Migraine epidemiology and comorbidities in Southern Israel: a clinical database study in a universal health coverage setting

Clinical setup

The structure of the health care system in Israel is based on a universal coverage system providing primary care through 4 health maintenance organizations (HMOs). The National Health Insurance Law mandates that all citizens residing in Israel join 1 of 4 official non-profit HMOs that are prohibited by law from denying membership. The Clalit Health Services (CHS) divides Israel into a number of geographic regions, and residents within each region have similar access to health services. To eliminate interregional heterogeneity [12], we included study patients residing in the southern district of Israel, the Negev region. The largest city in that region is Beer-Sheva, which is considered the capital of the Negev region. Overall, 8.2% of the Israeli population live in this region; 75% are Jewish and 25% are Bedouin. Municipal communities within the southern district are ethnically homogenous.

CHS, the largest Israeli HMO, is also the largest health care provider in the Negev region, covering approximately 67% of its 730,000 residents (and 50% of the total population in the country), with primary clinics available in every city, town, or settlement. Soroka University Medical Center (SUMC) is a tertiary 1100-bed medical center, with > 65,000 hospitalizations and about 200,000 emergency department visits annually, and is the largest regional hospital in southern Israel. SUMC is also a part of the HMO's hospital network. This unique setup of 1 hospital in 1 large geographic area facilitates a close follow-up and population-based assessment, with minimal referral bias, such that virtually no patients are lost to follow-up.

Study population and data collection

We performed a population-based, retrospective, observational cohort study. Adult (≥ 18 years) patients with migraine were identified in the computerized database for the southern district of the CHS. Patients with migraine were identified based on recorded physician diagnosis (International Classification of Diseases, Ninth Revision [ICD-9]) of migraine (with or without aura) and/or claims for specific anti-migraine medication (triptans) between 2000 and 2018.

We built our cohort in two steps: we first identified patients with a diagnosis of migraine made by a physician (16,675 patients with or without a triptan prescription), and then from those without a diagnosis made by a physician, we identified patients with a triptan prescription (14,091 patients). A high percentage of patients with a recorded diagnosis of migraine were also treated with triptans (71.8%). The study population’s flow is shown in Fig. 1.

Fig. 1figure 1

Physician-assigned migraine diagnosis was given by either a primary care physician or neurologist. Migraine diagnoses assigned by primary care physicians were found to be highly reliable. The Landmark Study demonstrated that clinic-assigned diagnosis of migraine was validated by an expert panel based on diary data in 98% of cases [13].

Triptans are migraine-specific abortive drugs [14]. The only labeled indication for this drug class in Israel is the acute treatment of migraine. Triptans are sometimes used off-label for the treatment of cluster headache [15]; however, its prevalence (0.1% of the general population) is negligible compared to that of migraine. Therefore, we considered triptan-prescribed patients as being diagnosed with migraine even when such a diagnosis was not recorded.

Patient characteristics, including demographics (gender, age at diagnosis, ethnicity, family status, education, social state score, and immigrant status) and clinical history (comorbidities, medications, hospitalization notes, diagnostic imaging, and primary care physician visits), were collected from a central computerized database of the CHS system.

For socioeconomic characteristics, we specifically used the socioeconomic index (SEI), scoring each municipality according to 14 variables (average monthly income, vehicle class, percentage of new vehicles, percentage of people with a high school diploma, students, percentage of work searchers, percentage of people with minimal monthly income values, people with more than double the average monthly income, median age, dependency ratio, percentage of families with ≥ 4 children, percentage of unemployment benefit recipients, beneficiaries of income support, and recipients of old-age pension) on a scale of 1 to 255 (where 1 is the lowest score). Based on this score, municipalities were aggregated into SEI clusters of 1 to 10.

Statistical analysis

We present data summaries of the main variables using descriptive statistics in the form of means and standard deviations for normally distributed quantitative variables, medians and ranges for non-normally distributed quantitative variables, and distribution in percentage for qualitative variables.

For univariate analyses, we used appropriate statistical tests. A chi-square test was used for categorical variables, with a Fisher’s exact test when needed. Continuous variables were compared using t tests for normally distributed variables and a Mann–Whitney U test for non-normally distributed variables. Univariate analyses were mostly used for the analysis of initial datasets that consisted of personal data records.

To inform current management practices and potential gaps in the management of migraine in this clinic population, we stratified the migraine population by the source of the migraine diagnosis (physician diagnosis and triptan prescription) and compared the rates of use of acute specific (triptan; Anatomical Therapeutic Chemical [ATC] Code NO2CC01-7) and non-specific medication (combination pain drugs including Acamol Focus, Excedrin, Migraleve, and Rokacet Plus; ATC Code NO2BE72, NO2CX50 and opioid drugs; ATC Code NO2AJ17, NO2AX02 and NO2AA55). Medication rates represent acute use at least once during the study period for each indication.

In order to estimate the diagnosis rates in southern Israel compared to the prevalence data accepted in the literature [1], we used indirect age adjustment. The study population was stratified according to age categories (5-year intervals). In each age stratum, we reported the number of observed migraine patients at the end of 2018 (end of study period), the total number of individuals (residents in the southern district insured by the CHS), the observed point prevalence of migraine per 10,000 adults, as well as the standard, age-specific prevalence of migraine per 10,000 adults derived from available tables of the reference population. The expected number of migraine patients in all age categories was calculated by multiplying the standard prevalence of migraine by the number of individuals in each age stratum. This number represents the prevalence of migraine that the study population would have experienced if it had the same age-specific prevalence rates as the reference population.

We estimated the annual migraine incidence rate per 1000 adults in the entire population over the 18-year period using the total number of incident cases in the at-risk population. Then we calculated the incidence for each interval of 5 years of age and for each gender. The population at risk of migraine comprised all subjects who had neither a recorded physician diagnosis (ICD-9) of migraine (with or without aura) nor a claim for a specific anti-migraine medication (triptans).

To compare the prevalence of comorbidity between the migraine and non-migraine samples, we used two approaches. First, we matched migraine patients 1:2 with non-migraine controls by gender, age, and primary clinic (patients are assigned to clinics based on place of residency, which correlates with socioeconomic status) and performed a univariate analysis for each comorbidity. Second, multivariable binary logistic regression models were performed to assess differences in the likelihood of each comorbid condition in the entire study population (a total of 465,750 patients) as a function of the presence of migraine diagnosis, adjusting for gender, age, and primary clinic. Odds ratios (ORs) and 95% confidence intervals (CIs) are shown along with P values.

For all analyses, a 2-sided P value < 0.05 was to be considered statistically significant. All analyses were performed using RStudio, version 1.4.1717.

Ethics approval

The study was approved by the SUMC ethics committee, reference number 0284-19. All clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki. The ethics committee approval exempted the study from informed consent due to the retrospective data collection that maintained subject confidentiality. Informed consent was waived by the institutional review board at SUMC. Patient records were anonymized and deidentified prior to analysis.

留言 (0)

沒有登入
gif