staff.

Conclusion

Simple interventions, such as regular monitoring of nursing observations in children coupled with associated training of nurses in interpreting the results, have resulted in a statistically significant reduction in child mortality rates in a rural hospital in Uganda. This is a cheap intervention that does not require specialist equipment and can be implemented in countless other resource poor settings.