(Last updated on 2021-11-10)

1. Analysis sample

1.1. Attrition of eligible clients

*Among all clients at baseline
**Among those who received methods/prescription at baseline

Number of observations by survey
indicator BFP2 KEP2 NGKanoP2 NGLagosP2
Number of cllients 911 4115 695 509
Number of cllients eligible for phone follow up 798 3898 663 454
Number of cllients who completed phone follow up interview 619 3313 462 351

1.2. Background and service characteristics by attrition

Lost clients tend to be:
* less educated (significant in all surveys)
* more likely to be new users at baseline (BF & KE)
* younger (KE & Lagos)
* LARC users (BF)

Also, lost clients tend to have reported:
* higher QCC at baseline, but only in BF and KE. Also the difference is way too small to be meaningful

  • Dark blue: average QCC among all 10 items
  • light blue: average QCC among 5 items about information
  • medium blue: average QCC among 5 items about (dis)respect and interaction

2. Transition from baseline to 6-month phone follow up

2.1. Outcome at 6-month follow up

  • Woman-level transition - slightly more complex than episode-level transition, although the follow-up period is relatively short
  • Responses to the below questions used to categorize women, based on their intention (explicit and implicit) and contraceptive use/dynamics since the baseline:
  1. Currently pregnant
  2. Started using the baseline method
  3. Still using the same method Currently using any method
  4. Stopped because she got pregnant
  5. Stopped because she wanted to get pregnant
  6. Stopped because of perceived low risk
  • Five categories are:
    1. Unintended pregnancy. This includes: (1) those who have had method failure AND currently pregnant; and (2) those who discontinued because of perceived low risk (i.e., “Infrequent sex/husband/partner away” or “Difficult to get pregnant/menopausal”) but currently pregnant (unmet need in the past 6 months)
    2. Discontinued in need. This includes: (1) a very small number of women who had method failure, but not currently pregnant, and not using any methods; and mostly (2) those who discontinued but did not mention the following three reasons - “Wanted to get pregnant”, “Infrequent sex/husband/partner away” or “Difficult to get pregnant/menopausal”. (unmet need currently)
    3. Discontinued with no need
    4. Switched
    5. Continued. Continued users mean women who “are still using the [baseline method].” Thus, it is possible some of them might have switched to something else and then switched back.

A small number of women were dropped since their intention was unclear or responses were inconsistent.

2.2. Overall transition

2.2.1. Pooled
2.2.2. By survey

Burkina Faso

Kenya

Nigeria, Kano state

Nigeria, Lagos state

2.3. Detailed transition

Baseline methods are presented on the left, with methods at the 6-month follow-up on the right side. Contraceptive methods are listed according to the order of method effectiveness. Modern and traditional methods are presented in blue and green shades, respectively. Discontinuation at the follow up is presented in red and orange.

2.3.1. Pooled
2.3.2. By survey

Burkina Faso

Kenya

Nigeria, Kano state

Nigeria, Lagos state

3. Considerations in building research question(s)

3.1. Sample distribution by outcome

Five categories are qualitatively different, and multinomial regression would be a good approach to see if QCC predicts outcomes. But, sample size limits feasible approaches…

Number of clients by survey and outcome
Outcome BF KE NGKano NGLagos
Unintended pregnancy since the baseline (whether currently or not) 1 23 1 2
Discontinued without reporting “no need”* 31 109 35 19
Discontinued due to “no neeed” ** 19 96 25 10
Switched to any contraceptive method 68 408 37 42
Continued using the same method from baseline 501 2682 363 277

3.2. Pathway from QCC to postive outcomes?

What are reasons for women to discontinue?

43-57% of women who started using the baseline method reported that they had side effects with the method. What are they?

4. Research questions

4.1. Question 1: Does better QCC at baseline predict no unmet need (categories 3, 4, and 5)?

No. 

Also, see excel file.

4.1.1. But, perhaps only for those who have side effects?

Not really…

4.2. Question 2: Among those who experienced side effects with the baseline method, does better QCC at baseline predict seeking help for side effects?

No. See excel file.

4.2.1. Perhaps only for those who have specific side effects?

Yes - though feasible to assess only in Kenya. See excel file.

4.2.2. Then, how about no unmet need among those restricted sample, in Kneya?

No. See excel file.

5. Discussion

Why no relationship?

  1. Conceptual flaw?
  • Purpose of QCC as a metric. QCC is not necessarily developed to predict contraceptive use outcomes.
  • Experience (measured via QCC) and clinical outcomes may not correlate.
  1. Analysis issue?
  • Lost-to-follow-up women: Perhaps QCC is more important for less educated or younger women?
  • Small N? But, the box plots suggest that even with a large sample size (i.e., even if we reach statistical significance), the difference would not be meaningful.
  • Different classification of QCC? Various metrics were tried, but none changes the story.
  • Appropriate variables are missing? Reported experience at baseline and outcomes at follow-up are somehow endogenous??

Interpretation?
* Good experience of care (measured via QCC) is important, regardless of clinical/FP outcomes.
* In Kenya, better QCC might have caused more care seeking for specific side effects. But, “advising about side effects” isn’t just about seeking care elsewhere.