Tree-based subgroup discovery using electronic health record data: heterogeneity of treatment effects for DTG-containing therapies

The rich longitudinal individual level data available from electronic health records (EHRs) can be used to examine treatment effect heterogeneity. However, estimating treatment effects using EHR data poses several challenges, including time-varying confounding, repeated and temporally non-aligned measurements of covariates, treatment assignments and outcomes, and loss-to-follow-up due to dropout. Here, we develop the subgroup discovery for longitudinal data algorithm, a tree-based algorithm for discovering subgroups with heterogeneous treatment effects using longitudinal data by combining the generalized interaction tree algorithm, a general data-driven method for subgroup discovery, with longitudinal targeted maximum likelihood estimation. We apply the algorithm to EHR data to discover subgroups of people living with human immunodeficiency virus who are at higher risk of weight gain when receiving dolutegravir (DTG)-containing antiretroviral therapies (ARTs) versus when receiving non-DTG-containing ARTs.

A pseudo-likelihood method for estimating misclassification probabilities in competing-risks settings when true-event data are partially observed

Outcome misclassification occurs frequently in binary-outcome studies and can result in biased estimation of quantities such as the incidence, prevalence, cause-specific hazards, cumulative incidence functions, and so forth. A number of remedies have been proposed to address the potential misclassification of the outcomes in such data. The majority of these remedies lie in the estimation of misclassification probabilities, which are in turn used to adjust analyses for outcome misclassification. A number of authors advocate using a gold-standard procedure on a sample internal to the study to learn about the extent of the misclassification. With this type of internal validation, the problem of quantifying the misclassification also becomes a missing data problem as, by design, the true outcomes are only ascertained on a subset of the entire study sample. Although, the process of estimating misclassification probabilities appears simple conceptually, the estimation methods proposed so far have several methodological and practical shortcomings. Most methods rely on missing outcome data to be missing completely at random (MCAR), a rather stringent assumption which is unlikely to hold in practice. Some of the existing methods also tend to be computationally-intensive. To address these issues, we propose a computationally-efficient, easy-to-implement, pseudo-likelihood estimator of the misclassification probabilities under a missing at random (MAR) assumption, in studies with an available internal-validation sample. We present the estimator through the lens of studies with competing-risks outcomes, though the estimator extends beyond this setting. We describe the consistency and asymptotic distributional properties of the resulting estimator, and derive a closed-form estimator of its variance. The finite-sample performance of this estimator is evaluated via simulations. Using data from a real-world study with competing-risks outcomes, we illustrate how the proposed method can be used to estimate misclassification probabilities. We also show how the estimated misclassification probabilities can be used in an external study to adjust for possible misclassification bias when modeling cumulative incidence functions.

Modeling the causal effect of treatment initiation time on survival: Application to HIV/TB co-infection

The timing of antiretroviral therapy (ART) initiation for HIV and tuberculosis (TB) co-infected patients needs to be considered carefully. CD4 cell count can be used to guide decision making about when to initiate ART. Evidence from recent randomized trials and observational studies generally supports early initiation but does not provide information about effects of initiation time on a continuous scale. In this article, we develop and apply a highly flexible structural proportional hazards model for characterizing the effect of treatment initiation time on a survival distribution. The model can be fitted using a weighted partial likelihood score function. Construction of both the score function and the weights must accommodate censoring of the treatment initiation time, the outcome, or both. The methods are applied to data on 4903 individuals with HIV/TB co-infection, derived from electronic health records in a large HIV care program in Kenya. We use a model formulation that flexibly captures the joint effects of ART initiation time and ART duration using natural cubic splines. The model is used to generate survival curves corresponding to specific treatment initiation times; and to identify optimal times for ART initiation for subgroups defined by CD4 count at time of TB diagnosis. Our findings potentially provide 'higher resolution' information about the relationship between ART timing and mortality, and about the differential effect of ART timing within CD4 subgroups.

Factors Associated with Late Engagement to HIV Care in Western Kenya: A Cross-Sectional Study

BACKGROUND

Late presentation of patients contributes significantly to the high mortality reported in HIV -care and treatment programs in sub-Saharan Africa.

METHODS

A cross-sectional study was conducted to assess factors associated with late engagement to HIV care at the Academic Model Providing Access to Healthcare in western Kenya. Late engagement was defined as baseline CD4 ≤100 cells/mm³.

RESULTS

Of the 10 533 participants included in the analysis, 67% were female and mean age was 36.7 years. Overall, 23% of the participants presented late. Factors associated with late engagement included male gender (adjusted odds ratio [AOR]: 1.54, 95% confidence interval [CI]: 1.35-1.75), older age (AOR: 1.62, 95% CI: 1.02-2.56), and longer travel time to clinic (AOR: 1.18, 95% CI: 1.04-1.34).

CONCLUSION

Nearly one-quarter of HIV-infected patients in our setting present with advanced immune suppression at initial encounter. Being male, older age, and living further away from clinic are associated with late engagement to care.

Active tuberculosis is associated with worse clinical outcomes in HIV-infected African patients on antiretroviral therapy

Objective

This cohort study utilized data from a large HIV treatment program in western Kenya to describe the impact of active tuberculosis (TB) on clinical outcomes among African patients on antiretroviral therapy (ART).

Design

We included all patients initiating ART between March 2004 and November 2007. Clinical (signs and symptoms), radiological (chest radiographs) and laboratory (mycobacterial smears, culture and tissue histology) criteria were used to record the diagnosis of TB disease in the program’s electronic medical record system.

Methods

We assessed the impact of TB disease on mortality, loss to follow-up (LTFU) and incident AIDS-defining events (ADEs) through Cox models and CD4 cell and weight response to ART by non-linear mixed models.

Results

We studied 21,242 patients initiating ART–5,186 (24%) with TB; 62% female; median age 37 years. There were proportionately more men in the active TB (46%) than in the non-TB (35%) group. Adjusting for baseline HIV-disease severity, TB patients were more likely to die (hazard ratio – HR = 1.32, 95% CI 1.18–1.47) or have incident ADEs (HR = 1.31, 95% CI: 1.19–1.45). They had lower median CD4 cell counts (77 versus 109), weight (52.5 versus 55.0 kg) and higher ADE risk at baseline (CD4-adjusted odds ratio = 1.55, 95% CI: 1.31–1.85). ART adherence was similarly good in both groups. Adjusting for gender and baseline CD4 cell count, TB patients experienced virtually identical rise in CD4 counts after ART initiation as those without. However, the overall CD4 count at one year was lower among patients with TB (251 versus 269 cells/µl).

Conclusions

Clinically detected TB disease is associated with greater mortality and morbidity despite salutary response to ART. Data suggest that identifying HIV patients co-infected with TB earlier in the HIV-disease trajectory may not fully address TB-related morbidity and mortality.

Risk factors for death in HIV-infected adult African patients receiving anti-retroviral therapy

OBJECTIVE

To determine risk factors for death in HIV-infected African patients on anti-retroviral therapy (ART).

DESIGN

Retrospective Case-control study.

SETTING

The MOH-USAID-AMPATH Partnership ambulatory HIV-care clinics in western Kenya.

RESULTS

Between November 2001 and December 2005 demographic, clinical and laboratory data from 527 deceased and 1054 living patients receiving ART were compared to determine independent risk factors for death. Median age at ART initiation was 38 versus 36 years for the deceased and living patients respectively (p<0.0148). Median time from enrollment at AMPATH to initiation of ART was two weeks for both groups while median time on ART was eight weeks for the deceased and fourty two weeks for the living (p<0.0001). Patients with CD4 cell counts <100/mm3 were more likely to die than those with counts >100/mm3 (HR=1.553. 95% CI (1.156, 2.087), p<0.003). Patients attending rural clinics had threefold higher risk of dying compared to patients attending clinic at a tertiary referral hospital (p<0.0001). Two years after initiating treatment fifty percent of non-adherent patients were alive compared to 75% of adherent patients. Male gender, WHO Stage and haemoglobin level <10 grams% were associated with time to death while age, marital status, educational level, employment status and weight were not.

CONCLUSION

Profoundly immunosuppressed patients were more likely to die early in the course of treatment. Also, patients receiving care in rural clinics were at greater risk of dying than those receiving care in the tertiary referral hospital.

Causes of delay in diagnosis of pulmonary tuberculosis in patients attending a referral hospital in Western Kenya

OBJECTIVE

To determine the length of delays from onset of symptoms to initiation of treatment of pulmonary tuberculosis (PTB).

DESIGN

Cross-sectional study.

SETTING

Chest/TB clinic, Moi Teaching and Referral Hospital (MTRH), Eldoret, Kenya.

SUBJECTS

Newly diagnosed smear positive pulmonary tuberculosis (PTB) patients.

RESULTS

Two hundred and thirty patients aged between 12 and 80 (median; 28.5) years were included in the study. They comprised 148 (64.3%, median 30 years) males and 82 (35.7%, median 28 years) females. One hundred and two (44%) came from urban and 128 (56%) came from rural setting covering a median distance of 10 (range 0-100) kilometres and paying Kshs 20 (range 0-200) to facility. Cough was the commonest symptom reported by 228 (99.1%) of the patients followed by chest pain in 214 (80%). The mean patient delay was 11 +/- 17 weeks (range: 1-78 weeks) with no significant difference between males and females, the mean system delay was 3 +/- 5 weeks (range: 0-39 weeks). The median patient, health systems and total delays were 42, 2, and 44 days respectively for all the patients. Marital status, being knowledgeable about TB, distance to clinic and where help is sought first had significant effect on patient delay.

CONCLUSION

Patient delay is the major contributor to delay in diagnosis and initiation of treatment of PTB among our patients. Therefore TB control programmes in this region must emphasise patient education regarding symptoms of tuberculosis and timely health seeking behaviour.

Nasopharyngeal Streptococcus pneumoniae among under-five year old children at the Moi Teaching and Referral Hospital, Eldoret, Kenya

OBJECTIVES

To determine the prevalence, risk factors and antibiotic sensitivity of streptococcus pneumoniae carried in the upper respiratory tract of children.

DESIGN

A cross-sectional study on consecutive clients.

SETTING

Maternal Child Health Clinic (MCH) at Moi Teaching and Referral Hospital (MTRH) in western Kenya.

SUBJECTS

Seventy eight of children attending Maternal Child Health Clinic between March 10th 2003 and July 11th 2003.

MAIN OUTCOME MEASURES

Upper airway carriage status, ventilation, housing, age, illness, sensitivity patterns.

RESULTS

Fifty six percent were boys; the median age was six months (range 1-42 months). Streptococcus pneumoniae carriage rate was in 28 (35.9%) cases. Fifty two percent of S. pneumoniae were resistant to penicillin, 25% to ampicillin and 78% to cotrimoxazole. There was significant association between the type of floor with pneumococcal carriage (p = 0.009) with people living in earth floor houses being five times more likely to be pneumococcal carriers as compared to those living in cement floor houses.

CONCLUSIONS

A significant resistance of S. pneumoniae to penicillin, ampicillin and cotrimoxazole was found. Earth floored houses may increase susceptibility to upper airway S. pneumoniae carriage.

RECOMMENDATION

Similar studies should be conducted in other parts of Kenya in order to learn about susceptibility patterns and associated risk factors, including floor type, in the country and tailor better treatment regimens.