Abstract
Objective: We performed a prospective study of amyotrophic lateral sclerosis (ALS) in North Canterbury, New Zealand, from 1985 to 2006, to ascertain the incidence of ALS over that 22-year period, and to detect patterns of change in incidence. We also aimed to identify factors that influenced survival.
Methods: A prospective database of all patients seen at the Department of Neurology at Christchurch Public Hospital formed the basis of this study. Additional cases were identified through hospital coding data and from neurologists’ private practice records. Kaplan-Meier life table analysis and Cox proportional hazards analyses were used for the survival analysis. Poisson regression and capture-recapture techniques were used to analyze incidence data.
Results: ALS incidence rates steadily increased by 3% per year over the 22 years, from 1.6 to 3.3 per 100,000 per year. Older age, bulbar symptoms, and male sex adversely affected survival. The median survival from diagnosis was 17.6 months and from symptom onset 27.6 months. Contemporary supportive therapies such as noninvasive ventilation and percutaneous endoscopic gastrostomy did not extend survival. There was no disease clustering and no clues to etiology were revealed.
Conclusions: We report the highest recorded incidence of amyotrophic lateral sclerosis (ALS) to date, with the incidence of ALS in Canterbury increasing over the 22 years of the study. We were unable to confirm improvement in survival using contemporary supportive therapies and confirmed older age, male sex, and bulbar onset as adverse prognostic factors. The increasing incidence is not explained by aging of the population.
GLOSSARY: ALS = amyotrophic lateral sclerosis; CDHB = Canterbury District Health Board; CPH = Christchurch Public Hospital; EEC = El Escorial criteria; ICD = International Classification of Diseases; KM = Kaplan-Meier; MND = motor neuron diseases; NDD = neurology department database; NIV = noninvasive ventilation; PMA = progressive muscle atrophy; UMN = upper motor neuron.
Amyotrophic lateral sclerosis (ALS) is one of the motor neuron diseases (MND). It is a relentlessly progressive and ultimately fatal condition affecting the motor neurons of the brain and spinal cord, with patients often dying within 3 years of diagnosis.1–3
Studies undertaken principally in northern hemisphere countries have demonstrated a consistent annual incidence rate between 1.5 and 2.5 per 100,000 population.4–11 There have been suggestions that the incidence of ALS has been increasing over time.12 However, this is not supported by longitudinal assessments.13–15 Few studies have been performed in the southern hemisphere.16
Incidence increases with age with a peak in the sixth and seventh decade. Several studies have supported differences in incidence based on ethnicity but significant methodologic issues have hampered interpretation of this work.17 Despite the introduction of percutaneous endoscopic gastrostomy (PEG) feeding, noninvasive ventilation (NIV), and the use of riluzole, most studies have not shown improved survival.15,18,19 Subspecialty clinics have demonstrated a modest slowing of progression in ALS, but have been unable to attribute this improvement to the use of these therapies.20,21
Mean survival from symptom onset varies from 25 to 39 months.2,6,10,22–24 Adverse prognostic factors include older age and bulbar onset.
The development of uniform guidelines for the diagnosis of ALS (El Escorial criteria [EEC]) have allowed direct comparisons between studies of ALS incidence.25
We undertook a 22-year prospective study of ALS in New Zealand to determine the incidence in a southern hemisphere region over an extended period. We also assessed factors affecting survival including newer supportive therapies, and examined for case clustering that may provide etiologic clues.
METHODS
Source population.
The study population was people resident in the North and Central Canterbury Health District of the South Island of New Zealand from 1985 to 2006. Canterbury is the largest province within New Zealand, with a population of 520,280 in 2006. It is split into the North and Central Canterbury Health district, with a population of 466,494, and the South Canterbury Health District with a population of 53,877.
Subjects.
All patients included in this study were diagnosed with MND by a consultant neurologist between January 1, 1985, and December 31, 2006, and were confirmed to have met the EEC for definite or probable ALS by chart review in 2007. All patients were analyzed for survival and prognostic factors but only those resident in North and Central Canterbury were included in the incidence study. For the purposes of this study, progressive bulbar palsy was included with ALS.
Data sources.
There were three main sources of recruitment for this study. The Christchurch Public Hospital (CPH) neurology department database (NDD) prospectively recorded all neurologic disorders seen by a consultant neurologist from 1985. The NDD was established by author P.J.P. for the specific purpose of undertaking analyses such as this. In total, seven consultant neurologists working within CPH in this time period had access to this database and coded for MNDs and nearly 80% of the patients identified came from this source. These seven neurologists represented all neurologists practicing in the area over the study period. Personal databases of two neurologists’ (from CPH) private practices and from the CPH neurophysiologist’s laboratory records were included. A search of hospital discharge data from all Canterbury District Health Board (CDHB) centers from 1992 to 2006 searching for motor neuron disease, ALS, and progressive bulbar palsy using the International Classification of Diseases (ICD)-10 coding system was also performed.
The records of all patients were evaluated and the data were compiled using Microsoft Access 2003. Demographic data collected included sex, place of birth, ethnicity, and employment history. The year of diagnosis was recorded for incidence. The delay from onset of symptoms to diagnosis and time from diagnosis to death or study end was also noted. Symptoms at onset and symptoms and signs at diagnosis were recorded. Investigations, including MRI, CSF examination, CK levels, nerve conduction studies, and EMG studies, were recorded. The use of supportive treatments, particularly PEG feeding and NIV and specific treatments such as riluzole, were also recorded.
Population data for the analysis of incidence came from national population census data recorded in 1986, 1991, 1996, 2001, and 2006. As we did not have specific population figures for each year of the study, we interpolated between census points to define a best estimate of the population for each year including best estimates of gender and age proportions. Only patients who had resided within North and Central Canterbury prior to the onset of symptoms were included in incidence statistics.
Statistical analysis.
Kaplan-Meier (KM) life table analysis was used to assess all survival analyses. Cox proportional hazards analyses were used to model factors associated with prognosis and Poisson regression was used to model factors influencing incidence. Routine diagnostics were carried out to ensure model validity.
To estimate the proportion of cases possibly missed, we employed capture-recapture analysis, a statistical technique that estimates the percentage of missed cases where more than one independent data source is used.26 We selected the last 10 years of the study as nearly all cases were notified from the ICD-10 hospital coding and the NDD during this period with only a small number notified from private practice (two), and the ICD-10 coding system was implemented by the DHB at this time point. The two data sets used met the criteria for independence of datasets.26
All analyses were undertaken using Stata version 10.0.
RESULTS
Patients analyzed.
A total of 393 patients were identified from all sources as having a diagnosis of MND. Of these, 275 patients were confirmed as having a diagnosis of MND and 244 were confirmed as definite or probable ALS by EEC. Of the 244 patients with ALS, 183 came from the NDD, 34 came from private practice databases, and 27 came from the CPH ICD coding. These 27 cases had usually been seen by a neurologist in consultation while the patient was managed by a non-neurologic service. A total of 215 patients lived within North and Central Canterbury at the time of symptom onset and were included in the incidence analysis.
The main source of notifications for this study was the NDD. This contributed 233 patients with MND, of whom 183 were found to have definite or probable ALS. Reasons for exclusion included other MNDs including spinal muscular atrophy and progressive muscular atrophy,27 primary lateral sclerosis,3 or possible ALS.11 Follow-up data on many of these cases were incomplete and therefore they were excluded from the study.
Of the 118 misclassified patients, the majority came from hospital coding and largely represented misclassification of poststroke pseudobulbar palsy as MND.
Demographics.
The gender ratio was 1.2 men: 1 woman (134/110). The mean age at disease onset was 64.6 years for both genders. The mean age at diagnosis was 65.5 years. The mean age at death was 67.9 (68.1 men, 67.5 women). In total, 198 patients had a recorded date of death; 9 had no recording and 30 were alive on December 31, 2006.
In 10 cases, a family history of ALS was established (4.1%), and in two cases an association with frontotemporal dementia was noted.
There was no association between employment or area of residence (rural or urban) with ALS, and the predominant ethnicity affected was New Zealand European, reflecting the ethnicity of the Canterbury population. Only two Maori individuals developed ALS during the study.
There were no evident clusters of disease.
Clinical features.
A total of 237 cases had complete recordings of symptoms at onset, symptoms and signs at diagnosis, and disease type. Of these, 162 had a spinal onset (68.4%) and 75 (31.6%) had a bulbar onset of disease. By diagnosis (median time 10 months) 114 (48.1%) had mixed bulbar and spinal disease, while 88 (37.1%) had spinal disease and 35 had a solely bulbar disease (14.8%). Principal symptom at onset of disease is displayed in table 1.
Table 1 Principal symptom at onset of amyotrophic lateral sclerosis in 237 patients
At diagnosis, 139 patients had upper motor neuron (UMN) signs (hyperreflexia, increased tone, or extensor plantar responses). At some stage all cases developed UMN signs.
Survival.
Overall survival of the whole cohort from diagnosis is displayed in figure 1A. Median survival as measured from the KM life table analysis was 17.6 months from diagnosis. Median time from symptom onset to diagnosis was 10 months (range 1–48), giving a median overall survival of 27.6 months. Fewer than 5% of cases survived longer than 60 months.
Figure 1 Survival of patients with amyotrophic lateral sclerosis
(A) Kaplan-Meier (KM) survival curve for 244 cases of amyotrophic lateral sclerosis. (B) KM survival curves according to gender (p = 0.13). (C) KM survival curves according to age at diagnosis (p < 0.001). (D) KM survival curve according to bulbar symptoms at onset (p = 0.02).
Survival by gender is shown in figure 1B, survival by age at diagnosis in figure 1C, and figure 1D compares survival by the phenotype at disease onset (bulbar vs spinal). These data for survival from symptom onset are displayed in figure e-1 on the Neurology® Web site at www.neurology.org.
To assess whether year of diagnosis affected survival, we performed survival analysis on cohorts of patients diagnosed over 5-year intervals (data not shown). There was no correlation of improved survival with year of diagnosis (p = 0.45). Therefore, the introduction in the late 1990s of NIV, associated with the increased use of PEG feeding and riluzole therapy, has not influenced the overall survival. However, only seven patients had received riluzole therapy as it is not publicly funded in New Zealand. A total of 57 PEG tubes were inserted, the first being in 1990. The average time from diagnosis of ALS to PEG tube insertion was 10.3 months. The average time from PEG insertion to death was 7.4 months.
A multivariable Cox proportional hazards model was employed to determine whether age group at diagnosis, gender, bulbar symptoms, or year of diagnosis influenced survival. The only independent risk factor was age group <50 at diagnosis (p < 0.001). When the same analysis was performed on patients over the age of 50, male gender (p = 0.038) and bulbar symptoms at onset (p = 0.045) were significantly associated with decreased survival and this effect became more significant as age increased.
Incidence.
A total of 215 cases of ALS were resident in North and Central Canterbury at the time of symptom onset. We did not include patients with progressive muscle atrophy (PMA)/spinal muscular atrophy or possible ALS in the analysis; including these cases increased the incidence rates by approximately 0.4 per 100,000 per year. Between 1985 and 2006, the population of the region increased from 370,998 to 466,404, a 25.7% increase. The population aged over 65 increased from 44,223 to 62,688, a 41.7% increase. Table e-1 displays the population denominators and table e-2 shows the number of patients according to year of symptom onset and diagnosis by age group and sex.
Figure 2 shows the age-specific incidence rates for 10-year age groups over the 22 years, demonstrating a steady increase in ALS incidence with age peaking in the 70–79 age group for both men and women.
Figure 2 Age- and sex-specific incidences of amyotrophic lateral sclerosis in North and Central Canterbury over 22 years
Figure 3 shows the incidence per year for the whole group and those under 65 and those over 65. The overall incidence of ALS in the study region has steadily increased from 1985 to 2006, increasing on average by 3.3% per year from just 1.6 per 100,000 in 1985 to 3.3 per 100,000 in 2006. The fit lines indicate significant correlation between incidence and increasing year for the overall group and for those under and over 65. The correlation was not significant for those over 65, reflecting the overall smaller population.
Figure 3 Incidence overall and incidence in those over and under 65
One possible reason for the increasing incidence could be a result of the aging population over the course of the study. In 1985, ∼12.3% of the population were aged over 65, but by the end of the study, this proportion had increased to ∼13.8%. To investigate this, we calculated age- and gender-adjusted incidence rates for each year in the study, referenced to the European population.27 The expected number of cases per year was regressed over time. The age and gender standardized incidence rate increase persisted, growing by 3.3% per year from 1.5 per 100,000 to 3.0 per 100,000.
Overall, there is a significant increase in the incidence rate ratio of ALS diagnosis, IRR = 1.03 (95% CI, 1.01–1.05) with each study year to double by the end of the study. When the same analysis was run for symptom onset rather than diagnosis, the effect attenuated slightly, IRR = 0.027 (95% CI 0.005–0.50). This may be due to the fact that we will have missed some cases with symptom onset who were not yet diagnosed at the end of the study.
Over 65s were 9.5 times more likely to contract ALS (95% CI, 7.08–12.20) (p < 0.0001) than those in the younger age group. Additionally, there was no evidence to suggest that the rate of increase due to age was mediated by sex (p = 0.76), although being female was protective, IRR = 0.66 (95% CI, 0.51–0.87). When the data were age and gender standardized referenced to the European population,27 the effects attenuated slightly, but remained significant. The IRR 1.02 (95% CI 1.01–1.05) increased with each study year. Age over 65 continued to increase the risk of contracting ALS, by a factor of 9.36 (95% CI, 7.28–12.04) and being female (IRR 0.72, 95% CI 0.56–0.93) was protective.
Capture-recapture analysis.
As the two main sources of notification (the hospital coding system and the NDD) demonstrated the best concordance for the last 10 years of our study, with little notification from private neurologists after 1997, we performed capture-recapture analysis to estimate the number of cases of ALS that may have been missed.26 Table 2 demonstrates the breakdown of the imputed unidentified patients missed by both datasets by age group and gender. These data suggest that the putative patients not picked up by our databases could, at most, represent nine patients over 10 years.
Table 2 Capture recapture data breakdown of the imputed unidentified patients missed by both datasets by age group and gender, 1997–2006
DISCUSSION
Our study has demonstrated the highest reported incidence of ALS. Furthermore, we have also demonstrated a rising incidence over our 22-year study period, increasing from 1.6 to nearly 3.3 per 100,000 per year (the average incidence for the 22 years being 2.25 per 100,000 per year). This increase was seen in both men and women and is not explained by the aging of the population as we originally suspected and persisted when the data were age and gender standardized referenced to the European population.27
Studying rare conditions such as ALS where the reported incidence rate has varied between 1.5 and 2.5 per 100,000 can be difficult.12 We found variation in the yearly incidence rates for Canterbury between 0.25 and 4 per 100,000 per year; however, in the over 65 age group the variation was far greater, between 2 and 20 per 100,000 per year, reflecting the significantly smaller population denominator. The wide range of the variation in incidence in such an uncommon disorder as ALS emphasizes why a longitudinal study with a significant length of time is needed to confidently determine trends. The approximately 400,000 population studied is also large enough to provide the power to determine trends in incidence over time.
Our case ascertainment was as complete as possible, and if there are missed cases these would most likely be in elderly patients where a diagnosis of ALS was suspected in a number of cases but further investigation/review were declined or the person died soon afterwards without a confirmed diagnosis being made. It would be highly unlikely in our community that a person with symptoms suggestive of ALS was not seen by one of the seven neurologists except in the situation outlined above. All neurology care was centralized in one institution over the whole study period and the majority of cases were prospectively recorded, which may have improved ascertainment and therefore the observed incidence. However, despite the existence of a prospectively entered database (the NDD), this study also relied on a retrospective review of case notes and hospital coding, which may have been incomplete.
The findings of our study reveal an increasing incidence of ALS over time in contrast to the most recent studies examining trends in incidence.13,16 While this increase could have been spurious if ascertainment was incomplete, Poisson regression analysis indicates that the rate of increase has been consistent over time for both young and old, men and women and in the early and late periods of the study. Underascertainment would have reduced the overall incidence rather than increased it. Misdiagnosis and inclusion of other MNDs could potentially have increased the incidence; however, this is unlikely, as all cases were recorded by a consultant neurologist experienced in the diagnosis of ALS and all cases were confirmed in 2007 as meeting the EEC for definite or probable ALS. Where diagnostic uncertainty existed the case was excluded and this could therefore have reduced our incidence rates. Unlike some other studies, we did not include possible ALS or PMA cases, which if included would have increased our incidence rates by around 0.4 per 100,000 per year.
Capture-recapture analysis suggests that our incidence figures are a slight underestimate, with potentially one person per year over the past 10 years not picked up by our databases.
There was no evidence for migration into the area to access care, as all cases were referred for diagnosis by primary care physicians to hospital clinics or private practice neurologists.
There was no evidence for clustering in any occupational, toxin exposure, or racial group, and migration did not appear to be a factor.
Our cohort behaved in the same manner as other similar population-based studies as regards survival, clinical features, and prognostic factors, indicating that we had observed a cohort of persons with typical ALS.22–24
There was no evidence that NIV, PEG, or riluzole affected survival, with survival after gastrostomy echoing previous studies.28 However, we were not able to measure quality of life and the use of riluzole in our community was extremely limited due to funding.
Our cohort is unique for its longevity, high incidence rate, and the evidence presented for an increasing incidence of ALS over time. The observational nature of this study does not provide any clues as to the factors associated with this increasing incidence nor the high incidence reported here. It does, however, raise the question as to whether local environmental or genetic factors are driving the increase and therefore further investigation is warranted.
ACKNOWLEDGMENT
The authors thank Research Associate Professor Richard Jones for his help in establishing and maintaining the Neurology Department Database (NDD).
Footnotes
-
Supplemental data at www.neurology.org
Disclosure: The authors report no disclosures.
Received May 14, 2008. Accepted in final form August 26, 2008.
REFERENCES
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
