(Updated: 2020-11-10 21:30:06 EDT)

Quality of Care Data from PMA’s Client Exit Interviews

  • See CEI sample methods and characteristics here.

1. Descriptive analysis: levels and patterns of quality

  • All estimates are unweighted.
  • Estimates based on a small denominator (i.e., <25) are suppressed.

1.1. Technical quality

Technical elements
* Explained: (1) how the method works, (2) side effect, (3) what to do with side effects, & (4) when to follow up
* Discussed: (1) other methods, (2) STI dual protection, (3) preferred method, (4) switching in the future, and (5) pros and cons of the method.
* Counseled about adherence among pill/injectable clients (see 2.1.3)

(3 items for MII noted in blue)

1.1.1. Average number of elements explained/discussed (0-9)

1.1.2. Percent of clients reporting specific items

1.1.3. Counseling among pill/injectable users
  • Counselled on higher chances of pregnancy if she does not take the pill every day or if she is >1 month late for shot?

1.2. Communications quality - cross cutting

Cross cutting between technical and experiential processes
* Communications: (1) clear/very clear information, (2) allowed to ask questions, and (3) all questions answered in a way she understood

1.3. Experiential quality

Experiential quality questions
* Polite (very polite vs. very polite/polite)
* Satisfied (very satisfied vs. very satisfied/satisfied)
* Refer
* Return

“Very satisfied” pattern

1.4. Outcome: did she receive what she wanted?

Survey questions

  • What were you given at your family planning visit today?
  • During today’s visit, did you obtain the method of family planning you wanted?
  • Why didn’t you obtain the method you wanted?

But, then some women reported having received what she wanted, even though she did not receive any methods or prescription…

Reasons for not receiving what she wanted

  • Potential provider bias in two responses, but not specific enough: “Provider recommended a different method” and “Not eligible for method”
  • No pattern by marital status/age for the potential bias
  • Most cited reasons are stock-out

2.Individual-level analysis

Many potential questions for either research papers or DYK briefs (although not necessarily mutually exclusive). But, before then, two statistical questions:
* Does PMA plan to construct analytical weights?
* What would be analytical recommendations to deal with clustering, considering the relatively large number of clients?

2.1. Can we select/reduce MII items - to measure “informed choice”?

PCA analysis suggest that all nine items pretty much equally contributes to measure one main component (presumably technical quality). Similar across countries, when PCA conducted by country.

PCA of nine items, using pooled CEI data

2.2. Factors associated with technical quality score

Motivation: to identify potential target facilities/clients to improve the technical quality

  • Communications?
  • methods - specifically LARCPM?
  • Facility structure/capacity (e.g., readiness)? This should not be related, unless it’s confounded with omitted variables (e.g., heath worker competencies, supportive supervision in the facility).
  • Background characteristics?

See regression output: output_QoC_miiscore_date.xml: multivariate regression analysis, using pooled data with country-dummy variables, adjusted for age and education, and adjusted for clustering on facility-ID (xtreg …, i(facility_ID))

NOTE: technical quality from the informed choice perspective. May or may not be associated with “outcome” (e.g., continuation), which should be examined with panel data.

2.2. Factors associated with “very satisfied”

Motivation: is satisfaction a truly useless question? When we examine only “very satisfied”, does it vary by plausible factors? Well, also, is this a right question?

  • Receiving what she wanted (universal) - excluded
  • Technical quality score?
  • Communications?
  • methods - specifically LARCPM?
  • Facility structure/capacity (e.g., readiness)?
  • Background characteristics?

See regression output: output_QoC_verysatisfied_date.xml: multivariate regression analysis, using pooled data with country-dummy variables, adjusted for age and education, and adjusted for clustering on facility-ID (xtlogit …, i(facility_ID))