The Systems Analysis and Improvement Approach: specifying core components of an implementation strategy to optimize care cascades in public health.

BACKGROUND: Healthcare systems in low-resource settings need simple, low-cost interventions to improve services and address gaps in care. Though routine data provide opportunities to guide these efforts, frontline providers are rarely engaged in analyzing them for facility-level decision making. The Systems Analysis and Improvement Approach (SAIA) is an evidence-based, multi-component implementation strategy that engages providers in use of facility-level data to promote systems-level thinking and quality improvement (QI) efforts within multi-step care cascades. SAIA was originally developed to address HIV care in resource-limited settings but has since been adapted to a variety of clinical care systems including cervical cancer screening, mental health treatment, and hypertension management, among others; and across a variety of settings in sub-Saharan Africa and the USA. We aimed to extend the growing body of SAIA research by defining the core elements of SAIA using established specification approaches and thus improve reproducibility, guide future adaptations, and lay the groundwork to define its mechanisms of action. METHODS: Specification of the SAIA strategy was undertaken over 12 months by an expert panel of SAIA-researchers, implementing agents and stakeholders using a three-round, modified nominal group technique approach to match core SAIA components to the Expert Recommendations for Implementing Change (ERIC) list of distinct implementation strategies. Core implementation strategies were then specified according to Proctor's recommendations for specifying and reporting, followed by synthesis of data on related implementation outcomes linked to the SAIA strategy across projects. RESULTS: Based on this review and clarification of the operational definitions of the components of the SAIA, the four components of SAIA were mapped to 13 ERIC strategies. SAIA strategy meetings encompassed external facilitation, organization of provider implementation meetings, and provision of ongoing consultation. Cascade analysis mapped to three ERIC strategies: facilitating relay of clinical data to providers, use of audit and feedback of routine data with healthcare teams, and modeling and simulation of change. Process mapping matched to local needs assessment, local consensus discussions and assessment of readiness and identification of barriers and facilitators. Finally, continuous quality improvement encompassed tailoring strategies, developing a formal implementation blueprint, cyclical tests of change, and purposefully re-examining the implementation process. CONCLUSIONS: Specifying the components of SAIA provides improved conceptual clarity to enhance reproducibility for other researchers and practitioners interested in applying the SAIA across novel settings.

Validity and item response theory properties of the Alcohol Use Disorders Identification Test for primary care alcohol use screening in Mozambique (AUDIT-MZ).

BACKGROUND: No validated tools exist to screen for substance use or dependence in Mozambique. The aim of this study was to validate the Alcohol Use Disorder Identification Test (AUDIT) for use in primary care settings in Mozambique. METHODS: The study administered a final adapted Mozambican 10-item AUDIT (AUDIT-10-MZ) to 502 individuals from antenatal, postpartum, and general outpatient consultations in three Ministry of Health primary health care clinics in Sofala Province, Mozambique. The study evaluated the AUDIT-10-MZ against the MINI 5.0-MZ as a gold standard diagnostic tool. RESULTS: Using the MINI 5.0-MZ, 16 (3.2%) of the sample tested positive for alcohol dependence and 3 (0.6%) tested positive for harmful alcohol use. The full AUDIT-10-MZ had acceptable internal consistency (α = 0.74); however, the shorter AUDIT-C-MZ had a higher alpha value than the full AUDIT screener (α = 0.79). The AUDIT-10-MZ performed well for screening in primary care, achieving areas under the receiver operating characteristic curves (AUROCs) of 0.94 (95% CI: 0.91, 0.96) for alcohol dependence. The AUDIT-C-MZ also performed well with an AUROC of 0.88 (95% CI: 0.80, 0.96) for alcohol dependence. Using a cut-off of ≥6, the AUDIT-10-MZ achieved a sensitivity of 68.8% and specificity of 92.0% for screening for alcohol dependence; a cut-off of ≥3 for the AUDIT-C-MZ achieved a sensitivity of 56.3% and specificity of 90.7%. CONCLUSIONS: Both the AUDIT-10-MZ and AUDIT-C-MZ are valid instruments for screening for alcohol dependence in Mozambique. The AUDIT-C-MZ performed particularly well and providers could use it as a brief screener in primary care settings. Optimal cut-points will depend on weighing false positives and false negatives but could be employed at ≥ 6 or ≥ 7 for the AUDIT-10-MZ and at ≥ 2 or ≥ 3 for the AUDIT-C-MZ. Future implementation research is needed to examine how best to integrate screening for substance use or dependence in primary care settings in Mozambique and other similar LMICs.

The prevalence of hypertension and its distribution by sociodemographic factors in Central Mozambique: a cross sectional study.

BACKGROUND: Hypertension (HTN) is a major risk factor for cardiovascular diseases, and its prevalence has been rising in low- and middle-income countries. The current study describes HTN prevalence in central Mozambique, association between wealth and blood pressure (BP), and HTN monitoring and diagnosis practice among individuals with elevated BP. METHODS: The study used data from a cross-sectional, representative household survey conducted in Manica and Sofala provinces, Mozambique. There were 4101 respondents, aged ≥20 years. We measured average systolic and diastolic BP (SBP and DBP) from three measurements taken in the household setting. Elevated BP was defined as having either SBP ≥140 or DBP ≥90 mmHg. RESULTS: The mean age of the participants was 36.7 years old, 59.9% were women, and 72.5% were from rural areas. Adjusting for complex survey weights, 15.7% (95%CI: 14.0 to 17.4) of women and 16.1% (13.9 to 18.5) of men had elevated BP, and 7.5% (95% CI: 6.4 to 8.7) of the overall population had both SBP ≥140 and DBP ≥90 mmHg. Among participants with elevated BP, proportions of participants who had previous BP measurement and HTN diagnosis were both low (34.9% (95% CI: 30.0 to 40.1) and 12.2% (9.9 to 15.0) respectively). Prior BP measurement and HTN diagnosis were more commonly reported among hypertensive participants with secondary or higher education, from urban areas, and with highest relative wealth. In adjusted models, wealth was positively associated with higher SBP and DBP. CONCLUSIONS: The current study found evidence of positive association between wealth and BP. The prevalence of elevated BP was lower in Manica and Sofala provinces than the previously estimated national prevalence. Previous BP screening and HTN diagnosis were uncommon in our study population, especially among rural residents, individuals with lower education levels, and those with relatively less wealth. As the epidemiological transition advances in Mozambique, there is a need to develop and implement strategies to increase BP screening and deliver appropriate clinical services, as well as to encourage lifestyle changes among people at risk of developing hypertension in near future.

Validity and item response theory properties of the Patient Health Questionnaire-9 for primary care depression screening in Mozambique (PHQ-9-MZ).

BACKGROUND: Depression is one of the leading causes of disability in Mozambique; however, few patients with depression are identified in primary care. To our knowledge, there are no validated tools for depression screening in Mozambique. The aim of this study was to validate the Patient Health Questionnaire-9 (PHQ-9) for use in primary care settings in Mozambique. METHODS: The PHQ-9 was adapted using a structured multi-phase process led by a team of bilingual experts followed by a review by lay individuals and pilot-testing including cognitive interviews. The final Mozambican PHQ-9 (PHQ-9-MZ) was applied among 502 individuals randomly selected from antenatal, postpartum, and general outpatient consultations in three Ministry of Health primary healthcare clinics in Sofala Province, Mozambique. The PHQ-9-MZ was evaluated against the MINI 5.0-MZ as a gold standard diagnostic tool. RESULTS: The majority of participants were female (74%), with a mean age of 28. Using the MINI 5.0-MZ, 43 (9%) of the sample tested positive for major depressive disorder. Items of the PHQ-9-MZ showed good discrimination and factor loadings. One latent factor of depression explained 54% of the variance in scores. Questions 3 (sleep) and 5 (appetite) had the lowest item discrimination and factor loadings. The PHQ-9-MZ showed good internal consistency, with a Cronbach's alpha of 0.84, and an area under the receiver operating characteristic curve (AUROC) of 0.81 (95% CI: 0.73, 0.89). The PHQ-2-MZ had an AUROC of 0.78 (95% CI: 0.70, 0.85). Using a cut-point of ≥9, the PHQ-9-MZ had a sensitivity of 46.5% and a specificity of 93.5%. Using a cut-point of ≥2, the PHQ-2-MZ had a sensitivity of 74.4% and a specificity of 71.7%. Increasing the cut-point to ≥3, the PHQ-2-MZ has a sensitivity of 32.6% and a specificity of 94.6%. CONCLUSIONS: The PHQ-9-MZ and PHQ-2-MZ emerge as two valid alternatives for screening for depression in primary health care settings in Mozambique. Depending on program needs and weighing the value of minimizing false positives and false negatives, the PHQ-9-MZ can be employed with cut-points ranging from ≥8 to ≥11, and the PHQ-2-MZ with cut-points ranging from ≥2 to ≥3.

Cascade Analysis: An Adaptable Implementation Strategy Across HIV and Non-HIV Delivery Platforms.

BACKGROUND: Cascades have been used to characterize sequential steps within a complex health system and are used in diverse disease areas and across prevention, testing, and treatment. Routine data have great potential to inform prioritization within a system, but are often inaccessible to frontline health care workers (HCWs) who may have the greatest opportunity to innovate health system improvement. METHODS: The cascade analysis tool (CAT) is an Excel-based, simple simulation model with an optimization function. It identifies the step within a cascade that could most improve the system. The original CAT was developed for HIV treatment and the prevention of mother-to-child transmission of HIV. RESULTS: CAT has been adapted 7 times: to a mobile application for prevention of mother-to-child transmission; for hypertension screening and management and for mental health outpatient services in Mozambique; for pediatric and adolescent HIV testing and treatment, HIV testing in family planning, and cervical cancer screening and treatment in Kenya; and for naloxone distribution and opioid overdose reversal in the United States. The main domains of adaptation have been technical-estimating denominators and structuring steps to be binary sequential steps-as well as logistical-identifying acceptable approaches for data abstraction and aggregation, and not overburdening HCW. DISCUSSION: CAT allows for prompt feedback to HCWs, increases HCW autonomy, and allows managers to allocate resources and time in an equitable manner. CAT is an effective, feasible, and acceptable implementation strategy to prioritize areas most requiring improvement within complex health systems, although adaptations are being currently evaluated.

Linkage to HIV care and antiretroviral therapy by HIV testing service type in Central Mozambique: a retrospective cohort study.

BACKGROUND: Access to antiretroviral therapy (ART) has increased dramatically in resource-limited settings since its introduction a decade ago. However, ART coverage remains low in countries with the highest disease burden, which may be partially explained by poor testing to care linkages. HIV testing service may impact early attrition in the HIV treatment cascade. METHODS: A retrospective cohort study was conducted in 18 clinics in central Mozambique using routine patient data and monthly reports. Patients referred from voluntary counseling and testing (VCT) were compared with those referred from prevention of mother-to-child transmission (PMTCT) for 3 outcomes: (1) enrollment at an HIV clinic ≤30 days after testing HIV positive, (2) CD4 test ≤30 days after enrollment, and (3) ART initiation ≤90 days after first CD4 test. RESULTS: Patient retention in the HIV care system dropped at each step from HIV testing to ART initiation. Enrollment in HIV care was not significantly different between PMTCT and VCT [risk ratio (RR) = 0.84, 0.72 < RR < 1.02]. Women tested in PMTCT were less likely to have a CD4 test ≤30 days after enrollment when adjusting for age, education level, and marital status (adjusted RR = 0.84, 0.70 < RR < 1.00), and were less likely to initiate ART ≤90 days after their first CD4 test when adjusting for age, education, and marital status (adjusted RR = 0.56, 0.44 < RR < 0.71). CONCLUSIONS: Poor linkages between HIV testing and care hamper efforts to improve coverage for HIV care and treatment services. Increased loss to follow-up among women diagnosed in PMTCT relative to VCT is worrisome and merits further qualitative research and programmatic attention.

Willingness of US men who have sex with men (MSM) to participate in Couples HIV Voluntary Counseling and Testing (CVCT).

BACKGROUND: We evaluated willingness to participate in CVCT and associated factors among MSM in the United States. METHODS: 5,980 MSM in the US, recruited through MySpace.com, completed an online survey March-April, 2009. A multivariable logistic regression model was built using being "willing" or "unwilling" to participate in CVCT in the next 12 months as the outcome. RESULTS: Overall, 81.5% of respondents expressed willingness to participate in CVCT in the next year. Factors positively associated with willingness were: being of non-Hispanic Black (adjusted odds ratio [aOR]: 1.5, 95% confidence interval [CI]: 1.2-1.8), Hispanic (aOR: 1.3, CI: 1.1-1.6), or other (aOR: 1.4, CI: 1.1-1.8) race/ethnicity compared to non-Hispanic White; being aged 18-24 (aOR: 2.5, CI: 1.7-3.8), 25-29 (aOR: 2.3, CI: 1.5-3.6), 30-34 (aOR: 1.9, CI: 1.2-3.1), and 35-45 (aOR: 2.3, CI: 1.4-3.7) years, all compared to those over 45 years of age; and having had a main male sex partner in the last 12 months (aOR: 1.9, CI: 1.6-2.2). Factors negatively associated with willingness were: not knowing most recent male sex partner's HIV status (aOR: 0.81, CI: 0.69-0.95) compared to knowing that the partner was HIV-negative; having had 4-7 (aOR: 0.75, CI: 0.61-0.92) or >7 male sex partners in the last 12 months (aOR: 0.62, CI: 0.50-0.78) compared to 1 partner; and never testing for HIV (aOR: 0.38, CI: 0.31-0.46), having been tested over 12 months ago (aOR: 0.63, CI: 0.50-0.79), or not knowing when last HIV tested (aOR: 0.67, CI: 0.51-0.89), all compared to having tested 0-6 months previously. CONCLUSIONS: Young MSM, men of color, and those with main sex partners expressed a high level of willingness to participate in couples HIV counseling and testing with a male partner in the next year. Given this willingness, it is likely feasible to scale up and evaluate CVCT interventions for US MSM.