Quantitative analysis

To understand the implementation outcomes and their moderators as reported by participants in the survey we had first to define and measure the outcome and explanatory variables before proceeding to statistical analyses. The outcome variables included in our study were: adoption and FOI.

In line with our previous work [19], we measured adoption as the presence or absence of and functioning of the redesigned CHF structures, availability status and use of guidelines and related processes, and job description. The structures considered include CHF governing board, CHF management teams and the conduct of the scheduled or ad hoc meetings of these structures.

We measured FOI as adherence to the program model as originally stipulated in the Redesigned CHF standard operating procedures (CHF-SOP) and related operational document, such as planning manuals, supervision guidelines (movement plans) and financial management guidelines. To measure FOI we assessed adherence to procedures related to the seven functions of the redesigned CHF reflected in our TOI.

The questionnaire consisted of several items related to each of the outcome variables (adoption and FOI). To provide a clear picture about larger theoretical constructs under study and not to focus on single items included in the questionnaire, not meaningful in and out of themselves, composite indices were calculated based on an unweighted sum of scores of the items. Items with high face validity that measure the same construct were selected [42]. Thereafter a composite score (sub-scale) or index was obtained from the sum scores of the items. Sub-scales related to adoption were adoption of CHF structures, adoption of CHF guidelines, and adoption of the CHF processes and adoption of job-description.

Subscales related to FOI (measured as adherence) were 1) recruitment and training; (2) materials for the program; 3) remuneration of actors; 4) monitoring progress; 5) addressing CHF benefits; 6) promotion to attract enrolments; and 7) addressing the quality of health care.

It is worth to note that overall indices (composite scores) for adoption and FOI were obtained as sum scores of the sub-scales related to the adoption and FOI respectively.

Thereafter, using the median as a cut off point, adoption and FOI indices were categorized into high and low scores.

Explanatory variables here also called “modifiers of the implementation outcomes” were socio-demographic factors (Age, categorized into <=40 years and >  40 year, Sex categorized as male and female, Occupation categorized as CHF service providers, CHF fund managers and others, Level of education categorized as secondary and below and post secondary, and years of schooling categorized as <=15 years and >  15 years), Time worked in the scheme categorized as <= 20 months and >  20 months, and Time stayed in the district categorized as <=7 years and >  7 years.

In our statistical analyses, we were aided by STATA version 13 (STATA Corporation, College Station, TX, USA) through which both descriptive and detailed analysis techniques to determine the relationship between variables were applied.

Descriptive analysis based on frequency, percentages and measures of central tendency or dispersion (mean, median or standard deviation and range) were used to understand the situation related to the adoption and implementation of the scheme. We used a chi-square test to determine the association between implementation outcomes and modifiers and a p-value of less than 0.05 was considered significant. The unit of analysis were the respondents of the questionnaire.

The document review checklist consisted of several items grouped together per theoretical constructs intended to be measured. We assigned an arbitrary weight of score to each item. The weight scores were from 1 to 10 (or 10–100 in percentage). The summation of each item score lead to a composite variable, namely adoption and FOI index for adoption items and FOI items respectively. The analysis of data from the document review was mainly descriptive with an explanatory component in the qualitative part of the study. We conducted descriptive analysis based on frequency, percentages and measures of central tendency or dispersion (mean, median or standard deviation and range) to understand the situation related to the adoption and implementation of the scheme. The unit of analysis were the districts.