Cost-Effectiveness Analysis of Clinical
Diagnosis versus Clinical Diagnosis
plus Rapid Diagnostic Tests in Malaria
diagnosis at lower health centres in
Kitutu Freddy Eric
6th Annual Scientific Conference
September 22nd 2010
Kabira Country Club
Cost-effectiveness analysis is a full economic evaluation where both the
costs and consequences of use of clinical diagnosis and malaria Rapid
Diagnostic Tests (RDTs) by clinicians in malaria management are
Rapid diagnostic tests are devices that use immunochromatographic
methods to detect Plasmodium-specific antigens in a finger prick sample
of whole blood.
Effectiveness is the proportion of patients correctly treated/managed,
according to the test result of the malaria diagnostic strategy
Lower health centres refer to health centres level II and level III as
classified in the Uganda Health System Structure
• Malaria is still leading cause of disease burden accounting for 25-40%
of all outpatient attendances, 9-14% of admissions and a case fatality
rate of 3% in Uganda (Republic of Uganda 2008).
• Constrained healthcare budgets and introduction of ACTs as first line
treatment of malaria has brought to the fore the debate of diagnostic
strategies for malaria.
• Efficacy studies comparing malaria RDTs with expert microscopy
show sensitivity (85-95%), specificity (80-100%). Hence, MoH is
advocating an expanded role for parasitological diagnosis of malaria.
• Lubell Y et al recognize that standard economic analysis have shown
malaria RDTs to be cost-effective, without due regard to impact of the
health workers’ response to the malaria diagnostic test results. 3
• In preparation for RDT implementation in Uganda, MoH has called
for further operational research in cost-effectiveness of malaria RDTs
in Uganda (Republic of Uganda 2008).
• This study was done to identify, measure, value and compare the costs
and consequences of the clinical diagnosis versus clinical diagnosis
plus malaria RDTs in malaria diagnosis at health centres in Uganda.
• It provides information to feed into the decision making process on
strengthening malaria diagnosis at lower health centres in Uganda.
• Cost-effectiveness analysis using a decision analytic model in TreeAge
• This study was piggy-backed on a quasi-experimental study done by
• Parameters such as sensitivities and specificities of diagnostic tests that
could not be got from the study were obtained from published
literature and expert opinion.
• Evaluation of costs and consequences was done from a provider
Description of model
• Simple analytic model employing a decision tree of two event paths
• Patients at the lower health centres either followed the clinical
diagnosis strategy or the clinical diagnosis plus malaria RDTs
• Costs, probabilities and effectiveness were determined from primary
study database, published literature or expert opinion
• Tree Age Pro Health care 2009 software was used for decision
• PPP_defence\Model structure_treeAge.doc
Description of diagnostic strategies
Clinical diagnosis (Standard of care)
• Patient arrives at health facility and meets records assistant
• Records assistant gives patient number & collects patient information
• Patient is examined by clinician; history taking on signs & symptoms,
• If patient has any classical signs of malaria, they are clinically
diagnosed to have malaria and treated
Description of diagnostic strategies 2
Clinical diagnosis + malaria RDTs (Competing alternative)
• Patient undergoes process similar to standard of care
• Clinician requests for malaria RDT on patient
• A sample of whole blood is obtained from finger prick of patient onto
RDT strip which has been open for less than 2 minutes
• Health worker observes for colour change on the RDT strip and
interprets results and treats the patient
Assumptions made in the model
• The model is assessing the impact of the strategies on a single episode
• If a patient was diagnosed and prescribed treatment, it was assumed
that the patient received a full course of treatment from the health
• If a patient was wrongly diagnosed and given treatment, he/she did
not cure and had to seek treatment again from a health facility
• Health centres that participated in the study received their laboratory
supplies through Malaria Consortium and medicines from National
• Primary study was health facility based in three facilities from each of
four malaria transmission zones.
District Zone Prevalence
Iganga Hyper-endemic <70%
Mubende Meso-endemic 20-70%
Kapchorwa Hypo-endemic >20%
Mbale/Jinja towns Urban and peri-urban
• Health facilities in study were staffed by clinical officers, midwives,
• Most malaria episodes were diagnosed and managed at community
level and health centres level II and III
• Laboratory supplies (RDTs, cotton & gloves)
• Medicines for treatment
• Programme costs (training, support & supervision, distribution of
• Intangible costs
• Health workers time, health facility infrastructure, water and
electricity were not evaluated
• Probability (Positive, treated); (Positive, not treated)
• Probability (Negative, treated); (Negative, not treated)
• Sensitivity and Specificity of diagnostic strategies
• Proportion of malaria patients correctly managed
• Costs collected & analyzed from provider perspective
• Data collection forms were pre-tested
• Estimates of clinician time, medicines, RDTs laboratory supplies, and
programme costs got from the database of primary study by Malaria
• Costs of medicines were obtained from National Medicines Stores
• Sensitivities & specificities of diagnostic strategies with respect to
expert microscopy were obtained from published literature or expert
• Each variable was defined and given a variable name prior to creating
a code sheet.
• Data was double-entered in Epidata 3.1; errors were identified and
• All changes in the original data were documented and adequately
• The dataset was regularly backed-up.
• A separate dataset was created for analysis.
• Analysis was done in SPSS version 10 and cost analysis was done in
Microsoft Excel 2007 and TreeAge Pro 2009.
• The original data, final database and study analyses were archived and
Characteristics of hypothetical cohort
There was significant difference in the populations in the two
Cost of resources in RDT use at lower
health centres in Uganda
The total cost of implementing malaria RDTs was USD 10,447, and more
than 50% of this cost was due to programme costs.
Model structure & transitional
Model structure PPP_defence\Model structure.doc
The Clinical diagnosis strategy has an ICER of USD 17,306 per patient
correctly managed as compared to the clinical diagnosis plus malaria
RDTs. Hence, the clinical diagnosis strategy is the less cost-effective
Two-way sensitivity analysis 1
Figure 3. Cost of false positives treated versus probability of positive
and treated 20
Two-way sensitivity analysis 2
Figure 3. Cost of true positive patients treated versus probability of
positive and treated 21
Two-way sensitivity analysis 3
• For all probabilities of patients positive and treated in both strategies
and costs of true positive and treated, the CD + RDT strategy
remained the cost-effective choice.
• CD + RDT strategy remained cost-effective even when costs of false
positives and treated were varied.
• However, the CD + RDT strategy was the optimal choice in diagnosis
and treatment of malaria up to a threshold probability of P (Positive,
treated) of 0.8. It also became less cost-effective with increase in cost
Cost of resources in RDT use at lower
health centres in Uganda
• Total cost of implementing malaria RDTs was about USD 10,500,
more than 50% was due to programme costs. It implies widespread
implementation of malaria RDTs can reduce this cost.
• Micro-costing was used as it estimates the opportunity cost more
closely (Gold 1996)
• Some costs were not analyzed because they were common to both
strategies or were likely to confirm the result (Drummond et al 2004)
• Some costs could have been an over-estimate because of their source,
Malaria Consortium (partner of MoH in development).
• Cost of treatment may be an underestimate as a some patients get
additional health care from private sector (MMV 2008)
• Resources identified for analysis were similar to those identified by
other researchers Shillcutt 2008, Bualombai 2003, Chanda 2009
• The CD strategy had an ICER of USD 17,306 per patient correctly
treated as compared to the CD + RDTs strategy.
• Further evaluation of dominance; CD + RDTs strategy dominated
CD strategy under deterministic conditions but was not statistically
significant under stochastic analysis
• This could have been due to time frame of analysis, or the physical
units used for effectiveness
• However, marginal cost and effectiveness were used in analysis
• Two-way sensitivity analysis where costs for true positive cases and
false positive cases were varied against the probability of positive cases
getting treatment; CD + RDTs strategy remained the optimal choice
over the ranges.
• Similarly, other studies found that malaria RDTs were more likely to
be cost-effective in areas of low to moderate malaria transmission and
high clinician adherence to malaria test results (Shillcutt 2008,
• However, these results applied to public health facilities and may not
be applicable to private and informal health sectors.
• Viewpoint of study was health care provider; MoH Uganda and
• Some parameter such as sensitivity of clinical diagnosis was estimated
from expert opinion since it was not available in literature. Specificity
of CD was assumed to be zero
• Effectiveness measure was an intermediate one.
• Limitations; it was assumed that only febrile patients ended up in
study and some prescribers may have altered their practice because of
study (Hawthorne effect)
• The cost of implementing malaria RDTs at lower health centres in
Uganda was USD 10,447.0 and more than 50% of this cost was due
to programme costs.
• The CD + RDT strategy is more cost-effective than CD strategy at
lower health centres in Uganda.
• This study provides more evidence for MoH in advocating for Clinical
diagnosis plus malaria RDTs strategy at lower health centres whilst
advising clinicians to adhere to the test results.
• Further research should be done to answer questions of cost-
effectiveness of malaria diagnostic strategies in private and informal
health facilities, at lower health centres in different malaria
• Also a generalized cost-effectiveness analysis of malaria diagnostic
strategies is warranted.
• Supervisors and Makerere CEU
• Malaria Consortium, district and facility authorities
• Department of Pharmacy, Mak