**4. Discussion**

We detected *a statistically significant association for the probable drop in HIV-risk after age of 45 among hysterectomised women*. The finding translates to an average drop of about 30% in HIV-risk for each annual increase in life expectancy after the age of 45 years. From an evolutionary perspective, our findings are likely to be a reflection of *antagonistic pleiotropy* theory of ageing. It is evident in literature as part of a nonadaptive evolution of ageing process, as per reproductive senescence assumption. However, a word of caution - *it was a statistical estimate, out of secondary data analysis!* We therefore in a pioneering move, call for biologically plausible prospective study designs, to substantiate the exact cause of our current observed statistical puzzle!

By reflecting the nature of our target population, we derived *a possibility for hormonal interplay that potentiate HIV-entry in women during their reproductive age; and significantly drop upon cessation of menses or removal of their endocrine uteri*. Several studies in the past have suggested potential roles of female reproductive hormones in facilitating HI-viral entry and proliferation in female epithelial tissues [22–27]. For instance, Aaron Weinberg and his colleagues showed for the 1st time back in 2003, that HIV-1 induced β-defensin expression in human *oral epithelial cells, with subsequent HIV-1 replication blockage, by the β-defensin 2 & 3, via direct interaction with virions, and also through modulation of CXCR4-tropic* HIV-1 isolates [24]. To our views, Weinsberg's findings were novel even though the target was oral (rather than genital) epithelia [24]. Likewise, Morrison and colleagues performed an individual data 18-prospective studies incorporating 2-stages random effect meta-analysis with 43613 women-years of

follow-up, that resulted to an adjusted Hazard Ratio (aHR) of 1.5 (95% C.I.: 1.24-1.86) upon usage of Depot Medroxyprogesterone Acetate (DMPA) [22]. Morrison study controlled for incident known biological and behavioural risk factors [22]. Progesterone (and not oestradiol) was associated with increased risks in those studies [22, 23, 26, 27]. However, causality has not been established on this topic. We hope our study findings have added quantitative estimates of association between HIV and ageing endocrine processes in non-gravid female humans.

The estimated HIV burden (17.76%) reported in our study is higher than recently reported HIV/AIDS statistics in Tanzanian general population. Tanzania's HIV Impact Survey 2016–2017 reported HIV prevalence of 6.5% among women aged 15–64 years [28]. However, our study population was unlikely to be representative of all women in Tanzania. Another contrasting factor was our limited time of follow-up, in days rather than months or years, normally applied for HIV-seroconversion incidence studies. The zero incidence in HIV-seroconversion rate observed may also likely be a function of limited time of follow-up. Moreover, of special interest to our findings, was the link between removal of endocrine uterus (a marker of *ageing* process) and HIV-risk. On average, a linear association accounted for more than half the variation between HIVrisk and chronological age. Besides, the negative sign in the correlation estimate; signified a probable reduction in HIV-risk with increasing age. It was additional statistical evidence, besides the estimated odds ratio statistic found on the linear model. However, we wish to caution against potentiality for both Berkson's and ecological fallacies when generalizing our findings. We wish our findings to be taken as *a treasure hunt* rather than justifiable evidence at present. For instance, we see a *potential for malice*, upon generalisation of these findings at individual level. We wish to caution readers, that the statistics on reduction in HIV-risk with ageing was analysed at group level. Otherwise, the same finding has several alternative explanations.

First, although the current analysis involved all women who underwent hysterectomies during the study period, and hence the estimate unlikely to be due to sampling variability, we did not have a control group by design. Moreover, there was relatively fewer individuals, in old age (>65 years) category. The analysis was therefore underpowered for detection of the observation among *senior citizens* per se. Middle aged women constituted the majority in our study population. The young and middle age groups are evident to be the most affected by HIV in Tanzania [28]. However, that evidence is doubted in present day Tanzania [29], as HIV has achieved a stable chronic status at community level. Historical data on predominance of young/middle aged members on HIV statistics in Tanzania by their own do not rule out high HIV incidence/prevalence in old aged group. Thus, it is equally likely, that our current findings to be a *statistical artifact* than a real phenomenon. However, the fact that even with the notable *under-powered statistics*, the observation was still statistically significant; is worth *speculations* towards a probable real biological phenomenon. The view follows a common knowledge, that it is relatively difficult to attain statistical significance in under-powered dataset than adequately (over)- powered dataset. We therefore strongly call for biological and clinical research on this specific topic.

Likewise, Tanzania just like other sub-Sahara African countries has never included ≥ 65 years cohort in its national HIV/AIDS surveys. This is for a variety of reasons including assumption of HIV-infection as a disease of youth and young adults. That assumption is considered by authors as a complete myth at present. The reason for disputing that assumption as a myth has been published before [29]. Otherwise, data on HIV-statistics among ≥ 65 years in Tanzania are scanty. Of the few available ones, there is one with evidence that reported a *relatively* low point prevalence (2.1%) [30]. It constituted senior community-dwelling females in North-Eastern Tanzania [30]. Currently, we are hesitant to assume that single retrieved community prevalence study to be nationally representative. Given the obvious gap, it is naturally justified that further population-based studies are needed on this topic.

Moreover, there is a specific call for interventions targeting ageing process, and senior citizens morbid and mortal statistics the world over, to be derived from reliable and valid tools. For instance, there is growing evidence that most scales and indicators used for assessing senior citizens morbid conditions report indices with questionable reliability globally [31]. One member of our team has just published his findings [31], that showed the current global scales and indicators for assessing frailty to record reliability values that were lower than *greatest lower bound* (*glb*) reliability estimate [31]. Part of the challenge has been contributed by years-long tendency of editors and reviewers to consider Chronbach's alpha coefficient as *a sin qua non* for defining reliability index in scientific literature. We could not control all biases associated with our quantitative variables nor did we assume perfection in the literature cited in this chapter, against all systematic and measurement errors in them. The fact that the world is ageing fast, especially sub-Sahara African countries [32], calls for tools with appreciable precision and accuracy during data collection & reporting processes. Thus, studies on biological, clinical, public health, demographic as well as economic analyses of ageing processes are highly warranted globally. Our message given the current statistical findings, *some somatic maintenance properties are likely to be retained (reversed) after reproductive years in female humans!*

Likewise, we took a careful measure not to overlook the finding on HIV and marital stata. There was a rather strong ego to consider that finding as spurious, but indeed the decision was not supported by a logical flow of reasoning among investigators. In fact, we are still debating! Previous studies in similar settings yielded confusing results [33–35], with one study from a population based study in Tanzania that conferred an almost additional 50% risk (aOR: 1.49, 95% C.I.: 1.08–2.04) to remarried couples compared to single/cohabiting partners [35]. Otherwise, the fact that HIV-infection has been present in Tanzania since 1983 [29], and ante-retro viral drugs against HIV became available in mid-1990's [29]; justifies possibilities for residual vertical infection, among infants born with HIV from infected parent(s), to proceed to adulthood. Should this speculation be valid, assumptions related to exclusively acquired HIV risks after sexual maturity could be nullified. Evidently, a member of our research group once co-authored a population based study, among under-fives in a sub-urb of Tanzania, that realized potential sources of infectious ailments, via a domain of febrile illnesses [36]. However, details about the HIVassociated behavioural risks as well as impact of Ante-Retro-Viral drugs against HIV to babies born with the disease were beyond the objective of our current and previous works. Thus, we do believe there is a desperate need for future studies on the mechanisms behind preponderance of HIV infection among participants who reported in married category.

Lastly, much as we do believe our current findings to be reflective of antagonistic pleiotropy theory of ageing process, we wish to address some other important limitations of our findings. It has been a general consensus that women tend to outlive men in longevity studies [37–42]. Even though most scholars of the past (and likely the present?) still associate the reason(s) back to environmental causes [42]; there is a clear indication, that the concept to have an underlying biological and/or clinical causes [37–41]. Mysteriously to our current findings, there have been previous studies that reported female advantages in HIV-survival patterns whether or not on anteretroviral treatment in similar settings [43, 44]. Thus, female survival advantages tend to manifest both against HIV-risks as well as in HIV-infection whether or not the latter factor is associated with treatment. Besides, we could not set up an enough time follow-up study in order to arrive with our current conclusion. However, we do believe what we just showed (probably for the first time?) to the world to be an otherwise real biological phenomenon, that has been part of knowledge base in what is currently referred to as biogerontology.
