ADR occurrence rates in rural Tanzania were estimated at 11.6 and 25.0 due to exposure to SP in Rufiji and Kilombero/Ulanga respectively. Incidence due to use of AS/SP combination was substantially lower at 5.6 per 100,000 exposures. A number of reasons might have been responsible for this trend, including differences in quality of care, frequency of drug use between the sites and a relatively sporadic supply chain for the AS/SP combination drugs. At least in Kilombero and Ulanga district, there was a project engaged in improving access to anti-malarials which might have also influenced the use of sulphonamide containing anti-malarials amongst the targeted population [45
]. Our study design may also have been somehow biased as it was more geared to identify ADR cases caused by use of SP monotherapy than AS/SP combination drugs. Following the national anti-malarial policy change, there were anecdotal reports of ADR events caused by SP drugs resulting to general public outcry and resistance in using SP drugs for malaria treatment. Therefore, ADR SOs in districts with SP monotherapy enjoyed closer supervision and follow-up of their activities than those in Rufiji which includes the highly inaccessible Rufiji delta area where reporting and tracing cases for confirmation might have been more difficult. Seven of the 67 cases successfully traced and clinically classified died for reasons associated with ADR. Of those dying, four were recorded in Rufiji and three were from Kilombero and Ulanga districts. Estimations of ADR occurrence rates were based on cases classified as ‘probable’ ADR events only, and would have been higher if ‘possible’ ADR had also been included. It is also important to note that most of the ADR cases reported were associated with SP and SP/AS (but none for AS alone).
These findings may be compared with studies conducted in other malaria-endemic settings including Malaysia, Peru and Malawi [16
]. In the Malaysian study, ADR incidence was estimated at 2.4 per 100,000 exposures to SP. SP was used prophylactically by almost a half of those exposed and almost certainly involved repeated dosing [17
]. The study in Peru followed up patients exposed to the combination of SP plus artesunate in selected clinics for two years, with 8% reporting some associated ADR, although very mild reactions were included in the ADR definition [16
]. A study in Blantyre, Malawi beginning early 2001 used a definition more comparable to that used in this study, and estimated ADR associated with use of SP or trimethoprim-sulphamethoxazole (cotrimoxazole) at 0.3 to 8.4 per 100,000 exposures [30
]. ADR incidence among HIV positive persons was higher at 20.4 per 100,000 exposures. This study did not collect information on HIV status during the IMPACT-Tz study, however, other sources indicate that HIV prevalence rates at the time ranged from 4 to 7% across the study districts [46
]. The studies from Peru and Malawi relied on health facility-based surveillance only, with no active community surveillance. The authors of the Malawi study in particular acknowledged this limitation and suggested that that the number of ADR cases might have been under-reported, especially in areas where access to health facilities was difficult [30
]. Following the nature of active surveillance component of this study, it is not surprising that incidence rates are higher. Studies on healthcare-seeking behaviour in the three study districts have reported that government health facilities see less than 60% of all healthcare seekers with fever/malaria, with many patients purchasing drugs at drug shops, general shops and small kiosks [40
], an indication that reliance on passive surveillance only is likely to miss some cases. However, it is possible that the figures presented in this study are underestimated, as recall bias and the assumption of homogeneity in the level of exposure to these drugs may have led to under-reporting during active surveillance, especially of more minor ADR. It is also important to note that this study’s surveillance estimates were based on areas with routine demographic surveillance where a range of health interventions are continually evaluated. It is possible that exposure to anti-malarial drugs in the study sites was therefore higher than might be reported from the rest of the country.
Analysis of the household costs of treating probable and possible ADR events resulting from use of anti-malarials showed that despite these events being rare occurrences, their associated treatment expenses can have important economic consequences. Mean treatment expenditure for an ADR episode was US$24.45 with a median of US$10.00, with average expenditure rising rapidly with number of providers visited. Average household costs of care were ten-fold higher than those reported for treating an episode of fever or malaria [48
]. For example, studies conducted in the same districts in mid-2001 reported a mean household expenditure of US$0.30 and a median of US$0.06 (all in 2001 prices) per fever episodes [50
Furthermore, some patients with ADR endured lengthy periods of illness as indicated by median admission days of 7.5 with the longest hospital admission lasting 60 days. Data were not collected on the total number of disability days caused by ADR events, but it seems likely that the number of productive days lost was higher than those usually reported in most malaria studies which do not consider ADR [48
]. By the time this study was completed, of the 50 patients with complete cost information, five (10%) patients said they were yet to fully recover.
The costs for treating ADR resulting from use of SP monotherapy were found to be higher than those from ACT use. However, this may be confounded by the different health financing policies in place in the two study areas during the study period. In Kilombero and Ulanga Districts where SP monotherapy was routinely delivered as first-line treatment, government facilities had started implementing user fees for all patients except pregnant mothers and children aged below five years in early 2003. In contrast, in Rufiji District, where SP/AS was available, services at government-owned facilities were officially free until late 2005 when user fees were introduced for the first time.
There are indications that the costs associated with treatment of drug reactions placed substantial financial burden on the households concerned. Of the 37 ADR patients who paid money while seeking treatment, only six patients made payments through a health insurance scheme, an indication of very low health insurance coverage in these areas. Additionally, eight patients reported having to sell part of their farm harvests or household assets; and two had to borrow money to finance their treatment. An alternative way to understand the magnitude of the treatment burden placed on households because of ADR events is by comparison with total household per capita income or expenditure. Tanzania’s GDP per capita was estimated at US$326 in 2005 [51
], and median monthly household expenditure in the HDSS study sites ranged from US$77 to 96 with 75% of expenditure on food [34
]. Following the WHO criteria for catastrophic expenditures, households spending 50% or more of their non-food expenditure on healthcare are likely to be impoverished by the cost [52
]. Other studies have defined household expenditures as catastrophic whenever household healthcare spending exceeds 40% of income remaining after subsistence needs are met [54
]. In either case, these findings indicate that a majority of treatment expenditures on adverse events could be considered catastrophic.
Finally, it is important to note that full cost information was only available for 50 patients classified as ‘possible or probable’ ADR events. This is an inherent problem with studying rare events. Some respondents had to recall costs over a relatively long period compared to standard household surveys, which may have increased the likelihood of recall bias. However, perhaps reflecting the fact that ADR are unusual and possibly catastrophic events, during data collection interviewees did not have difficulties in remembering treatment sought and costs incurred. Meanwhile it is also important to note that five cases included in the costing analysis had not fully recovered on the day of interview. This may have slightly underestimated the total cost of their treatment. However, all except one had already obtained all their drugs for treating their conditions and were unlikely to visit healthcare facilities or drug shops for additional treatment to the same illness.
While these drugs were found to be generally safe, this study underscores the importance of improving ADR surveillance for anti-malarial drugs. The use of active surveillance especially in places with HDSS research platforms may play an important part in enhancing post marketing drug safety surveillance. Efforts to increase access to effective anti-malarials should go hand in hand with implementation of effective drug safety surveillance across nations [28
], especially given the new generation of anti-malarial drugs in the pipeline and concerns about fake and substandard anti-malarials [26