Among recently diagnosed HIV-infected women in the Democratic Republic of Congo, we found that complete uptake of recommended PMTCT services was considerably lower among women with high TD, even after adjusting for covariates. Further, in alignment with our hypothesis concerning the mechanisms by which CCTs affect health behaviors, we observed evidence suggesting that the negative effects of high TD on complete uptake of services were mitigated by CCT. Though our results suggest a similar antagonist of CCT on the negative effect of TD on retention in care, our study did not, however, have enough power to access modification of such a small effect.
Measures of both morbidity and mortality are believed to contribute to high levels of TD, as they reduce utility and opportunity for consumption in the future, respectively [33]–[35]. Thus, it was surprising that our analyses revealed no significant association between discounting and several factors that could affect one’s health and survivability, including wealth and food insecurity. Indeed, the negative association between food insecurity and high discount rate was contrary to what we expected given previous research [33]. One possible explanation for these findings is that, while poverty could lead to high discount rates, high discount rates could positively affect wealth accumulation (i.e., inverse causality). Though high levels temporal discounting has often been linked to behaviors that perpetuate the cycle of poverty and ill health among low-income populations [34], discounted utility can – to some extent – be fiscally prudent. The grounds for this prudency are: 1) positive interest rates (i.e., $55 received now would accrue interest and could be worth more than $60 in one year), and 2) future uncertainty (i.e., given the possibility of events that may threaten future utility, $55 today carries definite value, whereas the value of $60 in one year is uncertain) [36]. Thus, placing more value on immediate rewards relative to those received in one year could – to some degree – be a rational adaptive response, particularly in situations of extreme uncertainty.
Another possible explanation for the negative association between discounting and food insecurity, and the lack of association between discounting and wealth, may lie in the unique life-situation faced by our study population at enrollment: pregnancy. It could be possible that pregnant women experiencing scarcity may prefer to receive money at a later date, after their delivery, in order to ensure they have adequate resources to support their baby. Though poverty and food insecurity could lead to high discount rates, this effect could be tempered or counteracted by fear for the well-being of their unborn child. It is difficult to assess if this might be occurring, however, without data on parental concern or a comparison group of non-pregnant women.
With regards to the impact of TD on our key outcomes, the stronger association between TD and uptake of available services, relative to retention to care, could suggest that TD is more closely linked to decision-making upon attendance at clinic appointments, including the acceptance of all proposed services (e.g., provision of blood samples). As with retention, no study to date has assessed the impact of temporal discounting on uptake of PMTCT services, so we are unable to compare our results to other findings. Several studies, however, have assessed the effect of discounting on adherence to ART (i.e., taking the ART medication as prescribed): in a study of HIV-infected adults in Uganda, investigators Linnemayr and Stecher found that individuals with a high discount rate exhibited significantly lower medication adherence than those with a low discount rate [39]. Similarly, in a study of HIV-positive adults on ART in Kenya, Thirumurthy et al. reported lower medication adherence among those with a high discount rate, though this association was not statistically significant, and that high discount rates were associated with significantly higher risk of mortality [40]. Though not directly comparable to our results surrounding uptake of PMTCT services, these previous findings may also suggest that TD may influence individuals’ decisions surrounding HIV care, beyond attendance at clinic visits.
Acceptance of the necessary testing and treatment, in addition to retaining in care, represent crucial steps within the PMTCT cascade [41]. Uptake of these services may be especially challenging in the Democratic Republic of Congo, as many women may be afraid to publicly or privately acknowledge their diagnosis; HIV is highly stigmatized and is often perceived as a moral failing, thus accepting HIV-specific services (and acknowledging one’s diagnosis in doing so) can have damaging social and personal implications [42], [43]. Retention, by contrast, is measured by attendance at routine clinic visits for prenatal and infant care, and thus does not implicate an acknowledgement of – and reaction to – one’s HIV status. Strong sources of external motivation may then be necessary for many women to overcome their fears surrounding the acknowledgement of their HIV status via uptake of HIV-specific services, beyond retention in care.
One such source of motivation may be the coverage of transport costs upon arrival for routine clinic visits. In a previous study using the same data, Yotebieng et al. found evidence suggesting that CCT improves retention by mitigating the cost of clinic attendance [44]. Among women wishing to attend clinic visits as part of routine prenatal or infant care, but who are unable to do so because of travel difficulties, CCT could provide enough money to cover the cost of transport. Women exhibiting high levels of TD from the present may experience a strong impetus to accept HIV-related services, as they are offered immediate monetary incentive for these behaviors and may overweight the value of this incentive. The immediate costs of clinic attendance, too, may be over-weighted among these women, and so they may be more likely to accept HIV-services to recoup their losses. These sources of motivation, rooted in the provision of CCT, could potentially overcome the objections to acceptance HIV-related services among those who are afraid to acknowledge their HIV status.
Our findings signify qualitative trends that could provide evidence for our hypothesis; however, they do not reflect definitive findings regarding the mechanisms of CCT. We interpreted our results as indicating potential antagonism between high TD and CCT on uptake of services, despite the limited statistical power of this study. Though moderate in size, the parent study was designed to detect the main effects of CCT, not interaction between TD and CCT, and only a small proportion of the study population exhibited low TD.
Further, we were unable to directly assess whether participants displayed time-inconsistent (or present biased) preferences, which could be an additional barrier to behaviors involving immediate costs and delayed benefit [45]. Though high discounting from the present could be indicative of present bias, it may also reflect high levels of time-consistent discounting (i.e., general rates of discounting, rather than an inordinate preference for the present) [46], [47]. It should also be noted that our findings reflect the context, population and parameters of this specific CCT intervention, and may not be generalizeable to other interventions involving economic incentives. There is some disagreement as to the appropriate method to measure TD, especially within low-income countries. Most measures of discounting were developed for a Western setting, and few studies have evaluated the use of existing measures across different cultural and economic contexts. Some argue that effort tasks (i.e., where individuals are asked to choose between earlier or later activities) should be preferred over monetary tasks (i.e., where individuals are asked to choose between earlier or later amounts of money), especially in low-income settings [48], [49]. The rationale for this argument is that the monetary discount rates in low-income contexts may reflect transient economic needs and not general intertemporal preferences. Recent studies have provided compelling evidence that, for those in poverty whose income is variable and subject to shocks, measures of discounting may capture volatile economic circumstances and not entrenched temporal preferences [50], [51]. As we did not measure discounting levels before and after shifts in economic circumstances, we do not know whether our measures of discounting reflect intrinsic preferences or situational constraints. Nonetheless, our study is among the first to explore any potential mechanisms by which CCT are hypothesized to improve HIV treatment and prevention [52] and high monetary discounting, regardless of cause or permanence, could arguably have similar effects on behavior in the immediate future when monetary rewards are offered, as is the case in our study.