Scientific Productivity and Population Health: Correlating Research Output with Mortality Rates in Arab Countries

Introduction

The United Nations has declared 17 sustainable development goals (SDGs) for the year 2030, SDG 3 is to “ensure healthy lives and promoting well-being for all at all ages,” which includes reducing the global maternal mortality ratio, ending preventable deaths of newborns and children, ending epidemics of communicable diseases, and reducing mortality from non-communicable diseases.¹

Many studies have investigated the factors affecting changes in mortality rates, including the Human Development Index (HDI),^2,3 Gross Domestic Product (GDP),⁴ health expenditure,⁵ and research.^6,7

The impact of research has been shown to function through four avenues: Research related impacts, Policy Impacts, Service impacts, and Societal impacts. Policy, service, and societal impacts are most likely to be at work when looking at improvements in mortality, as they represent improvements in healthcare policy, healthcare quality, and knowledge and awareness in the population, all of which can impact mortality directly.⁸

Research output has been measured in many ways, including the total number of published articles,⁶ articles per million persons (PMP), articles per US$1000 gross domestic product,⁹ research funding, and the number of patients recruited for clinical trials.⁷ Weighted citation ratios have also been used and shown to be the most suitable measure of research output.¹⁰ However, they have rarely been used in studies analyzing country-level data, where such a measure would be ideal.

This study aimed to determine whether a significant correlation exists between research output (as measured by the number of articles published and their field-weighted citation impact) and the top ten causes of mortality and mortality per 100,000 individuals due to those causes.

Materials and methods

The analysis started with all Arab countries (members of the League of Arab States). Countries must have maintained at least a classification of Lower Middle Income according to the World Bank during the period–2000-2019. Comoros, Mauritania, Somalia, Sudan, Syria, and Yemen did not meet this criterion.

Countries also needed mortality data in the WHO database, and Palestine did not meet this criterion.

The final criterion was having at least one article published in each field per country; Djibouti and Libya did not meet this criterion.

The countries that remained in the analysis were Algeria, Bahrain, Egypt, Iraq, Jordan, Lebanon, Morocco, Oman, Qatar, Saudi Arabia, Tunisia, and the United Arab Emirates.

Mortality data for the top 10 non-injury-related causes of death per 100 000 and the population of each country each year were acquired from the WHO Global Health Observatory.¹¹ The top ten causes of death in 2000 were followed for the duration of interest (2000-2019) for all countries.

Search terms were prepared for each of the causes of death in these countries by referring to the WHO methods and data sources for country-level causes of death 2000-2019¹² and comparing with the relevant codes from the ICD-10 database.¹³

A Scopus search was conducted using Python 3.11.4,¹⁴ addressing the Scopus API service. The search included search terms in the “TITLE-ABS-KEY” field, and only articles with an affiliation country being the country of interest were included. Articles and reviews were selected by adding “ar OR re” into the “DOCTYPE” field. The subject area was limited to medicine by inserting the term “MEDI” into the “SUBJAREA” field. Arabic results were excluded using the “LANGUAGE” field to ensure that the results could be parsed. Articles in English were included, articles in French were also added for the countries of Algeria, Lebanon, Morocco, and Tunisia.

The search was further refined using the Scopus API to access the affiliation country of the first author and match it to the country of interest to increase the likelihood that a certain article would benefit the country of interest and not be a result of international collaboration in a study that does not investigate the public health situation of the country of interest. The search was conducted on September 28^th, 2023.

After the search, each individual list of EIDs was uploaded to SciVal, where key metrics were evaluated, namely, the number of articles published per year and the Field Weighted Citation Impact (FWCI) (excluding self-citations). The uploading and retrieval of SciVal data were performed on October 4^th, 2023.

The collected data were then imported into RStudio,¹⁵ and statistical analysis was conducted using the R programming language¹⁶ with the following packages: readxl, ggplot2, and ggpubr.^17–19

The mortality rate per 100 000, number of articles published, and the FWCI for those articles for each country were aggregated year by year.²⁰

The year-by-year aggregates for the number of articles and the FWCI were tested for correlation with the aggregated mortality numbers. This was done using Spearman’s correlation coefficient. This test was utilized as the data did not meet the assumption of Pearson’s correlation coefficient of being normally distributed.

To assess the trend in the mortality rate for each country, another correlation test was conducted to measure the relationship between the mortality rates per 100 000 and the year.

This study was exempt from IRB approval, as it was conducted on openly available data.

Results

Data from 13 countries were analyzed in this study, and 36 803 articles satisfied the inclusion criteria.

The results from the correlation tests for each country are shown in Table 1.

The countries showed great variability in their respective coefficients, with values ranging from -0.902 for Iraq to 0.899 for Lebanon for the relationships between the number of articles published, and from -0.788 for Saudi Arabia to 0.639 for Lebanon for the relationships of the FWCI.

Ten countries had statistically significant coefficients for the relationship of the number of articles published, while only six countries demonstrated statistically significant coefficients for the relationship of the FWCI.

The results of the correlation tests conducted on the mortality rates for all ten causes are shown in Table 2.

All but four countries showed a declining trend in their mortality rates, with six showing statistical significance and coefficients with magnitudes greater than 0.8.

The countries with an increase in their mortality rates had statistically significant relationships with strong correlations.

The results of the correlation tests between ischemic heart disease and stroke-specific mortality rate and time are shown in Table 3.

Six countries showed statistically significant positive correlations with IHD specific mortality, but only four did for stroke-specific mortality. The four that did (Algeria, Lebanon, Morocco, and Tunisia) were also members of the group with an increase in IHD mortality.

The number of articles per million persons (PMP) in 2000 and 2019 are shown in Table 4.

In 2000, most Arab countries were quite close in terms of articles PMP, with Kuwait being almost double the second highest, and Algeria and Iraq with articles PMP less than 1.

All countries have improved since then, with Qatar gaining the number one spot with a value almost double that of the second highest. Iraq has improved drastically, while Morocco and Algeria have not improved much, although a large improvement is noticed when looking at the percentage improvement.

The total number of articles published each year for all countries is plotted against the year in Figure 1 to show the trend in research output.

Figure 1. Annual total number of publications from 2000 to 2019.

A 560% increase was observed in the number of articles published on the top ten causes of mortality for each country. From 540 articles published in 2000 to 3567 in 2019.

Discussion

Research output in Arab countries has shown notable growth over the past 20 years, as shown in Figure 1.²¹ Along with this increase, mortality due to communicable diseases and maternal and newborn issues has decreased significantly, while mortality due to noncommunicable diseases has increased.²² The factors causing this phenomenon remain a topic of debate.

The correlation between the two measures of research output (number of articles published and FWCI) and mortality rate was measured for each country.

Most of the included countries (nine) showed negative correlations between the number of articles produced and the mortality rate and between the FWCI of the articles and the mortality rate. This positive impact on mortality rates has been reported in several previous studies. Are et al. showed that cancer-specific mortality rates in most countries worldwide were negatively correlated with the number of cancer-related articles published in those countries.⁶ Pons et al. showed that weighted citation ratios are good indicators of mortality in Spanish public hospitals.¹⁰

The exact mechanism of this relationship is still debated, with public health policy impacts and direct healthcare service impacts being proposed candidates.⁸ Mediation analysis studies have shown that for conditions such as ovarian cancer, increases in research output improve mortality rates through better utilization of surgical procedures and chemotherapy.²³

There were positive correlations between the number of articles published and mortality rates in Algeria, Lebanon, Morocco, and Tunisia, as well as between FWCI and mortality rates. The magnitudes of the correlation coefficients for the number of articles published ranged from 0.629 in Algeria to 0.899 in Lebanon ( Table 1). This indicates strong to very strong correlations between the two variables.²⁴ Only two of these countries had coefficients greater than 0.6 for the FWCI relationship but they all had positive coefficients.

These relationships were unexpected, as Are et al. showed that countries with an increase in cancer-specific mortality had a stronger negative correlation with the number of articles produced than those with a decrease in mortality.⁶

A possible explanation for these findings is related to the specific burden of the disease in these countries. Notably, these were the only countries in the analysis that displayed a significant increase in mortality due to both ischemic heart disease and stroke. This is an important observation, as ischemic heart disease was the number one cause of death for all countries included in the analysis, and stroke appeared as the 2^nd or 3^rd cause of death in almost all countries.

In the case of Morocco, Elyamani et al. reported “alarming” incidence rates for IHD and stroke as well as a large incidence of cardiovascular risk factors present in the population and an unprepared healthcare system.²⁵ Kharbach et al. described a long prehospital delay in patients with ischemic stroke.²⁶

In Tunisia, research output did not seem to be a major factor in this relationship. Instead, it seems that Tunisia is experiencing a major increase in coronary heart disease mortality due to a large increase in modifiable risk factors, such as systolic blood pressure, cholesterol levels, and BMI.²⁷ They also mentioned that efforts are focused on treatment and that they must become more comprehensive to alleviate this issue.²⁶

Lebanon is also quite similar, with risk factors such as smoking²⁸ and diabetes mellitus²⁹ being widespread in the population. Genetic predisposition is also prevalent in the Lebanese population, which has high rates of familial hypercholesterolemia²⁹ owing to the presence of founder mutations.³⁰ Another issue has been observed in Lebanese healthcare, with a workforce shortage of nurses impacting patient outcomes.²⁸ Halabi et al. also mentioned the presence of gaps in the adherence of primary care physicians to cardiovascular disease prevention guidelines and the need for standardized Lebanese cardiovascular disease guidelines.³¹

A possible explanation for the situation in Algeria is its very low research output compared with that in 2019, with 0.86 publications per million persons (PMP), which is an extremely low value. The next two countries with the lowest articles PMP were Morocco with 3.84 (a 346% increase over Algeria), and Egypt, with 14.38 (a 1 572% increase over Algeria). This may be an indicator of a low prioritization of biomedical research in the country or that the research being conducted does not align with the country’s disease burden, representing a “research waste” phenomenon.³² It is also plausible to assume that a certain minimum threshold for publications is necessary for the effects of research to be translated into real-world effects, and that Algeria has not yet reached that level.

This lack of research may also be a contributing factor to the relationship observed in Morocco, as it had the second lowest number of articles with a PMP of 3.84, which may indicate a similar problem in the output of research as Algeria, albeit to a lesser extent.

Conclusion

Overall, there appears to be an inverse relationship between research output and mortality rates in Arab countries, with many countries increasing their research outputs and achieving improvements in their mortality rates. Research quality, as measured by the FWCI, was also strongly inversely correlated with mortality rates, indicating the importance of this metric in research evaluation. The causative nature and mechanism of this relationship are not well understood and will require research to be conducted examining each country individually and including as many causes of mortality as possible while analyzing articles individually to ensure their relevance. As this relationship was not observed in countries seemingly suffering from certain non-communicable disease (i.e., cardiovascular disease) crises, it is imperative that these countries focus on improving their healthcare systems with what they already know (published literature) as well as working on specific national research agendas to ensure the targeting and improvement of their most burdensome conditions. It is also necessary to measure whether a certain threshold exists for the proposed positive benefits of research to manifest, which would be beneficial in designing the aforementioned national research agendas for each country.

Statements and declarations