No association between fertility desire and HIV infections among men and women: Findings from community-based studies before and after implementation of an early antiretroviral therapy (ART) initiation program in the rural district of North-western Tanzania

Background: Fertility is associated with the desire to have children. The impacts of HIV and antiretroviral therapy (ART) on fertility are well known, but their impacts on the desire for children are less well known in Tanzania. We used data from two studies carried out at different periods of ART coverage in rural Tanzania to explore the relationship between HIV infection and fertility desires in men and women. Methods: We conducted secondary data analysis of the two community-based studies conducted in 2012 and 2017 in the Magu Health and Demographic system site, in Tanzania. Information on fertility desires, HIV status, and social–economic and demographic variables were analyzed. Fertility desire was defined as whether or not the participant wanted to bear a child in the next two years. The main analysis used log-binomial regression to assess the association between fertility desire and HIV infection. Results: In the 2012 study, 43% (95% CI 40.7-45.3) of men and 33.3% (95% CI 31.8 - 35.0) women wanted another child in the next two years. In 2017 the percentage rose to 55.7% (95% CI 53.6 - 57.8) in men and 41.5% (95% CI 39.8 - 43.1) in women. Although fertility desire in men and women were higher in HIV uninfected compared to HIV infected, age-adjusted analysis did not show a statistical significance difference in both studies (2012: PR=1.02, 95%CI 0.835 - 1.174, p<0.915 and 2017: PR = 0.90 95%CI 0.743 - 1.084 p= 0.262). Discussion: One-third of women and forty percent of men desired for fertility in 2012, while forty percent of women and nearly half of men desired for fertility in 2017. The data showed fertility desire, in 2012 and 2017 were not related to HIV infection in both periods of ART coverage.


Introduction
Fertility desire in both men and women is associated with fertility 1 . Fertility desire has been measured differently across Sub Saharan Africa (SSA) depending on the context of the question. It is described as the desire to stop childbearing 2 , whereas in the other paper, it is described as the desire to bear a child, within a given time period, and usually in women 3 . Fertility desire is of great interest to demographers and scientists as it forecasts future population fertility and the incidence of unwanted pregnancies. Due to differences in measurements of fertility desire, fertility desire reports vary across SSA countries, ranging from 8% to 82% 4 . In recent studies in SSA , it was reported that One-third of Malawian women 5 , half of Ethiopian women 6-10 , and two-thirds of Ugandan women desired for one or more child in the future 11 . In 2014, Ugandan study, reported a higher fertility desire in younger women and women with low social economic status 12 , whilst another qualitative study in Zambia, showed a low fertility desire in women with reduced parity and in the family with bigger than expected family size 13 .
HIV infection has been considered the major factors to influence fertility behaviour and fertility desire in women of reproductive age. In recent years, studies conducted in SSA had showed a lower fertility desire in women living with HIV (WLHIV) compared to HIV-uninfected women [14][15][16][17] . Yeatman and colleagues conducted a qualitative study and found that self-assessed likelihood of HIV infection reduced desire for fertility in women, while women who had a positive HIV test result reduced their fertility desire. WLHIV hides their fertility desire to be known to society because of the fear for mother-to-child transmission of HIV 18 . Fertility desire in WLHIV is largely due to HIV associated stigma 19 , while other women fear the physical health consequences of pregnancy and childbearing while living with HIV 20 . However, some studies have shown that HIV infection does not diminish the desire to be pregnant and bear a child 21 . Therefore the impact of HIV infection on fertility is well known, but the impact of HIV infection on fertility desire is equivocal. Antiretroviral therapy (ART) availability has reduced the risk of vertical HIV transmission and improved the wellbeing of the WLHIV. However, ART usage may motivate WLHIV to have more children despite HIV positive diagnosis. Studies in SSA have shown that ART usage increased fertility desires [19][20][21][22] . ART had been said to resume the quality of life for people living with HIV (PLHIV) and enhance their desire for children [23][24][25][26] . In Tanzania, up to 2012, ART initiation was based on diminishing immune cells markers (CD4 counts), with only 14% of PLHIV receiving ART in 2012 28 . In 2013, the Prevention of Mother to Child HIV Transmission (PMTCT) program ensured lifelong free ART was given to all pregnant women diagnosed with HIV regardless of their disease stage, viral or CD4 cell counts 29 . In 2016, the universal HIV test and Treat (UTT) policy provided ART to all HIV-infected individuals regardless of their immune status 30,31 The impact of ART and/or HIV on population fertility rate and desire may be more pronounced in countries with high HIV prevalence, HIV testing rate, and ART coverage 32 .
In 2018, Tanzanian data shows that, for women aged 15 years and above, 82% have tested for HIV in the past year and 82% of those found positive have initiated ART 33 . It is unclear whether the increased ART availability due to earlier ART initiation and PMTCT option B plus has impacted on the fertility desire in WLHIV. In this analysis, we used data from two repeated community-based studies carried out in 2012 and 2017 to explore the levels of fertility desires and its association with HIV infection in men and women.

Ethical consideration
All participants who contributed data for analysis provided written informed consent for study participation and publication of the results. Parents or any care giver consented on behalf of study participants who were aged less than 18 years. Ethical approvals were obtained from the Lake Zone Institutional Review Board (MR/53/100/513), the Ethical Review Committee of Kilimanjaro Christian Medical College of the Tumaini University of Tanzania (certificate number 2440) and from the London School of Hygiene and Tropical Medicine. (LSHTM Ethics Ref: 8623).

Study setting and design
The 2012 and 2017 studies were conducted in a health and demographic sentinel surveillance population in the Magu District of north-western Tanzania (Magu HDSS). Magu HDSS has a population of 45,000 with the majority, dwelling in the rural areas, belonging to Sukuma ethnic group and of the Christian religion. The main economic activities are small-scale farming, livestock keeping, and petty businesses involving agricultural and livestock products 34 . The HIV Serological surveillance system (sero survey) is nested within the Magu HDSS, with details described elsewhere 34,35 . The codebook for 2012 and 2017 sero-survey datasets, user guide for 2012 and 2017 sero-survey datasets and the information sheet and consent form for sero-survey can be found in Extended data 42 .
After consent, sero survey participants responded to structured face-to-face interview, collecting quantitative information on; demographics, fertility desire, child-bearing and family planning.

Data and variables
Primary outcome variable was the fertility desire defined as the desire to bear one or more child in the next two years. Both women and men answered the questions "Would you like to have an (other) child?", "How soon would you like your next child to be born?" and "How many more children would you like to have?" Fertility desire variable was binary and was defined as a proportion with fertility desire (desire to bear one or more children in the next two years) against all aged 15-49 year's age. For comparison purpose, we restrict the age of men to be between 15-49 years.
The exposure of interest was HIV infection, and was abstracted from the stored HIV test results. HIV testing was done in serological surveillance system and measured through a standardized Tanzanian protocol for HIV testing 36 . Demographic details of the participants were collected through the standardized serological surveillance questionnaire. The questionnaire collected information on age, marital status, education level, residence, occupation, religion, ethnicity, and lifestyles which included alcohol drinking habit and Cigarette smoking. We also had variables to represent the past obstetric history of the women including the number of previous pregnancies. Data entry and management were done using the Census and Survey Processing System software (CSPro) version 6.3.

Statistical analysis
The descriptive analysis reported the prevalence of fertility desire in men and women with 95% confidence intervals (95% CI) in each study, both overall and by exposure variables. We computed the association between fertility desire and HIV infection by calculating crude and adjusted estimates of prevalence ratio (PR) with 95% CI by using the log-binomial regression. All analyses were done separately for 2012 and 2017 studies. Analysis was done using STATA, version 16.1 (StataCorp, College Station, TX) statistical package.

Results
There were 5221 and 5730 participants aged 15-49 years in 2012 and 2017 respectively. In the 2012 study, there were 3361 women (64.4%) of whom 434 (12.6%) were WLHIV and 1860 men (35.6%) of whom 156 (8.4%) were men living with HIV. In 2017 there were 3560 women (62.1%) and 2164 men (37.9%), of whom 257 (7.2%) women and 101 (4.7%) men were living with HIV. Details on the characteristics of men and women participants in each study are shown in Table 1.

Fertility desire in men and women
In the 2012 study, the overall percentage of the desire to have one or more child in the next two years in men and women was 43.0% (95% CI 40.7 -45.3) and 33.3% (95% CI 31.8 -35.0) respectively, while in 2017, the percentage desire for fertility rose to 55.7% (95% CI 53.6 -57.8) in men and 41.5% (95% CI 39.8 -43.1) in women. In both studies, fertility desires in men and women living with HIV were relatively lower compared to the fertility desires in HIV uninfected men and women. Fertility desire for women living with HIV was 29.6% (95% CI 25.3 -34.3) in the 2012 and 38.9% (95% CI 95% 32.7 -44.9) in 2017. Whilst fertility desire in men living with HIV was 48.1% (95% CI 40.0 -56.2) in the 2012 survey and 51.5% (95% CI 41.3 -61.5) in the 2017 survey (Table 2). The level of fertility desire, with or without HIV infection, showed that the desires in men were almost always higher than that of women in both studies ( Figure 1). Overall, the fertility desires tended to decrease with increasing age notably with a higher fertility desire in men than in women in both surveys ( Figure 2).

Association between HIV infection and women's fertility desire
Although women's fertility desire was higher in HIV uninfected women compared to WLHIV in 2012 and 2017, age-adjusted analysis, did not show significant statistical association in 2012 (PR=1.02, 95CI 0.860 -1.187, p = 0.847) and 2017    Table 5.

Discussion
In our study, one-third of women and forty percent of men desired to have one or more children in the next two years in 2012, while forty percent of women and nearly half of men desired to have one or more children in the next two year in 2017. In 2017, HIV infection did not have an impact on fertility desire in either men or women, and this did not differ by different periods of ART provision. Increased fertility desire was associated in women who earned money for the family, living in rural areas and lower educational attainment. Decreased fertility desire was associated with increasing age and men generally tended to have a higher fertility than women.
The data from 2012 relate to a time when ART was only being used by a few people. By 2017 increased access to ART through earlier ART initiation and PMTCT option B plus in pregnant women had been rolled out. However, the impact of HIV on fertility desire did not change over this time, indicating that the population effect during the period of widespread availability of ART only increased fertility desire in men and women, but may not the observed gap in fertility desire in positives and negatives men and women.
Our findings support others on showing the prevalence of women's fertility desire in SSA 7,8 . The systematic review conducted by Martins and his colleagues in 2019, showed the prevalence of fertility desire in SSA to vary greatly 4 , with some of the findings being consistent with ours 9,37 . However, some results are conflicting with ours by reporting a lower prevalence of women's fertility desire 38,39 and extremely higher prevalence than ours 40 . Due to great variation on fertility desire estimates, researchers have suggested two main sources of variations in measuring fertility desires in SSA: internal and external sources of variations. Internal sources of variation included characteristics and sizes of study samples, data collection methods, definition of fertility desires and/or its assessment methods. External sauces of variation included social demographic, economic and cultural characteristics of a locality or segment of the population whether it is richly or poorly resourced country 41 .
Compared to HIV-uninfected women, there was no evidence of higher fertility desire in WLHIV in the period of earlier ART initiation program implementation (2017). In the Ugandan study, Lindsay and colleges supported our results by reporting no statistical significant difference on fertility desire among ART/PMTCT service users and non-users (adj. PRR: 0.84, CI: 0.62-1.14) 22 . Although the findings may not be comparable with our findings, the two studies differ in two main aspects 1) Our study did not have individual level data on ART and 2) our study reported on HIV sero-positivity alone with no information on whether the participants knew their HIV status. Ugandan study compared individual level ART data and possibly, participants were aware of their HIV status.
In our study, the following factors were found to increase women's fertility desire: being in a position to earn money for the family, and history of previous pregnancy. However, we found a higher desire for fertility in women who attended secondary and tertiary education in the 2012 but disappeared later in the 2017. Among users of modern contraceptives, fertility desire was decreasing with increasing age. Several studies have reported a range of factors that explain changes in fertility desire, and some of their results correlated with ours 7-9 .
Our study's strengths lie in the adequate sample size to measure the change in fertility desire and reliable HIV sero-status results. Among study weaknesses -absence of a qualitative component to adequately and reasonably measure fertility desire: unavailability of individual-level ART data, fecundity for men and women and finally inability to look into the data longitudinally.

Conclusion
We have reported a percentage of fertility desire in WLHIV in the period before and after earlier ART initiation program implementation in rural district of North western Tanzania. There was no evidence to suggest the difference in fertility desire between WLHIV and HIV-uninfected, over two difference phases of ART availability. The knowledge on factors associated with changes in fertility desire will be used in developing patient-centred reproductive health care in Tanzania.
The healthcare services will include interventions against unplanned pregnancies for WLHIV and HIV-uninfected women and increased FP promotion campaigns in men and women.

Data availability
Underlying data The data that support findings of this study cannot be shared publicly, but will be available upon request and following approval by The Medical Research Coordinating Committee (MRCC) of the National Institute for Medical Research (NIMR) in Tanzania. MRCC demand that all data collected within Tanzania may not be transferred or shared without their permission and before the signing of a data transfer agreement as the only criterion to access the data, which in line with the Government data protection policy.
For Researchers who wish to meet the above criteria for access to the data they should use the contact details below to request the data: In the study setting and design section, information should include also qualitative information not only quantitative. For example, the response for the question "Would you like to have an (other) child?", the response yes or no is qualitative.
It is unclear why the age is restricted only for men? What about women? It should be interesting to indicate the age range for women and men as inclusion criteria, the range cannot necessarily be the same in both cases.
In table 1 and table 2, the information of dichotomic variables is redundant. Indicate for fertility desire only the number and percentage of "yes" and in the case of HIV status, only the prevalence of HIV-positive. The same for alcohol taking, cigarette smoking earning money and even been pregnant. These changes will make the tables smaller and easy to understand. Indicate in a first top row the total number of participants per group and sex.
In the conclusion section, FP abbreviation was not defined before.
English editing: One-third -one-third.

○
Desired for one or more child -desire one or more child in the future. General comment: The manuscript seems technically sound and addresses a timely title.

Specific comments:
Although it can be said that the work has academic merit, the type of epidemiological study design used is not clearly stated.
I think there are contradictory concepts in the methods section. For instance, as stated in the abstract the authors conducted secondary data analysis of the two community-based studies of 2012 and 2017 in the Magu Health and Demographic system site, in Tanzania. To the contrary, in the methods section of the body of the article, besides starting unusually with "Ethical Considerations" it was stated as "All participants who contributed data for analysis provided written informed consent for study participation and publication of the results. Parents or any care giver consented on behalf of study participants who were aged less than 18 years." It needs further clarity for better sense and feasibility on how the secondary data from HDSS/five years or older/can be conducted with obtained written consent from each participant or care giver.
If possible, it is good to get further confirmation from a biostatistician.
Unable to assess the availability of all the source data underlying the results.

Approved with Reservations:
The reviewer believes the article has an academic merit. Nonetheless, I think it needs some revisions for better sense and clarity specifically on clearly providing the epidemiological study design employed, and the controversy between using secondary data and obtaining written consent from each participant. Furthermore, the data seems obsolete or 5-10 years old/comparing 2012 with that of 2017 in 2022; may need strong justifications for using the research findings to the required interventions timely to solve the observed gap or problem.

Is the work clearly and accurately presented and does it cite the current literature? Partly
Is the study design appropriate and is the work technically sound? Partly

Are sufficient details of methods and analysis provided to allow replication by others? Partly
If applicable, is the statistical analysis and its interpretation appropriate? I cannot comment. A qualified statistician is required.