**3. Results**

We successfully analysed 40981 women-hours of follow-up. They consisted of all patients who underwent Total Abdominal Hysterectomy at Amana, Mwananyamala and Temeke regional referral hospitals from March to October 2017. In essence, none of the patient seroconverted against HIV during follow-up period, making an HIV-incidence of 0/40981 patient-hours. Their pre-operative HIV-serostata included 19 (17.76%) who were *reactive* on both screening tests, thereby considered as seropositive for HIV anti-IgG; 84 (78.5%) *non-reactive* on both screening tests and considered seronegative for HIV anti-IgG. Moreover, 3 (2.8%) patients refused screening. Therefore, their HIV-serostata remained unknown to study investigators. Likewise, 1 (0.93%) patient had discordant results between the two screening tests. All efforts to determine her HIV-serostatus during the time of hospital stay were deemed unsuccessful. All studied patients were black Africans by origin. The median age was 42 (IQR: 37.5–47) years. Serial correlation between HIV-serostata and age of patients yielded statistically significant findings (γ = �0.514, P < 0.001). **Figure 1** below highlights the pattern of the observed serial correlation between chronological age and HIV-serostata of participants.

To test for potential confounding, other variables (i.e. +/� of comorbidities, reported recent (past 1-month) sexual (oral, vaginal/anal) history, 1-month prior blood -(products) transfusion episode and place of residence) were also tested but yielded nonsignificant statistical evidence of confounding (χ<sup>2</sup> -Yates corrected = 0.3833, df = 3). Converesely, it is worth noting that we also found a just significant correlation between reported marital and HIV stata (see **Tables 1** and **2** below) on univariate analysis during initial data exploration.

#### **Figure 1.**

*Serial correlation between HIV serostata and age among women who underwent total abdominal hysterectomy in Dar es Salaam regional referral hospitals,Tanzania (March–October 2017).*


*NOTE: Cochran–Mantel-Hanzel corrected <sup>χ</sup><sup>2</sup> value for the association = 8.0176, df = 3. <sup>1</sup> N = 107 but 3 individuals refused HIV screening and 1 had discordant HIV results.*

#### **Table 1.**

*Distribution of HIV by marital stata among women who underwent total abdominal hysterectomy in Dar es Salaam regional referral hospitals, Tanzania (March-October 2017).*


*N.B.: The model fitness was deemed desirable with LR value of 8.1142 and df = 2. \**

*Age was coded in a binary fashion with the cut-off of 45 years (average time for natural menopause?) and found to attain linearity with the logit function. Thus, Age < 45 was used as a reference group.*

*\*\*Marital status was collapsed to married/non-married in order to attain model specification of linearity with the logit function. Here, non-married was the reference group.*

#### **Table 2.**

*Findings on the multivariable logistic regression analysis of HIV-stata on ageing among women who underwent total abdominal hysterectomy in Dar es Salaam regional referral hospitals, Tanzania (March-October 2017).*
