Sampling-Based Approaches to Improve Estimation of Mortality among Patient Dropouts: Experience from a Large PEPFAR-Funded Program in Western Kenya
2008

Estimating Mortality Among HIV Patients Who Drop Out of Care

Sample size: 8977 publication 10 minutes Evidence: high

Author Information

Author(s): Yiannoutsos Constantin T., An Ming-Wen, Frangakis Constantine E., Musick Beverly S., Braitstein Paula, Wools-Kaloustian Kara, Ochieng Daniel, Martin Jeffrey N., Bacon Melanie C., Ochieng Vincent, Kimaiyo Sylvester

Primary Institution: Indiana University Division of Biostatistics

Hypothesis

Can sampling-based approaches improve the estimation of mortality among patients lost to follow-up in HIV treatment programs?

Conclusion

The study found that mortality estimates based on passive monitoring may underestimate true mortality by up to 80%, and that statistical adjustments can provide more accurate evaluations.

Supporting Evidence

  • The study assessed the impact of loss to follow-up on estimating patient mortality among 8,977 adult clients.
  • Dropouts were more likely to be male and younger than non-dropouts.
  • Statistical techniques adjusted mortality estimates based on information from located patients.

Takeaway

When patients with HIV stop going to their doctor, it's hard to know how many are dying. This study shows that using a sample of these patients can help us get a better idea of the real number of deaths.

Methodology

The study used statistical sampling techniques and medical record infrastructure to trace patients lost to follow-up and estimate mortality rates.

Potential Biases

There is a risk that patients who are harder to locate may have higher mortality rates, leading to underestimation of true mortality.

Limitations

The study assumes that the sample of patients located is representative of all dropouts, which may not be true.

Participant Demographics

The study included 8,977 adult HIV-positive patients, with approximately 35% male and a median age of 35.5 years.

Statistical Information

P-Value

p<0.001

Confidence Interval

95% CI 1.3%–2.0%

Statistical Significance

p<0.001

Digital Object Identifier (DOI)

10.1371/journal.pone.0003843

Want to read the original?

Access the complete publication on the publisher's website

View Original Publication