The time trends shows a gradual decline in malaria morbidity and mortality in Mpumalanga Province over the past eight malaria seasons, representing an 85% reduction in the annual number of confirmed malaria cases and 74% in the number of deaths-attributed to malaria. A previous study, estimating the burden of malaria in Mpumalanga Province between 1987 and 1999, showed fluctuating trends coupled with flood-related malaria epidemics in 1996 and 2000 . Since 2002, the province has seen a marked decline in malaria cases and deaths, with no epidemics detected. The declining trend in malaria incidence demonstrated in this study is consistent with previously published results from other malarious provinces of South Africa; KwaZulu-Natal  and Limpopo .
Notable peaks in the number of notified cases were observed during the 2001/02, 2003/04 and 2005/06 malaria seasons. The 2001/02 peak was followed by a steep reduction of almost 70% in the subsequent season. This could be a result of the change in drug policy to combat parasite resistance to SP ; the re-introduction of DDT; and the Lubombo Spatial Development Initiative (LSDI), a joint development programme implemented to control malaria in Mozambique and Swaziland . The introduction of artemisinin combination therapy (ACT) for treatment of uncomplicated malaria in 2003 and 2004  could partly explain the upsurge in malaria cases during the 2003/04 malaria season. However, it is possible that drug policy was not the only change that was introduced during this period. Further research is required to explain this scenario. The 2005/06 pattern may be attributed to the integration of Bushbuckridge municipality into Mpumalanga Province; the abandonment of the requirement of entry visa between South Africa and Mozambique  leading to large population movement between the countries thus bringing about importation of malaria cases; and rainfall anomaly.
In this present study, the finding of peak malaria incidence in the young adult age group (15-44 years) may be related to outdoor behavioural risk factors such as leisure patterns and sleeping arrangements leading to exposure to infective mosquito bites . The low proportion of cases in infants and children further supports possible predominance of outdoor transmission since small children tend to spend more time indoors particularly during mosquito biting time. In Tanzania, it was found that high usage of intra-domiciliary vector control tools may have altered vector feeding patterns from indoor to outdoor transmission, suggesting the need for additional vector control tools that target outdoor biting mosquitoes, such repellents and larval control .
This study found significant differences (P < 0.001) in malaria incidence between males and females. This was similar to previous observations in Mpumalanga , KwaZulu-Natal  and Limpopo  Provinces. In some societies, men have a much greater risk of contracting malaria due to occupational reasons, particularly those that work in mines, fields or migrate to areas of high endemicity for work . The study conducted in Ethiopia provides evidence of the relationship between occupation and malaria risk; the authors found that highland migrant labourers were vulnerable to malaria while migrating to find agricultural work . A similar scenario may exist in Mpumalanga Province.
Analysis of the findings reveals that the burden of malaria in Mpumalanga Province is strongly connected to importation of malaria parasites by population movement as almost half (48%) of the cases reported in Mpumalanga Province were acquired in Mozambique. Trend analysis of these cases reveals that 74% were among males between the ages of 15 and 44 years; confirming that the majority of the imported cases were introduced by young adult males crossing the border to seek work opportunities in South Africa. In view of the high proportion of cases acquired in Mozambique, intensification of the regional cross-border and intersectoral collaboration approach is vital in order to lower the risk of re-importation of malaria infections.
Malaria transmission in Mpumalanga Province shows inter-annual variation from 2001 to 2009 with a distinct malaria transmission season, prominent peaks in January and February. The reasons for the January/February peaks could be attributed to favourable climatic conditions for malaria transmission (peak summer season) or introduced parasites by human immigration from various places of origin following the December holiday season. This finding suggests important implications for the control programme in terms of timing when directing efforts in controlling malaria.
Although a large proportion of the cases reported in Mpumalanga Province were imported, the province still accounted for half (50.1%) of the total cases notified, which indicates the recurrence of local malaria transmission. The incidence of malaria was found to be most pronounced in Ehlanzeni district (low altitude region) than in the high altitude districts (Nkangala and Gert Sibande), suggesting the effect of altitude on malaria transmission. The burden of the disease in the province lies in Nkomazi municipality by contributing 73% of the province's malaria cases. Malaria transmission in this area can be attributed to factors such as its sub-tropical weather conditions, close proximity to high malaria transmission parts of Mozambique as well as intensive agricultural practices. The link between malaria and agriculture has a long history, in particular irrigation, by creating suitable vector breeding sites and facilitating malaria transmission.
Malaria case fatality rates fluctuated over the entire study period. The overall CFR was 0.54% which is not much significantly higher than the national target. However, it is important for the province to further decrease the CFR to below 0.5% through improved case management. Unlike other African countries, where malaria is a major cause of infant and child mortality, in Mpumalanga Province malaria-attributed mortality was lower in infants and children. Severe illness due to malaria was higher among adults, with the CFR reaching 2.1% in those over the age of 65 years compared to 0.28% and 0.14% for under fives and five-14 age groups. The enquiry into all deaths-attributed to malaria in Mpumalanga Province in 1999, revealed that late presentation to health care facilities was strongly associated with increased mortality due to malaria . In another study in South Africa, it was found that co-morbid diseases, especially HIV co-infection and poor management of malaria-related complications led to mortality outcomes . This suggest the need to maintain sustainable training programmes for all health care workers in all levels of health facilities in both low and high risk transmission areas as well as community health promotion and education.
It can be reported that rainfall has played a significant role in the transmission of malaria in Mpumalanga Province. The transmission season followed a distinct rainfall pattern and fluctuated considerably from year to year according to rainfall variability, with heavy rainfall associated with increased number of reported cases. Similar results were reported in KwaZulu-Natal where they observed a direct and predictable relation between rainfall and malaria transmission .
It has been suggested that anthropogenic climate change is expected to directly affect the behaviour and geographical distribution of mosquitoes and the life cycle of the parasite and thus changing the epidemiology of the disease [38, 39]. Casman et al. , however notes that since climate can correlate with transmission intensity, it can greatly affect the success or failure of control and eradication programmes. The authors however point out that in fringe transmission areas like South Africa, malaria surveillance and control maybe sufficient to mitigate any increases in transmission brought about by climate change. Mpumalanga Province needs to maintain high quality surveillance system to facilitate immediate detection, notification and response to outbreaks that may be triggered by climate anomalies.
The elimination of malaria transmission in some temperate regions of the world during the eradication era in the 1950s to 60s was largely based on IRS, which illustrates its programmatic effectiveness in malaria control . Numerous studies have shown that intensive IRS campaigns have substantially reduced levels of infection and incidence of malaria . In some settings it has been reported that the IRS intervention was associated with marked decreases in malaria transmission by more than 50% .
A study conducted in Uganda to assess the impact of IRS on malaria morbidity after a single round of spraying with lambda-cyhalothrin found a consistent decrease in the number of patients diagnosed with clinical malaria in the first four months after IRS . In South Africa, marked reductions in the number of confirmed cases and deaths in Mpumalanga and KwaZulu-Natal Provinces were observed following the introduction of IRS campaigns in Mozambique and Swaziland through the LSDI. In KwaZulu-Natal Province, a significant reduction in the number of cases in most endemic areas of the province was reported following the re-introduction of DDT for vector control .
Although increased IRS coverage has significantly contributed to the visible decreasing trends in malaria morbidity and mortality, other factors such as the combination of preventative interventions (early diagnosis, case management and effective treatment with ACT), low numbers of mosquito vectors, improved education and awareness, and improving socio-economic indices, had played a role in the marked reduction in the burden of malaria in Mpumalanga Province.
Although IRS has had beneficial effects in the history of malaria control and prevention, there is increasing awareness and concerns from new scientific evidence regarding the safety of insecticides (i.e. DDT) on humans and the environment . More recently, the Pine River Statement revealed substantial evidence that DDT and DDE pose a serious risk to human health, particularly due to IRS for malaria vector control . Other studies conducted in South Africa reported several risks associated with non-occupational exposure, such as male reproductive effects [47, 48]. In view of these public health concerns, efforts needs to be directed towards the development of new tools for malaria vector control in order to minimize adverse health effects and halt endemic transmission.
Based on the findings of this study, the Mpumalanga Malaria Control Programme needs to consider the following in order to achieve the goal to eliminate malaria; (i) address importation of malaria cases through intensification of the regional cross-border collaboration efforts; (ii) develop strategies to interrupt local transmission and transmission risk through identification of asymptomatic infections and effective treatment of all infections before transmission can occur and; (iii) maintain strong surveillance and reporting systems through vigilant monitoring of the data collection (i.e. completion of notification forms) and capturing process in order to maintain a consistent information system which is a cornerstone of a successful malaria elimination programme.
It is acknowledged that this study has several limitations given that it relies mainly on routine surveillance data. Firstly, the problem with surveillance data is that the quality might be subjected to reporting inconsistencies and incompleteness, emanating from lack of systematic inclusion of data from other sources such as traditional healers, faith-based organizations and self-treatment cases. Secondly, over-reporting of malaria cases could have also occurred in the high-risk areas due to high awareness and advocacy regarding malaria among health care workers. Finally, it is possible that some confounding factors that may have influenced the changes in the burden of malaria were not addressed in this study, such as the effect of other preventive interventions, changes in vector population and socio-economic status.
Despite these limitations it can be noted that routinely collected data through the provincial surveillance system remains the basis for measuring malaria trends over time. The main advantage of the surveillance data is that it includes asymptomatic cases through passive case detection, thus including cases which might have gone unreported in health care facilities. In view of this, the surveillance system data has captured the picture of malaria transmission patterns in Mpumalanga Province and these findings can be used to strengthen malaria control efforts.