Burden and Trends of Symptomatic Sexually Transmitted Infections in Malawi From 2000 to 2021: Comparative Analysis of Survey and Case Report Data

Two data sources for surveillance of sexually transmitted infection symptoms in Malawi had similar spatial and demographic patterns but differed in the magnitude, trends, and relative burden of symptoms. Background In settings without etiologic testing for sexually transmitted infections (STIs), programs rely on STI symptom data to inform priorities. To evaluate whether self-reported STI symptoms in household surveys consistently represent the STI burden, we compared symptomatic infection rates between survey self-reporting and health facility case reporting in Malawi. Methods We analyzed self-reported symptoms and treatment seeking in the past year among sexually active adults from 4 Malawi Demographic and Health Surveys between 2000 and 2015. Bayesian mixed-effects models were used to estimate temporal trends, spatial variation, and sociodemographic determinants. Survey reporting was compared with health facility syndromic diagnoses between 2014 and 2021. Results In surveys, 11.0% (95% confidence interval, 10.7%–11.4%) of adults reported STI or STI-related symptoms in the last year, of whom 54.2% (52.8%–55.7%) sought treatment. In facilities, the mean annual symptomatic case diagnosis rate was 3.3%. Survey-reported treatment in the last year was 3.8% (95% credible interval, 2.3%–6.1%) for genital ulcer, 3.8% (2.0%–6.7%) for vaginal discharge, and 2.6% (1.2%–4.7%) for urethral discharge. Mean annual diagnosis rates at facilities were 0.5% for genital ulcer, 2.2% for vaginal discharge, and 2.0% for urethral discharge. Both data sources indicated a higher burden of symptoms among women, individuals older than 25 years, and those in Southern Malawi. Conclusions Survey and facility case reports indicated similar spatial and demographic patterns of STI symptom burden and care seeking, but implied large differences in the magnitude and relative burden of symptoms, particularly genital ulcer, which could affect program priorities. Targeted etiologic surveillance would improve interpretation of these data to enable more comprehensive STI surveillance.

remains the norm in most settings because of limited routine laboratory testing capacity. 1 Lack of case definition standardization and incomplete and aggregated facility reporting (without age or gender stratification) have hindered the use of syndromic case data in STI burden assessment. 2 Nationally representative household surveys, such as quinquennial Demographic and Health Surveys (DHS), 3 often measure self-reported STI symptom prevalence and treatment seeking, but without etiologic STI testing.However, like facility case reporting, STI symptoms are often nonspecific and may reflect several infectious or noninfectious causes, such as bacterial vaginosis, and also omit high proportions of asymptomatic infection. 4his study explored using self-reported data on symptomatic infections to monitor STI burden and trends.We assessed the consistency of STI determinants, trends, and spatial patterns from self-reported symptoms in household surveys and facilityreported syndromic cases in health management information systems in Malawi and compared the distribution of symptoms across both datasets.

Data Sources and Study Measures
We compared 2 data sources on symptomatic STIs in Malawi.The first was the Malawi DHS, conducted in 2000, 2004, 2010, and 2015-2016. 3Demographic and Health Surveys were nationally representative cross-sectional household-based surveys.Women aged 15 to 49 years and men aged 15 to 54 years were eligible to participate who were permanent residents or visitors in selected households.All men in one-third of the sampled households were selected for the male survey.Variable definitions were informed by the Guide to DHS Statistics (Table S1, http:// links.lww.com/OLQ/B26). 5utcomes analyzed were self-reported genital ulcer, genital discharge, and known STI (defined in the DHS as infection acquired through sexual contact, other than HIV) in the last 12 months, and, among those, self-reported care or treatment sought for an STI or STI-related symptom in the last 12 months.Among those reporting care or treatment, the sector (public, private, or other) was also assessed for 2010 and 2015-2016.Covariates, selected based on factors associated with STI infection in previous literature, included sociodemographic and behavioral characteristics, and STI-related knowledge. 6,7Sociodemographic variables were age (15-19, 20-24, 25-29, 30-24, 35-39, 40+ years), sex (male, female), place of residence (urban, rural), marital status (never married, formerly married, married or living together), education (none, primary, secondary or higher), and current employment.Behavioral variables were age at first sex (<15, 15-19, 20-24, 25+ years), number of sex partners in the last 12 months (none, 1, 2 or more), condom use at last sex in the last 12 months (yes, no or unknown), and ever tested for HIV (yes, no or unknown).Knowledge of STIs was measured as knowing diseases can be transmitted through sexual contact (yes, no or unknown).
The second data source was health facility-level aggregate STI syndrome case counts recorded in the Malawi Department of HIV and AIDS Management Information System (DHAMIS), a database recording public sector supported HIV and related services in Malawi, provided through clinics, health posts, dispensaries, health centers, and hospitals, among others.Each quarter, monthly STI case reports are manually aggregated from paper STI clinic registers and verified. 8We analyzed data between January 2014 and December 2021.During this period, reporting facilities increased gradually from 708 in 2014 to 794 in 2021.
Aggregate STI case reports were stratified by sex (male, female), age (0-19, 20-24, 25+ years), and HIV status (HIV-positive, HIV-negative, unknown).Diagnoses of STI, which were not age-stratified, included genital ulcer disease (GUD), urethral discharge syndrome (UDS), low-risk vaginal discharge syndrome (VDS), high-risk VDS, and lower abdominal pain, among others.Because diagnoses could include multiple syndromes, the number of diagnoses was typically larger than the number of clients (Fig. S1, http://links.lww.com/OLQ/B26).Outcome variables for this analysis were cumulative annual diagnoses of GUD, UDS, and VDS (low-and high-risk combined) to align with DHS outcomes.
Adjusted district-level population estimates by age group, sex, and year for 2014-2021 were used as denominators for STI diagnosis rates.The Malawi National Statistics Office projected population totals by district, sex, and 5-year age group from the 2008 household census, accounting for anticipated population growth.These annual projections were adjusted proportionally in each district, sex, and age group to align with national census results in 2018. 9,10alysis Demographic and Health Surveys respondent characteristics and self-reported STI and treatment seeking were summarized as frequencies and survey-weighted proportions with designbased standard errors and 95% confidence intervals (95% CIs).Treatment seeking by health sector was assessed using only the 2010 and 2015-2016 DHS, due to inconsistent questionnaire formulation in earlier surveys.
To identify factors associated with self-reported STI symptoms, we used 4 Bayesian logistic mixed-effects models to estimate the odds of self-reported genital ulcer and genital discharge in the last 12 months among ever sexually active male and female individuals.To identify factors associated with STI treatment seeking, 2 Bayesian logistic mixed-effects models were used to estimate the odds of treatment seeking among male and female individuals reporting an STI or STI symptom in the last 12 months.All models included random intercepts and slopes over calendar year for the 33 geographic areas (29 health districts and 4 metropolitan cities).Survey cluster-level random effects were included to account for the clustered sampling design. 5Fixed effects included covariates for sociodemographic and behavioral characteristics, and STI-related knowledge.The models did not use sampling weights.
Health facility syndromic case reports in DHAMIS were checked for consistency and completeness.Severely high outliers, identified as monthly cases 5 times larger than the third quartile for the relevant health facility, year, and diagnosis, were assumed to be reporting errors and replaced with the median number of diagnoses per month for the corresponding health facility and year.The annual rate of syndromic diagnoses was calculated per district and year by dividing total number of diagnoses of each syndrome across all ages by the adult population aged 15 to 54 years.
To assess similarities in the distribution and characteristics of syndromic diagnoses in health facility case reports and STI care seeking in the past year in household surveys, we estimated the probability of treatment of genital ulcer and genital discharge in the last 12 months among all adult DHS respondents.Bayesian logistic mixed-effects models included random intercepts and slopes over calendar year for each geographic area and fixed effects for year and age interacted with sex.Models were used to predict the proportion of all adults aged 15 to 54 years seeking treatment of each symptom in the last year at district, region, and national Results were presented as adjusted odds ratios (aORs) with 95% credible intervals (95% CrI).All regression models used diffuse priors.Analyses were performed in R version 4.2.1, using the rstanarm package. 11thical approval for this study was granted by the Imperial College Research Ethics Committee (ICREC No. 6365329).
Factors associated with reporting an STI-related symptom varied by sex (Table 1).Among women, reporting ulcer and discharge were higher among those formerly married than never married (aOR, 1.45 [1.23 ) with 2 or more sex partners in the last year were much more likely to report ulcer and discharge than those with no partners.As expected, the odds of reporting either symptom were higher among men and women who reported having an STI in the last year.
Nationally, the annual case diagnosis rate among adults aged 15 to 54 years between 2014 and 2021 was 3.7% for all cases and 3.3% for symptomatic cases.The rate of VDS diagnoses among adult women increased from 1.7% in 2014 to 2.4% in 2021.The rate of UDS diagnoses increased from 1.4% to 2.5%.Annual diagnosis rates of GUD consistently averaged 0.5% over time (Fig. 2A).Southern and Northern Malawi had the highest case rates across all 3 diagnoses (Fig. 2B).The mean case diagnosis rates were particularly high for VDS and UDS in Mwanza (6.7% and 3.6%) and Neno (5.4% and 5.2%) in the South and for UDS in Likoma (3.5%), a small island district in Lake Malawi (Fig. 2C).

Comparison of Facility-Reported Syndromic Case Diagnoses and Survey-Reported Treatment Seeking
In household surveys, most adults reporting treatment of genital ulcer and genital discharge in the last year were female In household surveys, nationally, 3.8% (95% CrI, 2.3%-6.1%) of adults reported care seeking for genital ulcer in the last year, 3.8% (2.0%-6.7%)for vaginal discharge, and 2.6% (1.2%-4.7%)for urethral discharge.Although lower, the annual rate of health facility syndromic diagnoses among adults was within the credible intervals of the survey predictions for urethral discharge and vaginal discharge (Fig. 3A).However, the average facility diagnosis rate for genital ulcer was much lower (0.5%) than treatment-seeking proportions reported in the surveys.

DISCUSSION
Nationally representative household surveys and health facility case reporting in Malawi indicated divergent rates, Figure 2. Annual rate of syndromic case diagnoses among adults aged 15 to 54 years during 2014-2021 at (A) national, (B) regional, and (C) district levels according to geographic proximity.Diagnosis rates were calculated using total reported syndromic diagnoses and sex-matched census population estimates per year per district.y Axes truncated at 6%; maximum rate of 9.7% (2014) for VDS in Mwanza and 7.2% (2019) and 7.3% (2021) for VDS and UDS in Neno.
distributions, and temporal trends in STI symptoms.In household surveys between 2000 and 2015-2016, around 11% reported having an STI or STI-related symptom in the last year, of whom around half sought treatment of their last infection, implying that 6.0% of adults per year present for STI care.In health facility case reports between 2014 and 2021, the mean annual symptomatic case diagnosis rate was only 3.3%.
Symptom distribution also differed between data sources.In surveys, genital ulcer and vaginal discharge were most common among adults seeking treatment.In national case reports, facility diagnosis rates were highest for vaginal discharge and urethral discharge syndromes and substantially lower for genital ulcer.This diagnostic distribution aligns with previous research among STI clients attending an urban and rural hospital in Malawi. 12,13Furthermore, surveys indicated a higher proportion of symptomatic women treated for genital ulcer than men.The absence of sex-stratified case report diagnoses precluded direct comparison, but prior research identified the opposite sex trend among STI clients. 12,13][16] Sexually transmitted infection care seeking outside predominantly public sector facilities reporting to the national health information system may partly account for differences in the magnitude and distribution of symptoms across data sources.In 2021, DHAMIS case reporting included 794 health facilities.The 2018 master facility list documented 963 public, private, and traditional STI service providers in Malawi, but this likely underestimated the count because of difficulties capturing unregistered or informal sources of care. 17Most survey participants reported seeking care in the public health sector, but more than 30% sought treatment from the private and traditional sectors.Men reported using the private sector more than women.9][20] Individuals with genital ulcer reported using other means of treatment, such as traditional healers, more often than those with other symptoms.Although the traditional health sector is widely used in Malawi, 21 differences in traditional care seeking by STI symptom have not been previously documented.Moreover, frequent antibiotic stockouts in the public sector may result in periods of increased reliance on private and traditional providers. 22We were, however, unable to evaluate these patterns.
Temporal trends differed between the data sources, although our assessment was limited by not having survey data after 2015 for direct comparison with health facility data.Although facility case diagnosis rates increased between 2015 and 2020, household surveys showed no significant change in STI symptom reporting and a decline in STI care seeking between 2000 and 2015.The trends could have reversed, particularly trends in care seeking coinciding with health system improvements supported through HIV service decentralization.One possible explanation for the rise in syndromic diagnoses is the large scale-up of HIV testing in Malawi since 2015, 23 which may have facilitated STI detection and management, underscoring the importance of integrated health service provision. 1The upcoming Malawi DHS will help assess if increasing facility diagnoses are reflected in self-reporting, and if so, whether they are attributable to higher rates of symptomatic infection or treatment seeking.
Despite discrepancies in overall levels, both data sources had similar demographic and spatial patterns, consistent with the burden of STIs.They indicated a higher burden of symptoms among women, individuals older than 25 years, and in Southern Malawi.Specifically, self-reported genital ulcer in surveys was highest among women 35 to 39 years old and in the Southern region, aligning with the prevalence of HSV-2 and HIV.HSV-2, the main cause of genital ulcer in SSA, 14 is more prevalent among women within this older age group because of the lifelong nature of infection. 24HIV prevalence in Malawi is also higher among women aged 35 to 54 years and in the South. 25These complementary trends suggest that household surveys can provide insight into the relative prevalence of symptomatic STIs.However, future analyses aiming to reconcile the burden and distribution of STI symptoms with STIs should incorporate factors influencing the etiology of these symptoms.For instance, a recent study among STI clinic attendees with genital ulcer in Malawi identified a substantial proportion of cases attributed to syphilis, indicative of shifts in genital ulcer etiology over time, and a considerably higher proportion of HSV-attributed cases among those HIV positive than HIV negative. 26istrict-level differences between survey self-reporting and facility case-reporting require further investigation to inform program planning and prioritization.Districts with high self-reported treatment but low facility diagnoses might rely more on the private and traditional sectors for STI service provision.Programs should examine this further and consider expanding public sector services in these areas to address potential unmet treatment needs.Furthermore, district-level variation was greater for self-reported treatment than facility case diagnoses.Surveys may thus better identify districts with a higher infectious burden for prioritized programming.
Limitations in both data sources may contribute to the identified inconsistencies.Self-reporting in household surveys may underrepresent symptomatic infection, because of recall and social desirability biases, 27 and potential lack of awareness of STI symptoms. 28,29Overreporting may have resulted from using nonspecific questions on STI symptoms.Our analysis of routine case reporting may have overestimated syndromic diagnosis rates among adults, because of limited stratification available in DHAMIS.We included diagnoses across all age groups, yet 8.8% of total diagnoses were among those younger than 19 years.In addition, clients may have been double counted if they sought care from multiple health facilities or experienced recurrent symptoms in a year.
In conclusion, in Malawi, where etiologic STI surveillance is not widely implemented, we identified nonnegligible discrepan- Burden and Trends of Symptomatic STIs in MalawiSexually Transmitted Diseases • Volume 51, Number 3, March 2024 levels for the period 2014-2021, assuming a flat trend after the year 2015 (the year of the final survey).To compare the extent of district-level variability between data sources, we calculated coefficients of variation for each symptom based on average district rates during 2014-2021.

Figure 1 .
Figure 1.Percentage of ever sexually active adult DHS respondents self-reporting (A) STI or STI-related symptoms in the last year during 2000, 2004, 2010, and 2015-2016; (B) treatment seeking for an STI or STI-related symptom in the last year during 2000, 2004, 2010, and 2015-2016 among adults who reported STI or STI-related symptoms; and (C) health sectors accessed for STI treatment in the last year in 2010 and 2015-2016.Vertical line segments in panels A and B represent 95% CIs.

Figure 3 .
Figure 3.Comparison of the prevalence of individuals with self-reported symptoms who sought treatment in the last 12 months and annual facility-reported syndromic diagnosis rates, among adults aged 15 to 54 years in Malawi during 2014-2021.(A), National-level comparison of annual rates per symptom.Dotted lines represent aggregate health facility case reports.Solid lines represent self-reported symptoms from household surveys between 2000 and 2015-2016, predicted to 2014-2021.Shading represents 95% CrIs around self-reported symptoms.(B) District-and regional-level comparison of mean rates per symptom.Thick solid bars represent aggregate health facility case reports.Points represent self-reported symptoms from household surveys between 2000 and 2015-2016, predicted to 2014-2021.Error bars represent 95% CrIs around self-reported symptoms.