Introducing rapid tests for malaria into the retail sector: what are the unintended consequences?

Theoretical background

Anthropological approaches to evaluation, with a commitment to inductive, holistic research, are well suited to analysing the consequences of programmes beyond those specified in proposals and protocols. It has been suggested by Kleinman17 that this type of analysis is one of the key contributions that medical anthropologists can make to global health.18 ,19 His interest stems from Merton's original formulation of the unanticipated consequences of purposive social action as the unforeseen effects of actions with a motive, a choice among alternatives, a goal and a process (such large-scale government programmes, but also complex intervention trials).16 ,17 Merton offers several reasons for their emergence including unforeseen shifts in context (such as ongoing economic or social changes); the immediacy of the interests of those formulating the programme; and the institutional values that shape the thoughts of programme managers or policymakers which make it hard to see beyond their particular paradigm.16 While this approach is useful in pointing to the complexity of introducing programmes to promote social change, it offers relatively little in terms of understanding how local, everyday practices shape interventions. Moreover, it offers no means of understanding how interconnections between the intended and unintended consequences of an intervention may come about and, importantly for this paper, how one may influence the other.

This paper draws on post-structuralist theories of the microprocesses of sociomaterial relations, sometimes referred to in the literature as theories of assemblage, to better understand and develop an analysis of the unintended consequences of introducing mRDTs into the retail sector. Drawing on the French philosopher Deleuze, these theories have been influential in anthropology and sociology for their ability to account for the interaction between society, science and technology and local processes of social change.20 ,21 This gives ideas of assemblage an obvious relevance to the analysis of the introduction of mRDTs and also provides a pertinent frame for understanding Ugandan drug shops more generally. Like other ethnographic and social analyses of drug sellers and informal private providers,22 ,23 the literature on Ugandan drug shops has shown these to be spaces that provide an awkward mix of services, combining disparate elements from different social arenas and in which those selling drugs do not necessarily adhere to a single logic (biomedical or economic, scientific or religious) but are more likely to draw on and display forms of practice that come from multiple institutional settings.3 ,24 Analyses that attend to the structural position of drug sellers tend to present them as existing in an odd and often insecure position, as liminal actors occupying in an in-between social space.3 Post-structuralist accounts attend to the ways in which social fields are assembled rather than structured and suggest that the analyst shifts the focus. Instead of mapping the contradictions that shoot through everyday practice, findings seek to account for the ways in which heterogeneous elements of the social and material world (from medicines, to social norms, desires, people, buildings and illness) come together and cohere, influencing and impacting on one another.25

In terms of accounting for the intended consequences of the introduction of mRDTs into drug shops, theories of assemblage draw our attention to the ways in which programmes or projects work by establishing relationships between the mRDT and the use of malaria medicines and that they fail when they do not. Yet, as this intended relationship is established, relations between the other social and material elements within the shop emerge and it is these processes (sometimes quite self-consciously arranged and organised by DSVs and their clients) that create the unintended consequences of the intervention.

A focus on the unintended consequences through the lens of assemblage theories therefore makes the connections (or sociomaterial relations) between already existing elements of the drug shop and those that are put in place through the actions of those implementing a project or programme the key sites for analysis. Through a focus on everyday practice, this paper analyses the processes through which mRDTs became part of the assemblages that constitute the drug shops in Mukono. Combining qualitative and quantitative data, it explores three sets of connections that were key to shaping the uses of the mRDT in the drug shops: local markets, regulation, diagnosis and trust; blood tests, disease and the generation of trust between DSV and their clients; mRDTs, pharmaceutical markets and the desire for effective medicine.

Study setting

Between October 2010 and December 2011, a cluster-randomised trial was carried out in 59 registered drug shops grouped into 20 geographical clusters in a periurban area in Mukono, Uganda. Staff from 65 drug shops had been invited to join the trial, though 6 shops failed to attend the training and so were excluded. Following invitation and consent to join the project, clusters of drug shops were randomised to receive training in one of two methods to diagnose malaria: 10 clusters were trained to use clinical signs and symptoms (the presumptive arm), and 10 clusters were trained to use mRDTs (the mRDT arm). All DSVs attended a 3-day training in taking patient history, recognising malaria, prescribing ACTs and rectal artesunate, when and how to refer patients into the health service. In addition, the DSVs in the intervention arm received an additional day’s training on when and how to use an mRDT. DSVs were also trained in record keeping and preparing blood slides for the reference microscopy conducted on each client, which were used by the trial to evaluate whether the treatment decisions were appropriate or not. All DSVs were given ACTs for free (artemether-lumefantrine) to sell to patients at an agreed low retail price (child: 1000 Uganda shilling, adult: 3000 Uganda shilling, equivalent to US$0.40 and US$1.20, respectively), disposable gloves, cotton wool, lancets, blood slides, safety boxes for sharps disposal, triplicate carbon copy patient registers for record keeping for the project, and to provide a referral form should the client be considered too ill to be treated in the drug shop. All DSVs in the mRDT arm were additionally provided with a continuous supply of mRDTs for free, and were asked to sell these to patients at a fixed price (500 Uganda shilling, equivalent to US$0.20). The DSVs in the mRDT intervention arm were provided with signs to place outside their shop advertising the availability of mRDTs. In both arms, community sensitisation through Village Health Teams informed local residents about mRDTs underlining the fact that not all fevers are malaria; that it was beneficial to test for malaria before providing treatment and that mRDTs were available locally from trained DSVs and public health facilities. A detailed description of the various elements of the intervention is provided in the main trial paper.15

Focus group discussions

Focus group discussions (FGDs) were conducted with three subgroups: drug shop clients, DSVs and health workers at local public facilities and a private not for profit hospital. In all, 21 FGDs were conducted (see table 1) and each FGD had at least 5 and no more than 13 participants.

The participants were given a description of the project when they were invited to participate in the FGD. On the day of the FGD, the project was described again and the participants were asked to sign a sheet to consent in the research. They were advised that at any point they could leave the FGD and that there was no requirement for them to stay. Focus groups were conducted in Luganda and/or English, and in all FGDs notes were taken and the discussion recorded digitally.

The FGDs were conducted in a Sunday school building except for two that were conducted at health facilities. Each focus group was attended by three social scientists acting as facilitator, note taker or organiser of the session. The topics discussed were local treatment-seeking behaviour, experiences at drug shops, experiences with rapid diagnostic tests and experiences with referral. The latter was a particular interest in the study and this focus on referral from drug shops into public sector health facilities was the reason behind the inclusion of health workers from local health centres and hospitals in the research. An emphasis was placed on describing and understanding actual experience rather than eliciting generalised or ideal statements. The facilitators also allowed conversations to develop between participants when important but unanticipated themes arose. Participants received money to cover the cost of their travel and refreshments.

Recordings were transcribed, translated in English and imported into Nvivo (QSR International, Cambridge, Massachusetts, USA), a qualitative data analysis software package for the aggregation of data by codes.

Coding and analysing qualitative data is an interpretive endeavour and each FGD was coded by one social scientist who had collected the data with close support and supervision from the anthropologist (EH). A coding scheme for each group was drawn up through a combination of themes emerging from the transcript coupled with predefined areas of interest identified from the main intervention (mostly the interaction between different components of the trial) and from anthropological theory on informal health providers. The coding scheme was adapted as necessary to reflect new themes emerging and the anthropologist provided input at the conceptual level of analysis. The main findings were discussed with the principal investigator of the main trial, the anthropologist and the project's social scientists during a series of face-to-face meetings that took place everyday over the course of a week.

Follow-up questionnaires

DSVs in all clusters recorded the names and contact details of clients who visited their shops for a fever or history of fever. Interviewers then visited DSVs according to a randomised schedule to identify recent clients and carried out questionnaire interviews (n=646) at the client's home after 4 days, and again between 10 and 14 days after visiting a drug shop. The questionnaires were conducted to elicit reports on the economic costs associated with treatment of the recent febrile episode at a drug shop, whether they had been offered and had accepted to be tested with an mRDT, their reported results; the type and quantities of drugs purchased and any subsequent costs over the next 10–14 days; and exploring perceptions of mRDTs, their knowledge and understanding of the tests, treatment-seeking behaviour and referrals. For clients who were under 18, questionnaires were conducted with the adult who attended the drug shop with the child; clients under 18 who attended the DSV alone were excluded from the study. Clients who did participate or consent to both questionnaires or did not meet the overall study inclusion criteria were excluded from the follow-up analysis. The analysis of open questions was coded by CH using Nvivo (QSR International, Cambridge, Massachusetts, USA) and, where appropriate, imported into Stata V.13 (StataCorp LP) and converted into ordinal or categorical variables for descriptive quantitative analysis in conjunction with quantitative responses.

Ethical clearance was granted by the London School of Hygiene and Tropical Medicine and the Uganda National Council for Science and Technology.

Context: diagnostic practice and the generation of trust

According to government policy, registered drug shops are places in which over-the-counter medication may be bought and sold. Legally, drug shops are distinct from private clinics and expected to provide a limited range of products for sale, which excludes prescription medication such as antibiotics. In Mukono, however, categories of drug sellers were fluid and clients were unable to categorise private sector providers effectively during FGDs. The fluidity of this categorisation was also reflected in the heterogeneous nature of the drug shops in the trial. Most DSVs reported that they provided diagnostic advice and sales of medicine only but at least 2 out of the 59 study shops also offered in-patient services and provided intravenous medication. At some, the registered owner (who by law had to have some formal biomedical training) would be in the shop for most of its opening hours, providing relatively constant access to a health provider.

As has been reported elsewhere among clients in Uganda, the care seekers attending shops in Mukono appreciated the good social relations often found in these spaces3 but maintained a generalised suspicion of DSVs’ motives and skills.27 DSVs could be respectful and kind; attentive to the psychological, social and spiritual needs of their clients; understanding of and responsive to their economic constraints (providing of credit and selling cheaper or incomplete doses) and keen to use their skill and expertise. As one woman who had visited a drug shop in the mRDT arm of the trial with her child explained, these social relations often revolved around a desire for powerful pharmaceutical drugs to manage illness in them and in their children.We had five thousand shillings (2 US$) in our pocket. But because the sickness was severe, she wrote us a bill of eighteen thousand shillings (7 US$). Yet, she had given us a plenty of drugs in a polythene bag. Then I told her, “I have come with five thousand [shillings],” and she said, “Aha, there is no problem. You can even come back tomorrow”, and she didn't even know us…[later in the discussion]…This health worker gives you drugs and you also look at it and say, “as I have never seen such a quantity of drugs!” and you simply say that. Oh, but the drugs are worth the money you pay!FGD with mothers in the mRDT arm of the trial

Attending a drug shop was also a risky act. Businesses were infrequently visited by government regulators. In all three groups of participants in the FGDs, the following narratives echoed: drug shops are often run by those who claimed expertise but had none, who knowingly sold stolen, dangerous, expired or counterfeit drugs and were only interested in making as large a profit as possible. Accounts of particular instances of receiving poor quality care at drug shops and its consequences for children's health were commonplace. As previously described by Birungi,28 injections were a popular form of medicine, a sign of good care but raised particular concerns about the infection and illness that could be caused when unskilled DSVs administered injections to small children. Moreover, in the context of ongoing concerns about how to pay for treatment, clients expressed concern that they could be overcharged for medication by exploitative DSVs.

Where some FGD participants considered it to be almost impossible to distinguish between skilled and unskilled, trustworthy and untrustworthy DSVs, others described how they acted to protect themselves against the risks associated with attending a drug shop. Previous treatment success, recommendations from neighbours and the DSV's diagnostic assessments played a central role in identifying an effective, good quality and caring provider. Below, a group of women described how they evaluated a DSV's skill through the questions that s/he asked about the child's symptoms.P6: When you look there, when you are just passing by, or maybe if you have had your child fall sick, some people base [their assessment of a drug shop] on the way the drug shop looks and then see whether they [the DSVs] are skilled, drugs are in plenty. But, when you pass there, [you may see that] the drugs that are there are few. Now you can despise the place yet in the real, actual sense she has the idea.SS1: What do you mean by that thing of ‘idea’?P2: What she wants to say is that she knows what she does.SS2: How can you come to know that someone is knowledgeable?P6: For me what I can explain, you can take a child and you tell her, “Health worker, I feel the child has high temperature.” You have not yet told her that it seems the child is sick with malaria or what but for her she starts asking you that…that is to say she knows the conditions of children. Most of the time she asks you, “Has he done this, do you feel his head is a bit hot (some high temperature), how was he at night, did he at any time have diarrhoea?” Okay things like that. That is to say, you have not told her but she says correct things and you tell her, “yes, health worker that is how it was”, before she measures his temperature to see the temperature…and she asks you about the condition of the child, or how he was, whether it is malaria or it is another thing.FGD with mothers in the non-RDT arm

The diagnostic process underway in these drug shops that existed prior to or without the introduction of mRDTs had a productive role that went beyond the imperative of identifying disease in order to provide treatment. In this heterogeneous and poorly regulated market, it was an important demonstration of medical knowledge, and acted on the relationship between the DSV and the client, engendering trust in a space in which questions around motivation of the vendor and the type of service that they provided were under constant scrutiny.

mRDTs’ impact on decision-making about where to seek care

According to the FGDs, mRDTs had not been available in the drug shops in Mukono prior to the intervention but when they were introduced, they were popular among DSVs and clients. Characteristics of the clients interviewed on day 4 after drug shop consultation are presented in table 2. Following refusal of consent (12) and loss to follow-up (130), clients or carers comprised 251 who had visited a drug shop in the presumptive diagnosis arm and 253 in the mRDT arm of the trial. In both arms of the trial, half of those seeking care and medication from a drug shop were children under the age of 5, at 49.4% in the presumptive arm and 51.8% in the mRDT arm. Of the 253 clients presenting with symptoms of a febrile illness in the mRDT arm, 97.6% reported that they had purchased an mRDT. There were occasional reports of mRDTs also being used in the presumptive arm (7/251 cases), findings that correspond to those from the FGDs. This suggests that certain DSVs were making the choice to purchase and offer these tests for sale (see table 2).

Almost three-quarters of these care seekers in the mRDT arm of the trial (72% of the 253 respondents) reported in the questionnaires that they knew about the availability of mRDTs prior to the recent visit to the shop. Of those, the majority (73% of the 172 respondents) considered that this had an impact on their decision-making and why they had chosen to seek care in a particular drug shop. During the FGDs, health workers also noted changes in treatment-seeking patterns.P4: It [the introduction of mRDTs] has not affected us badly because it has helped to reduce the number of patients [that we see] and they [the DSVs] also get some money…[laughter]…P2: In my opinion, if a person has a sick child and they go to the drug shop and they take off some blood in order to really confirm that she has malaria they get to know the truth. They [the client] are happy because instead of going to the hospital they are being treated nearer their home.FGD with health workers from HC II in the intervention arm siteP5: We now get fewer cases of malaria at our hospital. You know what that means for our income, which is a bit negative for us but for the community it's okay because they have the services nearer.FGD with health workers in HC IV and the not-for-profit hospital

As mRDTs were introduced into drug shops, the process of finding out about disease maintained its ability to engender trust between the vendor and the client and the benefits of mRDTs were interpreted through this lens. The first quotation below links this to a process of being able to see and then act on a disease, while the second connected the care available at drug shops to the larger health facilities.Okay for me the different thing that I have seen, we have been suffering and what made us like the hospital are the blood tests. Yet, in these drug shops, they would simply inject you without testing your blood to see what type of fever it is. But for me the different thing that I have seen now, [is that] they [DSVs] test blood. Now we have come to completely dislike the public sector hospital because that [blood testing] is what has been mostly taking us there.FGD with mothers in the mRDT arm of the trialP1: The majority of us used to go to those big health facilities but now we go there to drug shops and clinics and they give you treatment.P5: I think it [testing] has been good for them [the drug shops vendors] because the majority of people have trust in them. Previously, they would despise those clinics.Facilitator: Previously you never used to put trust in them. Why were you going to other health facilities in the past?P4: Blood, they never used to test blood.Many Ps: Hmmm, bloodFGD with mothers in the mRDT arm of the trial

Refracted through this new logic of testing before treating illness, previous diagnostic practices in the drug shop were conceptually reconfigured. The practice of describing and suggesting symptoms so that they could be matched to available medicines was often recast as guess work and was contrasted with the definitive knowledge that the mRDT provided.I think work has been simplified because those days, before this project started, we would just try to guess what's going on, how to treat the patient. But after testing, I think we know exactly what we are doing and even the patients themselves realize it and that's why they come. But of course, those days they [patients] would have that negative…what? [attitude]…but now they know if you go there, they test you and they give you…what?…[They] will cure you. Of course that confidence has increased in them. Ya, they are confident that they going to get what? [That they are going to get] okay.FGD with DSVs in the mRDT arm of the trial

The claims by DSVs and clients of the benefits of using mRDTs were significantly broader than those stated in the project protocol. A few clients reported that the mRDTs in the shops they visited diagnosed malaria, typhoid and syphilis (eg, “what came out of the test was that my child had malaria and syphilis,” FGD with mothers in the mRDT arm of the trial). Mostly, however, the capacity of the tests was not specified but was described by DSVs and their clients as creating a general change in practice that enabled DSVs to know the causes of and necessary treatment for ‘illness’ rather than just identifying those patients with/without malaria.Ever since we started this project, it has made us treat the patients well because before we have been treating a person simply. A person comes, “I am suffering from malaria.” You would touch on him, measure the temperature and then give him drugs. But now you treat something you know and he also leaves understanding that really, they have treated me because you tested his blood.FGD with DSVs in the mRDT arm of the trial

From the questionnaires in the mRDT arm of the trial, just under half who had received an mRDT reported that they had not seen the results of the test (see table 3). Moreover, when they reported on their interpretation of the role in the care-seeking process, 47% of the 253 care seekers reported that the use of a test either enabled the DSV ‘to know’ what was causing their illness or fever, or ‘to treat what they know’.

Reports of the use of the mRDT to enable a definitive diagnosis (ie, of any illness) emerged in focus groups with DSVs and clients. In the quotation below, a mother who attended a drug shop in the mRDT arm underscores how the mRDT was imagined as enabling decisive knowledge, the effective targeting of treatment and that this would provide better value for money than in a system where malaria was treated presumptively.That five hundred shillings [for the mRDT], and she treats what she has known. Because some time back someone could come and she (DSV) would say, “it seems that he has malaria,” and she would prescribe different drugs for you and she would tell you that you had to pay this and this amount of money. But now, she takes blood from you and she gets to know that she has tested your blood and you have malaria and the drugs she gives you are for malaria.FGD with mothers in the mRDT arm of the trial

The DSV and the client quoted above also provide details of what makes up a good visit to a drug shop, that the client has been tested and purchased effective medication. The final section of the results considers the ways in which mRDTs acted on clients’ desires for effective medicines and reshaped local pharmaceutical markets in Mukono.

mRDTs increase medicine sales in drug shops

Like many of those working on the introduction of mRDTs in other sectors of health systems in Africa, the researchers involved in the mRDT trial in Mukono were concerned with the ways in which DSVs would manage negative test results. A negative mRDT could mean that drug sellers would not make a sale, reducing their income, and disincentivise testing among DSVs. We have previously shown how RDT use can also affect a DSV's reputation.27 Negative results carry a risk of revealing diagnostic uncertainty which could be interpreted as a lack of knowledge and skill in managing a patient's illness. During the FGDs, some clients did express concern about the ability of the test to always provide an accurate result. During the FGDs with the DSVs, however, concerns about negative test results were relatively rare. A close look at the interpretation of the capacity of the mRDT, its perceived ability to create definitive knowledge around disease and the way in which its introduction set a new standard for care based on testing blood helps to make sense of this.

When DSVs described anxiety around the management of a negative mRDT result, this was most often linked to encounters with clients who were convinced that they had malaria and wanted to purchase an ACT, but what also appears to have been critical in managing a negative result was the action taken by the DSV following the generation of the result. Referring a client who was not considered to be dangerously sick, especially if the DSV did not provide any medication first, was considered to be economically perilous and also hazardous to their reputation and so was infrequent.27 Instead, DSVs could act on a negative test result by providing what was understood by the client to be effective treatment. Below, two different mothers from different focus groups explained a positive outcome of their trip to the drug shop using mRDTs with their children. In the first description, the child is diagnosed with malaria, and in the second the child tests negative to malaria but both received an injection and other medication.After testing the child's blood, she gave him an injection and ended there. The malaria was very severe and she gave him an injection. She even gave me tablets that were yellow and we went back home. She even told me that I have to give to the child plenty to drink.FGD with Mothers in the mRDT arm of the trialFor me, I took a child there [to the drug shop] who had diarrhoea and had fever. They tested him but there was no malaria. They gave him drugs and syrup and they also gave him four injections. They gave him four injections and he got well.FGD with Mothers in the mRDT arm of the trial

The imperative to receive medicines as an outcome of drug shop encounters in both arms of the trial was underscored by data from the questionnaires with care seekers, where 100% of those interviewed during follow-up reported having purchased some form of medication (table 4). Unlike in the FGDs, where medicines were often identified by their colour, whether they were in tablet form, syrup, injection or crushed into a drink, for the questionnaires clients were asked to specify wherever possible the type of medicine that they had purchased. In concordance with the overall findings from the trial,15 reported sales of ACTs were substantially lower in the mRDT arm than in the presumptive arm, though not entirely ruling out the sale of malaria medicines to patients with a negative test. In the mRDT arm of the trial, for the clients who reported having a positive mRDT result, the majority said that they had purchased an ACT (94%) either with (17.9%) or without (76.4%) an antibiotic. A similar percentage to those being treated for febrile illness in the presumptive arm purchased an antibiotic (18%; see table 4). As one of the FGD respondents explained, past experiences of giving malaria medication to her son during an illness episode continued to influence the clients’ preference and demands for particular medicines. Now, when you get to the health worker he tests and tells you there is no malaria and as a result you tell him, “Health worker, the other time you gave me these tablets and these tablets. So, still just give me maybe Camoquine or Chloroquine and I go and give [them to] him and I see if he will get better.”FGD with mothers in the mRDT arm of the trial

While the results of the FGDs suggest that there was a good deal of polypharmacy among those testing both positive and negative for malaria, the questionnaires suggested that the range of medicines sold to clients with a negative test result was greater than either those with a positive result or those who had not been tested at all. It appears that more mRDT-negative clients purchased an antibiotic compared with mRDT-positive or untested patients—37.6–18% and 15%, respectively. Overall, the median amount of money spent by clients in drug shops using mRDTs was higher than those in the presumptive arm and the difference was marginally greater among clients in the mRDT arm who received a negative test result.