Persistent headaches one year after bacterial meningitis: prevalence, determinants and impact on quality of life

Little is known on headaches long-term persistence after bacterial meningitis and on their impact on patients’ quality of life. In an ancillary study of the French national prospective cohort of community-acquired bacterial meningitis in adults (COMBAT) conducted between February 2013 and July 2015, we collected self-reported headaches before, at onset, and 12 months (M12) after meningitis. Determinants of persistent headache (PH) at M12, their association with M12 quality of life (SF 12), depression (Center for Epidemiologic Studies Depression Scale) and neuro-functional disability were analysed. Among the 277 alive patients at M12 87/274 (31.8%), 213/271 (78.6%) and 86/277 (31.0%) reported headaches before, at the onset, and at M12, respectively. In multivariate analysis, female sex (OR: 2.75 [1.54–4.90]; p < 0.001), pre-existing headaches before meningitis (OR: 2.38 [1.32–4.30]; p < 0.01), higher neutrophilic polynuclei percentage in the CSF of the initial lumbar puncture (OR: 1.02 [1.00–1.04]; p < 0.05), and brain abscess during the initial hospitalisation (OR: 8.32 [1.97–35.16]; p < 0.01) were associated with M12 persistent headaches. Neither the responsible microorganism, nor the corticoids use were associated with M12 persistent headaches. M12 neuro-functional disability (altered Glasgow Outcome Scale; p < 0.01), M12 physical handicap (altered modified Rankin score; p < 0.001), M12 depressive symptoms (p < 0.0001), and M12 altered physical (p < 0.05) and mental (p < 0.0001) qualities of life were associated with M12 headaches. Persistent headaches are frequent one year after meningitis and are associated with quality of life alteration. NCT01730690.


Introduction
Community-acquired bacterial meningitis (CABM) is a rare but very severe event.It occurs in 1 to 2 per 100 000 inhabitants in occidental countries each year [1,2], and its mortality rate is approximately 17% [2,3].Longterm complications such as cognitive impairment, motor sequelae, seizures and hearing loss are also described, and guidelines suggest to perform a neurological clinical examination before and one month after discharge in adults and children [4][5][6][7].Hearing tests during the first hospitalization and every 3 months in children are also mandatory [8].Recent data on long-term patients' assessment are limited, particularly in adults, and factors associated with long-term neurological or sensory outcomes are not well defined [5,9].
We recently performed a large, multicentric, prospective cohort study assessing the in-hospital and one-year outcomes after CABM [3].We found that the impact was significant in terms of handicap and quality of life one year after the meningitis episode.Persistent headaches seemed to be very frequent in surviving patients and to highly contribute to this negative impact on quality of life but were not specifically analysed during this first study.
Post-meningitis headaches are well defined in the international classification of headache disorders upgraded in 2018 (ICHD-3) by the international headache society, as headaches persisting more than 3 months after a meningitis episode [10].However, little is known about their epidemiology and determinants.Identifying preventable factors associated with the occurrence of persistent headaches (PHs) following CABM could be of great interest if headaches contribute to the lower quality of life experienced by patients one year after meningitis.
Thus, we analysed data from the French cohort to specifically study post-meningitis-related persistent headaches one year after CABM.The incidence, risk factors and potential impact on quality of life were assessed.

Study design
The French prospective cohort COMBAT is a multicentre prospective cohort involving 69 hospitals in France designed to assess the epidemiology of CABM and has been described in detail elsewhere [3].In brief, adult patients (≥ 18 years) with a confirmed diagnosis of CABM were included between February 2013 and July 2015.Epidemiological, clinical, biological, microbiological and therapeutic data were prospectively recorded at admission, during the in-hospital period, at discharge and at 12 months.The presence of pre-existing headaches before CABM was also recorded.Month 12 data were obtained using a standardized questionnaire during a specific consultation by telephone contact (with the patient, a relative or the general practitioner if the patient could not answer himself) or with the help of a paper questionnaire sent by mail if the patient was not able to answer by phone.The presence of headaches at 12 months and their intensity and frequency were systematically recorded.Patients were divided into two groups: patients with headaches at 12 months (PHs group) and patients without headaches (NH group, for No Headache).Among the PHs group, we identified the patient with or without pre-existing headaches.Other collected data were vital status, disability, general condition, neurological condition, depressive symptoms, hearing loss and health-related quality of life.

Headache definitions and neurological assessment at 12 months
Patients were considered to experience PHs if they reported any persistent pain or uncomfortable sensation of the head at month 12. Patients were asked about the intensity of headaches, which was dichotomized based on the values reported on a numerical scale ranging from 0 to 10.The intensity was categorized as mild (1-3), moderate (4-6), or severe (7)(8)(9)(10).Additionally, patients were asked about the frequency of headaches, which was dichotomized as occasional or regular, according to the third International Classification of Headache Disorders (ICHD-3) published by the International Headache Society (HIS) [10].
At discharge of meningitis hospitalisation and at month 12, neuro-functional disability was assessed by the Rankin-modified score as follows: 0 = no symptom or no symptom other than headaches, 1 = low disability, 2 = mild disability, 3 = moderate disability, 4 = moderately severe disability, 5 = major disability, 6 = death [11].At month 12, the following data were also assessed: Glasgow Outcome Scale (GOS) (1 = death, 2 = vegetative state, 3 = severe handicap, 4 = moderate handicap, 5 = good recovery) [12]; Depressive symptoms using the Center for Epidemiologic Studies Depression (CES-D) scale, with a threshold of 23 for women and 17 for men [13,14]; Quality of life using the SF12 score, with two composite scores: the Physical Component Summary (PCS) and the Mental Component Summary (MCS) HRQL scores; these scores range from 0 to 100, with a high value indicating good HRQL (healthrelated quality of life) [15,16].To better assess the impact of headaches on the neurological status, they were excluded from the Rankin-modified score and considered as minor symptoms for calculating the GOS score.If headaches were the only symptom reported, the GOS score was 5, and the mRS score was 0.

Statistics
First, a descriptive analysis was performed on the entire cohort of surviving patients with available data at 12 months.Categorical variables were summarized as counts (percentage), and frequency distributions were compared between PHs group and NH group with Chi-square test or Fisher's exact test as appropriate.Continuous variables were expressed as the median [IQR], and differences were tested between PHs group and NH group with the independent t-test for normally distributed variables or the Mann-Whitney U test otherwise.
Second, we analysed factors at the time of meningitis associated with persistent headaches at 12 months among following variables: the patient's background characteristics, initial clinical presentation, biological results at inclusion, causative microorganisms and the initial treatments.All variables with a p-value < 0.20 in the bivariate analysis were entered into a multivariate logistic regression model with a backward variables' selection approach and a significance level at 5%.We performed these analyses in the entire cohort and then in the population of patients who did not report pre-existing headaches.
Finally, we analysed in bivariate analysis the association between persistent headaches at 12 months and concomitant M 12 Glasgow Outcome Scale, depressive symptoms, and quality of life.All statistical analyses were performed using SAS version 9.4 software (SAS Institute Inc, Cary, NC).

Ethics and regulatory issues
The COMBAT study was registered with ClinicalTrials.gov (NCT 01730690) and received ethics approval by the Comité de Protection des Personnes Ile de France CPP 4 (IRB 00003835) (2012-16NI) and the CNIL (commission nationale de l'informatique et des libertés) (EGY/FLR/ AR128794).
At 12 months, 86/277 (31.0%) patients reported persistent headaches.Among them, 39 (45.4%) had preexisting headaches before the CABM event, while 46 (54.6%) did not report headaches prior to meningitis.These new-onset and persistent headaches affected 24.6% of the patients who did not have headaches before meningitis (46 out of 187).Headache frequency was available for 76 patients, with 53 (63.9%) reporting occasional headaches and 23 (30.3%) reporting frequent headaches.No difference was documented in terms of M12 headache frequency and intensity between the patients with or without pre-existing headaches.Forty-eight patients of 87 who reported preexisting headaches declared no longer suffering from headaches at M12.For the other patients with pre-existing

Bivariate analyses
The results of the bivariate analyses are presented in No association was found between M12 headaches and the meningitis causative microorganism, antibiotics used, corticosteroid used, and the time interval between hospital admission and antibiotic initiation.Among the different meningitis-related complications, only brain abscesses were statistically associated with a higher probability of 12-month headaches, documented in 8.1% and 2.6% in the PHs and NH groups, respectively (p < 0.01).

Multivariate analysis
In

Association between M12 headaches and concomitant patient characteristics
The presence of headaches at 12 months was significantly associated with an altered M12 neurological condition: the Glasgow Outcome Scale was more often altered (36.1% of patients with a score < 5 in the PH group vs 20.9% in the NH group, p < 0.01;  **** SF-12: the calculation of the scores is such that a higher score corresponds to a better health."Two composite scores can be derived from the SF-12 Health Survey: a Physical Component Summary (PCS) and a Mental Component Summary (MCS) HRQL score.These scores range from 0 to 100, with a high value indicating good HRQL.An individual was defined as having a "normal" physical (or mental) HRQL if his PCS (or MCS) was higher than the 25th percentile of the distribution of the score in the general French population of the same age group and gender, using an existing approach to clinically interpret the results»

Discussion
Our study provides interesting information regarding persistent headaches (PHs) 12 months after bacterial meningitis in adult patients: i) PHs are frequent, observed in 31% of patients in general and in 25% of patients who did not report pre-existing chronic headaches before meningitis; ii) female sex, pre-existing headaches, a high percentage of polynuclear neutrophils in the initial CSF, and the occurrence of a brain abscess during the clinical course of meningitis are associated with 12-month PHs; iii) PHs probably play a detrimental role in the quality of life of patients, since the patients experiencing PHs demonstrated worse HRQL SF-12 scores than patients without PHs.The prevalence of PHs after CABM is not well known thus far.Headaches are often mentioned as potential long-term sequelae in studies assessing the prognosis following CABM.However, they are frequently overshadowed by other neurological sequelae and are not specifically investigated.Only two studies published prior to 2000 specifically examined the occurrence of post-CABM headaches and reported rates ranging from 31% [17] to 46% [18] among patients.We found similar results, with almost one-third of patients suffering from PHs one year after a bacterial meningitis event.This is higher than the proportion of patients with PHs following other acute nontraumatic neurological events, as only 12% of patients experience PHs three years after ischemic or hemorrhagic stroke [19] and 25% after cervical artery dissection [20].However, since headaches are frequently reported in the general population [10,21], we tried to better characterize the responsibility of CABM for PHs in evaluating the frequency of incident PHs, noticed by patients without pre-existing headaches.Incident PHs were frequent and reported by 25% of these patients.We also investigated whether CABM could worsen headaches in patients with pre-existing headaches, but no differences in headache intensity or frequency before and after meningitis were noticed.Surprisingly, we found that a significant proportion of patients with pre-existing headaches (48/87) declared no longer presenting headaches 12 months after the event.One hypothesis is that CNS infectious episodes could influence the psychopathology of patients suffering from previous headaches and could modify the patient's perception of their headaches after this event.
Unfortunately, we did not find any amendable factor associated with PHs.The risk of PHs was lowered neither by a prompter diagnosis nor by optimal management of the CABM.Indeed, the time elapsed between admission and the first lumbar puncture did not influence the risk.This risk was also not lowered by rapid administration of the first dose of antibiotics.We found that PHs were more frequent among patients with pre-existing headaches and among females.These results are in line with the results of a previous study; Bohr et al. [17] found that women suffered more often from headaches than men after meningitis but without a clear explanation of this result.A high percentage of neutrophils in the initial CSF analysis was associated with a higher risk of PHs in our multivariate analysis.We were not able to clearly interpret this result, and it needs to be interpreted with caution.It could suggest a possible role of inflammation in the genesis of headaches, but CRP rates were identical between the groups, and the initiation of corticosteroid use after CSF analysis was not associated with a reduction in headache rates at month 12. Finally, the presence of brain abscess was strongly associated with PHs.Only very few patients experienced this complication, and we were therefore not able to demonstrate whether a specific management of brain abscess, e.g., surgical drainage, could lower the risk of PHs.PHs could be linked to sequel lesions, but no systematic imaging follow-up of brain abscesses was recommended in France during the study period, and we could therefore not confirm this hypothesis.
Bacterial meningitis is known to be associated with possible long-term cognitive impairments, mental retardation in childhood, deafness, and sleep disorders [2,[4][5][6][7][22][23][24][25] but not with PHs.Therefore, PHs may be considered as a less prominent sequelae to monitor during the standard patient follow-up after CABM.However, in our study, PHs were highly associated with an impaired quality of life 12 months after bacterial meningitis.We found a strong association between PHs and neurological disabilities, with worse GOS and Rankin scores in the PH group than in the NH group.There were also more patients with depressive symptoms in the PH group.Association does not mean causal relationship and based on our results, we cannot affirm that depressive symptoms were solely a consequence of headaches.However, headaches were not among the parameters used to calculate the depression scores in our study and the association found between headaches and depressive symptoms were very significant in our multivariate analysis.Thus, in line with previous studies [26,27], we think that PHs can have a detrimental role in mental and physical status, leading to depressive symptoms and, therefore, inducing an impaired quality of life.
Our study has several limitations.First, the questionnaire was not designed to specifically describe 12-month headaches.However, thanks to a prospective, systematic and codified interview at 12 months performed by a well-trained evaluator, we were able to obtain reliable data about headaches.Second, no amendable factors associated with PHs were documented in our study.Thus, we are unable to propose changes to the initial care of CABM that could lower the risk of PHs.Third, we were not able to describe the type of headaches declared by the patients and to characterize them as headaches migraines, cluster headaches or new daily persistent headaches.Only a prospective study focused on this topic could address this point.Fourth, our results cannot be generalized to tuberculous meningitis as we only included one patient with this particular form of meningitis.The pathophysiology of tuberculous meningitis is specific and differs significantly from what is observed with other bacterial infections.The risk factors for persistent headache in this specific context are yet to be determined.

Conclusion
Headaches persisting 12 months after a bacterial meningitis episode are frequent, with a potential impact on patients' quality of life.Such as rarely described in the literature, they seem to be an underestimated sequela.Similar to deafness and neurological sequelae, PHs must be systematically tracked during the follow-up of patients after CABM.A better understanding of their pathophysiology is needed to avoid them, detect them as early as possible, assist suffering patients and finally improve patients' quality of life.

Table 1
Comparison between patients with and without headaches at 12 months: determinants of headaches among baseline characteristics headaches and who still reported headaches after the CABM, neither the frequency nor the intensity was modified by the meningitis event (p = 0.49).

Table 3
** dexamethasone was used in 98% of cases (10mg*4 per day (every 6 h), during 4 days) Denominators correspond to patients for who data were obtained at baseline and at 12 months

Table 2
Factors associated with headaches in the multivariate analysisVariables included in the model: sex, antidepressant use before meningitis, headaches before meningitis, acute headaches at diagnosis, brain abscess, DIVC, total number of lumbar punctures, number of white blood cells in CSF in initial lumbar puncture, neutrophils percentage in CSF in initial lumbar puncture

Table 3
Impact of persistent headaches on neurological status